Proceedings of SEAGS-AGSSEA Conference 2 PDF

Proceeding 20th SEAGC - 3rd AGSSEA Conference in conjunction with 22nd Annual Indonesian National
Conference on Geotechnical Engineering. Jakarta - INDONESIA, 6 -7 November 2018. ISBN No. 978-602-17221-6-9
“Geotechnical Challenge for Mega

Infrastructures”
Editor : Masyhur Irsyam

Ikuo Towhata
Ramli Nazir
Benson Hsiung
Paulus P. Rahardjo
Pintor Tua Simatupang
Didiek Djarwadi
Hendra Jitno
Widjojo A. Prakoso
Agus Setyo Muntohar
Nurly Gofar
Farid Ma’ruf
Endra Susila
Teuku Faisal Fathani
Gouw Tjie Liong
Erza Rismantojo
HIMPUNAN AHLI TEKNIK TANAH INDONESIA

INDONESIAN SOCIETY FOR GEOTECHNICAL ENGINEERING (ISGE)
Basement Aldevco Octagon, Jl. Warung Jati Barat Raya No. 75
Jakarta 12740 - INDONESIA
PREFACE
Assalamu’alaikum Wr. Wb.
The A ssociation o f G eotechnical S ocieties i n S outheast A sia ( AGSSEA) co nsists o f t he
Southeast A sian G eotechnical S ociety ( SEAGS), V ietnamese S ociety f or S oil M echanics an d
Geotechnical En gineering ( VSSMGE), H ong Kong Geotechnical S ociety ( HKGES),
Geotechnical Society of Singapore (GeoSS), Thai Geotechnical Society (TGS), C hinese Taipei
Geotechnical S ociety ( CTGS), Indonesian S ociety f or G eotechnical En gineering ( HATTI) an d
Malaysian G eotechnical S ociety ( MGS). It w as d ecided a t t he G eneral C ommittee m eeting o f
SEAGS and A GSSEA on 31 M ay 20 16 a t K uala L umpur t hat t he 20 th Southeast A sian
Geotechnical Conference and the 3rd AGSSEA Conference (20SEAGC-3AGSSEA Conference)
is t o b e h eld a t P ullman Hotel i n Jakarta, from 6 -7 November 2018 . This i s t he f irst t ime in
history of SEAGS that the conference is held at Jakarta, Indonesia, therefore the conference is
also co njunction w ith 22 nd Annual Indonesian N ational Confe rence on G eotechnical
Engineering.
The t heme o f 20SEAGC-3AGSSEA Conference i s “Geotechnical C hallenge f or M ega

Infrastructure”, t o i ndicate how m ega i nfrastructure ra pidly c onstructed b y I ndonesian
Government. This conference will cover the wide range topics:
1. Foundation and its problem
2. Deep excavation, basement and tunnels
3. Unsaturated soil mechanics,
4. Sedimentary and residual soils
5. Soft soils and marine foundation
6. Geotechnical and earthquake engineering
7. Geotechnical instrumentation
8. Ground subsidence and sea-water intrusion
9. Vibration and earthquake effect to the structure
10. Soil improvement
11. Slope stability
12. Geo-environmental engineering
13. Finite element method
There w ill b e s pecial k eynote ad dress d elivered b y Ministry o f P ublic W ork an d Housing o f
Indonesia, 11 keynote lectures and 84 technical papers to be presented in parallel sessions.
Deep c ondolences for all v ictims d uring t he Lombok - Palu - Donggala ear thquakes.
Liquefaction, flows and deep slide shown as phenomenom during the disaster.
Special gratitude to Menard as main sponsored, Geotekindo and Bauer as co-sponsored and all
sponsored who participate in this conference.
Wassalamu’alaikum Wr. Wb.

Jakarta, 6 November 2018
Dr. Pintor T. Simatupang

Conference Chairman
i
MESSAGE FROM PRESIDENT OF INDONESIAN SOCIETY

FOR GEOTECHNICAL ENGINEERING (ISGE)
Dear our honorable guest,

Dr. B asoeki Hadimuljono-Minister of Public Works a nd Public Housing of t he Re public of
Indonesia, Prof. Cha rles N g - President of the International Society f or S oil M echanics an d
Geotechnical En gineering ( ISSMGE), honorable guests from South Eas t Asian Geotechnical
Society ( SEAGS) and Association o f G eotechnical S ocieties i n S outh E ast A sia ( AGSSEA),
board members of the Indonesian Society for Geotechnical Engineering (ISGE), Chairman of the
Engineering Commision of the Indonesian Academy of Sciences, and all Participants; Welcome
to our I nternational Confe rence. T his year w e he ld t he 20 th SEAGSC t ogether w ith 3 rd
AGSSEAC in conjunction with our 22nd ISGE annual conference.
“Geotechnical Challenge for Mega Infrastructure” become our theme to answer the challenge we
face together from current rapid infrastructure construction. Many geotechnical construction was
built in accordance with good soil mechanics and geotechnical understanding, but we need more
than that. We need all stages in construction industry can work together, from theory, practical
construction, de tailing w orks, c omprehensive s upervision, a nd of c ourse t he qua lity a ssurance
process. We need all parties to collaborate together, to move fast forward without no chance for
mistake nor fa ilure. No time for failure, ISGE/HATTI collaborate with all parties (national and
international) try to ensure there is only one way, and its forward moving.
We understand that there is still many holes in our c onstruction process, not only cause by man
made construction but also by the mother of nature. We saw how horrible the Donggala – Palu
earthquake l ast m onth an d w e m ourn f or t he cau sed; b ut as en gineer, we s tand; w e s eek t he
cause, we learn, we give suggestion and the more important, we stand still, not only as engineers
but also as a nation. We stand for the humanity-our pray for them.
Allow us to convey our gratitude to Menard, our m ain sponsored, Geotekindo and Bauer as our
co-sponsored a nd a ll s ponsored w ho p articipate i n t his event. A lso t o a ll s peakers, w riters,
participants, especially to all committee who works to their best for the success of this event.
Last b ut n ot l east, h opefully t his ev ent will bri ngs g ood a nd ne w poi nt of v iew s eeing t he
geotechnical engineering. Please enjoy the event.
Geotechnical in our heart!!!
Jakarta, 6 November 2018

Indonesian Society for Geotechnical Engineering
Prof. Ir. Masyhur Irsyam, MSCE., Ph.D.

President
ii
ORGANIZING COMMITTEE
Advisor : Prof. Ir. Masyhur Irsyam, MSE, Ph.D

Conference Chairman : Ir. Pintor Tua Simatupang, MT. Dr-Eng
Conference Co-chair : Prof. Ir. Paulus P. Rahardjo, MSCE, Ph.D
Secretary : Aksan Kawanda, ST. MT.
Treasurer : Ir. Idrus M. Alatas, MSc. Ph.D
Technical Papers : Prof. Ir. Masyhur Irsyam, MSE, Ph.D
Prof. Ikuo Towhata
Dr. Noppadol Phienwej
Prof. Chu Jian
Prof. Eun Chul Shin, Ph.D
Prof. Harianto Rahardjo
Prof. Dr. Ir. Ramli Nazir
Ir. Pintor Tua Simatupang, MT. Dr-Eng
Prof. Ir. Paulus P. Rahardjo, MSCE, Ph.D
Prof. Ir. Widjojo A. Prakoso, MSCE, Ph.D
Prof. Agus Setyo Muntohar, ST. M.Eng.Sc, Ph.D
Ir. Nurly Gofar, MSCE, Ph.D
Dr. Ir. Didiek Djarwadi, MSc
Ir. Hendra Jitno, MASc, Ph.D
Ir. Endra Susila, MT., Ph.D.
Dr. Ir. Gouw Tjie Liong, M. Eng
Ir. M. Farid Ma’ruf, MT. Ph.D
Event : Ir. Fauzi Buldan Y.

Ir. Wawan Kuswanda, MT.
Ir. Dandung S. Harninto, MT
Ali Iskandar, ST. MT.
Yunan Halim, ST. MT.
Dayu Apoji, ST. MT
Reza Ismaniar, ST. MT.
Andri Mulia, ST. MT.
Edwin L., ST. MT.
Laksamana M. Sakti, ST. MT
Merry Natalia, ST. MT.
Amelia Yuwono, ST. M.Kom
iii
Sponsorship : Anthony Setiawan, MBA

Marcello Djunaidy, ST. MT.
Secretariat : Josephine Aristiti Setyarini, ST. MT.

Aflizal Arafianto, ST.
Sugino
Publisher : Himpunan Ahli Teknik Tanah Indonesia

(Indonesian Society for Geotechnical Engineering)
Address : Basement Aldevco Octagon

Jl. Warung Jati Barat Raya No. 75
Jakarta 12740
Telp. : 021 - 7981966
Fax. : 021 - 7974795
Email : sekretariat@hatti.or.id; hattipusat@yahoo.com
Website : http://www.hatti.or.id
ISBN No. : 978-602-17221-6-9
iv
TABLE OF CONTENTS
Preface Committee Chairman …………………………………………………….. i
Message from President of Indonesian Society for Geotechnical Engineering ii

(ISGE) ......................................................................................................................
Organizing Committee ............................................................................................ iii
Table of Contents ..................................................................................................... v
Keynote Speakers :
1. Wastes Cover Wastes: A Novel, Scientific and Environmentally Friendly

Approach for Landfilling (C.W.W. Ng1, J.L. Coo1, H.W. Guo1 and B.W. Lu1 - 1Department
of Civil and Environmental Engineering, Hong Kong University of Science and Technology, HKSAR)
……………………………………………………………………………………………………................….… 1-4
2. Damages Associated with Geotechnical Problems in 2018 Palu Earthquake,

Indonesia (Masyhur Irsyam1,2,3,9, Nuraini Rahma Hanifa1,3, Andhika Sahadewa1,2, Agus Setyo
Muntohar2,4, Widjojo Adi Prakoso2,5,Dandung Sri Harninto2, Didiek Djarwadi2,3, M. Ridwan3,
Danny Hilman Natawidjaja3,6, Mudrikh Daryono3,6,Endra Gunawan1,3, Irwan Meilano1,3, Astyka
Pamumpuni1,3, Ariska Rudyanto3,7, Sigit Pramono3,7, Ramli Nazir81Institut Teknologi Bandung
(ITB)2Indonesia Society for Geotechnical Engineering (HATTI)3National Center for Earthquake
Studies (PuSGeN), Ministry of Public Work and Housing4Universitas Muhammadiyah Yogyakarta
(UMY)5Universitas Indonesia (UI)6Indonesian Institute of Sciences (LIPI)7Indonesian Agency for
Meteorology, Climatologyand Geophysics (BMKG)8Universiti Teknologi Malaysia
(UTM)9Indonesian Academy of Sciences (AIPI)................................................................................... 5-14
3. Piling Construction and Testing of Megastructures on Problematical Soil

1 11
Ground of Kazakhstan (A.Zhussupbekov and A.Omarov Department of Civil Engineering,
L.N. Gumilyov Eurasian National University, Astana, Kazakhstan (Prof. Askar
Zhussupbekov)…………..…................................................................................................................ 15-18
4. Serious Lessons Learnt from Big Project (Ikuo Towhata, Professor Emeritus,
University of Tokyo, Visiting Professor, Kanto Gakuin University) ………………………… 19-24
5. Stability Analysis of Buried Waste Water Pipeline in Soft Ground (Prof. Eun
Chul Shin-Korea University - South Korea)………………...................................................................................... 25-28
6. New Solutions to Geotechnical Challenges for Coastal Cities (J. Chu1, S.F. Wu1,
H. Chen1, X.H. Pan1, and K. S-L. Chaim2 1School of Civil and Environmental Engineering, Nanyang
Technological University, Singapore, 2Geological and Underground Projects Department, Build
Control Authority, Singapore) …………………………………………………......................................….. 29-34
7. Use of Strut Free Systems in Deep Excavations (Chang-Yu Ou1 and Aswin Lim2)
1NationalTaiwan University of Science and Technology, Taipei, Taiwan, ROC 2Department of Civil
Engineering, Universitas Katolik Parahyangan, Bandung 40141, Indonesia ……..................………… 35-38
v
8. Behavior of Single Pile and Pile Group Foundation for High Rise Buildings
on Expansive Soils (Prof. Paulus P. Rahardjo-Universitas Katolik Parahyangan,
Indonesia)………….............…………………………............................................................................. 39-45
9. Challenges in Design and Construction of Deep Excavation and Case

Histories (Dato' Dr. Ir. Gue See Sew. CEO of G & P Professionals, Malaysia)……........….......... 46-54
10. Increasing geotechnical challenges in the design and construction planning

of the Third Phase Bangkok MRT Underground (N. Phienwej1, A. Asanprakit2,
P. Kittiyodom2 and S.Timpong2 1School of Engineering and Technology, Asian Institution
of Technology, Pathumthani, Thailan d2Geotechnical and Foundation Engineering Co. Ltd.,
Bangkok, Thailand)................................................................................................................................ 55-60
11. Mega Reclamation Projects: Challenges and Lessons Learned in Soil

Improvement Works and Acceptance Tests (K. Yee1 and S. Varaksin2 1Association of
Geotechnical Societies in South East Asia (AGSSEA), Kuala Lumpur, Malaysia 2Apageo, Paris,
France).................................................................................................................................................. 61-74
Session I.A1 – Pile. (TC 212)
1. Design and Construction of Foundation System for Malaysia First

Drawbridge at Kuala Terengganu (Shaw-Shong Liew, Kuan-Seng Koo and Fong-Wah
Chee)…………………………………….…………….……………….......... 75-80
2. Advancement of Bearing Capacity and Settlement Analyses of Piled-Raft

Foundation (Sugeng Krisnanto, Iwayan Sengara and Fithrie Nur Adelina) …….…........... 81-84
3. Effect of Intermediary Weak Layer on The Behaviour of Piled Raft

(Venkatraman Balakumar, Huang Min, Erwin Oh and Arumugam Balasubramaniam)…............ 85-91
4. Analysis of In Situ Laterally Loaded Tests on Caisson Foundations

(Jiunn-Shyang Chiou and Cheng-Chang Tsai)…….......………………......................... 92-95
5. Study of Bored Pile Capacity in Klang Valley Residual Soil Based on Field-
Performance Data (Allan Chwee Yew Lun, Balakrishnan Etty Gaunder and Nazri
Ali).................................................................................................................. 96-101
6. Foundation Value Engineering & Underpinning Using Micropile for

Building Upgrading Works (Kai Ming Lee and Elly Norissya Mohd Said)................... 102-106
7. Interpretation on Performance of Two Drilled Shafts Subjected to Tensile

Loading Considering Concrete Cracking Effect (San-Shyan Lin, Tai-Hong Chen and
Chia-Hong Lai).................................................................................................... 107-110
8. Pile Settlement Uncertainty in Jakarta, Indonesia (Bondan Satria and Widjojo Adi
Prakoso)…….................................................................................................... 111-114
vi
9. Re-Evaluation of Pile Capacity due to Shallow Gas – A Malaysia Case

Study (Christian Hariady Girsang, Noorizal Nasri Huang, M Syazwan Kamil Abdullah, M Razi
Mansoor, Muhammad Joehan Rohani, Azam A Rahman and Wan M Marzuki Wan
Ismail)……...................................................................................................... 115-120
Session II.A2 - Pile & Instrumentation. (TC 212)
10. Application of Newly Developed Real- Time Website-base GIS

Monitoring in Tunnelling (Christian Luis, Johnny Huang and Bin-Chen Benson
Hsiung)............................................................................................................. 121-124
11. ERT-Based Leakage Tracing for Dam Safety and its Potential Sliding
Surface (Helsin Wang, Chih-Hsin Hu, Sheng-Hsiung Hsieh and Yao-Chu Tsai)..............…... 125-135
12. Dynamic p-y Curves for a Single Pile by 1g Shaking Table Tests (Sangseom Jeong).. 136-140
13. Design Method for Bottom Single Blade Steel Rotation Pile Foundation :
Case study in Vietnam (Duy Lam Dao and Thi Tuyet Trinh Nguyen)……….............…. 141-146
14. Active Shaking Tests of Pile Foundation Models in Dry Sand Ground
(Kohei Kenda, Anh Tuan Vu and Tatsunori Matsumoto).................................................... 147-157
15. Estimating Pile Axial Bearing Capacity by c-phi' Derived from

Pressuremeter Test (Tjie Liong Gouw)………………………………...……….. 158-167
16. Volume Measurement for Heavy Punding Induced Craters and Ground
Heave (Amalia Ula Hazhiyah, Hung-Jiun Liao and Chih-Jung Chien) ............................... 168-178
17. Potentiality of Boehmeria Nivea as Alternative Material in the Production

of Geotextile (Oliver Celis, Catalino Mendoza, Ernesto Villarica, Ederick Songahid and Ann
Krischel Hipolito) ................................................................................................ 179-183
18. A Multidisciplinary Ground Model Approach to Geotechnical and

Geohazard Site Apprisal for Large Infrastructure Developments
(D. Rushton) ...................................................................................................... 184-187
19. Prediction of the Geological Condition for Pipe jacking Base on the Data
Collected in the Shafts (Jian Shou Keh and Kan Tang Fu) ……............................... 188-192
Session III.B1- FEM. (TC 103)
20. A Case Study On Monitoring and Numerical Analysis of Groundwater

Variation and Inclinometer Displacement in Taiwan (Ching-Jiang Jeng and Chia-
Yu Yang) ........................................................................................................... 193-197
vii
21. Numerical Simulations of an Excavation Case in Jakarta by Hypoplasticity

Model for Clays (Melisa Kosasi, Fuchen Teng and Benson Hsiung) .............................. 198-203
22. 3D Finite Element Analysis of Deep Excavation in Central Jakarta using

Total and Effective Shear Strength Properties. (Tatag Yufitra Rus, Bin-Chen Benson
Hsiung and Kuo-Hsin Yang) …........………………………………..........……….. 204-212
23. Plane Strain Ratio and Waling Size Evaluation of Deep Excavation in
Kuala Lumpur Using 3D Finite Element Analysis (Jen Shen Ang, Bin Chen Hsiung
and Ching Hung).................................................................................................. 213-221
24. Assessment of Mechanical Behavior of Granular Soils Adopting Various

Plasticity Models (Deepa Patil, Anitha Kumari S D and Nimmy Mariam
Abraham)........................................................................................................... 222-227
25. Remediation of Oil Tank Using PLAXIS 3D (Anthony Gunawan)……..…..….. 228-233
26. Dynamic Analysis of Underwater Tunnels (Akhila Babu, Anitha S D and Nimmy
Mariam Abraham)……………………...……………………………………..... 234-237
27. 3D Finite-Element Analysis on Behaviours of Pile Group and Piled Raft

Foundation Models Subjected to Cyclic Horizontal Loading (Anh-Tuan Vu and
Tatsunori Matsumoto)….........…………………………………………………... 238-244
28. Finite Difference Analysis of Raft Foundations under Vertically Static

Loads (Der-Wen Chang and Hsin-Wei Lien)………………………………………... 245-250
Session IV.B2 - Soil Mechanics & Lab. (TC 101 & TC 102)
29. Determining Unsaturated Soil Properties Through Parameter Estimation

(Ibrahim Ibrahim, Didit Nur Arif and Nurly Gofar)….......……………………..……… 251-255
30. Stochastic Estimation of Consolidation Settlement of The Upper

Pleistocene Clay Layer in Osaka Bay Using a Particle Filter Method (Shotaro
Kubota1 and Kazuhiro Oda2) ………………………………………………….................…………………. 256-257
31. Volcanic Cohesive Soil Behavior under Static and Cyclic Loading (Wa Ode
Sumartini, Hemanta Hazarika, Takaji Kokusho and Shinichiro Ishibashi)............................... 258-264
32. Screw Driving Sounding Test for Soil Identification and Classification
(Aminaton Marto, Go Sakai, Naoaki Suemasa, Nor Zurairahetty Mohd Yunus, Siti Norafida Jusoh,
Nadiah Jamaludin, Muhammad Mustakim Ponimin and Muhammad Fakrulnizam Mohd
Tahir)............................................................................................................... 265-270
33. Argillaceous Rock Properties Changes Due to the Weathering Process (Idrus
M. Alatas, Masyhur Irsyam, Ramli Nazir and Pintor T. Simatupang) ..................................... 271-277
viii
34. Investigation of Aqueous Phase Liquids Migration in Double-Porosity Soil

under Isothermal and Non-Isothermal Effect (Loke Kok Foong, Prof. Ir. Dr. Ramli
Nazir and Assistant Professor Dr. Hossein Moayedi)…...............……….........………… 278-283
35. Lessons Learned from Pressuremeter Tests on Stone Columns (Richard Ong). 284-287
36. Correlations Between Gradation, Physical and Mechanical Parameters for

Material Embankments Reclamation (Herman Wahyudi and Yudhi Lastiasih) …...... 288-291
37. Characterization of Leachate Distributions at Ngipik Municipal Solid

Waste Disposal site in East Java - Indonesia (Ria Asih Aryani Soemitro, Dwa Desa
Warnana and Nila Sutra) ....................................................................................... 292-296
Session V.C1- Earthquake. (TC 203)
38. Evaluation of Remedial Works for a Spillway on Landslide-dammed Lakes

by an Earthquake, a Case Study in the Jiufengershan Landslide (Ihui Chen,
Sheichen Ho, Yushu Lin, Junyang Chen and Miaubin Su) …………......................……… 297-300
39. A Comparison Between VS30 Based- and Natural Frequency Based-Site

Amplification Factor for Three Different Types of Soil Classification
(Bonifacius Yogatama and Budiwan Adi Tirta) .............................................................. 301-306
40. Maps of Corner Period (Tc) of Response Spectra In City of Jakarta

(Delfebriyadi Delfebriyadi)...................................................................................... 307-309
41. Analysis of Bedrock Synthetic Ground Motion on Bandung City using

PSHA Method (Arifan Jaya Syahbana, Anggun Mayang Sari and Eko Soebowo) ............... 310-315
42. Influence of Cyclic Behaviour of Vibratory Pile Driving And Surging on

Pile Performance Observed in Model load Tests in Dry and Saturated Sand
Grounds (Shunsuke Moriyasu, Mako Aizawa, Tatsunori Matsumoto, Shun-Ichi Kobayashi and
Shinya Shimono)................................................................................................... 316-322
43. Dynamic Compaction at New Yogyakarta International Airport for

Liquefaction Mitigation (Ryan Rahmat Setiaji, Abi Maulana Hakim, Febrini Hartianty
Adinda and K.M. Abuhuroyroh)................................................................................ 323-328
44. Field Identification of Active Fault Nearby the Footprint of the Dam (Didiek
Djarwadi, D.H. Natawidjaja and M.R. Daryono)............................................................. 329-333
45. Seismic Analysis of El–Agrem Concrete Face Rockfill Dam (Merouane

Abdellaoui and Belkacem Moussai)............................................................................ 334-336
46. Effects of Vibrating Frequency of a Plate Compactor on Soil Density (Yung-

Show Fang)........................................................................................................ 337-340
ix
Session VI.C2 - Slope Stability. (TC 208)
47. Using Time Domain Reflectometry for Monitoring Slope Movement in the
Jiufenershan Landslide (Sheichen Ho, Ihui Chen, Yushu Lin, Junyang Chen and Miaubin Su)
....................................................................................................................... 341-345
48. Application of Data Mining Technique to Complement Photogrammetric

Roughness Data (Dong-Hyun Kim, Arumugam Balasubramaniam, Ivan Gratchev and Chul-
Ho Lee) ............................................................................................................ 346-349
49. Case Studies of a Partially Collapsed RS Wall at a Building Site (Balakrishnan

Etty Gaunder, Allan Chwee Yew Lun and Mohd Redzuan Ahmad) ........................................ 350-356
50. Numerical Simulation of Some Debris Flow Events in Central Java for
Predicting Run-out Distributions (Imam Achmad Sadisun, Rendy Dwi Kartiko and Indra
Andra Dinata) ..................................................................................................... 357-360
51. Channeled landslide Protection Using Flexible Barriers (Thomas Hangartner,

Prosida Rhapsody and Christophe Balg) ..................................................................... 361-367
52. Research on Failure of Aeolian Sand Roadbed Slope Through Laboratory

Static Load Test (Xiukun Dong and Liying Liu) ...................................................... 348-371
53. Stability Analysis of an Overall Failure Excavation Case in Hang Zhou

(Tuan Nghia Do) …………………......………………………………………... 372-374
54. Effect of the Initial Suction Boundary on the Slope Failure of Volcanic
Residual Soil (Agus Setyo Muntohar) ................................................................... 375-378
Session VII.D1 - Soil Improvement. (TC 211)
55. Applying In Situ Debris-cement Mixtures to the Ground Improvement of

Bank in Wild Creeks (Junyang Chen, Ihui Chen, Yushu Lin, Sheichen Ho and Miaubin Su).. 379-382
56. Dynamic Compaction of Lateritic Fill for Property Development (Richard

Ong and Marini Mardi)………………………………...……………………....… 383-387
57. Settlement Ratio Determination of Vacuum Preloading Soil Improvement

Technique (Case Study at Palindra Toll Project Section 1) (Herwan Dermawan,
Masyhur Irsyam, Bigman M Hutapea, Endra Susila, Rizal Sutjipto and Idwan
Suhendra).......................................................................................................... 388-391
58. The Settlement Evaluation of Improved Soft Clay Using LECA

Replacement Technique (Azhani Zukri, Ramli Nazir and Ng Kok Shien) …….....……. 392-397
59. Performance of Helix Piled Raft in Tropical Fibrous Peat Soil under Traffic
Loads (Ardy Arsyad, Ahmad Bakri Muhiddin and Lawalenna Samang) ............................... 398-401
x
60. Effect of Compaction on Liquefaction of River Sand and Sea Sand in Hai
Phong City, Vietnam (Chau Lan Nguyen, Quang Phuc Nguyen, Hai Ha Nguyen and Duc
Manh Nguyen) .................................................................................................... 402-408
61. Proposal of Permeability Evaluation Method of Suspension Grout (Kentaro

Uemura, Takamitsu Sasaki, Naoaki Suemasa, Kazuya Itoh, Koichi Nagao and Shunsuke Shimada) 409-414
62. Compacted Polymer-Enhanced Bentonite-Sand Mixture – Behaviour and

Potential Applications (Agus Setianto Samingan and Yulian Firmana Arifin) ................. 415-420
63. Geosynthetic Reinforced Road Structure as Fast Rehabilitation for a

Typhoon Disaster (Hermina Ho, Jeff Yang and Henry Sie) ......................................... 421-424
Session VIII.D2 – Soft Soil. (TC 204)
64. The use of the Observational Method in the Deep Excavations for the
Realization of a Residential Compound (Marco Carassini, Filippo Bucci and Andrea
Antiga) ............................................................................................................. 425-432
65. Reliability Assessment on Deep Braced Excavations Adjacent to High

Slopes in Mountain Cities (Runhong Zhang, Wengang Zhang, Zhongjie Hou and Wei Wang)
……………….......…………………………………………….........……... 433-438
66. Numerical Investigation of Wall Deflections Induced by Braced

Excavations in Sands (Hou Zhongjie, Zhang Wengang, Zhang Runhong and Wang Wei) .... 439-444
67. Singapore Downtown Line 3 - Tunnelling Challenges in Soft Soil and

Under Conserved Structures (Michael McGowan, Sofren Leo Suhaendi and Gordon Lee). 445-450
68. Seasonal Variation of Water Content and Pore-water Pressure Distribution

in Vegetated Soil Slope (Nurly Gofar, Harianto Rahardjo and Alfrendo Satyanaga) ........ 451-456
69. Strength Characteristics of Cement-treated Peat in Sumatera Island,

Indonesia (Hirochika Hayashi, Takahiro Yamanashi, Hijiri Hashimoto, Eddie Suraryo, Fahmi
Aldiamar, Maulana Iqbal and Dea Pertiwi) ….............................................................. 457-460
70. Basement Excavation in Soft Marine Clay in Bukit Tinggi, Klang (Sharmeelee
S.) …............................................................................................................... 461-464
71. Effect of Rapid Impact Compaction Energy on Unconfined Compressive

Strength Properties of Fill Clay Soil (Arifin Beddu, Lawalenna Samang, Tri Harianto
and Achmad Bakri Muhiddin) ….............................................................................. 465-468
72. A Study on Soft Soil Stabilization by Mass Stabilization Methods for

Construction of Dike and Rural Roads (Vinh An Phung, Van Tuan Le and The Quynh
Do) …............................................................................................................. 469-464
xi
Session IX.E1 – Deep Excavation & Tunnel. (TC 204)
73. Study for Structural Performance of Steel Sheet Piles Used for Retaining
Wall (Eka Susanto, Matsui Nobuyuki and Otsushi Kazutaka) …....................................... 475-481
74. Protection of the Existing Railway Tunnels from an Adjacent Deep

Excavation (Jung-Feng Chang, I-Chou Hu, Chun-Seng Hsu and Huei-Ting Chang)….......... 482-486
75. Bored Pile Retaining Wall Solutions for Earthquake Slip 6 AT Ohau Point,
Kaikoura, New Zealand (Senthuran Arulanantham, Anthony Fairclough, Sam Glue and Jody
Edwards)…………..........……………………………..………..................…. 487-493
76. Effects of Cross Wall to the System Stiffness of Deep Excavations in Clay
(Zih-Yun Wang, Bin-Chen Benson Hsiung, Hsii-Sheng Hsieh and Louis Ge)…...................…. 494-499
77. Ground Surface Settlement Induced by Diaphragm and Buttress Walls

Installation: Numerical Study (Aswin Lim and Pio Go Hsieh) …............................ 500-503
78. Evaluation of Surface Settlement and Lateral Displacement During Tunnel

Construction Using 3D Numerical Modelling (Fahmi Aldiamar, Desyanti Desyanti,
Masyhur Irsyam, Bigman M. Hutapea, Endra Susila, Riska Muslimah and Weni 504-508
Maulina)……………………………………..............……………………….
Session X.RE2 - Other Sessions & Indonesian Class
79. Aplikasi Metode Elemen Hingga dan Kriteria Runtuh Papaliangas Batuan
Berkekar : Problem-Problem Analisa Terowongan, Kestabilan Lereng dan
Wellbore Stability dalam Mekanika Batuan (Wilham George Louhenapessy)
....................................................................................................................... 509-514
80. The Application of Active Lime & Cement as Soil Stabilization Material
(Renggo Ginanjar, Wilham G. Louhenapessy, Asrul Ahdar) ….......................................... 515-518
81. Perbandingan Analisa Perkuatan Lereng dengan Jangkar : Program dan

Manual (Wawan Kuswaya and Wilham G. Louhenapessy) ............................................. 519-522
82. On the Weak Limestone Rock Slope Stability Analysis (Wilham George
Louhenapessy) ….............................................................................................. 523-530
83. Slope Stability Anayses using Probabilistic Approach (A Case study in

Grobogan, Central Java) (Febryansyah Dwi Riyadinata, Yusep Muslih Purwana, Niken
Silmi Surjandari and Noegroho Djarwanti) ….............................................................. 531-534
84. Perancangan Proteksi Galian Dalam Dengan Soldier Pile. Studi Kasus :
Underpass Pada Pusat Perbelanjaan di Kawasan Jakarta Selatan (Siska Rustiani
and Ryan Alexander Lyman) …................................................................................ 535-538
xii
85. Study of Slope Failure and Reinforcement Analysis to Restore and Increase
the Slope Stability in Flores, East Nusa Tenggara Using Plaxis 2D and 3D
(Rendra Priatno1, Ali Iskandar2, and Jo Lian Huat3) ……………………...............................……….. 539-543
86. Study on The Effects of Vibration due to Pile Driving by Empirical

Formulas and Continuous Monitoring (Stefanus Diaz Alvi1, Ricky Setiawan2, Andy
Sugianto3, Paulus P. Rahardjo4) ……………………………………………….....………………………… 544-548
87. Pengaruh Tahanan Ujung Tiang dan Faktor Aman Terhadap Lendutan Pelat
Terpaku Tiang Tunggal Pada Lempung Lunak (P. Anas1, O. Ferly2, M. Roza3)
………………………………………………………………………………………………..............………….. 549-543
88. Application of Distributed Fibre Optic Sensor (DFOS) in Bi-directional

Static Pile Load Tests (Lee Siew Cheng1, Tee Bun Pin1, Chong Mun Fai1, Hisham
Mohamad2 and Ang Koh An3, Paulus P. Rahardjo4) ………………………………...………………….. 554-560
89. Design of Simple Drapery Systems for Rock Cuts and Natural Slopes (Matteo
Lelli1, Alberto Grimod2, Riccardo Laneri1, Deny1, Vincent Setiawan2, Dinda Mutiara Savitri1)
………………………………………………………………………………………………….....................….. 561-568
90. Emergency Works and Landslide Remediation using Geosynthetics

Reinforced Soil Structures – Recent Indonesian Case Studies (Deny1, Matteo
Lelli2, Dinda Mutiara Savitri3) …………………………………………………………...…………………. 569-573
91. Geotechnical Properties of Cement-Stabilized Mine Tailings from Brgy.

Gango, Libona, Bukidnon, Philippines (Einstine M. Opiso, Reinerio P. Supremo,
Reymar T. Rejas and Jemima R. Perodes)……………………………………………...…………………… 574-581
92. A New Breakthrough – Application of Control Modulus Column for

Settlement and Stability Control under Soft Soil and High Embankment
Load at Pemalang-Batang Toll Road (Ryan Rahmat Setiaji1, Panji Utomo2, KM.
Abuhuroyroh3)….................................................................................................................................. 582-591
93. Estimasi Modulus Resilien dari uji CBR pada Tanah Lempung distabilisasi
Abu Vulkanis dan Kapur (Devi Oktaviana Latif - Dr.,ST.,M.Eng, Ahmad Rifa’i- Dr., Ir.,MT.,
Latif Budi Suparma –Ir.,M.Sc., Ph.D) ………………...................................…………………………….. 592-596
94. Seasonal Variation of Water Content and Pore-water Pressure Distribution

in Vegetated Soil Slope (N. Gofar1, H. Rahardjo1, A. Satyanaga1 - 1School of Civil and
Environmental Engineering, Nanyang Technological University, 50 Nanyang Avenue, Singapore.
639798) …………………………………………………………………………………………………………. 597-602
95. Efek EPS Geofoam sebagai Material Pengisi terhadap Nilai CBR
Laboratorium pada Tanah Kohesif Darah Bandung (Lestari, A. S 1 and Julian
Clementio 2 - 1 Dosen Universitas Katolik Parahyangan, Bandung, Indonesia, 2 Mahasiswa
Universitas Katolik Parahyangan, Bandung, Indonesia) ……………………...................………………. 603-606
96. Prediction of Excess Pore Pressure Due to Pile Driving Based on CPTu (A.
Arafianto1 and P.P. Rahardjo2, 1Engineering Faculty, Universitas Katolik Parahyangan Bandung,
Indonesia 2Engineering Faculty, Universitas Katolik Parahyangan, Bandung, Indonesia)................ 607-610
xiii
97. Increasing geotechnical challenges in the design and construction planning

of the Third Phase Bangkok MRT Underground (N. Phienwej1, A. Asanprakit2, P.
Kittiyodom2 and S.Timpong2 1School of Engineering and Technology, Asian Institution of
Technology, Pathumthani, Thailand 2Geotechnical and Foundation Engineering Co. Ltd., Bangkok,
Thailand)............................................................................................................................................... 611-616
xiv
Wastes Cover Wastes: A Novel, Scientific and Environmentally Friendly Approach

for Landfilling
C.W.W. Ng1, J.L. Coo1, H.W. Guo1 and B.W. Lu1
1
Department of Civil and Environmental Engineering, Hong Kong University of Science and Technology, HKSAR
E-mail: charles.ng@ust.hk
ABSTRACT: For environmental protection and sustainability, the use of plants and recycled wastes have been investigated in a landfill site
located at Xiaping, Shenzhen under humid climates. The main objective was to validate a novel vegetated three-layer landfill cover system
using recycled construction waste without the need of geomembrane in the field. One section was transplanted with Bermuda grass while the
other section was left bare. To assess the landfill cover performance, the test site was heavily instrumented and monitored for a period of 13
months under natural climatic conditions. The cumulative rainfall depth was about 2950 mm over the whole monitoring period. At the end of
monitoring period, the measured cumulative percolation was only 27 mm and 20 mm at the bottom of the bare and grassed cover,
respectively. It is evident that the vegetated three-layer landfill cover system using recycled concrete without geomembrane can be effective
in minimizing percolation at humid climates.
Keywords: three-layer landfill cover, recycled concrete, vegetation, field monitoring, water infiltration.
1. INTRODUCTION in the field. Unsieved completely decomposed granite (CDG) and

coarsely crushed recycled concrete (CC) was used for the top and
With an increasing population worldwide and a high urbanization intermediate layer while sieved CDG was used as the bottom layer.
rate mainly in developing countries, the production of municipal One section was transplanted with Bermuda grass while the other
solid waste (MSW) also increases. Many of these MSW are section was left bare (refer to Figure 1). To assess the landfill cover
construction wastes. Landfilling is perhaps the simplest, cheapest performance, the test site was heavily instrumented and monitored
and most cost-effective method to dispose MSW. In most for a period of 13 months under natural climatic conditions.
developing countries, a very high percentage of MSW goes to
landfill. Even in developed countries, many MSWs are also
landfilled. For example, more than half of the member states in the
European Union still dispose in excess of 50 percent of their waste
to landfills (EEA 2013). In the United States, 50 percent of total
waste generated are also disposed in landfills (US EPA 2015).
To minimize rainfall infiltration of water into the waste and
hence to minimize leachate to contaminate groundwater, most
modern landfill cover systems utilize geotextile composites and
geomembranes due to their low permeability. However,
geomembranes are highly susceptible to interface instability and
defects/holes which can compromise their reliability (Daniel 1994;
Koerner and Daniel 1997; Amaya et al. 2006).
Alternative cover systems such as cover with capillary barrier
effects (CCBE) have been proposed (Ross 1990; Khire et al. 2000; 20 m
Iryo and Rowe 2005; Bouazza et al. 2006; McCartney and Zornberg Field test
2010; Siemens and Bathurst 2010; Zornberg et al. 2010; Rahardjo et site
al. 2012). A CCBE typically consists of a layer of fine-grained soil
such as silt or clay over a coarse geomaterial. Field studies have
shown that CCBEs can be effective for arid and semi-arid regions in
minimizing rainfall infiltration into underlying MSW (Benson and 12 m
Khire 1995; Khire et al 1999; Khire et al. 2000; Zornberg and
McCartney 2003). However, the performance of CCBEs under
humid climates have not been satisfactory (Morris and Stormont Figure 1 The field test site at the Xiaping landfill, Shenzhen, China
1999; Khire et al. 2000; Albright et al. 2004; Rahardjo et al. 2006). (Ng et al. 2018a)
A new three-layer landfill cover system was proposed and verified
for humid climates theoretically and experimentally (Ng et al. 2015a 2. DESCRIPTIONS OF THE TEST SITE
and b; Ng et al. 2016). This new system is to add a layer of fine-
grained soil (i.e., clay) underneath a two-layer barrier with CCBE The test site was selected and constructed at the Xiaping landfill,
(i.e., a silt layer overlying a gravelly sand layer). Based on one- which is located in Shenzhen City, China. Currently, the Xiaping
dimensional (1D) water infiltration test in a soil column (Ng et al. landfill having a total area of 149 ha, is Shenzhen’s biggest landfill.
2016) and two-dimensional flume model tests and numerical The test site is located in a humid subtropical climate region, with
simulations (Ng et al. 2015a and b), it is found that no percolation approximately 80% of rainfall occurring between May and
was observed after 48 hours of constant water ponding, which is September.
equivalent to a rainfall return period of greater than 1000 years in Figure 2 shows the cross section of the test site. The landfill
Hong Kong. cover consisted of three-layers, namely a 0.8 m thick sieved CDG
To promote environmental protection and sustainability, the use (dry density of 1.73 Mg/m3), a 0.2 m thick recycled CC (dry density
of plants and recycled wastes as landfill cover materials have been of 1.89 Mg/m3) and a 0.6 m thick unsieved CDG (dry density of
investigated in a landfill site located at Xiaping, Shenzhen in a 1.77 Mg/m3) from the bottom to the top. The slope was 12 m wide,
humid climatic region of China (see Figure 1). The main objective 20 m long and it inclined at 30° to the horizontal. Half of the test
was to validate a novel vegetated three-layer landfill cover system site (6 m width) was transplanted with Bermuda grass turfs while
using recycled construction waste without the need of geomembrane the other half was left bare.
1
3. MATERIAL PROPERTIES 4. FIELD INSTRUMENTATION AND MONITORING

The CDG soil used to construct the three-layer landfill cover system The field performance was assessed by measuring percolation
was excavated from a slope near the test site. For the low through the landfill cover by lysimeters and also by monitoring the
permeability layer, the CDG soil was sieved to recover only the variations of pore water pressure and volumetric water content
fraction less than 10 mm. The recycled CC was sourced from a within the landfill cover under natural climatic conditions.
recycling plant in Shenzhen and delivered to the Xiaping landfill. Percolation through the bare and grass covered landfill cover was
The basic properties of the cover materials are summarized in Table monitored from June 2016 to July 2017. Six lysimeters (1 m
1. Figure 3 shows the particle size analyses which were obtained diameter each) for the bare and grassed landfill covers spaced at 5 m
from sieve analysis apart were installed at 1.8 m depth to monitor water percolation
through the three-layer landfill cover (see Figure 2). Each lysimeter
5
was connected to an independent drainage pipe to allow gravity flow
0.6
of the percolated water.
5 The variations of pore water pressure and volumetric water
0.4
within the landfill cover were monitored from May 2017 to July
5
0.8 2017. To assess the variations of pore water pressure and volumetric
5 water content in both the bare and grass covered three-layer landfill
cover, jet fill tensiometers (JFTs) and moisture probes were installed
at different depths (i.e., 0.2 m, 0.4 m, 0.8 m, 1.2 m and 1.6 m) within
Unsieved completely the mid cross-section of the slope.
decomposed granite
The JFTs fitted with pressure transducers were used to measure
Coarsely crushed concrete
Lysimeter pore water pressure within the range of 100 kPa to -90 kPa at an
Sieved completely
Jet fill tensiometer
decomposed granite accuracy of ± 1 kPa. Changes in volumetric water content were
Moisture probe Gas sampling measured using SM300 moisture probes. Before installation, all
Gas pressure transducer Settlement plate 30o Landfill intermediate cover
moisture probes were calibrated for the different cover materials. In
Note: addition, an automated weather station was installed on top of the
All dimensions are in meters
slope to measure the atmospheric parameters including rainfall,
Figure 2 Typical cross section view and layout of instrumentation relative humidity, air temperature, wind speed and wind direction. It
in the test site (Ng et al. 2018a) is noted that details regarding gas monitoring and settlement of the
landfill cover will not be described herein as they are beyond the
scope of this paper.
Table 1 Basic Properties of Soils and Construction Wastes Used
5. MONITORING RESULTS
Unsieved Recycled Sieved
PROPERTY
CDG CC CDG 5.1 Pore water pressure response
Unified soil classification system SC GP SC
Specific gravity, Gs 2.63 2.45 2.61 In this extended abstract, only monitored results from the grassed
Atterberg limits landfill cover (see Figure 2) are reported. Other results are reported
Liquid limit, LL 37 - 37 by Ng et al. (2018a). Figure 4 shows variations of measured pore
Plastic limit, PL 20 - 20 water pressure at different depths in the grass covered three-layer
Plasticity Index, PI 17 - 17 system from 1 May 2017 to 18 July 2017 during the wet season of
Standard compaction curve the year. As expected, the maximum changes in pore water pressure
Maximum dry density (kg/m3) 1860 1890 1820 occurred near the surface (i.e., 0.2 m depth) and the magnitude of
Optimum moisture content (%) 12.6 - 14.4
8.1x10
changes were much smaller at depths between 0.4 m to 1.6 m. Pore
Saturated water permeability (m/s) 5.7x10-5 7.5x10-2 -8 water pressures near the surface (i.e., 0.2 m depth) were the first
affected by the onset of rainfall significantly. At the start of the
monitoring period, the measured pore water pressure was about -25
kPa. The high negative pore water pressure in this shallow depth
was due to evapotranspiration. Other similar field measurements
were also reported by Lim et al. 1996; Garg et al. 2015 and Ng et al.
2018b. Upon monitoring under natural variations, it was observed
that pore water pressure at 0.2 m depth was -20 kPa on 24 May
2017. This was before the occurrence of a rainfall event with a total
depth of 78 mm. After the rainfall event, the measured pore water
pressure increased to -15 kPa. This retained high negative pore
water pressure also reduced water permeability (Ng and Menzies
2007). At 4 June 2017 (i.e., after 10 days of drying period with
minimal rainfall), the measured negative pore water pressure at 0.2
m depth was found to be -33 kPa. Even after the grassed slope being
subjected to a heavy rainfall event with a total depth of 149 mm,
pore water pressure was still retained at -17 kPa. During the entire
monitoring period (1 May 2017 to 18 July 2017), the presence of
Figure 3 Particle size distribution of the unsieved completely grass helped in retaining lower negative pore water pressure due to
decomposed granite (CDG), coarsely crushed recycled concrete transpiration, as compared to the bare soil slope (Ng et al., 2018a &
(CC) and sieved completely decomposed granite (CDG) 2018b).
2
It can be seen from Figure 4 that the variations of measured pore

water pressure at 0.4m or deeper were much less affected by
weather. The observed changes in pore water pressure range from -9
to -2 kPa during the wet season. This suggests that there was no
water infiltrated through the three-layer capillary barrier cover
system during entire monitoring period. More solid evidence can be
illustrated from the measurements by six lysimeters installed at the
bottom of the three-layer cover system.
Figure 5 Cumulative percolation in the bare and grass covered

three-layer landfill cover system from June 2016 to July 2017
3. CONCLUSIONS
A newly proposed three-layer landfill cover system was constructed
and tested in Shenzhen, China. Based on field monitoring of the
full-scale test over a period of 13 months (i.e., June 2016 to July
2017), the following conclusions may be drawn:
a) Under natural weather variations, pore water pressure at
Figure 4 Variations of pore water pressure at different depths in the shallow depth (i.e., 0.2 m) was mostly affected by rainfall
grass covered three-layer landfill cover system events. However, negative pore water pressure was
retained at all depths (i.e., 0.2m, 0.4m, 0.8 m, 1.2 m and
1.6 m) in the cover after heavy rainfall events.
5.2 Cumulative percolation measured by lysimeters b) At the end of 13-month monitoring period, the measured
cumulative rainfall was 2,950 mm, whereas the
Figure 5 shows the measured cumulative percolation in the bare and corresponding amount of percolation was 27 mm and 20
grass covered three-layer landfill cover from June 2016 to July mm for the bare and grass covered landfill cover,
2017. Cumulative percolation at three different locations (i.e., crest, respectively. The measured percolation in the new cover
middle, toe) in both slopes are included. For ease of comparison, the system clearly meets the recommended criterion of 30
measured cumulative rainfall depth with a total amount of 2,950 mm mm/year for conventional compacted clay covers in the
is also provided in the figure. During the first 6 months of US.
monitoring (i.e., June 2016 to November 2016), measured c) The results of the field monitoring validated the potential
percolation increased at a relatively steady rate and showed little use of a grassed three-layer landfill cover system using
variation in response to daily rainfall events, even those in excess of recycled concrete without geomembrane as a promising
200 mm (i.e., 19 October 2016). This indicates the effectiveness of alternative landfill cover system for humid climates.
the three-layer landfill cover system in preventing excessive
percolation through the cover. No sign of preferential flow was 4. ACKNOWLEDGEMENT
observed through both the bare and grass covered landfill covers. The authors would like to acknowledge the grant 51778166
However, following a long drying period (i.e. December 2016 to provided by the National Natural Science Foundation of China.
May 2017), percolation increased steadily by about 10 mm for both
landfill covers. This may be due to some desiccation cracks which
were observed on the soil surface at the end of this long drying 5. REFERENCES
period. Similar findings were also described by Albright et al.
Albright, W. H., Benson, C. H. and Gee, G. W. (2004). “Field water
(2006). At the end of 13-month monitoring period, the maximum balance of landfill covers.” Journal of Environmental Quality,
percolation measured for the bare and grass covered landfill cover, 33, Issue 6, pp2317-2332.
was 27 mm and 20 mm, respectively. Both landfill covers meet the Albright, W. H., Benson, C. H., Gee, G. W., Abichou, T.,
recommended design criterion of 30 mm/year for compacted clays McDonald, E. V., Tyler, S. W., and Rock, S. A. (2006).
(Benson et al. 2001). It is clear that the newly proposed three-layer “Field performance of a compacted clay landfill cover at a
landfill cover system using recycled concrete without geomembrane humid site.” Journal of Geotechnical and Geoenvironmental
can be effective in minimizing percolation at humid climates. Engineering, 132, Issue 11, pp1393-1403.
3
Amaya, P., Queen, B., Stark, T. D., and Choi, H. (2006). “Case McCartney, J. S., and Zornberg, J. G. (2010). “Effect of infiltration
history of liner veneer instability.” Geosynthetics and evaporation on geosynthetic capillary barrier
International, 13, Issue 1, pp36-46. performance.” Canadian Geotechnical Journal, 47, Issue 11,
Benson, C., Abichou, T., Albright, W., Gee, G., and Roesler, A. pp1201-1213.
(2001). “Field evaluation of alternative earthen final covers.” Morris, C. E., Stormont, J. C. (1999). “Parametric study of
International Journal of Phytoremediation, 3, Issue 1, pp105- unsaturated drainage layers in a capillary barrier.” Journal of
127. Geotechnical and Geoenvironmental Engineering, 125, Issue
Benson, C., and Khire, M. (1995). “Earthen final covers for landfills 12, pp1057 – 1065.
in semi-arid and arid climates.” Landfill closures, GSP No. Ng, C. W. W., and Menzies, B. (2007). Advanced unsaturated soil
53, R. Dunn and U. Singh, eds., ASCE, Reston, Va., 201-218. mechanics and engineering. Taylor & Francis, London and
Bossé, B., Bussiere, B., Hakkou, R., Maqsoud, A., and Benzaazoua, NY.
M. (2015). “Field experimental cells to assess Ng, C. W. W., Chen, R., Coo, J. L., Liu, J., Ni, J. J., Chen, Y. M. ,
hydrogeological behavior of store-and release covers made Zhan, T .L. T., Guo, H. W., and Lu, B. W. (2018a). “A novel
with phosphate mine waste.” Canadian Geotechnical Journal, vegetated three-layer landfill cover system using recycled
52, Issue 9, pp1255-1269. construction wastes without geomembrane.” Canadian
Bouazza, A., Zornberg, J. G., McCartney, J. S., and Nahlawi, H. Geotechnical Journal. Under review.
(2006). “Significance of unsaturated behaviour of geotextiles Ng, C. W. W., Coo, J. L., Chen, Z. K., and Chen, R. (2016). “Water
in earthen structures.” Australian Geomechanics Journal, 41, infiltration into a new three-layer landfill cover system.”
Issue 3, pp133-142. Journal of Environmental Engineering, ASCE, 142, Issue 5.
Daniel, D. E. (1994). “Surface barriers: problems, solutions and doi:10.1061/(ASCE)EE.1943-7870.0001074.
future need.” Proceedings of the 33rd Hartford Symposium on Ng, C. W. W., Leung, A. K., and Ni. J. J. (2018b). Plant-Soil slope
Health and Environment, Pasco, WA, pp441-497. interaction. Taylor & Francis. In Press
European Environment Agency (EEA). (2013). Managing municipal Ng, C. W. W., Liu, J., and Chen, R. (2015b). “Numerical
solid waste – a review of achievements in 32 European investigation on gas emission from three landfill soil covers
countries, Environmental Assessment Report No. 2, under dry weather conditions.” Vadose Zone Journal, 14,
Copenhagen. Issue 8. doi:10.2136/vzj2014.12.0180.
Garg, A., Coo, J. L., and Ng, C. W. W. (2015). “Field study on Ng, C. W. W., Liu, J., Chen, R., and Xu, J. (2015a). “Physical and
influence of root characteristics on soil suction distribution in numerical modeling of an inclined three-layer (silt/gravelly
slopes vegetated with Cynodon dactylon and Schefflera sand/clay) capillary barrier cover system under extreme
heptaphylla.” Earth Surface Processes and Landforms, 40, rainfall.” Waste Management 38, pp210-221.
Issue 12, pp1631-1643. Rahardjo, H., Santoso, V. A., Leong, E. C., Ng, Y. S., and Hua, C. J.
Iryo, T., and Rowe, R. K. (2005). “Hydraulic behaviour of soil- (2012). “Performance of an instrumented slope covered by a
geocomposite layers in slopes.” Geosynthetics International, capillary barrier system.” Journal of Geotechnical and
12, Issue 3, pp145-155. Geoenvironmental Engineering, 138, Issue 4, pp481-490.
Khire, M. V., Benson, C. H., and Bosscher, P. J. (1999). “Field data Rahardjo, H., Satyanaga, A., Leong, E. C., Santoso, V. A., and Ng,
from a capillary barrier and model predictions with UNSAT- Y. S. (2014). “Performance of an instrumented slope covered
H.” Journal of Geotechnical and Geoenvironmental with shrubs and deep-rooted grass.” Soils and Foundations,
Engineering, 125, Issue 6, pp518-527. 54, Issue 3, pp417-425.
Khire, M. V., Benson, C. H., and Bosscher, P. J. (2000). “Capillary Rahardjo, H., Tami, D. and Leong, E. C. (2006). “Effectiveness of
barriers: design variables and water balance.” Journal of sloping capillary barriers under high precipitation rates.”
Geotechnical and Geoenvironmental Engineering, 126, Issue Proceedings of the 2nd International Conference on
8, pp695-708. problematic soils, Petaling Jaya, Selangor, Malaysia, pp39-
Knidiri, J., Bussière, B., Hakkou, R., Bossé, B., Maqsoud, A., and 54.
Benzaazoua, M. (2017). “Hydrogeological behaviour of an Ross, B. (1990). “The diversion capacity of capillary barriers.”
inclined store-and-release cover experimental cell made with Water Resources Research, 26, Issue 10, pp2625-2629.
phosphate mine wastes.” Canadian Geotechnical Journal, 54, Siemens, G., and Bathurst, R. J. (2010). “Numerical parametric
Issue 1, pp102-116. investigation of infiltration in one-dimensional sand-
Koerner, R. M. and Daniel, D. E. (1997). Final Covers for Solid geotextile columns.” Geotextiles and Geomembranes, 28
Waste Landfills and Abandoned Dumps, American Society of Issue 5, pp420-474.
Civil Engineers, Reston, VA. U.S. Environmental Protection Agency (EPA). (2015). Advancing
Leung, A. K. (2016). “Grass evapotranspiration-induced suction in sustainable materials management: facts and figures 2013,
slope: case study.” Environmental Geotechnics, 3, Issue 3, Report EPA530-R-15-002, Washington DC.
pp155-165. Zornberg, J. G., and McCartney, J. S. (2003). Analysis of
Lim, T. T., Rahardjo, H., Chang, M. F., and Fredlund, D. G. (1996). Monitoring Data from the Evapotranspirative Test Covers at
“Effect of rainfall on matric suctions in a residual soil slope.” the Rocky Mountain Arsenal, Geotechnical Report, US
Canadian Geotechnical Journal, 33, Issue 4, pp618-628. Environmental Protection Agency, Region 8, December.
Zornberg, J. G., Bouazza, A., and McCartney, J. S. (2010).
“Geosynthetic capillary barriers: current state of knowledge.”
Geosynthetics International, 17, Issue 5, pp273-300.
4
Damages Associated with Geotechnical Problems in 2018

Palu Earthquake, Indonesia
Masyhur Irsyam1,2,3,9, Nuraini Rahma Hanifa1,3, Andhika Sahadewa1,2, Agus Setyo Muntohar2,4, Widjojo Adi Prakoso2,5,
Dandung Sri Harninto2, Didiek Djarwadi2,3, M. Ridwan3, Danny Hilman Natawidjaja3,6, Mudrikh Daryono3,6,
Endra Gunawan1,3, Irwan Meilano1,3, Astyka Pamumpuni1,3, Ariska Rudyanto3,7, Sigit Pramono3,7, Ramli Nazir8
1
Institut Teknologi Bandung (ITB)
2
Indonesia Society for Geotechnical Engineering (HATTI)
3
National Center for Earthquake Studies (PuSGeN), Ministry of Public Work and Housing
4
Universitas Muhammadiyah Yogyakarta (UMY)
5
Universitas Indonesia (UI)
6
Indonesian Institute of Sciences (LIPI)
7
Indonesian Agency for Meteorology, Climatology and Geophysics (BMKG)
8
Universiti Teknologi Malaysia (UTM)
9
Indonesian Academy of Sciences (AIPI)
1. INTRODUCTION The National Center for Earthquake Studies (PuSGeN) and the
Indonesia Association of Geotechnical Engineering (HATTI)
The Palu earthquake occurred on Friday afternoon, 28 September deployed a team to conduct field survey assessing impact of Palu
2018, at 18:02:44 local time (Central Indonesia Time, WITA) with Earthquake, from 8 to 15 October 2018. This paper presents findings
moment magnitude (Mw) of 7.4, centered 26 km north of Donggala, of field surveys with emphasis on geotechnical impact and fault
Central Sulawesi (Figure 1). The earthquake has caused strong rupture offset observed in the field.
shaking, resulted in a tsunami that hit Palu city and disastrous
liquefaction occurrences, particularly in Petobo and Balaroa areas.
2. SEISMOTECTONIC OF SULAWESI
Figure 2 shows the intensity map by the Meteorological,
Climatological and Geophysical Agency (BMKG). BMKG Sulawesi is located in the Sunda block adjacent with three plates;
automated modeling indicates intensity of VI-VIII in Palu city and the Australia plates, Philippines plate and Pacific plate, known as
Donggala regency, Central Sulawesi. Two-day following the triple junction. This triple junction results in a complex tectonic in
earthquake, BMKG updated the intensity to IX-X (i.e. extensive Sulawesi, accommodated by strike slip faulting and thrust faulting.
damage) based on modeling, data instrument, and macro-seismic In 2017, PuSGeN considered 50 active-fault segments in Sulawesi
survey of 30 September 2018. The Palu – Koro fault zone is and the North Sulawesi Megathrust in the north of Sulawesi (Figure
believed to be the source of the Palu earthquake 2018 (Mw 7.4), as 3). Similar tectonic condition in Sulawesi was also previously
observed from its epicenter location and hypocenter depth, and reported by Irsyam, et. al., 2010.
aftershock distribution (Figure 1). The seismicity in Sulawesi is influenced by the subduction along
the North Sulawesi Megathrust (Figure 3). The WBZ from the slab
in North Sulawesi extends to a depth of about 180 km. The major
structures in Central Sulawesi is the Palu – Koro fault system, which
extends NNW-SSE direction and cuts Sulawesi apart more than 300
km, from the North Sulawesi trench pass through Palu bay,
southward turn to the SE connect to the Matano and Lawanopo
faults and further eastward, both faults join to Tolo trench.
Sulawesi moves north-west, suggesting a strong impact from the
Australia plate (Figure 4). Velocities in the northern and southern
part of Palu – Koro show different magnitude, indicating the activity
of Palu – Koro fault zone. To the west of northeast Sulawesi, the
island moves westward, indicating an impact from the Pacific plate.
The Palu – Koro fault is a sinistral strike-slip fault (horizontal fault
striking) in a north-south direction, crossing Palu city. The
maximum slip-rate of Palu – Koro fault is 58 mm/year based on
geological study by interpretation of geological offset, such as river
offset (Daryono, 2016). Recent GPS study also suggests that the
slip-rate of Palu – Koro fault is about 40 mm/year.
Central Sulawesi region is tectonically active as shown in Figure
Figure 1 Epicenters of the Palu earthquake 2018 (Mw7.4) 5. Historical destructive earthquakes along Palu – Koro fault zone
and the aftershock distributions (red dots) along Palu – Koro fault occurred in 1907, 1909, 1937 dan 2012. Paleoseismologi study has
in Central Sulawesi (PuSGeN, 2018) been conducted by Daryono (2016) and obtained that previous
earthquakes occurred in 1909, 1468 and 1338. Abendanon (1917)
This natural disaster caused many losses of lives and other concluded that the 1907 earthquake was followed by a more
material damages. The National Agency for Disaster Management destructive earthquake 2 years later in 1909. Houses that survived
(BNPB) on 9 October 2018 reported that this earthquake has caused during the 1907 earthquake were mostly destroyed by the 1909
2.037 casualties, 671 missing people, and 152 buried bodies. Total earthquake. The damages run along Saluki up to Donggala region.
of 4.084 people are heavily injured, 74.044 evacuees, and 67.310 He reported a large cracking of 7 km with uplift of 1.0 m. Trenching
houses are estimated to be damaged. Although the most damaged in Onu village on Saluki segment showed evidence of sinistral slip
area is in Palu city, these reported numbers are also compiled from of 1.5 m and vertical slip of 1.5 m. Daryono (2016) suggested a
other areas, such as Donggala, Sigi, Parigi Mountong, and plausible recurrence interval of 130 years in Palu – Koro. During
Pasangkayu, West Sulawesi. 2017, two main earthquakes were recorded along the fault.
5
Figure 2 Location of the main shock of the earthquake on 28 September 2018 and its shakemap (BMKG, 2018). Left is the automatic
shakemap first generated by BMKG on 28 September 2018, Right is the corrected shakemap based on modeling, data instrument and
macroseismic survey as of 30 September 2018
Figure 3 Active fault in Sulawesi as in the National Earthquake Source and Hazard Map launched in 2017
(Irsyam, et al., 2017)
6
Figure 4 Left: Seismicity of the Sulawesi for Mw ≥ 4,5, right: tectonic and deformation pattern derived from GPS data in Sulawesi
(Irsyam, et.al., 2017)
Figure 5 Historical Earthquake in Central Sulawesi (Daryono, 2016). Focal mechanism from CMT catalogue
3. GEOLOGY SETTING AND SUBSURFACE The morphology of Palu area consists of plains, denudation, and
CHARACTERISTIC hills. Plains area is irregular and weak topography. This area is
formed by fluvial sedimentation. Based on regional geology, Palu
The geology of Palu area is shown in Figure 6. Evolution of area consists of alluvium deposits and beach deposit (Qap) ) that are
Neogene kinematics along the Palu – Koro fault was confirmed Holocene. The unit is composed of gravel, sand, mud, and coral
based on microtectonics approach, i.e., sinistral strike-slip due to limestone.
east-west compression, radial extensions caused by telescoping This unit is formed in shallow river, delta, and sea environments.
vertical movement of Neogene granitoid, and then left lateral with As this unit is a young sediment, the arranged layer has not
normal component displacement due to north-south extension/east- undergone the whole process of lithification or cementation process
west compression which actually is still active. Palu depression area (Soekamto, 1995).
is filled by mostly clay, silt, and sand deposits as alluvial deposit.
The composition of gravel is granitic fragment to the northwest.
West escarpment to the north consists of granite and granodiorite
units, and to the south consists of schist-phyllitic units. East
escarpment consists of molasses (Pramuwijoyo et al., 1997).
7
Figure 6 Geology Map of Palu, Sulawesi (source: Geological Agency of Indonesia)
4. SURFACE DEFORMATION DUE TO PALU

EARTHQUKAE 2018
Surface deformation due to Palu Earthquake 2018 was detected by
ALOS-2 Satellite from JAXA, which was analyzed by the
Geospatial Information Authority of Japan (GIS), by comparing
interferometry analysis of 21 August 2018 data and 2 October 2018
data. The INSAR data suggests uplift in Donggala area, and
subsidence Palu city, in the eastern part of the fault. The surface
deformation was also detected by USGS/NASA Landsat-8 imagery
of 16 September 2018 and 2 October 2018, which was analyzed and
mapped by Austin Elliott, COMET, University of Oxford. Data
analysis from landsat image suggest a maximum coseismic surface
deformation of 5 meter, in direction consistent with the sinistral slip
characteristic of Palu – Koro fault, in Palu city. This surface
deformation pattern was verified in the field survey.
The field survey conducted on 8 to 15 Octobers 2018 by Figure 7 Offset Rupture 4,8 m at desa Donggala Kodi, Ulujadi,
Natawidjaja, Daryono, Gunawan, Meilano, and Pamumpuni Palu city, Central Sulawesi
observed rupture horizontal offset as large as 4-6 meters and vertical
offset up to 30-50 cm running through Palu city from south to north
(Figure 7-8). The rupture went into the sea in the Palu bay, and is
observed again on the ground in Labuan Salumbone, Tawaeli,
Donggala, Central Sulawesi, with a much smaller offset of less than
1 meter. The offset of the rupture of Palu Earthquake was then
mapped according to field observation as shown in Figure 7.
Figure 8 Offset Rupture desa Bomba, Marawola, Sigi regency,

Central Sulawesi
8
5. GROUND MOTION 6. REPORTED BUILDING and INFRASTRUCTURE

DAMAGE
The ground motion of the Palu earthquake was recorded in 2 BMKG Building and infrastructure damages were reported just after the 7.4
stations, namely in Poso (PMCI) and in Sabang (MPSI) (Figure 9), earthquake. Examples of building and infrastructure damages are
while 1 station in Palu was reported broken due to the foreshock shown in Figure 10.
earthquake that occurred 2 hours before the mainshock 7.4.
Figure 9 Ground motion and response spectra recorded in Poso (PMCI) and in Sabang (MPSI) by BMKG
Figure 10 Building and bridge damages induced by earthquake in Palu
9
7. GEOTECHNICAL IMPACT 7.1. Ground Cracking

During Palu earthquake 2018, two major interesting phenomena Ground cracking were found along road near the coastline and
occurrences were observed, namely liquefaction induced ground mountainous areas after the earthquake. At the mountainous road,
failure and massive lateral spreading. These incidents resulted in the ground failure was identified. It is believed that the damages were
affected villages to be buried as well as shifted (Figure 11). The initially created by the mainshock and became worse with the
following sections describe liquefaction and lateral spreading in the following aftershock. In particular, the aftershock exacerbated
affected area in more detail. ground deformations and destroyed foundations, suggesting the
According to the Center for Groundwater Resources and possibility of the progressive ground failure.
Environmental Geology, most Palu area has a very high potential of Ground cracking was also observed at runway in Palu main
liquefaction with a liquefaction potential index (LPI) > 15 with a airport, which is located nearby Petobo area (Figure 12). It was
shallow ground water level of about < 12 m. However, many informed that the typical top soil layer is clayey and silty soils. The
geotechnical aspects after the earthquake are interesting to be ground water table is relatively deep. Thus far, this situation only
investigated in order to find the actual cause and mechanism as well generated a minor damage in the air strip, while in the other areas,
as to learn any other relevant lessons. The investigations that are more severe damages were observed.
currently conducted are important to support the government for
housing relocation and reconstruction phase. 7.2. Liquefaction and Lateral Spreading
Irsyam and Sahadewa have conducted a preliminary field survey
to the liquefaction potential of Palu city and surrounding area. The Just after Palu earthquake 2018, major damages are reported in four
survey indicated that soil type was dominated by sand and silt. As locations, namely Balaroa, Petobo, Jono Oge, and Sibalaya Selatan.
reported by local people, water table depth was shallow before the These areas are located from 1 km to 7.2 km from Palu – Koro fault.
earthquake. Combining the impacts of soil type and elevation of Although, data is still being collected and compiled, it is currently
ground water table, it can be concluded that the area is susceptible to believed that damages in these locations are mainly attributed to
liquefaction. The lateral spreading and mass failure at the Balaroa liquefaction and lateral spreading. Liquefaction is a phenomenon in
dan Petobo are possible triggered by the liquefaction since those which saturated loose sand is converted into a liquid-like material
villages were covered by saturated loose sand and silty sand deposit. and loosing shear strength during an earthquake.
Figure 11 Liquefaction induced lateral spreading in Petobo village (Digital Globe)
For certain condition, liquefaction can result in lateral spreading

where a large horizontal surface deformation is experienced by large
ground blocks above the liquefied layer. Commonly, this lateral
spreading occurs in relatively gentle slope surface. Lateral spreading
is considered as one of the major contributors on the failure of the
structures and buildings in the urban areas with high seismicity.
7.2.1. Balaroa
Soil failure in Balaroa area is located about 1 km to the west of Palu
– Koro fault. The crown of soil movement is located in Jalan
Gunung Gawalise, whereas the toe is located in Balaroa village. The
affected area is mainly residential housing comprising of 34.5 Ha
with perimeter of 2.5 km. Aerial view of damages in Balaroa is
presented in Figure 13. Field observation at Jalan Gunung Gawalise
where the crown is located showed that slope failure may also be
Figure 12 Longitudinal Cracks on the Palu airstrip attributed to rotational land slide.
10
North
Toe
Damaged road at
Jl. Gunung Gawalise
Crown
Figure 13 Estimated soil failure affected area in Balaroa
7.2.2. Petobo fault. Although this area has slight inclined ground surface, the
Petobo, a populated residential area was severely damaged due to occurrence of mass movement to downslope was triggered by
liquefaction. This area is about 7.2 km to the east of Palu – Koro liquefaction (Figure 14).
Figure 14 Liquefaction on the residential at Petobo

( Digital Globe)
Figure 15 The illustration of mass movement due to liquefaction in Petobo village

( Fraga LF )
11
7.2.3. Jono Oge residential housing, and roadway. Practically, damaged roadway
area turned into a large corn farm which initially is located nearby
Massive soil failure in Jono Oge village is located about 6.7 km to the crown. It should be noted that the roadway leading to this
the east of Palu – Koro fault. The crown of soil movement is located damaged roadway section and the surrounding area are also heavily
in the east nearby a water channel, whereas the toe is located in the affected by the earthquake. Damages in a gas station located about 1
west (Figure 16). In general, north boundary of the affected area is km north to the damaged roadway section is shown in Figure 17.
bounded by a river. The affected area is estimated about 185 Ha
with a perimeter of 8 km and is mainly comprised of corn farm,
North
Damaged road at
Jl. Poros Palu - Palolo
3115 m
Crown
Toe
800 m
Soil movement
direction
Figure 16 Estimated soil failure affected area in Jono Oge
Figure 17 A damaged gas station in Jono Oge
12
7.2.4. Sibaya Selatan located in the west (Figure 18). The affected area is estimated about
29 Ha with a perimeter of 2.2 km and is mainly comprised of paddy
Soil failure in Sibalaya Selatan village is located about 3.6 km to the farm, residential housing, and roadway. Figure 19 shows the impact
east of Palu – Koro fault. The crown of soil movement in this of Palu Earthquake 2018 of translated housing and football field in
location is located in the east nearby a sluice gate, whereas the toe is Sibalaya Selatan village.
North
Damaged road at
Toe Jl. Poros Palu - Kulawi
450 m
840 m
Crown
Figure 18 Estimated soil failure affected area in Sibalaya Selatan
Figure 19 Translated housing and football field in Sibalaya Selatan after Palu Earthquake 2018
8. CONCLUSION 9. REFERENCES
PUSGEN and HATTI have conducted a field survey to preliminary Daryono, M.R., 2016. Paleoseismology Tropis of Indonesia (Stdu
investigate the earthquake rupture offset and geotechnical impact of case: Sumatera Fault, Palukoro-Matano Fault, and Lembang
the Palu Earthquake 2018. The rupture offset was found as large as Fault). PhD Dissertation Institut Teknologi Bandung.
4-6 meters sinistral offset and about half meter vertical offset. The Irsyam, M., Sengara, I.W., Asrurifak, M., Ridwan, M., Aldiamar,
geotechnical aspects observed after the earthquake are mainly F., Widiyantoro, S., Triyoso, W., Natawijaya, D.H., Kertapati,
related to liquefaction and lateral spreading. In Palu earthquake, E., Meilano, I., and Suhardjono (2010), Summary:
liquefaction resulted in lateral spreading where large horizontal Development of Seismic Hazard Maps of Indonesia for
surface deformation is experienced by large ground blocks above Revision of Seismic Hazard Map in SNI 03-1726-2002,
the liquefied layer. This lateral spreading was observed in relatively reseach report submited to the Ministry of Public Works by
gentle slope surface. Team for Revision of Seismic Hazard Maps of Indonesia, July.
13
Irsyam, M., Widiyantoro, S., Natawidjaja, D.H., Meilano, I., Pramumijoyo, S., Indarto, S., Widiwijayanti, C. and
Rudyanto, A., Hidayati, S., Triyoso, W., Hanifa, N.R., Sopaheluwakan, J., 1997, Seismic Parameters of the Palu-Koro
Djarwadi, D., Faizal, L., Sunarjito (editor), National Team for Fault in Palu Depression Area, Central Sulawesi. Indonesia,
updating of Indonesia Earthquake Hazard Map 2017 (2017), Journal of South East Asian Earth Science,
“Earthquake source and hazard Map of Indonesia 2017”, Soekamto, R.A.B., 1995, Regional Geological Map of Palu Sheet,
Indonesia National center for earthquake Studies, Research and Indonesia, Scale 1:250,000, Geological Research Center,
Development Agency of Ministry of Public Work and Housing. Bandung.
ISBN 978-602-5489-01-3. Thein, P.Y., Pramumijoyo, S., Brotopuspito, K.S., Kiyono, J.,
PuSGeN (Indonesia National center for earthquake Studies) (2018), Wilopo, Furukawa, A., and Setianto, A., 2014, Estimation of
“Quick Report on Palu Earthquake 2018”, Indonesia National Seismic Ground Motion Induced by the 23 January, 2005
center for earthquake Studies, Research and Development Earthquake in Palu Region, Central Sulawesi, Indonesia.
Agency of Ministry of Public Work and Housing. Journal of Geological Resource and Engineering 4, 200-207.
14
Piling Construction and Testing of Megastructures on Problematical Soil Ground of

Kazakhstan
A. Zhussupbekov1 and A.Omarov1
1
Department of Civil Engineering, L.N. Gumilyov Eurasian National University, Astana, Kazakhstan
E-mail: astana-geostroi@mail.ru
ABSTRACT: During the last 20 years, many high-rise buildings supported by pile foundation are rising up in Astana, the new capital of
Kazakhstan. Following megaprojects are already completed: Ministry of Transportation and Communication, Expo-2017, Housing estate –
Izumrudny Kvartal (Emerald Square), Cultural and Entertainment Center – Khan Shatyr and so on. Many megaprojects are under
construction or in planning. One of the unique projects is the housing estate “Abu-Dhabi Plaza” which started on 2011 in Astana.
The article presents loading tests of large diameter and deep boring piles on the construction site Expo-2017 in new capital city of the
Republic of Kazakhstan. Results of static pile tests using the static compression loading test (by ASTM – SCLT), static loading test (by
GOST – SLT) and bi-direction static loading test (by ASTM - BDSLT or O-cell) methods are presented in this paper. Hereafter the results of
underground testing by the piles with the methods of vertical static testing are presented, which had been made on Expo 2017 projects,
buildings of Pavilion in Astana, Kazakhstan.
Keywords: pile, SCLT, SLT, O-cell or BDSLT, load, settlement.
1. INTRODUCTION The Features of high-rise buildings present high requirements to

the results of the EGS (engineering and geotechnical survey) and
High-rise buildings (buildings with a height of more than 75 m) should solve the following main tasks in their implementation:
pose new challenges for engineers, especially in the field of - study of the geological structure of the soil massif with large
calculations and design of above-ground structures, bases and volume (up to 60 m in depth and at least 2 foundation widths
foundations. Therefore, designers of both above-ground and beyond its contour)
underground parts of the building are forced to resort to more - reliable assessment of the hydrogeological and hydro chemical
complex methods of calculation and design. Especially this applies conditions of both the compressible soil massif, and in the
to geotechnicians, who are involved in the design of foundations for excavation zone and adjacent territory with the establishment of
high-rise buildings. their corrosive aggressiveness, in time;
By complexity, problematic design, erection, operation, impact - determination of deformation and strength properties of
on the environment and people, high-rises can be attributed to the dispersed and rocky soils at large ranges of voltage changes;
structures of increased danger and complexity. - instrumental observation and monitoring of deformations of the
Kazakhstan has its own modern experience in designing and soil massif of the basement foundation and the adjacent territory
erecting high-rise buildings above 75 m of unique structures, under static effects.
including the “Emerald Quarter” (210 m), “Northern Lights 1”
(180 m) in Astana, Railways Building (174/156 m), Transport
Tower (the building of the Ministry of Transport and
Communications) (155 m), Khan-Shatyr (150 m) (see Figure 1).
Figure 2 Project of Abu Dhabi Plaza in Astana

Figure 1 High-rise buildings above 75 m of unique structures in
Kazakhstan Before you start laying the foundation, you need to decide on its
technology and depth. It depends on the expected load on it and the
Note: a) Abu-Dhabi Plaza; b) Emerald Quarter; c) Northern features of the natural conditions, namely the type of soil and the
Lights 1; d) Railways Building Towers; e) Grand Astana Tower; depth of the groundwater. Abu-Dhabi Plaza residential skyscraper
f) Triumph of Astana; g) Astana Marriott Hotel; h) Transport Tower consists of 5 main towers:
(the building of the Ministry of Transport and Communications); -Block R- offices and living quarters;
i) Hotel “Kazakhstan” (Almaty city) and j) Khan-Shatyr. -Block O- office building;
-Block H- hotel and furnished rooms;
After the completion, Abu Dhabi Plaza will become the tallest -Block Y- offices of class «A»;
building in Kazakhstan and Central Asia, and the height of one of -Block Z-residential apartments.
the blocks of the complex will be 320 meters.
The architectural concept is shown in Figure 2, which represents 2. THE ANALYSIS OF THE DEFORMATION OF THE
the construction site - in the centre of which a skyscraper should BASE FLAC3D
rise.
This grandiose skyscraper will be the fourteenth tallest building The analysis of the deformation of the base is based on the
in the world. The architect of the project is famous British architect production of a load of its own weight of a single design in the
Norman Foster. horizontal plane of the vertical deformation, as shown below in the
The total territory of "Abu Dhabi Plaza" will be about 500 Figure 3.
thousand square meters, including 206 thousand square meters of The horizontal grillage plan is a form of a reference deformation
ground floor for parking and retail spaces. and is used to estimate column drafts in various places. The draft for
15
each stage of construction is calculated using the ratio of the vertical

reaction at each stage to the vertical reaction at the full design load. 3. SECOND PROJECT EXPO 2017
The block R and the basement tile of the general basement are
modelled taking into account the changes in thickness. Model The site chosen to accommodate Expo-2017 Astana is located 8 km
physical boundaries are set within the grid 9 to 21 and the grid from south of the old city centre of Astana and just 4 km from the new
A to M. Figure 3 and Table 1 shows the settlement of the contour of government block on the southern bank of the Ishim River. The
the site using the program FLAC 3D. exhibition area with a total area of 25 hectares is surrounded by a
territory of 149 hectares, intended for housing residential and mixed
buildings, auxiliary exhibition facilities and transport infrastructure.
The total area of the Exhibition Area is 174 hectares (see Figure 4).
Pavilion of Kazakhstan (Sphere) is the only building in the
world, which is a sphere of finished form with a diameter of 80
meters. Possessing unique design features, a given shape of the
building, as well as the functions of the exhibition building, the
pavilion at the same time serves as a prime example of the use of
renewable energy sources (Zhussupbekov A.Zh., etc (2017)).
Figure 3 Deformation base (with software program FLAC 3D)
The wind forces are based on the tower dynamic parameters,

such as the natural frequency and damping of the tower. The wind
loads indicated in the RWDI report were based on observational
data over a 50-year period. Multiplication by a safety factor 1.4 was
used to estimate the ultimate wind load.
Table 1 Draft of the plate foundation under full fixed and movable
loads Figure 4 General layout of the Expo-2017 Exhibition
Location: Model 1 (with software

program FLAC 3D) 4. ENGINEERING - GEOTECHNICAL DESCRIPTIONS
Core of the building (core) 37,4 mm OF CONSTRUCTION SITES EXPO-2017
(Block R) At the construction site, a complex of laboratory and field studies of
Column RC1 (Block R) 14 mm the soil base was also carried out. Based on the field description of
Uneven settlement (Block R) 23,4 mm the soils confirmed by the results of cone penetration tests and
laboratory tests, a division of the soils composing the site of
The tower’s wind loads fluctuate depending on wind speed, prospecting for engineering-geological elements in the stratigraphic
wind direction, and local effects such as swirls. The measurement of sequence of their occurrence was carried out (see Figure 5).
the wind shield for comparison with the model is not suitable, and Figure 5 shows the engineering and geological sections of the EGE.
will not provide any meaningful information for the purpose of
building evaluation. Nevertheless, the effect of the measured wind
speed and direction is considered when comparing the results of
deformation with the computational model for gravity loads. Since
the design of wind loads is based on the dynamic reaction of the
building, the frequency of the natural oscillation and the estimate of
the damping of the tower will be measured to compare the values
used to calculate the wind strength. The effect of temperature on the
final model will be limited to normalize the effect on the measured
data. Recorded ambient temperatures at the time of measuring the
settlements of the tower are necessary to ensure the normalization of
the results. The measured data must be collected in order to
minimize the effect of temperature effects throughout the tower, and
also to limit the effect of localized heating, the effect of solar warm
amplification, and the settlement of the sun during the day.
Localized temperature control at specific work sites below the
working front will be taken into account. The building models will
take over the temperature of the structural elements within all the
façade closed floors especially not heated to the same temperature
throughout the tower. If inaccessible, an external ambient
temperature sensor will be used during data collection. Structural Figure 5 Engineering-geological cross- sections in construction site
elements in unloaded floors opened to the environment will be Expo-2017, Astana, Kazakhstan
considered at ambient temperature during the collection of these
readings. Hydrostatic pressure under the slab foundation was not 5. STATIC PILE LOAD TESTING
considered. Seismic loads are not taken into account for this project,
since Astana is not classified as an earthquake zone. Static testing with Osterberg method (О-Cell testing) was carried
out for the test of deep foundations at the site of the construction of
16
this object. Four bored piles were subjected to static tests (O-cell holding time of intermediate loading stages was 30 minutes,
testing- 2 piles and SCLT- 1 pile and SLT by GOST-1 pile). unloading - 20 minutes. The time for maintaining peak loads was
The target of this tests was obtaining of bearing capacity of piles on 120 and 240 minutes, respectively (Zhussupbekov A.Zh., Lukpanov
problematical soils ground of Expo 2017 (Zhussupbekov and R.E., Omarov A.R. (2016)).
Omarov A.R. (2016))
5.1 Static tests in accordance with the requirements of GOST

(CIS Standard)
Static tests of soils for bored piles are carried out in accordance to
GOST 5686-94. Test was carried out after the pile concrete strength
had attained more than 80% of the design value. As part of the
installation for soil testing, static pressing forces should include
equipment:
- device for pile loading (jack);
- supporting structure or platform for perceiving reactive forces
(for example, a system of beams with anchor piles or a platform);
- device for measuring the settlement of piles during the test
(reference system with measuring instruments).
Before starting the tests, the wire should be subjected to
preliminary stretching for two days with a load at least four Figure 7 Results of static loading tests (SCLT and SLT)
kilograms. During the tests, the load on the wire should not be more
than one and a half kilograms. Limits of measurement and the price
of division of pressure gauges used to determine the load on the pile The bearing capacity of the tested piles with static vertical-
during testing are selected depending on the greatest load on the pile pressing forces, at the above construction site, was 12000 kN (see
provided by the test program, with a margin of at least 20 precent. Figure 7). It should be noted that even with a maximum test load of
Loading of the tested pile is performed evenly, without impacts, by 12000 kN, only the elastic operation of the pile in the ground is
load stages, the value of which is set by the test program, but it is manifested, as evidenced by a slight residual soil settlement after
taken no more than 1/10 of the maximum load on the pile specified unloading, which is 1.4 mm.
in the program. When the lower ends of the field piles are buried
into coarse-grained soils, gravel and dense sands, and clay soils of 5.2 Method O-cell piles static test in accordance with ASTM
solid consistency, the first three load stages are assumed to be equal According to the results of engineering and geological surveys,
to 1/5 of the maximum load in the program. At each loading stage of bored piles 31.5 m long and 1000 mm in diameter were used as
the full pile, the reports for all strain gauges are taken in the foundations. In order to control and evaluate the compliance of the
following order: zero report - before loading the pile, the first report bearing capacity of piles on the ground, the design loads were field
immediately after the load is applied, field this consistently four static tests by the Osterberg method.
reports with an interval of thirty minutes and then every hour before The peculiarity of the O-cell test method is that the load is
the conditioned deformation stabilization (Zhussupbekov A.Zh., applied not on the head of the pile, but in the body of the pile, where
Omarov A.R. (2016)). the jack (power cell) is installed, working in two directions. The
For the criterion of conditional stabilization of deformation power cell (O-cell jack) divides the test pile into two parts: the upper
during testing by the natural pile, the speed of the pile sediment at a (upper test element - UTE) and the lower (lower test element-LTE).
given loading stage is assumed to not exceed 0.1 mm in the last 60 The power cell (O-cell jack) is a system of calibrated hydraulic
minutes of observation if sandy soils or clay soils lie from the hard jacks combined into one module. The hydraulic jack is installed at a
to the turgid consistency under the bottom of the pile, the bottom depth of ½ the length of the pile - 16.8 m. The power cell is
end of the pile lies clay soils from the fleshy to the flowing connected by hydraulic hoses to the hydraulic pump located on the
consistency, then two hours of observation. The test load of the field ground surface (Zhussupbekov A.Zh., Omarov A.R. (2016)).
pile shall be adjusted to a value at which the total pile residue is not When designing the O-Cell test, special attention should be paid
less than forty mm. When the lower ends of the field piles are to the study of the geotechnical structure of the soil massif of the
deepened into coarse - clastic, dense sandy and clay soils of solid construction site, since the location of the jack in the body of the
consistency, the load should be brought to the value provided by the pile depends on the accuracy of the survey data, in particular the
test program, but not less than the one-and-a-half pile load-bearing results of assessing soil resistance. The correct decision to place the
capacity determined by calculation, or the design pile resistance of jack affects the quality of the tests carried out, since the
the material. differentiated determination of the load-bearing capacity
The bearing capacity of the tested piles with static vertical- components (along the lateral surface and below the lower end) is
pressing forces, at the above construction site, was 12000 kN reduced to the correct selection of an equal ratio of the lateral
(Figure 7). resistance of the soil along the upper element to the resistance below
The calculated permissible vertical-punching load on the pile, the lower end of the lower element of the experimental pile.
taking into account the safety factor γk = 1.2 according to paragraph
3.10 of SNiP RK 5.01-03-2002 "Pile foundations", it is
recommended to take equal 10000 kN (Zhussupbekov A.Zh., 5.3 Results of field trials using the Static Load Test and
Lukpanov R.E., Omarov A.R. (2016)). Osterberg methods
Figure 8 shows a comparison of the test results: the "load-sludge"
5.2 Pile Static Compression Load Test by ASTM (SCLT
curve obtained by the SCLT method and the equivalent "load-
method)
settlement" curve determined by the O-cell method. For the
Static compression loading testing was carried out in accordance to comparative criteria of SLT, SCLT and O-cell results fixes
ASTM D 1143-07 (2013). settlement of 10 and 14 mm had been taken (see Table 2).
Vertical static loading of piles using the SCLT method is one of
the most widely used field test methods for soil used to analyze pile
bearing capacity. In the first cycle, the experimental pile was loaded
to 6000 kN of the design load, in the second cycle to 12000 kN. The
17
GOST 5686-94. 1994. (1994). Methods for field testing by piles.

Gersevanov Research Institute of Bases and Underground
Structures (NIIOSP), Moscow.
Zhussupbekov A.Zh., Omarov A.R. (2016). Modern advances in the
field geotechnical testing investigations of pile foundations.
The 15th Word Conference of Associated Research Centers
for the Urban Underground Space. Saint-Petersburg, Russia.
Procedia Engineering, Volume 165, pp 88-95.
SNiP RK 5.01-03-2002 (2002) "Pile foundations". KAZGOR
Design Academy, Almaty.
Zhussupbekov A.Zh., Lukpanov R.E., Omarov A.R. (2016). Bi-
directional static load testing. Selected paper from the
proceedings of the fourth Geo-China International
Conference. Shandong, China. pp 35-42.
ASTM D1143M – 07 (2013) Standard Test Methods for Deep
Foundations Under Static Axial Compressive Load.
Zhussupbekov A.Zh., Lukpanov R.E., Omarov A.R. (2016).
Figure 8 Comparison of test results by SCLT, SLT and O-cell Experience in Applying Pile Static Testing Methods at the
methods Expo 2017 Construction Site. Scientific journal Soil
Mechanics and Foundation Engineering: Issue 4: New-York,
Table 2 presents a comparative analysis of the bearing capacity of USA,. pp 251-256;
piles, obtained by different methods in this research. Zhussupbekov A.Zh., Omarov A.R. (2016). Geotechnical and
Construction of pile foundations in difficult soils ground.
Table 2 Different tests Geotechnik-Tag in München Geotechnik Zusammen-wirken
von For-schung und Praxis, München, pp 41-48.
ID The value of Results fixes
the bearing settlement
capacity of 10 mm 14 mm
piles, Qd
Pile (SCLT by ASTM) 12000 kN 11788 kN -
Pile (SLT by GOST) 12000 kN 10630 kN 11814 kN
Pile (O-Cell -1) 29000 kN 18220 kN 23985 kN
Pile (O-Cell -2) 29000 kN 20535 kN 28385 kN
3. CONCLUSION
The overlay of the curves showed that the convergence of the graphs
is observed only at the initial stage of loading, then a change in the
trajectory of the SLT curve, characteristic of the creeping stage of
soil resistance, is observed, whereas the O-cell curve (at this stage of
loading) is more characteristic of the elastic resistance of the soil.
According to the results of the SCLT unloading curve, elastic work
of the soil is still evident. The reason for the abrupt change in the
trajectory of the SCLT curve, which is not characteristic of the
elastic work of the ground, is the holding time of the loading stages
(lower compared to the O-cell test method), which can also explain
the almost completely elastic work of the soil during O-cell tests.
When testing piles using the SLT method "from top to bottom", a
design load of 6000 kN corresponds to a draft of 2.09 mm, a
maximum test load of 12000 kN is a draft of 10.51 mm. It should be
noted that even with the maximum test load, only the elastic
operation of the pile in the ground is manifested, as evidenced by a
slight residual soil sediment after unloading, which is 1.4 mm.
When testing piles using the O-cell test, a maximum test load of
29000 kN corresponds to a draft of 18.35 mm (for the PTP-1 pile)
and - 14.40 mm (for the PTP-2 pile). During the testing of the piles,
both elastic and plastic deformation of the soil was observed, due to
a greater test load on the pile than in the SLT method.
4. REFERENCES
Zhussupbekov A.Zh., Omarov A.R., Zhukenova G.A. (2017). The
experience in applying of static load and O-cell pile testing
geotechnologies in problematical soil conditions of Astana.
International Mini symposium Chubu (IMS-CHUBU)-
Nagoya, Japan. Japanese Geotechnical Society Special
Publication. Vol. 5 No. 2 pp 167-171.
Zhussupbekov A.Zh., Omarov A.R. (2016). Geotechnical and
construction considerations of pile foundations in
problematical soils. Proceedings of the 8th Asian Young
Geotechnical Engineers Conference (8 AYGEC). Astana,
Kazakhstan. pp 27-32.
18
Serious Lessons Learnt from Big Project

Ikuo Towhata1
1
Professor Emeritus, University of Tokyo
Visiting Professor, Kanto Gakuin University
E-mail: towhata.ikuo.ikuo@gmail.com
ABSTRACT: Big project is an attractive idea for ambitious engineers. The present level of construction technology allows such big projects
to be carried out and our world may be drastically changed. The problem is that we still do not know how nature would respond to such a
significant human action. We need to know bad consequence was sometimes the case in the past. Although it is difficult, we need to have
some provision by which the worst and catastrophic consequence is avoided by stopping or cancelling the project.
Keywords: Project, nature, disaster, earthquake, slope failure
1. INTRODUCTION The problem was the high rate of water evaporation from the
th lake surface. Together with the significant reduction of water supply
The 20 Century enjoyed a significant development of construction from its two major rivers, Aral Sea started to shrink in 60s.
technology. The source of power changed from human and animal Nowadays, Aral Sea, which used to be the 4th biggest lake in the
muscle in the former centuries to mechanical engines that utilize world, is disappearing. Because the lake water was salty, the deposit
coal and oil. This change gave human an overwhelming power of salt on the dried lake bed is blown by wind and is causing health
above nature. It thus became possible to construct huge canals, man- problem among local residents (Whish-Wilson, 2002). It is not
made islands, bridges etc. possible anymore to turn reverse the clock because cotton
The situation until 1970s was full of optimism. Behind such a agriculture has been well established in the region and the
situation, however, many negative aspects of the overwhelming immigrant farmers cannot go back to Russia because it is a foreign
human power were emerging gradually. Their consequence became country after disappearance of USSR.
evident after 1970s. Consequently, many people such as naturalists There have been several efforts to save the lake from
started to insist on the need to stop big construction projects and disappearance. One of them was the idea to transport water from
protect nature. It is true that human has to be aware of the huge rivers in Siberia to the Aral Sea (Borovskiy, 1980; Aladin and
importance of conservation of nature and control human power not Potts, 1992; Elhance, 1997). Although the idea looked attractive to
to seriously damage the global natural system. On the other hand, some extent, the obvious problems were the difficulty to pump up
the world population is increasing rapidly towards 10 billion. water across high mountains on the way. Later, the environmental
Increased population needs food and drinking water. People desire issues became more important such as the effects of less water flow
healthful life, comfortable living conditions, and safety of life from into the Arctic Sea, and introducing different animal species from
natural disasters. To cope with this new situation, human has to keep those rivers into the Aral Sea. Finally, Siberia and Aral Sea area
seeking for better and wiser use of the natural system. Thus, big became different countries after disappearance of USSR and such a
construction projects will be necessary as well from now on. project became impossible.
However, it is important for human to keep in mind what mistakes The Aral Sea project had both positive and negative aspects. The
happened during the past big projects and try not to repeat them. positive one is the development of agriculture and the negative one
The present paper attempts to provide some information in this is the environmental damages which are disappearance of the Sea
regard. and salt-induced health problems. Kazakhstan constructed a huge
dam in the northern part of the Sea and produced fresh-water Little
2. ARAL SEA Aral Sea (Figure. 2). This may be a solution of the problem to some
The Aral Sea region of the Central Asia is wide in space and enjoys extent but not complete or final. Note that a similar problem is going
ample sunshine. It was supposed to be a wealthy agricultural zone if on in other places of the world. The Urmia Lake in Iran is shrinking
ample water is provided. It was decided in early 1960s to construct quickly as well because of the water of tributary rivers is sent to
big canals and long irrigation channels that transported water of irrigation.
Amu Darya and Syr Darya Rivers, which were two major tributaries
of Aral Sea. It was intended by this water supply to convert the
region from an arid area into a newly developed cotton agricultural
area (Figure. 1). Many farmers immigrated from present Russia and
started successful cotton agriculture.
Figure 2 Big Earth Dam to Create Little Aral Sea
Figure 1 Newly Developed Cotton Agriculture (Uzbekistan)
19
3. DIVERSION OF RIVER WATER 4. MANMADE ISLAND ON THICK SOFT SUBMARINE

DEPOSIT
Quick industrialization and modernization often sacrifice
environmental issues everywhere in the world. Water in lakes is Kansai Airport near Osaka, Japan, was constructed in the southern
polluted by careless release of waste water or prevention of clear part of Osaka Bay where the geological subsidence is most
water supply into lakes. Later, human realizes the problem and has significant and the seabed is composed of thick marine clay and
to repair the damage. This procedure may not be very easy but has Cenozoic deposits at hundreds of meters below the seabed. From the
to be executed as a responsibility of the present generation to the geotechnical viewpoints, this least site was chosen because other
future. candidate sites refused the airport project because of the possible
Taihu Lake in China was one of the examples of polluted lake. noise problem.
Recently, canals have been and are being constructed between the The consolidation settlement of the airport island was serious as
lake and the lowermost section of the Yangtze River. It is intended has been reported by many authors (e.g., Nakase, 1992; Furudoi and
thereby to accelerate the water circulation and improve the quality Kobayashi, 2009). The rate and extent of consolidation settlement
of the lake water (Figure. 3). were underestimated in the original design stage. Reasons for this
are :
- The consolidation in the deep layers (>300m deep) was thought
to be minor because of building experience in Osaka City.
However, the sizes of the airport island and ordinary buildings
were completely different, and the bigger island increased the
stress in such deep layers.
- During the design stage, there was not a reliable technology for
soil investigation at 500 or 700 m below the surface.
- All the minor sandy layers sandwiched between clayey layers
functioned as drainage channels and promoted consolidation
volume change in clay.
- Construction of the second-stage island would finally increase
the subsidence of the first-stage island because the second island
Figure 3 Canal to Promote Water Quality Improvement increases the stress level in the seabed under the first-stage
for Taihu Lake island.
As a provision for significant subsidence of the airport island,
The author cannot draw final conclusion Waon this project. building foundation is equipped with jack-up systems. This
What he can say is that the take-in and take-out of river water in the provision has already been fully activated and there is not much
downstream part of a river (near the sea) does not affect the water margin anymore. The recent problem is the insufficient height of
level and that the amount of diverted water is not much as compared seawall around the island. In September 2018, a typhoon induced
with the total flow rate; thus the environmental effect would be high tide and high waves. As a consequence, the major part of the
small. Certainly careful field monitoring will be important. island and terminal buildings were inundated (Figure. 5) and the
More interesting project in the same area is the South-North function of the airport was stopped. It is difficult to make the walls
Water Transfer Project. In the history of this region, a grand canal higher than present because aircrafts need sufficient safety space
was constructed during the Sui Dynasty (AD 581-618) and, in the from ground structures during take-off and landing. The original
later era, helped combine the north and south parts of China ideas in the design thus did not work well because of the uncertainty
economically. To date, the water is diverted from the downstream in ground conditions and an unexpectedly strong storm which may
part of the river (Figure. 4) and sent to the north where water be possibly related with the recent global climate change.
resource is not sufficient to maintain mega cities. Due to the same
reason as mentioned above (short distance to the sea and ample river
flow), the environmental effect would be minor. Care must be taken
of the effects of two more plans of water diversion from the
upstream and the middle parts of the river because water level
therein may be affected. Human does not know everything about
nature and there is always a possibility that unpredicted situation
may happen. Detailed monitoring is essentially important.
Figure 5 Kansai Airport One After Sea-Water Inundation

(see brown color of dead lawn)
Figure 4 Water Intake For South-North Water Transfer Project

(Yangzhou, China)
20
5. VAJONT DAM Concrete arch dam was completed in 1960 and impounding
started. Then, the feared left bank started to move gradually.
Construction of Vajont Dam in Northern Italy (Fig. 6) was planned Fortunately, however, this slope movement stopped when the
and designed in 1950s as an extremely important project as the reservoir water level was lowered. This experience gave an
energy resource that would contribute to the development of Italian impression that it is possible to safely control the slope movement
industries and economy. The dam site was situated in a narrow and by raising and lowering the reservoir water level. During the first
deep gorge and looked like an ideally efficient dam site: rigid rock and second raising, this idea worked well (Fig. 7). However, the
mass, narrow and high dam with good amount of impoundment, and third raising triggered very rapid slope movement. The gauge
significant height difference for hydropower generation. became narrower quickly, the reservoir water level came up very
There were several problems at the site (Nonveiller, 1987; fast, and it was impossible anymore to lower the water level. Finally,
Semenza and Ghirotti, 2000). In the valley, there was a trace of an the entire mountain slope jumped into the reservoir and induced
ancient slope failure that suggested possibility of slope failure. high tsunami. This wave overtopped the dam and destroyed villages
During dam construction, the left bank of the reservoir in the downstream area. The number of victims was more than 2000.
(Toc Mountain) started to move and cracks developed in the The causative mechanism was the existence of weak rock layer
mountain slope. This further suggested the risk of natural disaster at depth beyond the capability of subsurface investigation. Reservoir
but was considered minor. In those days, there was not a reliable water pressure propagated into this layer, reduced the effective
technology to collect rock specimens from the depth of hundreds of stress and shear strength, and made the entire slope unstable. Then
meters and measure their mechanical strength. Thus, safety factor of the question is whether or not the senior engineers who feared the
the slope was not precisely assessed. However, the experienced possibility of disaster were able to stop the project. Issues to be
senior engineers felt that something wrong may happen in the dam. considered are;
- This dam project was a national project that was expected to help
develop the economy of Italy.
- Huge money had been spent for the success of this project.
- In spite of fear, it was not possible to verify the risk of slope
failure because deep subsurface investigation was not possible.
- Accordingly, the project was continued as scheduled until the last
moment of tragedy.
Good engineers should not behave against their official missions
because of personal fear. Professional engineers have to respect
confidentiality and should not leak important information to
outsiders. Then how could experts then avoid the tragedy? The
Vajont tragedy still remains an important material for study on
responsibility of engineers.
6. RESERVOIR-INDUCED SEISMICITY
Figure 6 Vajont Dam in Italy The reservoir of Koyna Dam in India (Figure 10) is one of the
earliest examples of reservoir-induced seismicity where significant
earthquake was triggered by impoundment. The 1967 Koyna
earthquake that registered M=6.3 claimed more than 100 victims.
The focal depth of the quake was 30 km. Fig. 8 shows a correlation
between the elevation of reservoir water and the number of
associating earthquake swarms. There is a reasonable correlation
between them, while exhibiting some time delay. The causative
mechanism of reservoir-induced seismicity has been considered
two-fold: stress in earth crust induced by weight of reservoir water,
and/or the increased pore water pressure in unstable faults, leading
to reduced effective stress and shear strength. The author supposes
that the former is not likely the cause of fault rupture because
reservoir is always located at low places (bottom of a valley) and the
increased gravity at the valley bottom tends to equalize the surface
stress distribution. Hence, the subsurface shear stress decreases. In
contrast, the pore water pressure increases the chance of fault
rupture. This situation is same as the cause of slope failure at the
Vajont dam site.
The second well-known example of reservoir-induced seismicity
is that of the Nurek Dam of Tadjikistan for which the induced
earthquake was of M=4.6 at maximum, while the focal depth was 10
km or less (Simpson and Negmatullaev, 1981). The correlation and
Figure 7 Time history of water level in the Vajont Reservoir the reservoir water level (depth) is significant as illustrated in
and rate of slope movement (Semenza and Ghirotti, 2000) Figure. 9.
21
Figure 8 Correlation between number of earthquakes and level of Koyna reservoir water (Gupta, et al., 1969)
Reservoir water depth (m)
Number of
earthquakes in 10
days
Figure 9 Correlation between reservoir water level and frequency of earthquakes in Nurek Dam (Simpson and Negmatullaev, 1981)
Aswan High Dam (Figure. 11) is a good example of a “big”

project for which both positive and negative effects have been
discussed. The maximum magnitude of induced earthquake was 5.6
and the focal depth has been in the range of 0-10km or 15-25 km
(Foulger et al., 2018). Gupta (1992) quoted Simpson’s data on the
correlation between water level and number of earthquakes. It is
discussed therein that the occurrence of earthquakes is correlated
wither with the height of reservoir water level or its rate of change
(Figure. 8).
Apart from induced seismicity, the benefit of Aswan High Dam
(Benedick, 1979) is the production of renewable energy
(hydropower) as well as control on the water level in the Nile Delta.
The latter enabled stable farming and reduced the epidemic that used
to be very bad after floods. Although the natural supply of fertile
soil had been supposed to stop by the Dam, it was thought to be
compensated for by production of chemical fertilizers. The
Figure 10 Koyna Dam in India unexpected negative issues are more reliance of chemical fertilizers
and soil decay in addition to a limited extent of coastal erosion and
salt contamination of firm land. Occurrence of those unexpected
negative issues is seen here similarly to the afore mentioned big
projects.
There are controversies on possible correlation between the
gigantic Wenchuan earthquake of Mw=7.9 in 2008, China, and the
construction of Zipingpu Dam of which the reservoir impoundment
was completed in 2006. Because the epicenter is located close to the
reservoir and the first filling of the reservoir and the onset of the
earthquake were close to each other, there have been many
discussions for and against the hypothesis that the gigantic
earthquake was induced by the reservoir impoundment. Lei et al.
(2008) mentioned the increased microtremors near the reservoir.
Figure 11 Aswan High Dam in Egypt The same point was made by Liao (2009) and Klose (2012); see
Figure. 13.
22
Figure 14 Zipingpu Dam
Figure 12 Correlation between reservoir water level and number of

earthquakes for Aswan High Dam (after Simpson, 1989,
quoted by Gupta,1992)
Figure 15 Relationship between dam height and maximum

magnitude as well as depth of the focus in induced earthquakes
(after HiQuake database by Wilson et al., 2017)
More reservoir-induced earthquakes have been known to date

such as those in Lake Mead in USA (Carder, 1945) and
Hsinfengkian Dam in China (M=6.1; Wang et al., 1976) among
many others. See Gupta (1992) for more details. Figure. 15 was
drawn by using HiQuake database of induced seismicity (Wilson et
al., 2017). It is found here that the focal depth of the induced
earthquakes may be as shallow as less than 5 km. Hence, the
devastating effects at the ground surface can be significant in spite
of the relatively small magnitude and energy. To assess the risk of
Figure 13 Correlation between occurrence of minor earthquake induced seismicity, bore hole injection tests may be useful (Ohtake,
and reservoir water level of Zipingpu Dam (after Klose, 2012) 1974; Ma et al., 2015).
In contrast, Chen (2009) discussed the occurrence of fore-, main 7. EARTHQUAKES INDUCED BY EXTRACTION OF
and aftershocks and argued that the observed pattern of those shocks OIL AND GAS
is different from those of known induced seismicity. Gahalaut and
The most recent technology for extraction of oil and gas from the
Gahalaut (2010) calculated the stress that was induced by the
earth crust injects pressurized water and other liquids in order to
reservoir water to discuss that the induced stress is oriented in
facilitate the outward flow of the desired resources. The collection
opposite direction to trigger the earthquakes.
of shale oil relies on hydraulic fracturing and hence injects
To date, it is difficult to draw a definite conclusion on the cause
pressurized fluid into earth crust as well. It is obvious that such a
of the Wenchuan earthquake. However, this controversy in
procedure may induce earthquakes as was discussed in the previous
conjunction with other induced earthquakes recommends the
chapter (Hubert, M.K. and Rubey, 1959).
community to be more careful of the possible risk of damaging
One of the earliest earthquakes induced by liquid injection
induced earthquakes.
occurred near Denver, USA, in late 1960s when waste water was
being injected into the earth crust (Healy et al., 1969). Afterwards,
similar phenomena have been reported from such oil/gas projects as
in Colorado (Raleigh at al., 1970). Fig. 15 illustrates a correlation
between gas production and number of earthquakes in Groningen oil
23
field in the Netherlands. There is an evident correlation between the Foulger, G.R., Wilson, M.P., Gluyas, J.G., Julian, B.R., Davies, R.J.
amount of gas production and the rate of earthquakes. Because of (2018) Global review of human-induced earthquakes, Earth-
this trend and the seismic risk in the country, the gas production in Science Reviews 178: 438-514.
this region will be terminated in 2030 or so. In spite of the recent Furudoi, T. and M. Kobayashi (2009) Geotechnical issues and
decreasing trend, however, a relatively big earthquake of M=3.4 approach on Kansai International Airport project - Prediction
occurred in January, 2018. and performance of settlement -, Proc. JSCE C 65(4): 998-
1017.
Gahalaut, K. and Gahalaut, V.K. (2010) Effect of the Zipingpu
Gas extraction in Groningen and earthquake swarm reservoir impoundment on the occurrence of the 2008
Netherlands earthquake rate.qpc Wenchuan earthquake and local seismicity, Geophysical
60 Journal International 183: 277–285.
Gupta, H.K. (1992) Reservoir-induced earthquakes, Elsevier.
50 Gas production Gupta, H.K., Narain, H., Rastogi, B.K. and Moha, I. (1969) A study
(billion cubic m/year)
40 of the Koyna earthquake of December 10, 1967, Bulletin of
the Seismological Society of America 59(3): 1149-1162.
30 Healy, J.H., Rubey, W.W., Griggs, D.T. and Raleigh, C. B. (1968)
The Denver earthquakes, Science 161(3848): 1301-1310.
20 Hubert, M.K. and Rubey, W.W. (1959) Mechanics of fluid-filled
Number of earthquakes porous solids and its application to overthrust faulting, Bull.
10 (magnitude>1.5) in 9 months Geol. Soc. Am. 70): 115-166.
0 Lei, X., Ma, S., Ｗ en, X., Su, J. and Du, F. (2008) Integrated
2002 2004 2006 2008 2010 2012 2014 2016 2018 analysis of stress and regional seismicity by surface loading -
Year A case study of Zipingpu Reservoir, Seismology and Geology
30(4): 1046-1064 (in Chinese).
Figure 15 Correlation between gas production and number of Liao, Y. (2009) Verification of the effects of Zipingpu Dam
earthquake in Groningen Gas field (after NWO, 2017) reservoir on inducing Wenchuan earthquake, Glacier and
Earth (n Chinese).
Ma, X., Li, Z., Hua, P., Jiang, J., Zhao, F., Han, C., Yuan, P., Lu, S.
8. CONCLUSION and Peng, L. (2015) Fluid-injection-induced seismicity
The present paper addressed several recent big projects by which experiment of the WFSD-3P borehole, Acta Geologica Sinica
some negative aspects became evident. It is thought herein that, English Edition, 89(3): 1057-1058.
although civil engineering has been doing many good projects for Nonveiller, E. (1987) The Vajont reservoir slope failure,
the welfare of people, sometimes mistakes are made to cause Engineering Geology 24: 493-512.
unfavorable consequences. The reason for this is that nature is still Nakase, A. (1992) Settlement of Kansai International Airport Land,
more complicated than what human knows. To cope with this Proc. JSCE(454/III-20): 1-9.
difficult situation, it is important to continuously monitor what is NWO (2017) Aardbevingsrisico's in Groningen (Earthquake risks in
going on in nature and detect small phenomena that suggest more Groningen) Stand van zaken opvolging aanbevelingen (State
serious situations that are going to occur in near future. If the future of affairs follow-up recommendations).
situation seems serious, it is important to stop or cancel, in the Ohtake, M. (1974) Seismic activity induced by water injection at
extreme case, the project. The current society does not have a Matsushiro, Japan, Journal of Physics of the Earth 22: 163-
system that allows such a very serious decision to be made. 176.
Raleigh, C.B., Bredehoeft. IleaJy, J.H. and Bohn, J. (1970)
Earthquakes and waterflooding in the Rangely oil field, Geol.
9. REFERENCES Soc. Am. Abstract 2: 660-661.
Semenza, E. and Ghirotti, M. (2000) History of the 1963 Vaiont
Aladin, N.V. and Potts, W.T.W. (1992) Changes in the Aral Sea slide: the importance of geological factors, Bull. Eng. Geol.
ecosystems during the period 1960–1990, Hydrobiologia Env. 59: 87-97.
237(2): 67-79. Simpson, D.W. and Negmatullaev, S.K. (1981) Induced seismicity
Benedick, R.E. (1979) The High Dam and the transformation of the at Nurek Reservoir, Tadjikistan, USSR, Bulletin of the
Nile, The Middle East Journal 33(2): 119-144. Seismological Society of America 71(5): 1561-1586.
Borovskiy, V.M. (1980) The drying out of the Aral Sea and its Wang, M.R., Yang, M.Y., Hu, Y.L., Li, Z.J., Chen, Y.T., Jin, Y. and
consequences, Soviet Geography 21(2): 63-77. Feng, D. (1976) A preliminary study on the mechanism of the
Carder, D.S. (1945) Seismic investigations in the Boulder Dam area, reservoir impounding earthquakes at Hsinfengkiang, Scientia
1940--1944, and the influence of reservoir loading on local Sinica, 19(1): 149.
earthquake activity, Bulletin of the Seismological Society of Whish-Wilson, P. (2002) The Aral Sea environmental health crisis,
America 35(4): 175-192. Journal of Rural and Remote Environmental Health 1(2): 29-
Chen, H. (2009) Wenchuan earthquake was not induced by 34.
reservoir, World of Science 5: 51-52. Wilson, M.P., Foulger, G.R., Gluyas, J.G., Davies, R.J. and Julian,
Elhance, A.P. (1997) Conflict and cooperation over water in the B.R. (2017) HiQuake: The Human-Induced Earthquake
Aral Sea basin, Studies in Conflict & Terrorism 20(2): 207- Database, Seismological Research Letters 88(6): 1560-1565.
218.
24
Stability Analysis of Buried Waste Water Pipeline in Soft Ground
E. C. Shin and J. K. Kang1

1
Department of Civil and Environmental Engineering, Incheon National University, Incheon, Republic of Korea
E-mail: ecshin@inu.ac.kr
ABSTRACT: Recently, ground collapse are increasing due to the old pipelines buried under the ground like waste water pipeline. The
purpose of this research was to analyze the causes of pipe deformation during the test construction of the channel pipeline improvement
project with the nature river work. The particle distribution, physical and mechanical characteristics were investigated through laboratory,
and field test using the additional boring exploration. Stability analysis was performed by applying soil cement wall with high strength deep
cement mixing method(SCW+HDCM) and Sheet Pile with HDCM method to two points, No. 84+4.0 and No.100+0.0, which are the most
vulnerable to subsidence of pipeline. The soft ground layer was reinforced by HDCM method for 2m to improve the strength of the clay soil
and to prevent settlement, satisfying the allowable residual settlement of 100mm.
Keywords: Waste water pipeline, stability, temporary earth wall, ground improvement, differential settlement.
1. INTRODUCTION 2. EARTH PRESSURE DURING EXCAVATION ON

SOFT GROUND
Recently, ground subsidence cases are increasing due to the
structures buried under the road. Much of this is caused by the The active thrust on the bracing system of open cuts can be
deterioration of the waste water pipe and the leakage of pipelines. estimated theoretically by using trial wedges and Terzaghi’s general
The number of leakage incidents is increasing as the city is aged. wedge theory (1941). Triangular distribution earth pressure theory
The increase in the underground space utilization due to the used in the design of retaining wall is significantly different in case
overcrowding of the city is accompanied by the excavation and of retaining wall in soft ground. The larger the deformation behaves
backfilling of the ground, which weakens the ground and increases the smaller the earth pressure.
the risk of collapse of the ground. When determining the construction depth of the retaining wall and
In general, the deep cement mixing(DCM) column improving the the cross-section of the self-supporting sheet pile, the earth pressure
soft clay ground by mixing chemical stabilizer which consisted of mainly used for Rankine-Resal earth pressure calculation is mainly
cement and lime at the original site is used for the infrastructure used. In the case of assuming that the back ground of the retaining
construction. Deep cement mixing is used to reduce the generation wall is horizontal, ignoring the wall friction angle with the wall, the
of waste during soft soil improvement and achieve low noise in a main earth pressure and the passive earth pressure at the bottom of
short period of time. The fundamental improvement principle of the the excavation are expressed by the following Eq. (1) and (2),
deep mixing process is in the formation of a rigid hardened body respectively.
produced by the hydration reaction between the stabilizer and water.
The chemical reaction (pozzolanic reaction) between the product by
Pa = γt z w + γ' ( z − z w ) + qK A − 2c K A (1)
the hydration reaction and the marine clay material improves the
soft ground (Shin et al., 2009).
Deep mixing method started to be developed from a research
Pp = γt z w + γ' ( z − z w ) + qK P − 2c K P (2)
work by the Research Institute of Harbor Technology belonged to
the Ministry of Transport of Japan since 1976. At the same time,
lime column was developed and used by now in Sweden which is where, Pa is the main earth pressure at the depth of z, Pp is the
method of mixing soil in underground as injecting the powder of
quick lime into the ground through the air pipe with high pressure. passive earth pressure at the depth of z, γt is the wet unit weight of
In domestic study about deep mixing, since the SEC (special earth the soil, γ' is the unit weight of the soil in water, z is the depth to
concreting) method with which cement is used as hardening agent any point on the surface, z w is the depth from the surface to the
was introduced from Japan in 1985. It has been applied mainly to a
groundwater surface, q is the surface load on the surface, and ⏀ is
retaining wall, foundation for building, foundation of seawall or
the internal friction angle of the soil.
quay as a harbor construction. In the related research, Bergado et al.
(2002) studied rrecent developments of ground improvement in soft Experimental earth pressure distributions are presented based on
Bangkok clay. Kim et al. (2005) conducted a reliability analysis of actual field measurements, and Peck (1969) 's empirical earth
pressure distribution is the most used. These diagrams for cuts in
the external stability of the quay wall installed in the deep mixed
sand, soft to medium clay, and stiff clay are given in Figure 1.
soil. Park et al. (2006) studied reliability analyze with respect to
external stability of quay founded on deep mixing ground. Lee et al.
(2007) studied with respect to formation shape of cement mixing
bulb with construction condition of deep mixing method. Han et al.
(2007) studied about strength of cement mixing bulb by construction
condition of deep mixing method. Chon (2010) studied about
compressive strength characteristics for deep mixing method. Kim
et al. (2011) analyzed the effect of the deep ground mixing and sand
treatment method on the application of the lower ground and
retaining line. Recently DCM lift injection method has been applied
in Incheon coastal area (Park, 2017).
The purpose of this study was to analyze the cause of
deformation and differential settlement during the installation of
waste water pipeline around a natural river and to propose a
countermeasure through stability analysis. Figure 1 Pressure diagram for cuts in sand(a), soft to medium
clay(b), and stiff clay(c) (Das & Sobhan, 2014)
25
The transverse earth pressure starts from the stationary earth passing of sieve number 200. It was carried out water washing
pressure. When the wall is pushed to the excavation side, the earth method and hydrometer distribution test. The particle size
pressure decreases to the main earth pressure. If it is pushed to the distribution of the soil is as shown in Figure 4. As a result of particle
back side, the earth pressure continues to increase but the manual size analysis by unified classification method, the soil type of KG-1
earth pressure can’t be increased. In other words, the minimum and to KG-3 was CL, and KG-4 was identified as ML and KG-5 as GC.
maximum earth pressure limits are set. The ground modeling is
simulated by a spring, and the basic equation of carbon spring is
given by the following Eq. (3).
d4y Ap E p
Ew I w + y = pi ± k h y (3)
dz 4 Lp
where, E w , Lw are elastic modulus and moment of inertia of the

earth retaining wall and A p , E p , L p are cross sectional area,
elastic modulus and length of the supporting structure, respectively.
pi is initial at rest earth pressure ( σ 0 ), k h represents the horizontal
reaction force coefficient.
Using Eq. (3) in SUNEX ver. w6.16(Jang, 2015) and EXCAV Figure 4 Particle distribution of sample
ver. 2.51(Oh, 2004), which are currently used as commercial
software, stability of the wall is analyzed. The lateral displacement The preliminary consolidation load was 79.36 kPa to
of the wall at each step, the shear force and moment acting on the 101.40 kPa , the compression index was 0.302 to 0.4337, Moisture
wall, and the axial force acting on the support are obtained. Figure 2 unit weight of soil was 17.18 kN/m3 to 18.09 kN/m3, and the initial
shows the analyzing model using the equation. void ratio was 1.162 to 1.420. Over consolidated ratio was about 1.0
as a normally consolidated soil. Table 1 shows the results of the
consolidation test on the undisturbed samples taken from the boring.
Table 1 Test results of consolidation by laboratory test
Consolidation Test(ASTM D2435)

Preconsol- Comp- Unit
No. idatin ression Swelling weight OCR eo
load, index, index, c s rt
P c (kPa) cc (kN/m3)
KG-1 82.92 0.41 0.11 18.09 1.04 1.42
KG-2 80.32 0.43 0.07 17.18 1.09 1.30
KG-3 79.36 0.39 0.06 17.66 1.08 1.19
Figure 2 Schematics of elastic beam model KG-4 101.40 0.30 0.05 18.05 0.95 1.16
3. SUBSURFACE EXPLORATION ON SOFT GROUND In-situ test was performed. Cohesion of soil was measured with
AND SOIL CHARACTERISTICS a field vane tester in order to confirm the undrained shear strength of
The total length of this construction is 7.9km and 3.55km is undisturbed state and disturbed state. The test results showed that
overlapped with natural river construction. A total of three the boring depth was about 3.5~5.0, cohesion of undisturbed sample
investigations were conducted on the design subsurface exploration with depth was 21.4 ~ 23.3 kPa , cohesion of disturbed sample was
of the waste water pipeline. In this study, the existing ground 2.6 ~ 3.5, and the sensitivity ratio of each boring was 7.15 ~ 8.23.
surveys were combined and re-confirmed the soft ground layer On the design, cohesion was similar with additional survey as 22.0 ~
through additional drilling of 5 holes. Figure 3 shows sewer pipeline 35.3, but sensitivity was not considered. Soil samples of KG-1, KG-
construction map and situation of subsurface exploration. 2, KG-3, KG-4, and KG-5 were very sensitive. Therefore, it is
expected that the ground has large deformation or the settlement
possibility is high due to the ground disturbance during excavation.
The results are shown in Table 2.
Table 2 Result of vane shear test in the construction site
c ur
cu cu
Boring Depth (distur
(undistur (Design, Sensitivity
No. (m) bed,
bed, kPa) kPa)
kPa)
KG-1 3.5 22.9 2.8 22.3 8.08
KG-2 4.0 23.3 2.8 23.9 8.23
KG-3 5.0 21.4 2.6 22.0 8.17
KG-4 5.0 28.4 3.5 35.3 8.13

Figure 3 Overview of waste water pipeline in soft ground area 4. ESTIMATION OF SOFT GROUND SOIL PROPERTY
The sample was mostly fine grained soil which is over 50%
26
Cohesion and internal friction angle were compared and examined Structural analysis was carried out a beam on elasto-plastic
by Dunham, Terzaghi-Peck, Meyerhof, Osaki, Schmertmann, and foundation model. It is similar to the beam on winkler foundation
Hisatake using empirical formulas based on N values. The design used to design piles with a foundation or horizontal load. The wall
constants were calculated as shown in Table 3 based on the stability at the final stage was evaluated from the stepwise
laboratory test results of the drilled specimens. excavation analysis. It is assumed that the stress is redistributed due
to the empirical earth pressure over time after the excavation is
Table 3 Soil property of shear strength for each layer in soft ground completed. The stability of the earth retaining walls and support
materials to the earth pressure is evaluated.
Unit weight, Cohesion, Internal friction
Soil type
r t (kN/m3) c(kPa) angle(°)
Soft soil property was revised using the additional ground
survey of 5 boring data, the subsurface exploration data of basic and
Reclaimed layer 19.0 10.0 20 detailed design of the sewer pipeline and the subsurface exploration
data of the basic and detailed design of the natural river
Accumulation
17.0 17.0 5 improvement project. As a result, the soft ground layer was deeper
(clay, N≤4) than the original design ground survey with a maximum of 12m, and
Accumulation
17.6 40.0 5
the ground layer also changed. Based on the revised plan, three
(clay, 4<N≤10) weakest section sites were selected. EXCAV and SUNEX were used
Accumulation to evaluate the stability of the pipeline by prefabricated wall.
18.0 5.0 25
(sand)
Weathered soil 19.0 20.0 30 5.2 Stability analysis for deep wall after ground improvement
Stability of construction depth of temporary wall was analyzed.
Weathered rock 20.0 30.0 33
When the construction depth of temporary wall is 4.0m at the
maximum excavation depth of 5.5m, the safety factor of the
Soft rock 23.0 100.0 33
construction depth is calculated to be 2.46 which satisfy 1.2 as a
standard of safety factor. The results are shown in Table 6.
Table 4 shows that the pre-consolidation load was 70.6kPa to
Table 6 Safety factor of section no.18+11.00
85.1kPa, compression index was 0.325 to 0.522, swelling index was
0.06 to 0.124, consolidation coefficient was 1.96e-3cm2/sec to Constructi Safety factor
Excavation
8.09e-3cm2/sec, and initial void ratio values was 1.032 to 1.311. It Section
depth(m)
on
was applied in the design of sewer pipeline construction. depth(m) Result Minimum Standard
No.18+11.00 5.5 4.0 2.46 1.2
Table 4 Soil property on design around waste water pipeline
Initial
Preconsoli- Compress- Swelling Consolidation Safety for ground settlement was calculated. The section no.
void
Boring No. dation load, ion index, coefficient
ratio, 24+0.0 was 7.2m in the thickness of the soft ground, and no other
P c (kPa) index, c c cs c v (cm2/sec)
eo channel was buried in the vicinity. The section no.84+4.0 is 8.6m in
thickness of soft ground. An existing line appeared at a distance of
BH-1 70.6 0.522 0.124 1.96E-03 1.311 2.98m from the position where the sewer pipeline is constructed.
The section no.100+0.0 point was 12m in thickness of the soft
BH-2 77.1 0.347 0.060 4.44E-03 1.032 ground, and the soft ground thickness of the three sites was the
largest. Figure 5 shows the result of settlement with elapsed time.
BH-3 85.1 0.325 0.063 8.09E-03 1.195
In this study, consolidation data was revised using additional

boring and existing boring data for the soft ground settlement
sections. The average value showed little bit lager than that of
design value. The revised soil property was shown in Table 5.
Table 5 Revised soil property around sewer line on this research
Preconsoli- Consolida- Initial
Compre- Swelling
Boring dation tion void
Division ssion index,
No. load, coefficient, ratio, Figure 5 Ground settlement with elapsed time
index, c c cs
P c (kPa) c v (cm2/sec) eo
No.8+0 ∼ Before ground improvement, settlement of some area using K-
No. KG-1 82.9 0.410 0.118 3.148E-03 1.42
50+18.0 Embank ver. 3 did not satisfy the criteria on the road settlement
No.50+18
standard. Table 7 shows the basement settlement of pipeline and
∼ allowable settlement after ground improvement.
BH-3 85.1 0.325 0.063 8.090E-03 1.19
No.
125+4.0 Table 7 Reviewed allowable settlement on weakest soft ground
Diameter Settlement After ground improvement
After STA. No. 50+18.0, there is a pressure pipeline along the Section of pipe of base
waste water pipeline in the adjacent area, and the overburden load is (mm) (mm) Result(mm) Criterion(mm)
expected to increase due to the embankment construction in the
No.84+4.0 1,000 146.1 65.3 100
future. Also it was found that the depth of soft ground layer was
deeper than that of original design from STA. No. 69+ 4.0 to STA. No.100+0.0 1,000 109.7 48.5 100
No. 125 + 4.0.
5. COMPARISON WITH NUMERICAL ANALYSIS 6. DISCUSSION
5.1 Elasto-plastic modelling for temporary earth wall As a result of the analysis of the soft ground, it is confirmed that it is
highly sensitive soft clay. Therefore, it is suggested to improve the
27
ground using the high strength DCM method(HDCM). Figure 6 stability of the pipeline. Therefore, it is suggested that the
shows cross section of No.100+0.0 as original design section and improvement of the soft ground should be done as a way to
reinforced section for ground improvement. minimize disturbance to the lower part of the basement.
2.Stability analysis was performed by applying SCW + HDCM

and Sheet Pile+HDCM method to the No. 84+4.00 and No.100+0.0
sites as the representative sections most vulnerable to subsidence of
pipeline.
3.As a result of 2m strengthening with HDCM method, the soft

clay ground was improved and the strength increased and settlement
(a) Original design section (b) Reinforced section amount decreased. The settling amount satisfies the allowable
residual settlement amount, 100mm.
Figure 6 Cross section No. 100+0.0 with ground improvement
8. REFERENCES
For the construction of sewer pipeline on the deep soft ground,
combination method like SCW, SGP, and sheet pile with ground ASTM D2435. (2014) Standard test method for one-dimensional
improvement were proposed. The lower part of the pipeline was consolidation properties of soils. Annual Book of ASTM
stabilized by using HDCM. The SCW method is effective in passive Standards, 04.08., pp246-260.
earth pressure resistance, forming foundation that resists heaving, Bergado, D. T. and Lorenzo, G. A. (2002) “Recent developments of
and in the reclamation area. The SGP method is capable of forming ground improvement in soft Bangkok clay”, Proceedings of
a foundation that is resistant to heaving, and is inexpensive when the International Symposium on Lowland Technology, Saga
buried. The sheet pile has the advantage of being able to increase the University, Japan, pp17-26.
effect of passive earth pressure resistance and to form the foundation Chon, S. G. (2010) Compressive strength characteristics for deep
to resist the heaving. We propose a method to prevent disturbance mixing method, Ph. D. thesis of L. N. Gumilyov Eurasian
by using the semi-shield method in the section after No.100+0.0 National University, Republic of Kazakhstan.
where the soft ground depth is highly deep in the lower part of the Das, B. M. and Sobhan K. (2014) Principles of geotechnical
basement. Table 8 shows the result of safety review of the engineering, CENGAGE Learning.
construction depth safety of the representative section. Han, W. S., Kwon, S. H., Kim, S. M., and Lee, K. Y. (2007)
“Strength characteristics of improved soil on construction
Table 8 Comparison of safety factor on temporary wall type after conditions of deep mixing method”, Proceedings of Korean
soil improvement Geotechnical Society Conference 2007, pp923-928.
Jang C. S. (2015) Elasto-plastic analysis of step underground
Excav- Const-
Maximum Safety factor excavation(SUNEX) manual 13th edition for ver. w6.16.,
ation ruction
Division settlement
depth depth Geogroup Eng. Co., Ltd.
(mm) Result Decision
(m) (m) Kim, Y. S, Choo, J. H., and Cho, Y. S. (2011) “Applicability study
on deep mixing for urban construction”, Journal of the Korea
Original design 7.1 5.9 76.47 1.786 O.K Academia-Industrial Cooperation Society, Vol. 12, No. 1.,
pp500-506.
Reduce Kim, Y. S., Park, O. J., and Huh, J. W. (2005) “Reliability analysis
construction EXCAV 7.1 1.0 17.19 1.875 O.K
for the external stability of quay walls constructed on deep-
depth
cement-mixed ground, Proceedings of Korean Society of
(SCW+ SUNEX 7.1 1.0 31.34 2.13 O.K
HDCM) Civil Engineering Conference 2005, pp2104-2107.
Lee, K. Y., Kwon, S. H., Kim, S. M., and Han, W. S. (2007) “Shape
Reduce
construction EXCAV 7.1 1.0 18.88 2.188 O.K characteristics of improved soil on construction conditions of
depth deep mixing method”, Proceedings of Korean Geotechnical
(SGP+ Society Conference 2007, pp226-223.
SUNEX 7.1 1.0 28.89 2.43 O.K
HDCM) Oh, J. W. (2004) Program on the design of propped retaining walls
for excavation by numerical method(EXCAV) manual ver.
Reduce
construction
EXCAV 7.1 1.0 10.22 2.237 O.K 2.51, Korean Geo-Consultants Co., Ltd.
depth (Sheet Park, C. H. (2017) Stability improvement effect of quay wall
Pile+ foundation ground using DCM lift injection method, Ph. D.
SUNEX 7.1 1.0 21.45 2.43 O.K thesis of Incheon National University, Republic of Korea.
HDCM)
Park, O. J., Huh, J., and Kim, Y. S. (2006) “Reliability analysis and
sensitivity analysis of quay walls constructed on deep-
7. CONCLUSION cement-mixed ground”, Proceedings of Korean Society of
Civil Engineers Conference 2006, pp2278-2281.
This study carried out to find the causes of pipe deformation during Shin, E. C., Kang, J. K., Oh, D. H., Lee, B., Kim, B. H., and Jung, S.
the test construction in the overlapping section of the sewer pipeline P. (2009) “Effect of Chemical Improvement by Hardening
and natural river construction. It was proposed the countermeasures Agent on Soft Soil”, Proceedings of 2009 Conference Co-
for the application. In order to clarify the causes of pipe deformation Hosted by KISTEC & KGES, pp. 427-432.
and differential settlement, the present state of soft ground was
reviewed by collecting the drill log data. In addition, the distribution
and physical soil characteristics of the boring data were examined
through laboratory and field experiment test based on the 5 boring
data. The results of the causes and countermeasures of pipe damage
are presented as follows,
1. In case of high sensitivity, it may cause the settlement due to

ground disturbance at the extraction, which may damage the
28
New Solutions to Geotechnical Challenges for Coastal Cities

J. Chu1, S.F. Wu1, H. Chen1, X.H. Pan1, and K. S-L. Chaim2
1
School of Civil and Environmental Engineering, Nanyang Technological University, Singapore
2
Geological and Underground Projects Department, Build Control Authority, Singapore
E-mail: cjchu@ntu.edu.sg
ABSTRACT: Coastal cities like Singapore and Jakarta have undergone rapid economic development in the past decades. There is an
increasing demand for infrastructure upgrading. With climate changes and population increase, geotechnical design and construction are
getting more and more challenging for coastal cities. We need to develop new technologies and new solutions to tackle the challenges and
enhance our ability to fight disasters such as earthquake and flood. In this paper, four different solutions pertinent to geotechnical
engineering for coastal cities are presented. The first is the establishment of a web-based 3D geological map to make better use of geological
and geotechnical data and information for more reliable geotechnical design. The second is a new method for using dredged slurry or soft
materials for land reclamation, as there is a shortage of granular fill materials in many cities. The third is the technology for construction of
seawalls using suction caissons. The last solution is the biogas desaturation method for mitigation of soil liquefaction which is becoming the
most cost-effective solution for prevention of liquefaction for a large area.
Keywords: Coastal cities, geological model, geotechnical engineering, land reclamation, liquefaction, underground construction.
1. INTRODUCTION and eventually a safer and more cost-effective design and

construction for future underground projects.
Many big cities are situated along the coast. Examples in Asia
include Hong Kong, Ho Chi Minh City, Jakarta, Manila, Shanghai,
Singapore, Thailand, and Tokyo. To cater for further economic 2.2 3D Geological Modelling
development, there is an increasing demand for infrastructure The GeM2S developed is a web-based design tool for future
expansion or upgrading along the coast. With global warming and underground projects. This system enables a geotechnical engineer
climate changes, geotechnical design and construction for coastal to a) utilize both geological and geotechnical data to construct a
cities are getting more and more challenging. Some of the geo-model as part of desk study, identify risks and design a targeted
challenges include the need for more space and more land. This site investigation program; b) enhance the reliability of the design
requires large-scale land reclamation and underground development. for underground construction by accounting for the uncertainties and
For land reclamation, a more cost-effective and operation-efficient spatial variability in the geo-model and geotechnical design
way is to reclaim land along the coast. This involves the use of a parameters; c) provide a web-based system for geological
huge amount of fill materials. Sand as s traditional land reclamation formations and cross-sections with key engineering properties and
fill material is not always available. In this case, soft fill materials or geotechnical design parameters to be analysed and visualized in 3D
waste may have to be used. Then the placement and treatment of models, or be incorporated into the Building Information Modelling
soft fill materials become difficult. For underground development, (BIM); d) set up a site rating system as required by the project based
one of the major challenges is the uncertainties in the geological on the quantity and quality of the geo-data and spatial statistical
conditions. Another challenge facing coastal cities is sea level rise analyses to guide future site investigation for underground
and the increasingly frequent coastal disasters such as earthquake construction and highlight potential geological hazards.
and flood. This requires higher or stronger seawalls to be SubsurfaceViewer was employed as the 3D geological
constructed. Liquefaction of soil is often one of the main causes for modelling software. The software methodology of 3D geological
the losses in lives and properties during earthquake. Mitigation of modelling is based on a single simple philosophy—the construction
liquefaction over a large area is another technical challenge. In this of geological subsurface models has to proceed with an
paper, some of the solutions developed to overcome the above understanding of the complete geological sequence and the likely
challenges are presented. geomorphological evolution of the study area. The procedure for the
construction of the 3D geological model consists of the following
2. 3D GEOLOGICAL MODEL steps: 1) preparation of the geological data, 2) creation of geological
cross-sections and fence diagrams; and 3) construction of 3D
2.1 Background
geological models. More detailed descriptions are available in Pan et
It is well-known that geological uncertainties is one of the major al. (2018).
challenges for underground construction. One way to overcome this
is to establish a three-dimensional (3D) geological model using
available geological and geotechnical data. With a 3D model,
engineers and urban planners can comprehend and visualize the
subsurface conditions in a more systematic way. A research project,
named web-based 3D geo-data modelling and management system
(GeM2S), has been carried out by Nanyang Technological
University (NTU) together with the Building Control Authority
(BCA), Land transport Authority (LTA) and Urban Redevelopment
Authority (URA) in Singapore to manage and utilize the existing
shallow borehole data to build a 3D geo-model of the subsurface
layers. BCA will combine this 3D subsurface geology model with
information from the deep geological surveys to develop a whole 3D
geological model for Singapore. This model will become a valuable
resource for future underground development as it can lead to a
better understanding of the subsurface geology structures in
Singapore, reduction in the uncertainties involved in the design, Figure 1 Fence diagram for one area in Singapore
29
As an example, the fence diagram constructed for one area in surcharge, or improved dynamic consolidation methods may be used
Singapore is shown in Figure 1. Using this fence diagram, a 3D instead (Chu et al., 2009b; Varaksin and Yee, 2009). Another
geological model can be constructed as shown in Figure 2. method is to premix cement with soft soil or mixing cement and soil
on-site or to use a combined cement mixing and consolidation
method (Chu et al., 2005). However, the use of cement is expensive.
When there are million cubic meters of fill to be treated, a slight
reduction in the unit cost can lead to a huge saving. Clay slurry may
also be dewatered using the geotextile tubes method in which clay
slurry is pumped into tubes made of woven geotextile. The solids of
the soil are retained by the geotextile and the water is consolidated
under the pumping pressure or self-weight. However, this method
may only be useful when the amount of soil to be improved is small.
Rigid inclusions or composite foundation methods by using columns
and reinforcements such as stone columns or deep cement mixed
columns are another type of methods to treat soft ground (Chu et al.,
2009b). However, these methods may not be suitable when the soil
is too soft to provide enough lateral support.
An alternative to vertical drains is to combine vacuum
Figure 2 3D geological model for one area in Singapore preloading with horizontal drains. One application as proposed by
Chu et al. (2005) and Chu and Lim (2008) is to use binders treated
The 3D geological model can not only assist engineers to sewage sludge as fill materials for land reclamation as illustrated
understand or visualise the underground geological conditions better, schematically in Figure 4.
but also help to make better interpretation of geotechnical properties
of the soil or rock. For example, using the 3D geological model
established for Singapore, we could identify the ancient river
channels. One example is given in Figure 3 (Pan et al. 2019).
Knowing the positions and extent of the ancient river channels, we
will be able to construction more accurate soil profiles or ground
models to enable more reliable geotechnical design to be achieved
for underground construction.
Figure 4 Use of horizontal drains plus vacuum preloading for the

consolidation of sewage sludge for land reclamation
(Chu and Lim, 2008)
There are a number of advantages in the use of horizontal drains

with vacuum preloading. Firstly, consolidation can take place as
soon as the clay layer is placed. The conventional method using
vertical drains, on the other hand, will have to wait until all the fill
materials have been placed. Secondly, the horizontal drains
accelerate the sedimentation process of the clay mud layer and thus
shorten the construction time. Thirdly, the strength of the clay can
be increased before the next layer of clay is placed. Fourthly, with
Figure 3 Ancient river channel identified using 3D geological model the use of horizontal drains, all the fills placed on top becomes the
fill surcharge as well. Hence the fill surcharge load increases with
the height of the fills. Finally, for the improvement of the soft soil
3. NEW LAND RECLAMATION METHODS fill from the ground elevation, a relatively thick layer of fill needs to
Land reclamation has become more challenging in recent years due be topped up to compensate for the large settlement. The new fill
to the following factors: 1) the available granular fill materials are will induce new consolidation and further settlement. Thus, it takes
depleting and soft or excavated soil such as dredged clay slurry have extra time for consolidation. It also requires a fairly accurate
to be used instead. How to improve the soft soil to enhance its shear estimation of the settlement caused by fills placed at different times.
strength and reduce the ground settlement due to the load by the In the past, vertical drains or composite drains with a width of
upper structures becomes a challenge; 2) disposal of industrial waste 100 to 300 mm have been used as horizontal drains. One problem
such as sewage sludge or incinerated ashes becomes a problem due with the discrete use of horizontal drains is that the positions of the
to the lack of dumping ground. One solution is to use these waste as horizontal drains become uncertain after the fill is placed on top or
fill materials for land reclamation; and 3) the water depth in the as the soil is undergoing consolidation. This causes uncertainties in
areas where land reclamation to be carried out is getting deeper and the design, analysis and quality control. Furthermore, the placement
deeper. The supply of a huge amount of fill materials becomes a of horizontal drains may involve in the intermittent use of a barge
challenge. The large amount of settlement induced by the heavy fill during the projects which increases operational costs. To overcome
materials is another challenge. the above problems, a new product called Horizontal Drainages
Various methods have been adopted for the treatment of soft soil. enhanced Geotextile sheet (HDeG) has been developed by Chu and
A review of some of the methods is given in Chu et al. (2009a, Guo (2015). A schematic illustration of the HDeG is shown in
2009b). For large scale land reclamation, one of the most Figure 5. Using this product, the horizontal drains will be placed at
economical methods is preloading together with prefabricated more or less the same elevation and the intervals between the drains
vertical drains (PVDs) (Chu and Raju, 2012). However, for very soft are more or less maintained. The new product HDeG also made it
or slurry types of fill materials, the use of fill surcharge becomes possible for horizontal drains to be placed without the use of a barge.
difficult. Vacuum preloading, combined vacuum preloading and fill Using the HDeG, the positions of the horizontal drains will be more
30
predictable and the reliability of the design and quality control of the of breakwater on soft seabed or in deep water. Suction caissons are
construction can be improved. If there is a need to accelerate the sunk into seabed using suction until sufficient bearing capacity is
consolidation process even further, electrolytes as either anode or obtained. In this way, the treatment of soft seabed soil is not
cathode can be embedded into the HDeGs to create electro-osmosis required. This will save both time and construction cost. This
effect that can also be incorporated into the horizontal drains as method can also be used for disaster rehabilitation or restoration of
shown in Figure 6. failed seawalls. The concrete caissons can be fabricated using
standard modulus in a casting yard connected to the sea. The
fabricated caissons or structures can be towed into the seawall
location and installed into the seabed using suction. With proper
design, the side friction of the caissons will provide sufficient
bearing capacity to safeguard stability and serviceability for the
seawalls. After the installation of suction caissons, the seawalls can
be installed on top of the suction caissons. One example of using
suction caissons as the foundation for a breakwater is shown in
Figure 11. The detail of this project can be found in Yan et al.
(2009).
Figure 5 Schematic illustration of one of the designs of the

Horizontal Drainage enhanced geotextile sheet (HDeG)
Figure 6 Use of electro-osmosis together with HDeGs Figure 7 Use of sand key as foundation for a rock bund for
reclamation in Singapore (Chew and Wei, 1980)
The performance of the HDeG product has been evaluated using
model tests and the testing results indicate that the use of HDeG is
effective (Chu, 2016; Chen, 2018). It should also be noted that no
membrane is required in this method as was case for the model tests.
Thus all the problems related to the use of PVDs and vacuum
preloading, such as sand blanket, the formation of a working
platform and the possibility of punching of membrane, were
eradicated. This is the other advantage of using horizontal drains. As
a first practical application, the proposed method has been used in
the design for a project under tendering.
Figure 8 Use of PVDs (not shown) and the weight of the rubble
4. CONSTRUCTION OF SEAWALLS mount to consolidate the soft seabed soil (not shown) before
Seawall construction has become more important for coastal semicircular concrete caissons were placed to form a breakwater
protection against the effects of sea level rising and natural disasters. in China (after Yan et al. 2009).
Seawalls can also be a way to fight flooding. The Giant Sea Wall
Jakarta is one of the examples.
The traditional method for seawall construction is to use sand
key to support embankment as shown in Figure 7. However, the
sand key construction involves dredging of a large amount of soft
clay from the seabed and placement of an equally large amount of
sand fill. Both the disposal of the dredged clay and the supply for
sand fill become problems. An alternative method is to use the so-
called non-dredging seawalls, or in other words, to build seawalls
without dredging of soft soil. This requires the soft seabed soil to be
improved. The methods adopted to treat soft seabed soil in the past
include the use of PVDs plus preloading (see Figure 8), deep cement
mixing or stone columns as shown in Figures 9 for a sand bund
construction in Singapore (Leong and Raju, 2007).
When water depth is relatively deep, the above methods may no Figure 9 Use of stone columns installed into soft clay from a barge
longer be applicable. In this case, suction caissons may become a to support sand bund (after Leong and Raju 2007)
better foundation type for seawalls. The suction caissons are made
of one or a group of steel or concrete cylinders as shown in Figure However, when the water is deep, or in other words, the upper
10 as an example. Four suctions are used as a group and installed structures or caissons to form the seawall is too big or too heavy, it
together to allow the penetrations of the 4 cylinders to be adjusted so will require a special offshore crane to install the upper caissons.
that the four suction caissons can sink into the clay evenly without This can be difficult or expensive. An alternative is to assemble the
any titling. This method is particularly suitable for the construction upper caissons with the suction caissons in the casting yard
31
and install the whole segment as one piece as shown in Figure 12. 5. MITIGATION OF SOIL LIQUEFACTION
Upon reaching the site, the whole segment can be ballasted using
water to the seabed level before suction is applied to sink in the Liquefaction of granular materials is one of the major geohazards.
suction caissons. The sea wall formed with back fill behind is shown Conventional methods for mitigating liquefaction include soil
in Figure 13. compaction or deep cement mixing. However, both methods are
expensive for large-scale deployment. It is necessary to develop
methods for liquefaction mitigation to be effective as well as
economical. An experimental study for the development of a more
cost-effective approach for mitigation of liquefaction hazard is
presented in this paper.
It has been found by some researchers that inclusion of some gas
bubbles in a fully saturated sand can lead to an increase of shear
strength in cyclic triaxial tests (Yoshimi et al. 1989; Xia and Hu
1991). Use of the desaturation method to enhance the resistance to
liquefaction has been attempted before (Yang et al. 2004; Okamura
and Soga 2006; Yegian et al. 2007; He and Chu 2014).
A biogas-desaturation method has been developed in this study
as a liquefaction mitigation approach. The denitrification process is
employed to generate nitrogen gas bubbles in sand to reduce the
degree of saturation of a fully saturated sand and thus increase its
resistance against liquefaction. Denitrification is a microbially
Figure 10 Suction caissons used as foundation for seawalls facilitated process of nitrate reduction that produces molecular
nitrogen (N2) through a series of intermediate gaseous nitrogen
oxide products. The complete denitrification process can be
expressed as a redox reaction as Eq. (1).
(1)
A series of shaking table tests were also carried out and the
results also show that the pore pressure generation and settlement in
the bio-desaturated sand was largely contained by using the
proposed method. Figure 14 shows the development of pore
pressure ratio in sand with different degrees of saturation under the
same input acceleration of a = 1.5 m/s2. Ru is defined as the ratio of
maximum excess pore pressure generated by the cyclic load to the
Figure 11 Use of suction caissons for a breakwater construction initial effective overburden stress. The pore pressure increased
in China (after Chu et al. 2012) during the cyclic loading and dissipated afterwards. For fully
saturated sand, there was a considerable amount of increase in
excess pore pressure as the shaking took place. The pore pressure
ratio exceeded 0.9 which indicates that the liquefaction occurred (He
et al. 2013). The pore pressure generated in biogas desaturated sand
were substantially lower than that in fully saturated sand. When the
degree of saturation Sr dropped slightly to 90%, the increase in pore
pressure becomes insignificant. The maximum Ru ratio in the biogas
desaturated sample (Sr = 90%) was less than 0.2 which is far less
than a trigging value of Ru = 0.5 when liquefaction could occur (He
et al. 2013).
Figure 12 Towing of suction caissons with seawall structures into

the seawall site for installation
Figure 14 Change of pore water pressures in shaking table tests

for fully and desaturated sand with a degree of saturation of 90%
There is a significant difference in the vertical strain between the

saturated and the biogas desaturated sand specimen as shown in
Figure 15. When the sample was fully saturated, a considerable
Figure 13 Illustration of seawall formed using the proposed suction
settlement occurred which indicated that the sand sample liquefied.
caisson method with backfill placed behind the wall
When the degrees of saturation dropped to 90 percent, the
volumetric strain caused by the ground shaking were mostly
32
confined within only 1%. This is evident that the biogas desaturated drains accelerate the sedimentation process of the clay mud
sand had strong resistance to liquefaction. This finding corroborates layer and thus shorten the construction time. Thirdly, the
the description of Tokimatsu and Seed (1987), who reported that strength of the clay can be increased before the next layer of
volumetric strains observed in non-liquefiable soils were usually clay is placed. Fourthly, with the use of horizontal drains, all
less than 1%. the fills placed on top becomes the fill surcharge as well. Hence
the fill surcharge load increases with the height of the fills.
3) For seawalls to be constructed in relative deep water, the
suction caisson method is more cost-effective as treatment of
he soft seabed soil is not required.
4) The biogas desaturation method is effective for mitigation of
soil liquefaction as proven by the shaking take tests. When
there is ground water seepage, the stability of the gas bubble
may become a concern. In this case, the biogas desaturation
method can be combined with biosealing method to prevent the
migration of gas bubbles.
7. ACKNOWLEDGEMENTS
The support from a number of finding agencies in particular, the
National Research Foundation and the Ministry of National
Figure 15 Vertical strain development in sand under shaking for Development, Singapore, is gratefully acknowledged. The
fully and desaturated sand with degree of saturation of 90% contributions from several other researchers and students to the
studies presented in this paper are also gratefully acknowledged, in
When there is no or very small seepage flow in the soil, the particularly, Prof Jia He, Prof Wei Guo, Prof Hailei Kou, Dr Bing Li,
biogas desaturation method is sufficient for mitigation of Dr Chuangzhou Wu, and Dr Kok Pang Lam.
liquefaction hazards. However, when there is a relatively big
seepage in soil, the stability of the gas bubbles may become a 8. REFERENCES
concern. In this case, a combined biogas desaturation and Chen, H. (2018). “Prefabricated horizontal drains with vacuum
bioclogging or the so-called “combined biodesaturation and preloading for improvement of soft marine clay.” PhD first
bioclogging” method could be used instead (Wu, 2015). In this year report, Nanyang Technological University, Singapore.
combined approach, the purpose of bioclogging is to “block” the Chew, S.Y. and Wei, J., (1980) “Major Reclamation Scheme for
passage for small gas bubbles to aggregate into bigger bubbles so as Marina City, Singapore”. Coastal Engineering 1980, pp.
to pre-empt the conditions for gas bubbles to escape from the 2245-2260.
ground. The amount of biogrout required for bioclogging in this Chu, J. (2016). “Innovation in Soil Improvement Methods”, S L Lee
case is much less than that for biocementation for the purpose to Lecture, Proceedings 19th Southeast Asian Geotechnical
increase the shear strength of sand. In terms of cost-effectiveness, Conference & 2nd AGSSEA Conf., Kuala Lumpur 31 May –
the construction cost involved in the combined biodesaturation and 3 June.
bioclogging method for liquefaction mitigation will still be Chu, J., Bo, M.W. and Arulrajah, A. (2009a). “Reclamation of a
significantly lower than the cost in the solo biocementation method. slurry pond in Singapore”, Geotechnical Engineering - Proc
The detail of the combined biodesaturation and bioclogging method of ICE, 162(1), pp 13-20.
will be presented in a separate paper. Chu, J. and Guo, W. (2015). “A geotechnical apparatus and method
of usage”, SG Patent Application, Serial No. 10201509039Y
6. CONCLUSIONS on 3 November 2015.
Four different solutions pertinent to geotechnical engineering for Chu, J., Goi, M. H. and Lim, T. T. (2005). “Consolidation of cement
coastal cities are presented: 1) web-based 3D geological map to treated sewage sludge using vertical drains”, Canadian
make better use of geological and geotechnical data and information Geotechnical Journal, 42(2), pp 528-540.
for more reliable geotechnical design; 2) horizontal drain enhanced Chu, J. and Lim, T. T. (2008). “Use of sewage sludge and copper
geotextile method to use dredged slurry or soft materials for land slag for land reclamation.” Geotechnics of Waste
reclamation; 3) technologies for construction of seawalls using Management and Remediation, GeoCongress 2008, ASCE
suction caissons; and 4) biogas desaturation method for mitigation Geotechnical Special Publication No. 177, pp 352-359.
of soil liquefaction. The following conclusions can be made: Chu, J. and Raju, V. (2012). “Prefabricated vertical drains”. Chapter
1) The web-based 3D geological map allows all the existing 4 in Ground Improvement, pp.87-167.
geological data and information as well as geotechnical data Chu, J., Yan, S. W., Li, W. (2012). “Innovative methods for dike
and information to be systematically ultilised for better construction – An overview.” Geotextiles and
geological design. This system makes future site investigation Geomembranes, 30(2), pp 35-42.
much more effective and reliable by providing more data and Chu, J., Varaksin, S., Klotz, U. and Mengé, P. (2009b).
additional geological processes (such as ancient river channels “Construction pro-cesses”. State-of-the-Art-Report, Proc.
or igneous intrusion etc) for geological profiling. The data 17th Int. Conf on Soil Mech. and Geot. Eng., Alexandria,
interpretation also becomes more meaningful when integrating Egypt, 5-9 Oct., 4, pp 3006-3135.
the new site investigation data with the existing geological data He, J. (2013). “Mitigation of liquefaction of sand using microbial
and geological map. methods.” PhD thesis, Nanyang Technological University,
2) When there is a shortage of granular fill for land reclamation, Singapore.
soft slurry or other soft fill materials may have to be used. In He, J. Chu, J. and Ivanov, V. (2013) “Mitigation of liquefaction of
this case, the conventional PVDs plus surcharge or vacuum saturated sand using biogas”. Géotechnique, 63(4), 267-275.
method becomes inefficient or time consuming. The horizontal He J. and Chu, J. (2014) “Undrained responses of microbially
drainages enhanced geotextile sheet (HDeG) method offers desaturated sand under monotonic loading”. Journal of
several advantages over the conventional method. Firstly the Geotechnical and Geoenvironmental Engineering, 140(5),
HDeG method is much faster. The consolidation can take place pp503-515.
as soon as the clay layer is placed. Secondly, the horizontal
33
Leong, K.W. and Raju, V.R. (2007). “Use of vibro replacement Wu, S.F. (2015). “Mitigation of liquefaction hazards using the
method for the stabilization of an offshore slope on soft clay”. combined biodesaturation and bioclogging method.” PhD
Proc 16th Southeast Asian Geotechnical Conference, 8-11 thesis, Iowa State University, Ames, IA, USA.
May, Malaysia, pp547-550. Xia, H., and Hu, T. (1991) “Effects of saturation and back pressure
Okamura, M., and Soga, Y. (2006) “Effects of pore fluid on sand liquefaction”. Journal of Geotechnical Engineering,
compressibility on liquefaction resistance of partially 117(9), pp1347-1362.
saturated sand”. Soils and Foundations, 46(5), pp695-700. Yan, S.W., Chu, J., Fan, Q.J., and Yan, Y. (2009). “Building a
Pan X., Aung Z., Nyo A.K.K., Chiam K., Wu D.F., and Chu J. breakwater with prefabricated caissons on soft clay”.
(2019). "Interpretation of distribution of ancient rivers in Proceedings of ICE, Geotechnical Engineering, 162(1): pp3-
Singapore using 3D geological model." In the Eighth 12.
International Conference on Case Histories in Geotechnical Yang, J., Savidis, S., and Roemer, M. (2004) “Evaluating
Engineering, Philadelphia, PA, USA. liquefaction strength of partially saturated sand”. Journal of
Pan, X., Guo, W., Aung, Z., Nyo, A. K., Chiam, K., Wu, D., and Geotechnical and Geoenvironmental Engineering, 130(9),
Chu, J. (2018). "Procedure for Establishing a 3D Geological pp975-979.
Model for Singapore." In GeoShanghai International Yegian, ., Eseller-Bayat, E., Alshawabkeh, A., and Ali, S. (2007)
Conference, Springer, Singapore, pp81-89 “Inducd-partial saturation for liquefaction mitigation:
Tokimatsu, K. and Seed, H. B. (1987) “Evaluation of settlements in experimental investigation”. Journal of Geotechnical and
sands due to earthquake shaking”. Journal of Geotechnical Geoenvironmental Engineering, 133(4), pp372-380.
Engineering, 113(8), pp861-878. Yoshimi, Y., T., K., and Kohji, T. (1989) “Liquefaction resistance
Varaksin, S. and Yee, K. (2007). “Challenges in ground of a partially saturated sand”. Soils and Foundations, 29(3),
improvement techniques for extreme conditions: Concept and pp157-162.
Performance”, Proc. 16th Southeast Asian Geot Conf, 8-11
May: Kuala Lumpur, pp101-115.
34
Use of Strut Free Systems in Deep Excavations

Chang-Yu Ou1 and Aswin Lim2
1
National Taiwan University of Science and Technology, Taipei, Taiwan, ROC
2
Department of Civil Engineering, Universitas Katolik Parahyangan, Bandung 40141, Indonesia
E-mail: ou@mail.ntust.edu.tw; aswin_lim@yahoo.co.id
ABSTRACT: This paper presents the performance of cross walls, buttress walls and their combination in reducing wall deflection and
ground settlement in excavations. Three dimensional finite element method was used to evaluate the performance. Cross walls have a very
good effect in reducing the movement but expensive. Buttress walls have a moderate effect in restraining the wall deflection. With proper
combination of cross wall and buttress walls, both the wall deflection and cost can be reduced significantly, even to achieve a strut free
excavation condition.
Keywords: Excavation, cross wall, buttress wall, strut free system.
1. INTRODUCTION wall (Ht) was 57.5 m. To reduce the lateral wall deflection and
ground settlement induced by excavation, 3 cross walls of 1.0 m
The huge population in urban areas due to economic development thickness and 26 m intervals were constructed and their depths were
has caused situations such as much denser buildings and deeper between GL-1.5 m and GL-45 m in the north–south direction. The
foundation excavations than ever before. Construction disasters and cross walls between GL0 m and GL-1.5 m were backfilled with the
adjacent building damage often occur due to ground settlement as a in situ soil, those between GL-1.5 m and GL-22 m were cast with
consequence of deep excavation, which not only affects construction 13.7 MPa concrete, and those below GL-22 m were cast with
progress, but also increases public nuisance. The protection of 24.0 MPa concrete. The cross wall were demolished with excavation.
adjacent buildings has become a major concern for designers and Figures. 3(a) and 3(b) show that the analyzed wall deflections at
contractors of deep excavations. It is thus an important hazard SO-1 and SI-8, respectively, for stages 7 and 9. For comparison,
prevention task in geotechnical engineering to conduct a study on analysis with assumption of no cross walls was also performed. The
protection of adjacent buildings when deep excavations are carried maximum wall deflection at the midpoint of two cross walls was
out. The integrity of adjacent buildings can be protected by predicted to have a 67% reduction by the installation of such walls
underpinning. Nevertheless, it is not practical to underpin a large (SO-1) whereas 77% at the cross wall section (SI-8). Installation of
number of target buildings, either from a financial or political point cross walls can significantly reduce lateral wall deflections [3].
of view. Therefore, reducing excavation induced movements to a
tolerable amount by auxiliary measures is a viable alternative. This
paper presents the performance of cross walls, buttress walls and Cross wall
(dismantle along
GL+0m 1FL GL+0.2 with excavation)
combination of cross walls and buttress walls in reducing the wall 
Fill t =18.25 kN/m
GL-3.0m
3
B1FL GL-4.4
deflection or ground settlement. Moreover, a strut free excavation SI-8 SI-7 Cross wall
CL
B2FL GL-9.0
 t =18.05 kN/m3
system is presented using a combination of cross walls and buttress SI-6
  =30o
B3FL GL-13.4
B4FL GL-16.8
walls. Diaphragm
Cc =0.4
C s =0.04
B5FL GL-20.2
B6FL GL-24.8
SI-9 wall SI-5
Inclined
steel strut
2. CROSS WALL Cross wall
GL-32.6m
CL
SI-10  t =18.74 kN/m3 Cross wall
Buttress   =31o ~33o
The basic configuration of a cross wall is shown in Figures. 1(a) and wall
SI-4 Cc=0.3; Cs =0.03 GL-45.0
SI-2
1(b), which refers to the construction of a wall, connecting two SI-3
GL-51.0m
SM/GW
retaining walls opposite each other, prior to excavation. SO-1 SI-12  t =19.62 kN/m3 GL-57.5
  =35 ~38
o o
Scale Unit:meter
Inclinometer 0 5 10 20 30 (m) N>50
GL-66.7m
Bedrock
(a) (b)
Diaphragm
wall Figure 2 The UPIB excavation project (a) plan (b) profile
Buttress wall
Excavation Excavation
Cross wall bottom bottom
12 10 8 6 4 2 0 10 8 6 4 2 0 12 10 8 6 4 2 12 10 8 6 4 2 0
0 0
Buttress wall Cross wall Buttress
wall 10 10
20 20
Depth (m)
(a) (b) (c)

30 30
Figure 1 (a) plan (b) cross walls (c) buttress wall
40 40
The cross wall functions as a strut-like component, which exists 50 50

before excavation. Along with excavation, cross walls provide a Stage 7 Stage 9 Stage 7 Stage 9
60 60
powerful resistance to counteract the lateral displacement, so as to (a) (b)
resist the lateral earth pressure on the back of the retaining walls. In Field observation
3D FEM analysis (with cross walls)
theory, movement of the retaining walls near the cross wall will be Equivalent beam model (with cross walls) Plane strain analysis (no cross walls)
restrained during excavation, and the lateral displacement of Figure 3 Comparison of analyzed and measured wall deflection
retaining walls will decrease. Ground settlement outside the (a) SO-1 (b) SI-8
excavation will be reduced too, which therefore achieves the
3. BUTTRESS WALL
protection of adjacent buildings.
Figure. 2(a) shows the UPIB excavation, which was installed The basic configuration of a buttress wall is depicted as shown in
with cross walls in the north-south direction. The excavation was Fig. 4(a). A buttress wall is similar to a cross wall in terms of
completed using the top–down construction method (Figure. 2(b)). construction. It is a concrete wall perpendicular to the diaphragm
The thickness of the diaphragm wall (t) was 1.5 m. The depth of the wall constructed before excavation, but not connected to the
35
opposite diaphragm wall. The location of the counterfort can be

arranged either at the inner or the outer side of the retaining wall as
shown in Figure. 4(a).
Lbw
He
Buttress walls Hbw
Buttress
Buttress
wall
wall
Diaphragm wall Cross
Hcw
wall
Hp
(a) (b)
Figure 4 Locations of buttress walls (b) Combination of Cross

and buttress walls Figure 6 Wall deflections for excavation with inner and outer
buttress walls
The author and his group have demonstrated that if the buttress
walls were demolished with excavation, the main mechanism of
buttress walls in reducing the wall deflection comes from the 4. COMBINATION
frictional resistance between buttress walls and adjacent soil [4] The
The buttress and cross walls can be combined to obtain more effect
combined stiffness of buttress walls and diaphragm walls play little
contribution in reducing the wall deflection. Figure. 2 also shows in reducing the wall deflection and cost. Fig. 4(b) shows a possible
three buttress walls with 12 m in length and 15 m in length were of combination. The effect depends on the buttress wall length
allocated in the west and east sides, respectively. The analyzed wall (Lbw), cross wall depth (Dcw) and cross wall spacing (scw). Figures.
deflection and ground settlement agreed well with the monitored 7(a) and 7(b) show the wall deflection at the middle of two BW/CW
values (Figure. 5). The analyzed wall deflection and ground for demolished buttress wall and maintained buttress wall for the
settlement for the case of buttress walls were moderately smaller
28 m and 8 m spacing, respectively. The wall deflection can be
than those without buttress walls. Installation of buttress walls
certainly has some effects in reducing the wall deflection and reduced significantly [5].
ground settlement. The amount of reduction in wall deflection was
67.7 mm at SI-9 and 52.6 mm at SI-10, almost one third of wall
deflection and ground settlement reduced.
Deflection (mm) Deflection (mm)

120 80 40 00 120 80 40 0
0 0
Lbw =4m (M)
10 10
Lbw =4m (D)
20 20 3D FEM analysis(with BW & CW)
Lbw =0m
Depth (m)
Depth (m)
3D FEM analysis(without BW & CW)

30 30 No BW and CW
Measurement
40 40
50 50
60 60
70 70
(a) (b)
Figure 5 Measured and computed wall deflections at the final stage
for UPIB (a) SI-9 (b) SI-10
(a) (b)
The author and his group also demonstrated that when buttress
walls maintained during excavation, i.e., no demolish with Figure 7 Spacing of BW/CW (a) 28 m (b) 8 m
excavation, the combined stiffness of buttress walls and diaphragm
walls plays a major role in reducing the wall deflection and ground
5. STRUT FREE EXCAVATION SYSTEM
settlement and frictional resistance between buttress walls and soil
play little role. The retaining wall behaves as a rigid stiff retaining Different types of Cross walls and cross walls can be combined to
wall and all of the wall, from top to bottom would move laterally. have more effect in reducing the wall deflection and ground
Fig. 6 shows the analyzed wall deflections at the final stage for the settlement. Figure. 8 shows an excavation with different
cases of buttress wall demolished with excavation, inner buttress arrangements of buttress wall where the excavation depth was 9.2
wall but maintains during excavation and outer buttress walls. The m. The maximum wall deflection at the final stage was about 50 mm
wall deflection can be reduced to a certain amount but its effect is [6].
still less than those of cross walls.
36
195.7m L = 100 m
Outer buttress wall Type E AH-2
SID 08 SID 10
Length = 5m (33m)
SID 09
North
AH-1
(30m)
Spacing = 8.5m BH-16 SID 11
BH-17 BH-18
Length = 6m (25m)
(25m) (15m) Cross Diaphragm
Type B Length = 8.5m BH-15
wall wall
B = 50 m
Type D (15m)
Ribs wall
88.6m
Middle section
BH-9 BH-10 BH-11 BH-12 BH-13 BH-14
8m
(25m) (25m) (25m) (25m) (15m) (15m)
Type C Cap-
BH-7 Length = 9.7m Buttress
8m
BH-6 (15m) Type A
(15m) Ribs wall wall slab
Inner buttress wall Length = 4.8m
BH-1 SID 07 SID 01
A(25m) A’
9m
SID 02
4m
SID 03
BH-19
SID 06
BH-3
SID 05 SID 04
BH-4 BH-20 AH-4
BH-5 Corner section Middle section
AH-3 (15m)
(40m) BH-2 (25m) (25m) (31.15m) (31.5m)
(30m)
SM 04
SM 03
SM 05
(25m) 13 m 13 m 8m 8m 8m
Figure 10 Plan view of the RFD strut free excavation system
Figure 8 A strut free excavation system by combination of different
shapes of buttress walls 6. CONCLUSION
Figure. 9 shows the RFD system, which comprises four main This paper presents the performance of cross wall, combination of
structures; namely, diaphragm walls, rib-walls, cross walls, and buttress wall and cross wall in deep excavations. Cross walls
buttress walls; and a complementary structure; namely, the cap-slab. function as lateral struts but exist before excavation. Installation of
The characteristics of the RFD system were (1) forming a cross walls in deep excavations can reduce the wall deflection to a
continuous earth retaining wall by constructing diaphragm walls very small amount. However, use of cross walls in a very wide
along a circumference of the excavated zone; (2) forming a rigid and excavation would be costly. Buttress walls can provide moderate
fixed retaining wall system by a series of rib-walls and cross walls improvement in reducing the wall deflection or ground settlement.
as shown in Fig. 9(a); (3) forming a rigid retaining wall by buttress The mechanism of buttress walls in reducing wall deflections
walls and the cap-slab. Fig. 10 shows the maximum wall deflection mainly come from the frictional resistance between the side surface
at the center of the long side is 165.4 mm, which is slightly larger of buttress wall and adjacent soil when buttress walls demolished
than that of the top-down construction. However, the maximum wall with excavation. When buttress walls are allocated outside the
deflection at the center of the short side is 50. 3 mm, which is much excavation, i.e., outer excavation, or buttress walls inside excavation
small than that of the conventional top-down construction. The but not demolished with excavation, the combined bending stiffness
amount of the wall deflection is related to several factors such as the from both diaphragm and buttress walls is very large, so that the
depth of the cross wall, length of the buttress, and length of the wall can deform linearly as a rigid retaining wall. Combination of
excavation side. Interested readers can refer to the references [7, 8]. cross wall and buttress wall can have good effect in reducing the
wall deflection and ground settlement. With proper arrangement of
C buttress wall or cross wall in excavations can even reach a strut free
L
excavation.
Buttress wall
7. ACKNOWLEDGEMENTS
Cap-slab The authors would like thank Ministry of Science and Technology
A A
(MOST) of Taiwan for granting us research funding for many years.
Soil Excavated The work presented in this paper mostly come from my previous
zone
Master and PhD students, especially from the outstanding work of
Dr. Aswin Lim.
Ribs wall
B B
Diaphragm
wall Cross wall 8. REFERENCES
Ou, C.Y. and Wu, C.H. "Effects of soil grouting on deep excavation
(a) behavior in sandy soil", Journal of the Chinese Institute of
Civil and Hydraulic Engineering, 2(2), pp169-182. [1]
Lbw Hsiao, K.J. (2016) "Use of Equivalent Parameters for Ground
Ground surface C
L Ground surface C
L
Cap-slab Cap-slab Improvement Piles in Deep Excavation Analysis", Master
Diaphragm Buttress wall Buttress wall
Thesis, Department of Civil and Construction Engineering,
wall Diaphragm National Taiwan University of Science and Technology,
wall
Taipei. [2]
He
Hsieh, P.G., Ou, C.Y. & Lin, Y.L. (2013) "Three-dimensional
Final excavation level Final excavation level
numerical analysis of deep excavations with cross walls",
Dcw Cross wall Acta Geotechnica, 8(1), 33-48. [3]
Hsieh, P. G. and Ou, C. Y., and Hsieh, W. H. (2016) "Efficiency of
Hp Cross wall Ribs wall
excavations with buttress walls in reducing the deflection of
the diaphragm wall", Acta Geotechnica, 11, 1087-1102. [4]
Lim, A. (2016) "Investigation of Integrated Buttress and Cross
soil soil Walls to Control Movements Induced by Excavation", PhD
Dissertation, Department of Civil and Construction
(b) Engineering, National Taiwan University of Science and
(c)
Technology, Taipei. [5]
Figure 9 The RFD strut free excavation system
37
Lim, A. and Ou, C. Y. (in print, 2018) "Performance and three-

dimensional analyses of a wide excavation in soft soil with
strut-free retaining system: case study", International Journal
of Geomechanics, ASCE. [6]
Lim, A. (2018) "Investigation of Integrated Buttress and Cross
Walls to Control Movements Induced by Excavation", PhD
Dissertation, Department of Civil and Construction
Engineering, National Taiwan University of Science and
Technology, Taipei, Taiwan. [7]
Lim, A., Ou, C. Y. and Hsieh, P. G. (2018) "A novel strut-free
retaining wall system for deep excavation", paper submitted
to Acta Geotechnica. [8]
38
Behavior of Single Pile and Pile Group Foundation for High Rise Buildings on
Expansive Soils
Paulus P. Rahardjo1 and Prieschila Tamsir2
1
Parahyangan Catholic University, Bandung, Indonesia
2
PT Geotechnical Engineering Consultant, Bandung, Indonesia
E-mail: rahardjo.paulus@gmail.com teknik@gec.co.id
ABSTRACT: High rise buildings supported by piles are now increasingly constructed in Indonesia especially in West Java (at Cikarang and
Cibitung area) and West Surabaya. The cases in West Surabaya is interesting because the depth of the expansive clays is more than 80 m.
The characteristics of expansive clays are very specific due to its capacity of swelling and becoming soft upon absorbing water from their
current unsaturated conditions, and yet as clay material, most designers worry about the long term compression. This paper discusses the
results of observation of the single pile and group piles from design, construction and performance in short term and long term settlement of
a complex of high rise buildings with three towers of 51 stories. The results of soil laboratory tests were disturbed except for water content
and index properties, hence the design has rely on the results of insitu testing including SPT, CPT and Pressuremeter Test (PMT). The
estimated settelement of pile groups of 14-15 cm were in fact much less (only 30%) and the short term settlement dominate almost 90% of
the total settlement. This fact may be related to the swelling characteristics and unsaturated soil condition
Keywords : pile foundation, expansive soils, pile group, settlement
1. INTRODUCTION that rich in volcanic material. Boring results in this area obtained
brownish grey clay, which is welknown as expansive material.
Behavior of single pile and pile groups to support high rise buildings
is always of interest to the engineers, and more specific is because
the buildings are founded on expansive soils. The main objective of
this paper is to review the behavior of a single pile and pile group
under three towers. A case study is presented to illustrate the actual
behavior of single pile and measured settlement of group piles. This LOKASI PROYEK
project is a mixed used building that include Apartment, Hotel,
Office and Mall in west Surabaya with 51 stories of Tower and 7
layers of Podium. Three towers and the podiums were connected to
each other as shown on figure 1, i.e. Tower Anderson, Tower
Benson and tower Lavis (tower 7) which was constructed the last.
The authors have been involved in the design and following
construction and performance of the foundation construction as well
as the subsequent settlement
Tower 7
Figure 2 Geological Condition (Badan Geologi Indonesia)
Tower Anderson
2.2. Soil Condition
Soil investigation was conducted by PT. Data Persada at the project

site in September 2015 consisting of 3 boreholes @ 50 m (BH-02,
BH-04 dan BH-05), 3 boreholes @60 m (BH-01, BH-03 dan BH-
06) with N SPT values at interval 2 m and undisturbed samples for
laboratory tests. Pressuremeter tests were also conducted to
investigate the at rest soil pressure and stress strain behavior in
cylindrical expansion. Description of soil stratification and
engineering properties are described in this paper
Figure 1 : Description of the project site

TOWER 7
2. GEOLOGICAL AND SOIL CONDITION

2.1 Geological Condition
According to Surabaya Geological Map, Location of this study is a
part of Lajur Kendeng and occupy Lidah Formation (Tpl) which
consist of blue clay stone, black spot, loamy, solid, and harden when
dry and soften and swell when exposed to water.
Figure 1 shows project location in the geological map. The sediment
rock in the study location is from Pliosen age and Tersiary period
Figure 3 Location of Soil Investigation (phase 1)
39
Bor Log
Nsp Nsp Nsp
B1 t B2 t
B3
t
0 5 10 15 20 25 30 35 40 45 50 0 5 10 15 20 25 30 35 40 45 50 0 5 10 15 20 25 30 35 40 45 50
0 0 0
2
4
2
4
2
4
NSPT (blows)
0 10 20 30 40 50 60
6 6 6
8 8 8
10 10 10
0
12 12 12
14 14 14
16 16 16
18
20
18
20
18
20
2
22
24
22
24
22
24 4 B1 B2
6
26 26 26
Depth
(m)
Depth
28 28 28
Depth
Depth
(m)
B3 B4
(m)
(m)
30 30
8
30
32 32 32
34 34 34
36 36 36
10
38 38 38
B5 B6
12
40 40 40
42 42 42
14
44 44 44
46 46 46
48 48 48
50
52
50
52
50
52 16
18
54 54 54
56 56 56
20
58 58 58
60 60 60
62 62 62
64 64 64
22
Nsp B4
Nsp
B5
Nsp
t B6 Nsp 24
26
t t t
25 30 35 40 45 50 0 5 10 15 20 25 30 35 40 45 50 0 5 10 15 20 25 30 35 40 45 50 0 5 10 15 20 25 30 35 40 45 50
0 0 0
2
4
2 2
4 28
Depth (m)
4
30
6 6 6
8 8 8
10 10 10
12
14
12
14
12
14
32
16
18
16
18
16
18 34
36
20 20 20
22 22 22
24 24 24
38
Depth
26 26 26
Depth
Depth
(m)
(m)
(m)
28 28 28
30
32
30
32
30
32 40
42
34 34 34
36 36 36
38 38 38
40
42
40
42
40
42 44
44
46
44
46
44
46 46
48
48 48 48
50 50 50
52 52 52
54
56
54
56
54
56 50
52
58 58 58
60 60 60
54
62 62 62
64 64 64
= Clay ; Soft to medium Consistency 56

= Clay ; Stiff Consistency
58
60
= Clay ; Very Stiff Consistency 62
64
= Clay ; Hard Consistency 66
68
70
Figure 4 Soil Stratification from 6 boreholes
Figure 5 Profiles of N-SPT values of soil
Figure 4 shows soil stratification from 6 boreholes; in general the PL, W and LL vs Depth Plasticity Index
Plasticity Index
strata may be described as follows : 0

0 20 40 60 80 100 120
Plastic Limit
140
0
0 10 20 30 40 50 60 70 80
Liquid Limit
• The surface layer or first layer is soft to medium clay of

Water Content (%)
10-12 m thickness with N SPT varies from 2 – 10. This 5 5
layer may be thin sediment of part of the active zone of

the expansive clays that has been influenced by water 10 10
infiltration causing the clay to softens.

Depth (m)
Depth (m)
• The second layer may be described as stiff clay found at

15 15
depth 10-22 m and increasing values of N SPT 13-20

20 20
• The third layer is very stiff clay found at depth 20-50 m

and increasing values of N SPT 20-30. This layer was the 25 25
bearing layer of the foundation where the tip was designed

at 38-42 m depth Figure 6 : Water content and Index Properties
• The fourth layer may be described as hard clay found at All data show that the soil is highly plastic clays or clayey silts
depth 50-60 m and increasing values of N SPT > 30m. having liquid limits of 80-130% and plasticity index of 45-85%.
Those are very high values compared to general soil (figure 7)
The distribution and N SPT profile may be shown on figure 5 and we
can see that the strength is consistently increasing with depth. Water IP vs LL
table is not detected through the whole layer, and if water is found in 100
the borehole, they are trap water or perch water table. This condition 90
is favorable for the case of long term or consolidation settlement 80
since practically no significant pore pressure will be developed.

70
60
Plasticity Index
2.3. Results of Laboratory Test

50
CH
40
30
Laboratory tests conducted for the project include Index Properties, 20
OH and MH
strength by Triaxial UU & CU, consolidation test and swell test. CL

10 CL-ML
The data show that water content is found around 42% which is 0
ML and OL
closer to plastic limits and even in some depth less that the plastic 0 10 20 30 40 50 60 70 80
Liquid Limit(%)
90 100 110 120 130 140
limit, this and that tells why the soil consistency has fall in the stiff Figure 7 : Plasticity of the soils
to hard clays.
40
Liquidity Index Liquidity Index been based on SPT values and for the stress history and soil
-0.8 -0.6 -0.4 -0.2 0 0.2 stiffness, the data from pressurementers have been used.
0
Due to uninspected poor data on strength and stiffness of laboratory

tests, limited pressuremeter tests were conducted at 10m, 20m, 35m,
5 45m in BH-2 and at 12m, 20m, 35m, 45m in BH-4.
Pressuremeter Test (PMT) is the best geotechnical data developed
by expansion of cylindrical cavity at insitu soil condition whch
10
gained more popularity due to the fact that the tests are carried out
under insitu stress condition, water content, producing insitu stress
Depth (m)
strain behavior at elastic and plastic condition, fast and economical,

also direct use for design and may be done continuously. Parameters
15
obtain by pressuremeter test include
1. Po = ground pressure at rest (kg/cm2)
2. Py = yield pressure (kg/cm2)
20
3. Em = elastic modulus (kg/cm2)

4. Gm = shear modulus (kg/cm2)
25
5. Cu = undrained shear strength of ground materials
Figure 8 : Liquidity Index
2
(kg/cm ) (after Gibson & Anderson, 1961)
Liquidity index of the soils are mostly very low and in some cases
are less than 0. The yield stress ratio or overconsolidation ratio will
be very high and those, this phenomena will normally reduce the
settlement.
Swelling characteristics of clay

The swelling characteristics of clay is measured with oedometer,
where load is applied to the sampel up to the calculated overburden
pressure, and then swelling is allowed by addition of water. Under
the effective overbudern pressure the soil still capable of exceeding
the overburden pressure such as shown on figure 9. This swelling
characteristic can bring up low rise building less than five stories
and the swelling characteristic can result in reducing the settlement
of high rise building. This aspect is seldom measured or investigated
but in reality many buildings have been rise up including a four
storey buildings nearby
Figure 10 : Pressuremeter Test used for the project

Tabel 1 : Summary Pressuremeter data obtained at BH-2 and BH-4
P0 Py P1 km Em
Borehole No. Depth (m) rm (cm) SPT value
(kg/cm2) (kg/cm2) (kg/cm2) (kg/cm3) (kg/cm2)
10 0.58 7.27 8.4 63.5 354.66 3.72 7
20 1.02 10.64 15.26 58.82 310.49 3.52 23
BH-2
35 2.29 18.98 26.02 42.01 238.14 3.78 29
45 2.52 13.6 25.82 68.53 364.27 3.54 33
12 0.56 6.66 12.15 10.97 61.28 3.72 13
20 1.84 6.43 10.83 17.54 109.81 4.17 21
BH-4
35 2.66 15.8 25.8 38.23 204.51 3.57 27
45 3.77 12.73 24.78 57.73 327.36 3.78 34
Although only limited data obtained from PMT, the data has been
spread at different depth (figure 11) and may be used to estimate
the variation of the test results along the depth.
Py vs Depth P1 vs Depth Em vs Depth

Py (kg/cm2) P1(kg/cm2) Em(kg/cm2)
0 5 10 15 20 0 5 10 15 20 25 30 0 50 100 150 200 250 300 350 400
Figure 9 : sweling pressure charareristics 0 0 0
5 5 5
2.4. Pressuremeter Test Results 10 10 10
15 15 15
Problems with Expansive soils are due to the swelling and shrinking 20 20 20
characteristics. Once the sampel was retrived to the laboratory, there

Depth (m)
Depth (m)
Depth (m)
25 25 25
will be changes in water content or volume of the soils. Furthermore 30 30 30
if triaxial CU are conducted, the saturation takes very long time and 35 35 35
inclusion of water into the soil may change its behavior, the strength 40 40 40
and stiffness drops significantly. To overcome this problem, insitu 45 45 45

tests such as SPT, CPT and Pressuremeter Tests are carried out. In
practice this insitu tests have been the more reliable data for design
50 50 50
and deriving parameters for analysis. The foundation design has Figure 11 : Variation of PMT parameters with depth
41
The SPT and Pressuremeter modulus corelations has been well

developed such as shown on figure 12. The pressuremeter modulus
can be used for long term settlement since practically the modulus is
similar to drained modulus tested by triaxial test (Briaud, 1996 and
Roger Frank, 2013).
Figure 13 : Adhesion factor for pile friction (Kulhawy and Jackson,

1989)
For bored pile, similar method is done with an average adhesion

factor for friction as much as 0.55 (Reese and Wright, 1979)
The results are shown on figure 14
NSPT vs Depth Daya dukung Ultimate (kN) Daya dukung Ijin (kN)
Nspt
0 5 10 15 20 25 30 35 40 45 50 0 2000 4000 6000 8000 10000 0 1000 2000 3000 4000

0 0 0
Fig 12 Correlation of pressuremeter moduli (Em) and N

B1 B1 B1
5 B2
5 B2 5 B2
B3 B3 B3
B4 B4 B4
10 B5 10 B5 10 B5
B6 B6 B6
15 15 15
3. THE DESIGN OF PILE FOUNDATION

20 20 20
25 25 25
Based on the soil condition (medium to very stiff clay) and 30 30 30
Depth (m)
Depth (m)
Depth (m)
economical consideration, pile foundation has been selected to carry

35 35 35
40 40 40
the structure loads. However since the buildings are 51 storeys, the 45 45 45
authors have considered that the foundation shall be deep enough to 50 50 50
avoid excessive settlement. The main foundation system is spun

55 55 55
precast pile with 600 mm and 500 mm diameters. The installation

60 60 60
65 65 65
method is by used of push-in/jack-in pile. This jack-in method is to 70 70 70
avoid noise and vibration disturbance to the surrounding

neighborhood with additional benefit that we had known injection Figure 14 : Calculation for ultimate and allowable capacity of 600
force which is chosen as high as 250% of the work load or allowable mm diameter of spun pile
bearing capacity.
However, the development is very close to the surrounding 4. CONSTRUCTION OF PILE FOUNDATION
neighbour buildings, hence on the pheriphery or at the boundary, the
Figure 15 shows the installation of pile foundation using push
foundation selected is bored piles of φ800 mm diameter. In order to
in/jack in method where the pressure reach 250% of the design load
balance between the driven pile and the boredpiles, both foundation
and held 3 times 10 second at end of installation. This method has
system has been design to carry the same axial stiffness at their
proven to be reliable.
allowable load. The allowable load for driven piles is 200 tons and
based on the bearing capacity analysis the required length of the The problems of jack in method (as well as driven hammer) in
driven spun pile is 34-37 m. While for boredpile φ800 mm the same expansive soils are heaving of the neighbour and the problem of
axial stiffness is 35 m with allowable capacity of 300 ton. In some heave of the piles. The second problem was overcome by redriving,
area higher capacity of 400 tons for bored pile is also required and but we have to make sure that all length of piles can be exactly
calculated to be 43 m length pushed into the soils because the movement of the machine requires
that no portion of the pile shall be on ground level.
For cohesive soils, the general formula use for tip resistance is:
Qp = q ⋅ Ap
where : q =  D B 9 cu for D/B < 4
 4 
q = 9 cu jika D/B > 4
Ap = cross section area of pile
The friction resistance is calculated using adhesion factor as
suggested by Kulhawi and Jackson (1989)
Qs = α ⋅ Su ⋅ p ⋅ L
Where α = adhesion factor.
Su = undrained shear sttrength
p = pheriphery of piles
L = length of embedded pile Figure 15 : Pile installation method
The adhesion factor α have used the recommendation of Kulhawy
dan Jackson (1989).
42
Figure 16 : Preparation for Pile Injection Figure 20 : Concreting of boredpile with tremie
6. BEHAVIOR OF SINGLE PILE UNDER LOAD

The behavior of single pile is well predicted by the results of pile
load test. However this behavior only represent short term condition
of the piles and the effect of the settlement is only on a small scale.
The real condition has wider area and the impact will also be
different. However to a certain degree the single pile behavior may
be reflected in the group behavior
The folowing figure is typical results of pile load test in expansive
soils. For driven piles, the results may not be consistent due to the
condition during driving (such as heave or water penetration into the
gap between the soils and the piles). Three different results are
Figure 17 : Installation of Boredpile presented in figure 21. However for boredpiles, the results of pile
load test are more consistent (figure 22).
Pile Load vs Settlement Curve
Spun Pile dia. 600mm, Leff = 37m
Pile Load in ton
0 100 200 300 400 500

0
10
20
Pile Settlement in mm
30
40
Figure 18 : Use of casing for bored pile 50
60
Figure 21 : Typical results of pile load test for driven piles

SPI Phase 4
Pile Load vs Settlement Curve
Bored Pile dia. 800mm, Leff = 38m
Pile Load in ton

0 100 200 300 400 500 600 700
0
10
20
Pile Settlement in mm
30
Figure 19 : Installation of rebar for bored pile

40
50
60
Figure 22 : Typical pile test results for boredpile in expansive soils
43
7. SETTLEMENT ANALYSIS AND SETTLEMENT Potongan 3-3' Coordinates (m)
MEASUREMENT DURING CONSTRUCTION 0

0 20 40 60 80 100 120
Settlement Analysis 4
Settlement (cm)
Settlement analysis for this project have used two methods, the first 6
one is based on Poulos and Davis (1980) and the second analysis is 10
1
by Finite Element Method (computer software GTS Midas). The 12 400
method suggested by Poulos is based on the interaction between

14
piles and computed as follows : Figure 26 Settlement profiles cross section 3-3
xi ⋅ ∑ (Pj ⋅ α ij ) + x1 ⋅ Pi Potongan 4-4'
Si = 0 20 40
Coordinates (m)
60 80 100 120
where : 0
2
Si= settlement of pile i due to own load and other piles 4
Settlement (cm)
xi= settlement of single pile due to unit load (mm/ton) 6
1
Pi= load at pile no -i (ton)
8
400
10
Pj = load on pile no-j (ton) 12
αij= interaction factor between piles

14
Figure 27 Settlement profiles cross section 4-4

Interaction factor for friction pile and tip bearing piles are different.
Poulos derived curves for these interaction by assuming Poisson The settlement analysis is also carried out by finite element
ratio equal to 0.5. These interaction factors are for particular piles modelling, where the piles and raft foundation are modeled and soil
with different length/diameter. Poulos dan Mattes (1971) stated that structure interaction are also introduced.
αF are function of s/D, L/D, and K, where K is the ratio of the pile The model is shown on figure 28, where the towers are supported by
modulus and elastic modulus of the soils. The results of Poulos and the pile and raft foundation and the podium is supported by pile caps.
Davis method is shown as settlement of each pile and contour of The 3D model allow the interaction between pile, pile caps, raft and
settlement was plotted on figure 23 also the soil. The difference with Poulos and Davis Method is the
110.00
load carried out by pile cap and raft are not considered. If one wants
1 to exclude the raft caryiing load, then the upper soil layer may be
soften to reduce the portion of load to the raft. Figure 29 shows
100.00
distribution of settlement under the structures and the induced

90.00
settlement to the surrounding. Maximum settlement of the system is

80.00
about 15 cm.
70.00
60.00 4
50.00
3
40.00
30.00
20.00
10.00 2 2'
0.00
0.00 10.00 20.00 30.00 40.00
4'50.00 60.00 70.00 80.00 90.00
3' 1'
100.00 110.00 120.00 130.00 140.00 150.00 160.00 170.00 180.00 190.00 200.00 210.00 220.00 230.00
N600 spun 200t D600 BP 200t D800 BP 300t D800 BP 400t Series5 Series6 Tambahan D800 35m
Figure 23 : Results of Settlement Calculation by Poulos and Davis

(1980) method
The settlement profiles from Poulos and Davis Method are shown Figure 28 : Model of pile and raft on this project
by long section 2-2 and cross section 1-1, 3-3 and 4-4 as illustrated
in figure 24 – 27. Based on the cross section and longitudinal
section of the settlement profile, the slope may be presented. The
differential settlement must be limited to 1/300, and all the results of
calculation comply with this requirement.
Potongan 1-1' Coordinates (m)
0 20 40 60 80 100 120
0
4
Settlement (cm)
6
1
8
375
Figure 29 : Countour of settlement of the structure and the

10
12
1 1
surrounding
400 800
14
Figure 24 Settlement profiles cross section 1-1 The settlement profile can then be evaluated by looking at cross of
settlement profiles as shown on figure 30. It is shown that the results
Potongan 2-2'
0 50 100
Coordinates (m)
150 200 250
of finite element analysis are in line with the results of approach
0
Tower 7
Tower Tower using Poulos and Davis method.
2 Benson Anderson
4 The effect of the adjacent towers to the settlement is also shown on

Settlement (cm)
1
the plot, and differential settlement are more pronounced at the
location of the tower and the podium
8
625
10
12 1
600
14
Figure 25 Settlement profiles long section 2-2
44
Jarak vs Settlement Proyek SPI Phase 4

Potongan 1-1 (Potongan Memanjang)
Kulhawy, F.H., and Mayne, P.W.,1990, “Manual on Estimating Soil
-50 0 50 100 150 200
Jarak (m)
250 Properties for Foundation Design,” Report No. EPRI EL-
6800, Electric Power Research Institute, Palo Alto, CA
0
La Viz Tower
2
4 Poulos, H.G., 1977, “Estimation of Pile Group Settlements”, Ground

Settlement (cm)
Engineering, ch.10(2), pp.40 50

6
1
8
Poulos, H.G., Davis, E.H., 1980, “Pile Foundation analysis and

449
10 1
12
14
1000
Design”, John Wiley, New York
16 Rahardjo, P.P., Junita, Brotodihardjo, W., 1993, “Analisis Pondasi
Jarak vs Settlement Proyek SPI Phase 4
Tiang Tunggal dengan Beban Aksial Berdasarkan Metode
Potongan 2-2 (Potongan Melintang)
Jarak (m)
Pengalihan Beban”, Prosiding Pertemuan Ilmiah Tahunan I,
0
0 20 40 60 80 100 120
HATTI, Jakarta
La Viz Tower
2
4
Settlement (cm)
8
1
10
1 650
400
12
Figure 30 : Settlement profiles in Long Section and Cross Section

After all piles are installed, and pile cap completed, a number of
points are decided as points for settlement measurement. The
settlements were measured every week and the load of the structures
are indicated by the number of storeys constructed. The results are
plotted from time to time and the following figures are the
settlement of the structures.
Settlement for Tower 5
Figure 31 : Settlement of tower 5 (Benson)

This is very interesting topic since the settlement measurement give
much less settlement compared to the results of analysis by a factor
of more than 3. The short term settlement dominate the total
settlement by about 60-70% and long term settlement take very
short time in less than 7 months.
8. CONCLUSION SUMMARY
The study of the behavior of piles in expansive soils show that the
expansive characteristics influence the behavior of single pile and
group piles, in terms of there may be inconsistency on the single pile
behavior due to the construction histories. The predicted settlement
is much less than calculated which may be due to the heave and the
unsaturated soil condition. In general the results of insitu testings are
more reliable for design and analysis.
9. REFERENCES
Deep Foundation Research Institute (DRFI). (2017) Manual Pondasi
Tiang edisi 5, Universitas Katolik Parahyangan
45
Challenges In Design And Construction Of Deep Excavation

S.S. Gue, C.S. Gue and C.Y. Gue
E-mail: ssgue@gnpgroup.com.my
ABSTRACT: Due to scarcity of land, especially in urban areas, the need for basements to optimise the use of land has resulted in
construction of deep excavation works. Deep excavation works pose great challenges to geotechnical engineers, particularly in complex
ground conditions such as in limestone formation, where it exhibits notorious karstic features with irregular bedrock profiles, variable
weathering condition, cavities and slime zones. With proper geotechnical input, costly failures and delays associated with underground
works such as excessive groundwater lowering, occurrences of sinkholes, excessive ground settlements, etc. can be prevented. Suitable
temporary earth retaining system and rock strengthening works have to be properly designed to prevent such failures. This paper presents
design principles of temporary earth retaining stabilising system (ERSS) together with vertical rock excavation as well as three case studies
including from Mass Rapid Transport (MRT) projects, with various challenging geological formations found within the Klang Valley of
Malaysia; Limestone and Alluvium with high ground water table.
Keywords: Deep excavation; temporary earth retaining stabilising system, ERSS, limestone; Kenny hill, alluvium
1. INTRODUCTION information needed for analysis, design and finally producing

construction drawings.
With the expansion of underground transportation systems such
as the Klang Valley Mass Rapid Transit (KVMRT) in Malaysia, 2.1 Planning of Subsurface Investigation and Soil
various underground stations, tunnel escape shafts, tunnel Parameters
intervention and ventilation shafts need to be constructed. These
constructions involve deep excavation works. Deep excavation Proper planning and supervision of subsurface investigation (SI)
works pose great challenges to geotechnical engineers, are of utmost importance to the designer in order to produce a
particularly in complex ground conditions as well as in close safe and economical design for a deep excavation. A thorough
proximity of existing buildings in urban areas. Careful selection discussion of the planning of SI, field and laboratory tests may
of suitable retaining walls and support systems is crucial, taking refer to publications by Geotechnical Control Office of Hong
into consideration such as control of ground movement, lowering Kong (GEOGUIDE 2: Guide to Site Investigation, 2017) and
of the groundwater table, encroachment into neighbouring land, Clayton et al. (1995). Generally, the following soil parameters
etc. The design of temporary earth retaining stabilising system should be obtained from the SI:
(ERSS) requires meticulous analyses and evaluations of various a) Shear strength parameters of soil (φ’ and c’)
possible failure modes, such as overall stability, basal heave
failure, hydraulic failure, structural failure, etc. This paper b) Stiffness of soil (E’)
presents design principles and lessons learnt from temporary c) Permeability of soil (k)
earth retaining stabilising system (ERSS) together with vertical
rock excavation as well as three case studies including from Mass d) Groundwater level
Rapid Transport (MRT) projects, with various challenging
The above information is usually obtained from routine SI
geological formations found within the Klang Valley of
programmes except for soil stiffness which requires special
Malaysia; Limestone and Alluvium with high ground water table.
testing techniques and interpretation of results. The use of
pressure meter tests is recommended to obtain representative soil
2. DESIGN CONSIDERATIONS
stiffness values for design. Further discussion of the use of
Major design considerations for deep excavations revolve within appropriate soil stiffness values will be presented in the next
the following aspects: section.
a) Overall stability 2.1.1 Shear Strength Parameters
b) Basal heave failure The effective shear strength parameters of soil (φ’ and c’) are
commonly obtained from Isotropically Consolidated Undrained
c) Hydraulic failure Triaxial (CIU) Test with pore pressure measurements. If finite
d) Axial stability element method is used, understanding of the constitutive models
and numerical algorithms adopted in the finite element software
e) Finite element analysis is important in order to model the problem appropriately. The
f) Impact on buildings from ground movement following are recommended:
associated with excavation
a) Hardening soil model should be used to model
It is vital that preliminary analyses be carried out for various excavation problems, as the conventional Mohr-
options of walls and support systems to assess on the cost and Coulomb model is unable to model unload-reload
time of construction together with technical requirements on problems properly. Mohr-Coulomb model is based on
safety and its influence on adjacent structures before the selection elastic behaviour and is unable to model density and
of the final option to produce safe and economical design. shear hardening which renders it inaccurate for
deformation problems.
This paper updates some of the recent developments for the b) Modelling of excavation is recommended to be
design and construction of deep excavation works in Malaysia. performed in effective stresses and with effective
The flowchart for analysis and design of deep excavation works stiffness and strength parameters, if possible.
is updated from Tan et al. (2016), as shown in Figure 1. The
figure shows the routine from the beginning such as subsurface
46
1. Project information study 2. Site location study

 Determine depth/area/ sequence of excavation.  Study the topography, hydrography,
 Understand project requirements/ specifications. historical usage of site location.
 Study site condition surrounding excavation
site and identify critical
3. Subsurface information study buildings/structures/ utilities.
 Geological study.  Determine the limit of work boundary.
 Determine subsoil profiles.
Insufficient SI
 Determine groundwater level. Propose additional
 Determine soil/rock design parameters. subsurface investigation (SI).
Additional SI
Sufficient SI
4. Selection of suitable earth retaining system and construction sequence

 Select wall type and support system.
 Service period (temporary or part of permanent structure).
 Economic, workability and availability.
 Determine effective construction sequence (clearly specify in drawing).
 Risk assessment for the proposed options.
5. Ultimate limit stage (ULS) analysis

 Moderately conservative soil/rock design parameters.
 Highest possible groundwater level during service period.
 Consider unplanned excavation (0.5m or 10% of the total retained height/ retained height below lowest
support).
 Surcharge load (construction/operational load or minimum 10kPa).
 Determine wall depth and support levels for following checks:
 Wall stability check
 Vertical stability check
 Basal heave stability check
 Hydraulic failures check
 Calculate wall bending moment, shear force and support load (ULS)
6. Serviceability limit stage (SLS) analysis 7. Worst case scenario

 Moderately conservative soil/rock design parameters. (WCS) analysis
 Highest possible groundwater level during service period.  One strut failure.
 Surcharge load (construction/operational).  Accidental impact load
 Follow wall depth and support levels in ULS analysis. on strut.
 Evaluate deformation of retaining wall and retained  Flooded condition.
ground.  Calculate wall bending
 Carry out damage assessment of surrounding buildings, moment, shear force
structures and utilities. and support load (WCS).
 Calculate wall deflection, bending moment, shear force and
support load (SLS).
8. Structural design
 Determine design value (ULS x 1.0 or SLS x 1.4 or WCS x 1.05 whichever is critical) for wall and supports.
 Structural serviceability check.
 Structural durability check.
9. Design check and review Review

 Independent check and review as quality control procedures. comments
10. Instrumentation & Monitoring

11. Groundwater control
 Planning for instrument quantity and location.
 Planning for groundwater control scheme.
 Select suitable instrument.
a. Provisional recharge wells.
 Specify monitoring frequency.
b. Provisional relief wells (for temporary
 Establish monitoring triggering system with
uplift)
contingency plan.
12. Drawings preparation 13. Drawings check and

 Details of earth retaining structures and supports system. review
 Construction sequence.
 Instrumentation & monitoring scheme.
 Ground improvement and groundwater control scheme. 14. Drawings Issue for
 Protection works and advance work. construction
Figure 1 Flowchart for analysis and design of deep excavation works modified after Tan et al. (2016)
c) If
47
information on effective strength parameters is not

available, undrained strength parameters (c = cu, φ = 0,
ψ = 0) with effective stiffness parameters can be used.
Proper understanding of the constitutive soil models is
essential.
2.1.2 Soil Permeability

Soil permeability (k) is important to ensure the drained or
undrained behaviour of soil is modelled correctly. In-situ tests are
recommended in order to account for the complex soil
stratigraphy at site which may not be accurately measured in a
laboratory. Either rising, falling or constant head tests can be
carried out in-situ. The values obtained should be compared to
published values as a check to ensure the values obtained are
reasonable for a given soil condition. Figure 2, reproduced from
BS8004: 1986, is useful as a simple check.
2.1.3 Soil Stiffness & Small Strain Stiffness

Figure 4 Characteristic stiffness–strain behaviour of soil with
Empirical correlations are commonly used for estimation of soil typical strain ranges for laboratory tests and structures (Atkinson,
stiffness. This is because routine laboratory tests give soil 2000)
stiffness parameters that are significantly less than the stiffness
values derived from back analysis of field measurements. This is
primarily due to disturbance to the soil samples and also testing
at strain levels which are larger than the range which is
appropriate for retaining walls. This is illustrated in Figure 4
(Atkinson, 2000) which shows the strain dependent
characteristics of soil stiffness, is non-linear. It can also be seen
that typical ranges of strain for conventional laboratory testing
are larger than the range for retaining walls.
Though various empirical correlations are available to determine

small-strain stiffness for design, it should be noted that empirical
correlations are highly dependent on factors such as local soil
conditions.
For practical purposes, small-strain stiffness is probably most
reliably obtained using geophysical techniques through field and
laboratory, in which shear wave velocity can be measured as Figure 5 Field and laboratory methods to evaluate shear wave
shown in Figure 5, after Sabatini et al. (2002). Cross hole seismic velocity (Sabatini et al., 2002)
test is preferred as it measures the average shear wave velocity
laterally along the soil layers between two boreholes while the The input parameters for the small-strain stiffness model in a
downhole seismic test yields a shear wave velocity vertically typical model are as follows:
across different soil layers within a single borehole. While it is
theoretically possible to back calculate the shear wave velocities a) G 0 – maximum small strain shear modulus
for different layers with downhole seismic tests, making it the b) γ 0.7 – denotes the shear strain, at which the shear
cheaper option, but in the authors’ opinion, the most direct form
modulus G is decayed to 70 percent of its initial value
of measurement is a better method.
G0
Figure 2: Permeability and drainage characteristics of soil (BS8004:1986)

The above two
48
parameters would be able to define the entire stiffness each may have their own advantages and disadvantages. For
degradation curve. The values of G 0 can be obtained from example, Hardening Soil model is better suited for non-linear
measurement of shear wave velocity from the following formulation of soil behaviour in general (both soft soils and
relationship: harder soil types) as compared to the Mohr-Coulomb model.
However, the Hardening soil model does not capture creep
G 0 = ρv s 2 (1) (secondary compression behaviour). On the other hand, the Soft
Soil Creep model is able to capture the creep or time dependent
where, ρ is mass density of soil and v s is shear wave velocity of deformations; but it is only suitable for soft soils and not for
soil. other soil types.
Proper understanding of the limitations for each soil models are
In addition to using shear wave measurement, the maximum
essential. The quality of a result from numerical analyses
small strain-stiffness can also be estimated using empirical
depends on the user’s understanding of both the problem and the
correlations. Table 1 presents the typical range for G 0 for several
tool, such as the finite element software packages as well as the
generic soil types. The maximum small-strain shear modulus can
soil models within these software packages. One should be aware
be correlated to the SPT N 60 value and to the CPT q c value as
of the following factors which may affect the results of numerical
follows (Sabatini et al., 2002):
analyses :
G 0 = 15,560 (N 60 )0.68 (2) a) Locations of the boundaries of the problem. The
G 0 = 1,634(q c )0.25(σ’ vo )0.375 (3) problem boundary should be located far enough away
such that there is no stress rotation near the boundary.
where, σ’ vo is effective vertical stress of soil. Corresponding For undrained analyses, the extent of the model
units are all in kPa. required will be greater.
b) Details of mesh. Finer mesh is preferred to coarser
It should be noted that empirical correlations G 0 should be
meshing, especially if high strain gradients are
treated with caution. While it may serve as an initial estimate, it
anticipated, or for failure analyses. Nonetheless, the
is highly recommended to evaluate G 0 from shear wave velocity
refinement of mesh should be carried out selectively at
obtained from relevant field or laboratory tests which represents
critical areas within the deformation mechanism to
the actual site condition.
avoid unnecessarily high computation time.
The shear strain at which the shear modulus G is decayed to c) Long, thin partitions will lead to calculation instability.
0.7G 0 for normally consolidated soils can be calculated form the As such, the layout of the model and mesh should
following equation (Benz, 2007): avoid these closely spaced partitions.
d) Stages of construction. As soils are non-linear, stress
0.385 dependent materials, thus, proper modelling of the soil
�0.7 = [2�′ (1 + �
��2∅′) + ��′(1 + �0 )�
��2∅′] (4) at various stages from the past to its construction stages
4�0
need to be carried out.
The values obtained above should also be checked against values
given by Stokoe et al. (2004), where the authors proposed a e) Modelling of interfaces. Improper modelling or use of
interface reduction factors may lead to dangerously
linear increase of γ 0.7 from γ 0.7 ≈ 1 x 10-4 for PI = 0 up to γ 0.7 ≈ 6
unsafe design.
x 10-4 for PI = 100. Where PI is the plasticity index of soil.
f) Use of suitable constitutive soil models to model
Range of typical G 0 values for various soil types are provided in different geotechnical problems.
Table 1. It can be observed that a large range exists between the
lower and upper bound values for each soil type. Vardanega & g) Sensitivity of various soil parameters. For different
Bolton (2011) have shown that the value of G 0 is dependent on constitutive soil models adopted in different finite
multiple factors including PI, overconsolidation ratio (OCR), element method software packages, different soil
mean effective stress at point of consideration. This reiterates the parameters may have different effects on analyses
point that empirical correlations solely to SPT’N is subjected to results.
limitations.
3. CASE STUDIES
Table 1 Typical values of maximum small-strain shear modulus
(Sabatini et al., 2002) Three case studies including from Mass Rapid Transport (MRT)
projects, with various challenging geological formations found
Soil Type Maximum small-strain within the Klang Valley of Malaysia; Limestone, Kenny Hill
shear modulus, G 0 (kPa) formation and Alluvium with high ground water table are
Soft clays 2,750 to 13,750 discussed in the following sections.
Firm clays 6,900 to 34,500
Silty clays 27,600 to 138,000 3.1 Case Study of Deep Excavation for KVMRT in Kuala
Dense sands 69,000 to 345,000 Lumpur Limestone
and gravels
The Klang Valley Mass Rapid Transit (KVMRT) from Sg. Buloh
to Kajang (SBK Line) is the first MRT project in Malaysia. The
project comprises of a total of 9.5km long twin tunnels from
2.2 Important Considerations in Numerical Analyses Semantan to Maluri with seven (7) underground stations and
associated structures such as portals, ventilation shafts, escape
The use of numerical analyses such as finite element method for shafts and crossovers to be constructed over the Klang Valley
simulating deep excavation is common and routinely used by and Kuala Lumpur city areas. Tun Razak Exchange (TRX)
engineers of different levels of experience and expertise. Various Station (known as Pasar Rakyat Station during design
constitutive soil models such as Mohr-Coulomb, Hardening Soil, development), Cochrane Station and Maluri Station are
Soft Soil Creep, etc. are commonly used in practice. The underground stations located in the city area with excavation
characteristics of each soil model varies from one another where
49
depths of 20m to 32m deep respectively, in limestone formation.

TRX Station (refer to Figure 6) is the deepest station with
maximum excavation depth of 45m below ground serving as one
of the underground interchange station for the KVMRT Line 2
from Sg. Buloh – Serdang – Putrajaya (SSP Line), which is
currently on-going.
Figure 8 Geological map of Kuala Lumpur superimposed with

KVMRT (SBK Line) tunnel alignment
The selection of retaining wall system for the deep excavation

has considered the workability and suitability of the subsoil and
rock conditions. Secant pile wall was selected as the earth
retaining wall supported by temporary ground anchors. The
advantages of the selected wall type are
(a) Water-tightness to prevent groundwater draw-down at
Figure 6 Excavation works at TRX station the retained side;
(b) The ability to vary the pile lengths to suit the irregular
limestone bedrock profiles;
The Kuala Lumpur Limestone is well known for its highly erratic (c) Primary pile serves as reference for reinforcement
karstic features. Due to the inherent karstic features of limestone determination based on more accurate bedrock profiles.
bedrock, the depth of the limestone bedrock is highly irregular. The hard/firm secant pile wall consists of primary
Other karstic features include variable weathering condition, (female) piles cast first with concrete strength class
cavities and slime zones, which poses great challenges for C16/20 without reinforcement and followed by
construction of deep excavation works. Figure 7 shows the secondary (male) pile with concrete strength class
typical features of limestone formation (after Neoh, 1998). C32/40 with reinforcement. Figure 9 shows typical
arrangement of the secant pile wall.
Figure 9 Typical arrangement of secant pile wall

The secant piles sizes used for this project are 880mm, 1000mm,
1180mm, and 1500mm. The secant pile was generally designed
Figure 7 Typical features of limestone formation (Neoh, 1998) with an overlap of 15-20% of pile diameter. The extent of
overlapping of the secant piles are governed by pile installation
Figure 8 shows the Geological Map of Kuala Lumpur (Ref: Sheet verticality, pile deviation and pile depth (Gaba et al., 2003). After
94 Kuala Lumpur 1976 and 1993, published by the Mineral and reviewing the piles as-built performance, the recommended
Geoscience Department, Malaysia) superimposed with the tunnel overlapping values of secant pile wall are shown in Table 2
alignment. Three underground stations namely the TRX, where overlapping of up to 34% were specified to ensure water-
Cochrane and Maluri are underlain by the Kuala Lumpur tightness of the wall.
Limestone. The analyses of the retaining wall were carried out through finite
element method. Wall displacement, bending moment and shear
The overburden subsoil above Kuala Lumpur limestone generally force were obtained from the analyses for structural design. A
comprises of loose silty sand to sand materials with SPT’N’ load factor of 1.4 for bending moment and shear force were
values of less than 4. Average unit weight and permeability of applied for pile reinforcement design. The quantity of
subsoil are 18 kN/m3 and 1x10-5 m/s respectively. Interpreted reinforcement ranges from 0.5% to 4% of pile cross-sectional
effective shear strength from consolidated undrained triaxial tests area depending on the analysis based on different rock head
(CIU) is c’= 1kPa and φ’= 29º. Bedrock profiles of limestone level. 20kPa construction surcharge and 0.5m unplanned
formation are highly variable which ranges from 3m to 30m excavation were considered in ultimate limit state design.
below ground. Cavities, pinnacles and valleys are detected during Serviceability limit state analyses were carried out to ensure the
subsurface investigation works. ground deformations caused by excavation will not exceed
acceptable threshold limits of existing buildings and structures.
50
Table 2 Overlapping length of secant pile wall scenario (e.g. an accident during lifting operations, etc.). It can be
appreciated that this depends greatly on the anticipated so-called
Pile Pile Length Pile Length Pile Length accidental load conditions. The above-mentioned accidental
Diameter <8m <15m <25m impact load was recommended by CIRIA C517 (1999) but site-
(mm) (mm) (mm) (mm) specific risks and hazards may dictate different loads.
880 130 170 -
1000 150 200 340 Table 3 Partial load factors
1180 170 230 360
Load Case EL DL LL TL IL
1500 225 260 380
Working Condition 1.5 1.4 1.6 1.2 NA
Accidental Impact 1.05 1.05 0.5 NA 1.05
All secant piles were founded on competent bedrock with
One-strut Failure 1.05 1.05 0.5 NA NA
minimum rock socket length of 1.5m to 4.0m. The termination
Note:
criteria for rock socket are based on coring in competent bedrock
EL – Earth pressure and groundwater
with verification of point load index strength, I s(50) > 4 MPa
DL – Dead load
(equivalent to average UCS of 44 MPa). Should I s(50) falls below
LL – Live load
4 MPa, an equivalent rock socket length can be calculated to
TL – Temperature effect
compensate for the lower rock strength.
IL – Accidental impact load
Given that the rock head level in limestone formation can be NA – Not applicable
erratic with very steep inclined rock surface, one of the
construction issues that typical arises are on the length of rock
If the strutting design was found to be inadequate for the
socket. To prevent any disputes, the authors recommend the
considered impact load, then the strut is expected to fail resulting
following criteria where it must all be met before it is considered
in a one strut failure condition. The spirit of the one strut failure
as rock for claiming purposes:
as stated in CIRIA C760 (2017) refers to the potential loss of
(a) Change of tools from that of soil to rock support from a single prop rather than the loss of an entire level
coring/excavation tools. of props, unless it is considered as a physical possibility. The
latter is typically simulated in a plane-strain 2D analyses where a
(b) Rock head level is taken as a flat rock surface (i.e. not single prop level is deactivated instead of reducing the stiffness
inclined) that is verified at multiple points by means of of the props; accounting for the larger span between struts.
Kelly bar, etc.
It needs to be recognised that a one-strut failure is in fact a 3D
(c) I s(50) should be > 1.0 MPa problem. For an intermediate strut level, a single strut failure
It is important to ensure that the retaining wall is socketed into would cause the stresses to be distributed to the closest supports;
competent bedrock as the vertical rock excavation is just 1.25m adjacent struts to its sides and to struts above and below it. The
away from the retaining wall alignment. Support system will be redistribution of load is governed by the vertical and horizontal
installed in stages until reaching the bedrock level. A row of tie- stiffness of the wall, and to a lesser extent, the walers. This is
back rock bolts was installed above the bedrock level to enhance because the pressure acting on the wall must first be taken by the
wall toe stability. Toe stability check was carried out with some walls before it is transferred to the waler. A weaker wall could
modification which replaces passive resistance by tie-back force fail even if the walers do not.
to achieve minimum safety factor of 1.2. In addition, vertical Stiffness of the wall in each direction depends on the
stability was checked with resultant vertical load from ground reinforcement design and the spacing of supports. Since most
anchor pre-stress against the rock socket length. walls are constructed vertically with the main reinforcement in
Excavation was carried out in stages facilitated by installing the vertical direction coupled with generally smaller vertical
temporary ground anchors. U-turn ground anchors were used as spacings between levels in comparison to the horizontal spacings,
local authority requires removal of ground anchors after vertical stiffness of the wall would typically be dominant. Hence
construction. The anchor consists of a few pairs of strands with most of the load would be transferred vertically; to the struts
different unit lengths. Adopted strand diameter is 15.24mm with above and below.
U-turn radius of 47.5mm. Proof tests were carried out prior to the If a 3D analysis is not carried out, a conservative approach to
working anchor installation for design verification. Based on the cater for this is to ensure full distribution of loads vertically. This
proof test results, the recommended reduction factor due to is done by removing the strut completely in a 2D analyses. As
bending of strand at U-turn point is 0.65. Working loads of long as the wall stresses are still within limits for the
anchor ranged from 212kN to 1060kN with 2 to 10 strands. reinforcements (i.e. FOS 1.05), the resulting strut forces would
Typical designed pre-stress load is 60-80% of the working load then be conservative. In contrast, the force distribution would be
capacity. Generally, the anchor will be locked off at 110% of less representative if the stiffness of the struts (intermediate strut
designed pre-stress load. All anchors are subjected to acceptance layer) is reduced to account for a larger span for one strut failure.
test up to 125% of working load before lock-off. It is important
to clearly define in construction drawing the anchor working The only exception for this is with regards to the top-level strut
load, pre-stress load and lock-off load to prevent where it’s failure during the initial stages (i.e. when only the
misunderstanding and confusion during construction works. single strut is installed) results in a cantilever condition.
Designing for the wall to survive a cantilever condition with
The design criteria considered in strutting design are earth relatively lower risks at these shallow excavation depths would
pressures and groundwater, material dead load, 1.5 kN/m live be less feasible.
load, eccentric load, temperature effect (changes of 10°C),
accidental impact load (50kN in vertical direction; 10kN in To comply for a single strut failure condition, propping stiffness
horizontal direction), and one-strut failure. Recommended partial is reduced and waler effective length is increased accordingly.
load factors for strutting design are shown in Table 3. This is in-line with the concept proposed by CIRIA C760 (2017).
Ultimately it would result heavier design for the walers at this top
The requirement to consider accidental and one strut failure load level as it must be sufficiently stiff to redistribute the loadings
case is often a subject to debate. Nonetheless, the general idea is horizontally as it remains as the only propped level during this
to provide for a robust design which can prevent catastrophic or initial stage.
progressive failure in the event of a severe but transient load case
51
The loss of a strut single strut would effectively increase the grouting may be required after reviewing the grout intake from
loadings by 50% for the immediate adjacent struts. This would primary grouting. Rock fissure grouting is also adopted for base
be mostly covered by the partial factors of about 1.5 for the grouting at larger grout hole spacing. If any cavities are detected
working condition. Even if the strut was designed to have 100% during drilling or grouting works, compaction grouting with
efficiency in the basic load case, it would only need marginal cement mortar will be used as cavity treatment. It should be
increase its design to comply with the lower 1.05 partial factor noted that grouting techniques rely much on local experiences
for one-strut failure. and contractor workmanship. The recommended holding
pressures for fissure grouting in limestone in this case study are
However, the same cannot be said for walers. This is because the shown in Table 4.
resulting bending moment of a waler is related to the square of its
effective length. This essentially means that the effective span of Table 4 Holding pressure for fissure grouting
the waler doubles during a single strut failure, causing the
bending moment to increase by 400% as opposed to the 50% Depth (m) Holding pressure
increase in load for struts. The resulting waler design to comply (Bar)
with single strut failure would thus be significantly more 0 to 10 2 to 4
conservative. 10 to 20 6 to 8
20 to 30 10 to 12
In general situations, it can be inferred that a one-strut failure 30 to 40 14 to 16
load case would usually be more critical than the accidental load 40 to 50 18 to 20
case. Thus, if it is considered, accidental impact condition is no > 50 >22
longer relevant. The authors opine that if the adopted accidental Note: Termination criteria shall be satisfied with flow rate less
impact loadings are representative of a credible worst-case than 2 litres per minute or grout volume reaches 10m3 for every
scenario, then the risk of a single strut failure is as low as grouting zone in 5m depth.
reasonably practical (ALARP), where designing for one-strut
failure is not required. The Maluri station is located underneath one of the major public
roads in town. Besides the challenging limestone bedrock, major
The above discussion excludes the scenario where specific risks utilities diversion (e.g. high voltage electrical cables) and traffic
dictates very high accidental impact loads where it becomes less diversion were carried out in stages and carefully planned for
feasible to design the strutting to withstand it. In these cases, a installation of the secant pile wall. Temporary road decking with
one-strut failure check and design would be more appropriate. deck posts were installed concurrently with secant pile wall
Nonetheless for the KVMRT project, it was a client’s installation. Approximately 300m long and 21m wide road
requirement to check for both load conditions. decking covered up the top of the station and crossover area
It is essential to note that groundwater control is essential and it where excavation works beneath are executed concurrently.
is one of the important criteria to be considered in excavation One of the construction difficulties was pile installation under
works. Groundwater drawdown may lead to excessive ground existing electrical transmission lines with safe allowable working
settlement and occurrences of sinkholes surrounding the head room of only 13m. A modified low headroom machine was
excavation. Potential risk of excessive groundwater ingress into used for secant pile installation. In this condition, limit of drilling
excavation pit shall be evaluated especially in limestone size to small diameter is required to fulfil the capacity of the
formation. Natural features of solution channel with cavities and modified machine. Deck posts installation required high capacity
highly fractured limestone connected to excavation pit may cause rig with deep rock drilling which is beyond the typical machine
disastrous flooding inside the excavation pit. Therefore, grouting capacity. Deck posts are supported by four micropiles in a group
in limestone was carried out as a risk mitigation measure for to support the underneath existing electrical transmission line.
groundwater control. Schematic of the excavation works is Figure 11 shows the base slab casting at Maluri station with live
shown in Figure 10. Note that rock slope strengthening indicated traffic on top of the temporary road decking while Figure 12
in Figure 10 is provisional only. The actual locations and extent shows the excavation works with strutting support at the Maluri
of rock slope strengthening are determined after geological Portal. Vibrations from close proximity rock blasting (< 3m clear
mapping works and kinematic analyses. distance) from the deck posts were successfully controlled below
the conservative 25mm/s limit adopted for reinforced concrete
frame structures.
Figure 10 Schematic of excavation works

Grouting works are mainly carried out in limestone to reduce the
rate of groundwater inflow into excavation and reduce pathways
of water flow into excavation area. Rock fissure grouting was
carried out along the perimeter of excavation area to form curtain
grouting up to 10m below final excavation level. Fissure grouting Figure 11 Maluri Station base slab casting with live traffic on top
involves a single packer in ascending or descending stages in of the temporary road decking
order to inject grout suspension into existing pathways, fissures,
cavities and discontinuities within the rock formation. Additional
52
struts. Figure 14 shows the schematic of the temporary sheet pile

wall of the original design. The plan area involved was
approximately 20,000m2. The general depth of excavation was
7m, while at locations of lift pits the excavation was about 10m.
By the time 40% of the area was excavated to 7m below ground
level, the water level in the excavation pit has increased to about
4m below ground level. Subsequently, a stop work order was
issued when settlement and damage to the shophouses were
reported. Investigations were carried out to identify the causes of
the issues.
Original Design
Figure 12 Maluri Portal with strutting support

The above case study has shown the challenges of deep
excavation in limestone formation in general, focusing on
conventional linear or rectangular excavations. For smaller
excavations such as intervention, TBM launching or escape
shafts, a circular shaft will be much more economical since it
transfers the lateral forces through hoop stresses. This allows for
a more economical design with smaller diameter and a lighter
steel reinforcement. A circular shaft would also allow for an
unobstructed excavation area which translates to faster overall
construction progress. It should be highlighted that in these cases,
the overlapping of piles may be larger than the typical 15-20% in
order to maintain effective shaft thickness after considering
Figure 14 Schematic of the temporary sheet pile wall of the
deviations from setting out and verticality tolerances. As a
original design
contingency, provisions of ring beams should be included for
redundancy or when deviation exceeds the allowable limit. During the investigation, 26 number of boreholes were drilled
Further information on similar local experiences can be made to outside of the development area with the objectives of mapping
Tan et al. (2016) and Tan & Chow (2008). the subsoil profile, groundwater profile and soil properties within
the site. Piezometers were installed in most of the boreholes to
monitor the water profile. Settlement profiles along the lines
3.2 Case Study of Deep Excavation in Alluvium with High were also measured.
Groundwater Table
The results of the original ground investigation (see Figure 13)
The following is an interesting case study where a deep indicated the present of sand layers within the marine clay
excavation works has caused damages to shophouses within especially near the toe of the sheet piles, with a layer of sand of
300m away from the edge of the excavation works. The deep about 7m thick. The surface water profile during the investigation
excavation works were for the construction of a shopping mall as shown in Figure 15, indicated that the groundwater on the
with two levels of basement car park located in Penang, retained side has dropped significantly, especially near the
Malaysia. The site is situated within Quaternary deposits of excavation. This indicates seepage through the sheet pile wall.
marine clays, and is within the old river which has been
reclaimed. Figure 13 shows the soil profile within the site, where
interbedded layers silty sand can be seen in the marine clays.
Figure 15 Groundwater profile after excavation

Every drop of a metre of groundwater would increase the
effective overburden pressure by 10kPa which is equivalent to
about half a metre of compacted earth fill. Thus, it causes
Figure 13 Soil profile within the site immediate settlements and in cases of clays, consolidation
settlements. In addition to that, as the drop of water level reduces
with distance away from the excavation, the increase in effective
The main activities near the vicinity of the affected shophouses overburden and hence settlements, also reduces respectively,
were the basement excavation and piling works. The piles along inducing differential settlements. Ultimately, it was the
the perimeter of the site had been driven and a temporary differential settlement or distortion that caused the damages to
cofferdam using 15m deep sheet pile wall supported by internal the shophouses.
53
Remedial works were carried by installing an additional row of Tan, Y.C. & Chow. C.M. (2008) “Design of Retaining Wall and
sheet piles of about 30m deep, penetrating into the relatively Support Systems for Deep Basement Construction – A
impermeable clayey soil underlying the sandy soils to serve as a Malaysian Experience” Seminar on Excavation and
hydraulic cut off in conjunction with recharging wells. These Retaining Walls by IEM-HKIE. Petaling Jaya, Malaysia.
have effectively restored the water table to its original level. Tan, Y. C., Chow, C. M., Koo, K.S., and Nazir, R. (2016)
Further basement excavation was allowed to be resumed after the "Challenges in Design and Construction of Deep
monitoring confirmed the effectiveness of the remedial measures. Excavation for KVMRT in Kuala Lumpur Limestone
Formation", Jurnal Teknologi (Sciences & Engineering)
4. LESSONS LEARNT October 78:8-5 (2016), pp97-107.
The lessons learnt based on the two case studies above are Twine, D & Roscoe, H. (1999) Temporary propping of deep
summarised below: excavation- guidance on design (CIRIA C517).
Construction Industry Research and Information
(a) Sufficient and detailed SI is required to identify Association (CIRIA), London.
representative soil parameters, particularly in limestone Sabatini, P.J., Bachus, R.C., Mayne, P.W., Schneider, J.A. &
formation where rock level is very erratic in nature. Zettler, T.E. (2002) Geotechnical Engineering Circular No.
(b) Rock probing is crucial for the identification of any 5 – Evaluation of Soil and Rock Properties. Federal
inclined rock levels, potential rock fissures and cavities Highway Administration (FHWA), U.S. Department of
which can be dealt with via grouting at an early stage. Transportation.
(c) Rock head level and socketing criteria should be stated Vardanega, P.J. & Bolton, M.D. (2011) Strength mobilisation in
clearly in construction drawings with prevent disputes. clays and silts. Canadian Geotechnical Journal Vol. 48, pp.
(d) Design of strutting should at least account for 1485-1503.
accidental load impacts if not one-strut failure, where
deemed necessary.
(e) A single strut failure would have significant cost
implications particularly on the waler designs due to its
increased effective span. Therefore, it is possible for a
single layer strut failure approach to yield cheaper
overall design without changing the waler span. This is
not applicable for first layer struts where its failure
would imply a cantilever wall condition.
(f) Deep excavation can be carried out simultaneously
with live traffic on top of the temporary road decking
above the excavation.
(g) Groundwater drawdown induced building settlement
and damages could be as far away as 300m from the
excavation, particularly with soft soils such as alluvium
with interbedded sand layers. Proper design of
hydraulic cutoff is needed to prevent excessive drop in
groundwater table.
6. ACKNOWLEDGEMENT
Many parts of the work in this paper, including the case studies
were carried out together with the following colleagues, Ir. Dr.
TAN Yean Chin, Ir. CHOW Chee Meng, Ir. KOO Kuan Seng
and some others from G&P Geotechnics Sdn Bhd.
7. REFERENCES
Atkinson, J.H. (2000) Non-linear soil stiffness in routine design
(40th Rankine Lecture). Geotechnique Vol. 50(5):487-508.
Benz, T. (2007) Small-Strain Stiffness of Soils and its Numerical
Consequences. Universität Stuttgart, Germany.
British Standards Institution. (1986) BS8004:1986 Code of
Practice for Foundations.
Clayton, C.R.I., Matthews, M.C. & Simons, N.E. (1995) Soil
Investigation (2nd Edition). Blackwell Science, Oxford.
Gaba, A.R., Simpson, B., Powrie, W. & Beadman, D.R. (2003)
Embedded retaining walls – guidance for economic design
(CIRIA C580). Construction Industry Research and
Information Association (CIRIA), London.
Gaba, A.R., Hardy, S., Doughty, L., Powrie, W. & Selemetas, D.
(2017) Guidance on embedded retaining wall design
(CIRIA C760). Construction Industry Research and
Information Association (CIRIA), London.
Geotechnical Engineering Office (2017) GEOGUIDE 2: Guide to
Site Investigation.
Neoh, C.A. (1998) “Design & construction of pile foundations in
limestone formation”. Journal – Institution of Engineers,
Malaysia. Vol. 59, No. 1, pp. 23 – 38.
54
Increasing geotechnical challenges in the design and construction planning of the

Third Phase Bangkok MRT Underground
N. Phienwej1, A. Asanprakit2, P. Kittiyodom2 and S.Timpong2

1
School of Engineering and Technology, Asian Institution of Technology, Pathumthani, Thailand
2
Geotechnical and Foundation Engineering Co. Ltd., Bangkok, Thailand
E-mail: noppadol@ait.ac.th
ABSTRACT: The 3rd phase MRT underground project of Bangkok that is being implemented has faced increased difficulties in design and
construction owing to the congested corridor along the alignment and the phenomenon of groundwater rebound in Bangkok subsoils in the
aftermath of the end of land subsidence era from deep well pumping. The rebound is more than 10 meters from the situation during the
initial phase construction and thus it necessitates careful soil investigation and design approach to prevent instability of groundwater ingress
and uplift in bore tunnelling and excavation of station boxes. The tight space along the MRT underground alignment resulted in the twin
bored tunnels being placed at very small clearance in a few locations. It creates high risk in the shield driving, particularly in the saturated
sand layer. The works also involve tunnel driving at close proximity or under a number of existing buildings or road viaducts. In such cases,
various schemes of underpinning are employed, and one of them is strengthening of founding soil layers by cement jet grouting. An
intensive instrumentation program according to the observational design method is employed to safe guard against occurrence of any
undesirable instability in the excavations and impacts to the third party properties.
Keywords: MRT Underground, shield tunneling, deep excavation, underpinning, groundwater rebound, the observational method
1. INTRODUCTION 2. BANGKOK MRT SYSTEM

The 1st phase MRT underground project, i.e. Blue Line (Initial) was The MRT system of Bangkok, when fully implemented, is shown in
constructed in 1999-2003. It consisted of 18 stations and 22 km of Figure 1. As it has been planned, the underground portion
twin bored tunnels running along inner city roads. The tunnel was constitutes about 60 km in combined length, of which about one half
successfully bored with 4 EPB shields and the station box was have been completed. Besides the underground works of the
smoothly constructed using concrete diaphragm wall with the top Orange Line East that are being constructed, the Orange Line West
down excavation method. For the 2nd phase project, Blue Line and Purple Line South projects, which are to start the
Extension, the works began in 2011 and completed 2017. The 5.4- implementation by the end of the year, have the remaining
km underground section consisting of 4 stations and 4.8-km-long underground portion of the Bangkok MRT system. Works of the
twin bored tunnels were placed along narrow roads in the old city three underground MRT projects that are being or to be soon carried
quarter and crossed the main river of the city. In this phase, the out are challenging because of the very tight work spaces and the
shield tunnelling and station excavation experienced incidents of risen groundwater head in the subsoils.
severe difficulties associated with the impact of groundwater
rebound in the subsoils of Bangkok. The rebound was the
consequence of banning deep well pumping to solve land
subsidence problem in Bangkok area. At a section of shield
tunnelling, problematic groundwater ingress occurred through
segmented lining just installed while driving through a mixed face
of clay and sand layers; and in one of the station box excavation a
hydraulic uplift of the excavation base occurred that resulted in
flooding and significant complication in finishing the excavation
and casting the concrete base slab.
Currently, the 3rd phase underground project, Orange Line East, is
under construction. The underground section consists of 12.1 km of
twin bored tunnels and 10 stations placed along two of the most
traffic congested roads of the city where there exist elevated road
and expressway viaducts along the alignment. The piled foundation
of the viaduct complicates the alignment design of the MRT
underground tunnel and positioning the station boxes that result in
necessity in close proximity placement of the twin bored tunnels to
each other and to neighbouring structures at numerous locations. In
addition, the soil investigation reveals that the subsoil stratigraphy
consists of sand layer of larger thickness and higher rebounded
Figure 1 Bangkok MRT System
groundwater levels than that in the earlier phase in other zones of
the city. The design approaches of the underground works of the
3. SUBSOILS AND GROUNDWATER
project are outlined below. In addition, presentation is made on the
experiences in the design and difficulties in dealing with the
Bangkok is situated at the head of the Gulf of Thailand on the
situation of risen groundwater pressure in the sand layers in the
southern part of the low lying Chao Phraya plain. The vast flat plain
tunnelling and deep excavations just made in the 2nd phase – Blue
is carpeted with a thick soft marine clay layer which overlies a very
Line Extension.
thick series of alluvial soils of alternating layers of stiff to hard clay
and dense to very dense sands. In the upper 50 m zone within
which most foundation and excavation works are placed, the subsoil
profile is relatively uniform throughout the city area, which is an
55
ideal condition for tunnelling. However, the soft soil condition Variation of Piezometric Pressure Profile with
necessitates the use of shield method for tunnel construction. The Elevation
(Standpipe Piezometer)
shallow zone consists of a 12-15 m thick layer of soft to medium
stiff clay (Bangkok clay) that is followed by a layer of stiff to hard Short Term Design Piezometric Profile
clay and a layer of sand (the First Stiff Clay and the First Sand). 100 Long Term Design Piezometric Profile
Below them, alternating layers of stiff to hard clay and dense to very Existing Tender Orange Line (2005)
dense sand exist to a great depth. The sand layers are aquifers of Additional Tender Orange Line (2015)
Bangkok groundwater system. The subsoil stratigraphy along the

90
Post Tender Orange Line (2017)
Orange Line East that is under construction is shown in Figure 2.

80
Contract E1 Contract E2
70
Elevation, (m asl)
60
50
40
30
Figure 2 Soil profile along E1&E2 Contracts of Orange Line East

20
0 20 40 60 80
Piezometric Pressure (m)
It is well known that Bangkok had suffered land subsidence Figure 1 - Variation of Piezometric Profile with Depth
problem owing to excessive deep well pumping from the sand Figure 3 Piezometric pressure in the subsoils at different times
aquifers for water supply. The decline in groundwater head in
shallow aquifer layers was as much as 25-35 m from the ground
surface during the peak of the crisis in early 1980s, when the
subsidence occurred at a rate as high as 120 mm/year. After the
surface water supply was made adequate in most of the city area
around the turn of the century, the situation improved as deep well
pumping diminished. Since then the rebound of groundwater started
to develop and the subsidence ceased in most city areas. The
rebound continues till present and the accumulative amount is about
10-12 m during the last 15 years. At present the piezometric water
head in the 1st and 2nd sand layers of Bangkok where the MRT
underground works are involved with is about 13-14 m depth below
the ground surface that is 10-12 m higher than the condition when
the 1st Phase MRT Underground line was constructed (Figure 3). It
has been risen at a rate of approximately 0.8-1.0 m per year (Figure
4). Because the groundwater law that has been in effect since 2004
totally bans deep well pumping in the entire area of Bangkok and Figure 4 Changes in piezometric heads in Bangkok sand layers
surrounding suburbs, it is foreseen that the ground water heads in
the sand layers might rebound to the hydrostatic level in the near 4. INCIDENTS IN PAST MRT UNDERGROUND WORKS
future.
In the past the subsidence situation created complication of long Problematic incidents related to the impact of groundwater rebound
term settlement and differential settlement in the design of MRT in the underground works of the past MRT project are presented.
underground structures. On the other hand, the condition of the The recently completed underground structures of the Blue Line
lowered groundwater head in the sand layers resulted in ease in deep Extension were placed in the stiff clay and sand layers as shown in
excavations for station construction and shield tunnelling. Incidents Figure 5. The deepest bored tunnel section is at 37 m depth at tunnel
of hydraulic instability were rare in past deep excavations and invert at a location near the river crossing.
tunnelling. However, at present the situation has changed since the
groundwater head started to show significant rebound about 15
years ago. Difficult and problematic incidents were experienced in
a number of recent deep excavations of building construction as well
as in the works of the Blue Line Extension Project. In addition, the
rise of groundwater head was linked to incidents of difficulties in
EPB shield tunnelling in the city including a severe incident in the
2nd phase MRT Underground project that led to a six-month delay of
the stall and repair of the TBM.
Figure 5 Soil Profile along Blue Line Extension Underground
56
of the soil with the side of the diaphragm wall. The U-shaped
Facing the situation of higher water pressure in the 1st and 2nd grouted zone as shown in Figure 7 was adopted. The grouting work
sand layers than that indicated by early investigation, the detailed was launched underneath the ground surface from a platform at the
design of two of the three station boxes that were placed at large mid-height of the station height. The scheme was proven to be an
depths needed to be modified from those planned in the pre-tendered effective method to prevent the hydraulic uplift failure of the
design. This is to avoid problems of base instability during excavation base and it allowed a normal construction process of the
excavation from hydraulic uplift of the stiff clay at the excavation concrete base slab casting under a dry condition.
base. The excavation procedure adopted for the third station
remained the normal approach as the designer considered it was still
adequate. Unfortunately, after a portion of the base area was
excavated and concrete casting, the hydraulic uplift occurred in the
next stage of the base excavation that resulted in flooding and
difficulty in control of water ingress in the remaining stages of the
excavation. The work suffered almost 20 months delay in the
construction to fight with the persistent problem of groundwater
ingress despite extensive remedial grouting exercise. A brief outline
on the situations of hydraulic instability involved with the base
excavation design and construction of these stations is given in the
following sections.
4.1 Difficulty and problem in station excavations
The first station excavation of BLE that needed modification in the

design was Station BS12 which was 225 m long, 25.8 m wide and
32 m deep. The diaphragm wall was 1.2 m thick and 44.5 m deep.
The cross section is shown in Figure 6. The detailed soil profile
along the station excavation was derived from a comprehensive
subsurface borehole investigation made after the award of the
contract. A piezometric pressure was observed at approximately
15.0 m below the ground surface. Figure 7 U-shaped base grouted zone of BS12
The second station was Station BS13 which was 181 m long, 23.4
m wide and 30 m deep. The diaphragm wall was 1.2 m thick and 40
m deep. Cross section and subsoil is shown Figure 9. The
+101.5
piezometric head was observed at approximately 16.0 m below the
ground surface.
Soft Clay
GWL+86.3 +88.0
st
1 Stiff Clay
+77.5
1st Sand
+65.0
2nd Stiff Clay +63.0
2nd Sand
Figure 6 Cross section of BS12
Preliminary design analysis indicated that excavation to the base

would experience hydraulic uplift failure owing to the high
piezometric head in the 2nd sand layer underlying the thin 2nd stiff
clay layer below the base of the excavation. Factor of safety against
the failure considering the weight balance equilibrium was only 0.59
while the Outline Design Specification specified a minimum factor
of 1.03. Thus the potential basal hydraulic uplift failure was the key
factor in the design and construction and it necessitated an auxiliary Figure 9 Cross section of BS13
measures. Among three possible measures, i.e. extending tips of the
diaphragm wall to the 3rd stiff clay layer, the dewatering by deep Similar to the condition of BS12 excavation, potential base
well pumping and base grouting, the third method was adopted for instability against hydraulic uplift was faced. However, the situation
reason of the most technical viable and cost effective one. The entire was not as critical. Factor of safety against the failure considering
area of the excavation base was grouted. Two methods of grouting the weight balance equilibriumn was marginally at 0.96. In addition,
were adopted, i.e. chemical grouting by Tube-A-Manchette (TAM) the 2nd stiff clay layer at BS13 was not as thin as in the case of
and cement jet grouting. TAM grouting was chosen for the main BS12, (i.e. 6 m). Therefore, a different measures was adopted. To
part of station box excluding the north and south ends. Both minimize time and cost impacts on the construction work, the
methods were to create a watertight barrier in the 2nd sand to thicken normal approach without implementation of auxiliary ground
an impervious plug below the excavation base to counteract the improvement measures was considered and finally adopted. The
hydraulic uplift pressure as well as to improve side shear resistance fundamental of the method is that if the base excavation is made in
57
stages of a small sized excavation zone bounded by high soil BLE project that the EPB shield tunnelling had to be made through
surcharge + cast concrete slab around the perimeter of the the sand layer as deep as 37 m below the ground surface with the
excavation, the stability of the excavated base against the hydraulic piezometric head 10 m higher than that existed during the time of
uplift will be significantly assisted by the side shear resistance of the the first MRT shield tunnelling, an incident of serious water ingress
soil along the perimeter as compared with that of large excavation occurred in August 2013 during the early drive of an EPB shield
area. The Outline Design Specification stipulated at that in case through the interface between the 1st stiff clay and the 1st sand layer.
shear strength of base soil was considered, a minimum factor of The consequence was flooding and severe damage to the TBM that
safety was required at 1.20. This requirement could be satisfied for led to an interruption of the work for over 7 months. The incident
the soil and groundwater condition below the excavation base of the was an uncontrolled water ingress into the tunnel in the tail area of
station. The accurate determination of the density of the soil layers the EPB shield machine. Flooding and excessive movements of
and the spatial variation of the thickness of the stiff clay layer were segmental rings in vicinity of the shield were experienced. It was
a must in the adaptation of this method. The staged excavation reported the ingress initially occurred through a 20 mm wide gap
sequence adopted for BS13 excavation is shown in Figure 10. The formed at the circumferential joint of the key segment placed near
work was successfully completed as planned. the invert of the ring. The groundwater ingress from the 1st Sand
layer shortly led to piping of fine sand into the tunnel, and
subsequently excessive movements and distortion of several
segmental rings behind the shield. During the incident various
remedial grouting measures were urgently carried out to control the
ingress and avoid excessive distortion of the concrete segmental
rings that might ultimately result in a tunnel collapse.
It was obvious that the incident was caused by difficulty of the
EPB shield in controlling stability of the excavated tunnel face while
the tunnel was being excavated in the mixed face stiff clay and sand
layers. At that position the groundwater head in the sand layer was
17 m. The sand was in the lower part of the tunnel and as the
tunnel was progressively advanced, the portion of sand in face area
progressively increased as the TBM drive pitched down. This
condition was one of the most difficult tunnel grounds to deal with
in soft ground tunnelling. It was apparently that with high ground
water pressure in the sand layer in the excavated tunnel face, the
shield machine experienced difficulty in maintaining adequate face
Fig. 10 Sequence of staged base excavation of BS13 pressures. Thus over excavation would significantly occur. After a
careful examination on the TBM operation data it was believed that
Successful base excavation of the BLE Underground station was the triggering factor of the incident was an excessive displacement
not the case at BS11 Station which was the deepest excavation of the key segment of the segmental ring resulted from the over
among all MRT stations in Bangkok. The base of the excavation excavation of the mixed face soil by the TBM under the high
was at 32 m below the ground surface. Prior to the start of the groundwater condition.
excavation it was judged by the designer that the subsoil and
piezometric head in the sand layers data indicated that the base 5. ORANGE LINE EAST
excavation could be with adequate factor of safety against hydraulic
uplift. Unfortunately, after the excavation was made to the lower The 3rd Phase MRT Underground work that is under construction at
level of the bracing of the D-wall and base excavation and concrete this stage is a part of the Orange Line East project (Cultural Center
base slab casting was progressed from one end of the station box to to Minburi) that is implemented as design-build contracts. The
about 15 percent of the length of station, significant water ingress underground portion constitutes 12.1 km length out of the total 21.1
occurred that seriously affected the work. The whole station was km length of the line. There are 10 underground stations along the
flooded by 2-m-depth of water within 12 hours. Extensive underground section that the works are administered in 3 contracts
dewatering pump system was installed and a comprehensive (E1-E3). The underground line is mainly constructed by bored
program of chemical and cement grouting subsequently employed. tunneling using 3 EPB shields except for a short cut-and-cover
The amount of ingress was as high as 1400 cubic meter/hour at one startup section. Owing to the tight space along the corridor of the
stage. Groundwater ingress repeatedly occurred in the subsequent alignment of the line that traverse along two most congested roads
stages of excavation for concrete casting that altogether delayed the of the city that also have foundation of elevated roadway along the
work by 20 months. The investigation pointed that there might be a right or way, the MRT underground twin tunnels are mostly placed
condition of lateral non-continuity of the 2nd stiff clay below the in the vertically stacked configurations and in many places they
excavation base that in the initial design was considered sufficiently need to be placed at very close proximity to each other, particularly
thick and covering the entire area of the excavation base. This when leaving or entering the station boxes (Figures 2 and 11). The
incident of severe and prolonged groundwater ingress also led to tunnels also pass very near to or cross foundations of existing
subsidence problem of the surrounding ground and buildings that buildings or other structures, some of which need underpinning
necessitated extensive stabilizing grouting works throughout the works. The close proximity shield tunneling works for the deeper
period. section of alignment in the sand layer of high groundwater pressure
for the shallow section in soft clay layer necessitate special
4.2 Water ingress in EPB shield tunnelling considerations in the design and construction planning to avert
instability problems during construction and to ensure stability of
Prior to the tunnelling in the Blue Line Extension project, bored the completed structures in long term. Because of the adoption of
tunnelling by means of EPB shields and concrete segmented ring the vertically stacked configuration of the running tunnels most of
support had been commonly used with great success including those the underground stations are excavated at large depth. And with the
in the Blue Line Initial System project, in which the tunnelling was very extraordinary thickness of the 1st sand layer in the subsoils in
made deep in the sand layer at depth as much as 25 m below the the considerable part of the alignment, careful considerations are
ground surface. There had been no incidents of groundwater needed for the design and planning of the station excavation to avert
problem in EPB shield driving in Bangkok sands. However, in the potential basal instability problems as earlier experienced in the
previous project.
58
grouting (Figure 12). In addition, an extensive instrumentation

C&C Tunnel Bored Tunnel Bored Tunnel program is employed to closely monitor the response of the viaducts
and ground movements at all stages of the station excavation.
1400 m (approx.) 1400 m (approx.) 1900 m (approx.)
OR14
OR15
Tunnel Profile-Contract E1
Figure 11 Vertically stacked alignment of tunnels in Contract E1
5.1 Close proximity tunnelling
The design of the bored tunnelling and concrete segment lining has
to deal with the situation of very close proximity placement of the
twin tunnels in various places. The clear distance between the
tunnels is as close as 1.6 m which is only 25% of the tunnel
diameter (e.g. Figure 12). For such conditions where the tunnels are
to be excavated in the 1st sand layer, special measures are provided
to avert potential instability problems to the lining of the firstly
excavated tunnel from the TBM weight and any unexpected
excessive over-excavation during the excavation of the second
tunnel. The protective measures generally consist of ground
improvement by jet cement grouting. Where the ground space does
not permit the launching of ground improvement, other measures are
used. They include installation of steel ring bracing inside the
concrete segmental lining of the first driven tunnel, the careful
driving of the shield, and the close monitoring of the movements of
the lining and ground.
1.6 m
Figure 13 Examples of close proximity tunnel to existing piled

foundation and underpinning by jet cement grouting
Figure 12 Example of close proximity shield tunnelling
Close proximity tunneling of the works also need to deal with the
situation of potential impacts to the piled foundation of the existing
structures and vice versa the impact of the load to the stability of the
tunnel lining. Extensive numerical analyses are employed to check
the situation of all cases in details and necessary preventive and
corrective measures are provided in the design. In some cases, the
excavation may be made without the need for auxiliary measures,
but in some cases, ground improvement by jet cement grouting is
adopted (Figure 13). In the worst case, underpinning with pile
replacement needed adopted.
Besides the cases of close proximity tunneling, there is a case of
the station box excavation placed very close to the piers of the
elevated expressway at a clear distance between the diaphragm wall
and the foundation piles of the pier as small as 2.7 m. In this case
the impact assessment with the aid of rigorous numerical analysis
(PLAXIS) suggests implementation of the mitigation measure to Figure 14 Pre-installed struts for station box excavation next to
control the lateral movement of the viaduct pier and induced flexural expressway viaducts
stress in the piles within the allowable limits by using preinstalled
struts for the station excavation at two levels by means of jet cement
59
projects. publication in the SEAGS & AGSSEA Journal. We look

forward to receiving your paper contribution.
5.2 Underpinning work at crossing with existing Depot Access
A difficult and challenging underpinning work of the project is at
the crossing of the cut-and-cover tunnel construction with the
existing cut-cover-tunnels of the 1st Phase Blue Line Initial to the
depot yard. The existing tunnels are at shallow depth and have piled
mat foundation. Because the existing depot structure permits only 6
mm of movement of the rail structure, a careful consideration in the
planning and design of the underpinning works in the limited access
space is made. Methods of underpinning initially considered
included the use of pipe roof jacking technique, jet cement grouting
over the entire area, etc. However, due to the presence of many
piles underneath the existing depot structure, the pipe jacking was
not used. The underpinning scheme needs to employ combination of
methods including barrette pile installation, soil improvement by jet
grouting of large diameter columns and soldier pile wall excavation
underneath the existing structures, etc. The section of the crossing
is shown in Figure 15.
Figure 15 Underpining work at Depot Access crossing
5.3 Measures to deal with potential hydraulic uplift in

excavation of station boxes
After thorough assessment on the alternative measures to deal with
potential hydraulic uplift and water ingress problems at the base of
the excavation of various deep station boxes of the project, it has
been decided to adopt the option of positive cut-off of the
diaphragm walls by extending the tips deeper below the base of the Figure 16 Examples of designed diaphragm wall tips for uplift
excavation to the 3rd stiff clay layer. Although this measures calls prevention at two station boxes
for a much larger depth of the diaphragm wall than the base grout
plug alternative, it is judged that the ease and time duration of 7. REFERENCES
construction are more favorable. Figure 16 show examples of the
soil conditions and design of the diaphragm wall tip for uplift Ch. Karnchang Public Co. Ltd. (2013), Assessment of Grout Plug
prevention. Remedial Works at Sanam Chai Station (BS12), The MRT
Blue Line Extension Project: Contract 2.
6. CONCLUSION Ch. Karnchang Public Co. Ltd. (2013), Assessment of Staged
The MRT Underground works in Bangkok have faced increasing Excavation Works at Itsaraphap Station (BS13), The MRT
difficulties and challenges owing to the situation of rebound of Blue Line Extension Project: Contract 2.
groundwater level in the subsoils and the works in the remaining Ou, Chang Yu. (2006), Deep Excavation, 1st edition, London :
projects are situated in congested narrow roads of the city where Taylor & Francis
there are numerous obstructions along the corridor. Therefore, the
design and construction planning need to be carefully made to avert
undesirable instability incidents in both deep station excavation and
shield tunneling similar to that had recently occurred in the previous
60
Mega Reclamation Projects: Challenges and Lessons Learned in Soil Improvement

Works and Acceptance Tests
K. Yee1 and S. Varaksin2
1
Association of Geotechnical Societies in South East Asia (AGSSEA), Kuala Lumpur, Malaysia
2
Apageo, Paris, France
E-mail: kenny.yeeks@gmail.com1 / s.varaksin@apageo.com2
ABSTRACT: Singapore is the world’s second busiest port with an average of 60 vessels calling daily. As a testament to Singapore’s
commitment to maintain its position as a global maritime nation, the government is going full steam ahead with the multi-billion dollars
mega-port terminal development at Tuas. The Tuas Terminal will be developed in four phases over 30 years. Phase 1 reclamation of 294
hectares began in 2016 while the works contract for Phase 2 reclamation of 387 hectares was recently awarded in February 2018. When this
entire development is completed by 2040, it is expected to be the largest container terminal in the world. For the land-scarce small city-state,
creating new land by reclaiming it from the sea has played an important role in the country’s maritime industry. Its first reclamation project
for terminal development was in 1967. This was followed by a string of reclamation projects including Tanjong Pagar Terminal (80 hectares),
Keppel Terminal (105 hectares), Brani Terminal (84 hectares) and Pasir Panjang Terminal (456 hectares). While it continues to develop and
expand its maritime industry, it faces serious challenges to land reclamation due to limited availability of sand, among others. As sand is not
readily available in large quantity, it calls for innovative solutions to overcome material constraint. Prior to Tuas, in another mega terminal
development which was completed in 2015, the use of alternative reclamation fill materials was initiated to replace sand. Dredged materials
from deepening of basins and nearby fairway as well as excavated earth from land construction projects were reused as reclamation fill
materials. The use of these alternative materials, which would otherwise be disposed of, managed to save 45% in sand volume and saved
SGD470 million construction cost. Also, it reduced the need for disposal grounds for the dredged and excavated materials. Needless to say,
the innovations first deployed in this development are now being implemented to a greater degree in the upcoming Tuas Terminal. The
concept and application of the innovations used in this earlier terminal development are presented as a case study in this paper. This paper
describes some of the soil improvement techniques and acceptance tests used and it highlights the challenges and lessons learned.
Keywords: Land reclamation, Vibro compaction, Dynamic compaction, Dynamic replacement, Pressuremeter test, Cone penetration test.
1. INTRODUCTION Southeast Asia. By 1982, it achieved 1 million TEUs in a year for

the first time, and Singapore became the world’s busiest port by
When Singapore gained independence in 1965, it was a low-income shipping tonnage. In 1990, it crossed the 5 million TEUs mark and
country with limited natural resources that lacked basic became the world’s largest container port for the first time. Further
infrastructure. A few decades later, the picture could not be more expansion led to the opening of Keppel Terminal in 1991 and Brani
different. Today, the small city-state is the world’s second busiest Terminal in 1992. By 1994, Singapore achieved 10 million TEUs.
port after Shanghai. It is no surprise that Shanghai is the number 1 With the container handling volume growing exponentially
busiest port given that China has a massive population of 1.3 billion worldwide, shipping lines have been increasing the size of their
and it has the largest economy in the world – when measured in ships, doubling in size every eight years over the last two and a half
terms of gross domestic product (GDP) based on purchasing-power- decades. Recognising the need to ramp up capacity through larger
parity (PPP). According to IMF’s World Economic Outlook, April cranes, deeper berths and longer quay lengths to serve mega ships
2017, China’s GDP of US$23.19 trillion exceeds the U.S. GDP of that stretch over 400m in length and hold 18,000 standard-sized
US$19.42 trillion. The economic strength of China is supported by containers or more, the Pasir Panjang Terminal was opened in 2000.
the fact that seven of the world’s top ten busiest ports are in China. In 2005, Singapore surpassed 20 million TEUs in a year. Soon after,
Maritime is steeped in the lifeblood of Singapore, with its the government announced further expansion of Pasir Panjang
illustrious relationship tracing back to its humble beginning as a Terminal. The Pasir Panjang Terminal Phases 3 and 4 were opened
fabled and historic port-of-call and entrepôt. After playing a critical in 2015. In 2017, Singapore has handled over 33 million TEUs.
role in the country’s development from a third world to a first world Today, Singapore has once again betting big on future trends
nation within one generation, the maritime sector continues to be a that make or break economies – to develop a mega-size
significant engine of growth for the country’s economy, making up technologically advanced container terminal. It is one of the
7% of the country’s GDP and employing more than 170,000 people. country’s most ambitious projects yet, aimed at almost doubling the
Singapore accounts for one-seventh of the world’s container 33 million TEUs in 2017. The multi-billion dollars Tuas Terminal is
shipments and it is one of the world’s largest refrigerated container a major milestone in Singapore’s next generation container terminal
ports with connections to 600 ports in over 120 countries. It trans- development. It is set to be a showcase for the latest port technology
ships a fifth of the world’s shipping containers. In 2017, the port and systems. This development will be carried out in four phases
handled over 33.67 million TEUs (Twenty-foot Equivalent Units) over 30 years. Works for Phase 1 reclamation of 294 hectares started
only losing to Shanghai’s 40.23 million TEUs but exceeds in 2016. The works contract for Phase 2 reclamation of 387 hectares
Shenzhen, China (25.21 million TEUs); Ningbo-Zhoushan, China was recently awarded in early 2018. When the entire development is
(24.61 million TEUs); Busan, South Korea (21.40 million TEUs); completed by 2040, it is expected to be the largest container
Hong Kong (20.76 million TEUs); Guangzhou, China (20.37 terminal in the world with a total handling capacity of up to 65
million TEUs); Qingdao, China (18.26 million TEUs); Dubai, UAE million TEUs annually – more than the combined 50 million TEUs
(15.44 million TEUs); and Tianjin, China (15.21 million TEUs) – capacity of the current terminals at Tanjong Pagar, Keppel, Brani
thus, making up the world’s top ten busiest ports in 2017. and Pasir Panjang. When Tuas Terminal is completed, all the current
Singapore’s achievements did not happen by chance but based terminals will relocate and merge at Tuas. This consolidation of
on forward-looking policy. To maintain its position and to stay container port activities will not only result in increased efficiency
ahead of the competition, the government has carried out a series of in port operations due to the elimination of inter-terminal haulage
bold moves by building ahead of demand. Starting with the opening but also free up 925 hectares of waterfront land for development.
of the Tanjong Pagar Terminal in 1972, it made waves not just as Figure 1 shows an artist’s impression of the Tuas Terminal.
the country’s first container terminal, but also as the first in
61
saved SGD470 million construction cost. Also, it reduced the need

for disposal grounds for the dredged and excavated materials
Drawing upon the lessons and experiences gained from this
development, reusing alternative materials is expected to reduce the
quantity of sand required by up to 60% and will result in a potential
cost savings of some SGD1 billion in the upcoming Tuas Terminal.
Needless to say, the innovations first deployed in this development
are now being implemented to a greater degree in Tuas.
The concept and application of the innovative solutions used in
this mega terminal development are presented as a case study in this
paper. This paper describes the reclamation activities and some of
the soil improvement techniques and associated acceptance tests
used. It highlights the challenges and lessons learned.
Figure 1 Artist’s impression of the Tuas Terminal
Yet challenges loom: While Singapore continues to develop and

expand its maritime industry, the small city-state faces yet another
challenge – finding land for developments. Land is Singapore’s
most cherished resource and its dearest ambition. Past terminal
developments were all built on reclaimed land; including Tanjung
Pagar Terminal on 80 hectares of reclaimed land, Keppel Terminal
on 105 hectares of reclaimed land, Brani Terminal on 84 hectares of
reclaimed land and Pasir Panjang Terminal on 456 hectares of
reclaimed land (Figure 2). The upcoming Tuas Terminal Phases 1
and 2 also involve the reclamation of 294 and 387 hectares of land
respectively. Since the first reclamation works began in 1822,
Singapore’s land area has expanded by a whopping 25% from
58,150 to 71,910 hectares in 2017. In its 2013 Land Use Plan, the Figure 3 Transportation of dredged materials by barges (top) and
government has set a goal to reach a total land size of 76,600 dredged clay lumps (bottom)
hectares by 2030 for its planned developments – meaning Singapore
still has some growing to do. 2. CASE STUDY: PROJECT DESCRIPTION
Singapore’s Prime Minister Lee Hsien Loong opened this SGD3.5
billion mega terminal development in June 2015. Figure 4 shows the
aerial photograph of the new terminal built on 198 hectares of
reclaimed land. It consists of 15 deepwater container berths on
6,000m of quay length and up to 18m draft, designed to serve mega
container ships with capacities larger than 10,000 TEU. It has
container yard equipped with intelligent planning and operation
systems, as well as unmanned rail-mounted gantry cranes. With the
full operation of this terminal, it boosts Singapore’s annual handling
capacity by 40% to 50 million TEUs.
Figure 2 Reclaimed land for terminal developments
But there are serious challenges to land reclamation for the small
city-state – the availability of sand for reclamation. The problem is
Singapore has used so much sand that it has run out of its own, and
need to import sand from elsewhere to meet its massive land
reclamation needs. Skyrocketing price of imported sand just added
fuel to the fire. The situation got worst when Malaysia banned the
export of sand to Singapore in 1997; followed by Indonesia in 2007
and Vietnam in 2009. In 2016, Cambodia enacted a ban against
exporting sand to Singapore. Hence, aggressive land reclamation Figure 4 Aerial view of the new terminal development
programme using sand is no longer tenable. It calls for innovative
solutions to overcome material constraint. Construction works began in 2007. It showcased several
In 2015, a mega terminal development was completed using innovative engineering feats that were recognised for engineering
several innovative engineering solutions including the use of achievements thought to have the greatest economic, infrastructural
alternative reclamation fill materials to replace sand. Dredged and societal impact. In 2016, it was voted by the public as being one
materials such as marine clay from deepening of basins and nearby of Singapore’s Top 50 Engineering Achievements – making it a
fairway as well as excavated earth from land construction projects construction model for the upcoming Tuas Terminal.
(e.g. roads, MRT, etc.) were reused as reclamation fill materials Some of the innovative engineering solutions used during
(Figure 3). The use of these alternative materials, which would reclamation works include the use of large concrete blocks (caissons)
otherwise be disposed of, managed to save 45% in sand volume and to build the seawall and wharf structure instead of traditional piling
62
methods and the use of alternative reclamation fill materials instead

of sand. The construction works was completed in 2015 and the
terminals are in full operation since 2017. Figure 5 shows the key
reclamation activities that happened simultaneously and they are:
Figure 6 Extent of soil improvement works (PVD – vertical drains;

VC – vibro compaction; and DC – dynamic compaction)
(adapted from The Straits Times on 28-10-2016)
X
X
Figure 7 Usage of different reclamation fill materials
3. OFFSHORE SOIL IMPROVEMENT BELOW WATER

3.1 Caisson Seawalls
The reclamation works was designed based on the use of large
reinforced concrete caissons to form seawalls. The caisson seawalls
serve to retain the reclamation fill materials forming the land and to
function as the quay for the container vessels when the berths are
put into operation. The construction involved precasting the caissons
Figure 5 Reclamation activities (source: The Straits Times on 28-10-
using specialised machine and transporting the completed caissons
2016)
into docking position using a floating dock. As the caissons were of
standard sizes, an on-site fabrication yard was set up to allow round-
a) Activity A: Laying foundation for the caisson seawall.
the-clock concreting work. Under a factory-like environment, the
Soft clay was dredged from the seabed until the hard strata to
productivity and works quality of the caissons were increased. One
form sandkey trench. Sand was filled into the trench and
hundred and fifty caissons were constructed; each measuring 21 to
compacted. After sand compaction, rocks were laid and
32m in height and weighs between 8,800 tonnes to 12,000 tonnes.
subsequently compacted to form rock mound.
These caissons were among the largest in the world. Figure 8 shows
b) Activity B: Caisson seawall preparation and placement.
some of the completed caissons at the fabrication yard.
The construction of caisson units consisted of (i) casting of base
slab; (ii) casting of caisson wall; (iii) curing of completed
caisson unit; and (iv) towing of caisson unit out to sea and ready
for installation.
c) Activity C: Soil improvement works.
Vertical drains were used for the improvement of dredged
materials and excavated earth. They were installed to depths up
to 55m. Some 24 million meters of vertical drains were installed.
After installation of vertical drains, sand surcharging was carried
out to a fill height of 18m. After removal of surcharge fill, deep
compaction was carried out to compact the sand fill using vibro
compaction and dynamic compaction. Thereafter, surface roller
compaction was carried out on the good earth layer.
d) Activity D: Reclamation filling works.
Alternative fill materials (i.e. dredged materials from sandkey
trenches, deepening of fairways and basins; and excavated earth
materials from land-based construction projects such as roads, Figure 8 Completed caissons at the fabrication yard
MRT, etc.) were used as reclamation fill materials in addition to
sand. Cemented-mixed soil was also used for the construction of 3.2 Caisson Seawall Foundation: Construction, Design and
the “geo-bund” temporary edge structure. The “geo-bund” Performance Requirements
provided the necessary safety benefits for the adjacent dredging
work and construction of caisson seawall. The caisson seawalls were installed on rock (rubble) mound founded
Taking a cross-section X-X from Figure 5, Figures 6 and 7 show the on sandkey trench (Figure 9). The foundation works consisted of (i)
extent of the soil improvement works and the different reclamation dredging of sandkey trench to reach very stiff strata; (ii) filling
fill materials used respectively. Using these alternative fill materials sandkey trench with sand followed by deep compaction; (iii)
reduced the quantity of sand required by 45% and resulted in cost carrying out hydrographic survey after sand trimming and levelling;
savings of SGD470 million. and (iv) placing rocks to form rock mound followed by compaction.
63
After completion of the foundation works, acceptance tests were The results revealed a layer of “softened” sensitive soil of about
carried out. Upon achieving the acceptance criteria, the caissons 1.5m thick at the base of sandkey trench and dredged seabed. To
were installed, backfilled with sand and compacted. supplement the CPT tests, pressuremeter (PMT) tests were carried
out on jack-up pontoon at 15 different locations. The tests were
carried out in the sand fill and clay layer down to 33m or more
below water. A summary of the PMT results is shown in Table 1.
The EP/PL ratios of the disturbed materials (i.e. disturbed sand /
softened clay) were generally less than 7 indicating an altered or
disturbed state while the undisturbed stiff clay shows a ratio greater
than 16 indicating an over-consolidated state (D.60.AN, 1975).
Table 2 shows the estimated strength values – internal friction
angle () and undrained shear strength (cu) of the materials based on
PMT results of Table 1. The internal angle of friction () of 30o to
32o for disturbed sand was less than the design requirement of 35o
while undrained shear strength (cu) of 50 to 80 kN/m2 for softened
clay was also less than the design requirement of 250 kN/m2. Only
DREDGED LINE
the undisturbed stiff clay layer registered undrained shear strength
(cu) greater than the design requirement. These values of ɸ and cu
Figure 9 Cross-section of caisson on sandkey trench were estimated from the equations below (Amar et al, 1991):
PL *
The performance requirement for the foundation of the caisson cu  (1)
seawalls was a residual settlement after construction not exceeding
5.5
40mm. To achieve this requirement, the construction of the sandkey  φ o  24o 
 
trench called for complete removal of the underlying soft to firm *  4 
clay until very stiff clay strata (undrained shear strength (cu) of not PL  2.5* 2  (2)
less than 250 kN/m2) and total replacement with clean sand. The where PL* is the net limit pressure (in bars) which is equal to the
sand fill with fine content (soil particles finer than 75m) less than difference between the measured limit pressure (PL) and the
10% was then compacted to achieve a relative density of minimum horizontal pressure at rest.
70%. The design requirement was an internal angle of friction (ɸ)
greater than 35o and the acceptance criteria based on cone Table 1 Results of PMT tests after vibro compaction of sandkey
penetration tests (CPT) were cone resistance (qc) of 8 MPa at the
surface (i.e. top finished level of sandkey trench) increasing to 22 EP EY PL
Material EP/PL
MPa at depth greater than 20m. (MPa) (MPa) (MPa)
Deep vibro compaction was carried out to compact the sand fill. Disturbed
2.6 – 3.3 7.8 – 9.9 0.70 – 0.72 3.8 – 4.4
Vibroflots with maximum centrifugal force of 35.4 tons and sand
amplitude of 26mm at the tip were used. The compaction was Softened
0.7 – 1.2 1.4 – 2.4 0.49 – 0.51 1.5 – 2.3
carried out with a 3m overburden sand fill placed above the top clay
finished level of the sandkey trench. The compaction spacing was Undisturbed
36 – 39 36 – 39 2.05 – 2.26 17 – 18
3.5m triangular grid with compaction duration of 40 seconds at stiff clay
every 0.5m vertical lift increments.
After completion of vibro compaction and sand trimming, rocks Table 2 Estimated  and cu after vibro compaction of sandkey
of maximum 300mm size were placed to form mound ranging from Material EP/PL ɸ (o ) cu (kN/m2)
1 to 4m thick. The rock pieces were compacted to achieve design
requirement of an internal angle of friction (ɸ) greater than 45o. Disturbed sand 3.8 – 4.4 30 – 32
Softened clay 1.5 – 2.3 50 – 80
3.3 Unexpected Soil Softening Undisturbed stiff clay 17 – 18 > 250
Prior to the removal of the overburden sand fill after vibro Note: EP is the pressuremeter modulus; EY is the stiffness modulus;
compaction and before placing of the rock mound, CPT tests were PL is the limit pressure; ɸ is the internal angle of friction and cu is
carried out as routine acceptance tests. On a particular area at water the undrained shear strength.
depth of about 30m, non-compliance results (i.e. measured cone
resistances lower than acceptance values) were recorded. Figure 10 3.4 Remedial Solution
shows a set of non-compliance CPT results after compaction.
From the CPT results, the total thickness of disturbed sand and
softened clay layer ranged from 80 – 150cm. Based on the measured
“c-” values, a re-analysis was carried out for caisson seawall
settlement and stability. Figure 11 (left) shows the original design
assumptions (design requirements) compared to the measured field
values in Figure 11 (center). To meet the performance requirement,
it was necessary to reinstate (“improve”) the measured field strength
values back to the original design assumptions or equivalent. Figure
11 (right) shows the “target” revised design requirements.
A remedial solution consisted of a rock mat over reinforcing
rock columns was adopted. Rock columns were introduced to
Dredged Line improve the composite strength properties (c-of the softened clay
while the compacted rock mat was introduced to increase the
internal angle of friction () of the disturbed sand. It also served as a
load distribution layer to the reinforcing columns below. Dynamic
replacement was carried out to install the rock columns and dynamic
Figure 10 CPT results after vibro compaction of sandkey compaction was carried out to compact the rock mat.
64
D  n. W.H (3)
Fz 
F2  F1
D 2
 Z Z 
o
2
 F1 (4)
where D = depth of improvement / compaction (m)

W = weight of pounder (ton)
H = drop height (m)
n = constant (depending on soil type, degree of saturation)
Fz = improvement ratio at elevation z
F1 = maximum improvement ratio at surface
F2 = improvement ratio at depth of compaction
Z = elevation (m)
Figure 11 Comparison of design, average field and target values Zo = natural (original) ground level (m)
3.5 Principle of Dynamic Compaction / Dynamic Replacement The compaction is generally the highest below the pounder.
Maximum increase in density is at about one third of the depth of
Dynamic compaction is a soil improvement technique used for in- compaction from the surface. An increase of the penetration
situ densification (compaction) of granular soil (e.g. sand) by heavy resistance of 300 to 400% can be expected in clean sand and gravel.
impacts. It is done systematically in a pre-determined grid pattern. It Dynamic replacement provides an alternative to dynamic
consists of delivering high energy impacts at the ground surface by compaction when the saturated soil cannot be compacted due to low
repeatedly dropping 10 to 40-ton steel pounder from heights of 10 to permeability or excessive fine contents – generally greater than 10 –
40m using a crane (Figure 12). Deep craters up to 2m are formed as 15%. Dynamic replacement is an extension of dynamic compaction,
a result of the impacts. In loose sand, the heave around the craters is using similar plant and equipment (Figure 14). However, unlike
generally small. The craters are filled with sand after each phase. dynamic compaction which is a soil densification technique for
granular soil, dynamic replacement is a soil reinforcement technique
used mainly for cohesive soil (e.g. clay). Stiffer materials with
higher shear modulus are introduced into the soil mass as
reinforcing columns. The improvement of the cohesive soil mass is
derived from the structural aspect of the stiffer reinforcing columns
via a composite soil-column mass, interacting through friction and
adhesion. It increases the bearing capacity, reduces settlement and
improves stability. Backfilling materials for the columns can be any
free-draining, hard, durable, inert materials such as sand, stones or
even rock pieces (up to 300mm size). The volume of columns
usually represents about 15 – 25% of soil mass volume.
Dynamic replacement process starts out by producing a pilot
crater (“print”) with light pounding. The crater is then backfilled
with suitable materials that will lock together under subsequent
heavy pounding (Figure 14). This pounding process is repeated with
increasing compaction energy until a noticeable decrease in crater
Figure 12 Dynamic compaction rig and impact craters formation occurs. Since the column material is more permeable,
pore water pressure from the underlying and adjacent less permeable
The spacing between compaction points depends on the required cohesive soil will dissipate quickly. Hence, the columns being load
depth of compaction, grain size distribution, permeability and bearing columns (reinforcement) also serve as large vertical drains.
location of the ground water table. The initial spacing of the
compaction points roughly corresponds to the compaction depth. It
is necessary to use highest compaction energy, with the heaviest
pounder falling from maximum drop height for the early phases of
compaction in order to extend the compaction effect as deep as
possible. The compaction energy and the spacing of compaction can
then be reduced for subsequent phases thereby allowing adequate
compaction to be carried out at shallower depths (Figure 13).
Figure 14 Dynamic replacement rig and works procedure
Column diameter up to 2.5 to 3m is common for dynamic

replacement. The surface area of these columns is approximately 5
Figure 13 Different phases of dynamic compaction to 7 m2 with design bearing capacity up to 100 tons per column in
soft to medium stiff clay. Spacing of columns varies from 4.5 to 7m.
Empirical formulae for the estimation of depth of improvement While dynamic compaction (DC) and dynamic replacement
(D) and the degree of improvement relative to the depth (Fz) are (DR) technique share similar plant and equipment, there are
given below (Varaksin and Racinais, 2009): distinctive differences in their applications and works procedure:
65
a) DC creates a homogeneous soil compaction mass effect while Since dynamic compaction and dynamic replacement were
DR is a composite soil-column reinforcement effect (Figure 15); carried out below water, the conventional “solid” pounders shown in
b) DC uses the same soil material for compaction while DR uses Figure 16 were not suitable due to water buoyancy. A special
imported material “stiffer than the soil mass” for the columns; offshore pounder was designed and built for this purpose following
c) DC starts with maximum compaction energy to reach design a hydrodynamic impact study (Figure 17). The study was carried out
compaction depth and finished with reduced energy in the final to determine the impact speed below water; water resistance and
ironing phase while DR starts with low compaction energy to inertia; impact stress and durability of the pounder while providing
form the column “plug” and finished with increased energy to sufficient “punching” energy to create the rock columns and
compact the whole column material. “compaction” energy to compact the rock mat.
d) DC uses compaction pounder with larger surface area to avoid
“punching” into the soil while DR uses a punching pounder with
smaller surface area for more penetration effect (Figure 16).
Figure 15 Mass compaction of dynamic compaction (left) and

composite soil-column effect of dynamic replacement (right)
Figure 17 Hydrodynamic impact analysis of the offshore pounder
Based on the terminal velocity of 7 to 8 m/s obtained from the

hydrodynamic impact study, the optimized configuration and shape
of the offshore pounder is shown in Figure 18. It was a combined
design of a “DC-and-DR” pounder in a single unit with two
compaction “surfaces” – larger one for DC and smaller one for DR.
Figure 16 Punching DR pounder (left) and flat DC compaction

pounder (right)
3.6 Offshore Dynamic Compaction / Replacement Works

To improve the disturbed soil layers, a working platform of 1.8m
thick rock blanket was placed at the dredged line after removal of
sand in the sandkey trench. For ease of compaction below water,
granite rock pieces with 30% grading size between 150 to 200mm
and 70% grading size between 200 to 300mm well-graded were
used. Dynamic replacement (DR) process was first carried out on
top of the rock blanket to form rock columns 1.3m into the softened
clay layer. The rock columns were designed as 2m diameter
columns installed at a square grid of 4.5m corresponding to an area
replacement ratio of 15.5%. The compaction inside the rock
columns was carried out to achieve an internal angle of friction (ɸ)
of 45o. After completion of the DR process, the remaining rock
blanket was compacted by dynamic compaction (DC) process to
form the 1.3m thick compacted rock mat with an internal angle of
friction (ɸ) of 45o. Both DC and DR works were carried out about
30m below water. Figure 11 (right) shows the mechanical properties
to be achieved with this remedial solution. These targeted values
were used to re-analyse the caisson seawall settlement and stability.
3.7 Field Calibrations

Prior to actual production works, several field calibrations (also
known as field trials) were carried out to calibrate the field operation
parameters with the “target” revised design requirements. Field
performances were verified by field measurements of rock volume
used; surface elevations before and after DC/DR compaction; and
pressuremeter tests (PMT) before and after DC/DR compaction.
Cone penetration test (CPT) was not carried out due to the presence
of compacted rock and its grading sizes. Figure 18 Offshore DC-DR pounder
66
The weight of the offshore pounder was 37.5 tons with its • Based on echo sounding survey before and after DC/DR works,
smaller “punching” surface area of 1.72m x 1.72m (approximately 3 the difference in measured surface elevations was about 38cm.
m2) for dynamic replacement process and a larger “compaction” Considering 1m3 loose rock volume equivalents to 0.87m3
surface area of 2.6m by 2.6m (approximately 6.8 m2) for dynamic compacted volume, the net enforced settlement was about 33cm.
compaction process. The height of the pounder was fixed at 2.3m Hence, the calculated volume of a DR column was 6.7m3 and
determined from the required DR columns penetration. The ratio of the diameter of DR column was 2.5m.
solid surface area over open surface area was determined based on • Based on sea-divers’ “direct” field measurements, the diameter
minimum water resistance against net pounder weight. The pounder of the DR columns varied between 2.4 and 2.6m. These values
was made of special manganese steel plates to withstand the are in agreement with the diagonal length (= 2.43m) of a 1.72m
abrasion caused by the heavy impacts on the granite rocks. x 1.72m pounder – dimensions of the punching surface. Surface
Using this offshore pounder, field calibration tests were carried upheave of 10 – 20cm between the columns was recorded. The
out on an area of 22.5m by 22.5m (506 m2). Twenty-five test columns penetrated about 1.3m into the softened clay layer.
columns were installed at 4.5m square grid with 30 blows per From these observations, the calculated and measured diameters of
column. The drop height was 5m in water. Drop height greater than DR columns exceeded the target value of 2m. Hence, the field
5m was shown not to improve significantly the compaction as operation parameters were satisfactory for full production works.
terminal velocity of 7 – 8 m/s was reached. This was consistent with Figure 20 shows the probable soil profile before and after DC/DR
the results obtained from the hydrodynamic impact analysis. After works but before surface levelling. Due to larger columns, the area
the completion of the columns, the rock mat was compacted with 3 replacement ratio (ARR) had increased to 22% instead of 15.5%.
to 6 blows at an overlapping grid pattern as the final ironing phase.
Based on field measurements, the number of blows was plotted
against column penetration as shown in Figure 19. During “column
penetration” phase, the first 15 blows produced significant column
penetration with an average of 1.3m. After the 15 blows, additional
blows produced marginal penetration (< 10 cm). However, more
blows were needed to increase the density (compactness) of the
columns. During this “column compaction” phase, additional 13 to
15 blows were given until surface heaving around the columns was
recorded.
Figure 20 Probable profile before (top) and after DC/DR (bottom)
3.8 Execution of Full Production Works

Figure 19 Column penetrations vs. number of blows With the completion of the field calibration tests, full production
works was launched following the sequence below:
Pressuremeter tests (PMT) were carried out before DC/DR  Removal of sand from the sandkey trench.
compaction works (pre-DC/DR tests). After compaction works, - Production rate for removing and dumping of sand was about
post-DC/DR PMT tests were carried out inside the DR columns and 4,000 m3 per day.
in-between the DR columns. During the tests, visual observation on  Placement of 1.8m thick rock blanket as “working platform”.
the return of drilling fluid was recorded. When there was no return - Well-graded granite rocks of 30% grading size between 150
of drilling fluid, it confirmed that the PMT test was carried out to 200mm and 70% grading size between 200 to 300mm
within the confinement of the free-draining rocks (i.e. inside the DR delivered by flat-top barges with tugboats equipped with
columns). Otherwise, the PMT test was conducted in the impervious excavators and shovels.
clay layer (i.e. in-between the DR columns) as indicated by the - After completion of surface levelling and hydrographic
presence of returning drilling fluid due to undrained condition. survey, the rock pieces were placed using a tremie barge. The
The average pre-DC/DR pressuremeter modulus (EP) and limit rock pieces were dropped down the tremie pipe and free-fall
pressure (PL) for the uncompacted rock obtained from 9 different to the sea-bed. The depth of tremie pipe, tremie barge position
test locations were 2.4 MPa and 0.55 MPa respectively. The ratio of and the quantity of placed rocks were monitored and
EP/PL was 4.4 (i.e. < 6 suggesting a loose state) (D.60.AN, 1975). controlled to achieve the intended coverage.
The average post-DC/DR EP and PL values for the compacted rock - Soundings and interim hydrographic surveys were carried out
obtained from 14 different test locations were 9.7 MPa and 1.2 MPa regularly after rock placing.
respectively. The ratio of EP/PL was 8.1 (i.e. > 6 suggesting a - Production rate for rock placing was about 3,000 m3 per day.
compacted self-bearing condition) (D.60.AN, 1975). The average  Installation of DR rock columns and compaction of rock mat.
estimated pre-DC/DR internal angle of friction of the uncompacted - The locations of DR columns were pre-determined and logged
rock fill was about 42o while the average post-DC/DR internal angle on to the GPS system located inside the DC/DR base crane
of friction of the compacted rock fill was about 48o which exceeded cabin as part of the on-board computer system.
the target value of 45o. - The DR columns were installed in two phases. The area
The other physical measurements taken include: coverage was divided into blocks and the installation of
• Total volume of loose rock pieces placed over the field columns followed a zigzag line within each block.
calibration area of 506 m2 was about 900 m3. This is equivalent - The date, column position, initial and final rock elevations at
to 1.78 m3 per m2. each column locations were recorded by the on-board
67
computer system. Also, the operation parameters such as drop 0.8 – 1 MPa is adequate for self-bearing condition where settlement
height, drop speed, impact velocity and number of drops at will not occur under its own weight (D.60.AN, 1975).
each column location; the total compaction time and the net
“buoyancy” pounder weight were recorded.
- Production rate was about 350 m2 per day.
 Filling of sandkey trench with sand and subsequent compaction.
- The sand filling proceeded immediately after compaction of
the rock mat and acceptance tests.
- The sand layer was compacted by vibro compaction.
- Production rate was about 350 m2 per day.
 Acceptance tests
- Acceptance tests using PMT for compacted rock and CPT for
compacted sand were carried out. PMT tests were carried out
before and after DC/DR works. CPT tests were carried out
after completion of vibro compaction works.
- Visual inspections and field observations (including
underwater inspections), surveys, samplings and testing were
carried out at regular intervals.
The offshore DC/DR works was carried out using a 200-ton
Liebherr LB895 hydraulic crane, fixed onto a crane barge measuring
25m by 50m with 4-point anchoring system. The DC/DR rig was
equipped with GPS system and a quality control computer system
for compaction. Figure 21 shows the plant and equipment used for
the offshore DC/DR works. The works was completed in 2009. Figure 22 Operation of PMT test with STAF system
Figure 23 (a) PMT read-out unit; (b) PMT drilling rig; (c) jacked-up
pontoon; and (d) STAF drill bit
Figure 21 Offshore DC/DR rig on barge and offshore pounder
Table 3 Results of PMT tests after DC/DR works
3.9 Post DC/DR Tests Results and Discussions Material EP (MPa) EUR (MPa) PL (MPa)
Settlement and stability for the caisson seawalls were re-analysed Compacted rock 7.3 – 10.4 31 – 34 1.1 – 1.6
using the post-DC/DR strength parameters derived from PMT tests.
PMT tests were carried out using the STAF system. STAF – self- Table 4 Interpretation of PMT test results after DC/DR works
bored tube system, is a slotted tube technique with inside
Material EP/PL EUR/EP EY (Mpa) ɸ (o)
disintegrating tool and mud circulation. This method emplaced the
slotted casing tube while drilling to avoid borehole wall disturbance Compacted
6.7 – 8.4 3.5 – 4.2 29 – 42 47 – 49
(Figure 22). This was deemed to bring improvement to the operation rock
and the test results especially in non-homogeneous and over- Note: EP is the pressuremeter modulus; EUR is the unload-reload
consolidated soil. Figure 23 shows the PMT equipment deployed for pressuremeter modulus; EY is the stiffness modulus; PL is the limit
the offshore testing works. PMT tests were carried out to depths of pressure and  is the internal angle of friction.
up to 33m below water on a jacked-up pontoon. After DC/DR
works, PMT tests were carried out at 29 different locations which The internal angle of friction  (o) for the compacted rock was
included both static and cyclic tests. A typical cyclic PMT test estimated based on the net limit pressure PL* (bars) using the
results is shown in Figure 24. A summary of the results is given in relationship below (Yee and Varaksin, 2012):
Table 3. Table 4 shows the ratio of the unload-reload pressuremeter
 φ  40o
 

modulus (EUR) over the static pressuremeter modulus (EP). The  
EUR/EP value ranged from 3.5 to 4.2 which agreed well with the 
 7 

suggested value of 3.5 – 4 for compacted rock and PL greater than PL *  4 * 2.5  (5)
68
Equation (5) is an extension of Menard (1970) equation for 4.2 Potential Problems of Loose Sand
granular soil extending to  > 40o; presumably the indicative lower
D’Appolonia (1970) reported that loose sand is very unstable when
bound value for the internal angle of friction () for granite rock
subjected to even a modest shock and vibration. Such vibration
pieces of similar size which corresponds to measured PL values of
could emanate from pile installation or from seismic effects. It is
about 0.55 MPa (Varaksin, 2009).
prone to liquefaction. For small strain of the order of 10-5 to 10-3 the
The measured internal angle of friction () of the rock after minimum relative density to prevent liquefaction should be about
compaction was estimated to be between 47o and 49o with an 70% and that fine sand with a relative density less than 50% is
average value of 48.5o based on an average PL* of 1.2 MPa which subject to liquefaction during ground motions with acceleration in
satisfied the target requirement of 45o. excess of 0.1g.
Hydraulic sand filling by pipeline method or rainbow method
has large volume of water. The resulting fill structure is likely to be
loose and it will remain loose because of the capillary retention of
the sand which prevents the sand particles from rolling into a stable
and denser configuration. Typical estimated relative density of sand
fill above water level is 50% to 70% and below water level, it can be
as low as 30% to 50%.
While loose sand is not as compressible as soft clay, it continues
to settle under its own self-weight for many years after their initial
deposition. Creep is often estimated as 1% of the thickness of loose
sand fill per log cycle of time which is sufficiently large that it
cannot be ignored in the design of foundations especially for thick
fill as in this case study. Another inherent problem of loose sand is
sudden instability. Although it may be metastable and change state
readily, its very instability nature makes it possible to alter its loose
Figure 24 A typical cyclic PMT test results structure effectively. Shearing of the sand particles by vibration or
impact to form a denser and stable structure has been the most
3.10 Lessons Learned effective means for densifying loose sand.
Dynamic compaction and vibro compaction have found wide
Soil improvement on land above water is usually an easier task. acceptance, and numerous case histories have illustrated their
When it is carried out at great depth below water, the task becomes practical applications for densification of loose sand for higher
more difficult. It requires more rigorous analyses and more elaborate bearing capacity and lower compression. Dynamic compaction has
construction procedure with specialized plant and equipment. The been described above. Vibro compaction is briefly described below.
works procedure or methodology needs on-site adaptation to suit the
prevailing conditions. Quality control becomes more critical as often, 4.3 Principle of Vibro Compaction
it is beyond our normal “visibility” range.
Some of the challenges and lessons learned in this offshore Vibro compaction uses an electric or hydraulic powered vibroflot
DC/DR works include: suspended from a crane (Figure 25). The vibroflot consists of a
 Adaptation of coarser and heavier backfill materials (i.e. rock torpedo shaped horizontally vibrating probe that vibrates at a
instead of sand) for compaction under water due to water frequency of 30 to 50 Hz with amplitude between 8 mm and 48 mm.
buoyancy and “turbulence” effect during impacts and the By shearing of particles caused by the horizontal vibration, the loose
requirement to achieve a high internal angle of friction (); soil particles are re-arranged into a denser configuration.
 Adaptation of a “porous” and heavier DC/DR pounder for
underwater compaction due to water buoyancy and terminal
velocity of falling pounder and its limited drop height.
 Adaptation of appropriate quality assurance and quality control
tests e.g. using PMT test for rock, cyclic and static PMT test to
verify the compactness of rock pieces;
 Visual inspections and field observations above and below water
(using sea divers) during works are necessary especially for
offshore works to provide additional information to assess
“what-is-going-on” below water.
4. ONSHORE SOIL IMPROVEMENT ABOVE WATER

Figure 25 Vibroflot and different phases of vibro compaction
4.1 Reclaimed Land behind Caisson Seawall
For sand compaction, it is recommended to use low frequency
With the completion of the caisson seawall, sand was placed behind vibroflot with large amplitude to transfer the compaction energy
the seawall overlying the seabed. Further away from the seawall, generated by the vibroflot to the surrounding sand as efficiently as
dredged and excavated materials (alternative fill materials) were possible. This is achieved when the vibroflot is vibrating at 15 to 20
first placed on the seabed and followed by sand overlying it (Figure Hz in resonance with the surrounding sand. Resonance between the
7). The sand was obtained from marine borrow sources and vibroflot and the sand leads to amplification of the ground
transported to the reclamation site. The hydraulic placement of sand vibrations, as the vibroflot and sand move “in-phase” with little or
resulted in a loose state with CPT cone resistance (qc) as low as 3 – no relative displacements occurring. Water jet is used as a cutting
5 MPa. After sand filling, vertical drains were installed to accelerate medium to break up the sand mass during compaction. A water/air
the primary consolidation of the underlying cohesive materials mixture is better than water jet alone. However, the drawback of
followed by fill surcharge to reduce post construction settlement. using water/air jet is that the dissipation time may be longer and post
Deep compaction was carried out after the removal of surcharge fill compaction tests (e.g. CPT) may be influenced by air bubbles.
to densify the hydraulic sand fill to reduce creep and to mitigate Figure 26 shows a design chart with different vibroflot capacity (30
vibration-induced settlement and liquefaction. Surface roller – 235 HP units). With larger vibroflots, the spacing between
compaction was carried out for the upper layer of good earth. compaction points can be wider and productivity is improved.
69
raised to +3.1m CD (i.e. 0.45m above top finished level) to allow

for settlement compensation after compaction. The area of
compaction was approximately 320,766 m2 for vibro compaction
and 1,572,127 m2 for dynamic compaction. Hence, the total area for
deep compaction of hydraulic sand fill was close to 1.9 million m2.
Figure 26 Vibro compaction spacing chart
4.4 Onshore Vibro / Dynamic Compaction Works

Deep compaction was carried out behind the seawall using vibro
compaction and dynamic compaction to densify the hydraulic sand
fill. Figure 27 shows the extent of the deep compaction works.
Figure 28 Characteristics of vibro compaction and dynamic
compaction
Figure 27 Cross-section of deep compaction area
The selection of the deep compaction technique was based on

their inherent merits and constraints. Dynamic compaction has
higher productivity and hence, a lower operating cost compared with
vibro compaction. Hence, it is a financial advantage to use this
technique. However, due to higher energy impacts on the ground
surface dynamic compaction generates higher surface vibration than
vibro compaction. For the fear of possible damage to the caisson Figure 29 PPV measurements for dynamic compaction and vibro
seawall due to excessive ground vibrations, vibro compaction was compaction
carried out closer to the seawall. But, vibro compaction produces
lesser degree of compaction at shallower depth compared with 4.5 Performance Requirements
dynamic compaction due to lack of overburden pressure. Hence, an
overburden fill of 1.5m was placed above the top finished level so The performance criterion for deep compaction was relative density
that the upper layer can be sufficiently compacted using vibro (DR) of minimum 70%. The acceptance criteria based on cone
compaction. Figure 28 summarises the key characteristics of the two penetration test (CPT) was cone resistance, qc as shown in Table 5.
deep compaction techniques.
A field vibration monitoring program was carried out to measure Table 5 Acceptance cone resistance (qc) after compaction works
the ground vibrations in terms of the peak particle velocity (PPV). Description Min. qc
Different compaction energy per blow for dynamic compaction and At 1 m below surface 6 MPa
different capacity vibroflots for vibro compaction were used. For At 2 m below surface 9 MPa
dynamic compaction, an open cut trench of 1 – 1.5m deep was dug Between 2 to 3 m below surface 10 MPa
to cut-off the transmission of surface vibration to the seawall. The Between 3 to 4 m below surface 11 MPa
results are shown in Figure 29. A safe PPV of 4 mm/s for the Between 4 to 7 m below surface 13 MPa
seawall was adopted. A safe distance of 45m from the caisson
seawall was demarcated for vibro compaction. Beyond 45m 4.6 Field Calibrations
distance, dynamic compaction was carried out (Figure 27).
At the vibro compaction area, compaction commenced from For deep compaction works, preliminary compaction spacing and
elevation +4.5m CD to -2.0m CD i.e. 6.5m compaction. The working parameters are usually determined from past experiences
working platform was raised to +6.0m CD to allow for the 1.5m and empirical design charts. Prior to actual production works, these
overburden fill. At the dynamic compaction area, compaction parameters are established and confirmed following the completion
commenced at +3.1m CD with compaction measured from +2.65m of field calibrations (or “field trials”). The following parameters
CD to -2.0m CD i.e. 4.65m compaction. The working platform was were used in a calibration area of 30m by 30m (900 m2):
70
For calibration of dynamic compaction (DC) works: layer within the sand fill mass is not included in the net thickness.
- Working platform was at +3.1m CD (i.e. 0.45m above top Table 6 shows the measured enforced settlements.
finished level at +2.65m CD to allow for settlement).
- Spacing of compaction points was 6m square grid with two
phases of compaction.
- Phase 1 consisted of 8 blows with 300 ton.m energy per blow
(total energy per phase of 2,400 ton.m)
- Phase 2 consisted of 6 blows with 300 ton.m energy per blow
(total energy per phase of 1,800 ton.m)
- Total applied compaction energy of 117 ton.m per m2
- Treatment depth was 4.65m; measured from +2.65m CD to
-2.0m CD
For calibration of vibro compaction (VC) works:
- Working platform was at +6.0m CD (i.e. 1.5m above top
finished level at +4.5m CD as overburden fill).
- Spacing of compaction points was at an equivalent of 3.4m
triangular grid using an electric vibroflot of 130 kW; 30 Hz
frequency; 23mm amplitude with 30 ton centrifugal force. A
combination of compressed air (using a 12,000 l/min
compressor) and high pressure water jet (100 m3/h under 1.2
MPa pressure) was used as cutting medium. Figure 30 Pre and post CPT results for vibro compaction
- Compaction was carried out to reach amperage of 200 amps or
40 seconds per 50cm lift whichever comes first.
- Treatment depth was 6.5m; measured from +4.5m CD to -2.0m
CD
The above operation parameters were expected to meet the required
acceptance qc values.
4.7 Post Compaction Results

The post compaction results consisted of CPT results and enforced
settlement measurements.
4.7.1 CPT results
Typical pre and post-compaction CPT results for vibro compaction
and dynamic compaction are shown in Figures 30 and 31
respectively. Pre-compaction CPTs were carried out after the
removal of surcharge fill. Profiles of relative density of 30%, 50%
and 70% have been superimposed using the method of Baldi et al.
(1982). The relative densities below water level varied between 30% Figure 31 Pre and post CPT results for dynamic compaction
and 60%, confirming a loose state. Above water level, it was in a
denser state. The results also identified a layer of 0.5 to 1m thick Table 6 Measured enforced settlements after VC/DC compaction
dense sand at +3.5m CD i.e. about 1.5m above water level. This was
the working platform for vertical drains installation and construction Net thickness of Enforced Settlement
Methods
traffic to place surcharge fill – both had caused surface compaction. sand layer (m) (cm) (%)
The post-compaction CPTs were carried out 7 days after Dynamic compaction 3 – 5.1 25 – 40 7.8 – 8.3
compaction. Above water level where the sand was initially in a Vibro compaction 5 – 6.5 50 – 60 7.7 – 12
denser state, the compaction did not show much improvement. The
greatest improvement was below water level where the initial
4.8 Observations and Discussions
relative density was low ( 50%). However, it was noted that the
post-compaction qc values were at best only about 60% of the The unexpected large non-conformance of qc values after VC/DC
acceptance values – an unexpected large non-conformance! compaction was studied and presented below.
4.7.2 Enforced Settlements 4.8.1 Estimation of relative density after VC/DC compaction
During dynamic compaction works, subsidence craters up to 3.5m in Enforced settlement is a function of initial relative density, thickness
diameter were formed with penetration averaging 35cm per blow of sand layer and degree of improvement. From an assumed initial
using a 1.83m x 1.83m steel pounder weighing 15 tons falling from relative density, a final relative density can be estimated from the
20m height. Water was frequently observed inside the craters. enforced settlement. The limitation with this method is that the
During vibro compaction works, subsidence craters up to 2.5m in initial relative density has to be estimated, and the values for
diameter were formed around the shaft of the vibroflot. These minimum and maximum density may not be valid for the full
subsidence craters were formed as a result of the densification of thickness of the sand fill. Nevertheless, this method is a useful tool
sand. These craters were then backfilled with sand on the surface. to use and may provide an insight to the non-conformance.
Enforced settlement is the difference between the measured Undisturbed samples were collected after the removal of
surface elevations before and after compaction. It is expressed as a surcharge fill but prior to compaction. Laboratory tests were carried
percentage of the thickness of compactible sand layer. For dynamic out following BS 1377 Part 4 1990 (Methods of tests for soils for
compaction, the enforced settlement is computed from its working civil engineering purposes: compaction-related tests). The minimum
platform at +3.1m CD while for vibro compaction, it is computed dry density varied from 1.31 to 1.33 Mg/m3 with an average of 1.32
from +4.5m CD where the compaction commenced. The upper 1.5m Mg/m3 (12.9 kN/m3). The maximum dry density varied from 1.75 to
was not compacted as it was only an overburden fill. Compaction 1.83 Mg/m3 with an average of 1.80 Mg/m3 (17.6 kN/m3). The
stopped at -2.0m CD. Any non-compactible interbedded cohesive specific gravity was 2.65. Using Equation (6), the calculated
71
minimum void ratio varied from 0.45 to 0.51 with an average value For vibro compaction, the compaction grid of equivalent 3.4m
of 0.48. The calculated maximum void ratio varied from 0.99 to triangular spacing was reduced to 3.1m – a 17% reduction in area
1.02 with an average value of 1.01. The following equations were coverage per compaction point. Varying duration of compaction and
used to determine the final relative density: withdrawing criteria together with different combination of water/air
G s, γ w jetting pressures were carried out in the new field calibration.
e 1 (6) To account for “ageing” effect (i.e. increase of qc with time), CPT
γd
tests were carried out 7, 14 and 21 days after compaction.
γ d,max (γ d  γ d,min ) (7) All the above produced little improvement (< 10%) and did not
DR 
γ d (γ d,max  γ d,min ) meet the acceptance qc values. It was true that the operating
parameters were optimised. Inadequate compaction was not the
e max  e (8) cause for the non-conformance. The only probable reason left was
DR 
e max  e min down to the intrinsic property of the sand fill material itself.
VT (9) 4.8.3 Carbonate content
VS 
1 e
Sand with significant carbonate content usually has higher initial
where DR = relative density void ratios and lower dry densities as a result of its angularity, poor
e = void ratio grading and intra-particle porosity. The values of emin (0.48) and
Gs = specific gravity (= 2.65) emax (1.01); and d,min (12.9 kN/m3) and d,max (17.6 kN/m3) when
w = unit weight of water (= 9.8 kN/m3) compared with typical values of silica sand fit into this trend (Table
d = dry density 8). These initial void ratios are comparable with carbonate sand
d,max = maximum dry density (= 17.6 kN/m3) found in Hong Kong and Dubai where emin = 0.5 and emax = 0.95 are
d,min = minimum dry density (= 12.9 kN/m3) average values for carbonate content greater than 10% (Figure 32).
emax = maximum void ratio (= 1.01)
emin = minimum void ratio (= 0.48) Table 8 Typical void ratios and dry densities of sand (Das, 2008)
VS = volume of solid compared with hydraulic sand fill (of this case study)
VT = total volume
Void Ratio Dry Density (kN/m3)
Based on the assumption of 45% initial relative density and Soil Type
using the enforced settlements of Table 6, the calculated relative emin emax d,min d,max
densities after VC/DC compaction exceeded the 70% criterion Coarse sand 0.30 0.74 14.9 19.3
(Table 7). This contradicts the non-conformance described earlier Fine sand 0.40 0.85 14.1 18.5
and it could not possibly provide any reasonable explanation. Hyd. sand fill 0.48 1.01 12.9 17.6
Table 7 Calculated relative densities after VC/DC compaction Samples were taken for laboratory tests following BS 1377 Part 3
Enforced Relative Density 1990 (Methods of test for soils for civil engineering purposes:
Methods chemical and electro-chemical tests). Visual inspections revealed the
Settlement (%) (%)
Dynamic compaction 7.8 – 8.3 71.2 – 72.9 presence of calcitic shell fragments that were platy, flaky and
Vibro compaction 7.7 – 12 70.9 – 85.3 porous. The test results confirmed carbonate content ranging from
13.1% to 15.2%. Chang et al (2006) reported a carbonate content of
How reasonable is the assumption of 45% initial relative 8% to 16% in the direct-dumped sand and 4% to 12% in the
density? Radio-isotope (RI) cone penetrometer tests were carried out hydraulic filled sand used in Changi East reclamation. Figure 32
at an adjacent area. Relative densities below water were reported to shows carbonate sand found in Hong Kong, Changi (Singapore) and
be as low as 50%. Choa et al. (1979) reported relative density of Dubai. The test results agreed with the trend shown.
hydraulic sand fill ranging from 40% to 80% at the Changi Airport
reclamation. The lower values were obtained below water. Chu
(2011) reported minimum relative density of 45% to 50% for the
hydraulic sand fill at the Changi East reclamation. Figures 30 and 31
indicated relative density before compaction between 30% and 60%
below water when interpreted using the method of Baldi et al.
(1982). Hence, the assumption of 45% is not at all unreasonable.
emax=0.95
4.8.2 Increased compaction energy
Is the applied compaction energy adequate? Causes of an inadequate
compaction for vibro compaction works include larger spacing of emin=0.5
compaction points, insufficient water/air jetting pressure and quicker
withdrawal of vibroflot during compaction (inadequate compaction
time). For dynamic compaction works, it includes inadequate total
applied compaction energy per m2, compaction energy per blow and
insufficient rest period between phases of compaction.
To address these issues, additional field calibrations were carried
out. For dynamic compaction, a 3rd phase compaction was carried
out with additional 6 blows of 300 ton.m energy. The total applied
Figure 32 Void ratios of carbonate sand
compaction energy increased from 117 ton.m/m2 to 167 ton.m/m2 –
an increase of 43%. The heave and penetration tests showed the 4.8.4 Performance of carbonate sand
penetration per blow was about 10 – 20cm for the 1st two blows and
reduced rapidly to less than 10cm thereafter accompanied by The performance of carbonate sand is strongly influenced by the
substantial surface heaving around the craters. This suggested that crushability and angularity of the fragile carbonate grains, high
the energy saturation point has been reached and further increase in initial void ratio of the fill and cementation between particles.
the number of blows will only cause volumetric displacement Carbonate sand is susceptible to particle crushing and hence, it is
(surface heaving) with no further densification in the soil mass. more compressible than silica sand. For an equivalent relative
72
density, the qc values measured in carbonate sand will be less than in ratio (e) relative to its maximum (emax) and minimum (emin) values
silica sand. Vesic (1965) added 10% shells to silica sand and as expressed in Equation (8). The complication with the use of
resulted in a decrease of qc values by a factor of 2.3. The higher relative density is the difficulties in measuring density below water
crushability of the carbonate grains is responsible for the low qc and the uncertainty associated with the determination of emax and
values. Since grain crushing increases with increasing density and emin. Hence, direct testing of relative density measurements is
increasing pressure therefore, a larger ratio of qc(silica) / qc(carbonate) is seldom used these days in favour of correlations developed from in-
obtained for denser states and larger mean pressures. At a denser situ tests such as CPT.
state, silica sand shows significant increase in qc values since the Jamiokowski et al. (2001) reviewed test data obtained from
hard particles are forced aside during cone penetration while for calibration chamber tests (Figure 36). Supplemented with few data
carbonate sand, it implies grain breakage, fracturing and crushing as available on undisturbed frozen sand samples, it shows the relative
the fragile particles are pushed closer together as they approach their density against the log scale of normalised cone resistance is almost
minimum packing arrangement (i.e. emin). Figure 33 presents test a linear relationship but still exhibit quite a wide spread of data.
results that reflect this behaviour. When relative density is less than
30% (loose state), the measured qc is about the same for both
carbonate and silica sand. However, for relative density greater than
30% and increasing, the qc(silica) / qc(carbonate) increases significantly to
reach a maximum value of 3.5 when the relative density is
approaching 100%.
Figure 34 Cone resistance qc and relative density for vibro

compaction
Figure 33 Ratio of qc(silica) / qc(carbonate) against DR (Mayne, 2014)
In order to correctly assess the relative density of carbonate

sand, correction factors (also known as “shell factors”) are used to
increase the qc values of carbonate sand to equivalent values
representative of silica sand and thus, allowing for the use of
existing correlations (e.g. Baldi et al., 1982) developed for silica
sand. The correction factor given in Figure 33 is based on Kulhawy
& Mayne (1990) and Jamiolkowski et al (2001) and it is expressed
based on normalised cone resistance qc values:
q c (silica) 5 (10)
 6
q c (carbonate) 1  (D R /100) 4
Figures 34 and 35 show the post compaction qc values adjusted

with a correction factor of 2 following Equation (10) and Figure 33.
With this correction, the measured equivalent qc values satisfied the Figure 35 Cone resistance qc and relative density for dynamic
acceptance qc values. compaction
Also, profiles of relative density of 70% based on the method of
Jamiolkowski et al. (1985) which was developed for various
compressibility of sand are included in Figures 34 and 35. Sand of
high compressibility includes carbonate (calcareous) sand while
sand of intermediate (medium) compressibility includes siliceous
sand with approximately equal parts of quartz and feldspar. Sand of
low compressibility includes those of quartz. For high to
intermediate compressibility, the measured qc values meet the 70%
relative density criterion after compaction. Hence, the carbonate
content and its compressibility may have caused the unexpected
large non-conformance of the results.
4.9 Lessons Learned
The question is “Is relative density a suitable performance
(acceptance) criterion for deep sand compaction?” The degree of
compactness of sand has been traditionally expressed in terms of
relative density. In simple terms, it is merely the location of the void Figure 36 Relative density evaluations from CPT (Mayne, 2014)
73
For example, a normalised cone resistance of 100 would give a 6. ACKNOWLEDGMENTS

relative density ranging from 40% to 70% depending on the
compressibility of sand. Also, there is a big spread for the The authors wish to thank their former colleagues in Menard for
undisturbed samples e.g. a 40% relative density would have a their assistance with some of the works described in this paper. They
normalised cone resistance from 20 to 200. Hence, a qc reading can also wish to record their gratitude and admiration to the project
correspond to a large range of relative densities and this leads to owner, project consultants, project managers and main contractors
uncertainty in the estimation of a “true” relative density. who have given them the opportunity to work on this challenging
The uncertainty gets worse with carbonate sand. Although some project and wish to thank them for their insight and dedication to
correction factors have been proposed to “correctly” assess the excellence in dealing with the project’s technical challenges.
relative density of carbonate sand from existing standard
correlations for silica sand, these factors are varying over a range of 7. REFERENCES
values. This increases the uncertainty already associated with the Amar, S., Clarke, B.G.F., Gambin, M.P. and Orr, T.L.L. (1991). The
standard correlations. Therefore, the concept of relative density may Application of Pressuremeter Test Results to Foundation
not be suitable to be used in the specifications for deep sand Design in Europe. Part 1: Predrilled Pressuremeters / Self-
compaction in land reclamation projects especially sand fill with boring Pressuremeters. ISSMFE European Regional Technical
substantial carbonate content. Committee No. 4 – Pressuremeters. 48 p.
It would be more appropriate to derive “directly” the Baldi G., Bellotti R., Ghionna V., Jamiolkowski M. and Pasqualini
specifications according to performance criteria which are E. (1982) "Design parameters for sands from CPT",
established from the functional requirements of the project needs – Proceedings of the 2nd European Symposium on Penetration
defined by the allowable stress-induced deformation (stiffness), Testing, ESOPT II, Amsterdam, May, Vol. 2, pp. 427 - 432.
bearing capacity and stability requirements (strength) and Chang M.F., Yu G., Na Y.M. and Choa V. (2006) “Evaluation of
liquefaction potential. It is noted that relative density is an relative density profiles of sand fill at a reclaimed site”
intermediate parameter. Strength and stiffness are not always well Canadian Geotechnical Journal, Vol. 43, pp. 903 - 914.
represented by relative density. In contrast, the cone resistance (qc) Choa V., Karunaratne G.P., Ramaswamy S.D., Vijiratnam A. and
is essentially directly responding to both the strength and stiffness. Lee S.L. (1979) "Compaction of sand fill at Changi Airport",
The measurements of cone resistance, sleeve friction and pore water Proc. of the 6th Asian Regional Conference on Soil Mechanics
pressure can be used either separately or together to evaluate soil & Foundation Engineering, Singapore, Vol. 1, pp. 137 - 140
engineering parameters including state of stress, strength and Chu J. (2011) Personal Communications.
stiffness. Thus, it is far better to estimate the friction angle, stiffness D’Appolonia E. (1970) "Dynamic loading", Journal of Soil
and liquefaction directly from CPT results and not go through the Mechanics & Foundation Engineering, ASCE Vol. 96 No.
intermediary step of using relative density to estimate these soil SM1.
engineering parameters. Instead of stipulating the relative density D.60.AN (1975). The Menard Pressuremeter – Interpretation and
and the accompanying test values (e.g. CPT qc) in the specifications, Application of Pressuremeter Test Results to Foundation
it is far better to stipulate the required stiffness and strength Design. General Memorandum. Sols –Soils No: 26 45 p.
parameters. Recently, the state parameter (S) has found interest Das B. M. (2008) Advanced Soil Mechanics, Taylor & Francis,
because of its application to critical soil mechanics and a rational London & New York.
framework towards understanding of soil liquefaction problems. Jamiolkowski M., Ladd C.C., Germaine J.T. and Lancellotta R.
(1985) "New developments in field and laboratory testings of
5. CONCLUSIONS soils", State-of-the-Art Report. Proceedings of the 11th
ICSMFE, San Francisco, Vol. 1 pp. 57 - 153.
The quest for land is as old as time immemorial. Land is
Jamiolkowski M., LoPresti D.C.F. and Manassero M. (2001)
Singapore’s most cherished resource. Since its first land reclamation
"Evaluation of relative density and shear strength of sands
works in 1822, the small city-state has grown from 58,150 to 71,910
from cone penetration test and flat dilatometer test", Soil
hectares in 2017. It has set a goal to reach 76,600 hectares by 2030.
Behaviour and Soft Ground Construction, GSP 119, ASCE,
This whopping increase in land area by reclamation – dubbed by the
Reston, VA, pp. 201 - 238.
local media as a “gift from the sea”; brought many challenges and
Kulhawy F.H. and Mayne P.W. (1990) “Estimating soil properties
also opportunities for innovative solutions.
for foundation design”, EPRI Report EL-6800, Electric Power
Some of the challenges and lessons learned in soil improvement
Research Institute, Palo Alto, 306 p.
works and acceptance tests are highlighted in this paper. The
Mayne P.W. (2014) “Interpretation of geotechnical parameters from
following lessons learned by the authors have been discussed:
seismic piezocone tests”, Proc. 3rd International Symposium
 Adaptation of non-conventional materials to be used for soil
on Cone Penetration Testing, Las Vegas, Nevada, pp. 47-73
improvement works in deep water under the prevailing
Menard, L. (1970). Determination de la Poussee Exercee par un Sol
conditions to achieve the required performance criteria;
sur une Paroi de Soutenement. Centre d’Etudes Menard,
 Adaptation of “site-specific-designed” construction plant and Publication D.38. 1970.
equipment used for the soil improvement work in deep water; The Straits Times (2016) “Engineering feats at Pasir Panjang a
 Use of appropriate quality control (QC) in-situ tests (e.g. static model for Tuas” October 28, 2016
and cyclic PMT tests) to suit the type of materials to be tested; Varaksin, S. and Racinais, J. (2009) “Etude Des Paramètres
 Conduct visual inspections and field observations (especially in D’application de la Consolidation Dynamique et de ses
deep water) before, during and after works to provide additional Techniques Dérivées” Proceedings 17th ICSMGE, Alexandria,
information to supplement results obtained from in-situ tests. Egypt, pp. 2407 - 2410.
 Specify performance criteria that are established from functional Varaksin, S. (2009). Personal Communications
requirements of the project needs defined by soil engineering Vesic A.S. (1965) “Ultimate loads and settlements of deep
parameters. foundations in sand” Proceedings of the Symposium on
The authors believe that successful soil improvement requires a Bearing Capacity and Settlement of Foundations in Sand,
blend of relevant soil engineering parameters, informed analytical Duke University, Durham, pp. 53-68
modeling supported with appropriate methods of analysis, adequate Yee, K and Varaksin, S. (2012) “Ground Reinforcement in Deep
in-situ and laboratory tests coupled with field observations and Water” Proceedings of the International Conference on
performance measurements according to functional performance Ground Improvement and Ground Control, Wollongong,
criteria. Learning and heeding lessons learned from past projects are Australia 30 Oct – 2 Nov 2012, pp. 575 - 585
important additions to the list.
74
Design and Construction of Foundation System for Malaysia First Drawbridge at

Kuala Terengganu
LIEW Shaw-Shong1, KOO Kuan-Seng2, CHEE Fong-Wah3 and TAN Pei-Ying4
1,2,3 & 4
G&P Geotechnics Sdn Bhd, Kuala Lumpur, Malaysia
E-mail: gnp-geo@gnpgroup.com.my
ABSTRACT: This paper describes the design and construction of foundation system for Malaysia’s first drawbridge, which is located at the
river mouth of Kuala Terengganu River. The proposed drawbridge aims to as bridge link connecting the northern and southern areas of the
Kuala Terengganu River. With the 76m long main span of bascule bridge decks, it permits no obstruction to the navigation traffic along the
river. The design challenges of the drawbridge foundation with free-standing pile length above the river bed are its stringent foundation
displacement requirements for the operation of the mechanical bascule structures under the critical loading conditions (i.e. wind load, seismic
load, accidental vessel impact and hydraulic load in addition to the structural dead loads, live loads and other imposed operational loads) and
also compliances with maritime structure design code of practice throughout the required designed life period. In general, each of the
drawbridge tower structure is founded on 118 numbers of 1m diameter with vertical and raked pre-casted marine spun piles. A giant pilecap
with dimensions of 76m (long) x 28m (wide) x 4.0m (thick) was designed for load transfer between super-structure and foundation system.
During construction works, the validation of the pile performance was very carefully devised and implemented, which included reference
compressive static maintained pile load test, static lateral pile load test, Statnamic test, high strain dynamic pile tests on working marine piles
over the river. The design and construction processes of ensuring the pile head connection conditions achieving the design rotation stiffness
and the problems encountered during massive staged concrete casting of the giant pilecaps with submerged soffit will also be elaborated and
discussed in this paper.
Keywords: bridge foundation, drawbridge, marine structure, marine piling
1. INTRODUCTION RL+1.32 m (+2.77 m CD) and RL-1.45 m (0.00 m CD) respectively.

The riverbed level along the bridge alignment is at about RL-8.3 m
This paper shares the experience on design and construction of an (-6.85 m CD).
alternative marine foundation system learnt from Malaysia’s first Based on the Geological Map of Peninsular Malaysia, 8th
drawbridge project, which is located at the river mouth of Kuala Edition, published by Minerals and Geoscience of Malaysia in 1985,
Terengganu River, with the intended purpose of connecting the the site location is generally underlain by Quaternary Alluvium. The
northern and southern divided developments across the river. The site investigation program confirms that the alluvium deposits at this
overall length of the two three spanned integral bridge from both area generally consist of marine deposits with mainly dominant of
side of river banks approaching the 23 m wide bascule bridge deck sand and gravel, but also consists of clay and silt of intermediate
on two tower piers is 632 m. With the 76 m long main span of plasticity. The STP’N value of subsoil is generally less than 15 for
bascule bridge decks on river, it permits no obstruction to the daily 22 m depth below riverbed and hard residual formation stratum was
navigation traffic along the river. Figure 1 shows the overall layout reported at 21 m to 26 m below riverbed as revealed from the
plan and section view of the project. proposed boreholes, in which Figure 2 shows the interpreted subsoil
profiles of the riverbed.
Figure 2 Interpreted subsoil profiles
3. DESIGN CRITERIA
The design criteria of the foundation system mainly focus on
operation requirements of drawbridge structure, serviceability
Figure 1 Overall layout plan and section view conditions and durability of marine structure throughout the
designed life period. Interaction design coordination was conducted
The foundation system at marine portion consists of 4 piers, among bridge design engineer, operation system suppliers and
namely P3 to P6. Two heavy foundation at P4 and P5 are designed foundation design engineer to establish design criteria for the
to support 55 m high drawbridge towers and 76 m long bascule foundation design. As a result, the foundation system need to design
bridge decks and the end span of the integral bridge. for total settlement less than 10mm and horizontal displacement less
than 20 mm in any directions under service conditions.
2. HYDROLOGY AND GROUND CONDITIONS As the foundations structures are located at river mount near to
The site is located at Kuala Terengganu river mount and subjects to South China Sea and subject to daily seawater tide, therefore the
daily seawater tides from South China Sea. Accordance to the Kuala durability design shall according to the marine structure standard.
Terengganu standard port datum, mean sea level (MSL) of the The foundation structure design including pilecap are in compliance
project site was defined at RL0m (+1.45 m Chart Datum (CD)). with codes of practice (British standard) BS5400 - Steel, Concrete
Meanwhile, the highest and lowest astronomical tide level are at and Composite Bridge, BS6349 - Maritime Structures and BS8500 –
75
Concrete. Based on Table 1 in BS5400 part 4, the environment 5. FOUNDATION DESIGN

condition of the project is classified as extreme condition. Where the
concrete surface exposed to the abrasive action (high and low tidal The original tender design was cast-insitu bored piles with
cycle, and upstream flood storm) of sea water; thus, the design crack temporary cofferdam island as working platform for bored piles
width of the reinforced concrete shall limit to 0.1mm in service limit installation and pilecap construction. An alternative proposal with
state (SLS) condition. In view of the extreme condition, concrete for cost and time effective consideration was proposed using pre-
the pilecap also to design in accordance with BS8500-1. Exposure stressed spun pile with marine driven piling method. Upon
class of the pilecap condition are defined as class XS3 in which is considering the logistic and constriction feasibility, 1m diameter
subjected to tidal, splash and spray zones as describe in Table A.1 of with grade 80 pre-stressed spun pile (140 mm wall thickness with
BS8500-1. 7.4 MPa effective pre-stressed) up to 36 m long in single casting
The proposed alternative marine spun pile foundation design length was selected for this project. Pre-welding with extension pile
was also considered some extreme cases condition such as wind for pile length more than 36 m and pile shoe welding were carried
load especially wind load on bascule deck in lift-up condition, out at pile manufacturing factory and delivered to project site using
seismic load during design earthquake incident, accidental ship marine barge. The spun piles are installed by marine piling barge
impact onto foundation structure and hydraulic forces due to river using hammer drive in method.
storm flow. Based on the manufacturing product information, allowable
Other than design considerations mentioned above, construction axial load of 1m diameter marine spun pile is 6850 kN and
methodology and sequence were also evaluated to ensure that the calculated cracking bending capacity is 835 kNm. The estimation of
proposed method can meet to design requirements at all times. compression pile allowable geotechnical capacity is generally based
Quality assurance and control measures during construction work on the SPT-N profile using Meyerhof’s Method (1984). The
are also importance such as temperature control, provision of ultimate bearing capacity of pile is equal to the sum of the ultimate
construction joints for large pour concreting of the gigantic pilecaps. resistance of the base of the pile and the ultimate skin friction over
the embedded shaft length of the pile. The allowable geotechnical
4. ANALYSIS LOAD CASES capacity of a single pile is generally derived by applying partial
safety factors as follows: -
Several load cases in SLS and ultimate limit stage (ULS) were
provided by bridge engineer based on their superstructure
requirements. Table 1 summaries all the critical load cases in SLS (1)
for pile group design and Table 2 summaries all the critical load
cases in both SLS and ULS for pilecap design. (2)
Table 1 Critical Load Cases in SLS for Pile Group where,
fb = Base resistance (150 x SPT-N limit to 16,500kPa)
No. Load Case ϒfL considered in load combination fs = Shaft resistance (2.5 x SPT-N)
As = Surface area of pile shaft
DL SDL LL WL EL AL DF Ab = Surface area of pile base
1 SLS-1.0EL 1.0 1.0 1.0 - 1.0 - -
Estimated pile lengths below riverbed level were based on pile
2 SLS-1.0WL 1.0 1.0 1.0 1.0 - - - either terminated within hard stratum not beyond bearing materials
with SPT-N value more than 200 or resting on bedrock. Calculated
3 SLS-1.0AL 1.0 1.0 1.0 - - 1.0 - allowable geotechnical capacity based on available adjacent
borehole results are summarised in Table 3.
4 SLS-1.0DF 1.0 1.0 1.0 - - 1.0 1.0
Table 3 Calculated Allowable Pile Geotechnical Capacity
Estimated Pile
Table 2 Critical Load Cases in SLS and ULS for Pilecap Design Allowable Geotechnical
Pier Reference Length below
Capacity
No. Load Case ϒfL considered in load combination No Borehole Riverbed
(m) kN Shaft Base
DL SDL LL WL EL TL BH-5 25 5,851 15% 85%
1 ULS-1.2EL 1.2 1.2 1.2 - 1.2 1.2 P4
BH-6 22 5,966 16% 84%
2 ULS-1.2WL 1.2 1.2 1.2 1.2 - 1.2 BH-7 20 5,547 13% 87%
P5
3 ULS-1.4EL 1.0 1.0 1.0 - 1.4 1.0 BH-8 22 5,717 14% 86%
4 ULS-1.4WL 1.0 1.0 1.0 1.4 - 1.0
Single pile working load is 5500 kN with major contribution up
5 SLS-1.0EL 1.0 1.0 1.0 - 1.0 1.0
to 85% from base resistance. Unique pile shoe design was adopted
6 SLS-1.0WL 1.0 1.0 1.0 1.0 - 1.0 to enhance pile toe structure resistance. Thick steel end plate pile
shoe was designed to resist end bearing pressure of pile as per
Figure 3.
Where,
fL = partial load factor
DL = Dead load
SDL = Superimposed dead load
LL = Live load
WL = Wind load
EL = Earthquake seismic load
AL = Accidental load
DF = Water flow drag force
TL = Traffic load. Figure 3 Pile Shoe Details
76
Pile group analysis was carried out by using soil-structure 6. MARINE PILECAP DESIGN
interaction software, PIGLET developed by M.F. Randolph (2004).
Three dimensional pile group deformation as well as forces along The proposed pilecap for Tower Piers P4 and P5 are approximate in
the pile can be reasonably well estimated by adopting soil-structure capsule-like shape with long dimension of 76 m and width of 28 m
input parameters as summarised in Table 4. (see Figure 4). General pilecap thickness is 4m except the centre
part reduced to 2 m for accommodating the counterweight block of
bascule deck required for operation purpose. Concrete grade
Table 4 PIGLET Input Parameters
adopted is Strength Class C32/40 (cylindrical strength/cube
Input parameter Design value strength) with minimum characteristic cube strength of 40 N/mm2 at
28 days.
Soil Shear Modulus (Axial), G axial 11.6 MPa According to BS8500, exposure class of the concrete for marine
pilecap shall be classified as class XS3 (tidal, splash and spray
Soil Shear Modulus (Lateral), G lateral 2.8 MPa
zone). To obtain concrete grade C32/40 with 75mm for class XS3,
Soil Base Modulus, G Base 2.3 GPa cement type CEM IIB-V (Portland cement with 21-35% fly ash) or
cement type CEM IIIA (Portland cement with 36-65% ground
Pile Diameter, D 1m granulated blast furnace slag) shall be adopted as supplied concrete.
The proposed cement content and water ratio for the concrete mix
Pile Young Modulus, E axial 19.3 GPa design (C32/40 with 75 mm for class XS3) are as follow: -
Pile Young Modulus, E lateral 29.3 GPa
Table 7 Proposed Concrete Mix Design
Generally, Young Modulus of grade 80 spun pile is taken as 40 Proposed Requirement

Subject
GPa and conversion from spun pile to solid circular pile properties Design of BS8500
need to be calculated as equivalent parameters.
Ground Granulated Blast
The outcomes of the analysis results indicated that a total 118 60% 35% - 65%
Furnace Slag Content
nos. of spun pile are required in order to meet the serviceability
requirement. Figure 4 shows the piling layout plan of Tower Piers Cement Content (kg/m3) 415 > 360
P4 and P5. For enhancement of lateral pile group resistance, 32 nos.
raked piles (with inclination 1 Horizontal : 8 Vertical) in the pile Water Ratio 0.4 < 0.45
group for Tower Piers P4 and P5 were proposed during the pile
arrangement configuration. The maximum predicted vertical
settlement and lateral displacement in X-Direction (along bridge
alignment) and Y-Direction (perpendicular to bridge alignment) are
summarized in the Table 6.
Table 6 Summary of Pile Group Deformation Analysis Results

P4 P5
(mm)
No Load Case
z x y z x y
1 SLS-1.0EL 9.8 1.6 3.1 9.7 1.7 3.3

2 SLS-1.0WL 9.8 17.8 3.9 9.7 19.1 4.3
3 SLS-1.0AL 9.8 7.6 12.9 9.7 8.2 13.7
4 SLS-1.0DF 9.8 7.6 6.7 9.7 8.2 7.3
Where z, x and y refer to pile group displacements (in mm) in

vertical settlement, X and Y directions on plan as defined above.
By summarizing the pile group displacement analysis obtained
from each load case, it was concluded that the predicted
displacements for the final configured pile group arrangement after
numerous configuration trials are able to fulfil the foundation
displacement requirement where overall vertical settlement are less
than 10 mm and lateral displacement in X and Y direction are both
less than 20 mm as required.
Fixed pile head connection to pilecap was adopted in foundation
system to efficiently control the lateral displacement of foundation
system. Fixed pile head connection is defined as a connection details
with ability to develop and maintain the restraining end moment at
the connection portion under zero rotation condition at the
connection interface. For spun pile with hollow annulus section,
reinforced concrete pile plug is required to serve as bridging
connection element between spun pile and pilecap structure. The
tension reinforcements and anchorage length shall be provided to
attain the end moment reaction as expected. Typical pile plug details
is shown in Figure 5. Figure 4 Piling Layout Plan
77
Figure 5 Typical Pile Plug Details
7. DESIGN VERIFICATION TESTS

Design verification tests were conducted to verify the analysis and Figure 6 Maintained Load Test Result
design assumptions before commencement of production work. The
spun piles were designed to take lateral forces and therefore pile The LPT is able to verify the design assumptions on prediction
bending resistance required to be verified by tests. Pile structural of lateral pile group movement under service loading condition
bending test was carried out to obtain allowable pile structural (Serviceability Limit State Check). The tested pile length is 38 m
bending capacity as well as pile stiffness derived from pile long instrumented with inclinometer at centre of spun pile annulus
deflection profile under four-point flexural test. Calculated elastic for lateral displacement monitoring. In addition, linear variable
modulus (E) of Grade 80 concrete is 40 GPa and Flexural Stiffness differential transformer sensors were attached on a separate
(EI) is 1,436,000 kNm2. For the pile concrete durability test, Rapid reference column to measure the lateral pile head deflection. 5 nos.
Chloride Penetration Test (RCPT) in accordance with ASTM 1202- temporary pipe piles were installed in line with test pile as reaction
97 was assigned. The result showing the concrete used for spun pile piles. Figure 7 showing setting out arrangement and the LPT results
is in compliance with the project specification of charge passing not are presented in Figure 8. In the first loading cycle, the observed
more than 1000 coulombs, implying very low chloride ion maximum pile head deflection at the tested load of 137.8 kN was
penetrability of the concrete and thus is suitable to be used for the 147.53 mm. Upon unloading to zero, the pile rebounded to a
proposed site environment. For environment control, a layer of residual deflection of 18.32 mm. In the second loading cycle, the
bituminous coating was required for top 3m from the pilecap soffit observed maximum pile deflection at the tested load of 220.0 kN
level as additional protection. was 398.67 mm for the test pile.
Continue pile monitoring (CPM) test on pile installation was
performed to monitor the impact force and stresses during pile
installation until termination to establish the required set criteria. Set
criteria was also verified with the performance of mobilised capacity
in High Strain Dynamic Pile Test (HSDPT) during end of drive of
the test piles. As a result, pile termination criteria adopted for
vertical pile is 20 mm per 10 blows using 35 tons hammer with drop
height of 0.8 m. Whilst for raked pile is 10 mm per 10 blows after
considering hammer drop efficiency in incline position for the
necessary pile rake.
An instrumented non-working pile near to Pier P5 was carried
out in a maintained load test using reaction pile system with
additional deadload of barge over the reaction beam. The main
objective of the instrumented test pile is to establish and verify the
design assumption of adopted soil parameters. 15 nos. temporary
pipe piles were installed at surrounding of test pile as reaction piles. Figure 7 Lateral Pile Test Result
The test pile was expected to load test up to maximum compressive
load of 16,500 kN (3 times of pile working load) or failure load
whichever comes first. The result of maintained load test is shown in
the Figure 6. Pile top settlement at working load is 11.98 mm with
residual settlement of 1.94 mm. Pile toe settlement at working load
is 2.94 mm which correspondence to soil base modulus of 2.3 GPa
(approximate 1150 times of average SPT-N value for PIGLET
analysis). Back analysis from the MLT results for soil shear
modulus in axial is 11.6 MPa (approximate 1450 times of average
SPT-N value for PIGLET analysis).
The foundation for all over water Piers, P3, P4, P5 and P6 are
subjected to lateral load. Lateral shear modulus, Glateral adopted for
lateral deformation and also the later pile capacity in the foundation
design shall also be subjected to field test verification. Lateral pile
load test (LPT) had been performed to assess the pile performance
in both structural and geotechnical capacity as well as soil stiffness
and pile deformation under lateral action.
Figure 8 Lateral Pile Test Result
78
Back-analysis was carried out with computer software, PIGLET Pre-joined one-length spun pile was decided with construction
to compare the assumed subsoil lateral shear modulus with the and design consideration. From construction aspect, one-length pile
actual pile lateral deformation. It was observed that actual lateral is commonly use in marine piling with time saving for handling and
shear modulus, Glateral, of subsoil from lateral pile test is avoidance of inconsistency of site welding of pile joint. The
approximately 2.8 MPa (approximate 350 times of average SPT-N delivered pile lengths were in 32 m and 38 m as estimated from
value for PIGLET analysis), which is relatively lower than the adjacent boreholes information plus extra 3m for construction
assumed lateral shear modulus of 4.3 MPa in the design stage. tolerance.
With the two preliminary pile tests under compressive loading Some pile performance verification tests were conducted in
and lateral loading carried out, design assumptions and adopted river. Temporary work preparation for the pile tests were properly
subsoil parameters can be verified and calibrated for foundation planed with a safe temporary platform for the sole purpose of
design. With the obtained design parameters and review of the four testing. In general, working platform level is required to be above
river pier pile group performance, revised detailed design with Highest Astronomical Tide (HAT) level to be free from the tidal and
more piles was needed for Pier P3 to ensure compliance pile group wave impacts.
performance under the assigned loading. The pile foundations are designed with fixed head condition
During the working pile installation, 2 nos. of statnamic tests between pile head and pilecap. Load transfer element with
were carried out for Tower Piers P4 and P5 as part of the quality introducing 3.5 m pile plug inside the spun pile anulus of sufficient
assurance/ quality control (QA/QC) procedure. Outcomes of the anchorage reinforcements into pilecap. Effectiveness of pile plug
statnamic load test results showed mobilised test load with force performance is much dependent on installation workmanship. A
pulse duration satisfied the condition stated in ASTM D7383-10: (1) layer of weak cement laitance with primary cement/lime slurry will
The applied force shall exceed the pre-load for a duration time of at normally found around the pile inner anulus due to segregation of
least twelve times the test pile length (L) divided by the strain wave excessive cementitious slurry and aggregate compaction from the
speed (c), 12L/c, and (2) The applied force shall exceed 50% of the spinning process. Therefore, the contractor shall ensure proper
actual peak force for a minimum duration of four times L/c. removal of this weak cement laitance before casting of pile plug.
HSDPT had also been carried out as QA/QC procedure for During piles installation, unexpected weak soil stratum was
installed working pile. Total of 22 Nos. (9.3% of installed piles at encountered at Pier P4 causing longer pile penetration length than
P4 & P5) HSDPT were conducted with mobilised test load of up to estimated pile length. As such additional extension piles with on-site
2 times working load. pile joint welding was needed. In view of potential insufficient
Meanwhile, verification test for designed concrete is vital before bending capacity and flexural stiffness at pile joint area, longer
the concrete casting work on site in order to prevent concrete quality reinforced pile plug was proposed to enhance the pile joint capacity.
issue of the structure along the structure’s service life. Series of test After completion of piles installation works, an unforeseen
such as concrete slump retention test, concrete casting temperature incident case was encountered at Pier P5. A pile head was observed
monitoring, concrete trial mix and concrete cube strength test were missing during site inspection. As the regular river traffic for public
carried out prior to pilecap execution works on site. has to be uninterrupted between P4 and P5, it was suspected the pile
Large volume concrete casting (casting thickness more than 1m) was damaged by uninformed accidental ship impact occurred during
was anticipated at the initial stage of project and the temperature the period of off-construction works. Underwater inspection had
development during casting was always the major concern for the been carried out and the missing pile head was found broken and
pilecap construction work, thus, temperature monitoring was estimated the damage at 9.5m below pile cut off level (2.5 m below
proposed during initial stage concrete casting at Pier P3. Designated riverbed level). Remedial works had been explored with permanent
initial concrete temperature was checked with estimation of 12˚C steel casing extended up above water level. The material falls into
heat generated from every 100 kg/m3 Portland cement hydration pile annulus required to be cleaned up for placement of
process and thus determining 27˚C as the concrete initial reinforcement cages in the remedial works by tremie concreting in
temperature; to control maximum concrete temperature below 75˚C the cleaned pile annulus from pile toe to the pile cut off level.
during curing. However, the designed temperature is then further The proposed pile cap soffit level of Piers P4 and P5 is RL-0.5
verified with on-site concrete temperature monitoring. m, which is 0.5 m below the MSL. Underwater concrete casting for
Post construction instrumentation and monitoring was proposed the pile caps was anticipated. Several options had been explored and
upon completion of the pilecap casting works. Four (4) settlement proposed by contractor for underwater concrete casting method. The
markers were installed on each pilecap to monitor the foundation final adopted temporary work for pile cap concrete casting is
performance during construction stage of the superstructure. The watertight steel formwork mounted to the group piles with water
instrumentation monitoring results show maximum cumulative seals around the piles. Before placing of steel formwork, vertical
foundation settlement of 5 mm with substantial dead load existed on support at the piles is required to carry the vertical loading from
the bridge in January 2018. The instrumentation monitoring period temporary formwork and self-weight of the lean concrete mentioned
will be extended up to completion of superstructure construction later. The designed supports consist of prefabricated steel gripper on
works. each installed pile head and some secondary girders supporting the
steel form. In order to provide sufficient bonding between the piles
8. CONSTRUCTION and steel grippers, a layer of synthetic rubber was used. Bonding test
was carried at laboratory for design certification. The steel gripper
The proposed Piers P4 and P5 are located in Kuala Terengganu was locked by competent diver under water with bolt and nut
River and the piling works for the foundation system was installed tightening system. After that, steel formwork will progressively
by adopting marine piling method i.e. hammer driven from piling place over the support system with a layer of 150mm thick lean
barge. Several constructions aspects needed to be taken into concrete below pile cap soffit level. Dewatering from inner side of
consideration for the marine piling works such as correct positioning steel formwork was then carried out to create dry working condition
and stability of piling barge during pile handling and installation for pile cap concreting work. For safety consideration, the water
process under the impacts from active river flow, daily tidal changes pressure outside the steel formworks was considered at HAT level.
from South China Sea, unpredictable strong wind gust in the distinct Figure 9 shows the pile cap concrete casting preparation works.
monsoon season at east coast of Peninsular Malaysia. The pile cap thickness of the pile cap Piers P4 and P5 is 4 m
Aforementioned conditions have created work challenges and safety thick with total concrete volume of 7,025 m3 for each pile cap. With
issue of work, thus posing uncertainty in overall construction consideration of local concrete supplier availability and the friction
progress. Addition precaution measures and construction planning limitation of the steel gripper support, concrete casting had been
are considered to prevent any unwanted incident and delay during divided into 4 stages. Stage 1 – cast a layer 700 mm thick concrete
construction works. throughout entire pile cap footprint (concrete volume of 1,353 m3)
to serve as base support for subsequence concreting work. Stage 2 –
79
cast a layer of 1,300 mm thick concrete (concrete volume of 2,124

m3) on top of completed base layer. Stage 3 – cast a layer of 1,000
mm thick concrete (concrete volume of 1,774 m3) except centre
portion, in which a void was deliberately formed to house the
counterweight block of the bascule deck. Stage 4 – final cast with
layer of remaining 1,000 mm thick concrete (concrete volume of
1,774 m3) up to finished pile cap top level. By adopting stages
concrete casting method, cold joint between earlier cast concrete and
new cast concrete was expected. With that, the flexural horizontal
shear stress against the interface shear resistance capacity shall
consider in pile cap design, and thus horizontal shear resistance was
enhanced by placing shear key reinforcements across the entire cold
joint surface.
Pump concrete method was adopted for marine pile cap casting.
Pumped mixed concrete at batching plant was delivered to riverbank
via concrete truck. After that, the concrete was pumped to pile cap Figure 10 Concrete temperature monitoring
through a 100 m long steel pipe with temporary support over the
river. Two concrete pumps had been deployed on site where one 9. CONCLUSION
pump was on standby in case any pump breakdown to maintain
Alternative foundation design work was started since year 2014
uninterrupted concrete casting. Workability and pumpability of
when project was awarded. Regular meetings and coordination
concrete also been considered for large volume and long duration
workshops were conducted between bridge engineer and foundation
concrete casting. The supplied concrete workability had been
engineer to establish an appropriate alternative foundation system as
specified to have minimum 4 hours slump retention period with
well as foundation contractor for ease of construction for Malaysia
consideration of maximum 2 hours concrete delivery time and
first drawbridge project. Several loading cases were considered in
maximum 2 hours concrete pumping and placing time.
foundation design to meet the serviceability and operation
requirements of bridge superstructure. Environmental and marine
conditions and potential worst design condition of the project were
also taken as part of design and construction evaluation. Cost and
time effective foundation system with driven pre-stressed spun piles
was developed purposely for this project to save both time and cost
of the project. Physical foundation construction work was
commenced in early of year 2015 and completed in end of year
2016. Superstructure work including the integral bridge, bascule
deck and tower structure are in progress in year 2018. Figure 11
shows the latest overall site condition in Feb 2018.
Figure 9 Pile cap concrete casting preparation works
For concrete temperature control, heat generated during

hydration process shall be controlled not exceeded 70°C and change
in gradient not exceeded 20°C. Based on the temperature monitoring Figure 11 Progress in Feb 2018
result at P3 pile cap shown in Figure 10, heat generated for 2,350
mm thick pile cap is 40°C with 415 kg/m3 cement content. 10. REFERENCES
Therefore, the initial temperature during concrete mixing shall be British Standard Institute (BSI). (2006) BS8500: Concrete –
controlled below 30°C. From the concrete mix design, free water Complementary British Standard to BS EN 206-4.
requirement is designed to be 165 kg/m3 with about 40% of water British Standard Institute (BSI). (1988) BS5400: Steel, Concrete and
will be substituted by ice in order to lower down the initial Composite Bridges.
temperature of concrete. For quality assurance and quality control, Ken Fleming, Austin Weltman, Mark Randolph and Keith Elson,
initial temperature of concrete and slump had been measured and (2009) “Piling Engineering”.
recorded. Concrete temperature or slump test exceeding specified Liew S.S., Koo K.S., Chee F.W. and Tan P.Y, (2016) “Planning and
requirement was not allowed to proceed for concreting. Interpretation of Instrumented Lateral Pile Test Performance
with a Semi Restrained Pile Head Condition”
M.F. Randolph, (2004) “PIGLET Analysis and Design of Pile
Groups”.
80
Advancement of Bearing Capacity and Settlement Analyses of Piled-Raft Foundation

Krisnanto, S.1, Sengara, IW.2, Adelina, F. N.3
1
Assistant Professor, Bandung Institute of Technology (ITB), Bandung, Indonesia; Indonesian Society for Geotechnical
Engineering (ISGE/HATTI)
2
Professor, Bandung Institute of Technology (ITB), Bandung, Indonesia; Indonesian Society for Geotechnical Engineering
(ISGE/HATTI).
3
Research Student, Bandung Institute of Technology (ITB), Bandung, Indonesia.
1
E-mail: sugeng.krisnanto@ftsl.itb.ac.id
1
Website: http://personal.ftsl.itb.ac.id/sugeng/
ABSTRACT: Piled-raft consists of pile group and raft. From the pile group point of view, the presence of raft provides an additional bearing
capacity to the pile group. From the raft point of the presence of piles reduces the settlement of the raft. Bearing capacity and settlement of
piled-raft foundation involve a complex soil-structure interaction. Since the pile configuration, pile stiffness, raft stiffness, and soil properties
may vary, there are several combination of parameters involved in the piled-raft group foundation analysis. Several methods have been
proposed in the analysis of piled-raft foundation. The methods vary in the way of treated the piles, raft, and soil. Generally, the methods
consist of analytical method, numerical method, small-scale foundation test, and large-scale foundation test. This paper presents a review of
the advancement of bearing capacity and settlement analysis of piled-raft foundation. The present as well as the classical methods of analysis
are reviewed and discussed. The review shows that numerical method calibrated with small-scale laboratory lest is a promising method to
analyse bearing capacity and settlement behaviour of piled-raft group.
Keywords: Piled-raft group, bearing capacity, settlement.
1. INTRODUCTION corresponding settlement was then calculated. After the pile group
reach its ultimate capacity, the remaining load was carried by the
Pile is used to sustain load from the upper structure when the ground raft and the corresponding settlement was then computed. The load-
soil strength is low by transferring the load to a soil layer that has settlement curve consisted of two linear sections: (i) The first linear
enough strength (e.g. Poulos and Davis, 1980; Bowles, 1996; section corresponds to the load-settlement of pile group; (ii) The
Tomlinson, 1994). In the practice, the piles are arranged in a group second linear section corresponds to the load-settlement of raft.
with a pile cap. The pile cap can be located above ground surface or Randolph (1983) proposed an analytical method to analyse the
in contact with ground surface. load proportion between the pile group and the raft and the
In the case where the pile cap is in contact with ground surface, settlement of piled-raft group based on average behaviour. The
actually the cap can be considered as a raft. The pile cap may proportion of load carried by pile group and the raft can be
provide an additional bearing capacity to the pile group. Thus the computed incorporating the stiffness of the pile group and the
system can be considered as a piled-raft group. stiffness of the raft.
Piles can also be added to a raft foundation. In this condition, Randolph (1994) introduced a parameter of the stiffness of
piles work to reduce settlement of a raft foundation that has enough piled-raft and an interaction factor. The proportion of load carried
bearing capacity but has an excessive settlement. The system can by the pile group and the raft can be computed using these factors.
also be considered as a piled-raft group.
To obtain a general behaviour the bearing capacity and
3. NUMERICAL METHOD
settlement of piled-raft group is difficult for the following reasons: (i)
A large number of combination of pile group configuration can Poulos (1991) idealized the raft in a piled-raft group as strip and the
occur, (ii) A large number combination of pile group-raft stiffness piles in a piled-raft group as springs. A computer program GASP
can occur, (iii) The high load that is required to perform a full-scale (Geotechnical Analysis of Strip with Piles) was developed to
load test, (iv) Different stress-strain behaviour between a small-scale accommodate the idealization.
and a full-scale tests. Lee (1993) performed a piled-raft numerical analysis. Lade and
This paper presents a summary of the advancement of bearing Duncan-Chang stress-strain model were used to model the soil. Both
capacity and settlement analyses of piled-raft foundation. Several piled-raft and raft foundations were analysed. Settlement of piled-
methods of analysis are reviewed and discussed. Finally the raft group and bending moment of raft were obtained in this analysis.
prospective method of analysis is recommended. Poulos (1994) performed numerical analyses of interaction
among piles and between pile group and raft. The analysis was
2. ANALYTICAL METHOD based on elastic condition. The analysis was incorporated in a
computer program GARP (Geotechnical Analysis of Raft with Piles).
Several analytical methods have been proposed to design piled-raft. Ta and Small (1996) used finite element method to analyse
Hain and Lee (1978) analysed the raft as a flexible elastic plate. The piled-raft group. The analysis was a 2-D analysis incorporating the
piles were modelled as compressible friction piles and the soil were interaction factor. Therefore, the 3-D condition can be
modelled as elastic material. The effect of the piles to reduce accommodated with lesser computation effort. The method was able
settlement was analysed. to model a piled-raft group in layered soil.
Davis and Poulos (1972) introduced an analytical method to Long, D. C. (2016) performed analysis on a building founded on
calculate settlement of piled-raft group. The calculation of piled-raft group. The analysis was performed using a commercial 3-
settlement of piled-raft group was based on the calculation of pile D finite element software. In the analysis, it was assumed that the
group settlement by considering the effect of pile cap. In the load was carried by the raft first. After the raft reached its ultimate
calculation of pile group settlement, the effect of settlement of one capacity, the remaining load was then carried by the piles.
pile to other piles in the group was considered. The method was
developed based on an elastic linear assumption. Thus, the piles, the
pile cap, and the soil were in an elastic condition. Davis and Poulos
(1972) also proposed a method to generate a simplified load-
settlement curve of piled-raft group. In the proposed method, it was
assumed that the load was carried by the pile group first and the
81
4. ANALYTICAL-NUMERICAL METHOD 8. NUMERICAL ANALYSIS ON ACTUAL PILED-RAFT

GROUP
Clancy and Randolph (1993) proposed an analytical-numerical
method to calculate load carried by pile group, load carried by raft Naylor and Hooper (1974) performed numerical analysis to predict
and settlement of piled-raft group. Piled-raft group was modelled the short- and long-term behaviour of piled-raft on London clay.
numerically. The interaction factors among piles were calculated Sengara (1997) performed analysis on an actual piled-raft group.
analytically. Settlement of piled-raft group can be obtained using The load-settlement of single pile was measured in the field. A
this method. numerical analysis was then used to simulate the load-settlement
curve. The simulated load-settlement curve was close to that
5. SMALL-SCALE LABORATORY MODEL TEST obtained from field measurement. Although the analysis was
performed on the load-settlement of single pile, the method of
Sengara et al. (2018) performed a small scale laboratory tests. It was obtaining interface parameters for numerical analysis was thorough.
found that the load applied to the piled-raft group was shared
between the pile group and the raft. In addition, it was observed that
9. NUMERICAL METHOD SUPPLEMENTED BY
the higher the number of piles, the lower the settlement of piled-raft
LARGE-SCALE FIELD TEST
group. This trend indicated that the presence of piles reduces the
settlement of the raft. Hussein (2018) performed a large-scale field loading test on piled-
raft group. The large-scale piled-raft group was then simulated in a
6. NUMERICAL METHOD SUPPLEMENTED BY 3-D numerical model. The numerical model was calibrated to obtain
SMALL-SCALE LABORATORY TEST close results to the large-scale model.
Long, P. D. (2011) performed laboratory tests of piled-raft small 10. DISCUSSION
scale model. Based on these result, several buildings on piled-raft
group were analysed with a commercial software. In the analysis it The comparison of the previous analysis method is shown in Table
was assumed that when a piled-raft group was loaded, the load was 1. Generally the method of analysis of piled-raft group can be
carried firstly by the piles until the ultimate bearing capacity was categorized as: (i) Analytical method; (ii) Numerical method; (iii)
reached. After the piles reach the ultimate bearing capacity, the Small-scale laboratory model test method; (iv) Large-scale field
remaining load was carried by the raft. model test or instrumentation on an actual piled-raft group. Several
methods of analysis consist of more than one method.
7. ANALYTICAL METHOD SUPPLEMENTED BY Analytical method provides a strong theoretical basis. However,
SMALL-SCALE LABORATORY TEST a closed form solution is not always available to accommodate a
complex pile group configuration. In addition, it is not easy to vary
Nguyen et al. (2013) proposed an analytical method to design piled- pile group configuration in an analysis of piled-raft group using this
raft group. The interaction factor was proposed to calculate method.
settlement and piles bending moment. The method was then
calibrated with small-scale laboratory tests and centrifuge tests.
Table 1 Comparison of the previous methods of analysis

Basis in Development of the Method
Large-scale
Field Model Test /
Method Small-scale Lab.
Analytic Numeric Instrumentation
Model Test
on Actual Piled-
Raft Group
Naylor and Hooper (1974) 
Hain and Lee (1978) 
Davis and Poulos (1972) 
Randolph (1983) 
Poulos (1991)  
Clancy and Randolph (1993)  
Lee (1993) 
Poulos (1994a) 
Poulos (1994b)
Randolph (1994) 
Ta and Small (1996) 
Long, P. D. (2011)  
Nguyen et al. (2013)  
Long, D. C. (2016) 
Hussein et al. (2018) 
Sengara et al. (2018) 
82
Considering both the technical performance of analysis and the

cost of analysis, numerical method calibrated with small-scale
laboratory tests is a promising method to analyse bearing capacity
and settlement behaviour of piled-raft group. A laboratory small-
scale test can be used to calibrate the numerical model. The
numerical analysis can then be used to analysis several pile group
configurations as well as the pile group and raft stiffness’s in a
piled-raft group. A scaling analysis (e.g. Altaee and Fellenius, 1994)
can be performed to convert the small-scale stress-strain condition
to the full-scale stress-strain condition.
A preliminary flow diagram for the analysis using numerical
method calibrated with small-scale laboratory test is shown in
Figure 1. The piled-raft group parameters and soil mechanical
parameters are first obtained. The load-settlement curve and the load
distribution along the piles are then obtained. The numerical model
is then created incorporating the piled-raft group, soil, and soil-pile
interface parameters. The numerical analysis is then performed and
the soil-pile interface parameters are then adjusted so that the
deviation between the load-transfer and load distribution along piles
between the laboratory test results and numerical analysis results are
within the deviation criteria. Once the deviation criteria are satisfied,
the soil strength and interface parameters can be used to analyse
piled-raft group model with different configurations and dimensions.
Laboratory test is one important key in the proposed method of
analysis shown in Figure 1. How far the piled-raft group
configurations and dimensions in the numerical model can be
extended from the calibrated numerical model need to be
investigated. Further research is still required related to these
matters.
11. CONCLUSIONS AND RECOMMENDATIONS

1. In general, the methods of analysis of piled-raft groups consist
of four categories: (i) Analytical method; (ii) Numerical method;
(iii) Small-scale laboratory model test; (iv) Full-scale field
model test or actual instrumentation on actual piled-raft group.
2. A method of analysis of piled-raft group can be consisted of one
category or combination from two or more categories.
3. Considering both the performance of analysis and the cost of
analysis, numerical method calibrated with small-scale
laboratory tests is a promising method to analyse bearing
capacity and settlement behaviour of piled-raft group.
4. A flow diagram for the analysis using numerical method
calibrated with small-scale laboratory test is proposed from this
study. The analysis start with obtain the piled raft group
parameters (dimension, mechanical properties, and pile load
transfer curve) and soil mechanical parameters (shear strength
and stress-strain relationship), followed by create a numerical
model. The pile interface mechanical parameters are then
adjusted to fit with the results of small-scale laboratory test.
5. Further research is still required to obtain more laboratory small-
scale data as well as to investigate how far the piled-raft group
dimensions in the numerical model can be extended from the
calibrated numerical model.
Figure 1 Preliminary flow diagram for the analysis using numerical 12. REFERENCES
method calibrated with small-scale laboratory test
Altaee, A. and Fellenius, B. H. (1994) “Physical modeling in sand”,
Canadian Geotechnical Journal, Vol. 31, No. 3, pp. 420-431.
In the numerical methods, the pile group configuration can be
Bowles, J. E. (1996) Foundation Analysis and Design, Fifth Edition,
varied and analysed easily. Thus, this method can accommodate the
McGraw Hill, New York, 1175 p.
weakness of the analytical method. However, to obtain
Clancy, P. and Randolph, M. F. (1993) "An approximate analysis
representative input parameters in the numerical analysis is not easy.
procedure for piled raft foundations", Int. Journal for
A calibration process between the numerical analysis and full-scale
Numerical and Analytical Methods in Geomechanics, Vol.
or small-scale model tests is required to obtain the representative
17, pp. 849-869, DOI: https://doi.org/10.1002/nag.
parameters.
1610171203.
Full-scale model test is the technically best among the methods
Davis, E. H. and Poulos, H. G. (1972) "The Analysis of pile-raft
considered in this paper. This method can simulate the stress-strain
systems", Australian Geomechanics Journal, Vol. G2, No. 1,
condition in the actual foundation. However, the method costs the
pp. 21-27.
most among the methods considered in this paper. The amount of
material and the instrumentations require high cost.
83
Hain, S. J. and Lee, I. K. (1978) "The analysis of flexible raft-pile Proceedings of the Seventh International Conference on Beer
systems", Géotechnique, Vol. 28, No. 1, pp. 65-83, DOI: et al. (Eds.), Balkema, Rotterdam, Vol. 1, pp. 183-191.
https://doi.org/10.1680/geot.1978.28.1.65. Poulos, H. G. (1994) "An approximate numerical analysis of pile-
Hussein H. H., Karim H. H., Shlash K. T. (2018) "Theoretical raft interaction", Int. Journal for Numerical and Analytical
verification for full-scale tests of piled raft foundation", Methods in Geomechanics, Vol. 18, Issue 2, pp. 73-92, DOI:
Advances in Analysis and Design of Deep Foundations, https://doi.org/10.1002/nag.1610180202.
GeoMEast 2017, Sustainable Civil Infrastructures Abu- Poulos, H. G. and Davis, E. H. (1980) Pile Foundation Analysis and
Farsakh M., Alshibli K., Puppala A. (Eds). Springer, Cham, Design, John Wiley & Sons, New York, 397 p.
pp. 200-212, DOI: https://doi.org/10.1007/978-3-319-61642- Randolph, M. F. (1983) Design of Piled Raft Foundations, Dept. of
1_16. Engineering Cambridge Univ, 34 p.
Lee, I. K. (1993) "Analysis and performance of raft and raft-pile Randolph, M. F. (1994) "Design and Performance of a Piled Raft
systems", Proc. 3rd Int. Conference on Case Histories in Foundation", Proceedings of the 13th International
Geotechnical Engineering, St. Louis, Missouri, pp. 1331- Conference on Soil Mechanics and Foundation Engineering,
1345. New Delhi, Vol. 5, pp. 61-82.
Long, D. C. (2016) "Prediction of piled raft foundation settlement – Sengara, I. W., Roesyanto, Krisnanto, S., Jayaputra, A. A., and
a case study", Geotechnical Engineering Journal SEAGS & Irsyam, M. (2018) "Settlement and bearing capacity study on
AGSSEA, Vol. 47, No. 1, pp. 1-6. small-scale piled-raft foundation in sand," accepted for
Long, P. D. (2011) "Piled raft - A new foundation philosophy for publication in 2018 in Geotechnical Engineering, South East
high rise buildings", Proc. 1st International Conference Asian Geotechnical Society (SEAGS) & Association of
Geotechnics for Sustainable Development - Geotec, Phung Geotechnical Societies in South East Asia (AGSSEA),
(Eds.), Construction Publishing House, Hanoi, pp. 267-276. Bangkok, Thailand.
Naylor, D. J., Hooper, J. A. (1974) "An effective stress finite Sengara, IW. (1997) "FE analysis and field load test performance of
element analysis to predict short and long-term behavior of a drilled shaft for 52-story tower foundation design", Proc.
piled-raft foundation on London clay", Proc. Conference on Third Asian Young Geotechnical Conference, Singapore, pp.
Settlements of Structures, Cambridge, pp. 394-402. 295-304.
Nguyen, D. D. C., Jo, S-B., and Kim, D. S. (2013) "Design method Ta, L. D. and Small, J. C. (1996) "Analysis of piled raft systems in
of piled-raft foundations under vertical load considering layered soils", International Journal for Numerical and
interaction effects", Computers and Geotechnics, Vol. 47, pp. Analytical Methods in Geomechanics, Vol. 20, No. 1, pp. 57-
16-27, DOI: http://dx.doi.org/10.1016/j.compgeo.2012.06. 72,DOI:https://doi.org/10.1002/(SICI)1096-9853(199601)20:
007. 1<57::AID-NAG807>3.0.CO;2-0.
Poulos, H. G. (1991) "Analysis of piled strip foundation", Tomlinson, M. J. (1994) Pile Design and Construction Practice,
Computer Methods and Advances in Geomechanics: Fourth Edition, E & F N Spon, London, 411 p.
84
Effect Of Intermediary Weak Layer On The Behaviour Of Piled Raft

V. Balakumar1, Min Huang2, Erwin Oh3 and A. S. Balasubramaniam4
1
Simplex Infrastructures Limited, Chennai, Tamil Nadu, India.
2
Arup Geotechnics, Australia.
3
Griffith School of Engineering, Griffith University, Australia.
4
Griffith School of Engineering, Griffith University, Australia.
E-mail: vb_kumar2002@yahoo.com; m.huang@outlook.com; y.oh@griffith.edu.au; a.bala@griffith.edu.au
ABSTRACT: Piled raft foundations are being used now extensively to support structures on problematic soils successfully in the last two
decades. Large numbers of milestone works have been done but all these works appear to have used only homogeneous layers of sand or
over consolidated clay. Few works have covered the behaviour of the piled raft in compressible clay but perhaps not in an extensive manner.
Naturally available deposits many times have compressible layer in between two relatively stronger layers. The present work studies the
effect of such intermediary compressible layers on the performance of the piles within a piled raft. The study has been carried out using
PLAXIS 3D and the paper presents the results of the study.
Keywords: PLAXIS 3D, compressible layer, pile group.
1. INTRODUCTION produce an optimum foundation design satisfying both the safety

and the serviceability requirements.
The main requirement in the design of a foundation for any structure
is that it has to satisfy the defined requirements of bearing capacity
2. NEED FOR THE STUDY
and the settlement both total and differential. Practically most of the
international codes of practice quantitatively recommend the The awareness of settlement based design has increased among the
permissible settlement for all types of structures based on the designers particularly after the vast improvement that has taken
performance requirement or what is known as serviceability place in the field of computational tools like FEA and the associated
requirements. The first option for the designers to satisfy the above software. The present day designers have started accepting the fact
requirements had always been raft; this option may satisfy the that settlement based design can lead to considerable economy
bearing capacity requirements in most of the cases but in the case of without sacrificing the safety and serviceability of the structure.
structures sensitive for settlement raft may not become an ideal Addition of piles below the raft and connected to it has proved itself
solution. The option under such cases would be to go for deep piles to be a successful and a viable solution in bringing down the
which are designed assuming that the entire structural load will be settlement level very close to the permissible settlement value. In
taken by the piles and the presence of raft and its capability to addition the system has a higher bearing capacity that the unpiled
transfer the load to the ground had been ignored. While such a raft at any given settlement level. Further considerable economy has
design satisfies all the safety requirements, it may not satisfy the also been achieved compared to the traditional fully piled system.
economic requirements. Further such a design contradicts the The combined foundation system addressed as piled raft has been
behaviour predicted when the raft is seated on competent ground successfully used in supporting a number of tall and super tall
(Balakumar and Anirudhan, 2010). It has been established that in the structures in the last two decades.
case of large pile groups with smaller dia piles with closer spacing However the combined piled raft foundation system has been
covered by a raft which would be seated on a competent ground the viewed with suspicion mainly because of the following reasons:
percentage of the load shared by the raft would be of the order of 1. The main problem the designers were facing was that many
20% or even more (Mandolini et al.,2017 ; Balakumar and traditional methods of analyses could not be applied since they
Anirudhan, 2010) . It is evident that ignoring the presence of the raft require a high level of extrapolation and approximation which
and its contribution in transferring the load to the competent ground were far beyond the comprehension of past experience.
cannot be justified from engineering principles. When the ground 2. The behaviour of piled raft had been studied by most of the
has adequate bearing capacity and settlement alone is a problem in researchers adopting homogeneous over-consolidated clay or
providing a large group of piles, the number of piles is governed by loose to medium dense sand. In practice such a homogeneous
the geometry of the foundation leading to an uneconomical design profile does not exist
with a very high factor of safety not justifiable from an engineering 3. It is strongly believed that presence of compressible strata at near
point of view to reduce the settlement. raft base (need not be below the raft) is not a favourable
The concept of introducing deep foundation elements namely condition for piled raft.
piles with the raft in a strategic manner as settlement reducers to The combined piled raft foundation system is a three
reduce raft settlement was developed by Burland etal., (1977) and dimensional interaction problem. Such a problem requires a detailed
subsequently many researchers have studied the behaviour of this three dimensional analyses. As Russo has pointed out, to move from
combined foundation system comprising of raft, piles and the soil the traditional capacity based design to settlement based design the
namely the combined piled raft foundation system adopting various method of analyses must be capable of taking into account properly
methods like analytical modelling (Cooke etal., 1981; Clancy and the soil structure interaction with in the foundation systems that is
Randolph,1993; Poulos etal.,1997), small scale model studies (Kim needed. Further the accuracy of analyses depends upon the accuracy
etal.2002; Weisner and Brown,1978, Balakumar and with which the in-situ geotechnical parameters are evaluated and the
Ilamparuthi,2004) and by monitoring the real-time piled raft skill in the use of software.
supporting tall structures. The piled raft foundation system is an A complex problem like a piled raft can become further
intelligent geotechnical concept which uses the deep foundation complicated due to the presence of a clay layer at an intermediary
elements placed strategically to enhance the overall stiffness of the level. The complexity of interaction among the various constituent
raft to carry a much higher load at any settlement level compared to elements has made it necessary to adopt a detailed three dimensional
the un-piled raft at the same settlement. In the recent past such a analysis. But the presence of intermediary compressible deposits can
development has enabled the designers to shift their design approach influence the pile soil pile interaction process which will influence
from the bearing capacity basis to a settlement based design process. the load sharing behaviour of the pile group. The influence of the
Such a change has resulted in using the piles in a strategic manner to intermediary compressible layer can also depend upon its location
85
below the raft The additional complexity generated by the presence layer is stiff clay followed by weathered rock in gravel form Here
of a clay layer at lower levels is the reason for the present study. the N –value is over 50. Pile load tests data are very scarce in
Surfers Paradise. The Authors have done the analysis and will
3. PREVIOUS WORKS compare with available pile load tests data in the second phase.
Based on all the milestone data number of structures have been
successfully supported on piled raft and also have been monitored.
In most of the cases the strata has been considered as over
consolidated clay or medium dense sand. The study on the
behaviour of piled raft on layered soil appears to be very limited
(Chow and Small,2008 ; Yamashita etal., 2010) had used the piled
raft with ground improvement and also supporting of tall structures
supported on soft ground etc has been discussed and published
results are available. However in reality there are number of
instances wherein piles may have to pass through intermediary
compressible layers as seen in the case of many sites in Brisbane
Gold Coast areas (Oh etal, 2008; Moyes etal.,2006, Min.J.Huang
2006). In such cases many structures have been supported on a piled
raft but appears not to have been monitored Therefore It is felt that a
Figure 1 Typical Soil Profile in Surfers Paradise,
more detailed study needs to be done on the effect of such an
Gold Coast
intermediary compressible layer sandwiched between denser layers,
on the behaviour of piled raft. A detailed study was planned and is
being carried out by the above research group of Griffith University RL (m) BH3 BH4
adopting analytical modelling and this paper presents the results of
part of the study. Further studies are going on. ±0.00
Layer 1: Medium Dense Sand
4. METHOD OF ANALYSIS
Complex three dimensional problems like a piled raft cannot be
Layer 2: Very Dense Sand
analysed by common traditional methods mainly because they

Layer 3: Stiff Peat
require a very high level of extrapolation and approximation which

are far beyond the comprehension of past experience The analysis of Layer 4: Very Dense Sand
the interaction among the constituent elements becomes favorable in
the case of sand and medium stiff to stiff clay.But the presence of
soft clay like peat can take away this advantage. The role of
analyses in the design process becomes clear only when the design Layer 5: Stiff Clay
objectives are clearly established. The facets of analyses such as

identification of appropriate parameters and a clear understanding of
empirical methods play a very important role. The essentials of
ROCK
analyses have to take into account the soil - structure interaction Scale 1:400
0 3 6 9
within the foundation. Keeping the above in mind, for the present *Layer 6: Clayey Sand and Layer 7: Gravel Sand are missing in this site
study, analytical method the finite element method and the software
Figure 2 Artique Project- Soil Profile along Section
PLAXIS 3D have been used in the rigorous analysis.
(Huang, 2006)
4.1 Basis of Model Definition Depth (m) BH1 BH2
The selection of model was done based on the problem requirement. WT.L
Layer 1: Medium Dense Sand
Since the present study attempts to find the behaviour of the piled
raft with an added complexity, the model has to be computationally
simple and at the same time must be able to produce the desired Layer 2: Very Dense Sand
results with minimum computational time .Hence a simple square
model has been assumed for the study. The model is briefly
discussed in section 4.3
Layer 5: Stiff Clay
4.2 Geotechnical Model Layer 6: Clayey Sand
The geotechnical model is based on the profiles presented by Layer 7: Gravel
various authors (Oh etal, 2008; Moyes etal.,2006, Min.J.Huang

2006). The Soil profile in Surfers Paradise Gold Coast is carpetted ROCK
with a layer of sand extending up to 13 m or more and is an ideal 0 2 4 8
place to use piled raft foundation. Larich of PTY limited has Scale1:100
*Layer 3: peat is missing

discussed about the soul project in surfer’s paradise QLD, Australia.
This profile did not have peat in between. Table 1 presents the Figure 3 Q.1 Tower- Soil Profile (Huang, 2006)
geotechnical model that has been used in the analysis. It is seen that,
in this particular site, peat layer is absent indicating that the presence 4.3 Piled Raft Model
of peat layer varies from site to site. A typical profile is presented in
Figures 1 to 3The geotechnical profile used for this analysis is given Figure 5 presents the model. The model comprises of 6 X 6 pile
in Figure 4. The profile comprises of medium dense to dense sand group with 4D spacing (D is the diameter of the pile) having an area
with N values range from 10 to 40.This layer exists from ground ratio of 6.1%. The area ratio is defined as the ratio between the total
level to 12.5m to 13m. Below this layer is the problematic peat layer cross sectional area of the piles provided to the plan area of the raft.
which is compressible and its N-value ranges from 2to 5.Below this
86
It has been observed that the piled raft with an area ratio of around settlement of the piled raft reached 35mm in the case considering
6% produces the best performance (Balakumar, 2008). the intermediary compressible layer.
Table 1 Geotechnical Model
Figure 6(a) FEA model
Figure 4 Geotechnical Parameters
The raft is 1m thick and the piles are 750 mm dia and 28 m long Figure 6(b) Deformation contour
namely 1.5 times the raft width. The length of the pile has been
taken on 28 m so that there will be sufficient length of pile available 5. LOAD SETTLEMENT RESPONSE
below the compressible layer. The piles are fully fixed to the base of In general in the case of homogeneous layer the applied pressure
the raft. Figure 6 a presents the finite element model and 6b the from the raft enhances the confining pressure below the pile group
deformation contour. The various layers have been identified by the by the inter -granular friction. This enables the pile group to take a
respective Es, density and Poisson’s ratio etc have been derived higher load by friction and once the friction is overcome the system
from the correlations given by Poulos (1988). The parameters are loses its stiffness gradually and then rapidly.(Balakumar , 2008) as
presented in Table 1.No mesh refinement was needed. The piles shown in Figure 7.
have been modelled as beam elements.Four typical piles have been
identified for the study. They are central pile (Pile 1), outer pile in
8MM RAFT 10MM DIA. PILE - 6D SPACING - LOOSE SAND
LOAD, kN
line with the central pile on the X-axis (pile 2),corner pile as pile 3 0 1 2 3 4 5 6 7 8 9 10
and the peripheral pil below the central pile on Y-axis downwards.
0
d = 10mm
4 t = 8mm
S=6d
SETTLEMENT
B = 200 mm
8
SETTLEMENT, MM
NT, mm
12
16
20 Loose
Med dense Bed density
Dense
24
Figure 7 Load Settlement Response of Piled Raft

on Homogeneous Layer
Figure 5 Piled raft model
In order to understand the effect of intermediary compressible
4.4 Loading layer on the behaviour of piled raft, the load settlement response of
piled raft on layered soil with intermediary peat layer is compared
In the present study the loading is restricted to vertical loading only. with the load settlement response of the piled raft on a homogeneous
The loading was applied in the form of pressure load till the layer. The load settlement response of piled raft on a homogeneous
settlement reaches nearly 2% of the raft dimensions. The final layer is taken from 1g tests conducted on a square piled raft ( with a
loading was 645 kN /m2. The loading was applied in steps till the pile- raft area ratio of total cross sectional area to the area of the raft)
87
of 5% which is close to the area ratio of the model studied now thereafter the fall is rapid. But it can be seen that the behaviour is
namely 6% The details of the 1g model tests on square piled raft more uniform and the ratio of axial stress to tip stress works out to
with 4D pile spacing on a poorly graded sand bed is discussed in 20%.
detail elsewhere (Balakumar and Ilamparuthi,2005). It is assumed
that this small variation may not affect the comparison of load
settlement response. Figure 6 presents the load settlement response SQUARE RAFT - LENGTH RATIO Vs SHAFT STRESS (20mm)
Shaft stress N/mm2
of piled raft on a homogeneous layer and Figure 8 presents the load 0.00 0.40 0.80 1.20 1.60 2.00 2.40 2.80
settlement responseof piled raft of the present study. of piled raft of 0.00
the present study.
0.20
Length 3
ratio 0.40
0.60
0.80 Central pile

Inner pile
1.00
Outer pile
1.20
Figure 9 Axial Stress Distribution in the Pile

without Peat Layer
Figure 8 Load Settlement Response of Piled Raft
It can be seen that in the case of piled raft on a homogeneous

bed the response is well defined. In this case, till the load reaches
25% of the final load the stiffness exhibited is very high and then as
the load increases the rate of reduction rapidly increases. But in the
case of the layered soil with peat almost upto 40% of the applied
load the stiffness of the piled raft is high and when the load further
increases the rate of reduction of the stiffness increases. When the
load level increases beyond 60 to 65% the rate of reduction of
stiffness is rapid. Although the behaviour is not as well defined as in
the case of homogeneous layer, the response is similar and the effect
of the compressible layer gets pronounced only after the load
reaches beyond 60% of the maximum load applied. The top sand Figure 10 Axial Stress Distribution in the Pile
layer whose state of compaction increases with depth offers with Peat Layer
adequate frictional resistance preventing the additional stresses
generated by the compressible clay affecting the load settlement
response. This amounts to the fact that when the upper layers are
competent the compressible layer does not appreciably influence the
load settlement response. It can also be seen that at the final load the
piled raft with peat layer settles more by 20.5% at the centre, 23% in
the periphery and in the corner the percentage variation is 24.6%.
Although the variation appears to be small it indicates that there will
be an increased differential settlement when there is an intermediary
compressible layer.
6. AXIAL LOAD DISTRIBUTION IN THE PILE

The axial stress distribution which is an important parameter in the
design of piles from the capacity point of view, and together with
the distribution of bending moment will influence the structural
design provisions. These two aspects are studied in comparison with Figure 11 Axial Stress Distribution
the data obtained by Min Huang(2006).The load sharing behaviour
of a typical square piled raft obtained through the 1g model tests is Comparing the above behaviour with axial stress distribution in
also considered. Figure 9 and Figure 10 presents a typical axial the piles passing through the compressible layer the axial stress
stress distribution in the pile without a compressible layer and a distribution exhibits a non-uniform reduction with the depth as in
typical distribution considering the compressible layer. In the case the Figure 11. There are three stages of stress variation. The
of the pile group passing through a homogeneous layer the shaft variation between top to 13m is treated as stage 1, from 13m to
stress distribution indicates a gradual reduction in the stress upto around 20m as stage 2 and 20m to 28 m as stage 3. The shaft stress
0.6L and then a rapid reduction beyond 0,8L, where the axial stress is evaluated in four piles of the group namely central pile, corner
reduces to a negligible level The ratio of the tip stress to head stress pile, one peripheral pile on the X –axis, and one peripheral pile on
is of the order of 20% confirming that maximum part of the load is the Y axis. The reduction in the shaft stress reduces gradually. But
transferred by friction. Min J. Huang,s analyses (Balakumar the rate of reduction reduces from 12.5 to 13 m level from the top
etal.,2008) also presents the similar behaviour as seen in the Figure and extends upto 23m (from 0.47L upto 0.8L)Beyond 0.8L as in the
11. Upto a depth of 0.8L the reduction in axial stress is gradual and previous cases the rate of the drop is more rapid. But the ratio of the
88
tip stress to the head stress to works out to 45% to 53% the pile head , stress at 0.5 L – stress at 0.25L/ stress at 0.25L and so on)
maximum being at the central pile and the minimum being at the is plotted against length ratio to study the effect of layer thickness.
corner pile. The reduction ratio works out to 85 % upto 13 m level; Similarly to study the effect of layer consistency the analyses was
from 13m to 22.5m the reduction in stress with reference to the head repeated for three different N-values namely N=4, N=8 and N=12.
stress is 71% which reduces to around 45%. The trend remains same In the same manner as said earlier the stress reduction ratios are
although the magnitude may vary. This indicates that the frictional tabulated and for one typical pile a plot is made between the stress
resistance of the weaker layer is enhanced by the stronger layers reduction ratio and length ratio for discussion.
above and below the compressible layer.
6.1 Discussion
In the case of load transfer mechanism there are three types of
interaction namely raft - soil soil-pile, pile –pile and raft- pile,and
raft -raft When there is an intermediary compressible layer there
appears to be an interaction between the upper layer (medium dense
to dense sand) and the lower layer namely stiff clay which causes
perhaps an enhancement of the frictional resistance of the weaker
layer in between. There is always a confusion particularly when the
soft layer is found below a denser layer in computing the negative
friction in the sense that when the compressible layer reference to
the pile the upper layer also settles down and hence the upper sand
layer also can contribute for additional drag load. This behaviour is
very evident from the shaft stress distribution from the rate of
reduction in the axial stress with depth. Although the pile group in a
piled raft is intended to be primarily a settlement reducer the pile
design has to recognize this excess load. It is also to be noted here
that the ratio of tip stress to head stress varies from 40% to 45% Figure 12 Bending Moment Distribution with compressible Layer
caused by perhaps the increased frictional resistance of the
compressible layer due to the stronger upper and lower sand layers. Bending Moment (kNm/m) Bending Moment (kNm/m)
-180 -120 -60 0 60 120 180 -100 -50 0 50 100
7. BENDING MOMENT BEHAVIOUR

0 0
2 2
In general the piles are always designed for combined axial load and
bending moment due to the lateral loads. The bending moment due 4 4
to axial load is very small, but when the pile passes through 6 6
Depth (m)
compressible layer this section is subjected to higher bending

Depth (m)
Pile 1 Pile 2
moment, which is not so in the case of pile group passing through
8 8
homogeneous layer. Care has to be taken to consider the bending 10 10
moment value at the section passing through compressible layer. It 12 t=0.25m
is seen from the Figure 12 that the bending moment exhibits both
12 t=0.25m
t=0.4m
t=0.4m
positive and negative bending moment unlike in the of pile group

t=0.8m
14 t=0.8m
t=1.5m 14
t=1.5m
through uniform homogeneous layer as presented by Min. Huang

t=3.0m t=3.0m
16
16
(2006), and shown in Figure 13.This variation in bending moment is Figure 13 Bending Moment Distribution without compressible
mainly due the differential settlement as seen in the settlement Layer
profile. Further the value of bending moments in the pile sections is
far higher and hence the piled raft design with pile group passing 8.1 Variation of Layer Thickness
through compressible layer has to be analysed for all the conditions
of loading so that the pile element can be designed for axial stress Table 2 presents the variation of stress reduction ratio α with pile
and the bending moment. Once the peat layer is crossed the length ratio for the various thickness of the compressible layers
structural provisions can be reduced to the extent of the axial force namely t = 2.5m , t=5m and t=8m.The table presents the variation
requirement. for four different selected piles.
The maximum positive bending moment as can be seen at 14m
Table 2 Variation in the Shaft Stress reduction ratio( %) at various
is of the order of 200 kNm which is not seen in Figure 8.Keeping
layer thickness
the axial stress distribution and bending moment together it can be
said that as pile is subjected to a higher bending moment in the
section inside the compressible layer. Hence the structural provision
can become higher than the conventional requirement in particular
the corner and the peripheral piles. In such cases it appears that
higher diameter pile may have to be used.
8. PARAMETRIC STUDY
In the present study, the influence of two important factors relating
to the compressible layer, namely the thickness and the consistency To discuss the results, the variation of the stress reduction ratio
on the behavior of the pile group of piled raft need to be considered. to the length ratio is presented for the central pile in Figure 14.
The effect of the layer thickness was studied by repeating the
analyses for three different thicknesses of the layers namely 2.5m,
5m, and 8m. The shaft stress distribution over the length of the piles
is plotted as typically shown in Figure 10. The stress reduction ratio
α which is the ratio of the variation in the stress between two
successive points ( eg : stress at pile head – stress at 0.25L / stress at
89
It can be seen that the variation in the shaft stress reduction ratio is
gradual and uniform indicating that N- Value has no significant
influence the shaft stress reduction and the variation is as per the
general behavior. In other words the shaft stress reduces over the
length uniformly unlike in the case of variation in the layer
thickness.
9. CONCLUSION
In order to study the applicability of piled raft under any practical
geotechnical conditions the effect of intermediary peat or
Figure 14 variation of the stress reduction ratio to the length ratio compressible layer was studied. Such profiles are widely seen in the
Gold Coast area as furnished by various authors mentioned in Table.
It can be seen that the rate of increase in the stress reduction Accordingly the study was carried out numerically using PLAXIS 3-
ratio is rapid as the length ratio of the pile increases when the layer D considering a general soil profile. A parametric study was also
thickness is 2.5m. As the thickness increases, the rate of increase in conducted by varying the thickness of the compressible layer and its
the stress reduction ratio reduces between the length ratio 0.5 and consistency. The conclusions are as follows:
0.75 and then it increases. This indicates that the compressible layer 1. The presence of peat layer induces a higher differential
present in this region settles, generating the negative friction which settlement. The settlement varies by 20 to 25% from the centre to
causes an increase in the shaft stress locally. In all the other piles the edge.
also the same trend exists. It can be said that as the thickness of the 2. From the study of axial stress and bending moment variation in
layer increases, the compressible layer contributes for the negative the vicinity of the compressible layer, the fall in the axial stress
skin friction locally, increasing the shaft stress at that section. distribution is influenced by the drag force developed by the peat
layer. The drag force adds an additional load on the pile
8.2 Effect of Variation in the State of Compaction irrespective of its location. In other words the piles are subjected
to additional axial load in the section of the pile passing through
In the same manner as in the previous case, the analyses was the compressible layer.
repeated for three different cases of N-Values namely N=4 ( soft), 3. It is seen from the ratio of tip stress to head stress; the ductile
N=8 ( medium stiff) N=12 (stiff) representing three different behaviour of the pile group is affected by the presence of the peat
consistencies of the compressible layer. The rate of variation of the layer.
shaft stress reduction ratio α is presented in the cases of four 4. From the parametric study it is seen that the thickness of the
selected piles in Table 3. compressible layer has a higher influence on the behaviour than
Table 3 The rate of variation of the shaft stress reduction the state of compaction.
ratio α(%) 10. REFERENCES

Anirudhan.I.V. and Balakumar. V (2010), ‘Pile Foundation as
L 0.2 L 0.5 L 0.75 L 1L
Settlement Reducer for Large MS Storage Tanks,’ Indian
Geotechnical Conference, 2010 – Bombay, India, pp.
1 6.2 16.7 26.6 51.7
Balakumar.V. (2008) Experimental Studies on Model Piled raft on
2 7.2 18.1 27.7 55.3 Sandand Field Study of Proto Type Behaviour- Ph.D Theses,
N=4
3 7.4 21.2 31.6 63.5 Anna University , Chennai.
4 6.7 18.14 30.1 55.2 Balakumr V. and Ilamparuthi K. (2004), ‘Laboratory Study on the
1 6.22 17.8 28.5 53.5 Behaviour of Piled Raft on Granular Soils,’ Proc. 15th South
2 7.5 19.4 29.8 57.3 East Asian Geotechnical SocietyConference 2004, Bangkok,
N=8 Thailand, pp. 293-298.
3 7.5 22.9 34.1 66.1
Balakumar. V and I.V. Anirudhan (2011), ‘Piled Raft Behaviour –
4 6.8 19 31.8 57.3 Model Studies and Field Performance,’ Indian Geotechnical
1 6.3 18.2 29.5 54.4 Conference, 2011 – Kochi, India, pp.
32.3 Balakumar. V and I.V. Anirudhan (2012), ‘Piled Raft Behaviour –
2 7.5 20.5 31.6
N = 12 (59.1) Generalization for Design,’ Indian Geotechnical Conference,
3 7.5 23.6 35.8 67.9 2012 – New Delhi, India, pp.
Balakumar V. Kalaiarasi V. and Ilamparuthi K. (2005)
4 6.9 19.6 33.16 38.46
“Experimental and Analytical Study on The Behaviour of
Circular Piled Raft on Sand”, Proc. 16th International
For the sake of discussion and better understanding, the rate of Conference on Soil Mechanics and Geotechnical
variation of the shaft stress reduction ratio α (%) is plotted against Engineering-2005, Osaka, Japan.
the length ratio and is presented in Figure 15 Balakumar V. Ilamparuthi K.and Kalaiarasi V. (2005), ‘Study on
Square Piled Rafton Granular Soil, ‘Proc. Indian
Geotechnical Conference 2005. Ahmedabad, India.
Burland J.B., Broms B.B. and de Mello V.F.B. (1977), ‘Behaviour
of Foundations and Structures’, Proc. 9 ICSMFE Tokyo 2,
pp. 495 – 546.
Cooke R.W., Bryden-Smith D.W., Gooch M.N. and Sillett D.F.
(1981), ‘Some Observations of the Foundation Loading and
Settlement of a Multi- story Building on a Piled Raft
Foundation in London Clay’, proc. Institution of Civil
Engineers, Part 1, Vol. 70, pp. 433-460.
Figure 15 variation of the shaft stress reduction ratio to length ratio
90
Clancy P. and Randolph M.F. (1993), ‘Simple Design Tests for Poulos H.G., Small J.C. Ta L.D., Sinha J. and Chen L. (1997),
Piled Raft Foundations’, Geotechnique, Vol. 36, No. 2, pp. ‘Comparison of Some Methods for Analysis of Piled Rafts’,
169-203. Proc. 14th Int. Conf. Soil Mech. Foundn. Engg- Hamburg – 2,
E.Y.N OH, M. Huang, C. Surarak, R. Adamec and A.S. pp. 1119-1124.
Balasubramaniam (2008), “Finite Element Modeling for Piled Poulos H.G -( 1998) The Pile - Enhanced Raft – An Economical
Raft Foundation in Sand”, Eleventh East Asia – Pacific Foundation System – Keynote Lecture 11TH Brazilian
Conference on Structural Engineering & Construction conference on soil mechanics and F oundation Engineering.
(EASEC – 11). Russo.G.(1998) Numerical Analyses of Piled Rafts,
Kim H.T., Yoo H.K. and Kang I.K. (2002), ‘Genetic Algorithm Intl.Jnl.Num.and Anl.Methods in GeoMech,Vol 22,pp.477-
Optimum Design of Piled Raft Foundations with Model 493.
Tests’, Journal of South East Asian Geotechnical society, pp. Weisner T.J. and Brown P.T. (1978), ‘Laboratory Tests on Model
1-9. Piled Raft Foundations’, Research Report 318, Sydney
Paran Moyes ; Harry G Poulos ; John C. Small Frances Badelow: University.
Piled raft Design Process for a High-rise Building on the Yamashita, K, Hamada, J., and Yamada, T. (2010),
Gold Coast, Australia. FieldMeasurements On Piled Rafts with Grid-FormDeep
Min.J.Huang -2006. Mixing Walls on Soft Ground, Geotechnical Engineering-
Min Huang. M (2006) Rafts and Piled Raft foundations at Surfers SEAGS, Vol. 42, No. 2, June, 2011.
Paradise Gold Coast, Australia – Analytical study using
PLAXIS Software – A theses submitted in Partial Fulfilment
partial requirement of the degree of Master of Civil
Engineering from Griffith University Gold Coast campus
Australia
91
Analysis of In Situ Laterally Loaded Tests on Caisson Foundations

Jiunn-Shyang Chiou1 and Cheng-Chang Tsai1
1
Department of Civil Engineering, National Taiwan University, Taipei, Taiwan
E-mail: jschiou@ntu.edu.tw
ABSTRACT: In this study, three cases of in-situ laterally loaded tests of caisson foundations are simulated. The differences of these three
cases are the application position of lateral loads and the scoured depth of the foundations. In the numerical model, a Winkler beam model is
adopted. The Winkler beam model utilizes the beam element to model the caisson and six types of springs to simulate the horizontal
subgrade reaction, the vertical shear stress on the front of the caisson, the horizontal shear stress and the vertical shear stress on the sides of
the caisson, and the normal stress and the horizontal shear stress on the base of the caisson. The simulation results show that the model can
properly capture the responses of the caisson foundations under lateral loading.
Keywords: Caisson foundations, In situ tests, Lateral loads, Winkler model.
1. INTRODUCTION 3. SOIL CONDITIONS

Caisson foundations are a commonly used foundation type of bridge The soil at the site, within a depth of 20 m, mainly consists of
structures. They generally have a large resistance due to a large gravels and cobbles with some sands or silts. The Standard
cross section and a large embedded depth. Compared to vertical Penetration Test (SPT) blow counts were generally larger than 50.
loading, lateral loading often governs final design due to a relatively The water level was about 3 m below the soil surface during the
low lateral capacity. In-situ load lateral tests on caisson foundations lateral load tests. The specific gravity of the soil solids was 2.75.
are very limited although the actual bearing behavior of foundations The field density tests and the sieve analysis indicated that the moist
is essential to verify the appropriateness analysis models in design. unit weight, the water content, and the void ratio of the soil were
In 2010, National Center for Research on Earthquake 22.66 kN/m3, 10%, and 0.31, respectively. The grain-size
Engineering (NCREE) of Taiwan conducted a series of in-situ distribution of the soil indicated the gravel content of about 89%, an
loading tests on the old Niu-Dou Bridge in Ilan Country, Taiwan. effective size D10 of 2 mm, a coefficient of uniformity Cu of 57.5,
The differences of these cases were the applied position of lateral and a coefficient of curvature Cc of 1.5. The soil is classified as
loads and the scoured depth of the foundations. For these tests, well-graded gravel (GW). According to field direct shear test, the
Chiou et al. (2012) used a Winkler beam model to simulate the friction angle and the cohesion for the peak shear strength were 37°
foundation load test. In their study, they proposed a method to
and 9.81 kN/m2, respectively, and the friction angle and the
determine the properties of Winkler springs for gravel ground.
Chang et al. (2014) used nonlinear fiber elements for the nonlinear cohesion for the residual shear strength were 31.8° and 0 kN/m2,
flexural behavior of columns to simulate the column load tests. respectively.
In order to investigate the foundation responses in the column
load tests in more detail, this study conducts numerical simulation to
analyze these tests.
2. OVERVIEW OF TESTS
The old Niu-Dou Bridge had two independent bridge structures, as
shown in Figure 1. Both bridge structures had seven spans. In the
test program, columns P2, P3, P4, and P5 of the left-side bridge
were tested, as displayed in Figure 2. The columns were supported
by caisson foundations. The foundations generally had a diameter of
4m and a length of 12m. The three column load tests were
conducted on P2, P3, and P4 and the foundation load test on P5. The
location of load application on P2, P3, and P4 were 9.54, 10.59, and
10.37 m, respectively, measured from the foundation top. The
column sections had a diameter of 1.8 m. These columns had Figure 1 Niu-Dou Bridge (Chiou et al., 2012)
different foundation embedment. The foundation of column P2 was
256.2 m
slightly exposed with a length of about 0.4 m. The P3 foundation
was exposed with a length of 1.2 m. The P4 foundation had a larger 36.6 m 36.6 m 36.6 m 36.6 m 36.6 m 36.6 m 36.6 m
exposed length of 4.0 m. The P5 foundation had an exposed length
of about 1 m.
For the column load tests, cyclic displacement controlled load
tests were performed on columns P3 and P4. A pseudo-dynamic test
and a single-cycle pushover test were conducted orderly on column
P2. A load-controlled monotonic lateral load test was carried out on
the caisson foundation of column P5.
12 m
P1 P2 P3 P4 P5 P6
Figure 2 Test specimens of Niu-Dou Bridge
92
4. THE ANALYSIS MODEL AND SIMULATION slightly damaged due to the former pseudo-dynamic test on it.
Figure 7 shows the comparison between the simulation curve and
This section further analyzes the column tests through numerical the pseudo-dynamic test: the initial portion of the simulated curve is
simulation to clarify the behavior of the foundations in the tests in very close to that of the pseudo-dynamic test. In Figure 6, the trend
more detail. of the simulated foundation displacement is consistent to that of the
4.1 Analysis model measured displacement: the simulation curve is a little softer than
that of the experimental one.
Chiou et al. (2012) applied a Winkler beam model to simulate the
lateral response of P5 foundation under loading. In this study, their
model was used to simulate the foundation behavior in the column
load tests. The Winkler-beam model adopted uses beam elements to
simulate the caisson body and spring elements to simulate the soil
reactions. To well simulate different soil reactions around the
caisson, a six-component Winkler spring model, as shown in Figure
5, is adopted. Six types of springs are utilized to simulate the
different components of the soil reactions acting on the caisson with
an equivalent rectangular section of width Be (perpendicular to the
direction of lateral loading) and length Le (parallel to the direction of
lateral loading). Springs kH and kSVB represent the horizontal
subgrade reactions and vertical shear stress levels on the front of the
caisson, respectively; springs kSHL and kSVL represent the horizontal
shear stress and the vertical shear stress on the sides of the caisson,
respectively; springs kV and kS represent the normal shear stress and
the horizontal shear stress on the base of the caisson, respectively. Figure 4 Analysis models for the column load tests
Since the test site was on gravelly soil, Chiou et al. (2012) proposed
a method which modified the load-displacement responses from the
plate loading tests and the direct shear tests to determine the load-
deformation characteristics of the springs. The details for the
method can be found in Chiou et al. (2012).
Beam elements are used to model the columns. In order to
simulate nonlinear flexural behavior of the columns, the distributed
plastic hinge model is applied. According to Chiou et al., (2009), the
properties of plastic hinges are set based on the moment-curvature
curves of the columns. Based on the above settings, the analysis
models for the column load tests using SAP2000 program
(Computer & Structures, 2002) are displayed in Figure 4.
Figure 5 Load-displacement curve at the top of pier of P2

Be
Le
V0
M0
H0
kSVL1
kSVB1 kSHL1 kSVB1
kH1
kSVL2
kSHL2
kSVB2 k SVB2
kH2
kSVL3
kSVB3 kSHL3 kSVB3
kH3
Figure 6 Load-displacement curve at the top of foundation of P2

kS
kV
Figure 3 Winkler spring model for soil reactions on caisson

[adapted from Japanese Specifications of Highway Bridges (JRA
2012)]
4.2 Analysis results of P2 column load test

In this test, the exposed depth of the foundation was about 0.4 m
(3.3% caisson length), and the point of load application on the
column was 9.54 m above the ground surface.
The results of analysis are shown in Figs. 5 and 6 for the lateral
displacement of the column and the foundation, respectively. In
Figure 5, the stiffness of the simulation curve is a little stiffer than
that of the experimental one. This is because the column was Figure 7 Load-displacement curve at the top of pier of P2
93
4.3 Analysis of P3 column load test

In P3 test, the point of load application on the column was 10.59 m
above the ground surface. The foundation embedment condition was
similar to P5 foundation (exposed length of 1.2 m (10% caisson
length)). Therefore, the foundation analysis model used for this test
is close to that for P5 test.
Figure 8 displays the analysis result of the lateral displacement
of the column. Compared with the experimental curve, they are in
good agreement. Figure 9 only displays the simulated load-
foundation displacement curve because of some measurement errors
on foundation displacement.
Figure 10 Load-disp. curve at the top of pier of P4
Figure 8 Load-displacement curve at the top of pier of P3
4.4 Analysis of P4 column load test Figure 12 Simulated load-displacement curves of columns P2, P3,
and P4
In P4 test, the exposed depth of the caisson was 4 m (1/3 length of
caisson). The point of load application on the column was 10.37 m
above the foundation top.
The analysis results of P4 test for the lateral displacement of the
column and the foundation are shown in Figs. 10 and 11,
respectively. In Figure 10, the simulation curve is close to the
experimental one. In Figure 11, the trend of the simulated
foundation displacement is consistent to that of the measured.
From the above simulations, it is seen the model adopted can
reasonably capture the response of the behavior of the columns and
their foundations in the tests.
With the analysis results, Figure 12 compares the lateral load-
displacement curves of columns P2, P3, and P4. As observed in the
test results, the overall curves are close. Their stiffnesses are a little
different, but their strengths are close. However, as shown in Figure Figure 13 Simulated load-foundation displacement curves of P2, P3,
13, their foundations have different responses. It can be seen in the and P4
figure that at the same lateral load P4 foundation has the largest
lateral displacement while P2 and P3 foundations have close lateral 5. CONTRIBUTIONS OF SOIL REACTIONS
displacements. P4 foundation has the lowest lateral stiffness and
In addition to the horizontal soil reaction in front of foundation kH,
strength because of the largest exposed length of foundation. As
the shear reactions around the caisson (kSVB, kSHL and kSVL) and the
those curves are compared with that of P5 foundation, the lateral
bottom resistance also contribute lateral resistance (kS and kV).
stiffness is the largest because of a pure horizontal load on the top of
Figure 14 (a) and (b) compares the horizontal resistance contribution
the caisson and a smaller exposed length.
of P2 and P4. The resistance of the caisson to the horizontal loading
mainly comes from the horizontal resistance of the soil in front of
the caisson. The horizontal side shear resistance of the caisson
94
contributes part of the resistance; however, its contribution reaches a 6. CONCLUSIONS

maximum at a small displacement. The shear resistance on the base
of the caisson acts in the reverse direction to the lateral loading due In this study, the six-component foundation model is used for the
to the different directions of movement between the upper and the foundation behavior and the distributed plastic hinge model for the
lower parts of the caisson. It is noted that for P4 foundation, with nonlinear flexural pier behavior. The behavior of the foundations
large exposed length, the lateral resistance in front of the caisson and pier in the column load tests with different degrees of
even exceeds the total applied lateral load to cancel out the reverse foundation exposure are reasonably captured.
base shear. The lateral responses of the foundations are influenced by the
foundation embedment and the location of load application. P4
foundation has the lowest lateral stiffness and strength because of
the largest exposed length of foundation. The lateral stiffness of P5
foundation is the largest because of a pure horizontal load on the top
of the caisson and a smaller exposed length.
(a)
(a)
Figure 14 Soil reaction contributions to horizontal loading at

different lateral displacements: (a) P2, (b) P4
For the moment resistance, with respect to the rotation center,

Figure 15 (a) and (b) compares the moment resistance contributions Figure 15 Soil reaction contributions to moment at different lateral
of P2 and P4 foundations, respectively. All the soil reactions around displacements: (a) P2, (b) P4
the caisson contribute to resist the moment loading. The soil
reactions from the front of caisson provide the maximum moment 7. REFERENCES
resistance, about 50-60% for both P2 and P4 foundations. The Computer & Structures, Inc (2002), SAP2000. Integrated software
horizontal shear resistance along the caisson shaft kSHL provides the for structure analysis and design [computer program].
secondary largest resistance, about 25-35% and 10-22% of total Computer & Structure, Inc., Berkeley, Calif.
applied overturning moment for P2 and P4 foundations, Chiou, J. S., Ko, Y. Y., Hsu, S. Y. and Tsai, Y. C. (2012) "Testing
respectively. The kSHL on P4 has a smaller contribution due to a and analysis of a laterally loaded bridge caisson foundation in
lesser embedded length. The moment resistances from kS are about gravel". Soils and Foundations, Vol. 52(3), 562-573.
12% and 16% for P2 and P4 foundations, respectively. The moment Chiou, J. S., Yang, H.H., and Chen, C.H. (2009) "Use of plastic
resistance from kV are about 4% and 12% for P2 and P4 hinge model in nonlinear pushover analysis of a pile".
foundations, respectively. With increasing exposed length, the base Journal of Geotechnical and Geoenvironmental Engineering
resistance provides more moment resistance. The vertical shear (ASCE), 135(9), 1341-1346.
reaction (kSVB+kSVL) also provides about 8-10% moment resistance Chang, K. C., Sung, Y. C., Liu, K. Y., Wang, P. H., Lee, Z. K., Lee,
for both P2 and P4 foundations although it does not directly provide L. S. and Witarto (2014) "Seismic performance of an existing
resistance to horizontal loading. bridge with scoured caisson foundation". Earthquake
Engineering and Engineering vibration, Vol. 13(1), 151-165.
95
Study of Bored Pile Capacity in Klang Valley Residual Soil based on Field-
Performance Data
Allan Y.L. Chwee1, E.G. Balakrishnan2, and A. Nazli3
1
Assistant Engineering Manager, GCU Consultants Sdn Bhd, Malaysia
2
Managing Director GCU Consultants Sdn Bhd, Malaysia
3
Faculty of Civil Engineering, Universiti Teknologi Malaysia, Malaysia
E-mail: allanchwee@gcu.com.my
ABSTRACT: Bored pile has been gaining its popularity in this recent years due to its ability to carry higher working load and higher lateral
force resistance. On top of that, bored pile is suitable to penetrate into shallow rock or hard layer. Many designers in the past would ignore
the end bearing contribution in bored pile due to the problem of soft toe at the base. This approach would be too conservative which would
result in longer pile length as the geotechnical capacity would only depends on shaft friction. In Malaysia practise, geotechnical capacity of
bored pile in residual soil is calculated from SPT-N correlation based on modified Meyerhoff approach. This paper presents the study of
relationship between pile capacity obtained from load test results and field performance data in Klang Valley geology. Extrapolation
technique using Chin’s method has been adopted in obtaining the ultimate shaft friction for non-fail load test. A total of 20 bored piles of
diameter varying from 900 mm to 1500 mm were tested on multiple geology formation in Klang Valley. From this research, the relationship
between shaft friction and SPT-N blow in three different types of geology such as Granite formation, Limestone formation and Kenny Hill
formation are explored. The findings showed that fsu/N ratio in Limestone formation and Kenny Hill formation are quite match and
agreeable with studied carried out by other authors. On the other hand, higher fsu/N can be obtained in Granite formation based on the
plotted graph. For the base resistance, interpretation of the collected load test results indicated a matching relationship between fbu and
Pb/Ptop in soil as well as rock where R-squared of more than 90% are obtained. The study of load test results have enable development of
shaft resistance correlation with SPT-N blow as well as correlation between Pb/Ptop and base resistance. This would allow designer to adopt
higher bored pile capacity which would result to shorter pile length and cost saving.
Keywords: SPT-N correlation, shaft friction, end bearing, load test and pile length.
1. INTRODUCTION length optimization if higher shaft resistance factor and end bearing
factor can be used for different layer of soil (Varies in SPT-N).
In Malaysia, the conventional geotechnical practice for pile This paper studied the relationship between shaft resistance/end
foundation design is based on working stress method where the bearing and SPT-N blow in different geology formation and discuss
derived allowable carrying capacity of pile (working load of pile) the possible correlation which can be developed from load test
shall always be greater than the unfactored column load. The results.
working load of a pile is obtained by applying safety factors (FOS)
to the ultimate geotechnical capacity of the pile. Conventionally,
2. LITERATURE REVIEW
different or same FOS will be applied to the shaft and base
resistance of the pile. The FOS and the combination of the different 2.1 General Geology of Klang Valley
FOS are depending on multiple factors such as the experience of the
designer to the similar soil or rock conditions, project design brief, Klang Valley is defined as an area in Malaysia which centered in
client requirement, etc. Technically, the FOS used in foundation is Kuala Lumpur and include the surrounding cities of Selangor state.
not a factor of safety on the strength of the foundation, but is a The largely Lower Palaezoic rocks extend southwards along the
factor to limit the settlement at serviceability based on experience of Western Belt into Selangor and the Federal Territory of Kuala
most soils for which relatively stiff and linear behavior will persist if Lumpur. The Kuala Lumpur Limestone is overlain unconformably
the stress levels are kept below about 30 to 50% of their ultimate by the more gently folded Carboniferous to Permian Kenny Hill
capacities (Atkinson, 2007). Formation, estimated to be about 1200 to 1500 m thick, that is
Commonly, the unit shaft friction and unit end bearing are exposed in several places notably towards the south of Kuala
obtained from correlation with SPT-N value primarily for residual Lumpur, from Petaling Jaya and Puchong to Kajang and Putrajaya.
type of soil. Many studies have been conducted and adopted in The Kenny Hill Formation named after Kenny Hill (Bukit Tunku) in
today practise where the pile geotechnical capacity is calculated Kuala Lumpur is a monotonous clastic sequence of interbedded
based on SPT-N correlation. Pile design is often verified through shale, mudstone and sandstone. The Kenny Hill Formation (Yin,
pile load tests and is essential to ensure that settlement of the pile 1976; Foo, 1983) shows both soft-sediment deformation structures
would comply with the permissible settlements as per the standards and multiple tectonic deformations.
or code requirement. On the other hand, the intrusion of granitic rock, mainly light-
In today construction, instrumented and working pile load test grey, coarse-grained biotite granite occupies the hills and the
such as static load test, high strain dynamic load test and even bi- undulating area around Kuala Lumpur except in the south. The
directional load test have been widely carried out. With the huge granite has intruded into the older sedimentary rocks and has caused
amount of load test data, study can be undertaken to refine the tilting and warping of the Kuala Lumpur Limestones and Kenny Hill
correlation based on field performance data. Commonly in residual Formation. The contacts between the granite and sedimentary rocks
soil, unit end bearing and unit shaft friction are obtained from are generally covered by thick alluvial or superficial deposits during
correlation with field performance data such as SPT-N value. Quaternary period (Komoo, 1989).
Considering the different types of soil in Malaysia, different
stiffness of residual soil could lead to different unit shaft friction 2.2 Existing Correlation with Field Data
value. With the huge pool of instrumented pile load test data in The current design practice in Malaysia is based on the conventional
Klang Valley, Malaysia, there is possibility of developing a realistic approach which is the working stress method. The working load
unit shaft friction and end bearing correlation to suit with Klang acting on the pile shall always be lesser than allowable carrying
Valley ground condition and further enhance the existing capacity of pile. Several correlation for calculating fsu and fbu using
correlation. Study can be carried out to assess the possibility of pile SPT-N values have been proposed by different authors. Meyerhof
96
(1976) suggested the ultimate unit end bearing, fb in a homogeneous The instrumented pile load tests were collected with the
granular soil for displacement piles as: following information:
a. Load settlement readings at the pile head.
fb = 40Nb (Db/B) < 400Nb (1) b. Instrumentation data comprising strain gauges (resistance wire
and vibrating wire types) and rod extensometers (tell-tale)
where Nb is the average value of SPT-N at the pile base and Db/B is along the length of the piles.
the average depth ratio of the base into bearing stratum. c. Soil Investigation data for each pile test.
Other authors suggested a common equation as below.
3.2 Extrapolation of Data
fbu = Kbu x SPT-N (2)
Geotechnical capacity which consist of shaft friction and end
where Chiu and Perumalswamy (1987) recommended Kbu to be 50 bearing can be obtained from the instrumented pile load test.
and Toh et al. (1989) proposed Kbu values between 27 and 60. Generally, the ultimate capacity is obtained when the pile is loaded
The relationship for shaft friction in correlation to SPT-N values to 3 times of the working load or load to fail which would mobilize
is generally expressed as below. the pile. However, non-fail load test would not have mobilized shaft
friction and end bearing. Thus, the ultimate capacity would not be
fsu = Ksu x SPT-N (3) able to be determined. Hence, it is important to determine the
ultimate or limit load as accurately as possible. Several methods are
Meyerhof (1976) suggested Ksu to be 2.0 while Chiu and available to extrapolate the final value to obtain the peak value or
Perumalswamy (1987) have adopted 2.5. Toh et al. (1989) resistance of the load test results. Four common extrapolation
recommended Ksu ranges between 2.5 and 2.7 for N values up to methods available are Davisson’s, Hansen’s, Chin’s and De Beer’s
120 and a lower values of Ksu should be used for greater N values. methods.
Established local correlations uses 2.0 as Ksu with fsu limited to Davisson’s method allow the engineer to extrapolate the
200 kPa for shaft resistance and Kbu = 40 with fbu limited to 4000 maximum value with consideration to the length and size of the pile.
kPa for base resistance. On the other hand, Brinch Hansen’s method predict the failure based
on the assumption that hyperbolic relationships exist between the
load and the displacement. For Chin’s method, it made assumption
3. METHOD OF STUDY
that the load-settlement (Q vs Δ) relationship is hyperbolic and
This research focused on the study of bored pile geotechnical hence the inverse slope of a plot of Δ/P (vertical axis) vs P results in
capacity in relation to field performance data such as standard the failure value. Lastly in De Beer’s method, load settlement values
penetration test (SPT-N). This research also study the influence of were plotted on double logarithmic chart.
empirical correlation to the different type of rock formation. From the 4 methods discussed, Chin’s method is deemed to be
most suitable for extrapolation of data considering its simplistic
3.1 Data Collection approach and independent from any soil condition. Chin’s method
make it possible to predict the ultimate resistance even if the head
For the study of bored pile performance, instrumented pile load test settlement did not reach 10% of the pile diameter. However, the
results were collected throughout the Klang Valley area. The extrapolation value become more accurate when the head settlement
following data were collected and the summary is indicated in Table approaches 10% of pile diameter.
1. A proposal by Borel et. al (2004) in his research indicated that
 Data at 14 sites in Klang Valley were gathered. the Chin method over predicts significantly the shaft resistance
 20 numbers of Instrumented Bored Pile were collected with when the load is mainly resisted by the toe. The absolute error made
various geological formation. can exceed 30% of the ultimate resistance which is independently of
 Years of testing are from 2012 to 2017. the percentage of load carried by the shaft. Therefore, the value
extrapolated by Chin method has been reduced by 20% in order to
Table 1 Summary of Data Collection represent the actual ultimate capacity.
No Location Pile Year Formation
1 Site 1 3.3 Instrumented Bored Pile Load Test
PTP1 2012 Kenny Hill
2 Site 2 PTP1 2013 Granite Instrumented pile load test has been commonly adopted by designer
3 Site 3 PTP2 2013 Limestone and accepted by most of the client before commencement of actual
4 Site 4 RRSN-TP1 2013 Granite piling work in the recent years. The awareness of instrumented load
test has grown exponentially over the years. Client understand that
5 Site 4 PTP2 2013 Granite
instrumented load test is able to determine the capacity of pile and to
6 Site 5 TP1 2013 Granite verify the design parameters in relation to shaft and base resistance.
7 Site 5 TP3 2013 Granite Thus, optimization can be explored which would result in economic
8 Site 6 PTP1 2013 Granite design.
9 Site 6 PTP2 2013 Granite In the selection of instrumented pile load test, several general
10 Site 6 PTP3 2013 Granite rules were observed as below:
11 Site 7 BN-122 2013 Limestone  Pile used in the load test shall be constructed similarly with the
12 Site 7 working pile.
PTP1 2013 Alluvium
 The test pile could be smaller than the actual working pile but
13 Site 8 PTP1 2014 Limestone
limited to 50% of the working pile diameter.
14 Site 9 PTP1 2014 Limestone  Location of the test pile must be similar to the working pile’s
15 Site 10 P03 2015 Granite ground condition.
16 Site 11 TP1 2015 Kenny Hill  Test pile shall be carried out within the same site with the
17 Site 11 TP2 2015 Kenny Hill working pile.
18 Site 12 PTP1 2016 Kenny Hill  Instrumented pile shall be tested to 3 times working load or
19 Site 13 PTP1 2016 Granite failure. Otherwise, extrapolation of data shall be carried out.
20 Site 14 PSLT 01 2017 Granite All the test pile selected must be installed with instruments
where the arrangement generally consist of 4 nos of Vibrating Wire
97
Strain Gauges (VWSG) and Tell-tale Extensometer at each soil

layer. Figure 1 indicated the typical arrangement of the instruments.
Photo of instrumented piles and type of load test are indicated in
Figure 2, 3, 4 and 5.
VWSG LEVEL A
VWSG LEVEL B Figure 4 Bi-directional Load Test
VWSG LEVEL C
VWSG LEVEL D
VWSG LEVEL E
Figure 5 Kentledge (Static) Load Test
VWSG LEVEL F 3.3 Correlation between Pile Load Tests and Field Data
From the large pool of load test results, tabulation can be carried out
PILE TOE LEVEL
VWSG LEVELG to determine the relationship between shaft resistance and base
resistance with field performance data ie. borehole results. In the
process of analyzing the result, several assumption have been made.
Figure 1 Typical Arrangement of Instruments for Pile Load Test i. The piles were vertical and no eccentric loading.
ii. All the strains measured in the piles were due to
compressive force only.
iii. The cross sectional area and circumference of the test
piles were assumed to be constant through the pile length.
iv. Elastic modulus of pile is assumed to be equal to elastic
modulus of concrete material of the pile.
The procedures used in the data analysis of the instrumented
piles and tabulation of data are as follows:
i. Only ultimate/maximum shaft friction and end bearing
values are collected. For load test that have not reach
failure state, extrapolation using Chin’s method are
adopted to obtain the maximum capacity.
ii. Load test selected must be fully instrumented with both
strain gauges and extensometers installed along the length
Figure 2 Vibrating Wire Strain Gauges of the piles.
iii. The depth of the strain gauges were measured from the top
of the pile.
iv. Graphs were plotted for ultimate end bearing against
Pb/Ptop, ultimate shaft friction/N against SPT-N and
ultimate end bearing/N against SPT-N based on different
types of geology.
4. RESULTS AND DISCUSSION

4.1 Relationship between Pile Capacity and Field-
Performance Data
From the collected load test results, studies were conducted to
correlate the pile geotechnical capacity in term of shaft friction and
end bearing with field performance data primarily standard
penetration test (SPT-N). Standard penetration test is the most
Figure 3 Reaction Pile Load Test common soil investigation method used in most of the development
site. Furthermore, designers in Malaysia have been using SPT-N
correlation based on modified Meyerhoff approach to obtain pile
capacity for many years.
98
4.1.1 Shaft Resistance 4.1.2 Base Resistance

In this study, correlation of SPT-N with shaft friction, fsu for 3 For end bearing, study has been carried out to obtain correlation of
different geology conditions are looked at to obtain a refinement to bearing capacity (fbu) for soil and rock with reference to Pb/Ptop.
current practice. Figure 6, 7 and 8 indicate the relationship between Two (2) graphs which show the relationship between Pb/Ptop and
fsu/N and SPT-N for granite formation, limestone formation and ultimate end bearing are shown in Figure 9 and 10 below.
Kenny Hill formation respectively.
Figure 6 Relationship between fsu/N and SPT-N for Granite Figure 9 Relationship between fbu and Pb/Ptop in soil
formation
Figure 7 Relationship between fsu/N and SPT-N for Limestone

formation Figure 10 Relationship between fbu and Pb/Ptop in rock
From the plotted graph in Figure 9 and 10, the summary of the
developed equation and R2 percentage are listed in Table 2.
Table 2 Summary of relationship between fbu and Pb/Ptop for soil

and rock
Material Relationship between R2 (%)
fbu and Pb/Ptop
Soil fbu = 154.69 (Pb/Ptop) 95
Rock fbu = 242.5 (Pb/Ptop) 98
For both condition, R-squared of more than 90% are achieved.
Figure 8 Relationship between fsu/N and SPT-N for Kenny Hill

formation
99
5. CONCLUSION
From the study done on the collected test results, the following
conclusion are drawn:
a. The subsoil profile in Klang Valley consist of varies geology
mainly Kenny Hill, Granite and Limestone formation. With
this varies formation, study has been carried out to determine
the influence toward the fsu/N ratio. From the study, Kenny
Hill and Limestone formation indicated almost similar value
and close to the existing correlation proposed by other
authors but granite formation provided a higher fsu/N ratio.
As Granite is an igneous type of rock while Kenny Hill and
Limestone are sedimentary type of rock, it is believed that
different type of rock formation would influence the fsu/N
ratio.
b. Some of the load tests have not reach the maximum capacity.
Extrapolation using Chin’s method deemed to be suitable to
Figure 11 Bar Chart for variation of Pb/Ptop according to load test obtain the ultimate capacity.
c. Relationship between fsu/N and SPT-N indicated that higher
fsu/N can be obtained for SPT-N < 50 while fsu/N ratio drop
From Figure 11, almost half of the collected load tests indicated to a consistent range of 2 and 3 for SPT-N > 50. Therefore,
Pb/Ptop ranges from 20% to 40% at 12.5 mm settlement while the higher fsu/N can be adopted for soil with SPT-N < 50 but
rest of the load tests showed Pb/Ptop less than 10%. An average verification using instrumented pile load test is highly
value of 12% is tabulated from the 20 numbers of load tests carried recommended.
out. d. For correlation of end bearing, 2 type of conditions are
considered which are pile in soil and pile in rock. Based on
4.2 Discussion the plotted graph between fbu and Pb/Ptop, the correlation of
From the graph plotted between fsu/N and SPT-N for three (3) end bearing are proposed as in equation 4 and 5. Both of the
different types of rock geology as in Figure 6, 7 and 8, a relationship correlation indicated high value of R-squared of more than
can be derived. All the 3 plotted graphs indicated higher fsu/N can 95% which indicate a good fit for all the data.
be obtained for SPT-N < 50. Once SPT-N > 50, the fsu/N ratio e. The obtained end bearing correlation with SPT-N blow for
decreased substantially which shown an almost consistent pattern. soil is within the recommended correlation by other authors.
This study indicates that fsu/N of more than 3.0 can be adopted in In addition, the finding also validate the existing correlation
soil with SPT-N < 50. For soil with SPT-N > 50, fsu/N became used in local practise.
almost consistent and fall within the range of 2 to 3. Therefore,
possibility of using higher fsu/N ratio can be considered for soil 4. REFERENCES
with SPT-N < 50. However, load test shall always be carried out to Abdelrahman, G. E., Shaarawi, E.M. and Abouzaid, K. S. (2003).
verify the assumption made in design. Interpretation of Axial Pile Load Test Results for Continuous
In term of base resistance, study has been carried out for two (2) Flight Auger Piles. Proc. Of the 9th Arab Struc. Eng. Conf.,
condition which are pile socket into rock and pile socket into hard Abu Dhabi, UAE.
layer of soil. Relationship between Pb/Ptop and SPT-N blows for all Atkinson, J.H. 2007. The Mechanics of Soils and Foundations. 2nd
the collected load test results have been plotted. Out of 20 numbers Edition, Taylor and Francis (Publishers)
of load test data, 16 number of load tests terminated at rock while Balakrishnan, E. G. (1994). Performance of bored piles in Kenny
the remaining load test terminated in soil. The findings show a good Hill Formation (weathered meta-sedimentary) in Kuala
match of data where the R-squared for both condition are above Lumpur, Malaysia. M.Eng. thesis, Asian Institute of
90%. The equation are as below: Technology, Bangkok, Thailand.
Balakrishnan, E. G., Balasubramaniam, A. S. and Noppadol Phien-
fbu = 154.69 (Pb/Ptop) for soil (4) wej (1999). Load Deformation Analysis of Bored Piles in
Residual Weathered Formation. J. Geotech. Geoenviron.
fbu = 242.5 (Pb/Ptop) for rock (5) Eng., Feb., 122-131.
Burland, J. F. (1973). Shaft friction of piles in clay—A simple
The bar chart as in Figure 11 indicated a variation of Pb/Ptop ratio fundamental approach. Ground Eng.,6(3), 30–32.
in between 0.5% and 40%. The average value for all the collected Bohn, C., dos Santos, A. L and Frank, R. (2016). Development of
load test data is calculated to be 12%. The average Pb/Ptop value Axial Pile Load Transfer Curves Based on Instrumented Load
obtained is almost in agreement to the industry practice where the Tests. Journal of Geotech. and Geoenviron. Eng.., ASCE,
end bearing capacity of bored pile is normally limited to 20% of 143(1).
working load. At Pb/Ptop = 12%, the fbu in soil is calculated to be Borel S, Bustamante M, Gianeselli L, Ponts et Chaussess LC.
1856 kN while fbu in rock is 2910 kN. Fbu of 1856 kN is correlated (2004). An appraisal of the Chin method based on 50
to be 37N in soil. This correlation is within the range recommended Instrumented Pile Tests. Ground Engineering.
by other authors such as Chiu & Perumalswamy (1987) and Toh et Bjerrum, L. (1953). Les pieux de fondation en Norvege. Ann. Inst.
al. (1989). In addition, 37N correlation is also in agreement with the Tech. Batiment Travaux Public., 6(63/64), 375–376.
local practise of 40N. Thus, this study validate the existing Chan, S. F. (1975). An experimental study of behaviour of end
correlation used for end bearing. bearing cast-in-situ piles. Proc., 4th Southeast Asia Conf. on
Soil Engrg., Kuala Lumpur, Malaysia.
Chang, M. F., and Broms, B. B. (1991). Design of bored piles in
residual soils based on field-performance data. Can. Geotech.
J., Ottawa, Vol 28, 200–209.
100
Chang, M. F., and Goh, A. T. C. (1989). Design of bored piles Meyerhof, G. G. (1976). Bearing capacity and settlement of pile
considering load transfer. Geotech. Engrg., Bangkok, foundations. J. Geotech. Engrg. Div.,
Thailand, 20, 1–18. 102(3), 195–228.
Chiu, H. K., and Perumalswamy, R. (1987). Foundation for Capital Randolph, M. F., and Murphy, B. S. (1985). Shaft capacity of driven
Square, Phase 1, Kuala Lumpur. piles in clay. Proc., Offshore Technology Conf., Offshore
Proc., 9th Southeast Asian Geotech. Conf., Bangkok, Thailand, 2, Technology Conference, Houston.
177–194. Samuel G. P. and Terry A. T. (1999). Extrapolation of Pile Capacity
Horvath, R. G., and Kenney, T. C. (1979). Shaft resistance of rock from Non-Failed Load Tests.
socketed drilled piers. Proc., Int. FHWA-RD-99-170, U.S. Department of Transportation, VA.
Symp. on Deep Foundations, ASCE National Convention, Atlanta, Tan B. K. and Yeap E. B. (1977). Structure of the Kenny Hill
182–214. Formation, Kuala Lumpur and Selangor.
Karlsrud, K. (2014). Ultimate Shaft Friction and Load-Displacement Geology Society Malaysia, Bull. 8, 127 – 129.
Response of Axially Loaded Piles in Clay Based on Toh, C. T., Ooi, T. A., Chiu, H. K., Chee, S. K., and Ting, W. H.
Instrumented Pile Tests. J. Geotech. Geoenviron. Eng., (1989). Design parameters for bored piles in a weathered
140(12), 04014074. sedimentary formation. Proc., 12th Int. Conf. on Soil Mech.
Komoo, I. (1989). Engineering Geology of Kuala Lumpur, and Found. Engrg., Rio de Janeiro, 2, 1073–1078.
Malaysia. Proceedings of the International Conference on Tomlinson, M. J. (1994). Pile Design and Construction Practise. E
Engineering Geology in Tropical Terrains, Bangi, Malaysia. & FN Spon., London., 99 – 165.
262 – 273.
101
Foundation Value Engineering & Underpinning Using Micropile for Building

Upgrading Works.
K. M. Lee1, E. G. Balakrishnan2 and E.N. Mohd Said3
1
Director, GCU Consultants (Johor) Sdn Bhd, Malaysia
2
Director, GCU Consultants Sdn Bhd, Malaysia
3
Geotechnical Engineer, GCU Consultants (Johor) Sdn Bhd, Malaysia
E-mail: kaiming.lee@gcujb.com.my
ABSTRACT: Jobs involving upgrading of old structures into new modern buildings are common in this country. This paper presents the
value engineering and underpinning design for an abandoned structure located in Southern Malaysia. The alternative design involved
optimization of foundation design by assessing the various design elements including the structural and geotechnical capacity of piles. The
common methods of piling design including the alpha and beta method etc. together with adopted parameters are also presented in this paper.
The acceptance criteria for pile load test commonly adopted in Malaysia will also be shared in for reference. The design of Micropile will be
elaborated for both structural and geotechnical capacity. Recommendations on design parameters and the relation with construction
constraints are also be highlighted. Besides, the common misconception on structural design of micropile will be discussed and pile load test
results are presented to substantiate the derived conclusion. Finally, there are many issues commonly faced in construction projects. Among
them, the limitation of SI in determining the pile length, the importance of proper construction planning, working under limiting headroom
and lesson learned will also be shared in this paper.
Keywords: Underpinning, micropile, structural, geotechnical
1. INTRODUCTION The existing building was a partially completed multi-storey

complex abandoned 20 years ago. The existing foundation
The scarcity of land has always been an issue especially in big cities comprised mainly of bored pile with diameter ranging between
and capitals that serves as the hub for major business activities. 600mm and 1700mm. Due to the higher loadings from the proposed
Johor Bahru as one of the largest city in Malaysia is facing the same upgrading works, most of the existing foundation needed to be
issue where land scarcity has recently becoming more prominent. upgraded as well. Besides, the changed in structural layout required
The scarcity of land in Johor Bahru city has resulted in many “land part of the building to be demolished and reconstructed. Based on
making” activities by reclamation. Besides, upgrading of the new layout plan, the Owner’s consultant proposed that bored
old/abandoned buildings has also been practiced due to high land piles and spun piles to be adopted for building extension areas and
demand. This paper will discuss on foundation value engineering also smaller structures located at the perimeter of the building.
and underpinning works of an abandoned building located in Johor Besides, underpinning of the existing structure will be carried out
Bahru. Due to the strategic location of the building, it has been using 250mm diameter micropile.
identified by the current Owner to upgrade the abandoned structure The newly appointed design and built Contractor has decided to
into a service apartment together with modern shopping centre. carry out a value engineering for the building design including the
The site is located at the city centre of Johor Bahru located at foundation works. During the value engineering exercise, two piling
approximately 350km away from Kuala Lumpur, Malaysia. The site options have been considered for the underpinning works. The
is believed to have underlain with old alluvium which is termed as options including the proprietary jack in pile system where the
the “Simpang Formation”. The deposit comprised of semi jacking will be carried out using the installed piles as the reaction
consolidated weathered coarse sand, sandy clay and also gravel. system and also the micropile that has more track records over the
The geological map and location of the proposed site is presented in jack in system have been evaluated. The contractor had eventually
Figure 1.1 for reference. A total of 20 nos. of borehole were also decided that the same piling systems to be adopted as proposed by
carried out on site to establish the soil properties and profiles on site. the Owner’s consultant. However, design optimisation was carried
out by adopting larger working load closer the pile structural
capacity of 0.25 times the concrete strength and also reduction in
Proposed pile length utilizing the individual borehole results together with
Site verification by pile load test. During the value engineering process,
the need to increase the API pile size was also identified according
to the structural design approach specified in FHWA. The design
approach, parameters including load testing results will be discussed
in the subsequent sections of this paper.
2. FOUNDATION DESIGN BY OTHERS

The proposed foundation design by the Owner’s consultant
comprised a combination of spun pile, bored pile and underpinning
using micropile. The information on existing pile design are
presented in Table 1 for reference. The layout plan for the proposed
development is also presented in Figure 1 to indicate the
complication of the foundation works especially working within
congested site and also under constraint headroom condition.
Figure 1 Geological map at site location
102
Table 1 Summary of Pile Sizes of Existing Design by Others Where,

 = Ks tan 
Pile Type Pile Size Pile Length Working Load Nq = bearing capacity factor, e.g. Berezantzev et. al, 1961
(mm) (m) (kN) σv’ = average effective stress along the shaft
Micropile 250 25.0 - 38.3 700 σvb’ = effective stress at pile base
Bored 600 29.0 -37.2 1,700 Ks = coefficient of earth pressure, e.g. Stas & Kulhawy, 1984
Pile 750 29.0 -37.2 2,310  = pile/soil friction angle, Stas & Kulhawy, 1984
800 29.0-37.2 3,020
900 29.0 - 37.2 3,820 Modified Meyerhof Method
1000 30.0 - 37.2 4,710
Shaft Resistance, fsu = KsN (5)
1100 30.1 - 41.5 5,700
1200 30.1 - 40.5 6,790 End Bearing, fbu = KbN (6)
1300 30.1 - 41.5 7,860
1400 31.0 - 44.0 9,230 Where,
1500 31.0 - 46.0 10,600 Ks = shaft resistance factor, e.g. 1.8-2.5 depending on soil
1600 31.0 - 48.0 12,060 type 
1700 31.0 - 49.5 13,620 Ks = base resistance factor, e.g. 40-45 for bored pile and
Spun Pile 300 31.0 800 depending on soil type
N = standard penetration resistance
Based on any of the above method, the ultimate pile capacity

can be derived and with appropriate safety factors, the working load
of a pile can be derived. The allowable pile working load can
generally be estimated as follow: -
or whichever lower (7)
where,
Qall = allowable Geotechnical Capacity
Qsu = ultimate Shaft Capacity, fsu As
Qbu = ultimate Base Capacity, fbu Ab
Fs = factor of Safety for Shaft Resistance
Fb = factor of Safety for Base Resistance
Fg = factor of Safety for Global Resistance
Figure 2 Site layout plan As/Ab = pile shaft area/pile base area
3.2 Structural Capacity of Pile

3. ALTERNATIVE DESIGN
For pile structural design, bored pile and spun pile adopted the
3.1 Geotechnical Capacity of Pile design value recommended in BS 8004 while the method
Several methods commonly used in Malaysia for estimation of pile recommended in FHWA has been adopted for micropile design.
geotechnical capacity include the alpha method, beta method and FHWA recommended that the structural capacity of an uncased pile
also the more frequent used modified Meyerhof method. The to be estimated from the steel pile and also the internal grout as
following outlines the concept of each method in high level for shown in equation 8. Besides, FHWA also stated an important fact
reference: that by excluding both the internal and external grout may result in
overconservative and uneconomical design.
Alpha Method
Pallow = 0.4fcAg + 0.47fyAs (8)
Shaft Resistance, fsu = α su (1) where,
fc = characteristic grout strength
End Bearing, fbu = Nc su (2) fy = yield strength of reinforcement
Ag = area of grout
where, As = area of steel
 = adhesion factor, e.g. Tomlinson, 1957
su = cohesion of soil from field/laboratory test 3.3 Final Design
Nc = bearing capacity factor, 6+L/d ≤ 9
L/d = pile length/size For the alternative design, foundation design using modified
Meyerhof method has been adopted. As spun piles are mostly
Beta Method driven to refusal, no optimization was carried out. For bored pile,
the pile capacity has been assessed with two (2) different sets of
Shaft Resistance, fsu =  σv’ (3) safety factors and the one with lower pile capacity will be adopted
as the final pile working load. Higher safety factor on end bearing
End Bearing, fbu = Nq σvb’ (4) was adopted for set 1 to cater for the uncertain base cleaning
especially for shorter end bearing pile while set 2 is mainly check
for longer pile which rely more on the shaft resistance that is more
cossistant.
103
Set 1 : Fs = 1.5; Fb = 3.0 3.4 Load Test Results

Set 2 : Fs = 2.0; Fb = 2.0
A total of seven (7) preliminary pile load tests up to three (3) times
Due to the small diameter of the micropile, base cleaning of the working load were proposed to be carried out at various locations on
pile remained more uncertain. Thus, the end bearing of the micropile the site. Maintained load test using both kentledge blocks and
has been omitted in the design. For shaft resistance, a safety factor reaction anchors were both adopted on site. Figure 4 shows the
of 2.0 has been adopted to derive the pile length. Based on the typical set up for maintained load test using ground anchors carried
adopted parameters, the results of the design are presented in Table out on site.
3.1 for reference.
Table 2 Summary of Pile Sizes of Alternative Design
Pile Type Pile Size Pile Length Working Load

(mm) (m) (kN)
Micropile 250 10.5 - 34.0 650
Bored Pile 600 15.0 - 31.5 1,700
750 10.0 - 30.5 2,250
800 19.5 - 28.0 2,350
900 14.0 - 37.0 3,600 & 3,820
1000 15.0 - 37.5 4,300 & 4,710
1100 15.0 - 42.0 5,400 & 5,700
1200 15.0 - 44.5 6,000 & 6,790
1350 34.0 - 40.0 7,300 & 8,500
Figure 4 Maintained Load Test Using Ground Anchors
1500 27.0 - 39.0 9,300
The maintained load test results for both the bored pile and
Comparing Table 1 & Table 2, it can be observed that the length micropile are summarised in Figure 5 and 6 for reference.
of bored pile has generally been reduced especially the lower range
pile length. Besides, some of the bored pile sizes have also been
reduced by adopting higher working load closer to structural
capacity. For micropile, the pile length and pile number have also
been reduced by lowering the redundancy in pile design. On the
other hand, the API pipe size of the micropile was increased from
88.9mm to 101.6mm according to the recommendation by FHWA.
Besides, to enhance the load transfer mechanism, the upper 3m
of the micropile has been cased with mild steel casing as shown in
Figure 3.
Figure 5 Pile Load vs Pile Top Settlement Curve for Bored Pile
Figure 6 Pile Load vs Pile Top Settlement Curve for Micropile

In Malaysia, the test piles are usually deemed acceptable when
the load test results meet the limiting settlement criteria of 12.5mm
and 38mm at one time working load and two times working load
respectively. The limiting criteria may alter to include elastic
shortening for long and slender piles. Based on the test results, all
bored piles managed to achieve the desired test load and meeting the
limiting settlement criteria. For micropile, only one (i.e. ULT 5) out
of three of the tested piles has achieve the required test load. Due to
the large pile top movement on piles, further assessment on toe
settlement for micropile was carried out and results are presented in
Figure 3 Cross Section of Micropile Figure 7 for reference.
104
Figure 8 On going Bored Pile Works on Site

Low head room and confined work space posted a very
challenging environment to the site workers. Movement between
existing columns, flooding of site and providing fresh air to workers
requires proper safety and operating procedures to ensure accident
Figure 7 Pile Load vs Pile Toe Settlement Curve for Micropile free site. Figure 9 & 10 showing the micropile rig working under
From Figure 3.4 and 3.5, it can be observed that ULT 7 with congested site and low head room condition.
smaller pipe size (88.9mm) managed to achieve a lower toe
settlement as compared to ULT 5. However, at higher induced load,
the pile failed to sustain the loading with relatively large pile top
movement. On the other hand, ULT 5 with larger pipe size
(101.6mm) managed to complete the entire load test cycle. In view
of the relatively large pile top movement and negligible pile toe
movement for ULT 7, the pile with smaller API pipe size is lightly
to have failed structurally.
4. CONSTRUCTION
The foundation construction for the project was carried out between
year 2014 & 2015. Soil investigation works were commenced
slightly ahead the foundation construction in order to enable the pile
design to be finalized. Generally, both bored pile and micropile Figure 9 Micropile Rig Working Under Limiting Head Room
activities started almost in parallel while spun piling works only
started midway through the piling construction. Bored piles were
mainly carried out at the building extension areas and also at
locations where demolishing of buildings was carried out.
Micropiles with smaller machines focused mainly within the
existing building areas due to congested site and limiting head room
condition. Spun piles were later carried out for smaller structures
positioned along the perimeter of the proposed building. Similar to
other constructions, many issues were encountered during the actual
physical works on site. Among those are lack of soil investigation
information for decision on pile termination, tight construction
schedule, congested and low head room and also sloping ground
condition etc.
As there was only limited SI carried out on site, the finalising of
actual pile length can only be confirmed by the site personal with
Figure 10 Micropile Rig Working Under Limiting Head Room
reference made to the nearest soil investigation information. Even
with the presence of geotechnical engineer on site, the identification As the site was surrounded by existing structures and roads, part
of suitable materials for pile termination on site has never been easy. of the site that was founded on higher ground needed to be trimmed
Thus, the termination of pile was purely based on the engineers on to facilitate the piling construction. The maximum difference in
site with feedback from the design office. At location where ground level was recorded to be approximately 7.5m. As such, cut
confirmation of suitable pile length was difficult, the approach by slope and temporary shoring works were required to facilitate the
lengthening the pile has to be taken. piling activities and site works. The slope was check to ensure that
Tight construction schedule has always been an issue for most no excessive movement that may result in damage on the completed
construction projects. As the construction involved demolishing part piles. The design was coupled with instrumentation monitoring
of the existing structure to cater for piling works, it requires proper during excavation to ensure nominal movement on the cut slope.
planning in order to ensure work continuity. Figure 8 showing the Figure 11 shows the cut slope formed during construction to
bored pile rig working on the partially demolished structure. facilitate the piling works.
105
engineering exercise including proprietary jack-in pile system where

installed piles are used as the reaction system for jacking. The final
adopted option involved bored piling at the extension areas,
micropile for the underpinning section of the building and also spun
pile for smaller structure positioned at the perimeter of the proposed
building.
It is important that sufficient SI to be planned such that
sufficient geological sections can be generate with reasonable
interpolation for foundation design. As SI costs is merely a fraction
of the total construction cost, sufficient SI with good engineering
judgement will result in economical foundation design. Foundation
design and checking with different sets of safety factors is a good
practice to prevent overlooking in foundation design especially
when dealing with uncertainties in construction methods.
The use of uncase grout section for structural capacity in
Figure 11 Cut Slope Near Piling Works micropile design especially in soil with low confining pressure will
As there were also nearby structures surrounding the site, likely to overestimate the pile structural capacity. For micropile
permanent retaining wall using contiguous bored pile (CBP) wall design, it is also a good practice to omit the end bearing of the pile
with strutting were also introduced to prevent movement of the due to the small pile diameter where base cleaning cannot be
surrounding structures. Figure 12 shows the layout plan of CBP ascertained. Finally, underpinning of foundations using micropile is
wall constructed on site. a proven solution with many successful case histories. The selection
of foundation solutions for underpinning shall take consideration of
other factors including the soil condition, availability of rigs,
schedule etc. and proper construction planning is necessary to
ensure on time delivery of a project.
CBR
wall 7. REFERENCES
Atkinson, J.H. 2007. The Mechanics of Soils and Foundations. 2nd
Edition, Taylor and Francis (Publishers)
Balakrishnan, E. G. (1994). Performance of bored piles in Kenny
Hill Formation (weathered meta-sedimentary) in Kuala
Lumpur, Malaysia. M.Eng. thesis, Asian Institute of
Technology, Bangkok, Thailand.
Figure 12 CBP Wall Layout Plan Balakrishnan, E. G., Balasubramaniam, A. S. and Noppadol Phien-
wej (1999). Load Deformation Analysis of Bored Piles in
Residual Weathered Formation. J. Geotech. Geoenviron.
5. DISCUSSION & LESSON LEARNED Eng., Feb., 122-131.
Burland, J. F. (1973). Shaft friction of piles in clay—A simple
Many methods are available for estimation of pile geotechnical
fundamental approach. Ground Eng.,6(3), 30–32.
capacity. The common used methods include the alpha method,
Symp. on Deep Foundations, ASCE National Convention, Atlanta,
beta method and also the Meyerhof method. The different methods
182–214.
are each suitable for different soil conditions. The alpha method
Karlsrud, K. (2014). Ultimate Shaft Friction and Load-Displacement
will be more appropriate for foundations under soft ground
Response of Axially Loaded Piles in Clay Based on
condition whereas the beta method will be more appropriate for
Instrumented Pile Tests. J. Geotech. Geoenviron. Eng.,
piles in cohesionless ground condition. The modified Meyerhof
140(12), 04014074.
method is suitable for both cohesive and cohesionless ground
Meyerhof, G. G. (1976). Bearing capacity and settlement of pile
condition but will have limitation when dealing with soft ground
foundations. J. Geotech. Engrg. Div., 102(3), 195–228.
with very low SPTN value (e.g. SPTN 0). The designer shall select
Randolph, M. F., and Murphy, B. S. (1985). Shaft capacity of driven
the method of design carefully by referring to sufficient SI
piles in clay. Proc., Offshore Technology Conf., Offshore
information with proven of load test results. The checking of
Technology Conference, Houston.
geotechnical capacity for bored pile with more than one set of safety
FHWA-RD-99-170, U.S. Department of Transportation, VA.
factors will help to prevent overlooking on design due to
Geology Society Malaysia, Bull. 8, 127 – 129.
uncertainties in construction method especially for shorter piles
Tomlinson, M. J. (1994). Pile Design and Construction Practise. E
which relies more on end bearing. For structural capacity of
& FN Spon., London., 99 – 165.
micropile, it is recommended that only the grout within the cased
section to be used for estimating the structural capacity of pile.
Including the grout for uncase section may overestimate the pile
structural capacity.
Sufficient SI with proper planning are necessary to achieve
optimise pile design. Besides, good construction planning is always
necessary to ensure smooth and safe working environment such that
the works can be completed on time.
6. CONCLUSION
Land scarcity especially in the city centre area has resulted the reuse
and upgrading of abandoned and old building structures. An
abandoned complex in Johor was identified to be upgraded to
service apartment with modern shopping centre. Several
underpinning options have been assessed during the value
106
Interpretation on Performance of Two Drilled Shafts Subjected to Tensile Loading

Considering Concrete Cracking Effect
San-Shyan Lin1, Tai-Hong Chen1 and Chia-Hong Lai1
1
Department of Harbor and River Engineering, National Taiwan Ocean University, Keelung, Taiwan 20224
Email:sslin46@gmail.com
ABSTRACT: In this paper, conversion of measured strain data into pile loads for tensile load testing of two drilled shafts is studied using the
secant modulus of concrete. A back analysis method, considering the possible effect of concrete cracking or slippage between steel-grout
interface, is used in converting the strain into pile loads. Subsequently, the t-z curves along shaft are obtained based on the pile loads interpreted
from back analysis.
Keywords: drilled shaft, tensile loading test, back calculation, hyperbolic model
1. INTRODUCTION Subsequently, the concrete modulus can be calculated as

Pile tensile load testing is used either to verify carrying capacity � ��
of the preliminary design or to determine ultimate frictional ��= �
= (3)
��×��
�
resistance of a pile foundation. In the latter case, strain gages are
often installed at selected depth along the pile for measurement of in which �� is the concrete stress. Multiplication of the concrete
strain distribution and subsequently converted into pile load. The stress and the steel stress by the concrete area and the steel area,
conversion procedure requires the property of the axial stiffness of respectively, the pile load can be calculated by summing these two
the tested pile. For tensile load testing, it is often carried out by values as given in Eq. (4)
holding the rebar cage on the pile top during testing. The testing
results may be affected by the bond strength between concrete and
steel interface. In the following, the pile refers to drilled shaft in this �= ��× ��+ ��× �� (4)
paper.
In this paper, two tensile load testing data are collected from The resulting modulus values given in Eq. (3) are often plotted
local contractors in Taipei. The secant modulus of concrete versus against strain, whose relationship is then modelled with a best fit
strain and the stress-strain relationship of the concrete at uppermost curve such as exponential or polynomial equation. In case of using
installed gage level of all cases are derived first for comparison. A exponential equation, the ��verses strain relationship can be
back analysis method using hyperbolic model is used in converting expressed as
the strain into pile loads.
��= �× ��
(��
⁡ ) (5)
2. The Secant Modulus of Concrete
where � and � are constants. Under any rebar strain gage level �
,
The secant modulus of concrete of the tested pile is often
the pile axial force can be expressed as
interpreted form collected strain data at the uppermost rebar gages
installed near the pile head. In order to calculate the stress level in a
drilled shaft, it is assumed that the measured strain from the gages ��= �
�{[�× ��
(��
⁡ )] × ��
+ ��× ��
} (6)
are representative of the entire cross section (Lam and Jefferis 2011).
In addition, it is also assumed that the modulus versus strain Once concrete cracked or slippage occurred at steel-grout
relationship at the uppermost gage level can be applied to the interface under tensile load, the tensile stresses resulted from tensile
remainder of the pile. On more assumption of the method is that the load will gradually carry by the steel reinforcement only. Hence, the
frictional resistance between the soil and the pile from ground readout from the rebar strain gages will deviate from the assumption
surface to the uppermost gage installed level is neglected under the of stain compatibility in Eq. (2), because the strain in steel
given applied load at head. In this regard, the uppermost set of rebar reinforcement is not equal to the strain in concrete.
gages need to be placed near the pile head to minimize the loss of
the pile load due to shaft resistance. In the meantime, it also needs 3. Back Analysis Method
to consider the end effect or the Saint Venant principle (Lam and
A hyperbolic model (Lin et al. 2007 ) is used to simulate the
Jefferis 2011). In general, the rebar gage is installed at depth
nonlinear behavior at the interface between pile shaft and
approximately equal to the width of the shaft. In Taiwan, most of
surrounding soil. Beyond this nonlinear behavior at the interface, the
the pile load testing installs the uppermost rebar gage at level of 1m
shear stress versus displacement behavior of the soil field is modeled
to 3m below pile head (Lin et al. 2007). The second gage level is
as linear behavior. A back analysis method (Xiao et al. 2003) is
installed at the cut-off level. The procedures of computing the secant
adopted to determine the required parameters based on the pile load
modulus of concrete are reviewed in the following:
test result.
The steel stress ��is given in the following equation
The assumed shear stress and relative displacement
relationship at the pile/soil interface is shown in Fig. 1 can be
��= ��× �
� (1) approximated by a hyperbolic equation having the form of:
where ��= elastic modulus of the steel and � �= measured strain ∆�

�
�=
�
(7)
from the rebar strain gage. The steel force � � can be obtained by �+�∆�
�
multiplication of the steel stress by the cross-sectional area of steel
��. Similarly, the concrete force � � is equal to the subtraction of Based on the instrumented rebar strain gage readout during the
the steel force �
� from the applied force at the pile head. pile load test, the pile load distribution along depth can be calculated.
Assuming stain compatibility, we have Subsequently, an spline interpolation function can be applied to fit
the measured data to obtain a load distribution along depth function,
�
�= �
� (2) which can then be used to determine the shear stresses along depth
at each loading stage using the following equation:
107
� =−
1 ̅ �(�)
��
(8) Class- Unit
�� 2�� SPT C
0 Depth Desc-ription ificati weight
N (kN/m2)
on (kN/m2)
in which Pi(z) is the Spline fitted function. The value of b at 0.0~
various depth can be calculated as �(� ) = ��/� �(�), in which �
Backfill SF 1.5~2 - -
� 2.8
is in general between 0.85 and 0.95. 2.8~ Very soft silty
The displacement at any depth of it’s corresponding loading CL 1.5 17.84 20.6
3.5 clay
step is calculated using the following equation: 3.5~ Very soft silty
CL 1~1.5 17.37 24.5
15.45 clay
� 1
� =� − ∫ � ̅ (�
)�� (9) 15.45~
�� 0 Soft silty clay CL 2.5~3 17.47 36.3
18.5
18.5~ Soft silty clay
in which �� and �� are the elastic modulus and the cross- CL 3~4.5 17.69 39.2
22.0 and some sand
sectional area of the pile, respectively. The value of � at various 22.0~
depth can be obtained using the following equation: Silty sand SM 9 - -
23.85
Andesite rock,
(� −�� )(1−�(�)� ) 20
��(�
)= ��
(10) 23.85~ gravel and
�� - ~ - -
38.6 silty sand
50/8cm
mixture
The initial shear stiffness of pile/soil interface is 38.6~
Sandstone - 50/5cm - -
39.3
��= 1/��(�
) (11) 39.3~ Sandstone
- 50/3cm - -
44.0 /Shale
4. Case Studies
The basic information on six collected tensile load testing is Table.3 Subsurface condition at D3 site
given in Table 1. These two piles were installed by reverse
circulation method. The ground conditions of the D1 and D3 sites Class- Unit
SPT C
are given in Table 2 and 3, respectively. The nonlinear stress- strain Depth Desc-ription ificati weight
N (kN/m2)
relationship of the concrete of the case examples at the uppermost on (kN/m2)
gage level is shown in Fig. 2. 0.0~
Backfill SF 11 - -
In order to determine a and b parameters given in Fig. 1, the 2.5
pile axial load along depth at each corresponding loading steps is 2.5~
Firm silty clay CL 4~5 - -
calculated first as shown in Fig. 3 and 4 for D1 and D3 piles, 4.5
respectively. However, the calculated pile axial force becomes 4.5~ Very soft silty
CL 1~1.5 17.63 45.1
unreasonably high at some gage levels when the applied loading 13.0 clay
becomes higher, as shown in these figures. The reasons for such a 13.0~
Soft silty clay CL 2~4 17.38 33.4
high axial load at these gages are most possibly affected by the bond 25.5
strength between concrete and steel interface, which may have 25.5~
Firm silty clay CL 5~8 17.53 43.2
caused slippage between mortar and steel interface or even caused 37.5
concrete cracking. Hence, the assumption of strain compatibility in 37.5~
Stiff silty sand ML 10~14 19.04- 76.5
Eq. (2) does not apply. The measured strain becomes taken by the 41.7
steel only instead of taking by both steel and concrete. It’s the 41.7~
Silty sand - 14 - -
possible reasonable to have such a high axial force at certain gage 43.0
levels. Andesite rock 50/13cm
An interpolation spline function is used to fit the measured data 43.0~
and some silty - ~ - -
to obtain a load distribution with depth shown in Figs. 3 and 4 for 70.5
sand 50/6cm
piles D1 and D3, respectively. The interpolation is based on the gage
readouts of the uppermost, the second set and the bottom gages for
the fitting. The axial load versus displacement relationships of both
piles is shown in Fig. 5.
Figures 6 and 7 show the results of the calculated initial
stiffness and ultimate strength at the pile/soil interface along shaft.
The t-z curves close to the pile head and close to the pile toe are
shown in Fig. 8 and 9, respectively.
Table.1 Tested pile information
Loading Construction
Pile Diameter Depth
Method Method
Reverse
D1 Tensile 1.5m 45.7m circulation
method
Casing
D3 Tensile 1.5m 49.7m
method
Table.2 Subsurface condition at D1 site Fig.1 Assumed shear stress vs relative displacement at the pile/soil
interface
108
Fig. 5 Load vs displacement relationship at pile head

Fig.2 Concrete stress-strain relationship
D1 (north)
Fig. 3 Axial load along shaft of D1 pile
Fig. 6 Predicted Ksi profile
D3 (south)
Fig. 4 Axial load along dept of D3 pile
109
5. Conclusions
Back analysis method was used in this paper to interpret the
required parameters based on tensile loading tests of two drilled
shafts. More reasonable axial force along pile shaft was fitted using
hyperbolic model. Subsequently, the t-z curves along shaft are
obtained based on the pile loads interpreted from back analysis.
Based on the back calculation of these two tested piles, the
following conclusions can be drawn:
a) The initial shear stiffness at soil/pile interface Ksi of D1 was
higher than that of D3.
b) The failure strength τf of D1 was also higher than that of D3
especially at depth 22m below ground surface.
c) Based on the t-z curves given in Figs. 8 and 9, the D1 pile was
not yet reached the ultimate value. However, hardening and
softening behavior was observed for the D3 pile close to head
and close to toe, respectively.
6. Acknowledgement
The present study was carried out as part of a research project
funded by Ministry of Science and Technology (NSC 102-2221-E-
019-028-MY3), Taiwan. The author is grateful for the financial
support.
7. References
Lam, C. and Jefferis, S.A., (2011) “Critical Assessment of Pile
Fig. 7 Predicted τf profile
Modulus Determination Methods,” Canadian Geotechnical
Journal, Vol. 48, No. 10, 1433-1448.
Lin, S.S., Wang, K.J., Hsieh, H.S., Chang, Y.H., and Huang, C.S.,
(2007) “Field Testing of Axially Loaded Drilled Shafts in
Clay/Gravel Layer,” Journal of GeoEngineering, Vol. 2, No. 3,
123-128.
Xiao, Z. R. Du, M.F., and Zhang, Z (2003) “A Back Analysis
Method to Determine the Parameters for the Pile Foundation
Analysis in Foundation: Innovations, Observations”, Design and
Practice, Thomas Telford, London, UK, pp.951-960.
Fig. 8 Interpreted t-z curves close to pile head
Fig. 9 Interpreted t-z curves close to pile toe
110
Pile Settlement Uncertainty in Jakarta, Indonesia.

Bondan Satria1 and Widjojo A. Prakoso2
1
Department of Civil Engineering, Universitas Indonesia, Depok, Indonesia
2
Department of Civil Engineering, Universitas Indonesia, Depok, Indonesia
E-mail: bondansatria.isaktil@gmail.com
ABSTRACT: Geotechnical engineering is one of the oldest branch of civil engineering, which continues to evolve along with the level of
human civilization. In the practice there are much uncertainty in the geotechnical field, both in terms of design and the inherent variability of
the soil material. This uncertainty, coupled with the limited soil data available, makes the geotechnical engineer assumed over estimates or
under estimate the soil parameter. The safety factor (SF) assumptions is one method to reduce the risk of such design error, but the FS can be
misleading as it is usually recommended or used without specific reference. One of the alternative approaches is the Reliability Based Design
(RBD), RBD is a design methodology that summarizes and incorporates key factors of uncertainty. RBD incorporates the cause and effect of
every variable and hence the possible consequences, making the eventual results more accurate. In this study the authors will discuss about
uncertainty analysis of pile settlement based on soil investigation data. The result is calibrated with the data of static axial loading test based
on existing projects in Jakarta. The use of RBD results in a more optimal and economical design, while maintaining the required safety
design consideration.
Keywords: Uncertainty, inherent variability, safety factor, reliability, pile settlement.
1. INTRODUCTION the pile. In this study is assumed that the soil around the pile skin
perimeter is bound or moped perfectly on the surface of the pile
Geotechnical engineering is one of the oldest branch of civil through friction and adhesion, each settlement of the pile is assumed
engineering, which continues to evolve along with the level of by the settlement of the soil around and at the bottom of pile. Based
human civilization. Before the 18th century the period of on classical soil mechanic “Poulus HG, Davis EH. Pile foundation
geotechnical engineering development was divided into 4 main analysis and design. New York: John Wiley & Sons; 1980,” soil can
phase, they are pre-classical, classical soil mechanic phase 1, be assumed to be linear elastic, the decline that occurs will be able
classical soil mechani phase 2 and modern soil mechanic to return to its original position with a force that is proportional to
(Skempton, 1985). In the practice, soil as heterogeneous material the modulus of soil elasticity and decreased magnitude
have much uncertainty in terms of design, the inherent variability of
the soil and also other factor such as design method, variations of 2. METHODOLOGY
loads occurring, variations in the strength of the soil material,
inaccuracies in designing, variations in method analysis, the quality The purpose of this paper to find the coefficient of
of the soil investigation, land degradation, soil expansion and uncertainty factors and determine the empirical relationship between
shrinkage and the use of construction methods . This uncertainty, one variable with other variables, key and factor so it is expected to
coupled with the limited soil data available, makes the geotechnical provide a coefficient factors that can make the calculation method of
engineer assumed over estimates or under estimate the soil pile settlement is more accurate. The field information and data such
parameter. Uncertainty in geotechnics has long been recognized as soil investigation data (SPT) and the pile static axial test of three
(ENR, 1963; Casagrande, 1965). Engineers understand directly or projects in Jakarta will be used as input, parameter and boundary in
indirectly that there is always the possibility of failure to achieve analysing the pile settlement in compression with spring stiffness
design objectives during the design life time. Therefore the safety method and elastic settlement analysis by Poulos & Davis, then to
factor (FS) is used in the design as a solution to reduce the risk of develop distribution data and uncertainty analyzes were analyzed
design failure, Safety factor (SF) generally used 2 ~ 3, this value is using random variable with 100x iteration and lognormal
generally adequately assumed in most foundation designs (Focht distribution for 50%, 100%, 150% and 200% from loading plan.
and O'neil, 1985). Safety factor serves as a solution to reduce design
failure, but (FS) can be misleading as it is usually recommended or 2.1 Pile Settlement Analysis
used without reference and is very subjective
One of the alternative approaches is Reliability Based Spring Method
Design (RBD), RBD incorporates the cause and effect of every The pile settlement formula with spring stiffness method is
variable, key factor of uncertainty and hence the possible calculated based refer to Skempton formula below :
consequences, making the eventual results more accurate. In this
paper, the authors will discuss about uncertainty analysis of pile Pile Stiffness :
settlement based on soil investigation data and the result will be
calibrated with the data of static axial loading test based on some (1)
projects in Jakarta, it is expected by using the key factors of Soil Stiffness :
uncertainty which is the initial step of the RBD will make the result
more optimal and economical design while maintaining the safety of (2)
design.
The paper will discuss the first step of RBD that is about (3)
the analysis of the uncertainty of the elastic settlement of pile (4)
foundation based on the soil data of field investigation and pile
static loading result as calibration on several projects in Jakarta, Where :
random number method and statistic and probabilistic method will
K = Stiffness of material
use to develop the data. When pile is loaded the settlement will
E = Modulus of elastiscity
increased. The settlement increase in the pile is caused by 2 (two)
A = Area of element
factors, the compression due to the structure of the pile itself and the
L = Length of element
settlement of the surrounding soil around pile and also the bottom of
f = Soil resistance
111
Askin = Parameter area of pile ̅ ∑

(10)
Atip = Tip area
Su = Shear undrained
√∑( ̅)
L = Pile settlement, use 1%d for friction and 4%d for tip (11)
Where rigid local coordinate matrices are used as below: (12)

̅
 EA EA  Where:
 f x1   L   u1 
    EA EAL   
s = Standard Deviation
f  u 2 
xi = X value at-i
 x 2  
 L L  n = Sample size
(5) ̅ = Mean
Which then assembled into global coordinates K system which is the
sum of Kpile and Ksoil with the following equation: Number of Iteration Cycle
{ } [ ] { } (6) The number of samples is plays an important rule to provide

accurate data. To obtain accurate analysis results in a statistical and
{ } [ ] { } (7) probabilistic method, a minimum size of the repetition or sample
process is adequate and can represent the model to be analyzed.
Poulos and Davis

Estimation of pile settlement on homogeneous soil will be used in
this study based on the following formula:
(8)
Where :
L = Settlement at the top of pile
F = Loading on the pile
Es = Modulus of soil elastiscity
d = Pile diameter
Ip = Settlement influence factor based on soil poison ratio
Random Variable
Figure 1 N-Sample vs COV
Random variable is a variable that has a probable value or chance is
a numerical result of a random phenomenon. The probability
characteristics of the random variable have a probability 3. RESULT AND DISCUSSION
distribution, which determines the possibilities occurring in the
intervals determined by the key factors and the key descriptions Spring Method
specified. Random variable can analysis by formula below, but for Based on the analysis from the spring method, there is has graphic
this paper author use excel formula that refer to algoritma Mersenne correlation between Su and N-SPT value for the three projects
Twister to generate radom variable. reviewed based on 50%, 100%, 150% and 200% from loading plan,
the analysis result have been verified using pile static axial loading
( ) (9) test result.
Where :
α = Multiplier
b = Increment
m = Modulus
Statistics and Probability
Statistics is a collection of data, information or results of application

of statistical algorithms in the form of numbers arranged in the form
of tables or lists and or diagrams that describe or relate to a
particular problem, while probability is an opportunity or being
probable or something to happen or be is a way of expressing
knowledge or believing that an event will take effect or has
occurred.
Figure 2 Pile Load Test vs Su/N-SPT (Spring Method)
This paper will used lognormal distribution, where the distribution
of lognormal associated with the normal distribution. Lognormal
distribution in the simplest form is a density function of a random
change whose logarithm follows the normal distribution law. The
basis for calculating variant and standard deviation is the desire to
Based on the figure 2 above, author have the data below :
know the diversity of a group data.
Table 1 Correlation Value of N-SPT vs Su for Spring Method
112
Loading 50% 100% 150% 200% Table 3 Mean, Standard Deviation and COV Resume analysis
Mean 11.87 6.33 5.30 3.47
STDV 3.76 0.06 1.44 0.59
COV 31.65% 0.91% 27.217% 16.90%
Poulos & David Method
Based on the analysis result from Poulus & Davis formula (Figure
3), provide the correlation between Es and N-SPT values for the
three projects reviewed for 50%, 100%, 150% and 200% from
loading plan and the analysis result have been verified using pile
static axial loading test result :
Figure 3 Pile Load Test vs Es/N-SPT (Poulos & Davis)
Based on figure 3 above, author have the data below :
Table 2 Correlation Value of N-SPT vs Es for Poulos & Davis

Loading 50% 100% 150% 200%
Mean 4.20 3.60 3.10 2.57
STDV 1.04 1.31 1.28 0.85
COV 24.86% 36.43% 41.18% 33.14% Uncertainty Factor
Based on the results of the design analysis, both using the method of
Resume for Data Analysis spring method and the elastic settlement formula by Poulos & Davis
The results of analysis from two different methods, soil data and pile obtained a value of COV smaller than the results of pile loading test,
loading test will be evaluated use statistic and probabilistic methods it is caused by the limited data of pile static axial test results.
for every phase of loading 50%, 100%, 150% and 200%, the result Therefore need an uncertainty factor that is incorporated into the
is resume in Table 3. analysis or formula in order to obtain the results of the analysis that
can cover the actual conditions on the field. Therefore author use a
constable X, where the constable X has an average value equal to 1
with a certain COV value so that it can cover the pile settlement
conditions based on the actual conditions.
Thus, the value of X constant for loading phase 50%, 100%, 150%
and 200% from loading plan below :
̅ =1
Xspring COV = 21.21% ~ 27.69%
̅ =1
Xp&d COV = 13.50% ~ 22.75%
113
4. CONCLUSION
Based on the information from the field data both soil

investigation data (SPT) and the static pile loading test of three
projects in Jakarta as an inputs in analyzing the reduction of the pile
settlement with the spring method and the elastic settlement by
Poulos and Davis and using the random variable probability method
with 100x iteration and lognormal distribution it can be made the
following conclusion:
1. The results of the analysis in this study have similar
results with pile loading test, but to obtain better results, is
require more data and distribution of samples are evenly
distributed in Jakarta (both ground investigation data and
pile testing).
2. The result of the analysis using spring method increased
COV along with increasing loading, whereas in Poulos
and Davis method got uniform COV for each loading with
the quantity ranged between 2 ~ 3 times from COV value
with spring method. This result explains that using the
assumption by the soil along the pole as homogeneous soil
in poulos & davis will increase the value of variability.
3. The correlation value of N-SPT with Su (undrained shear
strength) using spring method with 100 times cycle in this
study shows that the greatest variability occurs when the
load condition is 50%, and the smallest variability value
occurs at 100%. With variation values ranging between N-
SPT = 2.80 ~ 15.50 Su, for loads of 50%, 100%, 150%
and 200% from the loading plan.
4. The value of N-SPT correlation of Es (elastic modulus)
using Poulos & Davis formula with 100 times cycle in this
study shows that the greatest variability occurs when the
load condition is 150%, and the smallest variability value
occurs at 100%. With variations in value ranging between
N-SPT = 1.70 ~ 5.40 Es, for the load of 50%, 100%,
150% and 200% from the loading plan.
5. REFERENCES
Ang Alfredo H. S., and Tang Wilson H. (1975). Probability

Concepts in Engineering Planning and Design, Volume I & II
Basic Principles. Willey.
Kulhawy Fred H., Kok Kwang Phoon, Prakoso Widjojo A. and
Hirany Anwar. (2006). Reliability-Based Design of
Foundation for Transmission Line Structures. Cornell
University, Hollister Hall, Ithaca, NY, The 2006 Electrical
Transmission Conference.
Naghibi Farzaneh, Fenton Gordon A. and Griffiths D.V. (2014).
Prediction of pile settlement in an elastic soil. Elsevier –
Computer and Geotechnic.
Poulos H.G., Davis E.H. (1980). Pile Foundation Analysis and
Design. John Wiley & Sons.
Prakoso Widjojo A. (2016). Case Study on Variability in Soils and
Driven Pile Performance. Universitas Indonesia, Depok,
Indonesia.
Vesic A.F. (1977). Design of Pile Foundation. National Cooperative
Highway Research Program, Synthesis of Practice No. 42,
Transportation Research Board, Washington, D.C.F
Yudhi Lastiasih, Irsyam Masyhur dan Sidi Indra Djati. (2013).
Reabilitas Daya Dukung Pondasi Tiang Bor Berdasarkan
Formula Reese & Wright dan Usulan Load Resistance Factor
Design dalam Perencanaan Pondasi Tiang Bor Studi Kasus
Proyek Jakarta. Jurnal Ilmu dan Terapan Bidang Teknik Sipil
Badan Kejuruan Sipil Persatuan Insinyur Indonesia.
Bondan Satria, Widjojo A. Prakoso. (2017). Analisis Ketidakpastian
Penurunan Pondasi Tiang Pancang Berdasarkan Beberapa
Studi Kasus Di Jakarta, Indonesia. Tesis Departemen Teknik
Sipil Universitas Indonesia.
114
Re-Evaluation of Pile Capacity due to Shallow Gas – A Malaysia Case Study

C. H. Girsang1, N. N. Huang1, M. S. K. Abdullah1, M. R. Mansoor1, M. J. Rohani1, A. A. Rahman1 and W. M. M. Wan Ismail1
1
Group Technical Solutions, Project Delivery and Technology, Petroliam Nasional Berhad, Kuala Lumpur, Malaysia
E-mail: christian.girsang@petronas.com
ABSTRACT: Extensive geophysical and geotechnical surveys have been conducted at T field since 2006. The area has been known to show
anomalies that could potentially be shallow gas. The most recent geotechnical investigation took place in June 2016. This included one cone
penetration tests with pore pressure measurement (CPTU), new laboratory tests, four dissipation tests and five headspace tests. Dissipation
tests indicated a low to negligible amount of excess pore pressure. Headspace tests revealed that the gas most likely came from biogenic
source. Based on these new data, it was concluded that the effect of shallow gas on the integrity of the foundations is most likely negligible
and was not taken into consideration in the pile capacity analysis. The pile capacity was then re-evaluated using the NGI-05 method. Six
years of ageing was included in the calculations using the NGI-14 method as well as the effect of cyclic loading. The load was taken as the
largest axial pile head reaction for each pile. Taking the axial pile capacity including ageing and cyclic degradation give resistance factors for
axial loading between 2.07 and 2.97 which meet the requirements in ISO 19902:2007.
Keywords: shallow gas, ageing, cyclic loading.
1. INTRODUCTION 3. GEOTECHNICAL DATA

T field is located offshore Malaysia in 70-75 meter water depth. The 3.1 Borehole and CPTU Locations
structures in this field were installed in 2010. Extensive surveys
have been conducted since 2006. Geophysical surveys were Figure 1 shows the locations of the boreholes, gas seeps and seabed
conducted in 2006, 2010 and 2014; remotely operated vehicle depressions in relation to the platform structures.
(ROV) surveys were conducted in 2009, April and November of
2010, 2011, 2014 and 2015; and geotechnical site investigations All of the boreholes were drilled several years prior to any
were conducted in 2002, 2003, 2005, June, July, and August 2006 geophysical or ROV survey that identified the gas seepage and
and 2008 (Huang et al., 2018). craters. Therefore, it is possible that the gas observed emanating at
This area has been known to show anomalies t Crater 1 and Crater 2 could be filtering up to the surface through the
hat could potentially be shallow gas. Geophysical surveys old boreholes and CPTU holes.
identify several gas anomalies that extend to depths greater than 250
mbsf. Therefore, the gas is not from shallow sources, but is
generated at greater depths or at multiple depths before migrating
towards the surface.
The most recent geotechnical investigation took place in June
2016. This included one CPTU, new laboratory tests, four
dissipation tests and five headspace tests. The data from this
campaign was combined with the previous data to find out if there is
effect of shallow gas and then to re-evaluate the pile capacity taking
into account pile ageing and cyclic loading effects.
2. GEOPHYSICAL DATA
From geophysical surveys, it was found that the seabed near T
structures is irregular and mildly undulating due to the presence of
numerous small and large seabed depressions, isolated pockmarks
and pockmark clusters.
The T structures are located above what is believed to be a
shallow channel formation, one of which is directly beneath the
structures.
The geophysical surveys have identified numerous seabed
depressions near the existing T structures. Overall, the depressions
vary between 1.7 m and 4.0 m deep. The geophysical surveys
revealed that these types of seabed depressions are related with the
shallow channels and that shallow gas from deeper source migrates
up towards the seabed from the channel edges.
The geophysical surveys also noted numerous gas seeps within
T area. Within larger seabed depressions, gas seeps were scattered
randomly. Near T structure, gas was identified emanating from two
craters: Crater 1 (GS-1) with size of 12 m x 8m x 2.3m and Crater 2
(GS-2) with size of 7 m x 6 m x 1.7 m. The distance of these craters Figure 1 Borehole, gas seepage, seabed depression and platform
to the structure is 6.6 m and 35 m respectively. locations (not to scale)
115
3.2 Soil Layering 3.4 Gas Encountered during Drilling

Figure 2 presents a sketch showing soil layering as found in all A pilot hole drilled to 155 mbsf in June 2006 for the BH TB-DP
boreholes. There are two clayey silt layers reported in the borehole campaign did not show any presence of shallow gas. However, a
logs nearest to the T structures. It is possible that these two clayey soil investigation conducted at BH TB-CP1 location with a borehole
silt layers extend across the area where the foundations are located. drilled to 250 mbsf between July and August 2006 reported that
shallow gas was encountered at 170 m but the amount was minor
and drilling was proceeded to 250 m as planned.
3.5 Dissipation Tests

The June 2016 site investigation campaign conducted dissipation
tests at 67.5 mbsf, 86.4 mbsf, 110.6 mbsf and 114.5 mbsf. The main
purpose of dissipation tests is to get feedback on whether pore
pressure in excess of hydrostatic exists.
Figure 4 shows the measured pore water pressure (u2)
normalised by the vertical effective stress ('v) versus time for the
four dissipation tests. For tests at 86.4 mbsf and 110.6 mbsf, the
measured pore water pressure at the end of the test is less than 10%
of the vertical effective stress. For tests at 67.5 mbsf and 114.5
mbsf, the measured pore water pressure at the end of the test is less
than 20% of the vertical effective stress, and the pore pressures
appear as if they would continue decreasing if the test had not been
stopped. These tests all indicate a low to negligible amount of
excess pore pressure.
Figure 2 Soil layering
3.3 Soil Characteristics

Figure 3 shows the undrained shear strength (su) profile with depth
for BH TB-CP1, BH TB-KP, and CPTU TBCP-A1. The su profile
was based on CPTU measurements using Nkt values of 15 and 20, as
well as results from unconsolidated undrained (UU), consolidated
isotropically undrained (CIU), consolidated anisotropically
undrained (CAU) and direct simple shear (DSS) tests conducted
onshore. The su values from the CPTU for all three site
investigations give similar results. The laboratory tests are
consistent with the CPTU data in the upper 75 m, below which the
laboratory data tend to be lower than the CPTU data. This is due to Figure 4 Results of dissipation tests
sample disturbance increasing with depth caused by stress relief.
3.6 Headspace Tests
The June 2016 site investigation campaign also conducted five
headspace tests. The purpose of headspace testing is to detect the
presence of gas in collected soil and water samples. 'Headspace' is
the gas space above the sample in a chromatography vial. Volatile
sample components diffuse into the gas phase, forming the
headspace gas. The gas is then analysed in a laboratory to determine
what types of gas are present. The types of gas present and their
properties can indicate if the gas is from shallow biogenic sources or
from a hydrocarbon reservoir.
Table 1 gives the results in percent by weight. The samples were
all taken from depths corresponding to the clayey silt layers reported
in the boring logs, which are shown in Figure 2. It is expected that
these clayey silt layers would have higher concentrations of gas than
clay layers because they are more permeable.
Table 1 Headspace Test Results in % by Weight

Depth H2S CO2 N2 Methane Ethane Pro-
(mbsf) pane
65.1 0 0.11 94.34 5.55 0 0
69.1 0 0.16 97.82 2.02 0 0
73.1 0 0.13 97.44 2.43 0 0
85.1 0 0.16 97.40 2.44 0 0
Figure 3 Undrained shear strength with depth 89.1 0 0.05 95.42 4.53 0 0
116
The results show that a majority of the gas detected is nitrogen CPTU data with Nkt = 15 and 20. At greater depths the results from
(N2), followed by methane and a small amount of carbon dioxide UU tests show large scatter, but tend to give lower shear strength
(CO2). The tests detected no hydrogen sulphide (H2S), ethane or values than the CPTU results. The available DSS and CAU tests
propane. The percent of nitrogen is high because the samples were indicate that the Nkt for triaxial compression is likely closer to 15
flushed with nitrogen before being sealed. The samples were flushed than 20. The design profile of su,UU was selected based on the
with nitrogen to remove any gases present in the air that might be available data and engineering judgment.
sealed in with the soil sample. The methane gas most likely comes
from biogenic sources, whereas ethane, propane, and other higher
order hydrocarbons most likely come from a hydrocarbon reservoir.
Because there is no indication of ethane or propane, it was
concluded that the gas is most likely from biogenic sources.
The volume of soil sample and differences in permeability
introduce uncertainty to the amount of gas in the headspace. These
tests should, therefore, only be used as a qualitative indication of
whether gas is present in the sample and whether this gas is from a
biogenic source or from a hydrocarbon reservoir.
Based on the data shown in Sections 2 and 3, it is believed that
the effect of shallow gas to the integrity of T2 platform foundations
at T field is most likely negligible and was not taken into
consideration in the pile capacity analysis.
4. PILE CAPACITY
T2 platform is an eight legged jacket structure with 1829 mm
diameter piles made up of five sections. Wall thicknesses range
between 45 mm and 60 mm. Pile depth is 134 mbsf at piles A1, A2,
B1 and B2 and 142.5 mbsf at piles A3, A4, B3 and B4. An MHU Figure 6 Submerged unit weight and Plasticity Index profile
800S hydraulic hammer was used to drive the piles. From the as
built drawings, it was shown that all piles have a batter of 7.125
(1:8). The configuration of T2 platform piles is shown on Figure 5
below.
Figure 7 (a) Effective vertical in-situ stress (’v) and effective

preconsolidation stress (’vc) profile; (b) OCR profile
4.2 Static Reference Capacity

The axial static capacity of the piles was calculated using the NGI-
05 method (Karlsrud et al., 2005). The NGI method was selected in
order to capture changes in capacity correlated with plasticity. For
clays with plasticity higher than approximately 40%, the NGI-05
method give skin frictions that are similar to the ISO 19902 method
Figure 5 Configuration of T2 platform showing installation barge (ISO, 2007).
The craters are reported to be up to 2 m deep. This has been
4.1 Soil Parameters for Pile Capacity included by assuming a general scour level of 2 m. Table 2 present
the resulting axial capacities in compression and tension.
Based on the summary of soil conditions outlined in Section 3, soil
parameters were selected specifically for use in pile capacity Table 2 Static axial capacity. Weight of soil plug and pile not
assessments. included.
Figure 6 shows the profiles of submerged unit weight (γ') and
Plasticity Index (PI). Figure 6a shows the preconsolidation stress Pile Tip Friction/ Capacity
(σ'vc) measured in oedometer tests and correlation from CPTU tests Penetration Resistance MN
with σ'vc = k*Qt where k = 0.2 and Qt is the normalised cone mbsf MN
resistance. A k value of 0.2 is on the low side (Lunne et al., 1997), Qs Qb Qtotcompr Qtottens
but it appears to give a good fit to the oedometer data in the upper A3/4 142.5 95.2 6.9 102.1 95.2
soil layers. The data is plotted in terms of overconsolidation ratio B3/4
(OCR) in Figure 7b. A1/2 134.0 83.4 6.4 89.8 83.4
Figure 3, as presented in Section 3.3, shows the undrained shear B1/2
strength profile with depth, including laboratory tests and from Note: Qs = shaft friction; Qb = tip resistance
117
4.3 Effect of Ageing The assessment of the effect of cyclic loading on the ultimate
pile capacity was done by evaluating the cyclic load capacity for
Practical experience has shown that the skin friction on piles various load paths with the computer program PAX2 (NGI, 1994).
increases with time, even after full dissipation of the excess pore The degradation of soil resistance for the piles on the tension
water pressures generated by the pile driving. Comprehensive study side will be greater than for the piles in compression because parts
of pile ageing effects is presented by Karlsrud et al. (2014). of a pile in tension will experience two-way cyclic loading (i.e.
Skov and Denver (1988) proposed the following equation: reversal of direction of shear stresses) during the storm. However,
the available load data indicate that the piles do not experience
(t) = (t0) * [1.0 + 10 * log10 (t/t0)] (1) tension. Therefore, only piles in compression are analysed further.
Pile at leg A4 has the largest maximum storm load and pile at
where, leg B4 has the largest cyclic amplitude. The two piles are analysed
(t) : local skin friction at time t for cyclic effects in PAX2. The calculations in PAX2 evaluate the
(t0) : local skin friction at reference time t0 axial capacity of the pile by following a number of load paths to
t : time since pile driving failure.
t0 : reference time since pile driving The results of cyclic analyses show that cyclic loading may
10 : dimensionless soil parameter determined from pile field reduce the soil resistance for piles supporting T2 jacket by around
tests. 11% for the storm loading with 3-hour peak storm duration and the
most relevant loading direction.
This parameter expresses the increase in skin friction for a 10- Figure 8 compares the same PAX2 results with the data of
fold time increase. observed cyclic capacities for pile load tests on clay (Andersen et
For piles in clay, the reference time t0 is taken as 100 days since al., 2013; Karlsrud et al., 1993). The comparison shows that the
almost all pile tests used to calibrate the different calculation cyclic axial pile capacity estimated with PAX2 is quite consistent
methods were fully consolidated after 100 days. with observations from high quality cyclic pile load tests in clay.
The ageing factor Δ10 is a soil parameter that in principle could
be determined from load tests on piles loaded to failure at different
times after driving. The few tests carried out on identical piles, but
at different times, indicate that previous load tests on piles in clay
lead to Δ10 values that are too high. For piles in sand, the opposite
effect is observed.
Based upon the results presented by Karlsrud et al. (2014), the
Δ10 values for piles in clay are calculated as:
10= 0.05+1.3*(1.0 – PI/50%)2*OCR-0.5 (2)

OCR  (4 * su/’v)1.25 (3)
where su is the undrained shear strength from UU tests.

The pile tests in clay that form the basis for these
recommendations involve piles with an age of up to 10-15 years.
The skin friction increase due to ageing cannot continue forever. It
is, therefore, only 50% of the ageing effect was included for time
higher than 5000 days.
Table 3 presents the axial capacities including six years of
ageing based on the axial capacities reported above and the ageing
calculated using the Karlsrud et al. (2014) method.
Table 3 Static Axial Capacity Including Effect of Six Years of

Ageing. Weight of soil plug and pile not included.
Pile Tip Friction/ Capacity
Penetration Resistance MN Figure 8 Comparison of PAX2 results for cyclic capacity T2 piles
mbsf MN with cyclic pile load test for clays
Qs Qb Qtotcompr Qtottens
A3/4 142.5 111.6 6.9 118.5 111.6 5. RESISTANCE FACTORS
B3/4 Table 4 Axial Capacity and Resistance Factors for Storm Loads
A1/2 134.0 98.3 6.4 104.6 98.3
B1/2 Leg Tip Pile Capacity Load Resist-
Depth MN MN ance
It can be concluded that by applying effect of six years ageing, mbsf Qtot W1) Sum Factor
there is an increase of 16.1 to 16.5% and 17.2 to 17.9% in R
compression and tension respectively. A1 134.0 93.1 -2.3 90.8 30.8 2.94
A2 134.0 93.1 -2.3 90.8 36.3 2.50
4.4 Effect of Cyclic Loading A3 142.5 105.5 -2.4 103.1 42.8 2.41
The effect of cyclic loading should be quantified in axial pile A4 142.5 105.5 -2.4 103.1 48.9 2.11
capacity design, especially when also accounting for positive effects B1 134.0 93.1 -2.3 90.8 30.6 2.97
like ageing. Several factors should be considered when assessing the B2 134.0 93.1 -2.3 90.8 36.1 2.52
cyclic axial capacity of a pile. These include the type of loading, the B3 142.5 105.5 -2.4 103.1 42.9 2.40
cyclic soil properties, variation of soil strength and stiffness, pile B4 142.5 105.5 -2.4 103.1 49.9 2.07
flexibility and initial stress distribution.
118
1)
In compression, submerged weight of pile steel offshore engineering and other applications”, ISSMGE-TC
Table 4 lists the resistance factor for each leg based on the axial 209 Workshop-Design for Cyclic Loading: Piles and Other
capacity and loads of the piles. The capacity includes six years of Foundations. Paris, 4 September 2013.
ageing as well as degradation due to the cyclic nature of loading Huang, N. N., Girsang, C. H. and Mansoor, M. R. (2018)
whilst the load is the maximum storm load for each pile. “Managing shallow gas for offshore platforms via fit-for-
It can be seen from Table 4 that all piles have a resistance factor purpose solutions: a Malaysian case study”, Offshore
greater than 2.0 and as such meet the requirements in ISO Technology Conference Asia, paper OTC-28470-MS, Kuala
19902:2007. Lumpur, March 2018.
International Standard Organization. (2007) Petroleum and Natural
6. CONCLUSION Gas Industries - Fixed Steel Offshore Structures. European
Standard EN ISO 19902, CEN, December 2007.
T structures were installed in T field in 2010. This area has been Karlsrud K., Clausen, C. J. F. and Aas, P. M. (2005) “Bearing
known to show anomalies that could potentially be shallow gas with capacity of driven piles in clay, the NGI Approach”, Proc.
geophysical surveys identify several gas anomalies that extend to International Symposium on Frontiers in Offshore
depths greater than 250 mbsf. The most recent geotechnical Geotechnics, Perth Sept. 2005, A. A. Balkema Publishers,
investigation near T structures was carried out in 2016 comprised ISBN 0 415 39063 X.
one CPTU, new laboratory tests, four dissipation tests and five Karlsrud, K., Kalsnes, B. and Nowacki, F. (1993). “Response of
headspace tests. Headspace tests revealed that the gas most likely piles in soft clay and silt deposits to static and cyclic axial
came from biogenic source whilst from dissipation tests it was loading based on recent instrumented pile load tests”, Volume
concluded that only low to negligible amount of excess pore 28: Offshore Site Investigation and Foundation Behaviour,
pressure exists. As a result, pile capacity analysis does not take Society for Underwater Technology, pp549-584.
excess pore pressure into consideration. Karlsrud K., Jensen, T. G., Wensaas, E. K., Nowacki, F. and
The axial pile capacity of eight piles at T2 platform was re- Simonsen, A. S. (2014) “Significant ageing effects for axially
assessed using the NGI-05 method. This method was selected to loaded piles in sand and clay verified by new field load tests”,
capture changes in capacity correlated with plasticity. The effect of
Offshore Technology Conference, paper OTC-25197-MS,
ageing as well as cyclic loading were included in the calculations. Houston, May 2014.
By applying effect of six years ageing, there is an increase of pile Lunne, T., Robertson, P. K. and Powell, J. J. M. (1997) “Cone
capacity of 16.1 to 16.5% and 17.2 to 17.9% in compression and penetration testing in geotechnical practice”, Blackie
tension respectively. However, by applying the effect of cyclic
Academic & Professional, 312p.
loading, the soil resistance may be reduced by 11%. As a result, the
Norwegian Geotechnical Institute. (1994) “Verification and user’s
resistance factors for axial loading for storm loads was found to be manual for computer program PAX2, version 2.1”, NGI
between 2.07 and 2.97 which meet the requirements in ISO report 525286-1, 10 June 1994.
19902:2007.
Skov, R. and Denver, H. (1988) “Time-dependence of bearing
capacity of piles”, Proc. 3rd Int. Conf. on the Application of
7. REFERENCES Stress-Wave Theory to Piles. Editor B. H. Fellenius, Ottawa,
Andersen, K. H., Puech, A. A. and Jardine, R. J. (2013) “Cyclic Canada, pp879-888.
resistant geotechnical design and parameter selection for
119
120
Application of Newly Developed Real-time Website-base GIS Monitoring in

Tunnelling
L. Christian1, B. C. B. Hsiung2, and J. Huang3
1
Geotech Science Co., Ltd, New Taipei City, Taiwan
2
National Kaohsiung University of Science and Technology, Kaohsiung City, Taiwan
3
Geotech Science Co., Ltd, New Taipei City, Taiwan
E-mail: luis@geotech-science.com
ABSTRACT: In this paper, application of a newly developed real-time website-base GIS (Geographic Information System) automatic
monitoring platform in tunnelling is introduced. Some important functions in this platform have included a user-friendly operation interface,
automatic generation of induced settlement contour and its impacts on adjacent structures together with presentation of ground profile close
to the construction site etc. under a well-recognized GIS, such as Google map. By having these details, potential risk during construction
could be categorized and managed. Furthermore, as all data are provided real-time, site condition could be shown immediately and necessary
engineering judgement are thus eligible to be made without any delay. The application of this system in three tunnelling case histories are
demonstrated; the 1st case is tunnelling in central Kaohsiung which is twin-bored tunnel with depth of tunnel centre is approximately at 15 m
to 17 m beneath a 20 m wide main road in the city. The 2nd one is a case in Taipei metro system which has both cut-and-cover and NATM
tunnels, attentions were put on its application in follow-up of displacement changes as well as pore pressure during the construction. The 3rd
case is TBM tunnelling in a ground consists of very thick clay, occasionally with layers of highly permeable sand and high ground water
level in crowded environment city. It aims to completely monitor and control the surface settlement in order to eliminate impacts on adjacent
traffic, structures and daily life of residence. By applying this newly developed platform, risk of failures could be minimized, as it provides a
clear overall view of geotechnical measurement taken on the site from time to time.
Keywords: real-time website-base GIS automatic monitoring, risk categorized and management, tunneling, settlement
1. INTRODUCTION number of each instrument which can be selected later on (also

shown in Figure 1). By selecting any one instrument, traditional
Nowadays, due to the fast development of technology for both displacement versus time curve is still eligible to be indicated. In
hardware and software computer, lots of geotechnical monitoring order to maintain data reliability and avoid any man-made error,
results intended to be presented in more “user-friendly” style, real-time monitoring instrumentation together with data
especially using graphical skill and technology to build a monitoring transmission logs are strongly recommended to be adopted to collect
platform to be competent with neighbourhood area, such as deformation and stresses caused by construction activities.
Geographic Information System (GIS), used to provide a clear However, it is also possible to use manual instrument to collect data
overall view (Player, 2006). In addition, ground profile is always needed in said platform once the project scale is not large enough
critical to any underground construction activity in urban area, such and the client also has a concern in budget issue.
as deep excavations and tunnelling have a need to present ground
condition in the same platform. Therefore, it aims to briefly
introduce a newly developed real-time website-base GIS platform
which is eligible to present both ground profile along the route of
tunnel in the city and displacements induced by tunnel construction
activities in this paper. The visualized presentation of monitoring
results is also different from traditional one which is capable of
showing induced settlement contour. By doing so, it helps the
engineers to evaluate risks of adjacent ground and structure during
the tunnelling.
2. BACKGROUND OF THE WEBSITE-BASE GIS

PLATFORM
As indicated previously, this newly developed website-base GIS
platform shall be eligible to has functions as (1) indication of ground
profile for any location along the tunnel route as long as borehole
log information is available; (2) a “user-friendly” presentation of
monitoring data which shall not be a time-history for any single
instrument, has to be an overview of the whole construction package Figure 1 Instruments installed on site with Google MAP
at any time, any stage and any location; (3) a self-alert/warning
system shall be included and (4) the platform which has to be As shown in Figure 1, it has three options for user to choose
competent with widely accepted GIS software, such as Google Map which are “Map function”, “Map type” and “Marker List”. Using
or equivalent ones. these options, the user can freely choose the addressed site area, the
Considering four key functions stated above, the platform thus mode of the map, color and symbol of each instrument associated
be developed. Following are step-by-step operation of this website- with actual needs.
base platform. As the platform has a function of self-alert/warning, values of
Once the user starts to use this platform, the details of all alert/action levels shall be given before the start of the project and
instruments location together with neighborhood environment of the warning message shall be shown once any instrument reaches
site shall be presented in any GIS system, such as Google Map, as alert/warning level, as shown in Figure 2. Since none of instrument
shown in Figure 1. At the same screen, the user is able to see
121
indicates any measured deformation/stress currently beyond alert In order to generate a full view of ground profile, a full ground
level, “no available data in table” is therefore presented. profile was generated by the platform automatically instead of
Since the platform has to be user–friendly, the output can be manually. By doing so, the opportunity of man-made bias or error
possibly presented in the platform directly (refer to Figure 3) or into should be eligible to be reduced. In addition, potential ground risks
formats of EXCEL, CSV or PDF. In Figure 3, it shows name and could also be predicted and prevented in advance. Figure 5 shows
number of instrument, last measured value and related date as well generated ground profile (light colour means “silty sand” and dark
as alert level etc. colour means “silty clay”), it is aware that more reliable and
As explained previously, the platform has a function of contour intensive site investigation program as well as data should assist to
interpretation which can possibly categorize influence zone induced help to generate a more reliable ground profile. Limited and poor
by tunnel construction. Further details will be presented in later quality site investigation would not be able to achieve the purpose
section. stated above in the aspect of risk assessment.
Figure 2 Warning message from the system
Figure 4 Soil logs from boreholes close to the site
Figure 3 Details of instruments and measurements shown in the

platform
3. APPLICATIONS
Several examples are presented in this paper in order to demonstrate
the performance of newly developed website-based platform. Figure 5 Generated ground profile
3.1 Tunnelling in Kaohsiung

Deformation induced by tunnel construction was further
A 5.6 m of internal diameter twin-bored tunnel was constructed in examined. Different from previous product, the newly developed
central Kaohsiung, Taiwan and centre-to-centre distance is platform not only provides the function showing a conventional time
approximately 12 m at west end of the tunnel and gradually versus displacement curve of a single instrument, as shown in
increases to 14 m at east end of the tunnel. 25 cm thick, 1.2 m width Figure 6, but also could provide an overall deformation contour to
reinforcement concrete tunnel segments were used to construct the directly indicate the impacts on adjacent ground and structures (refer
tunnel and depth of tunnel centre is approximately at 15 m to 17 m to Figure 7; buildings with dark colour mean those has more than 10
beneath a 20 m wide main road in the city. The length of a single mm of observed building settlement). Although settlement of
tunnel is approximately 300 m. several buildings are more than 10 mm, the maximum tilting of the
Soil logs taken from locations close to tunnel site were presented building is only 3.3/10000, much less than the alert level (1/500) so
in Figure 4 (Hsiung et al., 2016). As shown in Figure 4, the main no immediate risk is seen and no additional protection measure
soil strata at tunnel depth has SPT-N value less than 20 and sand should be undertaken under this situation. Tunnel construction here
content of soils is very high, up to 70 to 80%. Groundwater level is only induces limited displacement and actually the road is quite
also high, approximately 3 to 4 m below surface level. wide so located buildings are a little bit away from the tunnel.
122
Figure 9 Influence zone made recommended by Clough and Schmit

Figure 6 Conventional presentation of monitoring record (1991)
3.2 NATM and Cut-&-Cover Tunnel in Taipei
In order to construct an above-ground depot for metro system in
Taipei, a cut-&-cover tunnel plus one NATM tunnel has to be
constructed for the use of cross-over and storage track of the depot.
Further, a historical building is located on top of slope close to the
tunnel and necessary slope-cut has to be conducted in order to
construct the cut-&-cover tunnel so the website-based platform is
considered to be adopted to control site activities, induced
deformations, pore pressure and stress as well as potential risks.
Figure 10 shows the site condition and earth anchors were
installed as additional support at the section for cut-&-cover tunnel
close to the portal of NATM tunnel. However, conventional
horizontal strutting system was applied for the rest section of cut-&-
cover tunnel.
Figure 7 Presentation of monitoring record using newly develop

website-based platform
By comparing details of cross section of tunnels (refer to Figure
8) and suggestions shown in Figure 9 made by Clough and Schmit
(1981), the influence zone shall be less than 8 m away from
boundary of the tunnel which is consistent with results shown in
website-based platform.
Figure 10 Site condition of NATM and cut-&-cover tunnel

As indicated previously, several modes can be selected in this
newly developed platform; Figure 11 thus presented those locations
of instruments with various backgrounds, such as road, terrain and
Figure 8 Details of cross sections of tunnels in transverse direction hybrid etc. so the user can choose anyone suitable.
123
Figure 13 Accident caused by tunnel collapse
Figure 11 Instruments with various backgrounds

As mentioned earlier, in this tunnel project, the monitoring of
historical building on top of the slope close to the tunnel is a key
issue in this project, it thus seen from Figure 12 that many
settlement points were installed on slope where immediately close to
the place that located historical building. A combination of Real–
Figure 14 Real-time monitoring instrument installed on site
time monitoring and manual monitoring were applied for the
measurements of these instruments so any movement or potential 4. CONCLUSION
risk of said historical building is thus eligible to be indicated in
advance and necessary protection or patch could be delivered on Application of a newly developed real-time website-base GIS
time. Except settlement markers, instruments installed on site (Geographic Information System) automatic monitoring platform in
include piezometer, inclinometer, tiltmeter, load cells, observational some tunnelling projects is discussed. By applying this newly
wells and crack meters on structures. developed platform, the risk of failures could be minimized, as it
provides a clear overall view of geotechnical measurement taken on
the site from time to time.
5. REFERENCES
Clough, E. J. and Schmidt, B., 1981, “Design and Performance of
excavation and tunnels in soft clay”, in soft clay engineering,
edited by Brand, E. W. and Brenner, R. P., pp. 600-634.
Hsiung, Bin- Chen Benson, Yang, Kuo- Hsin, Wahyuning Aila and
Hung, Ching (2016), “Three- dimensional effects of a deep
excavation on wall deflections in loose to medium dense
sands”, Computers and Geotechnics, Volume 80, 138- 151
Player, R. S. V., (2006) “Geographic Information System (GIS) Use
in Geotechnical Engineering”, GeoCongress 2006.
Figure 12 Instruments installed on the site

3.3 Accident Caused by Tunnelling in Taipei
As shown in Figure 13, an unexpected collapse was caused by
tunnel construction in urban area of Taipei city and the depth of
tunnel centre is approximately 40 m (refer to Figure 13). In order to
fully understand any change of ground and adjacent structures, real-
time monitoring instrumentation were installed on site immediately,
as presented in Figure 14. Once this newly developed platform can
be applied in the beginning of the project, it shall be much easier for
all parties involved to realise what happens on site at any specific
time and specific area which can assist to follow up and then decide
protection measures shall be undertaken immediately.
124
ERT-Based Leakage Tracing for Dam Safety and its Potential Sliding Surface
H. Wang1, C. H. Hu2, S. H. Hsieh2, and Y. C. Tsai2
1
Facility Management Research Center, CECI, Taipei, Taiwan
2
HCK Geophysical Company, Taipei, Taiwan
E-mail: herschel39@gmail.com
ABSTRACT: Electrical resistivity tomography (ERT) has been efficiently applied to geological investigation, mining exploration,
groundwater probing, and pollution prevention for years. In this paper, the on-site investigation was required to inspect the possible
groundwater leakage through an earth dam during either its full-level storage period or empty storage period. A potential seepage belt could
gradually develop along the residual permeable sandy/gravelly layer, an old abandoned river channel. A downhole ERT-based brine tracing
technique was conducted on the slopes nearby a spillway and a dam levee widening. The brine seepage flow observation could rapidly
reflect the possible direction and path of leakage flow developed in an earth dam in few hours. Monitoring the ERT distribution variations
can effectively identify the leakage spatial path, accumulation zone, and its potential sliding surface on dam slopes.
Keywords: Electrical resistivity tomography, leakage, brine tracing, dam, potential sliding, seepage.
1. INTRODUCTION vary with their water content. The resistivity response from fresh
water usually overlaps the resistivity ranges of soils. However, sea
Watertightness is one of key factors for planning dam engineering. water, as a better conduction medium, provides much lower
The watertightness investigations cover the seepage conditions on resistivity values below the resistivity ranges for both soils and
not only an earth dam itself but also the dam foundations. Seepage rocks. This implies that infusing brine, like a penetrant, into
is highly associated with permeability in geomaterials, geological subsurface can be a potential method to trace groundwater flows
structures, groundwater conditions, sinkholes, and mining pits when using ERT inspection.
(Hung, 1991). In general, sandstone and limestone are attributed to
relatively high permeable geomaterials. A bedding plane could be
regarded as a relatively high seepage channel when dipping to the
downstream side. Fractured rocks nearby faults or fault belts, and
rocks with tension cracks in anticline crests could lead to the
leakage at dam foundations. Reserving water could impact on the
original hydrogeological condition and promote groundwater to
outflow into adjacent low water table watersheds.
A complete watertightness investigation, including the items of
groundwater table and pressure, consists of geophysical inspection,
geological investigation, borehole exploration, and piezometers
(Hung, 1991). Electrical resistivity tomography (ERT in brief
below), one of near-surface geophysical inspection techniques, has
been used for groundwater flow or seepage investigations for years
(Schuster and Krizek, 1978; TGS, 2011; Wightman, et al., 2003).
In this paper, the ERT method is applied to inspect the possible
leakage through an earth dam in Taiwan. Modified from its Figure 1 A schematic representation for ERT inspection
traditional applications, an ERT-based brine tracing technique is
introduced to efficiently identify the seepage trend of groundwater
Table 1 Resistivity range of common rocks, soils, and waters (after
in potential sliding surface both nearby a spillway slope and on a
Loke, 2000)
dam widening slope.
Material Resistivity (ohm×m)
2. OVERVIEW OF ELECTRICAL RESISTIVITY Rocks
TOMOGRAPHY Granite 5×103–106
Slate 6×102–4×107
The ERT inspection technique can trace geological layers, faults,
Marble 102–2.5×108
groundwater pollutants, carves, mining, burial layers, or landslide
Sandstone 8–4×103
mass. The principle of ERT is to develop an artificial potential field
Shale 20–2×103
by probing one pair of current electrodes around a target zone (as
Limestone 50–4×102
shown in Figure 1). Two extra electrodes are used to measure the
Soils
ground potential difference. The measured apparent resistivity
Clay 1–100
image is usually displayed with a visible-light mode, which
Sand (dry) 5×103–2×104
corresponds to the resistivity values varying from 2,000 to less than
Sand (saturated) 2×102–103
1 ohm×m (Loke, 2000; Society of Exploration Geophysicists of
Gravel (dry) 2×104–8×104
Japan, 2014). The resistivity values for sediments composed of silt,
Gravel (saturated) 103–5×103
sand, and rock are more than 5 ohm×m, corresponding to the pink,
Alluvium 10–800
red, orange, yellow, or green zones. Water, pollutants, or metal
Waters
form low resistivity-content zones, corresponding to the light blue
Groundwater (fresh) 10–100
and grey shades.
Sea Water 0.2
The on-site resistivity values are highly associated with mineral
composition, grain size, mineral formation, water content, and ion
concentration (Loke, 2000; Society of Exploration Geophysicists of There are 4 commonly-used inspection types, including pole-
Japan, 2014). Table 1 shows the typical resistivity value ranges for pole array, pole-dipole array, dipole-dipole array, and Wenner-
geomaterials and water. Rocks usually have relatively high Schlumberger array (Loke, 2000; Society of Exploration
resistivity values. Soils have a wide variety of resistivity ranges and Geophysicists of Japan, 2014). Each inspection type has different
125
resolutions along the probing array direction and also expresses findings promote the reservoir agency to engage in the positioning
various inspection depth. investigation of leakage on the dam mass.
In practice, choosing ERT deployment is highly associated with An ERT investigation line is chosen at distances of 0k+603 m~
subsurface characterization and environment (Loke, 2000; Society 0k+783 m at the downstream side along the dam top, the most
of Exploration Geophysicists of Japan, 2014). Inspection depth and possible leakage position (Figures 2 and 3). The total investigation
resolution are in terms of electrode spacing and layout length. length is 180 meters. The inspection is conducted at both the high-
Furthermore, a trial and error process on sectional image is used to level storage and low-level storage periods. The on-site images in
determine the optimal layout deployment. In general, the smaller the reservoir show that the water level and waters coverage have
electrode spacing is arranged, the higher image resolution and the significant changes (Figure 4). If one carries out the ERT survey
shallower inspection depth are. Accordingly, the larger electrode along the dam at the different storage levels, such variations on
spacing is used, the lower image resolution and deeper inspection water head could likely lead to amplify the leakage effect on
depth are. resistivity distribution and probably help investigators identify the
leakage zone at the downstream side of the dam.
3. LEAKAGE TRACING CASES
3.1 Leakage throughout an earth dam Dam Top (Elevation 42 m)
The reservoir located in southern Taiwan was constructed in 1953.
Surrounding with scour-prone mudstone and silty sandstone, deposit
rapidly piled up to 3 quarters of its design volume found in 1996.
From 1997 to 2005, an improvement project was launched for
dredging the reservoir sediments, re-constructing the spill and
irrigation pipes through the earth dam, and dam mass improvement
(Sinotech, 2010). Since 2006, several piezometers, water level Clay
gauges, water level wells, inclinometers, flow weirs, seismographs, Sandy silt~ silty sand
and settlement points have been installed to monitor the dam Silty sand
conditions, including pore water pressure, water level, seepage Muddy sandstone
condition, displacement, and surface conditions (Chiou, et al., 2010; Silty mudstone
Liming, 2014; Sinotech, 2010). Coral stone
The 2,380-m long earth dam consists of one main dam (length
250 m) and two minor dams (right section length 300 m and left
section length 1,830 m) (shown in Figure 2). The main dam is Figure 3 Geological profile along the dam mass (after Sinotech,
composed of concrete core, compact soils, and rockfill. Similar to 2010)
the components in the main dam, the two minor dams are also
constructed with compact soils and rockfill except clay core. The
dam elevation is 42 m. The spill pipe intake and irrigation channel (a)
are also marked on Figure 2.
Spill Pipe Intake

Reservoir
(b)
Figure 2 Top view photo of reservoir and dam (modified from Spill Pipe Intake
Google Map) Reservoir
The geological conditions along the earth dam is plotted in
Figure 3. The clay core dam overlays a 10~20-m thick silty sand
layer which overlaps a monocline consisting of overlayered pervious
muddy sandstone and silty mudstone. The black dashed line also Irrigation Pipe Intake
indicates the concrete core zone.
The annual monitoring report for dam safety shows: (1) the
measured maximum settlement value less than tolerable settlement
10 mm/month; (2) the maximum lateral displacement less than 10
mm; and (3) eight leakage points found at the downstream side of
the dam at positions of 0K+475 m~0K+825 m (Liming, 2014; Figure 4 Field testing conditions: (a). full-level storage photo; (b).
Sinotech, 2010). The pervious geological condition and seepage
low-level storage photo
126
Figures 5(a) and 5(b) indicate the ERT inspection outcomes and grey shades, represent the silty sand and clay layers. The
during the full-level and low-level storage periods, respectively. significant resistivity variation zone is found at the measurement
The estimated geological layers and groundwater table are labelled positions of 30~70 m as shown in Figure 5(c). This pervious silty
with black and white dashed lines, respectively. The relatively high sand layer could be reasonably inferred as the potential leakage zone.
resistivity contour zones, corresponding to the pink, red, orange, In addition, an old landform sketch before the dam construction
yellow, or green zones, indicate the sand or gravel layers. The reveals that the meandering zone of the old river channel right
relatively low resistivity contour zones, corresponding to the blue overlaps this leakage zone.
Figure 5 Electrical resistivity tomography images: (a). Full-level storage period; (b). low-level storage period; (c). resistivity variations
127
3.2 Detecting the potential sliding surface along a spillway

An off-stream reservoir in southwestern Taiwan was constructed
during 1979~1987. The 1,535-m long earth dam, a height of 28
meters, is composed of selected impervious material core, compact
impervious material, riprap cobble and rockfill, and downstream
side gravel and sand filter. During the construction period, the dam
engineers found that the main dam foundation crossed the old Spillway
abandoned river channel. The soft ground could cause differential
settlement, initiate extra seepage, and undermine the stability and
safety on the main dam. In order to prevent from such damages, an
extra ground improvement on the dam foundation led to postpone
the time of completion more than 5 years.
A post-earthquake survey in 1999 indicated that the seepage
volume and turbidity had significant changes in nearby wells. The
post-event geological borehole exploration was conducted in the
dam safety inspection and engineers started to design the waterstop
grouting for mitigating the seepage through the dam (NCTU, 2005).
The recordings in inclinometers and settlement points also revealed WH-1
local deformation in the dam slope. A saturation condition was
frequently found on the dam rockfill toe. All of these evidences
implied that the seepage even piping (interior erosion) could exist in
this dam.
Its accessary concrete spillway is located at the west minor dam
as shown in Figure 6. Unfortunately, shallow sliding is frequently Line WH-1S
reported all the way around the spillway (NCTU, 2005). Figure 7
shows on-site slope failure images and temporary repair with plastic
canvases preventing from more surface erosion. The shallow Spillway
surface sliding removes more than 2-m deep top soils right beside
the spillway culvert walls. The soil lose also spreads to the substrate
soils under the spillway. Lack of substrate support leads to severe
cracking on the spillway bottom. Substrate refill and retrofitting
construction on the channel bottom are conducted for recovering the
original function of the spillway.
Figure 7 Slope sliding along the spillway structure
The around 19-m long geological column at borehole WH-1 is

shown in Figure 8. The geomaterials in the dam mass from the
ground surface are composed of top soil, weathering sandstone,
loose sandstone, and loose sandstone with mudstone, in series.
Figure 6 Top view photo of reservoir and dam (modified from

Google Map)
In order to know the possible sliding surface, the ERT-based

brine tracing technique is introduced to this manner. Two borehole
positions and two ERT investigation lines are also plotted in Figure
6. The borehole positions, including WH-1 and WH-2, are set at the
dam-top roadside grass. A 56-m long ERT survey Line WH-1S is
extended downward from borehole WH-1 and measures the sliding
mass conditions directly (see Figures 6 and 7). Another 70-m long Figure 8 Geological column at borehole WH-1
ERT survey Line WH-2S is extended from borehole WH-2
downward along the dam slope surface covered with bush and grass. The complicated 38-m deep column at borehole WH-2 is shown
Such an investigation pattern can not only provide the detailed in Figure 9. The geomaterials in the borehole log from the ground
geological information from the borehole samples but also transfer surface to its exploration depth are, in sequence, composed of
these two empty boreholes into the brine injection pipes for mudstone, loose sandstone with mudstone, loose sandstone,
following ERT tracing. mudstone, loose sandstone/mudstone (interlayered), mudstone, loose
sandstone with mudstone, loose sandstone, mudstone, loose
128
sandstone, and mudstone. The permeable and impermeable layers Figure 11 indicates the resistivity background image measured
alternatively occur throughout the dam mass from the ground along survey Line WH-1S. The dotted and dashed lines represent
surface. the layer boundaries defined by both ERT and borehole sampling
and borehole sampling only, respectively. The ERT background
image consists with the findings at borehole WH-1. The top soil
and weathering sandstone layers correspond to the light blue and
green zones, scattering from the ground surface to depths of 3~5 m.
The yellow, light green, and orange zones could indicate the
sandstone, which overlays the sandstone with mudstone, a relatively
low resistivity value zone.
Figure 12 represents the resistivity variation images (more than -
5%) along survey Line WH-1S up to 840 minutes. The resistivity
variations for brine tracing measure the resistivity differences
between the background and the influence from brine transmission
at specific moment. Infusing brine water could lead to the decrease
of resistivity values in the dam slope when it seeps through more
pervious layers. At the first 60 minutes, the intensive variations
occur at the depths of 2 m on the dam slope around the injection
borehole WH-1. Up to 240 minutes, the brine seepage flows
downward along the permeable sandstone layer. A significant brine
accumulation zone, marked as A, steeply seats upon a relatively
impervious layer. In future, this would possibly develop as a
potential sliding surface, about 3~5 meters deep, around the middle
part of the dam slope when extreme rain occurs. After 14 hours, the
brine water moves downward to the slope toe, labelled as B. This
could be another potential sliding surface.
Figure 13 shows the resistivity background image measured
along survey Line WH-2S. The dotted and dashed lines also
represent the layer boundaries defined by both ERT and borehole
sampling and borehole sampling only, respectively. The ERT
background image also consists with the findings at borehole WH-2.
The extremely low resistivity mudstone layer lays over the relatively
high resistivity sandstone layer, corresponding to the yellow zones.
Another mudstone layer overlays the sandstone interlayered with
mudstone, which covers over a mudstone layer.
Figure 14 presents the resistivity variation images (more than -
5%) along survey Line WH-2S up to 840 minutes. The resistivity
variations for brine tracing indicate the resistivity differences
between the background and the influence from brine transmission
at specific moment. Within the first 240 minutes, the relatively high
variations (-20% maximum) occur at the sandstone and mudstone
Figure 9 Geological column at borehole WH-2 interlayer, around at the elevations of 85~100 m and measurement
positions of 15~30 m. As the infusion time increases, the resistivity
Figure 10 shows the ERT-based brine tracing technique image at variation range expands outward, but limited. However, when
borehole WH-1. The brine water is continuously infused into the stopping injecting brine water into the dam mass (14 hours later),
dam mass through the borehole PVC pipe from water tanks. The the expansion phenomenon on the resistivity variation image stops
ERT instrument records the resistivity images before brine infusion and vanishes synchronously. This possibly reveals that the brine
and traces the resistivity variation images up to 840 minutes. water or groundwater is affected with near-borehole diffusion only.
The coverage on resistivity variation depends upon continuous brine
infusion. There is no significant direction or accumulation zone on
the seepage or groundwater flow along investigation Line WH-2S
installed on the dam slope. In fact, the dam slope is covered with
broomy bushes and grass. There is no significant scour or sliding
scars along survey Line WH-2S.
In brief, observing the brine seepage flow could rapidly reflect
the possible groundwater or leakage flow path in specific geological
layers and accumulation zone based on the ERT inspection
technique. Tracing the brine path can also provide the potential
sliding surface information on an earth dam mass in few hours.
Figure 10 ERT-based brine tracing conducted at borehole WH-1
129
WH1S-BK
WH-1
110
Top Soil+
Weathering Sandstone
105
Elevation (m)
100
Sandstone A
95
B Layer boundary defined by ERT and borehole

90 Sandstone with Mudstone
Layer boundary extended from borehole
Mudstone
85
0 5 10 15 20 25 30 35 40 45 50 55
5 2 35 50 65 90 12 15 20 26 32 45 80 140 ohm×
Distance (m) 0 Mudstone 0 0Sandstone
0 0 0 0 0 0
m
Wet Dry
Figure 11 ERT background image along investigation Line WH-1S
WH01-30-BK WH01-60-BK
105 105
Elevation (m)
Elevation (m)
100 100
95 95
90 90
85 85
0 5 10 15 20 25 30 35 40 45 50 55 0 5 10 15 20 25 30 35 40 45 50 55
Distance (m) Distance (m)
(a). ERT variation image after 30-min brine tracing (b). ERT variation image after 60-min brine tracing
WH-1 WH01-120-BK WH01-180-BK
105 105
Elevation (m)
Elevation (m)
100 100
95 95
90 90
85 85
0 5 10 15 20 25 30 35 40 45 50 55 0 5 10 15 20 25 30 35 40 45 50 55
(c). ERT variation image after 120-min brine tracing (d). ERT variation image after 180-min brine tracing
WH01-240-BK 840min-BK
105 105
Elevation (m)
Elevation(m)
100 100
95 95
90 90
85 85
0 5 10 15 20 25 30 35 40 45 50 55 5 10 15 20 25 30 35 40 45 50 55
Distance(m) Distance (m)
(e). ERT variation image after 240-min brine tracing (f). ERT variation image after 840-min brine tracing
-70 -60 -50 -40 -30 -20 -10 -2 % Resistivity Variation
Figure 12 ERT-based seepage brine tracing image along investigation Line WH-1S (variation more than -5% only)
130
WH2S-BK
WH2 Mudstone
115
Sandstone
110
105
Mudstone
Elevation (m)
100
95 Sandstone/Mudstone 150
45 Layer boundary defined by ERT and borehole
90 50 60
85 Mudstone Layer boundary extended from borehole
80
5 2 35 50 65 90 12 15 20 26 32 45 80 140 ohm×
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70
0 Mudstone 0 0Sandstone
0 0 0 0 0 0
m
Distance (m) Wet Dry
Figure 13 ERT background image along investigation Line WH-2S
WH2-30-BK WH2-60-BK
110 Elevation (m) 110

Elevation (m)
105 105
100 100
95 95
90 90
85 85
80 80
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70
WH2-120-BK WH2-180-BK
110 110
Elevation (m)
Elevation (m)
105 105
100 100
95 95
90 90
85 85
80 80
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70
WH2-240-BK WH2-840-BK
110 110
Elevation (m)
Elevation (m)
105 105
100 100
95 95
90 90
85 85
80 80
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70
Figure 14 ERT-based seepage brine tracing image along investigation Line WH-2S (variation more than -5% only)
131
3.3 Brine tracing seepage in a dam levee widening mass

The accessary dam levee widening is designed for increasing the
reservoir capacity described in section 3.2. Knowing the seepage
conditions, including position and direction, can provide appropriate
countermeasures for decreasing the seepage degree through the dam
mass. In response to the engineering need, borehole exploration and
ERT-based brine tracing technique are introduced to this seepage
investigation at the dam levee widening slope.
The borehole position and two ERT survey lines are plotted in
Figure 15. The borehole position is chosen at the downstream side
slope adjacent to the dam levee top. A 56-m long survey Line WH-
3H is set along the dam-top road and passes through borehole WH-
3. Perpendicular to survey Line WH-3H, another ERT survey Line
WH-3V is extended from borehole WH-3 downward along the dam
slope surface. Using such an investigation pattern can not only
identify the geological information from the borehole sample but
also transfer the empty borehole into the brine injection pipe for
ERT tracing.
Figure 16 Geological column at borehole WH-3
Figure 15 Borehole position and survey lines on a dam levee

widening slope
The 30-m long geological column at borehole WH-3 is shown in

Figure 16. The dam levee widening from the ground surface down
to 30 m deep is composed of covering concrete, loose sandstone,
loose sandstone/mudstone (interlayered), loose sandstone,
mudstone, loose sandstone with mudstone, loose sandstone, loose Figure 17 ERT-based brine tracing conducted at borehole WH-3
sandstone with shale, mudstone and loose sandstone. The
permeable and impermeable layers alternatively occur throughout Figures 20 and 21 also show the background and variation
the dam levee widening mass. resistivity images, respectively, along survey Line WH-3V. The
Figure 17 shows the on-site ERT-based brine tracing technique dotted and dashed lines represent the layer boundaries defined by
image at borehole WH-3. The brine water is continuously infused both ERT and borehole sampling and borehole sampling only,
into the dam mass from the borehole pipe with water tanks. The respectively. The background image also consists with the findings
ERT instrument records the resistivity images before brine infusion at borehole WH-3. Similarly, two loose sandstone layers
and traces the resistivity variation images for 400 minutes. correspond to the pink, red, orange, or yellow zones, scattering
Figures 18 and 19 present the resistivity background and surface to depths of 2~5 m and depths of 8~16 m, which overlays on
variation images, respectively, along Line WH-3H. The dotted and mudstone. In-between, the relatively low resistivity belt indicates
dashed lines represent the layer boundaries defined by both ERT and the sandstone interlayered with mudstone. The resistivity variations
borehole sampling and borehole sampling only, respectively. The for brine tracing measure the image difference between the
background image consists with the findings at borehole WH-3. background and brine infusing at specific moment (Figure 21). In
Two loose sandstone layers correspond to the pink, red, orange, or Figure 21, significant variation zones are identified on the pervious
yellow zones, scattering surface to depths of 2~3 m and depths of sandstone layer. The seepage moves downward to the slope toe in
6~13 m. In-between, the relatively low resistivity belt indicates the the SE-NW direction. The resistivity variations only exist within
sandstone interlayered with mudstone. The resistivity variations for the pervious sandstone layer and no significant resistivity aberration
brine tracing measure the image difference between the background is found in two mudstone layers as the infusion time increases. This
and brine infusing at specific moment (Figure 19). The resulting indicates that the seepage or groundwater belongs to inter-stratum
variations are still not significant after 300-min tracing. This water flow and is almost sealed within two relatively impervious
indicates that the main seepage does not transmit along the mudstone layers in this investigation coverage.
horizontal direction, i.e., parallel to the dam levee top road.
132
WH3H-BK
110
WH-3
Sandstone
105
Elevation (m)
Sandstone/Mudstone
100
Layer boundary defined by ERT and borehole
Sandstone Sandstone
95 Layer boundary extended from borehole
Mudstone
90 5 2 35 50 65 90 12 15 20 26 32 45 80 140 ohm×
10 15 20 25 30 35 40 45 50 55
0 0 0 0 0 0
m
Figure 18 ERT background image along investigation Line WH-3H
WH3-30-BK WH3-60-BK
105 105
E levation (m )
100 E levation (m ) 100
95 95
90 90
10 15 20 25 30 35 40 45 50 10 15 20 25 30 35 40 45 50
WH3-120-BK WH3-180-BK
105 105
E le v a tio n (m )
E le v a tio n (m )
100 100
95 95
90 90
10 15 20 25 30 35 40 45 50 10 15 20 25 30 35 40 45 50
WH3-240-BK WH3-300-BK
105 105
E le v a tio n (m )
E le v a tio n (m )
100 100
95 95
90 90
10 15 20 25 30 35 40 45 50 10 15 20 25 30 35 40 45 50 55
Figure 19 ERT-based seepage brine tracing image along investigation Line WH-3H (variation more than -2% only)
133
WH-3V-BK
110
WH-3
105
Elevation (m)
100
Sandstone
95
Sandstone/Mudstone
90 Sandstone Layer boundary defined by ERT and borehole
Sandstone Layer boundary extended from borehole
85
80
Mudstone 5 2 35 50 65 90 12 15 20 26 32 45 80 140 ohm×
0 5 10 15 20 25 30 35 40 45 50 55
0 0 0 0 0 0
m
Figure 20 ERT background image along investigation Line WH-3V
WH3V-90-BK WH3V-150-BK
105 105
Elevation (m)
Elevation (m)
100 100
95 95
90 90
85 85
80 80
0 5 10 15 20 25 30 35 40 45 50 55 0 5 10 15 20 25 30 35 40 45 50 55
105 105
Elevation (m)
100
Elevation (m)
100
95 95
90 90
85
85
80
80
0 5 10 15 20 25 30 35 40 45 50 55
0 5 10 15 20 25 30 35 40 45 50 55
105 105
Elevation (m)
Elevation (m)
100 100
95 95
90 90
85 85
80 80
0 5 10 15 20 25 30 35 40 45 50 55 0 5 10 15 20 25 30 35 40 45 50 55

Figure 21 ERT-based seepage brine tracing image along investigation Line WH-3V (variation more than -2% only)
134
4. CONCLUSIONS Hung, J.-J. (1991). An Outline of Elementary Engineering Geology,

Sino-Geotechnics Research and Development Foundation,
The ERT-based leakage tracing technique is applied to investigate Taipei, Taiwan. (in Chinese)
the potential seepage or groundwater flowing through earth dams in Liming Engineering Consultants (Liming) (2014). Safety Inspection
Taiwan. The conclusions are drawn as the followings: and Monitoring of Agongdian Reservoir in 2013, Report to
1. Under the different reservoir levels, a high water-level storage in Southern Region Water Resources Office, Water Resources
a dam provides a higher water head as a more intensive surcharge Agency, Ministry of Economic Affairs, Kaohsiung, Taiwan.
source. Significant variations on the resistivity space distribution (in Chinese)
are feasibly measured for tracing aberration leakage position on Loke, M. H. (2000). Electric Imaging Surveys for Environmental
the dam mass. and Engineering Studies— A Practical Guide for 2-D and 3-
2. The pervious silty sand is reasonably identified as the potential D Surveys, ABEM Instrument AB, Sunbyberg, Sweden.
leakage section, which is located at the old river channel when National Chiao Tung University (NCTU) (2005). Review of
constructing an earth dam. The previous river meandering causes Monitoring Problems and Non-Destructive Testing Methods
deeper disturbance on the silty sand feature. for Evaluation Dam Performance, Report No. MOEA-WRA-
3. A downhole ERT-based brine tracing technique plays an active 0940206, Water Resources Agency, Ministry of Economic
role to continuously detect the seepage flow path through the dam Affairs, Taichung, Taiwan. (in Chinese)
mass. Like a penetrant, the infused brine water will flow in the Schuster, R. L. and Krizek, R. J. (1978). Landslides— Analysis and
most possible leakage or groundwater path. Monitoring its ERT Control, Transportation Research Board Special Report 176,
distribution variations can effectively identify the leakage spatial Washington, D.C., U.S.A.
distribution and the major flow direction in an earth dam mass. Sinotech Engineering Consultants, Ltd. (Sinotech) (2010). Safety
4. Using brine path tracing method, one can rapidly identify the Inspection and Monitoring of Agongdian Reservoir in 2009,
accumulation zone of seepage in more permeable geological Report to Southern Region Water Resources Office, Water
layers and its potential sliding surface on dam slopes in few hours. Resources Agency, Ministry of Economic Affairs,
In future, dam engineers can use the ERT-based brine tracing Kaohsiung, Taiwan. (in Chinese)
technique to effectively sieve the possible leakage positions and Society of Exploration Geophysicists of Japan (2014). Application
take appropriate countermeasures preventing from further Manual of Geophysical Methods to Engineering and
deterioration. Environmental Problems, European Association of
Geoscientists & Engineers (EAGE), Houten, the Netherlands.
5. REFERENCES Taiwan Geotechnical Society (TGS) (2011). The Cause
Chiou, S.-J., Kao, H.-C., Tsay, M.-S., and Cheng, J.-Y. (2010) Investigation Report on Landslide at Mileage 3.1-km of
“Preliminary study on the application of QGIS in dam safety Freeway No.3, A Forensic Investigation Report to Ministry of
monitoring and inspection”, Proceedings of the 29th Survey Transportation and Communications, Taipei, Taiwan. (in
and Spatial Information Conference, Taipei, Taiwan. (in Chinese)
Chinese) Wightman, W. E., Jalinoos, F., Sirles, P., and Hanna, K. (2003).
Application of Geophysical Methods to Highway Related
Problems, Report No. FHWA-IF-04-021, Central Federal
Lands Highway Division, Lakewood, Colorado, U.S.A.
135
Dynamic p-y Curves for a Single Pile by 1g Shaking Table Tests

S.S. Jeong1 and H.S. Lim1
1
Department of Civil and Environmental Engineering, Yonsei University, Seoul, Republic of Korea
E-mail: soj9081@yonsei.ac.kr
ABSTRACT: In this study, dynamic p-y curves were suggested for pseudo-static analysis from 1g shaking table tests under various loading
frequencies. Based on the results of the model tests, the dynamic p-y curves are highly dependent on the relationships between the natural
frequency of the soil-pile systems and the loading frequency. The dynamic p-y curves were proposed as hyperbolic functions. The ultimate
soil resistance and the initial soil stiffness were developed as functions of the properties of the pile and soil. Based on comparisons with the
existing p-y curves, the proposed dynamic p-y curve was demonstrated to be capable of predicting the behaviour of a pile under dynamic
loads.
Keywords: Dynamic p-y curve, loading frequency, initial soil stiffness, ultimate soil resistance, Pseudo-static analysis.
1. INTRODUCTION variable differential transformers) located on either side of the pile

cap. In other to obtain the bending moments, strain gauges were
Seismic design has become increasingly important, because the installed along the pile. Accelerometers were installed on the
number of large-scale earthquakes has increased worldwide to superstructure and within the soil of to measure the natural
reduce the threat that earthquake-induced structural deformation and frequencies of the soil-pile system and the free field displacement.
damage. Pseudo-static analysis, which is a method of converting Sweep tests were conducted to evaluate the natural frequencies
dynamic loads into the equivalent static loads, is widely used in the of the soil-pile systems under various conditions. The input wave
seismic design of pile foundations. The p-y curves considering was as a sine wave and applied approximately 5 seconds. The
nonlinear behaviour of soil are most frequently used for pseudo- loading frequencies used in the tests were calculated from the
static analysis in practical engineering applications. However, the p- natural frequencies (fn) of the soil-pile systems. The natural
y curves, which were proposed by Reese et al. (1974) and API frequencies range from 0.4fn Hz to 1.6fn Hz. The amplitude
(American Petroleum Institute, 1987), derived from field tests accelerations ranged from 0.098g to 0.4g.
applying static and cyclic loads at the pile head. In other words, the
p-y curves do not properly consider soil stiffness and soil inertia
Unit : cm
effects under seismic loads. Therefore, the p-y curves are not : Strain gauge
: Accelerometer
suitable for application to seismic design of pile foundations. 3
: LVDT
Much work has been done on pile foundations subjected to G.L. 0
dynamic loads. Ting et al. (1987) noted that the secant slope of the G.L. -4 (2D) Dr 40, 80 %
dynamic p-y curve is highly dependent on the loading frequencies in G.L. -8 (4D)
Sponge G.L. -12 (6D)
dynamic pile load tests. Dou and Byrne (1996) showed that API p-y
t=5 G.L. -16 (8D)
curves underestimate the lateral soil stiffness or resistance under G.L. -20 (10D)
strong shaking intensities. The NCHRP (National Cooperative Pile
D = 2, t = 0.2
Highway Research Program, 2001) described that the dynamic G.L. -30 (15D) 70
behavior of the soil-pile interaction is closely associated with the
pile diameters, shear wave velocities of the soil and loading
frequencies. Additionally, the NCHRP suggested dynamic p-y G.L. -44 (22D)
curves by using a numerical analysis method that related the static

p-y curves to the dimensionless frequency. However, the
verification of the dynamic p-y curves was conducted under G.L. -59 (29.5D)
3
restrictive conditions by statnamic tests in which a lateral load was 9
applied on a pile head. El Naggar and Bentley (2000) observed that
the soil resistance under dynamic loading increases due to the Shacking direction
damping effect. They also noted that the dynamic p-y curves
depended on the loading frequency. Therefore, it is necessary to Figure 1 Sectional view of 1g shaking table test
study dynamic p-y curves for seismic design of pile foundations.
The objective of this study is to develop dynamic p-y curves for Table 1 Pile Properties
pseudo-static analysis by using model tests under various loading
frequencies based on the natural frequencies of soil-pile systems. Scaling factor
Pile Prototype Model pile
(Iai, 1989)
2. 1G SHACKING TABLE TESTS Diameter (cm) λ 91.44 2.0
Thickness (cm) λ 1.4 0.2
2.1 Test set-up and programs Pile depth (m) λ 1,710 64
The dynamic model tests were conducted by 1g shaking table tests. Flexural rigidity
λ4.5 842,983 0.31995
The size of the soil box was 1,200×600×800 mm with sponge pads (kN·m2)
50 mm thick on the sidewalls to reduce reflection waves during *Note: Scaling factor λ = 26.72
shaking (Fig. 1). The model pile was made of aluminium alloy with
a hollow circular section. The size of the model pile was applied the
similitude law proposed by Iai (1989). The properties of the model Table 2 Properties of Test Soil
pile are summarized in Table 1. Jumoonjin sand, characterized as
D10 D60 γd,max γd,min
clean and uniform sand, was used in the tests. The properties of USCS Cu Gs
(mm) (mm) (t/m3) (t/m3)
jumoonjin sand are listed in Table 2.
Fig. 1 shows the sectional view of 1g shaking table tests. The SP 0.38 0.49 1.59 2.65 1.62 1.36
displacement of the pile was measured by two LVDTs (linear
136
2.2 Determination of experimental dynamic p-y curves 14
The experimental dynamic p-y curves were derived from the 12
Natural frequency (Hz)

bending moments along the depth of the pile. The bending moment 10
profile along the pile was double differentiated and integrated to the
8
obtain soil resistance (p) and lateral pile deflection (ypile) according
to simple beam theory, as shown in Eq. (1) and Eq. (2): 6
4
d 2M ( z) Dr 80%
p (1) 2
dz 2 Dr 40%
0
M ( z)
y pile   (2) 0 0.1 0.2 0.3 0.4 0.5
dz Acceleration amplitude (g)
EI
Figure 3 Natural frequency of soil-pile systems at various
where p is the soil resistance, M(z) is the bending moment at depth acceleration amplitudes
z, ypile is the lateral pile deflection, EI is the flexural rigidity of the
pile and z is the distance along the pile. 60
G.L .-12cm(6D)
The bending moment of the pile foundation were calculated by
Eq. (3) using the stain measured by the strain gauges. 40 G.L.-8cm(4D)
)
m
EI (3)
/c 20 G.L.-4cm(2 D)
M  N
(
ec
y na
ts 0
is
where M is the bending moment, y is the distance from the neutral er
li -20
axis, E is the Elastic modulus of the pile and ε is the strain of the o
S G.L.-4 cm (2D)
pile.
G.L.-8 cm (4D)
-40
G.L.-1 2cm (6 D)
2.3 Test results
-60
The natural frequencies of the soil-pile systems were determined -0.1 -0 .0 5 0
Pile deflection (cm)
0 .0 5 0 .1
from sweep tests. Fig. 2 shows the typical fourier transform for a (a)
natural frequency of a soil-pile system using sweep tests results. As
shown in Fig. 2, the natural frequency of this soil-pile system is 80
10.86 Hz under an acceleration amplitude of 0.154 g in dense sand. 0.22g 0.3g
60
Fig. 3 shows change in the natural frequency with the 0.154g
0.4g
acceleration amplitude of the soil-pile systems. As the acceleration ) 40 0.098g
m
amplitude increases the natural frequency of the soil-pile systems /c
N
( 20
decreases due to the reduced elastic modulus of the soil. In addition, ec
the natural frequency of the soil-pile system for the relative density na 0
ts
of 40 %, is smaller than that for the relative density of 80 % due to is
re-20 0.4g
the increased stiffness in the soil with a higher relative density. li
oS 0.3g
Fig. 4 shows dynamic p-y curves under various conditions. As -40
0.22g
shown in Fig. 4(a), the dynamic p-y curves are flat near the surface -60 0.154g
and become stiffer with depth because the soil confining stress 0.098g
-80
increases. Fig. 4(b) shows that the lateral pile deflection increases -0.3 -0.2 -0.1 0 0.1 0.2 0.3
with the acceleration amplitude due to greater inertial forces. The Pile deflection (cm)
soil resistance also increases with the acceleration amplitude. (b)
However, the increments of the soil resistance decreased with the
acceleration amplitude. The secant slope of the dynamic p-y curve 80
decreases with an increase in the acceleration amplitude. As shown 60

1.2fn 1.1fn
in Fig. 4(c), The dynamic p-y curve is closely related to the ratio of 0.9fn 1.0fn
the loading frequency to the natural frequency. As the frequency )

40 0.8fn
ratio approaches 1.0, the lateral deflection further increases, and the m
/c 20
slope of the dynamic p-y curve decreases due to resonance. N
(
ec
na 0
0.3 ts
0.154g, 10.86Hz sie -20
r
il 1.2fn
o -40
A mplitude ra tio
0.2 S 1.1fn
1.0fn
-60 0.9fn
0.8fn
0.1 -80
-0.3 -0.1 0.1 0.3
0 (c)
0 5 10 15 20 25 30
Frequency (Hz) Figure 4 Experimental dynamic p-y curves: (a) with different depth
Figure 2 Typical Fourier transform for natural frequency of soil-pile (b) with various acceleration amplitudes (c) with various loading
system frequencies
137
3. PROPOSED DYNMAIC P-Y CURVES 3.1 Ultimate soil resistance

Based on the experimental dynamic p-y curves, the simplified Kim et al. proposed the ultimate soil resistance based on Rankine’s
dynamic p-y backbone curve was suggested for pseudo-static passive earth pressure. The best-fit ultimate soil resistance was
analysis under various loading frequencies. All of the peak points of determined by using Eq. (4).
the experimental dynamic p-y curves corresponding to the pu
maximum soil resistance, were collected at several depths and  AK p ' z n (4)
D
plotted on a p-y plane. The dynamic p-y backbone curves were fitted
to this data. where D is pile diameter, Kp is Rankine’s passive pressure
In this study, the general shape of experimental dynamic p-y coefficient, is effective unit weight, z is depth of soil and A, n are
curves for dry sand were fit mathematically by a hyperbolic function. curve fitting parameters. In order to determine the curve fitting
The best fit curve was the hyperbolic function of Kondner (1963): parameters A and n, Eq. (4) was normalized in logarithm as Eq. (5):
y (4) p
p log( u ) = n log( z ) + log( AK p γ' ) (5)
1 y D

K pu Fig. 6 shows the soil resistance in transformed coordinates from
where p is the soil resistance, y is the lateral pile deflection, K is the the model test results. From the linear regression analysis, empirical
initial soil stiffness and pu is the ultimate soil resistance. Generally, equations for the soil resistance were obtained as Eqs. (6) and (7):
the shape of the hyperbolic curve is controlled by the value of pu and The soil resistance for dense sand:
K. Therefore, these values were proposed from model tests. pu
Figs. 5 show the peak points of the experimental dynamic p-y = 7.10K p γ' z1.97 (6)
D
curves and the dynamic p-y backbone curves, which were
determined by regression analysis. As shown in Fig. 5, the results The soil resistance for loose sand:
indicate nonlinearity of the soil-pile system, and the hyperbolic pu
= 8.21K p γ' z1.88 (7)
function accurately reflects the experimental results. Additionally, D
the initial slope (K) and the ultimate soil resistance (pu) increased
with depth due to confining stresses of soil. In this study, the initial 2
soil stiffness and the ultimate soil resistance were suggested to be a 1.8 Dense sand (Dr 80%)
Y = 1.9676X -0.2932
function of the confining pressure and the Rankine’s passive 1.6 Y = log(p u /D), X = log(z)
pressure coefficient, respectively. 1.4 R2 = 0.9776
log(p u /D)
1.2
50
1
45 Dynamic p-y backbone curve at 2D depth
Measured data at 2D depth 0.8 Loose sand (Dr 40%)
40
Soil resistance (N/cm)
0.6 Y = 1.8791X -0.4472

35
Y = log(pu /D), X = log(z)
30 0.4 R2 = 0.9995
Loose sand (Dr 40%)
25 0.2
Dense sand (Dr 80%)
20
R2 = 0.9265 0
15
0 0.2 0.4 0.6 0.8 1 1.2
log(z)
10
5
Figure 6 Determination of ultimate soil resistance
0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 3.2 Initial soil stiffness
(a) The variation of the initial tangent modulus with confining pressure
50 is represented by an empirical equation proposed by Janbu(1963):
45
σ'
40 R2 = 0.9277 K = APa ( ) 0.5 (8)
Soil resistance(N/cm)
35 Pa
30
25 where Pa is atmospheric pressure (10.13 N/cm2), is confining stress
20 and A is a curve-fitting parameter.
15 A linear regression analysis was used to obtain the best-fit
10
Dynamic p-y backbone curve at 4D depth values for the parameter A.
5
0
Measured data at 4D depth Fig. 7 shows the initial soil stiffness in the transformed
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 coordinates from the model test results. Through a linear regression
analysis, the curve-fitting parameters of A were determined to be
(b)
50
1,414.8 and 995.94 for the relative densities of Dr 80 % and Dr 40
45 %, respectively. Based on these results, the empirical equation for K
40
R2 = 0.9750 can be rewritten as Eqs. (9) and (10).
Soil resistance(N/cm)
35 The initial soil stiffness for dense sand:

30 σ'
25 K = 1414 .8 Pa ( ) 0.5 (9)
20
Pa
15
10
The initial soil stiffness for loose sand:
Dynamic p-y backbone curve at 6D depth
5
Measured data at 6D depth σ' 0.5
0 K = 995.94 Pa ( ) (10)
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 Pa
(c)
Figure 5 Dynamic p-y backbone curve at dense sand: (a) at 2 pile
diameter; (b) at 4 pile diameter; (c) at 6 pile diameter
138
2500 Bending moment (N·cm)

Loose sand (Dr 40%)
-1000 0 1000 2000 3000 4000 5000
Dense sand (Dr 80%)
0
2000
Dense sand (Dr 80%) 10

K (N/cm2 )
1500 Y = 1414.8X
Y = K, X = P a(σ/P a)0.5
20
R2 = 0.9240
1000
Loose sand (Dr 40%)
30
Y = 995.94X
Depth (cm)
500 Y = K, X = P a(σ/P a) 0.5
R2 = 0.9632 40
0 50
0 0.5 1 1.5 2
Pa(σ/Pa)0.5 60
Figure 7 Determination of initial slope
70 Measured data
3.3 Application of proposed dynamic p-y curves 80 Reese p-y curve
API p-y curve

To verify the applicability of the proposed p-y curve more reliably, 90
other test results were compared with proposed dynamic p-y curves. Proposed p-y curve
Yang (2008) has conducted 1g shaking table tests in dry loose and (b) 100
dense sand. The test soil was Jumoonjin sand and the model pile Figure 7 Comparison of bending moment distributions results of
with hollow circular section was embedded in a soil box (1,800×600 Yang(2008) for application of proposed p-y curves
×1,200 mm) and the embedded depth of pile was 110 cm with a
diameter of 3.2 cm. An input sine wave of acceleration amplitude
0.154 g and frequency 6 Hz applied at the bottom of the soil box. 4. CONCLUSION
The proposed dynamic p-y curves were validated by pseudo- In this study, a new hyperbolic p-y curve was proposed under
static analysis with the FB-MultiPier program (2007). The analysis dynamic loads for piles embedded in dry sand that can apply to
was performed by applying the inertial force determined by the pseudo-static analysis. The experimental results confirmed that
measured acceleration at the surcharge mass and the surcharge mass. dynamic p-y curves are largely affected by the relationships between
Fig 7. shows the predicted and measured bending moment the natural frequency of the soil-pile systems and the loading
distributions in test piles. The resulting proposed p-y curves predict frequencies, the acceleration amplitude. Based on comparisons with
a trend of the measured bending moment distributions better than the existing p-y curves, the proposed dynamic p-y curve is shown to
the existing p-y curves. be capable of predicting the behavior of the pile under dynamic
loads. Therefore, the proposed p-y curves are appropriate to
Table 4 Pile Properties and Test Conditions (Yang, 2008) represent the behavior of the pile subjected to dynamic loads.
Pile Model Test Condition
pile 5. REFERENCES
Diameter (cm) 3.2 American Petroleum Institute (API). (1987) Recommended practice
Soil condition Dr 80%, 40% for planning, designing and constructing fixed offshore
Thickness (cm) 0.5 platforms, API Recommended Practice 2A (RP 2A), 17th
edition, pp 68-71
Frequency Bea, R.G., Littom, R., Nour-Omid, S., and Chang, J.Y. (1984) "A
Pile depth (cm) 110 6
(Hz) specialized design and research tool for the modelling of
Flexural rigidity Acceleration near-field soil interactions", 16th Offshore Technology
2764,424 0.154
(kN·cm2) (g) Conference, Houston, pp249-252.
Boulanger, R. W., Curras, C. J., Kutter, B. L., Wilson, D. W., and
Abghari, A. (1999) "Seismic soil-pile-structure interaction
Bending moment (N·cm)
-1000 1000 3000 5000 7000
experiments and analyses", Journal of Geotechnical and
0 Geoenvironmental Engineering, 125, Issues 9, pp750-759.
Broms, B. B. (1964) " Lateral resistance of piles in cohesive soils ",
10
J. soil mech. Fdns. Div. ASCE, 90, pp750-759.
20 Dou, H. and Byrne, P.M. (1996) "Dynamic response of single piles
and soil-pile interaction", Canadian Geotechnical Journal, 33,
30 Issues 1, pp80-96.
Depth (cm)
El Naggar, M. H. and Bentley, K. J. (2000) "Dynamic analysis for

40
laterally loaded piles and dynamic p-y curves", Canadian
50 Geotechnical Journal, 37, Issues 6, pp1166-1183.
Hoit M, Hays C, McVay M, Williams M. (2007) The FB-MultiPier
60 v4 program, Bridge Software Institute.
70
Iai, S. (1989) "Similitude for shaking table tests on soil-structure-
Measured data
fluid model in 1g gravitational field ", Soils and Foundations,
80 Reese p-y curve 29, Issues 1, pp105-118.
API p-y curve
Janbu, N. (1963) " Soil compressibility as determined by oedometer
90
and triaxial test", Proceedings of the European Conference on
Proposed p-y curve
100
Soil Mechanics and Foundations Engineering, Wiesbaden,
(a) pp19-25.
139
Kim, Y. H., Jeong, S. S. and Lee, S. J. (2011) "Wedge failure Reese, L. C., Cox, W. R. and Koop, F. D. (1974) " Analysis of
analysis of soil resistance on laterally loaded piles in clay", laterally loaded piles in sand", Proc. of the 6th Offshore
Journal of Geotechnical and Geoenvironmental Engineering, Technology Conference. Houston, Texas, pp473-484.
ASCE, 137, Issues 7, pp53-77 Rovithis, E., Kirtas, E., and Pitilakis, K. (2009) "Experimental p-y
Kim, Y. H., Jeong, S. S. and Won, J. O. (2009) "Effect of lateral loops for estimating seismic soil-pile interaction", Bulletin of
rigidity of offshore piles using proposed p-y curves in marine Earthquake Engineering, 3, Issues 7, pp719-736.
clay ", Journal of Marine Geosourses and Geotechnology, 27, Thavaraj, T., Finn, W. D. L., and Wu, G. (2010) "Seismic response
Issues 1, pp53-77 analysis of pile foundation", Geotechnical and Geological
Kondner, R.L. (1963) " Hyperbolic stress-strain response: cohesive Engineering, 28, pp275-786.
soils", Journal of Soil Mechanics and Foundation Division, Yang, E. K. (2008) " Evaluation of dynamic p-y curves for a pile in
87, Issues 1, pp115-144. sand from 1g shaking table tests", Seoul, Seoul National
National Cooperative Highway Research Program (NCHRP) (2001) University [Ph. D dissertation].
Static and dynamic lateral loading of pile groups, NCHRP Yang, E. K., Kwon, S. Y., Choi, J. I. and Kim, M. M. (2010) "
Report 461, Transportation Research Board – National Prediction of the natural frequency of pile foundation system
Research Council 2001, pp13-21. in sand during earthquake", Journal of the Korean
Novak, M. (1974) "Dynamic stiffness and damping of piles", Geotechnical Society, 26, Issues 1, pp45-54 (in Korean).
Canadian Geotechnical Journal, 11, pp574-598. Yoo, M. T., Choi, J. I., Han, J. T. and Kim, M. M. (2013) "
Dynamic p-y curves for dry sand from centrifuge tests",
Journal of Earthquake Engineering, 17, pp1082-1102.
140
Design Method for Bottom Single Blade Steel Rotation Pile Foundation:
Case Study in Vietnam
L.D. Dao1 and T.T.T. Nguyen2
1
Department of Civil Engineering, University of Transport and Communications, Hanoi, Vietnam
E-mail: daoduylam@utc.edu.vn
ABSTRACT: Bottom single blade Steel Rotation Pile (SRP) is a large diameter steel pipe pile with a helical blade welded to the edge with a
lot of advantages such as rapid construction, small construction area, especially less vibration, less noise, environment friendly and high
resistance which has been applied recently in the urban of Vietnam. However, how to design SRP foundation in ensuring the conformity with
the specification for bridge design in Vietnam is an important question and the new Design Specification for SRP has been edited. This
article presents the way to find out the method to design SRP foundation in Vietnam as a case study.
Keywords: Steel Rotation Pile, bridge foundation, design method in Vietnam.
1. INTRODUCTION 2. LRFD DESIGN METHOD

Steel Rotation Pile (SRP) is a large diameter steel pipe pile with a 2.1 ASD and LRFD methods
helical blade welded to the edge. During construction, with a casing
rotator for example, the pile is rotated in pressing and the blade on The design of the foundations has been traditionally based on ASD
the edge performs the digging that drives the pile into the ground as method, safety is achieved in the foundation element by restricting
a screw. Since the 1990s, this kind of foundation has been applied the estimated loads (or stresses) to values less than the ultimate
successfully for many bridges in Japan and in the other countries, resistance divided by a safety factor (SF). In ASD all of these loads
especially in urban area based on the big advantages, such as rapid are assumed to have the same variability. As a result, load factors
construction, small construction area, especially less vibration, less are not applied on the load combinations considered for either the
noise, environment friendly and high resistance. strength or service limit states. The factor of safety is a number
greater than unity. The SF provides reserve strength in the event that
an unusually high load occurs or in the event that the resistance is
less than expected. For the Service Limit State, unfactored loads are
used to calculate deformations, and these deformations are
compared to the maximum tolerable values.
ASD method has many limitations: does not adequately account
for variability of loads and resistances, the SF is applied only to
resistance; selection of a SF is subjective, and does not provide a
measure of reliability in terms of probability of failure, etc.
Figure 1 Steel rotation pile To overcome these deficiencies, LRFD method [1, 2, 6] was
developed from the 1950. The idea is that:
Recently in Vietnam, this new kind of the foundation was
applied for Hoang Minh Giam flyover and will be used for Ring Resistance ≥ Effect of Loads
Road No.3 part Mai Dich - Thang Long of Hanoi (Figure 2).
And the resistance side is multiplied by a statistically-based
resistance factor, f (value is usually less than one), the load
components on the right side are multiplied by their respective
statistically based load factors, γi, (values are usually greater than
one).
The process of assigning values to resistance factors and load
factors is called calibration. A design code may be calibrated by
using:(1) judgment, (2) fitting to other codes, (3) reliability theory,
or (4) a combination of approaches.
Calibration by judgment requires experience and the
Figure 2 Hoang Minh Giam flyover and Ring Road No.3
fundamental disadvantage of this method of calibration is that it
results in non-uniform levels of conservatism. Calibration by fitting
In Vietnam, bridge foundation is now designed based on Load
to other codes such as ASD can be used where there is insufficient
and Resistance Factors Design (LRFD) method [1, 16, 17, 18] with
statistical data to perform a more formal process of calibration by
the Specification for bridge design 22TCN 272-05 in section 10 [1]
reliability theory. This method was used for AASHTO LRFD old
but this section shall apply only for the design of spread footings,
version before 1998 and for example, gave the values of resistance
driven piles and bored piles foundations.
factors from 0.7 to 0.56 fitting to the SF from 2 to 2.5. However, the
In the world, SRP design is mentioned in some specifications [7,
research [6, 11] so that LRFD factors converted from SF of ASD did
10, 15] such as JRA 2012 of Japan which uses Allowable Stress
not provide the desired level of reliability and NCHRP of U.S
Design (ASD) method. However as mention below, ASD method
recommend to use reliability theory for (applied for AASHTO
cannot be used for bridge design in Vietnam.
LRFD new version from 1998 until now).
Therefore, the new LRFD method for SRP foundation design is
Therefore, for the assignment of load and resistance factors in SRP
required to be established [19], the key content will be discussed in
design, the calibration by reliability method should be selected.
this paper: load factors, new resistance factors, new bearing formula.
141
2.2 Basic equation of LRFD method area, the coastal part of Japan, with the surface layer is soft soil,
very soft clay and very deep bearing layer (Figure 3).
In the bridge design specification [1, 2], the basic equation is: In the US, the typical geological bearing soils are mainly sand
  Q   R
i i i n  Rr (1) and gravel. However, relatively depths are not so big, mainly under
20m. In Europe, the bearing layer is mainly rocky and small depths,
in which: mainly less than 20m.
So in general, the geological conditions of Japan are rather
 i = load factor: a statistically based multiplier applied to force similar to the geology of Vietnam when comparing to the US and
effects Europe conditions. Then the loading test data in Japan are quite
useful for calibration of SRP design factors.
 =resistance factor: a statistically based multiplier applied to
nominal resistance
i = load modifier: a factor relating to ductility, redundancy and
operational importance [1]
Qi = force effect
Rn = nominal resistance
Rr = factored resistance:  Rn
3. DESIGN METHOD FOR SRP

3.1 Experimental and geological data
For calibration of SRP design factors by reliability method, loading
test data shall be collected [7, 9, 12, 15, 16] There are many reliable
data of SRP loading test in Japan and one data in Vietnam (Hoang
Minh Giam flyover). However, a few loading test data of SRP can
be found in the other countries. Variety of data of SRP loading test
data are shown in Table 1:
Table 1 Editorial Instructions
Item SRP
foundation
Number of data: 23 reports
- Pile tip resistance 22
- Shaft resistance (sand, gravel) 37
- Shaft resistance (clay, sand-clay) 21
Pile diameter Dp (mm) 318～1600
Blade diameter Dw (mm) 1.5 Dp, 2 Dp
Depth L (m) 12.5～55.7
Below soil layers All Soil
Bearing soil layer Sand, gravel
Some typical data are shown in Table 2:

Table 2 Editorial Instructions
Bearing Pile Shaft Figure 3 Boring hole examples in Vietnam and Japan
No Dp Dw L
layer tip R R
(mm) (mm) (m) (type) (kN) (kN) 3.2 Loads and load combinations for SRP design
1 406.4 812 34.5 gravel 1978 1463
The loads for SRP design, of course, shall follow the design of
2 800 1200 15.7 sand 7620 1299
bridge superstructure as specified in the bridge specification [1, 2,
3 900 1350 55.2 gravel 6250 6599
18]. So it doesn’t need to mention more detail about the permanent
4 1000 1500 25.1 sand 11009 3478 and transient loads in this paper.
5 1600 altered For the load factors, LRFD method applies the factors for
2400 51.0 43330 13242
rock various permanent and transient load types using the equation (1).
Selection of the load factor(s) to be used is a function of the type of
Almost data were collected in Japan, so we should do some load and limit state being evaluated (a limit state is a condition
comparison of the soil conditions in Japan, Vietnam and other beyond which a foundation or structure component ceases to fulfil
countries [2, 4, 5, 8, 13, 15, 18]. its intended function).
In Vietnam, the soil type and depth of bearing layer are different The load factor, γi, chosen for a particular load type must
between the North and South. In Japan, the scope of the soil bearing consider the uncertainties in the: magnitude and direction of loads,
layer are also varied. location of application of loads and possible combinations of loads.
For example, the data of boring holes in the North of Vietnam To fulfil with the design of bridge superstructure, abutment and pier,
shows bearing layer is not so deep (from about 30-50m), as same as the load factors and load modifier for SRP design should be taken
in the Western Chugoku region of Japan. Besides, geology the same values with the above structures: The permanent and
conditions in the South of Vietnam is quite similar to the Tokyo Bay transient loads and forces shall be also considered following the
142
bridge design specification [1, 2]. The SRP foundation should be 0.16 6
optimal designed to support these loads.
5
In conclusion, the load factors, load modifier as well as load
0.12
combinations for SRP design in Vietnam shall be selected as
4
R e la t iv e f r e q u e n c y
N u m b e r o f ca se s
specified in the bridge design specification 22TCN 272-05.
0.08 3
3.3 New Resistance factors for SRP design
2
Resistance factors for SRP are not mentioned in the 22TCN 272-05
0.04
so need to be assigned by calibration of reliability method based on 1
experimental data [2, 5, 6].
The resistance factor of the foundation can be obtained from 0.00 0
formula (2) [5, 6, 8, 11]: 0.0 0.5 1.0 1.5 2.0 2.5 3.0
Ratio of loading test data over the design value of shaft for clay
  1  COV  COV  
0.16 8
 Q
2 2
R  D D   L   
QD QL
 L Q   1  COVR2   7
 (2)
 
0.12 6
 QD QD 
  QL  exp T ln 1  COVR2  1  COVQ2D  COVQ2L  
N u m b e r o f case s
5
 Q L 
0.08 4
in which: 3
0.04 2
φ = resistance factor ; R = resistance bias factor
1
COVQ  coefficient of variation (the ratio of the standard deviation to
0.00 0
the mean) of load ; COVR  coefficient of variation of 0.0 0.5 1.0 1.5 2.0 2.5 3.0
Ratio of loading test data over the design value of shaft for sand
resistance; T  target reliability index.
Figure 4 Data analysis for pile resistance
 D ,  L  dead and live load factor ; QD / QL  dead to live load ratio;
QD , QL  dead and live load bias factor Using the formula (2) and the defined method in the items 2.1
and 3.3, the main results are shown in Figure 4 for pile tip resistance
There is relationship between reliability index βT and probability and pile shaft resistance.
of failure Pf [3]. βT of SPT method was calculated and was found to The values of resistance factor corresponding to target reliability
be between about 1.5 and 3.0 [5, 11]. A reasonable value of target index, βT are calculated as in Table 3. Here the value at βT = 2.33
reliability index, βT, for single piles appears to be in the range of 2.0 will be chosen for the next step.
to 2.5, corresponding to Pf between 10-1 and 10-2.
Then a reduction factor RD can be calculated by the index from Table 3 Editorial Instructions
the confidence interval. This index is divided the lower limit of the
Number
confidence interval by the average value, and the reduction factor is Resistance factor
Content of data
the ratio of the index for in case of actual number of data and φ
n
enough data. Confidence interval is calculated following equation:
βT 2.00 2.33 3.0 812
Pile tip resistance 0.70 0.62 0.49 1200
CI    t (1.0,n  1)   / n (3) Shaft resistance 0.53
0.45 0.32 1350
(sand, gravel)
Equation (3) is for a confidence interval of 95% one-sided, using Shaft resistance 0.58
the ratio of the lower limit of the reliability interval for sufficient 0.48 0.33 1500
(clay, mixed clay)
data as reduction factor.
Each resistance factor is re-calculated by SRP loading data The values of reduction factor with the number of data defined
following LRFD methods. About shaft friction, each data of sand in Table 3, are shown in the Table 4 as follows:
and clay is evaluated individually. Shaft resistance is measured by
stress of pile for each layer, so it is able to get some data by one Table 4. Reduction factor
loading test (Figure 4).
Standard
Mean RD
0.16 6 Resistance deviation
 (lower CI)

5 Pile tip resistance 1.10 0.30 0.99
0.12
Shaft resistance
1.22 0.57 1.00
4
N u m b e r o f ca se s
(sand, gravel)
0.08 3 Shaft resistance
(clay, mixed 1.48 0.79 0.96
2 clay)
0.04
1 Finally, the recommendation resistance factor is shown in Table
5. This value is obtained by multiplying calculated resistance factor
0.00 0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
and reduction factor.
Ratio of loading test data over the design value of pile tip
143
Table 5. Recommendation resistance factor capacity of the pile shaft resistance. The mobilization of the tip
resistance requires a considerably larger settlement than the
Recommendation
Resistance φ × RD  mobilization of the shaft resistance. The effects of the negative shaft
resistance factor)
friction on the geotechnical bearing capacity are checked separately,
Pile tip resistance 0.61 ⇒ 0.60
when negative shaft friction is developed or the pile is plugged. The
Shaft resistance (sand,
0.45 ⇒ 0.45
gravel) possible corrosion of the pile does not lower the geotechnical
Shaft resistance (clay, ⇒ bearing capacity.
0.46 0.45 This instruction is mainly concerned with SRP, the geotechnical
mixed clay)
bearing capacity of the pile can be determined in many different
The uplift resistance factor can be also assigned by the same ways which can be roughly divided into direct and indirect methods.
method based on experimental data. Indirect methods include:
Finally, resistance factors for SRP at the strength limit state [19] - Static bearing capacity formulas
for static analysis shall be taken as specified in Tables 6, unless - Empiric methods based on the penetration resistance
regionally specific values are available. - Stress wave analysis without stress wave measurements.
Direct methods include:
Table 6. Resistance factors for SRP (single pile) - Dynamic test loadings
- Static test loadings.
Resistance In design stage the indirect methods are used in designing of the
Condition
factor pile dimension, penetration depth and the construction equipment.
Pile shaft
These are checked on the site using direct methods, usually with
resistance: all soil 0.45 dynamic test loadings.
Nominal Resistance in and gravel The bearing capacity of the SRP consists of the pile tip
Axial Compression, φst resistance and external shaft resistance.
Pile tip resistance:
0.60
sand and gravel 2.1.4 Geotechnical bearing capacity formula
Pile shaft The studies in Japan [9, 12, 13, 14] and item 3.1 have shown the
resistance: all soil 0.45 sufficient data and scientific basis for determining the bearing
Uplift Resistance, φupst and gravel capacity of Bottom single blade Steel Rotation Pile which is also
specified in the specification JRA 2012 [10].
Pile tip resistance:
0.60 The other studies such as in US, Euro and Australia [3, 4, 8, 15]
sand and gravel
are mainly for small diameter or multi-blades rotation (screw) piles
almost used for buildings, not for bridges. However, these studies
3.4. Bearing capacity of SRP also show the main role of screw blade (rotation wing) in the
3.4.1. Determination of new formula for bearing capacity resistance components of the rotation steel pile.
Consequently, in combination with the geological analysis in
The bearing capacity of a pile is determined as follows: the pile item 3.1, the using a part of the bearing capacity formula of JRA
must sustain with sufficient certainty loadings in different loading 2012 to determine the bearing resistance formula of SRP for LRFD
cases after driving, and settlements and horizontal movements must design is acceptable with the above scientific basic.
be within the permissible structural tolerances. The geotechnical bearing resistance formula for SRP is
The bearing capacity of the pile is determined either based on following:
the structural or geotechnical bearing capacity, and the smaller one
is chosen to the design capacity. RR=φRn= φpstRp+φsstRs (4)
The structural bearing capacity of the pile is determined by the
strength of the pile structure. The structural bearing capacity is in which:
checked for the axial loads, bending moments from the horizontal φpst = Resistance factor of pile tip for SRP, static analysis
loads, eccentricities loads. In addition to the requirements of the φsst = Resistance factor of shaft for SRP, static analysis
supported structure the bearing capacity of the pile should be Rp = Pile tip resistance (kN)
considered for buckling, additional loads, such as negative shaft Rs = Pile shaft resistance (kN)
friction and bending of the inclined piles due to the ground
settlements or bending caused by one-sided soil pressure or lateral Range of considering
resistance. In a completed structure the steel pipe pile is usually shaft resistance
filled with soil. The structural capacity of the pile is formed by the
bearing capacity of the steel pipe considering the corrosion
reduction. Permitted material stresses of the pile are determined on 1.0Dw
the basis of the pile material and soil conditions. In bouldery soil Bearing layer
conditions it may be appropriate to reduce the material stresses over around 1.0Dp
Pile tip depth
permitted in normal situations. The recommended minimum (End of Steel pipe)
thickness of the steel pipe pile driven from the upper head is 9mm
[18, 19]. The corrosion of the pile should be considered when
determining the long-term structural bearing capacity of the pile.
Cutout length
So for SRP foundation, the material resistance shall comply with
steel pile structure calculation by the current bridge specification [1,
18, 19]. In addition, this resistance in most cases is usually greater
than geotechnical resistance so in this paper, we pay more attention
Cutout portion
in the geotechnical bearing capacity.
Dwi
The geotechnical bearing capacity is determined according to the
ground conditions, construction and checking procedures. Dp
The geotechnical bearing capacity of the pile consists of the Dw
bearing capacity of the pile tip resistance, and of the bearing Figure 5 SRP dimension and resistance
144
The pile resistance can be referred from the bearing capacity Pile construction shall be finished on full examination of the
formula of JRA 2012. finishing conditions so as to ensure the bearing capacity. The
The pile shaft resistance can be determined by: finishing conditions should be determined based on results of test
piling operation, and are represented by such factors as embedded
Rs = qsAs (5) depth of the pile, dynamic bearing capacity. Embedded depth of a
pile specified in design documents or confirmed by results of trial
Where: piling test.
Rs：Pile shaft resistance (kN) For the uplift resistance of SRP, it should be estimated in a
qs：unit shaft resistance of pile (kN/m2) similar manner to determinate the shaft resistance as specified
As : Area of pile shaft surface   Dp L (m2) above, and it can be assumed that the steel blade behaves like an
Dp: Pile diameter (m), anchor as shown Figure 6.
L: Length of the section to take into account the shaft resistance of Pullout force
the pile upper 1Dw from pile tip (m), (m2) Ground surface
The nominal unit shaft resistance of SRP, in kN/m2, shall be

taken as: SRP Shaft resistance
For non-cohesion soils
L
qs = 3N (Maximum 150) (6)
For cohesive soils Pullout resistance
at steel blade
H
Bearing layer
qs = c or 10N (Maximum 100) (7) Blade
Where: Figure 6 Uplift of SRP shaft and steel blade
N: SPT blow count, uncorrected for overburden pressure
2
C : cohesion (kN/m ) 3. CONCLUSION
The pile tip resistance can be determined by: SRP foundation for bridge is suitable for transport works especially
in urban area with the key advantages: easy application in narrow
Rp = qpAp (8) space, fast construction, reduce vibration and noise, high quality
control, small influence to nearby monument, environmentally
Where: friendly, recyclable, etc. The efficacy is shown in many reel projects
Rp：Pile tip resistance (kN) including two overpass projects in the capital of Vietnam.
qp : Unit tip resistance of pile (kN/m2) For application of SRP foundation, the specifications for design,
Ap： Area of pile tip = Dw2/4 construction and acceptance are required. To fulfil with the LRFD
Dw: Blade diameter (m) design of bridge superstructure, abutment and pier, the load factors,
load modifier as well as load combinations for SRP design shall be
The nominal unit tip resistance of SRP in bearing layer, in selected as specified in the specification 22TCN 272-05. The
kN/m2, shall be taken as: resistance factor values for SRP can be assigned based on loading
For bearing layer of sand test data according to LRFD method, in compatibility with the
bridge design specification system in Vietnam. The normal
qp = 120N (Maximum 6,000, Dw is 1.5Dp) (9a) resistance for geotechnical bearing capacity can be referred from the
qp = 100N (Maximum 5,000, Dw is 2.0Dp) (10a) formula in JRA 2012 with the carefully study. These above things
are the most important contents of the new LRFD specification for
For bearing layer of gravel SRP design (TCVN 11520:2016) in the study case of Vietnam.
The Vietnamese case studies show that the new method based on
qp = 130N (Maximum 6,500, Dw is 1.5Dp) (9b) LRFD can be established and used effectively for SRP foundation
design instead of ASD method of Japan if need.
qp = 115N (Maximum 5,750, Dw is 2.0Dp) (10b) The uplift resistance, settlement and other checking for SRP
foundation LRFD method design as well as SRP design examples
N: SPT blow count at bearing layer, uncorrected value for calibration between the different specifications, the construction
requirements and technologies will be discussed in the next papers.
145
4. REFERENCES JRA, (2012) Specification for Highway Bridges, Japan.

M. Barker et al., (1991) National Cooperative Highway Research
22 TCN 272-05 Specification for Bridge Design, Vietnam (based on Program Report 343.
AASHTO LRFD 1998). Masahiro S. et al. ,(2000) Study on estimating the bearing capacity
AASHTO LRFD (1998, 2007, 2012, 2017) Specification for Bridge of pile considering the number of loading test, Ministry of
Design. Construction of Japan Report.
A.S. Nowak, K.R. Collins, (2000) Reliability of Structures, Mc NSSMC (2005, 2012) Technical report; manual and presentation of
Graw Hill. NS Eco-pile.
Eurocode, (2003) CEN, EN 1997 Geotechnical design. Ohsugi et al, (2005) Pile installation tests of screwed piles in a
Dao Duy Lam, and Nguyen Thi Tuyet Trinh, (2016) Research on calibration chamber, Journal of Structural and Construction
resistance factor of bearing capacity for Steel Rotation Pile Engineering, Architectural Institute of Japan.
foundation in Vietnam, Geotechnics for Sustainable P.J. Yttrup et al.,(2003) Ultimate strength of steel screw piles in
Infrastructure Development - Geotec Hanoi 2016, Phung sand, Australian Geomechanics, Vol38.
(edt). ISBN 978-604-82-0013-8 PMU Thang Long, (2016) Design Document and Loading test
Federal Highway Admin. U.S., (2001) Load and Resistance Factor Report of Hoang Minh Giam flyover, Hanoi.
Design (LRFD) for Highway Bridge Substructures-Reference PMU Thang Long, (2017) Design Document of Ring Road No.3
Manual. FHWA HI-98-032. part Mai Dich - Thang Long of Hanoi.
Haward A. Perko, (2009) Helical piles: A Practical Guide to Design TCVN 10834:2015 Steel pipe pile for bridge foundation - Design
and Installation. Specification.
ITST, (2014) Research theme DT144001. TCVN 11520:2016 Bottom single blade Steel Rotation Pile
Japan Institute of Country-ology and Engineering (JICE), (2014) Foundation - Design Requirements.
Construction Technology Review and Certification Report.
146
Active shaking tests of pile foundation models in dry sand ground

Kohei Kenda1, Anh-Tuan Vu2, Tatsunori Matsumoto3
1
Hazama-Ando Corporation, Japan
E-mail: shirasagisan5@gmail.com
2
Le Quy Don Technical University, Vietnam
E-mail: vuanhtuan@mta.edu.vn
3
Kanazawa University, Japan
E-mail: matsumoto@se.kanazawa-u.ac.jp
ABSTRACT:
In this research, behaviours of pile foundation models including piled raft and pile group subjected to dynamic loading were investigated
through small-scale experiments. Foundation models consisting 6 piles, with or without batter piles, were used in the experiments. They were
pile rafts (6PR and 6BPR) if the raft was in contact with ground surface, while they were pile groups (6PG and 6BPG) if the raft was not in
contact with ground surface. To create dynamic load acting on the foundation, a vibro-hammer, placed on the raft, was used. The vibro-
hammer can provide vibration load (active shaking) mainly in the vertical direction (called vertical loading) or in the vertical and horizontal
directions simultaneously (called combination loading) by rotating two discs of eccentric mass synchronously in opposite directions or the
same direction. Active shaking tests were conducted on 4 types of pile foundation models (6PR, 6BPR, 6PG and 6BPG) in a consistent dry
sand ground. It was found from the experimental results that the piled rafts are more effective foundation type to decrease settlement and
inclination under dynamic loading than the pile groups.
Keywords: Piled raft, pile group, dry sand, model experiment
1. INTRODUCTION 40 80 80 40 40 80 80 40
Pile foundations supporting structures such as bridges, buildings,
40
40
wind turbines etc. are often subjected to dynamic loads caused by 4 5 6 4 5 6
traffic load, wind load, or wave load. If excessive displacements
80
80
and/or inclination of a foundation structure are caused by the
1 2 3 1 2 3
dynamic loads, the structure could be damaged and could not be
40
40
used as in normal status. Hence, it is necessary to investigate
behaviours of pile foundations under dynamic loading. In this 240 240
research, behaviours of pile foundation models in dry sand ground
30
30
subjected to active dynamic loading were investigated through
experiments. 80 80 80 80
2. OUT LINE OF EXPERIMENTS 1.1 1.1

255
15
2.1 Pile foundation models
(in mm)
Pile foundation models used in Yesim et al. (2017), Vu (2017) and 20 20
Vu et al. (2018) were used in the experiments of this study. The pile
foundation models consist of 6 piles with or without batter piles and P1 P2 P3 P1 P2 P3
a raft as shown in Figure 1. The model pile is close-ended P4 P5 P6 P4 P5 P6
aluminium pipe having an outer diameter of 20 mm, a thickness of
1.1 mm and a length of 285 mm. In battered pile foundations (6BPR Figure 2 Dimensions of pile foundation models
and 6BPG), inclination angle of batter piles is set at 15 degrees. The
6-piled raft 6-battered piled raft
rectangular duralumin raft has dimensions of 240 mm in length, 160
foundation foundation
mm in width, and 30 mm in thickness. Pile positions are shown in
Figure 2.
Figure 3 shows cases of foundation type in the experiments. The G.L
foundation is piled raft (PR) if the raft base is in contact with the
ground surface. In the case of pile group (PG), a gap of 20 mm is set PR
between the raft base and the ground surface. Each model pile is
mounted with strain gauges along the pile shaft to obtain axial P1 P2 P3 P1 P2 P3
forces, shear forces, and bending moments of the pile. The positions P4 P5 P6 P4 P5 P6
of strain gauges are shown in Figure 4. Physical and mechanical
properties of the model pile are listed in Table 1.
6PR 6BPR
6-pile group 6-battered pile group
foundation foundation
G.L gap
= 20mm
P1 P2 P3
PG
P1 P2 P3
P4 P5 P6 P4 P5 P6
6PG 6BPG
Figure 1 Pile foundation models Figure 3 Cases of foundation type in the experiment
147
Shear strain
gauge
Axial strain Rear

t = 1.1 gauge
30 500mm
Level 1 20 20
Level 1
Sand particles 40
Level 2
adhered
40 120 Front 800mm
Level 3
530mm
L=255
40
Level 4 Level 2
40
Level 5 80
close-ended 40
Level 6 Level 3
35 35
15
(mm)
D = 20 Figure 5 Dimension of laminar box
P1-P3 P4-P6
In order to grasp the mechanical behaviours of the sand, a
Figure 4 Position of strain gauges series of triaxial CD compression tests were carried out (Vu et al,
2018). Five CD tests were carried out with different confining
Table 1 Physical and mechanical properties of model pile pressures, p0, of 7, 17, 27, 50 and 100 kPa. The results of the CD
tests are shown in Figure 6. The internal friction angle, p', at peak
Property Value strength is 42.8 degrees and the friction angle at residual state, r', is
Outer diameter, D(mm) 20.00 35 degrees. The initial stiffness, q/a, increases almost linearly
with increase in the square root of p0. It is seen from Figure 6b that a
Wall thickness, t(mm) 1.1 small amount of negative dilatancy occurs at a very early stage of
shearing followed by a large amount of positive dilatancy. Positive
Length from raft base, L(mm) 255
dilatancy behaviour weakens after the axial strain, a, exceeds about
Cross section area, A(mm2) 65.31 8%.
Silica sand
Moment of Inertia, I(mm4) 2926.2 500
p0= 100 kPa
Deviatoric stress, q (kPa)
Young’s modulus, Ep(N/mm2) 70267 p0= 100 kPa

400 cyclic
Poison’s ratio, v 0.31
300 p0= 50 kPa

2.2 Model ground
The sand used for model ground in the experiments was dry silica 200
sand #6. The physical properties of the sand are shown in Table 2. p0= 27 kPa
The model ground with a relative density, Dr of about 82% was
prepared in a laminar box shown in Figure 5. The model ground 100
consisted of 11 layers (10 layers of 50 mm and 1 layer of 30 mm). In p0= 17 kPa p0= 7 kPa
order to control density of the model ground, the sand was 0
compacted by tamping in each layer. 0 2 4 6 8 10 12 14 16 18 20
Axial strain, a
(%)
Table 2 Physical properties of silica sand #6
(a) Axial strain a vs deviatoric stress q
Silica sand
Property Value -12
p0= 7 kPa
Density of soil particle, s (g/cm3) 2.668
(%)
-10 p0= 17 kPa

Maximum dry density, dmax (g/cm3) 1.604
vol
-8 p0= 27 kPa
Volumetric strain,
Minimum dry density, dmin (g/cm3) 1.269 -6 p0= 50 kPa p0= 100 kPa
Maximum void ratio, emax 1.103 cyclic
-4
Minimum void ratio, emin 0.663
-2 p0= 100 kPa
Relative density, Dr (%) 82.0
0
Dry density, d (g/cm3) 1.533
2
0 2 4 6 8 10 12 14 16 18 20
Axial strain, a
(%)
(b) Axial strain a vs volumetric strain vol
Figure 6 Results of triaxial CD tests for the sand
148
2.3 Loading equipment and measurement items

Figure 7 shows the initial state of a pile foundation model. A
rotating discs
connection plate having a weight of 30N was attached on the raft
with screws. Four load cells were set on the plate to measure vertical
load on the raft, as shown in Figure 8.
In dynamic loading test, a vibro-hammer having a weight of
300N was placed on the raft. The vibro-hammer supplies the raft
with not only the dead load but also dynamic load by rotating two
discs having eccentric mass. Eccentricity of mass of each disc and
input vibration frequency can control magnitude of the dynamic
load. As shown in Figure 9, the vibro-hammer can provide vibration vertical loading
load mainly in the vertical direction (called vertical loading) or in eccentric mass
the vertical and horizontal directions simultaneously (called
combination loading) by rotating discs synchronously in opposite combination loading
directions or the same direction. Eccentric mass was kept constant in
vertical loading and combination loading tests. Hence the amplitude
of dynamic load depends on the rotation frequency of the discs.
Figure 10 shows an illustration of experimental setup.
Measurement transducers were accelerometer (Acc), laser Figure 9 Loading mechanism by vibro-hammer
displacement meter (Laser), encoder (ENC), load cell (LC), and
strain gauges of each pile. Four accelerometers, Acc1 to Acc4, were
placed on the raft to measure the accelerations in the vertical and Acc5 Acc6
horizontal directions. ENC2 ENC1
VL laser VR laser
Initial state Acc8

HL laser LC LC HR laser
150mm
Acc7
Acc4 Acc2
Acc3 Acc1
connection plate
530mm
Strain gauge
255mm
Dry slica #6
Dr = 82%
( ρd=1.533g/cm 3)
800mm
Figure 7 Pile foundation with connection plate
Figure 10 Experimental setup
2.4 Experimental cases

Figures 11 and 12 shows experimental cases of each foundation
model. In all the cases, firstly the foundation was vertically loaded
Vertical load using the self-weight of the vibro-hammer (Static loading test).
Next, dynamic vertical loading test was carried out by operating the
Front
Rear vibro-hammer. The input frequency, f, of the vibro-hammer was
increased from 0 Hz to 30 Hz at intervals of 5 Hz.
In the cases of PG, when the settlement of the raft reached about
10 mm, vertical loading was interrupted. Thereafter, dynamic
combination loading test was continued. In the cases of PR, dynamic
combination loading test was conducted after the end of vertical
loading test.
Apart from the dynamic loading tests mentioned above, sweep
tests of each model foundation were carried out to estimate the
natural frequency, fn, of each model foundation. In the sweep tests,
small amplitudes of vertical vibration was applied to the foundation
Figure 8 Setting load cells on connection plate using the vibro-hammer with increasing the rotation frequency. The
response vertical acceleration of the raft was measured. The natural
frequency of the foundation was obtained from the FFT processing
of the measured response acceleration.
149
Figure 12 shows the natural frequency of each foundation model. Figure 12 Natural frequency, fn, of each foundation model
The four foundation models have almost the same values (fn = 14 to
15 Hz).
3. EXPERIMENTAL RESULTS
In this particular paper, the experimental results of dynamic vertical
Test 1. Static loading by weight loading tests of 4 foundation models (6BPR, 6BPG, 6PR, 6PG) are
of vibro-hammer presented and discussed.
3.1 Behaviours of each foundation model

Test 2. Vertical loading
Figure 13 shows changes of the vertical force, FV, the horizontal
input vibration force, FH, and the input vibration frequency, fV, with elapsed time, t,
frequency in the case of 6BPR.
The vertical load on the raft, FV, was measured by the load cells
0Hz→5Hz→10Hz→・・・ (see Figure 8). The horizontal load, FH, on the raft was inertial force
calculated as the product of the horizontal acceleration and the mass
Test 3. Combination loading of the vibro-hammer.
Although vertical loading was intended in these experiments,
not only vertical load but also horizontal load acted upon the
foundation model. It is thought that the horizontal load was caused
by imperfect synchronization of two rotating discs having eccentric
Figure 11 Experimental cases of each foundation model mass. As the result, combination load was applied to the raft.
Figure 14 shows changes of fV and the vertical displacement, w,
of the raft with elapsed time in the case of 6BPR. Note that the
6-piled raft 6-battered piled raft vertical displacement after the static loading by the vibro-hammer
was set as 0. When fv = 10 Hz, no vertical displacement occurred.
This is reasonable, because the corresponding Fv was very small.
When fv was increased to 12 Hz, the vertical displacement started to
occur and increased with time, and terminated at a certain time
PR instant. Hence, fv was further increased to 25 Hz, but the foundation
was still stable without increment of w. When fv was increased to 30
Hz, the vertical displacement of the foundation started to increase
P1 P2 P3 P1 P2 P3 suddenly. When fv was decreased to 15 Hz, no more vertical
P4 P5 P6 P4 P5 P6 displacement occurred.
The corresponding results in the cases of 6BPG, 6PR and 6PG
are shown in Figures 15 to 20. The results of these cases were
6PR (fn=13.9Hz) 6BPR (fn=13.9Hz) similar to those in 6BPR. In all the cases, the vertical displacement
6-pile group of the foundation started to increase rapidly when the input
6-battered pile group frequency, fV, was increased to 30 Hz.
gap
= 20mm
P1 P2 P3 PG P1 P2 P3
P4 P5 P6 P4 P5 P6
6PG (fn=15.0Hz) 6BPG (fn=13.8Hz)
150
2000
6BPR 30
Vert. force, FV
Vert. frequency, fV(Hz)

1500
25
External force, F(N)
Hor. force, FH
1000 Dead load of 20
Vibro-hammer
500 15
0 10
5
-500
Vert. freq(Hz) 0
-1000
0 120 240 360 480 600 720 840
Time, t (s)
Figure 13 Time changes of input vibration frequency, fV, vertical force, Fv, and horizontal force, FH, in the case of 6BPR
-2
0 30
Vert. frequency, fV (Hz)

2
Vert. disp. , w (mm)
30Hz 25
4
6BPR
6 20
8
15
10
12 10
14 Vert. disp.
16 5
Vert. freq(Hz)
18 0
20
0 120 240 360 480 600 720 840
Time, t (s)
Figure 14 Time changes of input vibration frequency, fV, and vertical displacement, w, in the case of 6BPR
2000
30
6BPG
Vert. frequency, fV(Hz)
1500 Vert. force, FV

External force, F(N)
25
Hor. force, FH
Dead load of
1000 20
Vibro-hammer
500 15
0 10
5
-500 Vert. freq(Hz)
0
-1000
0 120 240 360 480 600 720 840
Time, t (s)
Figure 15 Time changes of input vibration frequency, fV, vertical force, Fv, and horizontal force, FH, in the case of 6BPG
151
-2
0 30

2 30Hz
25
4
6
6BPG 20
8 Interrupted loading
15
10
12 10
14 Vert. disp.
16 5
Vert. freq(Hz)
18 0
20
0 120 240 360 480 600 720 840
Time, t (s)
Figure 16 Time changes of input vibration frequency, fV, and vertical displacement, w, in the case of 6BPG
2000
30

Vert. force, FV
1500 6PR
External force, F (N)
25
Hor. force, FH
1000 20
500 15
Vert. freq(Hz)
10
0
5
-500
0
-1000
0 120 240 360 480 600 720 840
Time, t (s)
Figure 17 Time changes of input vibration frequency, fV, vertical force, Fv, and horizontal force, FH, in the case of 6PR
-2
30
0
2 30Hz
25
4 6PR
6 20
8 Vert. disp. 15
10
12 10
Vert. freq(Hz)
14
5
16
18 0
20
0 120 240 360 480 600 720 840
Time, t(s)
Figure 18 Time changes of input vibration frequency, fV, and vertical displacement, w, in the case of 6PR
152
2000
30

1500 6PG Vert. force, FV
External force, F (N)
Hor. force, FH 25
1000 20
Vert. freq(Hz)
500 15
0 10
5
-500
0
-1000
0 120 240 360 480 600 720 840
Time, t (s)
Figure 19 Time changes of input vibration frequency, fV, vertical force, Fv, and horizontal force, FH, in the case of 6PG
-2
30
0
30Hz

2
6PG 25
4
6 20
8 Vert. disp.
15
10 Vert. freq(Hz)
12 10
14
5
16
18 0
20
0 120 240 360 480 600 720 840
Time, t(s)
Figure 20 Time changes of input vibration frequency, fV, and vertical displacement, w, in the case of 6PG
3.2 Comparisons of behaviours of the foundation models Figure 24 shows the horizontal force, FH, in the cases of 6BPR
subjected to 30 Hz loading and 6BPG during 30 Hz loading. Amplitudes of FH of 6BPG are
larger than those of 6BPR. In contrast, amplitudes of vertical force,
In this section, the behaviours of each foundation model in loading FV, of 6BPG are smaller than those of 6BPR (see Figure 21). The
step of fv = 30Hz are compared and discussed. larger amplitudes of FH of 6BPG may be a reason for that the
Figure 21 shows the vertical force, FV, in each foundation model vertical displacement of 6BPG is relatively larger although
with elapsed time from the start of 30 Hz loading. amplitudes of FV of 6BPG are smaller than the other cases.
Figure 22 shows the increment of the vertical displacement, w, Figure 25 and Figure 26 show distributions of bending moments
of each foundation with time from the start of 30 Hz loading. of the piles in 6BPR and in 6BPG, respectively, at elapsed time of
Amplitudes of Fv in 6PR, 6PG and 6BPR were about 1600 N 10s from the start of 30Hz loading. Bending moments of the piles in
(Figure 21). The vertical displacement of 6BPR is the smallest 6BPG are smaller than those in 6BPR, although horizontal forces on
followed by those of 6PR and 6PG, indicating that the batter piles 6BPG are larger than those on 6BPR. Larger vertical load in 6BPR
and the raft base resistance suppress the vertical displacement. It compared with 6BPG (see Figures 13 and 15) is one of the reasons
may be judged from Figure 22 that inclusion of the batter piles is that caused larger bending moments in 6BPR. Another reason could
more effective to suppress the vertical displacement. be the effect of load transferred from the raft base to the ground in
Although amplitudes of Fv in 6BPG were smaller than those in the case of 6BPR. The load transferred to the ground increased
the other cases, w in 6BPG were comparable to those in 6PG. It is stresses and stiffness of the soil beneath the raft, which resulted in
difficult to explain this result definitely at this stage. larger bending moments in the piles of 6BPR.
Figure 23 shows the increment of inclination of the raft, , of It is also seen from Figures 25 and 26 that bending moments of
each foundation model with time during 30 Hz loading. Comparison the centre piles (P2 and P5) are very small in both of PR and PG.
of  in 6PG, 6PR and 6BPR indicates that the batter piles and the Figure 27 and Figure 28 show horizontal accelerations of the
raft base resistance suppress the inclination of the foundation, vibro-hammer, the raft and the ground surface in shaking of 6BPR
similarly to the effect for suppressing the vertical displacement. The and in shaking of 6BPG, respectively, with elapsed time from the
inclusion of the batter piles (6BPR) has a great effect on the start of 30Hz loading. Horizontal accelerations on the raft (Acc2 and
inclination reduction of the foundation. Acc4) of PR are smaller than those of PG. This result indicates that
153
raft base resistance is effectively mobilised to suppress the that in PG. This indicates that the influence of the shaking of the raft
horizontal acceleration of the raft. on the ground around PR is relatively larger than that around PG.
Focusing on the ratio of horizontal acceleration at the ground Similar result is found from the comparison of the experimental
surface to that on the raft, the ratio in PR is relatively higher than results of 6PR and 6PG as shown in Figures 29 and 30.
2000
1600 30Hz
6PG
1200
800 6BPG
400
Vert. force, FV(N)
0
0 2 4 6 8 10 12 14
Time, t (s)
2000 30Hz 6PR
1600
6BPR
1200
800
400
0
0 2 4 6 8 10 12 14
Figure 21 Time changes of vertical force, FV, during 30 Hz loading
-1
0
6BPR 6BPG
1 6PR 6PG
2
3
4
4.45
5
6
6.56
7 6.72
30Hz 7.20
8
0 2 4 6 8 10 12 14
Time, t (s)
Figure 22 Increment of vertical displacement, w, of each foundation model with time during 30Hz loading
(deg.)
-0.25 6BPR 6BPG

6PR 6PG
0.00
0.16
Increment of inclination,
0.25
0.46
0.50 0.59
0.75
0.87
30Hz
1.00
0 2 4 6 8 10 12 14
Time, t (s)
Figure 23 Increment of inclination, , of each foundation model with time during 30Hz loading
154
1600
30Hz
1200
800
Hor. force, FH(N)
400
-400
-800 6BPG 6BPR

0 2 4 6 8 10 12 14
Time, t (s)
Figure 24 Time changes of horizontal force, FH, during 30 Hz loading in cases of 6BPR and 6BPG
Figure 25 Distributions of bending moments of the piles in 6BPR at elapsed time of 10s from the start of 30Hz loading
Figure 26 Distributions of bending moments of the piles in 6BPG at elapsed time of 10s from the start of 30Hz loading
155
(m/s ) 50
6BPR Vibro-hammer(Acc5)
2
40 Raft(Acc2)
Raft(Acc4)
H
30
Ground surface(Acc7)
Hor. acceleration,
20
10
0
-10
-20
-30 30Hz
-40
0 2 4 6 8 10 12 14
Time, t (s)
Figure 27 Horizontal accelerations of vibro-hammer, raft and ground surface in shaking of 6BPR with elapsed time from the start of 30Hz
loading
(m/s )
50
6BPG Vibro-hammer(Acc5)
2
40 Raft(Acc2)
Raft(Acc4)
H
30
Hor. acceleration,
20
10
0
-10
-20
-30 30Hz
-40
0 2 4 6 8 10 12 14
Time, t (s)
Figure 28 Horizontal accelerations of vibro-hammer, raft and ground surface in shaking of 6BPG with elapsed time from the start of 30Hz
loading
(m/s )
50 6PR Vibro-hammer(Acc5)
2
40 Raft(Acc2)
Raft(Acc4)
H
30
Hor. acceleration,
20
10
0
-10
-20
-30 30Hz
-40
0 2 4 6 8 10 12 14
Time, t (s)
Figure 29 Horizontal accelerations of vibro-hammer, raft and ground surface in shaking of 6PR with elapsed time from the start of 30Hz
loading
156
(m/s ) 50 Vibro-hammer(Acc5)
2
6PG
40 Raft(Acc2)
Raft(Acc4)
H
30
Hor. acceleration,
20
10
0
-10
-20
-30 30Hz
-40
0 2 4 6 8 10 12 14
Time, t (s)
Figure 30 Horizontal accelerations of vibro-hammer, raft and ground surface in shaking of 6PG with elapsed time from the start of 30Hz
loading
A possible reason for this is illustrated in Figure 31. In the case 4. CONCLUSION
of PR, the ground beneath around the raft is hardened by the raft
base pressure. Hence, vibration of the raft is easily transmitted to the Series of active dynamic load tests on 6-pile foundation models
surrounding ground. In contrast, in the case of PG, the ground (with or without batter piles) in dry sand were carried out at 1-g
around the piles is softened by, for example, generation of gaps field to investigate the behaviours the foundations.
between the pile shaft and the ground. Therefore, the vibration of The experimental results indicate that the piled rafts are more
the surrounding ground is attenuated largely compared to that of the effective than the pile groups in suppressing vertical displacement
foundation. and inclination induced by vertical dynamic load. It is also
confirmed from the results that piled raft with batter piles is the most
effective type to reduce settlement and inclination.
5. REFERENCES
Unsever, Y., Matsumoto, T., Esashi, K., and Kobayashi, S. (2017)
"Behaviour of model pile foundations under dynamic loads in
hardened saturated sand", Bulletin of Earthquake Engineering. Springer,
area Netherlands, 15, pp1355-1373.
Vu Anh-Tuan (2017) "Experimental and Numerical Study on
Behaviours of Pile Group and Piled Raft Foundation Having
Batter Piles Subjected to Combination of Vertical and Cyclic
Horizontal Loading", Doctoral Thesis of Kanazawa University.
Vu Anh-Tuan, Matsumoto, T., Kobayashi, S., and Nguyen Tuong-
Lai (2018) "Model load tests on battered pile foundations and
(a) Piled raft finite-element analysis", Int. Journal of Physical Modelling in
Geotechnics, 18, Issue 1, pp 33-54.
softened
(b) Pile group

Figure 31 Illustrations of deformation patterns of the ground
157
Estimating Pile Axial Bearing Capacity by c- Derived from Pressuremeter Test
Tjie-Liong GOUW1
1
Associate Professor, Universitas Katolik Parahyangan, Bandung, Indonesia
E-mail: gtloffice@gmail.com
ABSTRACT: Due to its rather brittle nature, retrieving undisturbed samples of Jakarta cemented greyish stiff clay, often found at a depth of
30 to 120m, is very difficult. Good and reliable effective shear strength parameters, i.e., c’ and ' values, obtained from triaxial test are hardly
available. By modifying cavity expansion theory, Gouw (2017) was able to derive these effective shear strength parameters through
Pressuremeter in situ test stress strain curve. It was found Jakarta cemented clay exhibiting a drained behaviour when loaded. Its effective
cohesion, c’, values are linearly increasing with depths, averaging from around 95 kPa at 20 m to around 475 kPa at 100m depth, while its
effective friction angle ' values are within 20o – 30o, averaging to around 24o. The values found to be similar to the values derived from CIU
triaxial compression test from relatively good undisturbed samples. This paper presents the methodology in deriving the shear strength
parameters and then applying the derived Pressuremeter c’ and ' values to estimate the pile axial bearing capacity through finite element
simulation and comparing it with the commonly known SPT method applied in Jakarta.
Keywords: Pressuremeter, modified cavity expansion theory, effective shear strength parameters, pile axial capacity
1. INTRODUCTION
By far, Pressuremeter test is the only known in-situ geotechnical
testing device capable to generate a stress-strain curve of in-situ soils,
somewhat similar to the stress-strain curve obtained from triaxial or
direct shear test in soil laboratories. By simulating the Pressuremeter,
hereinafter abbreviated as PMT, test through modification of
cylindrical cavity expansion theory and matching the resulting stress
strain curve with the actual PMT data curve, Gouw (2017) was able
to derive effective shear strength parameters, i.e., c’ and ' values, of
Jakarta cemented stiff clay. His research showed that Jakarta
cemented clay, known of its rather brittle nature, exhibiting a drained
behaviour when loaded under the PMT test. The effective cohesion,
c’, values were found to be linearly increasing with depths, averaging
from 95 kPa at 20 m depth to around 475 kPa at 100 m depth, while
its effective friction angle ' values are within 20o–30o, averaging to
around 24o. The values found to be similar to the values derived from
CIU triaxial compression test from relatively good undisturbed
samples. This paper presents the PMT testing principle, the traditional
PMT parameters, the modified cavity expansion formulas used, a case
study in deriving c and  of Jakarta cemented clay, and application of
the values obtained to estimate pile axial bearing capacity through Figure 1 Schematic Diagram of Pressuremeter Test (Briaud, 2013)
finite element simulation, finally comparing the result with the
commonly known SPT method applied in Jakarta local practice.
2. PRESSUREMETER TEST AND ITS PARAMETERS

Pressuremeter test is conducted by inserting a cylindrical membrane
into a carefully prepared borehole to a determined test depth where
the cylindrical membrane is then pressurized against the borehole
wall and the subsequent volume expansion (Menard PMT) or the
radial expansion (OYO PMT) of the cylindrical membrane is
measured. Figure 1 shows the schematic diagram of PMT.
If the pressure is applied by pumping de-aired water into the
cylindrical membrane, the actual pressure or stress acting on the
borehole wall needs to be corrected against membrane resistance and
against the hydrostatic pressure from the manometer level to the
centre of the membrane. In Menard PMT the volume of expansion is
corrected against the expansion of the hose to deliver the water from
the control unit to the membrane. In OYO PMT, also known as
Elastmeter, the radius of expansion is corrected against the reducing
membrane thickness when pressurised.
The corrected volume or radius is then converted into radial
strain of the borehole wall. The resulting corrected radial stress strain
data is then plotted. Figure 2 shows the typical stress strain curve Figure 2 Pressuremeter Typical Test Graph
obtained from PMT test. (modified after Briaud, 2013)
158
Traditionally six parameters are obtained from the PMT stress- The expansion of cylindrical cavity can be divided into elastic and
strain curve, i.e.: Po, Py, PL, Km, Em, and G (Baguelin et al, 1972, 1978; plastic zone as illustrated in Figure 4. By using Mohr Coulomb failure
Gambin, 1980, 1995; Gambin and Frank, 2009; Clayton et al, 1982; criterion and radial stress vs modulus of deformation, depicted in
Briaud, 1992; Clarke, 1995). The parameters are described below Figure 5, Mecsi derive equations to calculate the cohesion, c, and
(refer to Figure 2 for some notations): friction angle, , of soils, from PMT test data. His equations are:
• Horizontal pressure, Po, is the pressure when the membrane first

touches the borehole wall, i.e. first point at the beginning of
linear or elastic part of PMT curve. This pressure is interpreted
as soil total horizontal pressure at rest, i.e.,
Po = ’vo ko + uo (1)
’ho = ’vo ko = Po - uo (2)
where ’vo is vertical effective pressure, ’ho is horizontal

effective pressure, ko is at rest horizontal earth pressure
coefficient, uo is hydrostatic groundwater pressure.
• Yield pressure, Py, is the end of the linear curve and the
beginning of the non-linear or plastic part of the PMT curve,
• Limit pressure, PL, is the ultimate pressure of PMT curve where

soil start to ‘flow’, i.e. radial strain keeps on increasing at
relatively constant presssure. In practice, limit pressure is hardly
achieved, and to obtain this PL value, the test curve must be
extrapolated in a logarithmic plot as shown in Figure 3 below,
Figure 4 Cylindrical Cavity Expansion Zone

(Modified after Mecsi, 2013)
Figure 3 Extrapolation of PMT Test Data to Obtain PL

(Modified after Baguelin et al, 1978, Ghionna et al, 1981)
• Horizontal subgrade reaction, Km, obtained through linear

part of the test curve, i.e.:
∆P P -P
Km = = y o (3)
∆R RPy -RPo
where RPy is cavity radius at Py and RPo is cavity radius at Po.
• Soil deformation or stiffness modulus, Em :
RPo +RPy
Em = (1+υ) Km (4)
2
where  is Poisson ratio of the soil, usually taken as 0.33.
• Shear Modulus, G :
Em
G= (5)
2(1+υ)
3. MODIFIED CAVITY EXPANSION FORMULAS Figure 5 Mohr Failure Criterion and Modulus of Deformation
The cavity expansion theory used in deriving the shear strength Relationship (Modified after Mecsi, 2013)
parameteres from PMT test curve is modified from Mecsi work
(Mecsi, 2013) which is elaborated below.
159
2.�
�� = (6) 4000
√� (i)
3500
1−��
��
�= (7)
1+��
��
3000
σ β
Es = Eo ( σ r )
Radial Stress, kPA

(8)
ref 2500
where Es is deformation modulus at a cavity pressure of r, Eo is 2000

deformation modulus at a reference pressure ref = 100 kPa as shown
in Figure 5, coefficient  is rigidity index. 1500
The radius where the soil is still in compression is defined as radius PMT DB-09/66
1000
of compression (plastic) zone, , and formulated as: Mecsi:
Mecsi:b = 0.5, c=0,
= 0.5, f= =
c=0, 32o32o
1+si n ϕ 500
2si n ϕ 39 42 44 47 49
σ r +c.cot ϕ
ρ = rc ( σ ) (9) Cavity Wall Radius, R (mm)
ρ +c.cot ϕ
4000
where rc = cavity radius at cavity pressure r and  is horizontal or
radial stress at boundary of compression zone which is defined as:
3500
σ ′ho σu
σρ ≈ [1+ξ-√ (1+ξ)2 -2(1-ξ)2βξ ] +σ′ho (10) 3000
β σ ′ho
Radial Stress, kPA
2500
The radial stress inside the compression zone (at radius r ≤ ) is:
2sinϕ 2000
c ρ 1+sinϕ c
σr = (σρ + ) .( ) - (11)
tanϕ r tanϕ 1500
The radial stress outside the compression zone (at radius r > ) is: PMT DB-09/66
1000
Mecsi:  = 0.9, c=0, = 21o

Mecsi_a=0.9_c=0_phi=21deg
ρ 2
σr =(σρ -σ′ho ). ( r ) +σ′ho (12) 500
39 42 44 47 49
Cavity Wall Radius, R (mm)
The induced radial strain, r:
Figure 6 No Unique c -  values obtained by Mecsi Formulas
4000
1- 1-
σref σr σ ′ho
∆εr = (1-)E [( σ ) -(σ ) ] (13) σc 0.5
o ref ref Es =3500
Eo ( ) (8a)
100
The induced radial displacement Ur: • When3000

PMT stress level is above yield pressure P y,
Radial Stress, kPA
∆εr(i-1) +∆εr(i) aye aye

∆Ur = (r(i) -r(i-1) ) (14)
Esy = Eyo ( P )
2500 σcy
→ Esy = my Eo (
σcy
) (8b)
2
y Py
With the above formulas, it is supposed to be able to derive the c and 2000
 of clayey soils by matching PMT test data curve with the calculated Es = elastic soil deformation modulus at cavity pressure of c
radial stress strain curve, i.e. matching r vs r plot from PMT against Eo = Em = pressuremeter modulus as defined in equation (4)
1500
r vs r plot from the above cavity expansion formulas. Esy = plastic deformation modulus = cy/y = cavity pressure at plastic
part divided by its corresponding strain (from Pressuremeter
Gouw (2017) found that the above formulas could not match PMT test1000
data) PMT DB-09/66
data curve of Jakarta cemented stiff clay, especially in the plastic Eyo = my.Eo = my.Em Mecsi:  = 0.8, c=0, =18o
Mecsi_a=0.81_c=0_phi=18deg
phase of the curve, i.e. the part after yield pressure P y. To match the my = yield500factor
test data curve, many trial and error were done. However, every trial cy = cavity pressure
39 at and
42 above yield 44 pressure 47 49
could only partially match the PMT data curve and gave different set aye = rigidity factor after yield pressure
of , c and  values, i.e no unique values could be obtained. On the

same test data curve, each of the diagram in Figure 6 shows different To find both my and aye, equation 8b is normalized as follows:
values of rigidity index and c –  values! By modifying the aye
Esy σcy
deformation modulus function, i.e. modifying equation (8), Gouw = my ( P ) (8c)
(2017) was finally able to match the PMT test data curve and derive Em y
a more consistent values of c –  of Jakarta cemented stiff clay. The

modified formula is as follows: from PMT data calculate and plot Esy/Em vs cy/Py, the parameter my
and aye can then be obtained by running power function regression
• When PMT stress level is still within the linear range, i.e. within analysis. Figure 7 shows one of the plotted test data. In this case, my
Po to Py, equation (8) needs to be modified into: = 0.6151 and aye = -2.06. Once parameter my and aye are found,
substitute these parameters to equation 8b.
160
1.0 • 20 pre-borehole Pressuremeter tests conducted at cemented stiff

0.9 clay layers.
PMT DB-09/66 • A total of 123 undisturbed samples for laboratory index properties
0.8
tests, triaxial UU, triaxial CIU and consolidation tests.
0.7
Figure 9 to 10 show index and engineering properties of the
Esy / Em
0.6
subsoil. Stiff clay layer is found below 20m depth, it exhibits an
0.5 y = 0.6151x-2.06 increasing SPT blow counts with depth, bulk unit weights vary within
0.4
R² = 0.9249 16.5–18.5 kN/m3 (Figure 9). Plasticity index are mostly within 20 to
60%, water contents fall near the plastic limits, with liquidity indices
0.3
less than 0.30, an indication of stiff clay (Figure 10). Void ratios of
0.2 the stiff clay are found to be within 0.70-1.30, it has specific gravity
0.1
of around 2.63, and water content averaging around 35% (Figure 11).
0.0 SPT NN
SPT Bulk UnitUnit
Bulk Weight, g (kN/m
Weight, 3) 3
g (kN/m )
1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2.0 00 1010 2020 3030 4040 5050 60 60 0 02 24 46 68 810 10
12 12
14 14
16 16
18 18
20 20
22 22 0 20 42
 cy /Py 00 0 0 0 0
Figure 7 Finding my and aye from Pressuremeter Test Data -10

-10 -10-10 -10 -10
-20 -20-20 -20 -20

Figure 8 shows one of the results of PMT test curve matching with
-20
curve calculated from the modified equation (8), i.e. modified E -30
-30 -30-30 -30 -30
function or modified cavity expansion model. The result shows that -40 -40 -40
-40
when the stiff clay is still in linear “elastic” range, the shear strength
-40 -40
consists both cohesion and angle of internal friction (since the shear -50
-50 -50
-50
-50
-50
strength parameters are derived from Pressuremeter, it is notated as -60
-60
-60 -60
-60 -60
cPMT and PMT). However, once the soil entering non-linear plastic
Depth (m)
Depth (m)
Depth (m)
Depth (m)
Depth (m)
Depth (m)
-70 -70 -70
part, the stiff clay lost its cohesion (cyPMT = 0 kPa), and only the angle -70 -70 -70
of internal friction yPMT is working. It is also found that the angle of -80
-80
-80
-80
-80
-80
internal friction remains constant throughout the elastic and plastic -90 -90 -90
phase, i.e. PMT = yPMT. The same outcomes are found from all the -90 -90 -90
PMT test data. -100

-100
-100
-100
-100
-100
-110 -110 -110
-110 -110 -110
4000
-120 -120 -120
-120 -120 -120
-130 -130 -130
-130 -130 -130
3500
Figure 9 SPT Blow Counts and Bulk Unit Weight

3000
Radial Stress, kPA
2500
Calculated by Modified Cavity Expansion Wp Wn WL (%) Wp PlasticityWL
Wn Index,(%)
PI (%) Liquidity Index, LIIndex, PI (%)
Plasticity
0 10 20 30 40 50 60 70 80 90 1000 10 20 0 30 40
20 50 4060 70
60 80 8090 100
100 0.000.1 0.2 20
0.3 0.4 0.5400.6 0.7 60
0.8 0.9 1.0
Before yield: 0 0
80 100
0 00
2000
CPMT = 83 kPa - PMT = 21.2o
After Yield: -10 -10 -10 -10
-10
CyPMT = 0 kPa - yPMT = 21.2 o
-20 -20 -20 -20
-20
1500 -30 -30 -30 -30
-30
PMT DB-09/66
-40 -40 -40
-40 -40
"Modified Cavity_Expansion_Theory"
1000 -50
-50 -50 -50
-50
-60 -60 -60

-60 -60
Depth (m)
Depth (m)
(m)
Depth(m)
Depth (m)
500 -70 -70 -70

-70 -70
Depth
39 41 43 45 47 49
-80 -80 -80
-80 -80
-90 -90 -90
Figure 8 Good Match of PMT Test Data vs Modified Cavity -90 -90
Expansion Theory -100

-100
-100 -100
-100
-110 -110 -110

-110 -110
4. CASE STUDY ON JAKARTA CEMENTED CLAY

-120 -120 -120
-120 -120
-130 -130 -130

A case study was carried out at a project site at Bendungan Hilir, -130 -130
central Jakarta, where many high-rise buildings are located. The

following field and laboratory testings were carried out: Figure 10 Atterberg Limits and Liquidity Indices
• 21 deep borings carried out between 90 to 120 m depths. SPT tests

were taken at every 2 to 3.5 m intervals.
161
Specific Gravity, Gs Water Content, Wn (%) P'c (kPa)

Void Ratio, e Degree of Saturation, Sr Eoed (MPa)
2.2 2.3 2.4 2.5 2.6 2.7 2.8 0 10 20 30 40 50 60 70 80 90 100 0.0 0.5 0 1.0 5001.510002.015002.5 2000
3.0 25003.5 3000 40 50 060 2070 4080 6090 80
100100 120 140 0
0 0 0 0 0 0 0
P'c_Oedometer
-10 -10 -10 -10 -10 -10 -10
-20 -20 -20 -20 -20 -20 -20
-30 -30 -30 -30 -30 -30 -30
-40 -40 -40 -40 -40 -40 -40
-50 -50 -50 -50 -50 -50 -50
Depth (m)
Depth (m)
-60 -60 -60 -60
Depth (m)
-60 -60 -60
Depth (m)
-70 -70 -70 -70 -70 -70 -70
-80 -80 -80 -80 -80 -80 -80

OCR ≈ 2.0
-90 -90 -90 -90 -90 -90 -90
-100 -100 -100 -100 -100 -100 -100
-110 -110 -110 -110 -110 -110 -110

P'c =18z
-120 -120 -120 -120 -120
R² = 0.6334 -120 -120
-130 -130 -130 -130 Eff. Overburden Pessure -130 -130 -130
Water Content, Wn (%)
0.0 2.30.5
Specific Gravity, Gs Void Ratio, e
2.4 1.0 1.5 2.72.0
2.8 2.5 3.0
Degree
Water of Saturation,
Content, Wn (%) Sr
Figure 12 Pre-Consolidation
Void Ratio, e
Pressures and Oedometer Modulus
Degree of Saturation, Sr
0 50 60 70 80 90 100 2.2 2.5 2.6 0 103.5
20 30 4040 50 60 70
50 60 100 100 0.0
70 808090 90 0.5 1.0 1.5 2.0 2.5 3.0 3.5 40 50 60 70 80 90 100
00 0 0 0 0
-10
-10 -10 -10 -10 -10
-20
-20 -20 -20 -20 Drained Cohesion _TXCU
-20C'TXCU (kPa) Drained Friction Angle 'TXCU (degree) Undrained Coh
0 50 100 150 -30

200 250 300 0 10 20 30 40 50 0 50
-30
-30 -30 -30 -30
0 0 0
-40
-40 -40 -40 -40 -40
-10 -10 -10
-50
-50 -50 -50 -50 -50
-20 -20 -20
-60
-60 -60 -60 -60 -60
Depth (m)
-30 -30 -30

-70
-70 -70 -70 -70 -70
-80
-80 -80 -80 -80 -40 -80 -40 -40
-90
-90 -90 -90 -90 -50 -90 -50 -50
Depth (m)
-100
Depth (m)
-100 -100 -100 -100 -100

-60 -60 -60
-110
-110 -110 -110 -110 -110
-70 -70 -70
-120
-120 -120 -120 -120 -120
-80 -80 -80
-130
-130 -130 -130 -130 -130
-90 -90 -90
Figure 11 Specific Gravity, Water Content, Void Ratio and Degree -100 -100 -100
of Saturation
-110 -110 -110
C'TXCU (kPa) 'TXCU (degree) Undrained Cohesian_TXCU CuTXCU (kPa) Undrained Friction uTXCU (degree)
Figure 12 shows the pre-consolidation pressure and oedometer
Drained Cohesion _TXCU Drained Friction Angle
-120 -120 Angle_TXCU -120
modulus. The0 pre-consolidation
50 100 150 200 250 pressures
300 appear
0 10 increasing
20 30 with
40 50 0 50 100 150 200 250 300 0 10 20 30 40 50
0
depth. Comparing
0
with the corresponding effective
0
stresses, the over 0
-130 -130 -130
consolidation
-10 ratio of the stiff clay layers is
-10found to be in the order -10 -10
of 2.0. The effective and total shear strength obtained from triaxial
CIU tests -20
are shown in Figure 13. -20 -20 -20
-30 -30 -30 -30

Figure 14 shows typical PMT test data match reasonably well
with the curve
-40 derived from the modified-40cavity expansion theory -40 -40
described above. The black triangular dots show the PMT test data
point and -50
the dashed red lines show the curve
-50 obtained from modified -50 -50
cavity expansion theory. With this matching of curve, the c and 

Depth (m)
Depth (m)
-60 -60 -60 -60

values of the tested cemented stiff clay can be derived. Note that the
notation of-70 PMT DB-xx/yy in the graphs -70 means the PMT test -70 -70
conducted at borehole no xx at depth of yy meter. Figure 15 shows
-80 -80
the PMT parameters
-80
derived from the test data, all the notations on -80
the graphs-90are as defined before. The effective

-90 horizontal stress ’ho -90 -90
is obtained by subtracting PMT total horizontal pressure Po, with its
corresponding
-100 hydrostatic groundwater pressure,
-100 as formulated in -100 -100
equation -110
(2). It is important to show the value
-110
of effective horizontal -110 -110
stress here as it needs to be implemented in equations (10), (12), and
(13). -120 -120 -120 Figure 13 c’- ϕ’ and cu and ϕu-120
from Triaxial CIU Tests
-130 -130 -130 -130
162
2500 4500
4000
2000
3500
3000
Radial Stress, kPA
Radial Stress, kPA

1500
2500
2000
1000
1500
PMT DB-07/27 PMT DB-01/66

1000
500
"Mod. Cavity_Expansion_Theory" "Mod. Cavity_Expansion_Theory"
500
0 0
36 38 40 42 44 46 48 50 36 38 40 42 44 46 48 50
Cavity Wall Radius, R (mm) Cavity Wall Radius, R (mm)
1800
4500
1600
4000
1400
3500
1200
3000
Radial Stress, kPA
Radial Stress, kPA
1000
2500
800
2000
600
1500
PMT DB-08/43 PMT DB-06/92

400
1000
"Mod. Cavity_Expansion_Theory"
200 "Mod. Cavity_Expansion_Theory"
500
0
0
34 36 38 40 42 44 46 48 50
34 36 38 40 42 44 46 48 50
4000
4000
3500
3500
3000
3000
Radial Stress, kPA
2500
Radial Stress, kPA
2500
2000
2000
1500
1500
PMT DB-04/56
1000
1000 PMT DB-03/86
"Mod. Cavity_Expansion_Theory"
500 "Mod. Cavity_Expansion_Theory"
500
0
0
36 38 40 42 44 46 48 50
36 38 40 42 44 46 48 50
Figure 14b PMT Test Data Points (black triangular points) vs

Figure 14a PMT Test Data Points (black triangular points) vs
Modified Cavity Expansion Theory (dashed red line)
Modified Cavity Expansion Theory (dashed red line)
163
Po, Py, PL (kPa) 'ho(kPa) = P'o = Po - uo depth, with a value of around 95 kPa at a depth of 20 m to 475 kPa at
Em (kPa)
0 1000 2000 3000 4000 5000 0 100 200 300 400 500 600 a depth
0 of 100
20000 40000 m,60000
and 80000
it is 100000
clearly higher than the values obtained
0 0
from
0
CU triaxial test, be the undrained or drained cohesion. The lesser
-10
Po Py PL -10 values of cohesion from triaxial tests are generally attributed to the
-10
-20 -20
brittle nature of the Jakarta cemented stiff clay which tends to suffer
-20
micro cracks resulted from the sampling process by thin wall tube
sampler and during the preparation of the samples in the laboratory.
-30 -30 -30
-40 -40 The higher values of cPMT is attributed to the cemented nature of the
-40
-50 -50
Jakarta stiff clay.
-50
Depth (m)
-60 -60 -60
Depth (m)
Depth (m)
-70 -70 -70 cpeak (kPa) = c before Py  peak (degree) =  after Py

-80 -80 -80 0 100 200 300 400 500 600 0 10 20 30 40 50 60
0 0
-90 -90 -90
-10 -10
-100 -100 -100
-20 -20
-110 -110 -110
-30 -30
-120 -120 -120
-40 -40
-130 -130 -130 -50
-50
Depth (m)
Depth (m)
-60 -60
Em, Eoed (kPa) -70 -70
Thousands
0 50 100 150 -80 -80
0
-90 -90
y = -0.2106x
-10 -100 R² = 0.9099 -100
Eoed
Em -110 CPMT from PMT data -110
-20
Linear (Em)
 PMT from PMT Data
-120 C' from Triaxial CU -120
-30 ' from Triaxial CU
-130 -130
-40
cultimate (kPa) = c after Py  ultimate (degree) =  after Py
-50 0 100 200 300 400 500 600 0 10 20 30 40 50 60
0 0
-60
Depth (m)
-10 -10
-70 -20 -20
-30 -30
-80
-40 -40
-90
-50 -50
y = -0.0011x
Depth (m)
Depth (m)
-100 R² = -0.561 -60 -60
-70 -70
-110
-80 -80
-120
-90 -90
 rc > Py → cPMT = 0
-130 -100 -100
Py < rc < Py →  PMT constant
Figure 15 Pressuremeter Parameters and Oedometer Modulus -110 CPMT from PMT Data -110
-120 -120
C' from Triaxial CU ' from Triaxial CU
-130 -130
Figure 16 and 17 show the c and  values derived from PMT
data, notated as cPMT and PMT, plotted against effective (drained) and
total (undrained) c –  from CU triaxial test, respectively. It can be Figure 16 cPMT and PMT vs Triaxial Drained c’ – ϕ’
seen the c- values derived from PMT data by using modified cavity
expansion give a clear existence of soil cohesion when the stress From all the above phenomena, it can be concluded or at least
strain of the stiff clay is still within the linear “elastic’ range, i.e. cPMT postulated that for Jakarta stiff clay, at the initial stage of
and PMT are mobilized at the same time (since the c and  are derived Pressuremeter test the soil is in partially or near drained cohesion, as
from PMT, they are given PMT indices). However, once the stress the radial stress and strain reaches its yield pressure, P y, the stiff clay
level reaching and above its yield stress level the stiff clay losses the is already in fully drained cohesion. The explanation is: at the initial
cohesion (cyPMT = cultimate = 0), what remain thereafter is the angle of stage, while the radial stress tends to reduce the soil volume, the
internal friction which remain constant throughout all the stress level concurrent induced tangential strain will expand the soil radially,
(ϕyPMT = ϕ peak = ϕultimate). The same outcomes are found from all the therefore the soil is not in a fully compressive nature, but rather in a
PMT test data. This means Jakarta stiff clay exhibits no dilation radial and tangential ring like shearing nature. Consequently, at this
property (ϕpeak - ϕultimate = 0). stage the soil at least is in a partially drained condition. At and beyond
yield pressure, the induced tangential strain will be large enough to
Comparing Figures 16 and 17, from 27m to 97m depth the PMT cause spacings within the clay particles move to a larger distance one
values are within 21o – 33o and these values fall within the drained another and possibly creates micro cracks within the soil structure,
angle of internal friction (Figure 16) rather than the undrained angle hence the clay start to lose its cohesion and left only with its angle of
of internal friction (Figure 17) obtained from triaxial test. The results internal friction, at this stage the stiff clay is already in a fully drained
also show the cohesion parameter of Jakarta stiff clay increases with condition. This postulated phenomenon is illustrated in Figure 18.
164
cpeak (kPa) = c before Py  peak (degree) =  after Py

From 20 m to 100 m depth:
0 100 200 300 400 500 600 0 10 20 30 40 50 60
0 0 cPMT (kPa) = y (m) / 0.2106 (15)
-10 -10
EPMT or Em (kPa) = y (m) / 0.0011 (16)
-20 -20
-30 -30 where y is depth in m.

-40 -40
5. ESTIMATING PILE AXIAL CAPACITY

-50 -50
Depth (m)
-60
Depth (m) -60
-70 -70
The shear strength and the deformation modulus of the stiff cemented
-80 -80 clay obtained from PMT data are applied to estimate pile axial bearing
-90 -90
capacity through finite element analysis by using the axisymmetric
y = -0.2106x model in Plaxis 2D software. The input parameters are presented in
R² = 0.9099
-100 -100
Table 1. The finite element model is shown in Figure 19. Figure 20
-110 CPMT from PMT data -110
 PMT
shows the resulted pile load settlement curve. By applying the
from PMT Data
-120 Cu from Triaxial CU -120 ultimate load criterion set in the Indonesian Geotechnical standard
u from Triaxial CU
(SNI 8640:2017) which set the ultimate load as the load at pile head
settlement of 4% pile diameter, the ultimate pile capacity can be
-130 -130
cultimate (kPa) = c after Py  ultimate (degree) =  after Py estimated.

0 100 200 300 400 500 600 0 10 20 30 40 50 60
0 0 Table 1 Plaxis Input Parameters
-10 -10
-20 -20
-30 -30
-40 -40
-50 -50
Depth (m)
Depth (m)
-60 -60
-70 -70
-80 -80
-90 -90
 rc > Py → cPMT = 0
-100 -100
Py <  rc < Py →  PMT constant
-110 CPMT -110
from PMT Data
-120 -120
CU from Triaxial CU U from Triaxial CU
-130 -130
Figure 17 cPMT and PMT Triaxial Undrained cu – ϕu
Figure 18 Radial Expansion causing Micro-cracks
As found above, the strength parameters of the Jakarta cemented

stiff clay derived from the PMT tests, cPMT and PMT, together with Figure 19 Plaxis Finite Element Model
the PMT deformation modulus, Em, are linearly increasing with depth
and can be written as follows:
165
SPT Blow Count, N (blows/ft)
10
20
30
40
50
60
0
0
-5
-10
-15
-20
-25
-30
-35
-40
Deptn, y (m)
-45
-50
Figure 20 Pile Load Settlement from FEM Analysis -55
-60
4% of 1.5m pile diameter is 60 mm pile head settlement, from -65

figure 19, it can be found that the ultimate capacity of the pile is:
-70
Qult_PMT = 30,395 kN
-75
Figure 20 shows the idealised SPT profile to calculate the pile

axial bearing capacity from the following formula:
-80
-85
Qult (kN) = m Ns As + n Nb Ab (17)
-90
where m = 6 = friction coefficient, n = 40 = base coefficient, N s is

SPT blow count along the pile shaft, Nb is the SPT blow count at pile -95
base; As is the pile skin area and Ab is the pile base cross sectional -100
area.
Idealised DB-01 DB-02 DB-03 DB-04
Based on this approximate SPT formulas commonly adopted in DB-06 DB-07 DB-08 DB-12 DB-13
Jakarta practice, the ultimate bearing capacity of the same pile size Figure 21 Idealized SPT Blow Counts
found is:
Qult_SPT = 30,610 kN Further research is necessary to make sure whether the theory
derived in this study can be applied to estimate the strength
It can be seen the PMT and the SPT results give similar values parameters of other soil types. It will be good if PMT test data can be
of estimated pile axial capacity. done in conjunction with instrumented pile load test data tested to
failure, with this the theory can be further verified.
6. CONCLUDING REMARK 7. ACKNOWLEDGEMENT

To derived c and  values of Jakarta stiff clay from PMT data, Mecsi The author would like to thank Prof. Paulus. P. Rahardjo, and Prof.
model needs to be modified. The deformation modulus need to be A. Aziz Djajaputra for their valuable guidance during the research.
divided into two parts as written in Equation (8a) and (8b). With this To Prof. H. Moeno, R. Karlinasari PhD and S. Herina, for their
modified E function, cavity expansion theory can then be applied to feedbacks. To GEC and PT. Pondasi Kisocon Raya for providing
derive the shear strength parameters. necessary data for the research. Finally, high appreciation also
PMT test in Jakarta stiff clay initially exhibits partially drained attributed to Universitas Katolik Parahyangan for facilitating the
condition and then gradually become fully drained condition when research.
reaching and beyond its yield pressure. The c and ϕ values obtained
from Pressuremeter test are effective stress parameters. The 8. REFERENCES
Pressuremeter test can reveal the effect of cementation of Jakarta stiff
clay which appear in a higher value of cohesion which cannot be Baguelin, F., Jezeqel, J.F., Lemee, E., and Le Mehaute, A. (1972)
captured by triaxial test due to the difficulty in obtaining a good “Expansion of Cylindrical Probes in Cohesive Soils”, JSMFE,
‘really’ undisturbed Jakarta stiff clay samples by normal thin wall ASCE, 98; SM11. Proc. Paper 9377, pp1129-1142.
tube sampler. Baguelin, F., Jezequel, J.F., and Shields, D.H. (1978) The
The axial pile bearing capacity calculated by finite element Pressuremeter and Foundation Engineering, Trans Tech
method with strength and stiffness parameters derived from PMT test Publication, Switzerland.
is comparable with the calculated bearing capacity of SPT formula Briaud, J.L. (1992) The Pressuremeter, A.A. Balkema, Rotterdam.
commonly used in Jakarta’s practice. Briaud, J.L. (2013) Geotechnical Engineering: Unsaturated and
Saturated Soils, John Wiley & Sons, New Jersey, USA.
166
Clarke, B.G. (1995) Pressuremeters in Geotechnical Design, Blackie

Academic and Professional, London.
Clayton, R.I., Simons, N.E., and Matthews, M.C. (1982) Site
Investigation A Handbook for Engineers, Granada Publishing,
London
Gambin, M. (1980) “A Review of the Menard Pressuremeter over the
Last Twenty Years in Europe”, Sol Soils, 32, Paris.
Gambin, M. (1995) “Reasons for the Success of Menard
Pressuremeter”, Proceedings of Fourth International
Symposium on Pressuremeters, May 17-19, 1995, Sherbrooke,
Quebec, Canada.
Gambin, M. and Frank, R. (2009) “Direct Design Rules for Piles
using Menard Pressuremeter Test”, Foundation Design with
Menard Pressuremeter Test, French Contributions to
International Foundation Conggress & Equipment, Expo ’09,
pp3-10; also in ASCE Geotechnical Special Publication no.
186, pp111-118.
Ghionna, v., et al. (1981) Performance of Self-boring Pressuremeter
Tests in Cohesive Deposits, Report FHWA/RD-81/173/1981,
MIT, Boston.
Gouw, Tjie-Liong (2017) Shear Strength Derivation of Jakarta Stiff
Clay by Use of Pressuremeter Test based on Modified Cavity
Expansion Theory, PhD Dissertation, Universitas Katolik
Parahyangan, Bandung, Indonesia.
Mecsi, J. (2013) Geotechnical Engineering Examples and Solutions
Using the Cavity Expanding Theory, Hungarian Geotechnical
Society, Hungary.
SNI 8460:2017 (2017). Standar Nasional Indonesia - Persyaratan
perancangan geoteknik. Badan Standardisasi Nasional.
167
Volume Measurement for Heavy Pounding Induced Craters and Ground Heave
J. L. Hung1, U. H. Amalia1, H. C. Shih2, and J. C. Chih3
1
Department of Civil and Construction Engineering, National Taiwan University of Science and Technology, Taipei, Taiwan
2
Taiwan Building Technology Center, National Taiwan University of Science and Technology, Taipei, Taiwan
3
Li-Jia Engineering Company, Taipei, Taiwan
E-mail: amaliaula@gmail.com
ABSTRACT: Traditionally, the volume of dynamic compaction induced crater and ground heave is measured by means of level surveying
and ruler measurement. However, since ground heave around the crater and the shape of the crater itself are irregular, it is not only difficult
but also time-consuming to accurately measure the volume of crater and ground heave. This study proposes a method that adopts the up-to-
date image processing (photogrammetry) technology to more accurately measure the crater volume and the ground heave around it. A
commercial software, which is initially used for the drone, is used here to generate point cloud of the crater and its surrounding area using the
images captured with a video camera or smartphone. The accuracy of this method was calibrated with a known volume box in the laboratory
first before it was used in a field trial test. This paper will present and discuss the operation procedure and image processing of this method.
The crater volume measured from the photogrammetry method is compared with that measured from the traditional measuring method. It is
found that the volume of DC crater can be better approximated by cone shape crater than by truncated cone shape crater, which is commonly
used in the DC industry and can be seriously over-estimate the actual volume of DC craters.
Keywords: crater volume, ground heave, dynamic compaction, photogrammetry method.
1. INTRODUCTION
Ground improvement by heavy poundings like dynamic compaction
(DC) and rapid impact compaction (RIC) is a commonly used
method for the densification of in-situ sandy soil to a large depth.
After pounding, a crater on the ground is formed, and surface
heaved. To evaluate the effectiveness of pounding, the volume
change of crater and ground heave before and after pounding needs
to be measured. Traditionally, the volume of crater and ground
heave was measured using level surveying. However, since ground (a) Dynamic compaction (b) Rapid Impact
heaves around the crater and craters itself are irregular in shape, it is (DC) Compaction (RIC)
difficult to be accurately measured with traditional surveying
Figure 1. Dynamic compaction and rapid impact compaction
method. This study will use the up-to-date image processing
methods
technology (photogrammetry) to measure the crater volume and the
ground heave around it with reasonable accuracy. A commercial
software (Pix4Dmapper) which is initially used for the drone is
adopted here to generate point cloud of the crater using the images
captured with a smartphone camera. In this study, the accuracy of
this method was calibrated with a known volume box in the
laboratory first before it was used on a pilot test project of DC and
RIC densification on a land reclamation site, which was reclaimed
by hydraulic fill method.
2. TEST SITE CONDITIONS

The backfill material of the reclaimed site was mostly silty sand soil
pumped in by dredging boat from the nearby seabed. As shown in Figure 2. Electrical resistivity image profile (RIP) of
Table 1, from ground surface to GL-8 m, it was silty sand layer with reclaimed land at the test site
trace of gravel and shells and N = 2~20; from GL – 8 m to GL -16
m, it was silty sand layer with trace of clay and N= 4~16; from GL - The SPT-N values of the hydraulically filled sandy soil at GL -
16 m to GL -25 m, it was sandy silt layer with clay and fine sand 5 m ~ GL -9 m (= 4 ~ 10) was weaker than the soil near the surface
and N = 7~28; from GL -25 m ~ GL -49.4 m, it was silty sand layer (GL ~ GL -5m). Since the site is located right next to the sea, the
with N = 14 ~ 23. The hydraulically filled sandy soil (GL to GL -16 groundwater level was high and varied from GL -3 m to GL - 4.1 m
m) was loose and sensitive to vibration; it will be very likely to due to the tidal effect. The electrical resistivity image profile (RIP)
liquefy during earthquake shaking. To improve the engineering of the reclaimed land at the test site is shown in Figure 2. The silty
properties of the reclaimed land against earthquake-induced sand layer (backfilled sand layer) from GL -2 to -7 m has a much
liquefaction, the dynamic compaction (DC) and rapid impact lower electrical resistivity. It indicates a higher water content and
compaction (RIC) methods (Figure 1) were tested during the pilot looser density of this layer.
test of this ground improvement project.
2.1. Pounding Plan of the Pilot Test
Table 1. Simplified soil profile of the reclaimed land The pilot test site covered an area of 20 x 20m2. The layout of the
Depth Thick SPT-N 
pounding points is shown in Figure 3. The DC pounding work was
Soil  (t/m3)  divided into three stages to avoid excessive pore water pressure built
(m) (m) (Avg.) 
up in the ground. As planned, the first and second stages were to be
0-8 SM-1 8.0 2 – 20 (8) 4.8 ~ 33.6 1.52 ~ 2.13 30o
pounded 25 times per stage; the third stage was pounded 20 times.
8 - 16 SM-2 8.0 4 – 16 (9) 12.4 ~ 32.8 1.61 ~ 2.11 30o Pounding point was 10 m spaced at each stage. The pounding
energy per impact was about equal to 500,000 kg-m with a 26,400-
16 - 25 SM-3 9.0 7 – 28 (16) 11.7 ~ 39.7 1.63 ~ 2.11 32o
kg hammer and 19 m free fall (Chow, 1992). The hammer has a
25 - 49.4 SM-4 24.4 14 – 23 (22) 18.3 ~ 31.4 1.69 ~ 2.11 34o square shape footprint of 1.75 x 1.75 m in dimension and the height
168
of 2 m. After pounding, a crater was formed on the ground surface,

and its volume was measured by traditional measuring method (i.e.,
only measure the depth and top diameter of the crater with rulers)
and by the photogrammetry method.
Meanwhile, the rapid impact compaction (RIC) method was also
tested here. RIC is a track-mounted machine that imparts energy by
dropping an approximately 7,500 kg weight from a falling height of
about 1 m onto a 1.5 m diameter steel plate that is placed on the
ground surface. The time interval between each blow was 1.2-1.5
sec, and the impact rate was about 40-50 blows per min. In each
pounding points, several sets of pounding were performed. One set
of pounding means the hammer assembly has penetrated to the
predetermined depth, or the hammer blow has reached the maximum
number. In this test, the maximum blow number was set to 15 ~ 40
blows and the predetermined penetration depth is limited to 0.9 m.
RIC has its onboard data acquisition system that automatically
records information such as drop height, number of blows, and
penetration per blow during the compaction process (Simpson,
2008).
10 m 10 m
1st stage pounding

10 m
2nd stage pounding
3rd stage pounding

10 m
Figure 3. Layout of pounding points for different pounding stages
Table 2. Design requirement of SPT-N values after dynamic

compaction
Depth (m) SPT-N
GL ~ GL -5 m >16
GL -5 m ~ GL -8 m >19
GL -8 m ~ GL -10 m >21
The traditional survey work carried out on this site includes the
ground surface elevation survey around the crater and the volume of Figure 5. Field surveying work for ground heave and crater volume
the crater. The ground surface survey was carried out by a level
survey on the mark points set on the ground surface (Figure 5). The The evaluation of densification effect is typically done by using
crater depth might be measured when the hammer is still inside the the relationship among accumulated pound-in volume and heave
crater or lifted out of the crater. Its purpose is to calculate the heave volume (Note: plus backfill volume if backfill material is added
and sag volumes around the crater. The area that calculated in the during pounding) and some impacts. Tang (2016) measured the DC
crater volume approach is the area of a grey area in Figure 5. The crater volume in each impact pounding. As shown in Figure 6c, the
heave volume is calculated where the heave of the ground surface is heave volume remained unchanged until it reached the 27th impact;
significant. after 27th impact, the ground heave began to increase, and the
densification effect of DC began to decrease. This information can
 be provided to engineer to decide the optimum number of DC
Vh  h1 ( D 2  3 a D ) (1)
impact from the pilot test. However, there are two distinct problems
3
here. Firstly, it needs to do the survey work after each DC impact.
Where: As shown in Figure 6b and Eq. 5, to obtain the pound-in volume and
Vh = heave volume (m3) heave volume of the ith impact pounding, an ith run of survey needs
h1 = highest height of measured heave (m) to carry out, and the measurement results of (i-1)th impact pounding
D = distance of affected heave area (m) are needed. It is a very time-consuming surveying process for the
a = radius of the crater (m) pilot test. Secondly, it is not easy to measure the volume of crater
and ground heave with reasonable accuracy using the traditional
So the pound-in volume resulted by the DC pounding (Figure surveying method (Tang, 2016).
6a) can be obtained to evaluate the densification effect of DC to the
ground soil.
V p , i  (Vc , i  Vc ,i 1 )  Vh , i (3)
V p  Vc  V h (2)
Where:
Vp,I = pound-in volume at ith number of pounding
Where:
Vc,I = crater volume at ith number of pounding
Vp = pound-in volume
Vc,i-1 = crater volume at (i-1)th number of pounding
Vc = crater volume
Vh,I = heave volume at ith number of pounding
Vh = heave volume
169
surveying method for the crater. Preferably, this can be done with
simple devices and at least effort.
2.2 Photogrammetry method
Recently, many image processing programs are available for UAV
or drone to map specific area or landscape from the sky. These
programs use the photogrammetry technology to process the
captured images. The digital images of an area captured by drone
were put into the software to get the 3D digital mapping of the area
(a) including the elevation and plane dimension of the building or
landscape. The process of photogrammetry starts with capturing
images from at least two different locations, then common points are
identified on each image and rays line are developed from each
camera angle to points on the object. These rays are intersected to
produce 3D coordinates of the points of interest. (Li, 2016)
Area captured
Drone by drone
(b)
Landscape Area
Figure 8. UAV or drone used to capture pictures for the

photogrammetry
(c) This study attempts to apply the photogrammetry method and

use the commercial image processing software to the smaller object:
Figure 6. Measurement of heave volume and pound-in volume from the DC and RIC-induced craters. Digital images of the crater are
the site surveying after DC pounding (Tang, 2016) taken on site by a smartphone or digital camera. Then these images
put into the Pix4Dmapper software to generate a 3D point cloud
model of the crater. The volume of the crater can be computed using
the generated 3D model.
3. IMAGE PROCESSING
3.1 Point Cloud Model
The core of photogrammetry method is to generate a 3D object
model by using the captured surface images of the object. The image
processing software converts images into 3D models and point
clouds. Having the point cloud data (e.g., the crater), the volume of
the crater and the elevation change of the surrounding ground
surface can be calculated.
3.2 Image Acquisition Plan
To establish the point cloud of the object, the number of digital
images taken on the site should be from various angles to visualize
its point clouds. Then the point clouds are used to reconstruct the
model from the initial digital image. To obtain good images for
processing, digital images should be taken around the object. It
requires each image to overlap with its neighboring image to get
high accuracy results. The higher the overlapping image rate, the
more clear model can be constructed. The sufficient distance
(ground sampling distance, GSD) of taking digital images should be
carefully considered, and GSD serves as necessary measurement
Figure 7. Problems associated with crater volume measurement in calculation. Therefore, it is crucial to make a good image acquisition
the traditional measuring method plan to capture images at various angles and distances. (Pix4D, 2017)
The GSD is the distance between the center of two consecutive
As shown in Figure 7, the rim of crater appears on the ground pixels on the ground. It influences the accuracy and the quality of
is not the actual rim of the crater. Therefore, the measured crater the results as well as the details that are visible in the final results.
diameter will be larger than the actual crater diameter. As a result, it GSD related to the camera height and camera properties. Therefore,
will overestimate the crater volume and the densification effect of
the image acquisition plan should maintain at least two different
the DC. There is a need to have a more accurate and quicker camera height. While the image acquisition plan for the crater
170
depends on the GSD and overlap, the overlap of pictures depends on Figure 10. Generated point cloud model by the image acquisition
the affected area of the crater. In this case, crater volume and ground plan
heave will be measured; overlapped images should contain all the
area affected by the pounding. For crater volume and ground heave Following are the standard operation procedures (SOP) of digital
measurement, the following circular image acquisition plan is image processing adopted in this study:
recommended: (1) Input digital images
(1) Moving around the crater for the first time with a camera dip- Based on the image acquisition plan, digital images are
angle of 60o. obtained from the site. Input digital images to the Pix4Dmapper
(2) Moving around the crater for the second and/or third time while software. Select Processing option of 3D Maps, and start
reducing the camera height and the dip-angle in each round. processing.
(3) It is best to take an image every 5 to 10 degrees when walking (2) Processing Step 1: Initial Processing
around the crater and take pictures to ensure enough overlap. Select the Initial Processing option, and click start. When start
More images should be taken for shorter distances. Step 1, Pix4Dmapper computes the key points on the images. It
Recommended overlap is at least 75% of common points. The uses these key points to find matches between images.
camera should be at a constant height above the crater as much (3) Processing Step 2: Point Cloud and Mesh
as possible to maintain the GSD value. After Step 1 Initial Processing is completed, Step 2 Point Cloud
and Mesh can be processed. Select Point Cloud and Mesh
option. Click start to begin Step 2 processing. In this step, the
3.3 Digital Image Processing
point cloud is generated and can be visualized in rayCloud.
Since the images acquisition needs to change camera heights and (4) Scale the Model
angles, a length adjustable monopod was used to take pictures while After finish Step 2, the initial point cloud data are generated.
walking around the crater. The images taken must cover the whole However, these data need to be adjusted using the scale or
crater and also the possible heaving area around it. To enhance the ground control points (GCPs) on site. Once the point cloud has
stability of smartphone/digital camera and to assure better image been adjusted with the scale, it needs to be reoptimized. After
overlapping, a DJI OSMO Mobile® was used here. OSMO re-optimization, the processing of Step 2 is re-generated.
Mobile® can also maintain the camera angle by its built-in gimbal (5) Processing Step 3, DSM, Orthomosaic and Index
and facilitate the process of image taking on site. DSM (Digital Surface Model) is a 2.5D model of the mapped
area. Each pixel and each point of the vector point cloud
contain (X, Y, Z) information. Running Step 3 means that there
is information about the coordinate of the measuring object.
60°
After processing Step 3, the volume of the crater can be
obtained. Select the area of interest and click Compute.
Start
45°
h2 Image acquisition in
the site
Point of image taking
h1
Input images to
Area affected by pounding Pix4Dmapper
Crater Processing Step 1.

Initial Processing
Figure 9. Image acquisition plan for crater measurement Processing Step 2.

Point Cloud and
After mounting the smartphone/digital camera to the OSMO Mobile, Mesh
the joystick handle was used to control the camera function of the
smartphone. For the maximum use, the extension rod of OSMO Yes
Scale the
Mobile was also used to get the additional height of the images. The model
process of crater images acquisition includes (1) adjust and fix the First running Apply,
smartphone/camera angle, (2) hold the camera to the first height (h1), Step 2? reoptimize
(3) walk step by step around the crater and take digital pictures at point cloud
Set GCPs
Yes
each walking step. Carry out another round with different camera (optional)
angle and height. No
Processing Step 3.
DSM, Orthomosaic
Number of image and Index
input with different
height and camera Assign area to
dip-angle
calculate volume,
compute.
Finish
Figure 11. Standard operation procedure (SOP) of digital image

processing using photogrammetry software
171
4. VOLUME MEASUREMENT Table 3. The standard deviation of the box volume measurement
Standard % of
4.1 Calibration Box Volume Volume Mean Deviation
Deviation Error*
No.
To calibrate the accuracy of volume measurement, a box with (cm3) (cm3) (cm3) (cm3) (%)
known dimension is used here as the sample object to check the 1 2947.32 11.95
2 2939.70 4.33
accuracy of volume measured by this photogrammetry method. The
3 2880.56 2935.37 -54.81 42.13 1.46
dimension of the box is 28.5 x 16.5 x 6.8 cm (Figure 12) with the 4 2914.69 -20.68
inside volume of 2894.74 cm3. Unfortunately, the default unit 5 2994.58 59.21
“meter” used in Pix4Dmapper cannot be changed on the computer 42 . 23
* % of error   100  1 . 46 %  1 . 5 %
screen, and the output value of measured volume can only be 2894 . 74
displayed to two decimal places. As a result, the output volume of
the calibration box from Pix4Dmapper is 0.00 m3. It is unacceptable
4.2 Photogrammetry Measurement of Craters
for the volume measurement here. To deal with this problem, the
default unit of the software is disregarded, and the metric unit in the In the pilot test site, both DC and RIC craters were measured with
software is regarded as “cm”. Fortunately, this change on the default the photogrammetry method following the above mentioned image
unit did not affect the operability of volume calculation of this acquisition plan. Not only the crater volume but also the ground
software. Figure 13 shows the volume determined for the box using heave were measured. The captured images covered an area which
the Pix4Dmapper software. was large enough to cover the ground control points (GCPs) around
the crater and also the surrounding ground heave. In addition, both
the digital images before and after DC pounding were taken and
compared. GCPs adopted here were with known coordinates and to
6.8 cm be used for creating the initial coordinates and measure the elevation
change of ground surface caused by DC pounding.
16.5 cm Following are the steps of this photogrammetry procedure adopted
for the crater measurement:
(1) Taking images of initial site condition. In each impact
location, at least three points are marked on the ground surface
25.8 cm and used as GCPs. One of the GCPs is used as the benchmark
and assigned the initial coordinate of (0,0,0). The coordinates
Figure 12. Point cloud image of the calibration box of the other GCPs should be decided based on their distances
and directions to the benchmark GCP. It is essential to make
sure that these GCPs cannot be moved during DC pounding.
They are supposed to work as permanent GCPs (Figure 14).
Figure 14. Images acquisition process of the initial site condition

with its GCPs (Ground Control Points)
Figure 13. Volume measurement result of the calibration box
using Pix4Dmapper software (2) Taking images after-pounding. The images captured after
pounding should cover not only the crater but also the GCPs.
The volume of the box was measured five times by the Therefore, the elevation difference before and after pounding
photogrammetry method following the standard operation can be compared and calculated (Figure 15).
procedures (SOP) mentioned above. On Step 3, different measured
volumes of the box were found among the five measurements (Table
3). Such a difference in measured volume were mainly resulted from
the process of manual selection of box boundary from the point
cloud data. Slightly different box boundary may be selected from
different runs of measurement. The average volume obtained from
the output is 2935.37 cm3; while the actual inside volume of the box
is 2894.74 cm3. The standard deviation of the five-volume
measurements was 42.13 cm3. Compared to the actual volume of the
box, the error is only 1.5% of the box volume (Table 3). Such an
error should be acceptable for the heavy pounding induced crater
volume measurement. It confirms that this photogrammetry method
and Pix4Dmapper software are applicable to the volume
measurement of a smaller object (like the calibration box and the Figure 15. Images acquisition process after-pounding
crater) using a smartphone/digital camera.
172
(3) Processing images of the initial site condition. Input the measure the top diameter of the crate during traditional crater
images data of initial site condition to the software and run the measurement. The intersection of the cross was used to locate the
process. Get the point cloud data of the initial site condition. measuring location for the crater depth. The yellow ball inside the
Input the coordinate of benchmark GCP (0,0,0) and also the crater was used to keep the camera focus on the crater during images
other GCPs. Each GCP needs to be precisely clicked on at least capturing process.
three images of the point cloud. Input the adjusted GCPs then
re-optimize the point cloud (Figure 16).
Crater R1 Set 1
Figure 16. Setting the GCP point coordinate and selecting in

the at least three images
Crater R1 Set
(4) Set elevation of the initial site condition. After the point
cloud has been re-optimized, the elevation of the ground
surface can be set. Choose the area of interest which will cover
the crater and its surrounding area and export as shapefiles
(.shp). The elevation of the selected area will be set as the
Crater R1 Set 1
initial elevation for the calculation of crater volume and ground
heave afterward.
(5) Processing after-pounding condition. Input the images
captured after the first pounding. Since the GCPs do not move, Figure 18. (a) Photo taken at the site, (b) Point cloud model (bird’s
the GCP coordinates for after-pounding are the same as the eye view), (c) Point cloud model (cross view) of the third stage of
initial site condition. Repeat the process adopted for the initial RIC crater (R1) at first set
site condition to get the point cloud of the first pounding.
(6) Crater volume measurement. Import the shapefiles (.shp) of
the initial elevation and then compute the crater volume. At the
same time, the ground heave can also be calculated (Figure 17).
Crater R1 Set 2
(a) (b)
Figure 17. (a) Selected elevation area on initial condition that
exported as shapefiles (b) Volume measurement with imported
shapefiles after pounding
4.2.1 Point Cloud of Craters

Figures 18 and 19 show the pictures of the first stage RIC crater
after two sets of pounding. The RIC craters mostly have a shape Crater R1 Set 2
close to a conical shape which is different from the real shape of the
crater formed by RIC pounding. Because, after each set of RIC
pounding, the hammer assembly of RIC was lifted up from the
crater. In the process, it also brought out some soil with it and
formed a crater with an upside down conical shape.
Figure 20 and 21 show the pictures of third stage DC craters Crater R1 Set 2
and their point cloud models (Crater D3). It can be seen from the
picture that the shape of the crater is irregular, but is close to conical
or truncated conical shape. The conical shape assumption can be Figure 19. (a) Photo taken at the site, (b) Point cloud model (bird’s
confirmed by the cross-section of the point cloud model shown in eye view), (c) Point cloud model (cross view) of the third stage of
the figures. The white cross shown in the picture is the ruler used to RIC crater (R1) at the second set
173
4.2.2 Craters Volume Measurement

On the site, crater volume was measured with the traditional
surveying method as well as the photogrammetry method. For the
former, measuring rulers (Figure 7) was used to measure the depth
and top diameter of the crater; for the latter, smartphone or digital
camera was used to capture the images of the crater and the GCPs
around it (Figure 15). The craters were measured after the DC
hammer was lifted out. Without the hammer in the crater, the soil on
the side of the crater would fall off and changed the apparent shape
Crater D3 Pounding 2
of the crater from truncated conical to more or less conical.
Nevertheless, it was only the shape changed, and the volume of the
crater may not change much.
The dimension of RIC craters measured from the traditional
surveying method yielded the top diameters ranging from 1.5 to 3 m
and depth from 0.5 to 0.9 m. The top diameter of DC craters ranged
from 2.5 to 5.0 m and depth from 0.5 to 2.0 m. Since the RIC crater
is notably smaller than DC, it requires less amount of images than
those of DC crater.
For the photogrammetry method, several trial and error tests had
been performed. The first trial was to determine the adequate input
number of images while considering the processing time and
accuracy of crater volume measurement. The initial trial number of
input images starts from 40. However, 40 picture images were not
enough to generate a point cloud model and some of its details
outside the crater, where the GCPs were usually located, were
missing. Thus the ground heave outside the crater cannot be shown.
Until the number of input images reached the threshold value of 110
for DC and 70 for RIC, the point cloud model started to show the
complete crater details and ground heave around it. However, the
threshold input number of the images tends to differ from one crater
Figure 20. (a) Photo taken at the site, (b) Point cloud model (bird’s to another. Since each crater has its details and needs different
eye view), (c) Point cloud model (cross view) of the third stage of threshold images to generate the point cloud model, the processing
DC crater (D3) at second pounding time to generate the point cloud data varies. In addition, the
processing time also depends on the CPU of the computer and the
number of input images. Based on the experienced obtained from
this study, the processing time is around 20-27 minutes for RIC
craters and 30-45 minutes for DC craters (Note: 110 images needed
for DC and 70 images for RIC).
After the threshold number of images had been decided and the
point cloud and 3D models of the site could be generated and scaled,
then the crater volume could be computed. Since most of the craters
did not show significant ground heave, so the volume change mainly
Crater D3 Pounding 7 results from the forming of the crater. The traditional measuring
method only measures the depth and top diameter of the crater. It is
not easy to accurately measure the volume of the crater with such
limited data especially the shape of craters is irregular. To simplify
the volume calculation of the crater, the traditional measuring
method uses the following formulas to approximate the crater
volume. Among them, Eq. 4 formula is for the cone shape crater; Eq.
5 is for the truncated cone shape crater. As shown in Figure 5, Eq. 5
is the commonly used formula to calculate the volume of the
truncated cone shape DC crater.
 r 2h
Vc1  (4)
Crater D3 Pounding 7 3
 h 2
Vc 2  ( a  ab  b 2 ) (5)
3
Crater D3 Pounding 7 Where:
Vc1 = volume of the crater with a conical shape
Vc2 = volume of the crater with a truncated conical shape
r = radius of crater measured on site
a = on-site measured top radius of the crater
b = effective radius of hammer footprint = 1.0 m (converted
Figure 21. (a) Photo taken at the site, (b) Point cloud model (bird’s from square to circular shape by means of equivalent area
eye view), (c) Point cloud model (cross view) of the third stage of method)
DC crater (D3) at seventh pounding h = measured depth of the crater
174
The volume calculated from the photogrammetry method (Vc,p)

and field measurement (Vc1 and Vc2) are listed in Tables 4 and 5. In
total, 3 RIC craters (R1, R2, and R3) and 3 DC craters (D1, D2, and
D3) are listed here. R1 and D1 represent the 1st stage pounding
(refer to Figure 3) for RIC and DC; R2 and D2 are for the 2nd stage
pounding; R3 and D3 are for the 3rd stage pounding. In general, the
ground becomes stronger when the compaction proceeds from the
first stage to the third stage. So the crater volume reduces as the
impact stages escalate.
For the RIC cases, each crater was subjected to two impact sets;
each set has 15-40 blows. As mentioned before, the RIC resulted
craters can only be measured after completion of each set, and the
craters are generally in a conical shape. Therefore, only Vc1 formula
is used to calculate the volume of RIC craters based on the field
measured crater depth and top diameter from the traditional
measuring method. The point cloud data generated from
Pix4Dmapper software for the RIC craters are shown in Figure 22.
The comparison between crater volume determined from
photogrammetry method (Pix4Dmapper) and traditional measuring
method is shown in Figure 23.
Crater R1 Set 1
Crater R2 Set 2
Crater R3 Set 1
Figure 22. 3D model from point cloud data representative of each

stage RIC craters (cross views)
Table 4. Comparison of RIC crater volume between

Photogrammetry and traditional field measurement
Photogrammetry
Traditional Field Measurement
Measurement
Crater Heave Top Vc1  Vc, p Figure 23. Comparison graphs between photogrammetry and field
Crater No. Depth
Volume Volume dia. Vc1 measurement of RIC craters
(h) Vc , p
(Vc,p) (Vh,p) (2r)
For the DC cases, each pounding stage had 20-25 poundings, but
m3 m3 m m m3 %* only the first ten poundings are shown in Table 5. The DC hammer
R1 1 1.60 0.08 0.88 3.00 2.073
used in this case had a square shape footprint (Figure 24). However,
29.6%
during the free fall process, the hammer tends to rotate a little during
2 1.25 0.02 0.70 2.40 1.056 -15.6% each pounding and forms a more or less circular crater on the
ground surface after a few impacts (Figure 24). The crater resulted
R2 1 1.57 0.01 0.84 2.55 1.430 -8.9% by this DC hammer can be somewhat in a cone shape or truncated
2 1.67 0.05 0.80 2.93 1.798 7.7%
cone shape except for the 1st impact. As the number of impact
increases, the crater depth increases. However, the crater shape
R3 1 1.40 0.02 0.83 2.38 1.231 -12.1% remains close to conical. Together with the on-site measured depth
and a top diameter of the crater, Eq. 4 and 3 were used to calculate
2 0.95 0.16 0.53 1.85 0.475 -50.0% the crater volume. Craters (D1, D2, & D3) from different pounding
* “-” means crater volume is underestimated stages were chosen for comparison here.
175
Figure 25. 3D model from point cloud data representative of each

stage DC craters (cross view)
The methods (traditional measuring and photogrammetry) used

for the DC crater measurement are the same as those for the RIC
craters. However, the craters formed by DC tend to be much larger
due to larger pounding energy. Therefore, the area size to be
covered by the photogrammetry method is larger and more images
are needed to generate the point cloud model after each DC crater
measurement. The results of DC craters measurement are discussed
in the following section.
Table 5. Comparison of DC crater volume between photogrammetry

Crater D1 Pounding 5 and traditional field measurement
Photogrammetry
Traditional Field Measurement
Measurement
Figure 24. 3D model from point cloud data of a DC crater for Crater Crater Vc1  Vc, p V2  Vc , p
different pounding (birds-eye view) (contd.) Heave Top
No. Depth
Vol. Vol. Dia. Vc1 Vc , p Vc2 Vc , p
(h)
(Vc,p) (Vh,p) (2r)
m3 m3 m m m3 %* m3 %*
D1 1 1.42 0.24 0.44 2.10 0.508 -64.2% 1.598 2.3%
2 1.79 0.49 0.67 3.05 1.632 -8.8% 3.658 90%
3 2.19 1.00 0.92 3.35 2.703 23.4% 5.644 141%
4 2.74 0.51 1.05 3.30 2.994 9.3% 6.320 116%
5 4.04 0.24 1.13 3.40 3.420 -15.4% 7.064 63.7%
6 4.66 0.31 1.28 3.50 4.105 -11.9% 8.307 67.2%
7 4.95 0.42 1.31 3.65 4.569 -7.7% 8.983 70.6%
8 5.63 0.38 1.39 3.65 4.848 -13.9% 9.531 59.2%
9 6.11 0.46 1.44 3.80 5.444 -10.9% 10.42 60.7%
10 7.38 0.13 1.57 3.75 5.780 -21.7% 11.16 42.4%
D2 1 1.34 0.44 0.40 2.10 0.511 -61.8% 1.527 4.0%
2 2.76 0.24 0.72 3.05 1.150 -58.3% 3.087 2.7%
3 3.67 0.14 0.95 3.35 2.390 -34.9% 5.290 34.2%
4 4.51 0.10 1.16 3.30 3.287 -27.1% 6.955 44.1%
5 4.62 0.38 1.25 3.40 3.783 -18.1% 7.815 58.4%
6 4.82 0.73 1.34 3.50 3.937 -18.3% 8.220 59.6%
7 4.80 1.60 1.50 3.65 4.674 -2.6% 9.555 86.6%
8 5.97 1.30 1.51 3.65 5.296 -11.3% 10.39 63.7%
9 6.84 0.21 1.61 3.80 6.411 -6.3% 12.06 66.4%
10 7.02 0.29 1.63 3.75 5.842 -16.8% 11.38 52.5%
D3 1 1.47 0.21 0.68 2.24 0.893 -39.2% 2.632 63.5%
2 2.05 0.10 1.06 2.75 2.099 2.4% 5.121 131%
3 2.24 0.08 1.22 3.20 3.271 46.0% 7.065 194%
4 2.77 0.11 1.21 3.34 3.534 27.6% 7.395 150%
5 3.65 0.39 1.25 3.33 3.629 -0.6% 7.610 95%
6 3.85 0.19 1.42 3.35 4.172 8.4% 8.711 112%
Crater D1 Pounding 9 7 4.22 0.08 1.41 3.55 4.652 10.2% 9.322 107%
8 4.74 0.09 1.48 3.70 5.304 11.9% 10.33 105%
9 4.62 0.04 1.60 3.72 5.797 25.5% 11.25 129%
Figure 24. 3D model from point cloud data of a DC crater for
different pounding (birds-eye view) 10 5.15 0.15 1.56 3.77 5.805 12.7% 11.16 104%
176
* “-” means crater volume is underestimated measuring method can under-estimate the crater volume by 12~50%
(Crater R3 in Figure 23).
As for DC case, the crater volumes calculated from Eq. 4 and 5
are compared with those calculated by photogrammetry method. As
shown in Table 5 and Figure 26, the crater volume calculated from
Pix4Dmapper software is in better agreement with Eq. 4 for conical
shape crater with the maximum difference of 65%. The larger
difference appears in the initial impacts. As the number of impacts
increased, the crater shape became closer to conical; then the
difference can be reduced to about 20%. In comparison, the crater
volume calculated from Eq. 5 (Note: the most commonly used
formula used by the DC contractors) tends to overestimate the
Figure 26. Comparison between photogrammetry and field volume of DC crater by more than 65%. It implies that if the
measurement of DC crater volume (contd.) hammer was not in the crater during crater measurement, the
truncated conical shape assumed for the crater by Eq. 5 could be
quite different from the apparent crater shape and overestimates the
volume of crater. So if the crater volume is used to evaluate the
effectiveness of compaction as shown in Figure 6, such an
overestimation on crater volume should be taken into account.
Also shown in Figures 23 and 26, the RIC and DC induced
ground heave is not apparent in this pilot test. In other words, most
of the compaction energy had been absorbed by the ground. It
indicates that the in-situ ground and groundwater conditions
underlying the site are suitable for either DC or RIC method.
6. CONCLUSIONS
Among the parameters monitored for DC or RIC method, the
volume of the crater is the most significant one to evaluate the
effectiveness of pounding. This paper presents a photogrammetry
method which can carry out the volume measurement of the crater
and the surrounding ground heave with better accuracy. The
following conclusions are made based on the results of a field trial
test on a hydraulically filled reclaimed site.
1. Before the field test, an accuracy of 98.5% on volume
measurement of a calibration box with known volume had been
obtained using the photogrammetry method and Pix4Dmapper
software in the laboratory.
Figure 26. Comparison between photogrammetry and field 2. By using the smartphone or digital camera and the
measurement of DC crater volume photogrammetry technology, the point cloud data of the ground
surface can be established. Then, the point cloud data were
5. DISCUSSION ON MEASUREMENT RESULTS further calibrated with the ground control points placed around
the crater. The change of crater volume before and after each
It can be found from the RIC craters that for the first and second pounding can be determined using the calibrated point cloud
stages poundings, the crater volume determined from the traditional data with reasonable accuracy.
measuring method yields a larger volume for Set 1 and a smaller 3. By assuming a conical shape for the RIC crater, the crater
volume for Set 2 compared to those measured from Pix4Dmapper. volume calculated from Eq. 4 for the traditional measuring
Apart from the overestimation of top diameter by traditional method tends to underestimate the crater volume by about 20%
measuring method, one other possible reason for this phenomenon is compared to that measured with the photogrammetry method.
that the traditional measuring method used the ruler to measure the However, the conical shape may not be very representative to
crater depth (h). The contractor tended to push the ruler with force the real shape of the crater formed by RIC pounding due to the
into the bottom of the crater. So the measured crater depth (h) was disturbance caused by lifting up the hammer from the crater.
larger than that measured by photogrammetry method which only 4. DC crater volume measured by the photogrammetry method is
measured the surficial shape of the crater. Thus, the volume in good agreement with that calculated from Eq. 4 (for conical
calculated with either Eq. 4 or Eq. 5 showed a larger volume using shape crater). Although the overestimation by Eq. 4 is large in
the traditional method. After one set of RIC pounding, the crater the initial impacts, the difference is reduced to about 20% as
was backfilled with soil before performing next set of pounding. the number of impacts increased and the crater shape became
Since the ground became denser as the pounding increased, it closer to conical. In comparison, if Eq. 5 (truncated cone) is
became more difficult to push the ruler into the bottom of the crater. used to calculate the crater volume, more than 65%
As a result, the measured depth of crater was closer to the overestimation than that measured with photogrammetry
photogeommetry method, so was the crater volume approximated by method may result. However, it should be noted that the craters
Eq. 4 and 5. Therefore, the crater volume measured with traditional in this study were measured after the hammer was lifted up
method reduced and became less than that obtained from the from the crater and the side of crater might fall off to the
photogrammetry method (Figure 23). At the third stage, the above bottom and changed the apparent shape of the crater from
mentioned effect of over-estimating crater depth disappeared. The truncated conical to more or less conical (Figure 25). If the
volume (Vc1) of R3 crater became less than Vc,p for both set 1 and crater were measured with the hammer still in the crater, the
set 2. However, the assumption made on the shape of the crater may shape of the crater might more likely be in truncated conical
be the other source of volume underestimation. In other word, shape. Then the overestimation in volume by Eq. 5 may not be
assuming a conical shape for the crater and using the traditional as much as those shown in Table 5 and Figure 26.
177
7. ACKNOWLEDGMENTS
Li, L., Zhang, X., Chen, G. and Lytton, R. (2016). “Measuring
The Authors wish to thank the I-San construction company for
Unsaturated Soil Deformations during Triaxial Testing Using
carrying out the DC and RIC operation for the trial test of this crater
a Photogrammetry-Based Method”, Canadian Geotechnical
volume measurement program. The financial support from the
Journal, NRC Research Press, 53, pp. 476.
Ministry of Science and Technology (MOST) of Taiwan
Pix4D SA. (2017). Pix4Dmapper 4.1 User Manual, Switzerland, pp.
Government and Li-Jia Engineering Company are also greatly
11-46.
appreciated.
Simpson, L. A., Jang, S. T., Ronan, C. E. and Splitter, L. M. (2008).
“Liquefaction Potential Mitigation using Rapid Impact
8. REFERENCES Compaction”, Geotechnical Earthquake Engineering and Soil
Dynamics IV, ASCE, pp. 5.
Chow, Y. K., Yong, D. M., Yong, K. Y. and Lee, S. L. (1992). Tang, I. E. (2016). “Effect of Dynamic Replacement on the
“Dynamic Compaction Analysis”, Journal of Geotechnical Engineering Properties of Coal Ash Pond”, Master’s Thesis,
Engineering, ASCE, 118(8), pp. 1141. NTUST, Taipei, Taiwan
178
Potentiality of Boehmeria Nivea as Alternative Material in the Production of

Geotextile
Oliver C. Celis1, Catalino Mendoza2, Ernesto Q. Villarica1, Ann Krischel Hipolito1, Alfred C. Santos1, Nichols P. Manarang1,
Mary Jane Lusung1, Christian Ivan L. Magistrado1, Ederick T. Songahid1, Jefferson V. Cuellar1, John Mark G. Evangelista1,
Ralph Allen M. Cotelo1
1
Civil Engineering Department, College of Engineering
2
Natural Science Department, College of Arts and Science
Our Lady of Fatima University, Philippines, 2000
Email: revilo.silec@gmail.com1, catfish3789@yahoo.com2, ernestovillarica08@gmail.com3, annshipolito@gmail.com4,
zionengr@gmail.com5, jeffersoncuellar143@gmail.com6
ABSTRACT: The study focused on the potentiality of Boehmeria Nivea as alternative material in the production of geotextile. The
researchers used Ramie's bark in the production of geotextiles and series of tests were performed at the Philippine Textile Research Institute
(PTRI) of the Department of Science and Technology (DOST) located at Taguig, Metro Manila, to compare ramie geotextiles and
commercially available coconet geotextiles. Each test specimen was tested for their nominal thickness, mass per unit area, and tensile
strength. The results were analyzed and gave presentable results. After analyzing the results, the researchers reached at acceptable findings.
All the test conducted of ramie geotextile did meet the minimum tensile strength of the commercially available coconet geotextile for 400,
700, and 900 and considerably met the polyfelt specification filtration 58 for non - woven geotextile. Therefore, ramie fiber can be used as a
raw material in the production of a high strength and sustainable geotextile.
Keywords: Geotextile, Boehmeria Nivea, Ramie, polyfelt.
1. INTRODUCTION jute, coir, and wood shavings. Ramie (Boehmeria Nivea), a plant
that is vastly found in the Philippines especially in Mindanao and
Geosynthetics have been characterized by the American Society some parts of Batan and Babuyan Islands. It has a great potential
for Testing and Materials (ASTM) committee D35 as “a planar as an alternative material for geotextile production due to the
product manufactured from polymeric material used with soil, high tensile capacity of its fiber.
rock, earth, or other geotechnical engineering related material as Ramie are subtropical bast fibers, which are obtained from
an integral part of a human-made project, structure, or system” their plants five to six times a year. The fibers have silky luster
(https://www.slideshare.net/pparida/geosynthetics). They are and have white appearance even in the unbleached condition
used for purposes of separation, reinforcement, drainage, and (http://textilelearner.blogspot.com/2012/12/ selection-of-fiber-
filtration. According to Dr. Robert M. Koerner, “there are eight for-geotextiles.html). They consist of pure cellulose fiber which
types of geosynthetics: geogrids, geotextiles, geonets, possess highest tenacity among all plant fibers”. Thus, ramie can
geomembranes, geosynthetic clay liners, geopipe, geofoam, and be considered a great source of natural fiber.
geocomposites.”(http://www.acegeosyntheticsecopark.com/geosy Textiles are primarily and traditionally used for the design
nthetics#WHAT%20IS%20GEOSYNTHETIC?). Among these and production of garments but recently natural fibers and
geosynthetic products, the commonly used product is geotextiles. textiles were used as an alternative to synthetic fiber to promote
Geotextiles have been defined by the American Society of environmental awareness and conservation for the benefit of the
Agricultural and Biological Engineers (ASAE) as, “a fabric or generations ahead of us. Knowledge with regards to the
synthetic material placed between the soil and a pipe, gabion, or utilization of natural-fiber geotextiles should be further
retaining wall to enhance water movement and retard soil encouraged and recognized.
movement, and as a blanket to add reinforcement and The study aimed to determine the potentiality of Ramie
separation."(http://www.apparelsearch.com/education/research/n (Boehmeria Nivea) as an alternative raw material in geotextile
onwoven/2001_kermit_duckett/education_research_nonwoven_g production.
eotextiles.htm). Geotextiles are commonly utilized in
geotechnical engineering to reinforce soil and to construct strong
2. METHODS AND MATERIALS
bases for roads. Geotextiles are the most popular type of
geosynthetic material due to their versatility and affordability 2.1 Preparation of Sample
compared to other types of geosynthetics.
In civil engineering applications, the geotextiles used are The natural fiber used in this study came from Boehmeria Nivea
generally polymeric material. They offer longer life span and are (Ramie) a wild plant generally abundant in Barangay Manila de
not subject to biodegradability. However, synthetic geotextiles Bugabos, Agusan del Norte, Mindanao and other parts of the
are not eco-friendly and could possibly create environmental Philippines. Initially, the Ramie’s bark were taken from the stem
problems in the long run. Also, the raw materials used in of the plant. The collected stems were then cleaned and prepared
synthetic geotextiles are not readily accessible. Thus, prices of for the succeeding processes.
raw materials needed for production do not reduce resulting into The bark were removed with by means of peeling using a
expensive synthetic geotextiles. Due to the problems in relation knife, precautions were done to avoid damages in the fibers.
to the use of synthetic geotextiles and with the awareness and Cellulose layer of the bark were then removed and cleaned in
efforts to provide sustainable development, the use of preparation for drying the fiber.
biodegradable natural geotextiles has emerged. Natural fiber- Ramie’s fiber were then sun-dried to remove excess water.
based geotextiles are widely accepted because they are eco- Oven-drying can also be performed if available to thoroughly
friendly, renewable, economically viable, abundantly available, remove excess moisture of the extracted fiber.
very cost-effective and cheaper in comparison to their synthetic Figure 1 shows the process flow of ramie geotextile
counterparts. production.
Natural fiber-based geotextiles are commonly in the form of
179
are the weft or filling. The strategy in which these strings are
interwoven influences the quality of the textile.
Collection of
Ramie’s bark 2.2 Equipment
Several materials and equipment were used in the preparation of
sample specimen needed for testing the properties of the natural
fiber geotextile.
The J.A. king pneumatic sample cutter SASD-692 was used
Extraction of Fiber to cut eight (8) circular test specimens that were used for the test
From Ramie’s bark for mass per unit area and nominal thickness. (Laroza, J.P et al.
2014).
Eight (8) test specimens were tested using the Zwick/Roell
tensile strength tester Z050 (CRE) using a 50 kN full safe load
with a wide-width strip method in order to determine the average
tensile strength of sample.
Drying of Ramie’s bark The SDL digital thickness gauge M034A with 25 cm2
Fiber pressure foot area subjected to 2 kPa was used to find the
nominal thickness of the eight (8) circular test specimens by the
J.A. king pneumatic sample cutter.
2.3 Experimental Procedure

Spinning of Fiber The experimental procedure was conducted by Engr. Jun Cometa
Into Rope at Philippine Textile Research Institute (PTRI) under the
Department of science and Technology (DOST) Philippines. One
(1) test samples of Ramie’s bark geotextile 2000 mm x 1000 mm.
From 2000 mm x 1000 mm ramie geotextile, eight (8) circular
specimen is required for the test for mass per unit area and
nominal thickness test using the J.A king pneumatic sample
Weaving of Rope to desired
cutter SASD – 692 and SDL digital thickness gauge M034A with
Geotextile
25 cm2 pressure foot area subjected to 2 kPa pressure
respectively. Eight (8) specimens were subjected to tensile
strength test using the zwick/roell tensile strength tester Z050
(CRE) with 50 kN full scale load
Nominal thickness is one of the considerations in measuring
Natural Fiber Geotextile from how good and reliable one fiber to be a geotextile material. It
Ramie’s bark provides the results needed for the variation of the tensile
strength test. It is also one of the basic properties used to control
the quality of geosynthetics.
The nominal thickness is determined by placing a sample of
Figure 1 Process flow of ramie geotextile production the geotextile on a plane reference plate and applying a pressure
of 2 kN/m2 through a circular pressure plate with a cross-
The sample specimen were made using a 8 ft x 4 ft wooden sectional area of 2500 mm2. A vernier gauge measures the
frame for weaving the geotextile into desired dimension, 3 wood distance between the reference plate and pressure plate. The test
nails, hammer, scale, and tape. is useful for quality control and classification of geotextiles.
The dried extracted fibers were span into yarns. The This test method is used to determine if the geotextile
procedure was done by means of wheel spinning, manually by material meets specifications for mass per unit area. This test
hand or by mechanical spinning. Based on research, the latter method can be used for quality control to determine specimen
method can produce a good quality of yarn. A good quality of conformance to standard specifications. This measurement
yarn can be identified be means of physical and mechanical allows for a simple control of the delivered material by a
features, e.g. thickness, appearance, colour, proper twist, comparison of the mass per unit area of the delivered material
strength, fineness, texture, etc. In this study, hand spinning and the specified mass per unit area.
method was used by the researchers for producing of yarn. The Furthermore, the mass per unit area of a geotextile is
ramie’s bark fiber was tied into the wall and spanned the fiber determined by weighing test specimens of known dimensions,
manually by hand to produce an estimated 400 meters of fiber cut from various locations over the full width of the laboratory
rope needed to produce a 2000 mm x 1000 mm ramie’s sample. The calculated values are then averaged to obtain the
geotextile samples. mean mass per unit area of the laboratory sample.
After spinning and producing an estimated of 400 meters Tensile strength is also one of the considerations needed to
rope ramie’s fiber, the next process was to weaved the fibers to measure how good and reliable one fiber to be a geotextile
form the by-product which is the ramie’s geotextile. Weaving the material. It also measures the greatest longitudinal stress
fiber was done in normal process like any other textile. It is substances (geotextile) can bear without tearing apart.
finished by intersecting the longitudinal strings, the twist that was In this study, width wide tensile strength test is used. A
tossed crosswise over with the horizontal strings. This is a specimen of the geotextile, at least 200 mm wide, is clamped
common method of textile production in which two particular within the compressive jaws of a tensile testing machine which is
arrangements of yarns or strings were intertwined at right points capable of applying the load at a constant rate of strain. During
to frame a textile and fabric. Alternate techniques are composed loading, a load-strain curve is plotted and, from this, the
of weaving, bind making, felting, and meshing or plaiting. The maximum load, breaking load and the secant modulus at any
longitudinal strings are known as the twist and the parallel strings specified strain may be determined.
180
The tensile strength of geotextiles and related materials is a 3.1.3 Tensile Strength of Ramie Geotextile (Machine
very important property as virtually all applications rely on it Direction and Cross-Machine Direction)
either as the primary or secondary function. This test is useful for
quality control and can also be used for design purposes Figure 3 presents the graphical representation of the Tensile
Table 1. Shows the number of specimens required for every Strength Machine Direction (MD) of Ramie Geotextile based on
test. the test results.
Table 1. Number of Test Specimens
Geotextile Nominal Tensile strength Number of
sample thickness test test
test/Mass per specimens
unit area MD CMD
Ramie 8 4 8
It can be seen from the table that the total number of

specimens for testing was eight (8). This specimens were divided
according to the test performed namely the nominal thickness
test, mass per unit area and tensile strength test.
3. RESULTS AND DISCUSSION Figure 3 Tensile Strength, MD of Ramie Geotextile (kN/m)

This section presents the data collected from the test results. The
Figure 3 presents a graphical comparison of the average
data were used to create tabulations, graphical diagrams, analysis
tensile strength of Ramie Geotextile which is about 21.8 KN/m
and interpretation of data gathered to determine the viability of
(blue bar) after 4 trials were made.
ramie as alternative geotextile for slope protection.
Figure 4 presents the graphical representation of the tensile
strength CMD of ramie geotextile based on the test results.
3.1 Properties of Ramie Geotextile
3.1.1 Nominal Thickness
Figure 2 presents the graphical representation of the nominal

thickness of Ramie Geotextile based on the test results.
Figure 4 Tensile Strength, CMD of Ramie Geotextile (kN/m)
Figure 4 presents the comparison of the average tensile

strength of Ramie Geotextile Cross Machine Direction which is
about 24.4 KN/m (blue bar) after four (4) trials were made.
Figure 2 Nominal Thickness of Ramie Geotextile 3.2 Comparison of Properties of Ramie Geotextile and
Coconet Based on Test Results.
The above figure presents the average nominal thickness of
ramie geotextile which is about 13.39 mm (blue bar) and 3.2.1 Nominal Thickness of Ramie Geotextile and Coconet
comparing to no. of trials (8 trials) done in test specimens Geotextile.
(orange bar). Based on results, Ramie Geotextile has larger cross For verification of the strength and potentiality of Ramie
sectional area compared to the commercially available Coconet Geotextile, comparisons were made against commercially
Geotextile. available geotextile Coconet. Figure 5 presents the graphical
comparison between the average nominal thickness of ramie
3.1.2. Mass Per Unit Area geotextile and coconet geotextile.
Table 2 presents the mass per unit area of Ramie Geotextile.
Formula:
4
mx 10 (1)
M 
A
Based from the data drawn, Table 3 shows the average mass
per unit area of Ramie Geotextile was 1684.73 g/m2 after 8 trials
were made.
Figure 5 Nominal Thickness Test Result
181
3.3 Comparison of Average Tensile Strength of Ramie

The above figure shows that the average nominal thickness of Geotextile to Polyfelt Specifications for Non–Woven
Ramie Geotextile is larger than the Coconet Geotextile which is Geotextiles.
about 13.39 mm and 10.00 mm respectively.
The Ramie Geotextile filtation capacity was also compared with
the Polyfet Specification for non - woven geotextile.
3.2.2 Mass Per Unit Area of Ramie Geotextile and
Figure 8 presents the average tensile strength of ramie
Coconet Geotextile.
geotextile as compared to Polyfelt Specification for Non –
The mass per unit area of Ramie Geotextile and Coconet Woven Geotextile. Ramie Geotextile passed the specification for
Geotextile were also compared. The value for Ramie Geotextile Polyfelt F58 especially in CMD, considering that F58 is the
was affected since it has a larger cross sectional area than the rearmost value of Polyfelt Specification for Non – Woven
Coconet. The value signifies that the geotextile can cover more Geotextile with 24.0 KN/m tensile strength. Based on Figure 8, it
area and hold more soil in place than the Coconet Geotextile. can be concluded that Ramie Geotextile is classified as F-58
based on the Polyfelt Specification for Non-Woven Geotextile
CMD.
Figure 8 Comparison of the Average Tensile Strength of Ramie

Figure 6 Mass Per Unit Area Test Result Geotextile
Figure 6 shows the comparisons of the average mass per unit 4. CONCLUSIONS
area of Ramie Geotextile and Coconet Geotextile. Based on the
results, the average mass per unit area of Ramie Geotextile Based on the results of experiments and investigations, the
exceeded the mass per unit area of three types of Coconet researchers arrived at the following conclusions.
Geotextiles with a value of 1684.73 g/m2 against Coconet 400, The average nominal thickness of geotextile is 13.3 mm and
700 and 900 with a mass per unit area of 1284.73, 984.73 and the average tensile strength of ramie geotextile MD, CMD was
784.73 respectively. 21.8 and 24.4 kN/m respectively. This outcome will be of help in
calculating the cost appropriation of the finished product.
3.2.3 Tensile Strength of Ramie Geotextile and Coconet Based on test results, the average tensile strength of ramie
Geotextile. geotextileis higher than average tensile of coconet geotextile. In
addition, Ramie Geotextile is more capable of resisting greater
Tensile Strength Test were conducted to evaluate the capacity of tensile force for a thinner sample than that of coconet geotextile.
the Ramie Geotextile and compared it with the tensile strength of Lastly, Ramie Geotextile meet the category of Filtration 58 of
the commercially available Coconet Geotextile. Polyfelt Specification for Non-woven Geotextile, thus can be
Figure 7 presents a comparison between the average tensile concluded that it can be utilized in different geotechnical
strength of Ramie Geotextile against Coconet 400, 700, 900 engineering application such as slope protection.
Geotextiles (MD and CMD). Based on the results, the average
tensile strength of Ramie Geotextile for both Machine Direction 5. REFERENCES
and Cross Machine Direction was higher than the tensile strength
of Coconet 400, 700, and 900 Geotextiles. The average tensile Department of Public Works and Highway Standard
strengths of Ramie Geotextile MD and CMD were 21.8 and 24.4 Specification for Highways and Bridge and
kN/m respectively. Airpots 2013, Item 622 Coconet Bio-Engineering
Solution, p. 514.
Huang, H. and Gao, X., Geotextiles, Viewed, January 2018,
<http://www.apparelsearch.com/education/research/non
woven/2001_kermit_duckett/education_research_nonw
oven_geotextiles.htm>
Kentucky Transportation Cabinet. (n.d.). Slope Protection.
Retrieved June 28, 55552015, from 5555transportation.
ky.gov/EnvironmentAnalysis/Environment%20Resourc
es/3-5555Slope%20Protection.pdf
Khan, A. J. (2010). Quality Control of Jute Geotextiles &
Development of Testing 5555Facilities. Bangladesh:
Bangladesh Jute Research Institute (BJRI).
Koerner, R. M., What is Geosynthetics?, Viewd February 2018,
Figure 7 Tensile Strength Test Results <http://www.acegeosyntheticsecopark.com/geosyntheti
cs#WHAT%20IS%20GEOSYNTHETIC?>.
182
Laroza, J.P. et. Al., Viability of Banana Sheath as Natural Fiber Ministry for the Environment. (2009, August 24). Soil intactness
Geotextile, Batangas State University, 2014. of erosion-prone 5555land. Retrieved June 28, 2015,
Mahuya Ghosh, (2009). SUITABILITY OF NATURAL FIBRES from new Zealand. Govt.nz: 5555www.mfe.goct.nz/
IN GEOTEXTILE 5555APPLICATIONS. Retrieved environmental-reporting/land/erosion-risk/
June 28, 2015 , from Scientist, Geotech Cell, Parida, P. (2009). A Seminar on Geosynthetics. Viewed January
5555Indian Jute Industries’ Research Association, 2018,
Kolkata–700088, https://www.slideshare.net/pparida/geosynthetics.
India:5555http://gndec.ac.in/~igs/ldh/conf/2009/articles Selection of Fiber for Geotextiles: Fibers Used in Geotextiles,
/T07_40.pdf Viewed January 2018,
http://textilelearner.blogspot.com
/2012/12/selection-of-fiber-for-geotextiles.html
183
A Multidisciplinary Ground Model Approach to Geotechnical and Geohazard Site

Appraisal for Large Infrastructure Developments
D. Rushton
East Point Geo
E-mail: davidrushton@eastpointgeo.com
ABSTRACT: A major geotechnical challenge for large and mega infrastructure developments is to economically appraise the geotechnical
conditions of the development area early enough for concept design, with sufficient resolution for detailed design and with sufficient spatial
coverage to provide flexibility for layout changes. In active geological settings, characterisation and mitigation of potential geohazards
presents a significant additional challenge. This paper describes the use of a multidisciplinary ground model as a tool to support both
geotechnical site appraisal and geohazard appraisal. The combined use of a geographical information system with a ground model is also
described, to manage the spatially extensive data associated with large infrastructure developments and to perform geotechnical and
geohazard spatial analysis over large areas. Examples of successful application of this approach from major offshore developments are
described and the importance of a competent and experienced geoteam of discipline specialists is highlighted.
Keywords: Ground model, geographic information system, geohazard, spatial analysis
1. INTRODUCTION 2. MULTIDISCIPLINARY GROUND MODEL

Large infrastructure developments can cover several hundreds of 2.1 Data Collation
kilometres in the case of transport routes, pipelines and cable
corridors, or hundreds to thousands of hectares for major building 2.1.1 Desk Study
complexes such as power stations, ports and subsea oil and gas A desk study is an essential first phase to the development of a
developments. multidisciplinary ground model and involves the identification and
A major geotechnical challenge for such large or mega analysis of all publicly-available data of relevance to the foundation
infrastructure developments is the practical and economic appraisal zone of the planned infrastructure. The desk-study should focus on
of the geotechnical variability of the development area sufficient for building a first iteration of the ground model, which is likely to be
design. In active geological settings, characterisation and mitigation lower resolution than the requirements of the final foundation design
of potential geohazards presents a significant additional challenge. but is an initial framework to be iteratively refined. An initial desk
The combination of a detailed understanding of past and present study-based ground model would typically divide the study area into
geological processes with traditional ground investigation broad terrains of similar ground conditions where terrain boundaries
approaches allows a full three- or four-dimensional understanding of are usually defined by significant geological features such as major
the geotechnical conditions of a site to be efficiently characterised in slope breaks, surface fault expressions, textural boundaries, etc.
a multidisciplinary ground model. In addition to characterising the present-day conditions, which
Rigorous data management using a geographical information represent a ‘static’ point in time, the desk study should also consider
system (GIS) allows all relevant data to be hosted and manipulated the geomorphological process history of the site to characterise the
in an intuitive spatial domain that highlights the geographical dynamics of the site and build an understanding of the geotechnical
interrelations between different datasets and between geotechnical conditions expected to remain for the life of the planned
datasets and the planned infrastructure. GIS-based spatial analysis infrastructure, those which may be associated with past processes
techniques can be employed to rapidly perform geotechnical that are no longer active and those which may arise due to new
analysis or geohazard assessment, such as foundation suitability processes not previously experienced at the site.
mapping, slope stability screening or seabed sediment transport A multidisciplinary approach is recommended, even at the desk
analysis, over very large areas. study phase. Analysis of desk study data should involve geologists
The development and of a multidisciplinary ground model and and geomorphologists, to interpret data from multiple sources and at
its use for site characterisation, gap analysis, visualisation and variable resolution, into a common format for integration in a
communication are described in this paper. The application of a ground model. Extrapolation between datasets may be required,
GIS-based ground model to geotechnical site appraisal is also which is best-performed by discipline specialists. Geotechnical
discussed and is of specific relevance to large infrastructure engineers and geohazard analysts should also be involved at the
developments where efficiencies are required over traditional initial desk study phase to focus the ground model on its end
methods due to the scale of the site. The use of a ground model for applications to inform foundation design and geohazard assessment.
geohazard appraisal is presented as a spatially consistent and Key datasets of relevance at the desk study phase will include
resolute way to identify and characterise geohazards, understand terrain elevation data, regional geological maps (e.g. solid geology,
their potential interactions with infrastructure, and mitigate surficial sediments, geological structure), aerial / satellite imagery
accordingly. In this way, geohazards are managed in a risk-based for onshore developments, regional seismic lines (more readily
framework consistent with other routine project risks. Examples of available in offshore settings), existing boreholes and other intrusive
successful application of the ground model approach are presented data, wave and tidal current data for offshore developments, etc.
from major offshore development projects and the importance of a
competent and experienced geoteam of discipline specialists is 2.1.2 Data Acquisition
highlighted.
The development and requirements of a ground model is well A desk study ground model will form the basis of an initial gap
described by other authors (e.g. Fookes, 1997, Knill, 2003, Evans, analysis to define data acquisition requirements. To maintain a
2010). This paper highlights the diverse use and application of a ground model approach to characterising the site, data acquisition
ground model for large infrastructure projects from both a should cover locations of currently planned infrastructure and the
geotechnical and geohazard assessment perspective. surrounding area, to ensure flexibility for revisions to the
infrastructure layout and to provide a wider geological and
geotechnical context for geohazard assessment. The economic
argument to focus acquisition on the current infrastructure layout
184
may provide short-term savings, but the flexibility afforded by a level of data manipulation and combined data and spatial query is
wider site understanding can save significant costs associated with not possible unless data are integrated within a GIS-based ground
late-stage remobilisation of data acquisition equipment if more data model.
are required. Non-spatial data such as reports, photographs, diagrams, etc. can
Newly acquired data are typically used to update the ground be made accessible via hyperlinks from a relevant geographical
model to reflect the improved understanding of the site. Therefore, location within the GIS ground model to provide single-interface
an integrated multidisciplinary approach should be applied to data access to all information.
acquisition planning, monitoring and results interpretation to
optimise the resulting ground model iteration. For example, this 2.3 Visualisation and Communication
includes the specification of combined sample acquisition for
geotechnical testing, geochronological dating and sedimentological A multidisciplinary ground model is an effective way to visualise
logging, either from within a single composite borehole or via the the three-dimensional spatial relationships between datasets and to
specification of clustered boreholes to target multiple requirements understand the link between the terrain, subsurface, geotechnical
from the same location. Where combined geophysical and conditions and geohazard processes. Communication of this key site
geotechnical data acquisition is planned, boreholes and in situ tests understanding is possible via both two-dimensional and three-
should be sited to lie on geophysical acquisition lines for optimal dimensional GIS interfaces. Typically, specific features or processes
correlation and interpolation potential. will also be illustrated using annotated conceptual block models
As well as the potential to overlap geotechnical and geohazard which make use of a graphical approach to clearly communicate
data acquisition targets, geohazard-specific data acquisition may observations and interpretations (Figure 1).
also require a focus on areas distal to the development area. For
example, a pipeline route may be planned along shallow gradient
valley floor, surrounded by steeper slopes of the valley sides.
Without proper characterisation of the valley sides, which may
include data acquisition, the geohazard risk associated with slope
instability cannot be fully assessed.
Whilst careful data acquisition planning helps avoid costly late-
stage remobilisation of data acquisition equipment, it can be
beneficial to adopt a phased approach whereby an initial limited
quantity of ‘reconnaissance’ data are acquired and used to make an
interim ground model update, from which a further gap analysis and
detailed acquisition scope are created.
A further alternative is to ensure the results of the data
acquisition are reviewed by a multidisciplinary team, or ‘geoteam’
(described later in the paper), as the acquisition progresses. This
approach avoids multiple iterations of data acquisition but provides
flexibility for changes to the data acquisition scope in almost real-
time. New data are compared with predictions from the existing Figure 1 Example of a conceptual block model which uses a
ground model; where conditions are consistently as predicted the graphical approach to clearly communicate key observations and
density of data acquisition locations can be reduced and conversely interpretations
where conditions are not as expected additional data can be
specified. Specification of additional data locations during data 3. GEOTECHNICAL SITE APPRAISAL
acquisition can be expedited by pre-defining option locations before 3.1 Soil Model
mobilisation.
A predictive soil model focused on infrastructure engineering
2.2 Geographical Information System Framework requirements is the primary interpretive output from a
multidisciplinary ground model for geotechnical engineering
A successful multidisciplinary ground model requires integration of application. The predictive soil model comprises soil units and soil
data from each discipline in a common framework. All disciplines provinces. Soil units are used to divide the three-dimensional zone
of ‘geo-data’ for infrastructure developments have a position in of interest into volumes of similar soil properties and geotechnical
geographical space and therefore a GIS platform provides an conditions, with characteristic geotechnical parameters defined for
intuitive spatial framework in which to integrate all data of each unit. Soil provinces are surface zonations which group areas
relevance to a ground model. with a similar sequence and thickness of soil units between ground
A GIS has a primarily map-based interface built from an ordered level and the depth of interest below ground level. Soil unit
set of layers which clearly demonstrate the spatial interrelations boundaries are interpolated between boreholes and, where
between datasets and the planned infrastructure. Continuous surface geophysical data are available and if acoustically significant, can be
data are represented as grids and triangular irregular networks mapped as continuous surfaces across a development area (Figure 2).
(TINs), and discrete features as points, lines and polygons (typically This simple but versatile approach allows the geotechnical
known as vector layers). All vector layers have associated attributes conditions to be predicted anywhere in a development area with as
which are used to capture tabular detail about each feature and much accuracy as the available data will allow and with
vector layers are typically symbolised to highlight a specific geologically-informed interpolation where data are absent.
attribute.
Most GIS software has a powerful set of in-built tools and can 3.2 Geotechnical Application
typically be customised to develop additional tools. In addition to
The predictive soil model can be used directly to provide soil
basic navigation tools, some of the most useful tools for GIS-based
parameters for foundation design associated with a fixed
ground model manipulation allow users to make composite spatial
infrastructure layout and, if sufficiently detailed, will include lower
and attribute queries. A well-structured GIS ground model can be
bound parameters for capacity calculations, upper bound parameters
used to make queries such as “select all locations within 2000m of a
for installation calculations, etc. However, where the infrastructure
planned structure where the soil friction angle is greater than 30° at
layout is not defined, spatial analysis informed by the GIS-based
less than 10 m below ground level” or “select all samples on slopes
ground model can help refine the layout for optimal geotechnical
greater than 10° with a unit weight greater than 21kN/m3”. This
placement.
185
areas of hard ground near seafloor which would cause significant

problems for the shallow foundations of the planned infrastructure.
The hard ground typically caused a high amplitude acoustic
response and always caused a high cone resistance CPT response.
The GIS-based ground model spatial analysis was combined with a
novel use of Receiver Operating Characteristics analysis (Fawcett,
2005) to quantify the confidence and the chance of false positives
associated with an acoustic amplitude above which hard grounds
were expected.
4. GEOHAZARD APPRAISAL
4.1 Geohazard Model
A geohazard model is the primary interpretive output from a
multidisciplinary ground model for geohazard appraisal and
assessment. The format of the geohazard model will vary according
to the application, but typically includes a geohazard database and
register.
The geohazard database records the number of observations of
each geohazard occurrence in the ground model datasets, and
records key metrics associated with each event. For example, a
database of past offshore landslides would be built from
observations on hillshaded seafloor elevation data and sub-bottom
profiler or seismic data, where available, to characterise the number
of events visible, the depth below ground level (and hence inferred
age) of each event, and key metrics such as the length, width,
thickness, orientation, etc. Where possible, regional
seismostratigraphy and geochronological dating will be used to
Figure 2 Example soil model for a wind farm development refine estimates of the age of events, with the aim of determining a
comprising soil units visible on the cross section and soil provinces rate of reoccurrence.
visible on the plan view map (Clare et al, 2012) The geohazard register documents all geohazards considered to
be credible at the site, summarises the properties of each geohazard
Rushton et al (2017) present examples of geospatial analysis, based on observations and metrics from the geohazard database,
which includes a method developed to derive GIS-based lateral pile captures initial estimates of the likelihood of each geohazard and, if
capacity and map the required monopile length sufficient to provide possible, makes some inference of the consequence of the each
capacity for an offshore wind turbine across an entire licence zone. geohazard impacting the planned infrastructure. This geohazard
The method involved development of a regular mesh of analysis register is an important step in documenting which geohazards
points and use of the GIS-based ground model to assign design soil require further assessment in a full quantitative risk assessment
parameters and lateral load-displacement (p-y) soil springs to each (QRA) and which geohazards are considered of sufficiently low
mesh node, for use in a one-dimensional finite element analysis of concern to the planned infrastructure to not require further
the monopile (Figure 3). An iterative procedure was performed to consideration.
determine the required monopile lengths: if the monopile was able
to withstand the applied lateral load and moment, and remain within 4.2 Quantitative Risk Assessment
the prescribed displacement and rotation tolerances, the pile length A QRA takes the largely qualitative geohazard register and
was reduced. As soon as either of the tolerances were exceeded, the advances the assessment of the main threats by further quantified
previous monopile length was taken as the required length. analysis, which should include consideration of all components
described in Table 1.
Table 1 Quantitative Risk Assessment Considerations
Component Description
P(event) Probability of a geohazard event
P(spatial) Spatial probability: infrastructure exposure to
a geohazard event and likelihood of it being
hit given an event occurs
P(hit) P(event) × P(spatial)
P(damage|hit) Probability of infrastructure damage in the
event of it being hit by a geohazard event.
This is assessed via vulnerability analyses
P(damage) P(hit) × P(damage|hit)
Risk P(damage) × Consequence, where
consequence may be in terms of health and
Figure 3 Graphical depiction of ground model sampling via a safety, financial, reputational or
regular mesh in GIS to provide inputs to an automated sequence of environmental and is typically assessed by the
monopile lateral capacity analyses infrastructure developer
In another offshore example for a major oil and gas company, a Probability of a geohazard event may be inferred from rates of
very shallow GIS ground model was built to integrate cone reoccurrence as described in the Geohazard Model section. In
penetrometer test (CPT) data with the reflected amplitude of the addition to reliance on such historical frequency estimates, it is
acoustic signal from a geophysical source. The aim was to locate typical to perform forward modelling to estimate the likely future
186
P(event) as a result of natural processes and possibly modified by 6. CONCLUSION

the planned infrastructure development. Rates of reoccurrence and
forward modelling results should be compared and reconciled as a Data acquisition, geotechnical site characterisation and appraisal,
sense check. and geohazard assessment have been routinely performed as part of
Again, a GIS-based ground model can be used as an input to infrastructure development projects for decades. However,
GIS spatial analysis to determine P(event) over a large development requirements to deliver large and complex infrastructure projects
area. Mackenzie et al (2009) describe spatial analysis of landslides efficiently, reliably and safely requires traditional approaches to
with the use of a ground model to provide the geotechnical soil evolve.
condition inputs. Dimmock et al (2012) describe a similar GIS- For the geotechnical aspect of an infrastructure development, a
based approach applied probabilistically and with pseudostatic multidisciplinary ground model is a versatile tool that can bring
earthquake loading to inform a QRA for a deep water subsea gas repeated time and cost efficiencies to a project. Infrastructure
development project in the Mediterranean Sea. Rushton et al (2015) developments in geohazardous settings can also benefit from the
describe further advances landslide spatial analysis via the understanding of the geological processes afforded by the integrated
incorporation of the shear band propagation mechanism, rather than ground model approach.
limit equilibrium. The example presented by Rushton et al (2015) is GIS is an intuitive and spatially resolute software platform with
for a new deep water oil facility in the Caspian Sea. Figure 4 which to build and manage a ground model. Its strength in
demonstrates graphically how geotechnical properties are extracted visualising and querying large volumes of data is of importance for
from the ground model as pixel surfaces which are then combined large infrastructure developments. The spatial analysis capability of
using spatial analysis to perform an infinite slope assessment for GIS, when coupled with a detailed ground model, can bring
each pixel and generate a result of factor of safety for each pixel. significant efficiencies to geotechnical and geohazard analysis.
This approach has been successfully employed for several large
offshore developments, delivered by a competent and experienced
geoteam of discipline specialists.
7. REFERENCES
Clare, M., Rushton, D. and Balthes, R. (2012) “A ground model-
based approach to efficient assessment and management of
risk for pile installation and behavior”. Proceedings of the
Hans Lorenz Symposium, Hamburg, Germany.
Dimmock, P., Mackenzie, B. and Mills, A. (2012) “Probabilistic
Slope Stability Analysis in the West Nile Delta, Offshore
Egypt”. Proceedings of the 7th International Offshore Site
Investigation and Geotechnics Conference: Integrated
Technologies – Present and Future, London, Volume 1, pp
535-542
Evans, T., G., (2010) “A Systematic Approach to Offshore
Engineering for Multiple-Project Developments in
Geohazardous Areas”. Keynote Paper, Proceedings of the 2nd
International Symposium of Frontiers in Offshore
Geotechnics, Perth.
Fawcett, T. (2006) “An introduction to ROC analysis”. Pattern
Figure 4 Graphical depiction of GIS-based slope stability Recognition Letters 27, pp 861-874.
assessment which uses geotechnical properties extracted from the Fookes, P., G. (1997) “Geology for engineers: the geological model,
ground model as pixel surfaces and generates a further pixel surface prediction and performance”. Quarterly Journal of
of factor of safety Engineering Geology and Hydrogeology, 30, pp 293-424.
Spatial probability is typically a function of the size of the
Hill, A., Prakash, A., Hampson, K., Gray, T., Rushton, D.,
geohazard event relative to the size of the infrastructure and
Mackenzie, B. and Puzrin, A. (2015) “Geotechnical
therefore is usually calculated within the same GIS framework as
Challenges in the Caspian Sea”. Proceedings of the Offshore
the ground model. The other QRA components are largely assessed
Technology Conference, OTC-25988-MS.
outside of the ground model in a risk workbook, but the interim
Knill, J., L., (2003) “Core values: the First Hans Cloos Lecture”.
result of P(hit) and the final results of P(damage) and Risk are most
Bulletin of Engineering Geology and the Environment, 62,
effectively communicated as mapped layers hosted in the GIS
pp 1-34.
ground model and visualised relative to the planned infrastructure
Mackenzie, B., Hooper, J. and Rushton, D. (2009) “Spatial analysis
layout (e.g. Hill et al, 2015),
of shallow slope instability incorporating an engineering
5. GEOTEAM geological ground model”. Proceedings of the 4th
International Symposium of Submarine Mass Movements and
The approach described for the development of a multidisciplinary Their Consequences, Austin, Texas.
ground model for application in geotechnical and geohazard site Rushton, D, Gray, T., Puzrin, A. and Hill, A. (2015) “GIS-Based
appraisal relies heavily on large and diverse datasets, powerful GIS Probabilistic Slope Stability Assessment Using Shear Band
software applications and cutting-edge analysis. However, the most Propagation”. Proceedings of the Offshore Technology
important component in the development of a multidisciplinary Conference, OTC-25871-MS.
ground model is a competent and experienced geoteam of discipline Rushton, D., Jones, L., Nott, T. and Rattley, M. (2017) “Geospatial
specialists able to work collaboratively and combine their expertise. Analysis for Geotechnical Design”. Proceedings of the 8th
The specific skillsets within a geoteam will depend on the International Offshore Site Investigation and Geotechnics
requirements of each individual project, but may comprise Conference: Smarter Solutions for Future Offshore
geologists, geomorphologists, geotechnical engineers, geophysicists, Developments, London, Volume 1, pp 231-238
risk analysts, sedimentologists and facility engineers.
187
Prediction of the Geological Condition for Pipe jacking Based on the Data Collected
in the Shafts
Keh-Jian Shou1, Fu -Kang Tang1
1
Department of Civil Engineering, National Chung-Hsing University, Taiwan
E-mail: kjshou@dragon.nchu.edu.tw
ABSTRACT: No -Dig method often encounters difficult geological problems, which makes the excavation in difficulty and thus affecting
the construction schedule. Central Taiwan area is majorly covered by gravel alluvium, in which the strata contains large gravels. Those large
size gravels often result in excessive wear and increase in construction cost. In this paper, we took samples from the launch shaft and arrival
shaft pipe-jacking and carried out sieve analyses. Then we transformed the size distribution into borehole data by the newly derived formula.
The obtained borehole data was applied to the T-PROGS geostatistic analysis. The purpose of the geostatistic analysis is to building a 3D
geological model from data from pipe-jacking shafts, then we can obtain the geological and size distribution along the pipe-jacking
alignment. In this study, the analysis was performed for different scenerios and the impact of material changes on the pipe jacking rate. The
results show that the change of gravel particles larger than 20 cm is highly correlated with pipe jacking rate. The results suggest that it is
feasible to apply the image sieve analysis in the geostatistcs analysis.
Keywords: pipe jacking, geostatistics, geological condition, T-PROGS, Markov chain model.
1. INTRODUCTION complex strata conditions are simulated based on transition

probability theoretical equations.
For a no-dig project, it is not easy to explore the geologic condition The study of Li, Shao, Jin, and Cui (2009) presented that the
around the construction area in advance, due to environmental modelling process should also consider the heterogeneity of hydro-
factors such as heavy traffic, residential area, and high density of geological structure. The spatial distribution pattern of hydro-
existing pipelines. This can lead to many unexpected situations geological parameters and the level of sophistication determine the
during construction, such as cutter head damage due to excessive accuracy of the simulation. For the sedimentary rock formation, the
wear caused by large size gravels and collapse of soft soil at the spatial distribution pattern of the rock is the most important control
cutter head of shield machine. Nevertheless, it is common to factor. Using a large amount of drilling data, the researchers
conduct geologic investigation in a test pit or in the launch shaft employed the Markov chain transition probability sequence with an
directly. indicator kriging model as the primary method for the analysis.
By using the geostatistical approach, we can deduce geologic Theodossiou and Latinopoulos (2009) used the T-PROGS in GMS
conditions and the parameters of the strata formation encountered to conduct a probability analysis on the distribution of soil in an
during the excavation. This way, pipe jacking preparations can be aquifer, and similarly used the input of existing drilling data into T-
more efficiently made in terms of machine parameters, staff PROGS to calculate a three-dimensional transition probability
scheduling, and intermediate shaft location, etc. This study applied a model. The results obtained were used for further modelling of the
geostatistical method by entering drilling data into the software T- aquifer.
PROGS (Transition Probability Geostatistical Software) to simulate
and obtain the three-dimensional geological model between the
3. METHODOLOGIES
launch and arrival shafts.
Since this study focuses on the pipe jacking in the gravel 3.1 Sampling and Sieve Analysis Testing
formations of Taichung area, part of the work site was rededicated
to finding information on distribution of particle size within the The laboratory test was based on the Taiwan specification
strata formation, which can help to re-adjust work parameters. Soil CNS10989 on "the method to obtain the laboratory testing sample
samples obtained from the site can be immediately analysed through from the bulk sample from the field", and the testing specification
sieve analysis. Then a conversion of particle size distribution can be CNS486 on “the method for sieve analysis of fine and coarse
input into the software for strata formation simulation. It is hoped aggregates". Sieve analysis can be appropriately applied to soil
that through the effective use of estimation and simulation, this particles of a size greater than 75 μm (# 200 meshes). For particles
method can better predict the strata formation encountered during a smaller than 75 μm, a hydrometer test must be used. The study was
pipe jacking, such that the operation of the construction can be more designed to use sieve analysis testing for particles of a size larger
efficiently and more economically. than 75 μm. The test aimed to enable soil samples to pass through a
series of sieves of various mesh sizes, to obtain the weight of
2. LITERATURE REVIEW particles within each particle-size category, and to draw the particle
size distribution curve.
Jones, Walker, and Carle (2002) used the transition probability
approach under the software MODFLOW during random
3.2 Conversion of Particle Size Distribution
simulation. Transition probability geostatistics is superior to the
traditional indicator kriging method in a number of ways, such as its Considering a three-dimensional penetration through a volume of
ability to use a simpler and more intuitive approach to enable a gravelly soil, we have to assume that the penetration may not go
better understanding of the relationships among various geological through the center of every single particle, but instead through at a
frameworks, and in its ability to simulate a number of geological point that is 1/8 of the projected area at the bottom. Looking down
correlations in more details. Weissmann (2005) used the geological from the z-axis, the centroid position was as shown in Fig. 1(a), and
model in the Kings River alluvial fan as an example to outline the side view was as in Fig. 1(b). Let R ' equal the cut-through
various geological features according to data from drilling, the
geophysical log, and rock core classification. The geological model length, R equal the radius of the spherical particle, and y equal the
was simulated by the transition probability method, in conjunction distance CO from the center of 1/8 of the sphere to the center of the
with the Markov chain. Seeboonruang (2006) studied the lower
sphere to obtain R. Here, we must first find y . First of all, by
reaches of the Chao Phraya River in central Thailand, where
looking from the direction of the z-axis, the sphere is divided by the
188
x and y axes into four sectors of equal area at the xy plane of the T-PROGS is composed of many sub-programs, including
projected area (z = 0). The equation for the sector area is: GAMEAS for calculation in bivariate statistics (transition
probability and indicator variogram), MCMOD for development of
 R 2  y 2  Markov chain model, TSIM for three-dimensional and cross-
Ac  (1) correlation conditional simulation, etc. The theoretical background
4 of T-PROG is briefly described as below.
Transition probability tjk (h) is defined (Carle and Fogg, 1996)
as follows:
tjk(h)=Pr{ k occurs at x+h | j occurs at x} (4)
where x is a specified location; h is the separation vector; and j and

k are independent geological units, rock formations, or marked
categories of materials.
In the classification of variables the Markov chain (Carle and

Fogg, 1997) provides a simple and powerful stochastic
mathematical model. In time series applications, the theoretical
Markov chain model assumes that the future depends on the present,
not the past. The assumed occurrence of a spatial Markov chain
depends entirely on the most recent data. The reason that the
Markov chain model has become an applied method in geostatistics
Figure 1 Top view (a) and side view (b) of a penetration through a is because it provides the most direct method of development that
sphere uses regional patterns to describe all the spatial cross-correlations.
The sector area Ac is integrated along the z-axis and divided by 4. RESULTS AND DISCUSSION
1/8 the volume of the sphere to obtain the distance from the center
of sphere to the three axes; which, after some calculation 4.1 Analysis of the Actual Cases
adjustments, we obtain y =3R/8, as shown in Fig. 1(b). Applying The analysis of the pipe jacking project with a configuration plane
the geometry, we obtain R' (distance EF from the starting point to diagram, as shown in Fig. 2. The diameter cross-section corresponds
the end point). Applying the Pythagorean Theorem, we obtain: to an actual drill machine head of 0.6 m as actual case 1
(extrapolated) and actual case 2 (interpolated). The sieve analysis
 
R 2  EF  y
2 2
(2) data of shaft Bi43 and shaft Bi46 were used to establish the 3D
geological model and estimate the probability of jacking through
large-size gravel particles. Then these probabilities of encountering
Substituting the known
y =3R/8 into Eq. (2) and obtaining: the large-size particles (greater than 20 cm) were used to explore
their correlation with the advancing rates of actual pipe jacking
work.
55
EF  R '  R 2  (3R )  R (3)
8 8
which is the penetrating length of 1/8 sphere. The total penetrating

length is the combination of the two 1/8 spheres (1/4 sphere). Thus,
it becomes 2R'1.854R. This shows that the penetrating length
formed by penetrating through the center of the 1/4 sphere is shorter
than the radius 2R of the sphere. Since the cut-through did not go
through the center of all the particles that were cut, but instead only
went through them at the center of their 1/8 sphere, the size of these
penetrated-through particles were underestimated. Figure 2 The alignment of the study cases
The particle size distribution obtained from a sieve analysis is
referred to as the raw particle size distribution. Since the penetration Since the location and depth of the on-site sampling points are
of drilling will cause the value for the original effective particle size different, the modelling position was separated into two primary
to be underestimated with the ratio of Ru=1.854R/2R=92.7%. active wells, with the depth of each active well equal to 11 m. The
Multiplying this ratio to the raw particle size distribution, we can two active wells were further divided at 5.5 m in depth as the
obtain the adjusted particle size distribution, which represents the boundaries for upper and lower intervals, for a total of four section
particle size distribution of inside the drill hole, i.e. the core log. blocks, as shown in Fig. 3. The particle size distribution of the
Blocks I, II, III, and IV are represented by Bi43_-5.5 m, Bi46_-6.5
3.3 T-PROGS Analysis m, Bi43_-9.0 m, and Bi46_-8.5 m, respectively. The sieve analysis
results were then converted to the adjusted particle size distribution,
Transition Probability Geostatistical Software (T-PROGS) is a as shown in Fig. 4. For a better practical application, the particle
geological statistical method based on transitional probability size were classified into three categories: (a) Particle size greater
(Carle, 1999). A combination of an indicator co-kriging model and a than 200 mm; (b) Particle size less than or equal to 200 mm but
3-D Markov Stratigraphic sequence model is created to simulate the greater than 20 mm; and (c) Particle size less than or equal to 20
spatial variability of the model. mm.
189
4.2 Comparison with the Advancing Rate
The pipe jacking project records for the sections Bi42 to Bi43 and
Bi43 to Bi46, including daily work record, the history of pipe
jacking, and the number of jacked pipe count, were used to obtain
the advancing rate, as shown in Fig. 5~6. The data were also
converted into advancing rates in hours per pipe jacked. Such that
we can analyse the relationship between the number of hours spent
and the change in the geological materials can be obtained.
However, the figures show that the advancing rate increases as the
end of the work nears.
Figure 3 The data blocks of the pipe jacking shafts.
(a)
(a)
(b)
(b)
Figure 5 Progress and advancing rate of pipe jacking section Bi43 to
Bi46
(c)
(d)
Figure 4 The gradations of the four data blocks, (a) Bi43 -5.5 m, (b)
Bi46 -6.5 m, (c) Bi43 -9 m, and (d) Bi46 -8.5 m (a)
190
(b)
Figure 6 Progress and advancing rate of pipe jacking section Bi42 to
Bi43 Figure 8 The results of three-dimensional geological modelling for
case II, section from Bi42 to Bi43.
The three-dimensional geological model for case I, i.e., section
from Bi43 to shaft Bi46, was illustrated in Fig. 7, with a regression 5. CONCLUSIONS
trend line of particle size change. Comparison of the results found
that the tendency of large size particle (greater than 20 cm) to
gradually decline (representing a decreasing number of grids for The alluvium in Taichung area is mainly gravel deposits, containing
particle size greater than 20 cm) is very consistent with the tendency a substantial amount of large sized gravel, which has a significant
of the advancing rate to increase. In contrast, grids for particle size impact on the success of any no-dig project. This study conducted
between 2 cm and 20 cm tend to rise at the same position an on-site sieve analysis of geomaterial launch and/or arrival shafts.
(representing an increase in finer materials). This finding is quite After obtaining sieve analysis data, a derived formula is applied to
consistent with fact that the reduction of large pieces of gravel in the adjust and convert the data into drilling core log. Then a GMS sub-
strata, leading to an increase in advancing speed. program T-PROGS is used to perform three-dimensional geological
Similarly, the material change for the section from Bi42 to Bi43 modelling to analyse the vertical and horizontal material changes.
in the three-dimensional geological model (Fig. 8) added to the The analysis results were compared with pipe jacking data of the
trend line obtained Fig. 5-6. Comparison of results showed that the study cases.
position of grids greater than 20 cm tended to increase gradually Comparison of the simulation results with advancing rates of
(represent an increasing number of grids greater than 20 cm), which actual cases showed that the trends in material change and
is very consistent with the tendency of the position of the drilling advancing rate have a high degree of correlation. When the number
rate to increase (representing a decrease in drill rate speed). of grids of size greater than 20 cm decreased, and those grids of less
However, the trend of grids for size between 2 cm and 20 cm is to than 20 cm but greater than 2 cm increased, the advancing rate also
decline at the same place (representing a decrease in finer increased. Conversely, when the number of grids of size greater than
materials), which is quite consistent with the increase in large size 20 cm increased, and those grids of less than 20 cm but greater than
gravel in the strata, leading to a decrease in advancing speed. 2 cm decreased, the advancing rate also decreased. They highly
correspond to the actual pipe jacking situation: when the cutter head
of a shield machine encounters large sized gravel rock, its speed will
slow down due to cutting wear.
From the diameter point of view, the larger the diameter cross
section, the greater the number of grids encountered of size greater
than 20 cm; the smaller the diameter cross section, the smaller the
number of grids encountered of size greater than 20 cm. Since the
parameters for the shafts of the two study cases are the same, the
trends in material change in the longitudinal section from the two
cases are consistent. The results show that the material change in
grids of size less than 2 cm in each case is scattered. The reason for
the inaccuracy might be due to the size of the analysis grid, i.e., the
unit grid size of 0.2 m by 0.2 m by 0.2 m far exceeds the particle
size of less than 2 cm.
Figure 7 The results of three-dimensional geological modelling for

case I, section from Bi43 to Bi46.
191
Li, C.Q., Shao, J.L., Jin, P., Cui, Y.L., 2009. Conditional simulation
of hydro-geological structure and its application in plain
6. REFERENCES
area: A case study of North China Plain. Geoscience,
Vol.23, No.1, pp.137-143.
Carle, S.F., Fogg, G.E., 1996. Transition probability-based indicator
geostatistics. Mathematical Geology, Vol. 28, Seeboonruang, U., 2006. An application of transition probability
No.4,pp.453~476. approach to geostatistical simulation: A case study in the
Lower Chao Phraya Basin, Thailand. Proceedings of the
Carle, S.F., Fogg, G.E., 1997. Modeling spatial variability with one 2nd IMT-GT Regional Conference on Mathematics,
and multidimensional continuous-lag Markov chains. Statistics and Applications Universiti Sains Malaysia,
Mathematical Geology,Vol.29,No.7, pp.891~917. Penang.
Carle, S.F., 1999. T-PROGS:Transition Probability Geostatistical Theodossiou , N., Latinopoulos , D., 2009. Economic aspects of the
Software, Version 2.1, delineation of well head protection areas under conditions
of uncertainty. Proceedings of the 2nd International
Hydrologic Sciences Graduate Group, University of California, CEMEPE & SECOTOX Conference, Mykonos.,
Davis. pp.309~314.
Jones, N.L., Walker, J.R., Carle, S.F., 2002. Using transition Weissmann, G.S., 2003. Application of transition probability
probability geostatistics with geostatistics in a detailed stratigraphic framework,
MODFLOW. Proceedings of Model CARE 2002, Prague, Czech University of New Mexico, USA.
Republic, pp.359~364.
192
A Case Study on Monitoring and Numerical Analysis of Groundwater Variation and

Inclinometer Displacement in Taiwan
Ching-Jiang Jeng 1 and Chia-Yu Yang 1
1
Department Environmental and Hazards-resistant Design, Huafan University, New Taipei City, Taiwan
E-mail: jcjhf@cc.hfu.edu.tw
ABSTRACT: This study examined monitoring and numerical analysis results of a dip slope case in New Taipei City, Taiwan. The
displacement data collected over the years by the inclinometers are used to study the creep of the slope, and combined with the influence of
the typhoon rainfall. The time lag and groundwater level changes and the effectiveness of the implementation of two catchpits in the slope
are reviewed. The Shape Acceleration Array (SAA) measurement is taken as the reference basis for hazard prevention and maintenance of
the slope. Geostudio is applied to analyze the relationship between rainfall infiltration, groundwater level, and the displacement of the entire
slope.
Keywords: Groundwater variation, inclinometer, SAA, numerical analysis.
The main exposed stratum under the site is the Mushan
1. INTRODUCTION Formation, with the bedrock being mainly interbedded with
sandstone and shale. It is a dip slope striking toward the east,
This study deal with the data obtained from monitoring system of dipping southward about 10°–20°. Huang and Jeng (2004) pointed
dip slope area at the Huafan University in northeastern Taiwan as out that through the observation and comparison of drilling cores
shown in Figure 1. and ground resistance images, as well as topographical
characteristics, there are two local small faults: a south-trending
Nanshihkeng Fault and a northwest-trending A Fault, as shown in
Figure 3.
Figure 1 Location of the study site
Jeng and Sue (2016) analyzed the monitoring data collected

from more than 300 settlement and displacement observation marks
on this site and compared them with the displacement recorded by
the inclinometers, concluding that a rise in the groundwater table
caused by typhoons is the most critical factor in slope stability.
Therefore, several countermeasures, including catchpits with
horizontal drainage pipes, were recommended. The Japan
Association for Slope Disaster Management (JASDiM)
recommended the threshold values of slope displacement for
different sliding stages, which were used to define three ranges. Figure 2 Inclinometer distribution and 2-D resisstivity image
Generally, slope displacement can be distinguished into several profiles line
stages, in which the three stages are: “initial displacement,”
“constant velocity displacement,” and “accelerated increment 3. ANALYSIS OF MONITORING RESULTS
displacement.” Xu (2011) pointed out obvious stages characteristics 3.1 Groundwater Variation Caused by Rainfall
for the gradual evolution of slope variations. To classify a type of
displacement into one of the three stages, the s-t curve for the Four sets of auto-recording groundwater-level gauges are
relationship between displacement and time can be converted into a designed in this research area for observation as shown in Figure 2.
T-t curve. The curve after conversion has a unique and deterministic Additionally, Figure 4 illustrates one of the monthly rainfall data
tangent angle (). Normally, the tangent angle of the curve is greater and groundwater-level changes in 2016. The rainfall types can be
than 45 degrees when it enters the acceleration stage. classified as post-peak rainfall, pre-peak rainfall, and pre-peak
rainfall, respectively. Additionally, when rainfall occurs,
2. MONITORING SYSTEM AND SITE INFORMATION groundwater level also changes, although with some time lag. The
degree of groundwater level change is W1>W2>W3>W4, which
The monitoring system set up in this site includes inclinometers, indicates that degree of change of the groundwater level in
tiltmeters, crack gages, water-level observation wells, settlement and observation wells is different across different regions. The W1
displacement monitoring marks, rebar strain gages, concrete strain groundwater level change is particularly evident, with the
gages, and rain gages. All of these monitoring instruments Figure 2 groundwater level change uplifted by rainfall up to about 23 m,
are continuously operational in taking regular measurements (see while in W2, W3, and W4, the amount of change was about 8 m, 6
Jeng and Lin, 2011; Jeng and Jiang, 2013; Jeng et al., 2015, 2017). m, and 1 m, respectively. This indicates that the groundwater level
in the area of W1 was more susceptible to rainfall influence. A
193
comparison of different rainfalls shows that the peak rainfall of post- Figure 5 shows that the greater the peak rainfall, the shorter the
peak rainfall was greater than the peak of other types of rainfall. reaction time lag is for the groundwater level to rise. With respect to
During the typhoon season, rainfall will cause greater rise of the the rainfall type perspective, the peak rainfall of the pre-peak
groundwater level, and the lag time between the peak rainfall and rainfalls is concentrated less than 20 mm and has a long
peak groundwater level will significantly shorten. When the groundwater-level time lag. The peak rainfall of the middle-peak
groundwater level in W1 rises to about ground level (GL) -30 m, rainfalls ranges from 20 mm to 40 mm, and the peak rainfall of the
W2 and W3 will then increase more noticeably. post-peak rainfalls is concentrated at around 50 mm with a short
time lag. It indicates that if this study area has rainfall greater than
50 mm, it should be closely observed and the observation time
should be stretched to about 40 hours. Further, Figure 6 shows that
for greater peak rainfall, the groundwater level will significantly
increase, and the post-peak rainfall will cause the most significant
groundwater level rise. According to the auto-recording
groundwater-level gauge, the groundwater rise amount was between
1 and 23 m. Lastly, Figure 7 shows that when the total average
cumulative rainfall to peak point exceeds 200 mm, the groundwater
level in auto-recording groundwater-level gauge in W1 will increase
up to 20–23 m.
As mentioned above, when the peak rainfall is greater, the time
lag of the groundwater level rise will be shorter. The peak rainfall of
the post-peak rainfall type is greater than that of other rainfall types,
which makes the change of its groundwater level significant. Thus,
observation for up to 40 hours is needed for the post-peak rainfall.
The data from the auto-recording groundwater-level gauge shows
that the groundwater level W1 is most affected by rainfall, with a
change between 1 m and 23 m. A groundwater-level change of 0.6–
5.8 m was found in the auto-recording groundwater-level gauge in
Figure 3 Site geological map W2, 0.25–3.1 m in W3, and 0.1–2 m in W4.
201609Time history (hr)

180
Time lag of the peak groundwater
0
0 100 200 300 400 500 600 700 800
60 160
level respond in W1 (hr)
-5 w1
140
-10 50 w2
120
-15 w3
w4 100 Later type

-20 40
Groundwater level (m)
80 Forward type
Hourly rainfall (mm/hr)
-25
-30 30 60 Center type

-35 40 Linear(w1)
-40 20 20
-45
0
-50 10 0 20 40 60 80
-55 Peak rainfall intensity (mm)
-60 0
Figure 5 Groundwater level and time lag of rainfall peak diagram of
Figure 4 Groundwater level and rainfall duration curves in automatic groundwater-level gauge W1.
September 2016.
25
Groundwater-level change caused by rainfall can vary due to (1)
rainfall type, (2) rainfall amount, (3) rainfall time lag, (4)
Groundwater riseup value (m)
20
groundwater-level gauge region, and (5) seasonal factors. Analyses
of these factors are illustrated in Figs. 5–7, which contain only
analysis data from observation well W1. These figures are, in this 15
post-peak type
order, the groundwater level and rainfall peak time lag diagram, the pre-peak type
peak rainfall and groundwater-level rise diagram, and the 10
medium-peak type
accumulated rainfall and groundwater-level rise diagram for W1 linear regression
groundwater-level gauge from 2013 to 2015. Figure 5 is the 5
groundwater level and rainfall peak time lag diagram of automatic
groundwater-level gauge W1. The x-axis is the peak rainfall, and the 0
y-axis is the time lag for groundwater level to reach its peak. Figure 0 20 40 60 80
6 is the peak rainfall and groundwater level rise diagram of Peak rainfall (mm)
automatic groundwater-level gauge W1. The x-axis is the peak Figure 6 Peak rainfall and groundwater level rise diagram of
rainfall, and the y-axis is the amount of groundwater level rise. automatic groundwater-level gauge W1.
Figure 7 is the accumulated rainfall and groundwater-level rise
diagram of automatic groundwater-level gauge W1. The x-axis is
the accumulated rainfall, and the y-axis is the groundwater-level rise
amount. The round dots are post-peak rainfalls, the square dots are
pre-peak rainfalls, and the triangle dots are middle-peak rainfalls.
There are two types of accumulated rainfall: accumulated up to the
peak (dark-colored symbols) and accumulated for the entire rainfall
(light-colored symbols).
194
implementation of the catchpits in 2016, the lowest curve has

SIS-40 decreased significantly to −52.03 m. The highest groundwater level
25 also dropped from −22.67 m to −27.84 m. This is due to the
implementation of the two catchpits at the site, which significantly
20 Segment 1 reduced both normal groundwater level and the highest groundwater
y = 0.0341x - 8.6631
level during rainy and typhoon seasons.
15 R² = 0.7358 Segment 2
10
Linear regression of
segment 1
5
0 segment 2
0 100 200 300 400 500 600 700 800
-5
Figure 7 Accumulated rainfall and groundwater level rise diagram
of automatic groundwater-level gauge W1.
3.2 Inclinometer Displacement Triggered by Typhoons

Figure 8 is the SIS-40 displacement and cumulative rainfall amount
from typhoon rainfall. The x-axis is the cumulative rainfall amount
from typhoon rainfall, and the y-axis is the displacement caused. Figure 9 Time graph of groundwater level and rainfall amount from
Different inclinometer displacements from different typhoons in the auto-recording groundwater-level gauge from April to July 2016.
past years are presented in the graph. Once the displacement starts to
accelerate, the accumulated rainfall is treated as an alert rainfall
threshold value. Figure 12 shows the typhoon cumulative rainfall
alert of SIS-40 has a threshold of about 350 mm. The typhoon
cumulative rainfall alert threshold values from the other nine
inclinometers are summarized and shows that the alert cumulative
rainfall threshold for heavy typhoon rainfall causing campus slope
displacement is approximately 315–495 mm.
SIS-40
25
20 Segment 1
y = 0.0341x - 8.6631
15 R² = 0.7358 Segment 2 Figure 10 Comparison of groundwater level before and after
catchpits implementation.
10
segment 1 3.4 Discussion of SAA displacement measurement results
5
Linear regression of This study sought to understand slope displacement caused by
0 segment 2 typhoon rainfall, so the collection of rainfall data and SAA
0 100 200 300 400 500 600 700 800 displacement data continued until February 2017 to observe and
-5 analyze the relationship between typhoon rainfall and displacement
Figure 8 SIS-40 displacement and cumulative rainfall amount from of SAA. In this paper, only part of the data was taken and enlarged,
typhoon rainfall. shown in Figs. 11–12. The x-axis is the date, the left y-axis is the
rainfall amount, and the right y-axis shows the accumulated
3.3 Comparison of Groundwater Level Change to Explore displacement of inclinometer SIS-11 and inclinometer SIS-20.
the Effectiveness of Catchpits As shown Figures 11 to 12, there were two typhoons in August
2015, the medium typhoon Soudelor and the strong typhoon Goni.
Figure 9 is the relation graph of rainfall data and groundwater level Accordingly, SIS-11 had displacement of 6.048 mm and 1.489 mm,
from the auto-recording groundwater-level gauge from April to July respectively, and the displacement of SIS-20 was 4.839 mm and
2016. The x-axis is time, the left y-axis is the water level, and the 1.64 mm, respectively. In October 2016, there was light typhoon
right y-axis is the rainfall amount of the day. The groundwater level Aere along with its peripheral circulation. SIS-11 had a consequent
from the auto-recording groundwater-level gauge in W1 has a red displacement of 11.321 mm, while SIS-20 had a displacement of
curve, the water level in W2 has a green curve, the water level in 3.384 mm. The above data are plotted in Figure 13 as an SAA
W3 has a purple curve, and the groundwater level in W4 has a light displacement increment graph during typhoon rainfall.
blue curve. The dark-blue curve shows the daily rainfall amount Figure 13 shows that when the typhoon cumulative rainfall
reaches 184 mm, significant displacement will occur; when the
from April to July. The figure shows that, similar to aforementioned
cumulative rainfall reaches 300 mm or more, or when the peak
data, W1 is much more affected by the rainfall amount than the
rainfall exceeds 156 mm, the displacement will begin to accelerate,
other three auto-recording groundwater-level gauges. The data from
producing displacement larger than 3 mm. This result agrees with
the auto-recording groundwater-level gauge in W1 from 2013 to that in Section 3.2 of this study. Therefore, the study sets typhoon
2016, as shown in Figure 10, shows that the normal groundwater rainfall of more than 300 mm as the alert threshold. In other words,
level from the auto-recording groundwater-level gauge in W1 has a
lowest value of about −43.67 m from 2013 to 2015. After the
195
if the weather forecast reports a typhoon rainfall of more than 300

mm, then the displacement situation should be closely monitored.
200 165
SIS-11A Cumulated displacement

180
160 160
140
Rainfall (mm)
120 155
100
(mm)
80 150 Rainfall
60 SIS11A_12.5m-X
40 145 SIS20_15.0-X
20
0 140 Figure 14 Cross-section profile of SEEP/W.
2015/8/15
2015/8/23
2015/8/11
2015/8/13
2015/8/17
2015/8/19
2015/8/21
2015/8/25
2015/8/27
2015/8/29
2015/8/1
2015/8/3
2015/8/5
2015/8/7
2015/8/9
Table 1 Input Parameters for numerical analysis

Date
Sandstone
Colluvium Gravel Shear
Figure 11 SIS-11 and SIS-20 displacement versus rainfall amount in Soil layer
soil layer
and shale
layer zone
August 2015. interbedded
Deformation Linear Interface
Elastoplastic Elastoplastic
200 10/6~10/16 190 model elasticity elements
SIS-11A Cumulated Displacement
180 Peripheral circulation of Typhoon Aere 188

Elastic modulus
160 186 3 × 104 3.225 × 105 3.753 × 106 3748
140 184 (kPa)
Rainfall (mm)
120 182 Poisson’s ratio 0.334 0.28 0.23 0.334

10/5~10/6
100 180
(mm)
Typhoon Aere
80 178 Unit weight
Rainfall 19.31 25.52 23.86 23.3
60 176 SIS11A_12.5m-X
(kN/ m2)
40 174 SIS20_15.0-X Cohesion, C
20 172 18.5 41.8 38.7 0
0 170
(kN/ m2)
Internal friction
2016/10/1
2016/10/13
2016/10/25
2016/10/3
2016/10/5
2016/10/7
2016/10/9
2016/10/11
2016/10/15
2016/10/17
2016/10/19
2016/10/21
2016/10/23
2016/10/27
2016/10/29
angle, Φ 29.6 32.13 32.74 23

(degrees)
Date
Figure 12 SIS-11 and SIS-20 displacement versus rainfall amount in This study used rainfall data during typhoons in August 2015
October 2016. and October 2016, the amount of groundwater-level variations from
four auto-recording water-level gauges, as well as two auto-
14 recording inclinometer SAAs, to observe and analyze the
relationship between rainfall amount during typhoon, slope
SAA Displacement increatement (mm)
SIS-11A Peak rainfall

y = 0.4616e0.0042x
12
R² = 0.5387
displacement, and groundwater-level variation. The results are
SIS-20 Peak rainfall aggregated to show the trend. Two typhoons struck in August 2015,
10 y = 0.433e0.0097x the typhoon Soudelor during 8/6–8/9, and the typhoon Goni during
8
R² = 0.4204 8/20–8/23. Soudelor generated a rainfall amount of 342.5 mm, an
SIS-11A Cumulated SIS-11A displacement increment of 6.048 mm, and an SIS-20A
rainfall
6 displacement increment of 4.893 mm. Typhoon Goni generated a
4
SIS-20 Cumulated rainfall amount of 30.5 mm, an SIS-11A displacement increment of
rainfall 1.489 mm, and an SIS-20A displacement increment of 1.64 mm. A
2 Average peak rainfall southern low-pressure system (typhoon Aere) generated a rainfall
amount of 794 mm, an SIS-11A displacement increment of 11.321
0 mm, and an SIS-20A displacement increment of 3.384 mm. The
0 200 400 600 800 Average cumulated
rainfall
rainfall amount and actual displacement data were used as a
Rainfall induced by typhoons (mm) reference in the GeoStudio software simulation analysis. Figure 15
shows the case of SIS-20 and SIS-11 inclinometer displacement
Figure 13 SAA displacement increment during typhoon rainfall. increment diagram during one of the typhoon Soudelor. These
figures demonstrate the displacement result simulated by this model
4. NUMERICAL ANALYSIS OF GEOSTUDIO has the same displacement curve trend as the actual monitoring data.
SIMULATION This study adopted the Morgenstern-Price extreme balance
In this study, the geological and hydrological data of the site were method to calculate the safety factor of the sliding surface. Figure 16
entered into GeoStudio software to carry out SEEP/W, SLOPE/W, shows the profile of SLOPE/W analysis. The most critical potential
and SIGMA/W numerical analysis. The simulation results were sliding surface existed between Jue-zhao building and Wu-ming
compared to the monitoring data to analyze the effect of rainfall on building is shown in the profile. The results of safety factor variation
the slope. The initial condition setting: boundary conditions of during typhoon are shown in Figure 17. Figure 17 shows that during
rainfall infiltration are shown in Figure 14, wherein AB is a the Typhoons Aere , although the safety factor of the slope will drop
boundary of the lower constant head slope, and the total head height to about 0.6, the minimum safety factor for rainfall can still be
is set to H = 387.8 m; CD is the head boundary of the upper slope, maintained at 1.2. This illustrates that even though typhoon rainfall
and the total head height is set to H = 534.5 m. The upper and lower will cause a certain amount of slope displacement at the campus, its
boundaries of the head setting mainly use the average groundwater safety factor is still higher than the general suggested value of 1.2;
level from historical monitoring data. The BC is the impervious in other words, there is not yet immediate danger of sliding.
boundary (Q = 0); AD is the boundary of rainfall infiltration. Table However, further studies should investigate the impact on safety
1 shows the input parameter for the analysis factors in situations of higher typhoon rainfall through actual
follow-up test results and feedback analyses.
196
2. After the implementation of the catchpits, the lowest

groundwater level has decreased from −43.67 m in 2015 to
−52.03 m in 2016. The highest groundwater level also dropped
from −22.67 m to −27.84 m. Therefore, the normal groundwater
level and highest groundwater level has been significantly
decreased during the rainy seasons and typhoon seasons by the
draining effectiveness of the two catchpits.
3. When the typhoon cumulative rainfall reaches 184 mm,
significant displacement will occur; when the cumulative rainfall
reaches 300 mm or more, or when the peak hourly rainfall
exceeds 100 mm, the displacement will begin to accelerate,
producing displacement of larger than 3 mm. Therefore, the
study sets typhoon rainfall of more than 300 mm as the alert
threshold.
4. The typhoon Soudelor simulation shows that the displacement
Figure 15 SIS-20 displacement increment (left) SIS-11 displacement result simulated by this model has the same displacement curve
increment (right) of simulation of the typhoon Soudelor. trend as the actual monitoring data.
5. The result agrees with the results mentioned above; generally,
only when the rainfall exceeds 77 mm will the displacement be
driven. In other words, simulation results from the software
using cumulative rainfall and peak rainfall greater than 60 mm
will be more important and more meaningful than simulation
results from simulating a single peak rainfall.
6. The slope model analytical results show that during the
Typhoons Aere, although the safety factor of the slope will drop
to 0.6, the minimum safety factor for rainfall was still maintained
at 1.2, higher than the general suggested value of 1.2. This means
that there is not yet any immediate danger of sliding.
6. REFERENCES
Huang, C. S. and Jeng, C. J.: A supplementary geological survey
Figure 16 Profile of potential sliding surface analysis and analysis of the Talun area around the Huafan University.
Journal of Art and Design of Huafan University, (1), 59–69,
2004.
Japan Landslide Association for Slope Disaster Management
Factor of Safety (JASDiM): Essential implementation of technical design for
2 landslide countermeasures, (2), 22, 1978 (in Japanese).
1.8
Typhoon Aere Jeng C. J., and Lin, T. A.: A case study on the in-situ matrix suction
1.6 monitoring and undisturbed-sample laboratory test for the
Factor of safety
1.4 unsaturated colluvium slope. Soils and Foundations, 51, (2),

1.2 321–331, 2011.
1 Jeng, C. J. and Jiang, J. R.: Research on serial behaviors of
Factor of Safety
0.8 colluviums slope from rainfall infiltration caused ground
0.6 water variation to the slope stability and displacements.
0.4
Journal of Art and Design of Huafan University, (8), 17–31,
0.2
2013.
0
0 100 200 300 400
Jeng, C. J. and Yang, C. Y.: A case study on groundwater variation
typhoon duration, hours and slop displacement induced by rainfall. Journal of Art and
Figure 17 Variation of safety factor in typhoon simulation. Design of Huafan University, (10), 141–156, 2015.
Jeng, C. J. and Sue, D. Z.: Characteristics of ground motion and
threshold values for colluvium slope displacement induced by
5. CONCLUSION heavy rainfall. Nat. Hazards Earth Syst. Sci., (16), 1309–
Based on the results, the following conclusions can be made: 1321, 2016.
1. The groundwater-level W1 is most affected by the rainfall, with Xu Q.: Early warning and emergency disposal of landslide disaster,
a variation of 1–23 m. The groundwater level variation was 0.6– state key laboratory of geohazard prevention and
5.8 m for W2, 0.25–3.1 m for W3, and 0.1–2 m for W4. When geoenvironment protection: An academic report in National
the peak rainfall is higher, there is less delay of groundwater Chung Hsing University, Taiwan, 2011.
level rise, and the peak rainfall of post-peak rainfall is larger than
that of other rain types, which makes the groundwater-level
variation large. Thus, observation up to 40 hours is needed for
post-peak rainfall type.
197
Numerical Simulations of an Excavation Case in Central Jakarta by Hypoplasticity

Model for Clays
Melisa Kosasi1, Fuchen Teng2, Benson B. C. Hsiung3
1
Department of Civil and Construction Engineering, National Taiwan Univ. of Science and Technology, Taiwan;
2
Department of Civil and Construction Engineering, National Taiwan Univ. of Science and Technology, Taiwan;
3
Department Civil Engineering, National Kaohsiung Univ. of Science and Technology, Taiwan;
E-mail: melisakosasi9@gmail.com; FTeng@mail.ntust.edu.tw; benson@cc.kuas.edu.tw
ABSTRACT: Numerical simulations of a top-down constructed excavation in Central Jakarta was carried out in this study. The excavation
supported by the diaphragm wall has been constructed in a medium-stiff Central Jakarta clay. Measurements on wall deflections were made
during the construction. The numerical simulations were conducted by using a hypoplasticity model (HC model) for clays which is capable
of modeling small strain non-linearity, soil anisotropy, and recent stress-history effects. Properties of the medium-stiff Jakarta clay were
collected and studied carefully to calibrate the HC model. Numerical simulations with consideration of pre-excavation recent stress history
effect is thus suggested herein and parameters of HC model used for excavations in Central Jakarta are thus recommended also.
1. INTRODUCTION of a very thick (more than 100 m) greenish silt with consistency
from very stiff to hard. Laboratory test results of Jakarta clay such
Jakarta is a rapidly developing megacity that demands to maximize as index properties, oedometer tests, consolidated undrained tests
the use of underground space. More and more large deep and seismic down-hole test had been taken. Field measurement of
excavations are under construction and their deformations will affect wall deformation is also collected. Soil properties for Central Jakarta
adjacent infrastructures such as buildings, road, bridges, and public clay at the site are shown in Figure 1.
utilities. To properly estimate the interactions among the affected The depth of the excavation is 24.85 m, with 1.2 m thick and
soil, excavation support system and adjacent structures in these 33.7 m deep of diaphragm wall. The groundwater level was at a
projects, advanced soil constitutive models must be used to depth of 2.8 m below the ground surface (GL-2.8 m). A road
represent soil behavior during design. Features of soil behavior that embankment with total length 66 m is located at one side of
affect the calculation of ground movements including soil embankment. Figure 2 shows the excavation profile.
nonlinearity at very small strains, stiffness anisotropy, soil
compressibility and recent stress history effects should be well
simulated in the soil model.
A non-symmetric excavation in Central Jakarta was used as a
case history. A 3.1 m height of embankment with width about 66 m
was located at one side of the excavation. The construction of the
embankment altered the stress-history of soils subjected to
excavation activities later. The so-called recent stress history effect
is defined as the pre-excavation events that generated the current
state of effective stresses of the clays. The effect of recent stress
history of soils have been studied mostly at the laboratory scale for
London clay (Atkinson et al. 1990), Bothkennar clay (Smith et al.
1992), Boston blue clay (Santagata et al. 2005), and Chicago clays
(Cho 2007, Finno and Kim 2012). Significant differences on soil
stiffness in the small strain range were found as a result of those
experimental programs, and it was concluded that the observed soil
responses depends on the recent stress history of the material prior
to shearing.
Figure 1. Soil properties
An advanced soil constitutive model, hypoplasticity model for
clays (HC model) developed by Masin (2005, 2013 and 2014), is
adopted in this study. HC model is capable of capturing the features 3. SOIL PARAMETER
that represent the small strain stiffness of soils, stiffness anisotropy The basic hypoplasticity model have 5 parameters: ’, *, *, N and
and recent stress-history effects. (Arboleda et al 2017, Teng et al. pp. These parameters are similar to parameters used in Modified
2018). Numerical simulations of the hypothetical excavation made Cam Clay model. The model is not based on a conventional
through the Central Jakarta clays are presented to illustrate the effect elastoplastic framework decomposing strains in elastic and plastic
of pre-excavation recent stress history effects on the computed components, and consequently does not involve complex yield-
ground movements. The performance of HC model, with well- surface and plastic-potential definitions. It captures the nonlinear
calibrated parameters, on the magnitude and distribution of the behavior of soils at large strains in the framework of critical-state
computed excavation-induced ground movements is presented soil mechanics by means of the Matsuoka-Nakai shape of the
herein. critical-state locus in stress space. HC model used herein is
enhanced by intergranular strain concept which can represent the
2. PROJECT BACKGROUND soil behavior at small strain. Since this model is using critical soil
Soil in Jakarta is generally composed by quaternary and tertiary mechanic concept, the friction angle is determined by stress at
deposit (Firmansyah & Sukamta 2000). Quartenary deposit formed critical state. There are 9 additional soil parameters for intergranular
by volcanic ash which divided into 3 layers: 3-5 m thick of upper strain parameter: G, E, , Ag, ng, mrat, R, r, and . Table 1
lahar; alternate silty clay, silty sand and sandy silt; and about 5 m summarized the parameters of the hypoplasticity model for clays.
thick of lower lahar that consists of cemented silty sand. Tertiary
deposit is located 35 m below the ground surface. This layer consists
198
Figure 2. Excavation profile
Table 1. Parameters for HC model Table 2. parameter of Central Jakarta Clay for HC model
No. Symbol Parameter name Upper Lower
1 N Position of normal compression line No. Symbol clay clay Source
2 * Slope of normal compression line Value
3 * Slope of unloading line 1 N 1.52 1.285 Oedometer tests
4 ’ Critical-state friction angle 2 * 0.126 0.098 Oedometer tests
5 pp Stiffness fitting parameter 3 * 0.018 0.011 Oedometer tests
Anisotropy coefficient of shear, 4 ’ 38 39 CU tests
6 GE Optimization
Young and Poisson's moduli 5 t 15 5
7 Ag Stress dependency of Gvh on mean from CU test
8 ng normal stress 6 pp 0.1 0.1 (Masin 2014)
Very small strain behaviour upon 7  15.46 16.67 Test data
9 mrat
strain path reversals 8 e0 1.81 1.62 Test data
10 R Size of elastic range 9 G 1 1 Assumed value
Material constant controlling the rate 10 Ag 15500 15500 Downhole tests
11 r of evolution of the intergranular strain 11 ng 0.46 0.46 Downhole tests
tensor (Arboleda et al
Material constant controlling the rate 12 mrat 1 1
12  2017)
of degradation of the shear stiffness
13 5E-05 5E-05 (Masin 2014)
13 OCR Over consolidated Ratio
14 r 0.08 0.08 (Masin 2014)
Since available data for Central Jakarta Clay at small strain level is 15  0.9 0.9 (Masin 2014)
limited, some parameters were evaluated from several references (as listed Optimization
16 OCR 6.25 3.37
in Table 3). G is set to be 1 to ignore the anisotropy behavior. The from CU test
parameter mrat is an intergranular strain parameter that controls the very
small strain behavior of clays upon strain path reversals. A value of 1 is
chosen for this parameter as a material independent constant and the very 3.1 Parameter *,*, and N
small shear stiffness was controlled with the parameters Ag , ng and the
remaining three intergranular strain parameters (Arboleda et al 2017). R as Thirteen oedometer test results around excavation site were
size of elastic range was taken equal to 5E-5 (Masin 2014). In fact, collected to determine *,*,  and OCR. *,*, Nare obtained
Central Jakarta Clay could be further categorized into 2 layers: upper clay from ln(1+e) vs ln p space from oedometer test results. * and , *
and lower clay. Some parameters calibrated in this simulation mainly are slope of normal compression line and unloading line
focus on lower clay layer because this layer dominates to wall
respectively as shown in Figure 3.
deformation. Table 2 summarized parameters used in this study.
199
Figure 3. Definition of parameters N, and *

(a) (b)
In order to evaluate the sampling disturbance, specimen quality
designation (SQD) value of 13 oedometer data results were
calculated by method propose by Terzaghi et al. (1996). This
method proposed volumetric strain measured at ’v during
oedometer test as a criteria for determining SQD value. Figure 4
shows SQD calculated for these data. SQD result shows poor
result, around D to E. Based on recommendation proposed by
Terzaghi et al. (1996), soil samples with SQD in range D to E
would not produce reliable estimates of ’p . Poor SQD value will
result in decreasing ’p .
Oedometer test data with volumetric strain at ’v0 below 10
was chose to determine *,*, and  parameter used in this
simulation. From data chosen with SQD below 10, parameter
*,*, and N are determined by average value for each layer.
Figure 5 a, b, c, d shows the figure of various data of *,*, and
OCR versus depth.
(c) (d)
Figure 5. Oedometer test results
3.3 Parameter ’ and t

CU test results were collected and used to determine values of ’,
t and over consolidated ratio (OCR) parameter. ’ is friction
angle at critical state. t is shift of mean stress due to cohesion.
For upper clay layer, ’=38o and t =15 are used in this
simulation based on CU test result. The critical state friction angle
for upper soil layer is high. This value shows even for upper clay
layer, the soil characteristic is a stiff clay layer. From oedometer
test result, OCR for upper clay layer is equal to 3.37. Lower clay
layer is a stiff clay soil with used parameters are ’=39o, t =5 and
OCR=6.25. OCR for lower clay layer is higher than obtained
from oedometer test result. This value was adopted because it has
good alignment with calibration result using HC model which will
Figure 4. Volumetric strain at (’v) vs. Depth for specimen be explained in section 3.6.
quality designation (SQD)
200
3.4 Parameter Ag and ng

Central Jakarta clay was formed primarily from volcanic ash.
Parameter Ag and ng define the stress dependency of Gvh on mean This soil type shows very strong structure even with no history
normal stress as proposed by Masin (2014): that soil has been subjected to high overburden pressures in the
 p'  (1) past. A clay layer formed by volcanic ash in New Zealand
Gvh  pa Ag  ng
 pa  (Jacquet 1990) shows a high OCR values around 17. Wallace
where pa= reference pressure parameter (1 kPa), Gvh = very small (1993) and Millar (1986) proposed the chemical composition
shear modulus in which the double indices refer to the (Iron oxide) of clay may play an important role to the OCR value.
propagation direction and polarization of a shear wave, and Based on the calibration of CU test, a OCR of 6.25 was obtained
p’=mean stress. Figure 6 shows range of Gvh/pa vs. p’/pa from to yield a reasonable undrained shear strength for Central Jakarta
down-hole seismic test result. Ag and ng are determined according clay. Figure 8 shows the comparison of CU test between
to the regression line. laboratory test result and simulation result.
Figure 8. Comparison of CU test result for lower clay layer from

laboratory and HC model
4. NUMERICAL MODEL OF EXCAVATIONS

In this excavation simulation, soil layers were divided into 2 soil
layers. The upper clay, from ground surface level (GL) 0.0 m to
GL -3.8 m is a stiff clay layer with ’=38o. The lower clay layer,
from GL -3.8 m to GL -65m, is a very stiff clay layer with ’=39o.
There are 11 phases in this simulation as listed in Table 3.
Figure 6. Down-hole seismic test result (Shear modulus at small Excavation zone is located right next to a 3.1m height of road
strain vs. mean stress in terms of reference pressure) embankment with width is about 66 m. This road embankment is
later connected to a bridge. Because the embankment was built
3.5 Parameter r and  for transportation purpose, the original form of embankment area
is assumed as a horizontal ground surface which later is loaded
r is material constant controlling the rate of evolution of the and consolidated by this embankment. This road embankment is
intergranular strain tensor.  is material constant controlling the located at central Jakarta that had been built over decades. The
rate of degradation of the shear stiffness. These parameters is construction of this embankment is expected has a recent stress
usually calibrated by optimisation of CU test. r and in this history effect to the wall movement. In this simulation, the
simulation are 0.08 and 0.9, respectively. These values were embankment is consolidated for 20 years in order to make the
suggested by Masin (2014) for stiff clays. excess pore pressure fully dissipated. This consolidation step is
followed by the installation of the diaphragm wall and four
3.6 Calibration of Oedometer and CU test result excavation stages.
Fig 4. Shows one oedometer calibration simulation for *,*, and
Table 3. Stage construction
N. The purpose of this calibration is to confirm HC model
represent the same soil behavior with laboratory test result. One Phase Stage Construction
0 Initial phase
calibration result of oedometer test shown in Figure 7 has good
agreement with laboratory test result. 1 Consolidation of embankment for 20 years
2 Install diaphragm wall
3 Excavate to GL. -4.9m (1st excavation stage)
Oedometer test 4 Install B1F at GL. -3.9m
Hypoplasticity model
5 Excavate to GL. -11m (2nd excavation stage)
6 Install B1F slab at GL. -10.2m
7 Excavate to GL. -16.9m (3rd excavation stage)
8 Install B2F slab at GL. -16.1m
9 Excavate to GL. -24.85m (4th excavation stage)
10 Install bottom slab GL. -24.05m
Finite element mesh of the excavation case is shown in Figure

9. Dimension of numerical model in horizontal and vertical are
247 m and 65 m, respectively. The excavation width is 20.5 m.
The embankment is symmetry with total length and height of
embankment are 66 m and 3.1 m. The height of embankment
Figure 7. Comparison of oedometer test results with computations inclines from zero to 3.1 m with horizontal length equal to 11 m
from HC model from excavation side. The left boundary is at a distance of 80 m
201
from the diaphragm wall and the right boundary is at a distance of

80 m from the end of embankment. The left and right vertical
boundaries were restrained from horizontal movement and the
bottom was restrained from both of the vertical and horizontal
movements.
Figure 9. Finite element mesh of excavation case

at Central Jakarta
The stiffness of structural parameters is reduced by 20%, (a)1st excavation stage (b) 2nd excavation stage
considering that the stiffness of the concrete retaining wall
reduces
when subjected to large bending moment of diaphragm wall
causes the occurrence of the crack in the concrete. The axial
stiffness of the concrete floor slabs is also reduced by 20%. Table
4 contains material properties of structures used in numerical
simulation and t, E and means thickness, elastic modulus and
Poisson ratio.
Table 4. Material properties of structures

Structure Type t (m) E (Mpa) 
B1F slab 0.8 21000 0.15
B2F slab 0.4 21000 0.15
B3F slab 0.4 21000 0.15
B4F slab 1 21000 0.15 (c)3rd excavation stage (d) 4th excavation stage
D-Wall wall 1.2 21000 0.15
5. NUMERICAL SIMULATION RESULTS

Wall deflections induced by excavation were monitored by
inclinometers. Correction of inclinometer data result had been
done as suggested by Hsiung & Hwang (2009). The inclinometer Figure 10. Deformations of diaphragm wall at no-embankment
reading can be trusted once the tip of inclinometer properly side at various stages
embedded into a stable stratum to avoid the inclinometer tip move
together with diaphragm wall. Corrections had been calculated by Maximum wall deformation from simulation result of final
using B1F slab as referred point at first excavation stage. Other excavation stage at embankment side is larger around 10 mm than
inclinometers data at B1F slab which move inward to excavation that at no-embankment side. The possible reason is even the
zone after first excavation stage had been moved fit to the embankment has already been consolidated for 20 years and
reference point. excess pore pressure calculated in simulation is very small, under
At no-embankment side, corrected maximum deformation of 1kPa, the embankment load still has an effect to wall deformation.
diaphragm wall based on inclinometer data result at final It is aware that a significant difference is seen in the aspect of
excavation stage after completion of slab and 1 month after are 25 lateral wall movement measured for the time after the completion
mm and 42 mm, respectively. At embankment side, corrected of B4F slab and 1 month after and the reason has to be further
maximum deformation of diaphragm wall at final excavation explored in the future.
stage after completion of slab and 1 month after are 22 mm and
45 mm, respectively. Figure 10 shows wall deflection of
simulation result and field measurement at no-embankment side.
Wall deformation of simulation result at first excavation stage
show slightly different with field measurement while at second
and third excavation stages show good agreement with field
measurement. For last exacavation stage, there is deformation
difference between deformation at completion of slab and 1
month after completion of slab. Maximum wall deformation of
simulation result is 33 mm.
Figure 11 show inclinometer data result and simulation result
of embankment side of diaphragm wall. Wall deformation at 1st
excavation stage shows slightly deformation difference to field
measurement. For 2nd and 3rd excavation stages, maximum wall
deformations of simulation result are 5-7 mm larger than field
measurement. Maximum wall deformation at final stage is 43
mm.
(a)1st excavation stage (b) 2nd excavation stage
202
Atkinson, J. H., Richardson, D., and Stellabrass, S. E. (1990).

“Effect of recent stress history on the stiffness of
overconsolidated soil.” Geotechnique, 40(4), 531-540
Cho, W., (2007). “Recent stress history effects on compressible
Chicago glacial clay.” Ph.D. thesis, Northwestern Univ.,
Evanston, IL.
Finno, R. J., and Kim, T. (2012). “Effects of stress path rotation
angle on small strain responses.” J. Geotech Geoenviron.
Eng., 10.1061/(ASCE)GT.1943-5606.0000612, 526-534
Firmansyah, I., & Sukamta, D. (2000). Common Practice
Basement Construction in Jakarta-Indonesia. ACF Symposium
Technical Report, 28-39.
Hsiung, B. B. C., & Hwang, R. N. (2009). Correction of
inclinometer readings for movement at tips. Geotechnical
(c)3rd excavation stage (d) 4th excavation stage
Engineering, 40(2), 39-48.
Jacquet, D. (1990). Sensitivity to remoulding of some volcanic
ash soils in New Zealand. Engineering Geology, 28, 1-25
Masin, D. (2005). A hypoplastic constitutive model for clays.
International Journal for Numerical and Analytical Methods
Figure 11. Deformations of diaphragm wall at embankment side in Geomechanics, 29(4):311-336
at various excavation stages Masin, D. (2013). “Clay hypoplasticity with explicitly defined
asymptotic states.” Acta Geotechnica, 8(5), 481-496
6. CONCLUSION Masin, D. (2014). “Clay hypoplasticity model including stiffness
Based on laboratory data, field measurement collected and finite anisotropy.” Geotechnique, 64(3), 232-238
element analyses, the conclusions of this study are as following: Millar, P. 1986 Taranski Brown Ash A Discussion on the
1. Oedometer test results show that: the range of * for Central Influence of Iron Oxide Bonding on Engineering Performance
Jakarta Clay is from 0.08 to 0.16; * is in a range of 0.005 to Ministry of Works and Development, Central Laboratories,
0.025; N is in a range of 1.2 to 1.6. Actually, Central Jakarta Report No. 2-86/8
Clay could be further categorized into 2 layers: upper clay PLAXIS 2D [Computer software].
and lower clay.
Santagata, M., Germaine, J. T., and Ladd, C. C. (2005). “Factors
2. Calibration of CU test show good agreement with ’=39o for
lower clay layer. This indicates that lower clay layer is a stiff affecting the initial stiffness of cohesive soils.” J. Geotech.
clay layer with high friction angle. From this calibration, it is Geoenviron. Eng., 10.1061/(ASCE)1090-
also indicated OCR for lower clay layer is at high value, 0241(2005)131:4(430), 430-441
equal to 6.25 in consideration of volcanic structure. Smith, P. R., Jardine, R. J., and Hight, D. W. (1992). “The
3. The deformation analysis of diaphragm wall due to yielding of Bothkennar clay.” Geotechnique, 42(2), 257-274
excavation histories have a better agreement at earlier Teng F.-C., Arboleda-Monsalve L. G., Teng F.-C., and Finno R.
excavation stages. Simulation on pre-excavation activity by
(2018). “Numerical Simulation of recent stress-history effects
advanced HC model is applicable to Central Jakarta Clay and
yield a good simulation results. on excavation responses in soft clays.” J. Geotech. Geoenv.
4. Research about small strain behavior in Indonesia is still Eng., 10.1061/(ASCE)GT.1943-5606.0001921
limited. As a pioneer study for HC model on Central Jakarta Terzaghi, K, Peck, R. B., & Mesri, G. (1996). Soil mechanics in
Clay, this study provides a range for parameters. Further engineering practice (3rd ed.). New York: Wiley.
research in small strain behavior is highly demanded for the Wallace, K.B. (1973). “Structural behavior of residual soils of the
advanced soil model. continually wet Highlands of Papua New Guinea.”
Geotechnique, 23, 471-494
7. REFERENCES
Arboleda-Monsalve L. G., Teng F.-C., Kim T. and Finno R.
(2017). “Numerical Simulation of Triaxial Stress Probes and
Recent Stress-History Effects of Compressible Chicago
Glacial Clays.” J. Geotech. Geoenv. Eng.,
10.1061/(ASCE)GT.1943-5606.0001684, 04017029
203
3D Finite Element Analysis of Deep Excavation in Central Jakarta

using Total and Effective Shear Strength Properties
Tatag Yufitra Rus1, Bin-Chen Benson Hsiung2 and Kuo-Hsin Yang3
1
Department of Civil and Construction Engineering, National Taiwan University of Science and Technology, Taipei, Taiwan
2
Department of Civil Engineering, National Kaohsiung University of Science and Technology, Kaohsiung, Taiwan
3
E-mail: tatagyufitrarus@gmail.com
ABSTRACT: This paper presents a study on 3D finite element analysis of a large-scale deep excavation in Central Jakarta. The soil, Central
Jakarta clay, is modeled by both Hardening Soil and Mohr-Coulomb model in Plaxis 3D. The total and effective stress analyses under undrained
conditions were used to model the short-term undrained conditions of the clay during excavation. The input soil properties were established
using data based on an extensive site investigation program including in situ and laboratory tests, and empirical relationships with standard
penetration numbers (SPT-N). The top-down construction method and a supporting system of concrete slabs are also considered in the
numerical model. The numerical results are compared with measured wall deformation from inclinometers at a certain distance from the corner
of excavation. The simulations using drained and undrained soil shear strength (Undrained A and B functions in Plaxis 3D) are also compared
and discussed.
Keywords: Jakarta, deep excavation, 3D finite element, total and effective stress analysis, wall deformation.
1. INTRODUCTION To raise the stiffness of the retaining wall, 414×405×18×28 steel

h-beams were designed as the kingposts in the middle of the
Jakarta is the capital city of Indonesia has a population of more than excavation area at 3.0 m intervals. The function of kingpost is to hold
10 million continues to increase every year. The increase in the weight of reinforced concrete slabs. The h-beams were inserted
population in Jakarta would be directly proportional to the need of the 4.0 m into bored piles with a diameter of 1.0 m and an embedded
increase in infrastructure services in Jakarta and use of deep length of 14.5 m.
excavation is expected to be an essential scheme for private or public The diaphragm wall and kingposts were constructed before the
sectors. The role of the deep excavation in Jakarta is thus becoming soil was excavated to ground level (GL) −1.5 m. In the next phase, a
important. deck slab with a thickness of 0.4 m was installed. The second
The wall displacement induced by deep excavation is the critical excavation stage involved removing the soil to GL −5.28 m and then
factor in designing the deep basement. However, limited references constructing a top slab at GL −3.88 m. Subsequently, the soil was
to high-quality soil properties data and few of the large-scale deep excavated to GL−11.38 m and a middle slab with a thickness of 0.4
excavations that have been studied are the challenges in Central m was placed at GL−10.38 m. The fourth and fifth excavations were
Jakarta. Moreover, most of the valuable information obtained is excavated after the middle slab installed. In the last construction
documented by the local language which raises the difficulty of deep stage, the bottom slab with a thickness of 1.0 m was installed at
excavation research in Central Jakarta. GL−17.73 m. Table 1 details the construction phases and time
The 3D FE analysis is technically used to study 3D wall behavior sequences of the excavation for this study.
by the concept of corner effect which was first proposed by Ou et al.
(2006) to indicate the location of the plane-strain of the diaphragm
wall. The study of PSR for the Central Jakarta has been established
using Undrained B (Hsiung et al. 2018) and was evaluated by
comparing the result with using Undrained A to examine the reliable
effective stress parameter in Central Jakarta clay.
This paper presents a well-documented case of large-scale deep
excavation in Central Jakarta clay using total and effective stress
parameter by 3D FE analysis. The excavation is practically
completely placed in the thick layer of Central Jakarta clay. Detailed
information corresponding to the structure details, soil conditions, in
situ and laboratory soil tests, construction sequences, and monitoring
data are introduced. The input soil properties were established using
data based on an extensive site investigation program including in situ
and laboratory tests, and empirical relationships with standard
penetration numbers (SPT-N) and laboratory Triaxial tests. The
results of numerical analysis by Plaxis 3D are also compared with
previous results of Undrained B conducted by Hsiung et al. (2018)
and discussed with the field measurement data to obtain the reliability
of the parameters that were used in the simulations.
2. PROJECT DESCRIPTION AND SITE EXPLORATION

2.1 Details of excavation site condition
A deep excavation in Central Jakarta was selected for the case study
and numerical analysis. Figure 1 shows the cross-section of this study.
The construction was performed using a top-down method with five
excavation stages, supported by four-level reinforced concrete slabs
with various thicknesses. The maximum excavation depth was 19.03
m in the final excavation stage. The excavated pit was retained using
a 1.0 m thick and 24.2 m deep diaphragm wall. Figure 1 Cross-section and soil profile of the excavation
204
Table 1 Stage construction of the excavation Wall deformation (mm) Wall Deformation (mm)
-20 0 20 40 -20 0 20 40
Elapsed 0 0
Stages Construction sequences 1st exc
days 1st exc
1 Diaphragm wall installation 101 5 2nd exc 5
2 1st excavations to the depth of GL. −1.5m 18 2nd exc
3 Deck slab installation at ground level (slab 20 10 3rd exc
10
thickness, t=0.4 m) 3rd exc
Depth (m)
Depth (m)
4 2nd excavation to the depth of GL. −5.08m 30
5 Top slab construction at GL. −4.18m (slab 14 15 7.8 mm
15
18 mm
thickness, t=0.8 m)
6 22
th
5 exc 20
3rd excavation to the depth of GL. −11.18m 20
7 Middle slab construction at GL. −10.48m 20 5th exc
(slab thickness, t=0.4 m) 25 25
8 4th excavation to the depth of GL. −17.53m 28
9 28 Toe movement = 10.2 m
5th excavation to the depth of GL. −18.93m 30 30
10 Bottom slab construction at GL. −18.13m 21 (a) (b)
(slab thickness, t=1.0 m)
Figure 3 Monitoring data of wall deformation DW-15 (a) before
2.2 In situ observation and instrumentation correction; (b) after correction
Several monitoring instruments were installed around the excavation
site, including inclinometers, surface settlement points, observation Wall deformation (mm) Wall deformation (mm)
wells, rebar stress transducers, and kingpost strain gauges to monitor
-20 0 20 40 -20 0 20 40
wall displacement during construction. The observation well 0 0
summarized the groundwater level generally located at depth 2 m
beneath the ground inside the excavation zone. Then, all monitoring st
1 exc 1st exc
data were carefully monitored to obtain the reliable and representative 5 5
2nd exc
data were chosen for this study. However, only the representative
2nd exc 3rd exc
measurement from inclinometer would be presented. The
10 3rd exc 10
inclinometers were inserted on the left and right sides of the
Depth (m)
Depth (m)
diaphragm wall. Figure 2 shows the locations of the inclinometers in

the excavation area used for further discussion and analysis. 15 10 mm 15 13.54 mm
Some methods have been suggested for the correction of
inclinometer readings to overcome wall toe movement. Inclinometer
readings can be corrected by referring to the lateral movement at the 20 20 5th exc
5th exc
B1F level during each excavation process as suggested by Hwang and
Moh. (2007) & Hsiung and Hwang. (2009). This method considers 25 25
the increment of the inclinometer reading at the specific point (i.e.,
the B1F level) for each excavation phase has a positive value, Toe movement = 3.54 m
showing that the wall at the certain point regularly moves forward 30 30
(toward the excavation site). (a) (b)
Figure 4 Monitoring data of wall deformation DW-171 (a) before

correction; (b) after correction
2.3 Soil exploration

2.3.1 In situ and laboratory soil tests
Geotechnical characteristics of Central Jakarta soil data was obtained
in underground section CP106 included in situ and laboratory tests.
The site exploration work included 10 Boreholes (BHs), Standard
Penetration Tests (SPT) and shear wave velocity (S-Wave velocity)
located nearby each Section of the excavation as shown in Figure 7.
Profile of soil characterization in Central Jakarta is displayed in
Figure 6.
Figure 2 Plan view of the excavation (Hsiung et al. 2018) The soil in Central Jakarta is mainly dominated by clay/silt
consisted of soft clay to very stiff clay as shown in Figure 6 (a) and
The inclinometer readings may show a negative value if the (b). The soil is simplified as 2 layers which are upper soil layer and
inclinometer is embedded in the wall toe movement moves toward lower soil layer as shown in Table 2 and 3. Upper soil layer is from
the retained soil side. Under those conditions at each excavation the ground surface to 7.95 m deep beneath the subsurface and
phase, all wall deflection reading curve must be corrected in parallel continued by lower soil layer to hard soil layer at a depth of 40 m.
so that the wall deflection curve at the certain point can return to the Moreover, the soil unit weight is divided into two, 14 kN/m3 and 18
same magnitude at previous excavation phase. Figure 3 and 4 present kN/m3 as detailed in Table 2 and 3.
field measurement of lateral wall deformation before and after Figure 6 (d) displays the relationships of the natural water content
correction. (ω) and the Atterberg Limit test results corresponding to depth. The
natural water content was in the value range of 34.6–89.3% and was
205
close or higher than the plastic limit, with a liquidity index (LI) of SPT-N S- wave velocity (m/s)
0.19–0.63. A gradual decrease in the LI relating to depth was also 0 20 40 60 0 300 600
observed in this excavation, suggesting that shear strength may rise 0 0
with depth. A high void ratio was connecting with the high water BH-11
content as shown in Figure 6 (c), especially for the soil layer Soft BH-12
5 5
occupying top 10 m. BH-13 BH-15
Permeability is one of the most significant index properties of soil Very stiff
in the advanced simulation of deep excavation. In situ permeability 10 10
tests, by the falling head method, have been performed in selected
boreholes (BH15-17). After firstly observing the initial level of the Hard
groundwater table, then the casing is loaded with water until 15 15
achieving the top of the casing pipe. The water drop-down is
Depth (m)
Depth (m)
measured at particular time intervals until attaining the stable or up to
the first initial water level. The permeability test results represent that 20 20
Stiff
the hydraulic conductivity of the in-situ soil in the research Section is
in the range of 3×10−8 to 7×10−8 m/s. Hard
25 25
Average
Fine Content (%) value
Sand Content (%)
30 30
0 50 100 0 50 100 Very stiff
0 0
35 35
5 5
10 10 40 40
(a) (b)
15 15
Depth (m)
Depth (m)
Figure 7 In-situ test information (a) SPT-N

20 20
The SPT-N value that was used in this study was the average
value of BH-11, BH-12, and BH-13 due to close to the location of the
25 25 excavation site. The N values remain lower than 10 to the depth of
approximately 7 m below the surface and become stiffer reaching 50
30 30 to the depth of 13 m then consistently in the average of 30 from 20 to
40 m shown in Figure 7 (a). Moreover, the S-wave velocity curve line
BH-11
35 35 of BH15 looks similar the SPT-N curve value in the range from 136.4
BH-12
m/s to 355 m/s displayed in Figure 7 (b). Those values support the
BH-13
40 40 classification of the soil layer into upper soil layer and lower soil
(a) (b) layer.
2.3.2 Soil modulus and shear strength properties

Void Ratio PL, ω, LL (%) The undrained shear strength (Su) of the clay was collected from
0 2 4 0 50 100 Triaxial CU tests and validated through several empirical methods for
0 0 estimating the Su of the clay. An empirical equation suggested by
Muir Wood. (1983) correlates Su with the LI. The empirical equation
5 suggested by Muir Wood. (1983) is as follows.
5
Su = 170 × e−4.6LI (kPa) (1)

10 10
Another empirical equation discussed in Hettiarachchi and

15 15 Brown. (2009) associates the relationship of Su with the SPT-N. The
empirical equation suggested by Kulhawy and Mayne. (1990) is as
Depth (m)
Depth (m)
20 20 follows.
25 25 Su = 6 × N (kPa) (2)
Figure 8 compares the Su profile calculated from LI (the green

30 30
shaded area), the SPT-N (red dash line), and the Triaxial UC test
results (hollow circles). The Su values obtained from these three
35 35 methods display a similar trend. For depths of 15–30 m, the Su value
increased linearly with depth and could be approximated as 0.32 𝜎′𝑣
40 40 (where 𝜎′𝑣 is the effective overburden pressure). For simplicity, the
(c) (d) Su of the soil for the numerical simulation was determined using Eq.
(2) and the results are listed in Table 2. Regarding effective soil shear
Figure 6 Profiles of borehole log information (a) Sand content; (b) strength of clay, few high-quality, and reliable CU tests were
Fine content; (c) Void ratio; (d) water content, plastic limit, and delivered. Therefore, the effective friction angle of the soil that was
liquid limit (Hsiung et al. 2018) selected in this research was simplified to upper layer 38o and lower
layer 41o of effective friction angle as displayed in Table 3 for
206
Undrained A purposes. The K0 values were defined by using Soil modulus, E50 (kPa)
equations from Jaky. (1944) for normally consolidated soils and from 0 200000 400000
Mayne and Kulhawy. (1982) for overconsolidated soils, as shown in 0
Interpreted from PMT
Eq. (3).
CU test
2800N
K0(NC) = 1−sin ϕ' (3a) 5
4000N
K0(OC) = (1−sin ϕ') OCR (sin ϕ') (3b)

10 E50 = 2800~4000N
Where K0(NC) and K0(OC) are the at-rest earth pressure
coefficient for normally consolidated and overconsolidated soils, 15
respectively, ϕ' is the effective friction angle, and OCR is the
overconsolidation ratio. The investigated clay reported that the soil is
normally-consolidated clay to slightly over-consolidated clay.
Depth (m)
20
Therefore, the value of K0(NC) in Plaxis 3D follows the default
setting by Jaky. (1994) equation.
25
The soil modulus measured from PMT was expected to be close to
the initial soil modulus (Ei). For comparison, the measurements were
converted to the soil modulus at 50% stress level (E50) by using Eq.
30 E50 = 833 (z+76)
(4), assuming a hyperbolic curve for the stress-strain relationship.
2−𝑅𝑓
𝐸50 =𝐸𝑖 (4) 35
2
Where Ei is the initial soil modulus and Rf is the failure ratio,

40
which is assumed to be 0.9 for clay. Figure 9 presents a comparison
of E50 versus depth with various approaches and tests. As suggested
by Hsiung. (2009) and Yong. (2015), the soil modulus for clay is E =
45
4000 N (kPa). Moreover, the Architectural Institute of Japan. (2001)
suggested that E = 2800 N (kPa) can be applied to all soils. The 𝐸50 Figure 9 Undrained shear strength with depth
values determined from PMT appear located in between of those (Hsiung et al. 2018)
empirical equation values and become linearly associated with depth.
The linear regression line for 𝐸50 obtained from PMTs is expressed Z indicates the soil depth in meters in Figure 9. The E50 values
as follows. obtained from the CU tests tend to be at the lower bound of the E 50
𝐸50 = 833 (z + 76) (kPa) (5) values compared to those determined using PMT and estimated from
SPT-N (Figure 9). The low E50 values may have been attributable to
the sample disturbance and the quality of the soil sample. Table 2 and
3 summarize the E50 values determined from PMT as shown in Eq.
Su (kN/m2)
(5) as a soil parameter to be used in Undrained A and Undrained B
0 50 100 150 200 250 for Hardening Soil and Mohr-Coulomb.
0
UC test 3. FINITE ELEMENT METHOD
Kulhawy & Mayne 3.1 Numerical analysis
5
(1990)
Su = 6N (kPa) A three-dimensional finite element (FE) analysis or benchmark
analysis, was conducted to simulate the performance of the deep
10 excavation in this research. The FE software Plaxis 3D was chosen as
Su = 6N (kPa) a numerical tool for the 3D simulations. Figure 11 displays the 3D FE
Muir Wood (1983) model of the benchmark analysis. The dimension of the finite element
model was 182 m×100 m×40 m. A half of excavation area with a
Depth (m)
Su = 170×e-4.6LI (kPa)
15
length of 40 m was performed by symmetric model to represent
the excavation case (E-E) which is located in the plane-strain
condition of the excavation. This was because the influence of corner
20 effects becomes inconsiderable once the distance is more than 30 m
Kulhawy & Mayne (1990) from the corner. At the distance of 30 m away from the corner, the
Su = 6N (kPa) wall is technically under plane-strain conditions and the wall
25
≈ 0.32'v deflection remains almost the same (Ou. 2006, Ou et al. 1996, and
Hsiung et al. 2016). Thus, the total length of the excavation in the
design was 80 m. Moreover, shortening the excavation length can
reduce the dimension of the 3D model, saving computational time and
30 cost without affecting the numerical results.
Ten construction phases were performed with the total of 5
excavations and 4 slabs as listed in Table 1. The groundwater table
35 was initial set 2.0 m below the ground surface and lowered to 1.0 m
below excavation surface at each excavation phase. The distance from
Figure 8 Undrained shear strength with depth the lateral boundaries of the model to the diaphragm wall in the
(Hsiung et al. 2018) excavation was 80 m, which was about four times the excavation
207
Table 2 Input parameter of Mohr-Coulomb using Undrained A and B analysis
Depth 𝜸 c' Su 𝑬𝟓𝟎 𝑬′𝟓𝟎 E'

Consistency ϕ' N
(m) (kN/m3) (kPa) (kPa) (kPa) (kPa) (kPa)
0-7.95 Soft 14 6 38 4 24 66619 57759 86638
7.95-9.95 Very stiff 18 46 41 21 126 70763 61352 92028
9.95-19 Hard 18 46 41 39 234 75366 65342 98013
19-22.5 Stiff 18 46 41 12 72 80593 69874 104811
22.5-24.5 Hard 18 46 41 34 204 82884 71860 107790
24.5-38.95 Very Stiff 18 46 41 22 132 90172 78179 117269
Note: m = 1.0; Rf = 0.9; v' = 0.3; vur = 0.2; 𝐸′50 = 0.867 𝐸50 and 𝐸′ = 1.5 𝐸′50
depth. Standard fixed conditions were used to the FE model; Two undrained functions, namely Undrained (A) and Undrained
horizontal movement was fixed at the lateral boundaries, and both (B), were specified in Plaxis 3D: Undrained (A) requires effective
horizontal and vertical movements were fixed at the bottom boundary stress parameters for both soil modulus and shear strength, whereas
of the model. The soil model, structural elements (diaphragm walls Undrained (B) is performed using the effective soil modulus and
and floor slabs), and soil-structure interface elements are discussed in undrained soil shear strength.
the following Section. The explanation of the undrained soil shear strength was
discussed in Section 2.3.2. The numerical analyses results were
3.2 Input parameter performed using PMT as shown in Eq. (5) were compared with field
measurement wall deformation
3.2.1 Soil parameter
As noted, in the Plaxis 3D manual, when selecting Undrained B,
Mostly, previous researchers in Indonesia often used plasticity index the soil modulus in HS model becomes stress-independent. It is
(PI) correlations to obtain the effective friction angle for Central entirely different with Undrained A which needs to convert the value
Jakarta clay as shown in Figure 10 that was conducted by Bjearum 𝑟𝑒𝑓
of the drained reference soil modulus (𝐸50 ) by Eq. (7).
and Simons. (1960). This concept is applicable for soft clay, but
misleading with the Central Jakarta clay which possibly gives the 𝑟𝑒𝑓 𝑐′ 𝑐𝑜𝑠 𝜙' + 𝜎 ′ 3 𝑠𝑖𝑛 𝜙'
𝑚
underestimated value of strength parameter. Moreover, establishing 𝐸′50 = 𝐸50 ( ) (7)
𝑐′ 𝑐𝑜𝑠 𝜙' + 𝑃𝑟𝑒𝑓 𝑠𝑖𝑛 ϕ'
the correlation value of soil stiffness corresponding to the depth has
to be appropriate to achieve the better result compared with field 𝑟𝑒𝑓
Therefore, E′50 can be used directly as 𝐸50 in the Undrained B
measurement.
model because the value of effective soil strength becomes
In this research, Hardening Soil (HS) result was compared with
meaningless. However, the stress dependency of the soil modulus was
Mohr-Coulomb to evaluate the effective soil strength used in
still modeled by manually inputting various soil modulus values for
Undrained A with previous research conducted by Hsiung et al.
soil layers to consider the change of the soil modulus with depth.
(2018) using Undrained B performed by Plaxis 3D. Table 3 shows the 𝑟𝑒𝑓
input soil parameters in the HS model for the FE analyses. The soil Table 3 lists the 𝐸50 values used in the simulation. As known in the
strength is divided into two layers. The upper layer performed with 6 Mohr-Coulomb soil model which only has a single soil modulus value
kPa of effective cohesion and 38o of effective friction angle obtained that can be estimated to be 𝐸′50 = 1.5 E′ in this model. The HS model
from high-quality of CU test. The lower layer used 46 kPa of effective allows to input separate soil modulus values to distinguish the soil
cohesion and 41o of effective friction angle. The effective stress behavior under loading and unloading conditions. According to Lim
analysis under undrained conditions was simulated to model the et al. (2010) and Calvello and Finno. (2004), the reference modulus
short-term undrained conditions of the clay during the excavation for unloading/reloading and oedometer loading were estimated to be
𝑟𝑒𝑓 𝑟𝑒𝑓 𝑟𝑒𝑓 𝑟𝑒𝑓
process. The assumption of soil undrained conditions can be selected 𝐸𝑢𝑟 = 3𝐸50 and 𝐸𝑜𝑒𝑑 = 0.7𝐸50 . Similar procedures were suggested
by low soil permeability, as described previously. Moreover, the real by Surarak et al. (2012) to determine input soil parameters in the HS
construction revealed that no pumping required, which is suggested model. Moreover, the sensitivity of each parameter on the wall
that the soil in the research remained soil to undrained conditions. deformation of an excavation was discussed by Gebreselasse and
Kemfert. (2005).
3.2.2 Structural properties and interface elements

The retaining wall was 1.0 m thick and 24.2 m deep as built in the
structure design. Corresponding to the American Concrete Institute,
Young’s modulus of concrete (Ec) can be obtained as follows:
E = 4700 fc′ (MPa) (8)
Where f′c (MPa) is the standard compressive strength of the

concrete. Considering the overlapping of unit weight and volume
between soils and other materials (e.g., concrete and steel), the soil
unit weight should be reduced in both of concrete and steel materials.
Ou. (2006) suggested the stiffness (EI) of a diaphragm wall is usually
Figure 10 Relationship between plasticity index and effective reduced by approximately 20–40% to consider defects and cracks in
friction angle (Bjearum and Simons 1960) the concrete. Therefore, 70% of wall stiffness was applied.
208
Table 3 Input parameter of Hardening Soil using Undrained A analysis
𝒓𝒆𝒇 𝒓𝒆𝒇 𝒓𝒆𝒇

Depth 𝜸 c' 𝑬𝟓𝟎 𝑬′𝟓𝟎 𝑬𝟓𝟎 𝑬𝒐𝒆𝒅 𝑬𝒖𝒓
Consistency ϕ' N
(m) (kN/m3) (kPa) (kPa) (kPa) (kPa) (kPa) (kPa)
0-7.95 Soft 14 6 38 4 66619 57759 287650 201355 862949
7.95-9.95 Very stiff 18 46 41 21 70763 61352 126885 88820 380655
9.95-19 Hard 18 46 41 39 75366 65342 111639 78147 334917
19-22.5 Stiff 18 46 41 12 80592 69874 99693 69785 299079
22.5-24.5 Hard 18 46 41 34 82883 71860 95616 66931 286848
24.5-40 Very Stiff 18 46 41 22 90172 78179 85654 59958 256963
Note: m = 1.0; Rf = 0.9; v' = 0.3; vur = 0.2 and 𝐸′50 = 0.867 𝐸50
Figure 11 3D Finite element model of benchmark analysis
Table 4 shows the input material properties of the diaphragm wall Plaxis 3D has the interface menu to design the plate elements.
used in Plaxis 3D. According to Ou. (2006), the axial stiffness of floor Interface elements were used to model the correlation between the
slabs in the top-down construction method must also be reduced by soil and structural elements such as the diaphragm wall and bottom
20% of the total stiffness considering the construction process. This concrete slab. The value of the interface reduction factor (Rinter)
is because the compressive strength of the constructed concrete might influences both the stiffness and strength of the interface. According
differ from the design strength during the construction; defects and to Ou. (2006), Rinter = 0.67 is a typical value for the interface between
cracks in the concrete can also be considered through this approach. clay and concrete interface.
The slab thickness for each floor level and the parameters for the slabs
are listed in Table 5.
4. RESULTS AND DISCUSSIONS
Table 4 Input parameter of diaphragm wall 4.1 Comparison of prediction and field measurement
Figure 12 describes the comparisons of the measured and predicted
Parameter Symbol Value Unit wall from several soil constitutive models included HS Undrained A,
Compressive strength of concrete f'c 21 Mpa HS Undrained B, MC Undrained A and MC Undrained B. Central
Thickness d 1 m Jakarta excavation was selected to model the simulation and it is
located in the plane-strain condition as shown in Figure 3.
Young's modulus E 21700 Mpa
As displayed in Figure 12, the wall deformation kept raising
Young's modulus 70% 70%E 15200 Mpa corresponding to the excavation phase until reaching the maximum
Unit weight 𝛾 6 (kN/m3) value in the last stage of the excavation. The comparison indicates
that the input parameter used performed well to all soil constitutive
Poisson’s ratio v 0.15 - models. However, the 3rd excavation shown in Figure 12 (c) describes
that the numerical analysis was overestimated compared with field
Table 5 Input parameter of concrete slabs measurement data. Numerical analysis results in 5th excavation as
displayed Figure 12 (d) are located between the field measurement
Slabs d (m) v 80%E (MPa) data both sides (DW-15 and DW-171).
The predicted wall deformation for soil near wall toe of HS results
Deck slab 0.4 0.15 17400 was lesser than MC in the below of excavation zone because the
Top slab 0.8 0.15 17400 stiffness moduli in the HS could take the elastic unloading process
𝑟𝑒𝑓
Middle slab 0.4 0.15 17400 due to the excavation process by 𝐸𝑢𝑟 together with comparatively
Bottom slab 1 0.15 17400
short embedded wall depth. Moreover, the simulation results of
209
Wall displacement (mm) Wall displacement (mm)

0 10 20 30 0 10 20 30
0 0
1st excavation 2nd excavation
Top slab
GL. -1.5 m
5 5
GL. -5.28 m
(a) (b)
10 10
Measurement
Depth (m)
Depth (m)
DW-15
15 15
Measurement
DW-171
HS
20 Undrained A 20
HS
Undrained B
25 MC 25 (c) (d)
Undrained A
MC
Undrained B Figure 13 The shading of relative stress at different excavation
30 30 stages from MC Undrained A: (a) 1st stage; (b) 2nd stage; (c) 3rd
(a) (b) stage; (d) 5th (final excavation) stage
Wall displacement (mm) Wall displacement (mm) 4.2 Stress path

0 10 20 30 0 10 20 30 The soil effective stress path (ESP) was conducted to learn the
0 0
3 excavation
rd 5 excavation
th behavior of soil nearby the excavation location. The location of ESP
Top slab Top slab was pointed in several locations namely A, B, C, D, and E. Each point
of the effective stress path was investigated at different excavation
5 5
stage to observe the p'-q value. The Mohr failure envelope can be
modified into p'-q diagram called Kf line which is a tangent line of the
Middle slab Middle slab
10 10 Mohr circles which as shown in Eq. 10.
Technically, the purpose of observing ESP is to study the
influence of unloading process during excavation based on numerical
Depth (m)
Depth (m)
GL. -11.38 m
15 15 analyses at every stage toward the soil strength. Eq. 8 and 9 can be
Bottom slab rewritten regarding of the effective stress. Moreover, Eq. 10 can
express the relationship between the Kf line and the Mohr failure
20 20 envelope:
GL. -19.03 m
(𝜎 ′ 𝑉 +𝜎 ′ 𝐻 )
p' = (kPa) (8)
25 25 2
(σ′ V −σ′ H )
q= (kPa) (9)
30 30 2
(c) (d)
sin ϕ' = tan ψ (10)
Figure 12 Comparison of predicted and measured wall All those ESP points were taken at every excavation depth and the
displacement from several soil constitutive models at the excavation toe of diaphragm wall as shown in Figure 14. Each point was
at various excavation stages: (a) 1st stage; (b) 2nd stage; (c) 3rd stage; observed for every stage of excavation. Points E and F were observed
(d) 5th (final excavation) stage at the end of the embedded wall to represent the effective stress path
because in the last excavation process, shown in Figure 13 (d),
Undrained A for both HS and MC recorded the consistent result and illustrates the highest stress occurred.
located in the between of measurement data which mean the effective The stress path values are corresponding to the location of
soil strength and total stress parameter are similar and further excavation depth and stress path points. The Kf line was obtained by
discussion will be carried out in the later section. the high-quality data of effective strength. When the stress path points
Figure 13 shows the shading of relative shear stress for each are above the excavation depth, then the value of overburden stress
excavation process. The relative shear stress is the ratio of the would increase and would be in reverse when the stress path points
mobilized shear strength and the maximum shear strength which under the excavation depth as shown in Points A, B, and C. Those are
illustrates the proximity of the stress point to failure envelope. The due to the horizontal stresses 𝜎𝐻 in the retaining zone decreases when
figure used MC Undrained A to see the evolution of relative shear the soils are excavated above the stress points and 𝜎𝐻 should increase
stress in different phases. The shading of relative stress at 1st stage when the soils are excavated below the stress path points. Moreover,
represented if the wall had no experience of high ground pressure in the point E shows that the overburden pressure increases
against the wall as shown in Figure 13 (a). The transformation corresponding to the excavation stages due to unloading process.
occurred in Figure 13 (b) and (c) which the wall has taken on the soil Figure 15 indicates that those values, based on numerical analysis
pressure. Eventually, in Figure 13 (d) indicates that the soil elements results, were located below the Kf line and far enough to reach the Kf
around the excavation site and the embedded length of diaphragm line which indicates that the models were in the safe zone and
wall in plane-strain location became critical in the design of the appropriate for the structure design. This indicates that by having a
numerical analysis. high-quality data of effective strength parameter and selecting
stiffness moduli are necessary to obtain the consistent simulation
210
results of Undrained A and Undrained B, respectively. However, clay during the excavation process. The following conclusions were
unless having a good quality test result, relative loading between soil illustrated by the results of this research:
stress status and failure envelope may not be the same which could
lead totally different results from Undrained A and Undrained B 1. The ground condition in Central Jakarta generally dominated by
analyses. Then, HS model with Undrained B analysis still can give clay/silt which consisted of soft clay to very stiff clay. The natural
acceptable results and it means as long as a set of proper soil stiffness water content of the clay in the range of 34.6 to 89.3% with LI of
can be given, HS model with total stress parameter can give a between 0.19 and 0.65. The SPT-N values were lower than 10 to a
reasonable result. depth of 7 m (soft clay) and continuously in the average of 30 at
depth 20 m (stiff clay).
2. The soil model was separated into upper soil layer and lower soil
layer to simplified the input soil parameter. The upper layer
performed with 6 kPa of effective cohesion and 38o of effective
friction angle. Moreover, the lower layer used 46 kPa of effective
cohesion and 41o of effective friction angle.
3. For Central Jakarta clay, a linear relationship between the soil
modulus (from PMT) and depth was suggested to use E 50 = 833
A (z+76), were measured in kilopascals and meters. This empirical
equation results were located in between of soil moduli obtained
based on 2400N and 4000N.
4. The results of numerical analyses were compared between
predicted and measured. The results indicate that the effective soil
strength parameter used in Undrained A was reliable to be used in
B the analysis because the wall deformation results of predicted,
located close to both sides of measurement data and Undrained B,
were consistently good in numerical performance.
5. The stress path was conducted to confirm the result of diaphragm
wall in the excavation site. The effective stress path points based
on numerical results were located below the Kf line indicate that
Undrained A and Undrained B could be performed equally well to
C be used.
Eventually, the effective (Undrained A) and total (Undrained B)

D ∗ E shear strength properties results show that they could be performed
and selected equally well to design the deep excavation in Central
Jakarta. However, considering the high-quality data of effective soil
strength and selecting soil Young’s modulus are fundamental to
obtain the rational result.
Figure 14 The location of selected stress points 5. REFERENCES

Architectural Institute of Japan, 2001. Recommendations of Design
of Building Foundation. Japan (in Japanese).
150
A Calvello, N., Finno, R., 2004. Selecting parameters to optimize in
B model calibration by inverse analysis. Comput. Geotechn. 31,
100
𝐾𝑓 line 5th excavation 410–424.
C
5th excavation Gebreselessie, H., Kempfert, G., 2005. Sensitive study of the
D 1st excavation (D)
3rd excavation
hardening soil model parameters based on idealized
𝐾0 line
q (kPa)
50 E excavation. In: Proceedings of 11th International conference on

1st excavation (E) computer methods and advances in geomechanics, Torino,
5th excavation
1 excavation
st Italy, pp. 321–328.
0 2nd excavation Hettiarachchi, H., Brown, T., 2009. Use of SPT blow counts to
5th excavation 5th excavation estimate shear strength properties of soils: energy balance
approach. J. Geotechn. Geoenviron. Eng. 135 (6), 830–834.
-50 Houston, W.N., Mitchell, J.K., 1969. Property interrelationships in
0 25 50 75 100 125 150 175 sensitive clays. ASCE. J. Soil Mech. Found. Divis. 95 (4),
p' (kPa) 1037–1062.
Figure 15 The effective stress path at the excavation area Hsieh, P.G., Ou, C.Y., Lin, Y.K., Lu, F.C., 2015. Lessons learned in
design of an excavation with the installation of buttress walls.
4. CONCLUSIONS J. GeoEng. 10, 67–73.
Hsiung, B.C.B., Hwang, R.N., 2009. Evaluating Performance of
Diaphragm Walls by Wall Deflection Path. SEAGS. Special
This paper presents a case study and numerical simulations for a
Issue on Excavation and Tunneling in Geotechnical
large-scale deep excavation to obtain the reliable total and effective
Engineering, pp. 81–90.
soil strength based on the comparison of wall deformation between
Hsiung, B.C.B., Wang, C.L., Lin, H.T., Chen, C.H., 2013. Design and
predicted and measured results in Central Jakarta excavation, Central
performance of a large scale excavation adjacent to sensitive
Jakarta, Indonesia. Various soil constitutive models were established
structures in urban area. In: Proceedings of the 2nd
in numerical simulations included Hardening Soil Undrained A and
International Conference on Geotechnics for Sustainable
B and Mohr-Coulomb Undrained A and B. Thereafter, the effective
Development-Geotec. Hanoi 2013, Hanoi, Vietnam.
stress path was conducted to check the behavior of Central Jakarta
211
Hsiung, B.C.B., Yang, K.H., Aila, W., Hung, C., 2016. Three-
dimensional effects of a deep excavation on wall deflections in
loose to medium dense sands. Comput. Geotechn. 80, 138–151.
Hsiung, B.C.B., Yang, K.H., Aila, W., Ge, L., 2018. Evaluation of
the wall deflection of a deep excavation in Central Jakarta using
three-dimensional modeling. Tunnelling and Underground
Space Technology. 72 (2018) 84-96.
Hwang, R.N., Moh, Z.C., 2007. Deflection paths and reference
envelopes for diaphragm walls in the Taipei Basin. J. GeoEng.
1, 1–12.
Jaky, J., 1944. The coefficient of earth pressure at rest. J. Soc.
Hungarian Archit. Eng. Budapest Hungry 355–358.
Kulhawy, F.H., Mayne, P.W., 1990. Manual on Estimating Soil
Properties for Foundation Design. Electric Power Research
Institute, Palo Alto, California.
Lim, A., Ou, C.Y., Hsieh, P.G., 2010. Evaluation of clay constitutive
models for analysis of deep excavation under undrained
conditions. J. GeoEng. 5, 9–20.
Lim, A., Ou, C.Y., 2017. Stress paths in deep excavations under
undrained conditions and its influence on deformation analysis.
Tunnelling and Underground Space Technology. 63 (2017)
118-132.
Likitlersuang, S., Surarak, C., Wanatowski, D., Oh, E.,
Balasubramaniam, A., 2013. Finite element analysis of a deep
excavation: A case study from the Bangkok MRT. Soils
Foundat. 53 (5), 756–773.
Mayne, P., Kulhawy, F.H., 1982. K0-OCR relationships in soil. J.
Geotechn. Eng. Divis. 108 (GT6), 851–872.
Moh, Z.C., Hwang, R.N., 2005. Geotechnical considerations in the
design and construction of subways in urban areas. Seminar on
recent developments on mitigation of natural disasters, urban
transportation and construction industry, Jakarta, Indonesia.
Muir-Wood, A., 1983. Index properties and critical state soil
mechanics. Paper presented at the Proceeding of the
Symposium on Recent Developments in Laboratory and Field
Tests and Analysis of Geotechnical Problems, Bangkok,
Thailand.
Ou, C.Y., Chiou, D.C., Wu, T.S., 1996. Three-dimensional finite
element analysis of deep excavations. J. Geotechn. Eng. 122
(5), 337–345.
Ou, C.Y., Shiau, B.Y., Wang, I.W., 2000. Three-dimensional
deformation behavior of the Taipei national enterprise center
(TNEC) excavation case history. Canad. Geotechn. J. 37 (2),
438–448.
Ou, C.Y., 2006. Deep Excavation: Theory and Practice. Taylor &
Francis, Netherlands.
Schweiger, H.F., 2009. Influence of constitutive model and EC7
design approach in FEM analysis of deep excavations. In:
Proceedings of ISSMGE International Seminar on Deep
Excavations and Retaining Structures, Budapest, Hungary, pp.
99–114.
Surarak, C., Likitlersuang, S., Wanatowski, D., Balasubramaniam,
A., Oh, E., Guan, H., 2012. Stiffness and strength parameters
for hardening soil model of soil and stiff Bangkok clays. Soils
Foundat. 52 (4), 682–697.
Wang, J.H., Xu, Z.H., Wang, W.D., 2010. Wall and ground
movements due to deep excavations in Shanghai soft soils. J.
Geotechn. Geoenviron. Eng. 136, 985–994.
Yang, K.H., Uzuoka, R., Thuo, J.N., Lin, G.L., Nakai, Y., 2017.
Coupled hydro-mechanical analysis of two unstable
unsaturated slopes subject to rainfall infiltration. Engineering
Geology. 216 (2017) 13-30.
Yong, K.Y., 2015. Learning lessons from the construction of
Singapore Downtown line (DTL). In: Proceedings of
International Conference and Exhibition on Tunneling and
Underground Space, Kuala Lumpur, Malaysia.
212
Plane Strain Ratio and Waling Size Evaluation of Deep Excavation of Kuala Lumpur
Using 3D Finite Element Analysis
Jen Shen Ang1*, Bin Chen Benson Hsiung1, Ching Hung2, A. J. Li3, J. W. Chen2
1
Department of Civil Engineering, National Kaohsiung University of Science and Technology, Kaohsiung, Taiwan
2
Department of Civil Engineering, National Cheng- Kung University, Tainan, Taiwan
3
Department of Construction Engineering, National Taiwan University of Science and Technology, Taipei, Taiwan
E-mail: 1102102171@gm.kuas.edu.tw
ABSTRACT: This paper presents a study based on 3D Finite Element Analysis of deep excavations in Kuala Lumpur, Malaysia. The
ground in Kuala Lumpur mainly consists of a layer of highly permeable sand, occasionally with some thin layers of clay above rock which
includes interbedded sandstone, siltstone, shale and mudstone for so called “Kenny Hill Formation” or limestone with sinkholes for so called
“KL limestone formation”. A common feature is the depth of rock varies in a widely range which leads to challenges of design and
construction of excavations. By using 3D finite element analysis together with observational data, it aims to explore plane strain ratio (PSR)
of deep excavation of Kuala Lumpur which can indicate impacts from the corner effect associated with distance to the corner. Influences on
PSR from certain factors, such as centre- to- centre distance of struts and hard soil stratum/rock depth are also covered and examined in this
study. Moreover, individual secant bored piles (SBP) wall have to be adopted for retaining structure of deep excavation in Kuala Lumpur as
hydraulic- grab type diaphragm wall can’t be constructed in very hard rock, such as limestone and thus steel waling has to be installed in
order to connect each SBP, not only providing the function of strut- wall connection. It is not possible to apply 2D analysis to examine
waling size in the aspect of function of SBP connection so 3D analysis has to be conducted instead for said purpose. The evaluation of
waling size is thus included in this study.
Keywords: Deep excavation, Kuala Lumpur, 3D finite element analysis, Plain Strain Ratio, SBP wall, walling size
1. INTRODUCTION width and length of the basement excavation are approximately 43

m x 78 m in plain view (Figure 1). The maximum depth of
Kuala Lumpur is the capital city of Malaysia. Due to the fast excavation is 13.2m. 20m deep of diaphragm wall is supported by
development urban area and limited urban public transport service one level of H-section steel struts. There are ten inclinometers in the
in Kuala Lumpur, additional underground space is required for both site and the location are as figure 1. Kenny Hill formation is residual
private and public uses in the city. The behaviour of walls induces soils which derived from weathered sedimentary rocks and
by deep excavation had been studied but limited studies have been commonly found in major part of Kuala Lumpur city centre (KLCC)
reported for excavations in Kuala Lumpur, Malaysia, especially in area. The sequence of Kenny Hill Formation along the alignment
the aspects of PSR (Plain strain ratio) and evaluation of waling comprises of interbedded to be the Upper Palaeozoic age. This
details where deeper basements for the use of both public and formation has undergone mild and regional metamorphic materials,
private sectors in the future. Furthermore, available and reliable soil such as siltstone and shales/mudstone overlain by stiff over-
data is limited and few of the large-scale deep excavations in the consolidated soils predominately of sandstone/siltstone to quartzite
city have been well documented. These are all obstacles to increase and schist/phyllite respectively. The ground condition in this study
the difficulty of studying deep excavation in Kuala Lumpur area. consisted of upper recent alluvium deposits about 8m depth which
3D wall behaviour has been studied by using 3D (three- underlain by residual soils and weathered rocks of Kenny Hill
dimensional) FE (Finite element) analysis. The concept of plane formation. The ground condition at the site consists of a 7.5 m thick
strain ratio (PSR) was first proposed by Ou et al. (2006) and it is the of silty sand layer underlain by residual soils of Kenny Hill
ratio of the maximum wall deflection at a section of a wall where formation. Upper recent alluvium layer is low SPT- N values
distance (d) from the corner to the maximum wall deflection at the Skeptom (1986), mainly consists of loose silty sand material. At the
section under plane strain conditions. The PSR was adopted in this deeper soil layer, the SPT- N value ranges from 15 to 200
study to validate the 3D wall behaviour of an excavation in drained blows/300mm and increase with depth (Figure 2). The excavation
material in KLCC (Kuala Lumpur City Centre) area. was completely embedded in drained material. The observation
One of the most commonly used additional supporting systems shows that the groundwater table at approximately 2.5m below
in conjunction with retaining walls for deep basement excavation is ground surface.
the steel strut- waler system (Chiew & Leow, 2006). The struts
usually consist of an H-section with walers laid across the walls to
ensure continuity. On the other hand, for strut-waler connection, the
webs of the strut and waler are in two different planes perpendicular
to each other. Strut-waler system was adopted in Kuala Lumpur due
to the construction of SBP wall. Since the primary bored pile has no
reinforcement so it can’t take much bending moment, it has to rely
on the waler to take bending moment and get better connection.
This paper presents two cases of deep excavation in Kuala
Lumpur. One of these two cases is named Case KLCC and the
ground is Kenny Hill Formation. The other case is located at Bukit
Bintang where the ground is Kuala Lumpur limestone formation
which named as Case Waler.
2. THE SITES AND PROJECTS BACKGROUND

The case history of excavation in Kenny Hill Formation (Case
Figure 1 Plan view of excavation shape and inclinometers
KLCC) which provided by Law et al. (2016). It is a three- level deep
arrangement of case KLCC (Law et al. 2016)
basement, the geometry of basement excavation is rectangular, the
213
machine) need to launch from this site, that is the reason there is one
section having a very large spacing of strut, which is 17m in order to
fit the whole TBM into the pit.
Figure 3 Development of karsts from i-geology

Figure 2 Excavation stages and soil profile in cross-section (Law et
al. 2016)
Case Waler is located at Bukit Bintang, the ground of this case is

Kuala Lumpur Limestone formation as described previously. Kuala
Lumpur Limestone Formation is composed of fine to coarse, white
to grey, predominantly recrystallized limestone and dolomites with
irregular level of rock below the alluvium and containing numerous
voids and solution channels. These features are consistent with
classification of Karst terrain according to Waltham & Fokes (2003).
Karst topography in limestone is formed by chemical dissolution
process when groundwater circulates through the limestone (figure
3), carbon dioxide from the atmosphere is fixed or converted into
the soil in aqueous state and combined with rainwater to form
carbonic acid, which can dissolve carbonate rocks. Simplified
subsurface profile of Kuala Lumpur Limestone formation. Case
Waler is one of the underground metro stations in Kuala Lumpur
and has to be excavated in Kuala Lumpur limestone formation
together with soil above. The maximum excavation depth of Case
Waler is 31.5m. There are 4 combination zones of the site, the
description of the zones as below are in plan view:
1. Zone 1: Trapezoidal shape with width of 23 m and the
zone is connecting with zone 4 and zone 2, the station end
is to enable TBM launching.
2. Zone 2: Rectangular shape with width of 23 m and the
zone is connecting between zone 1 and zone 3.
3. Zone 3: This zone is complex geometry and it connect
with zone 2. TBM launching will at the station ends with
23 m width and the entrance will locate beside the station
end.
4. Zone 4: This zone is one of the entrances of the
underground station with rectangular shape
Figure 4 shows the geometry of the case Waler excavation. The

pit was retained by SBP wall with the strut-waler supporting system,
including H-type steel props, ground anchors, H-type steel walers,
rock bolts. Curtain and base grouting were carried out to prevent
water ingress through rock fissures of sinkholes surrounding the
excavation area due to drawdown of groundwater. The embedded Figure 4 Geometry of case waler
depth of retaining wall of Case Waler is shallower than excavation
depth since the limestone was considered to be a very stiff material, 3. FINITE ELEMENT ANALYSIS METHOD
can be stable without any support system once the excavation
reaches the depth of rock. In this case, four TBM (Tunnel boring 3.1 Mesh Boundary of Numerical Modeling
214
The numerical modelling of Case KLCC is built associated with the of this model is used “normal” mesh setting. The model consists of
information which provided by Law et al. (2016), such as mesh 416,761 nodes and 284,581 ten-node tetrahedral elements.
boundary and geometry of deep excavation side. The groundwater
table is observed at approximately 2.5 m below the ground surface
and set the groundwater table into PLAXIS 3D. The mesh boundary
and phreatic level are shown as Figure 5. The total nodes in the 3D
model are 147,229 and 99,446 is the number of total ten-node
tetrahedral elements. The “normal” mesh is applied to the whole
model.
Figure 7 Mesh boundary of PLAXIS 3D in Case waler
Figure 5 Mesh boundary and phreatic level in PLAXIS 3D
Although the Law et al. (2016) has shown the arrangement of

lateral supporting system (strutting system) in horizontal and
diagonal, the exactly coordinate and spacing are not given.
Therefore, the strutting system is thus assumed to have a horizontal
spacing of 6 m in primary wall and 2 m spacing in complementary (a)
wall with perpendicular based on local engineering practice in
Malaysia and only one-layer strutting system in used Details of
strutting system in the model are illustrated in Figure 6.
(b)
Figure 8 Detail supporting system in PLAXIS 3D: (a) Plate element
(retaining wall); (b) Node to node element (struts)
Based on the interpreted subsoil parameters, the groundwater

table is approximately at 1 mbgl and set it in PLAXIS 3D for
simulation.
Figure 6 Dimension of retaining wall and struts spacing 3.2 INPUT PARAMETERS
For Case Waler, since it aims to evaluate whether the capacity of Hardening soil (HS) model is adopted to simulate the soil behaviors
waler is enough to take the stresses from retaining wall or not so the in Case KLCC. HS model is an advantage model for simulating the
geometry of whole excavation is not simulated by using PLAXIS behavior of different types of soil, both soft soils and stiff soils
3D in order to shorten time of generation of model and running time reported by Schanz (1999). Law et al. (2014) highlighted that the
of computer operation. A rectangular shape of 3D model is general problem in the analysis and design of deep excavations in
conducted to undertake the analysis. The mesh boundary of PLAXIS loose to medium dense sand with shallow meta-sedimentary hard
3D is length of 290 m (x), width of 200 m (y), depth of 104.5 m (z) layer which soil tests information is limited and low-quality
presented as Figure 7. The geometry of excavation is length of 120 sampling. Normally, the shear strength of soil can be obtained from
m and 30 m for width in plan view and the strut spacing of 6 m with laboratory triaxial tests and direct shear tests but not on the stiffness
an area of 17 m strut spacing (Figure 8). The purpose excavation of soil. Due to the reasons, one of the field tests which called
level is 31 mbgl and the toe of wall is located at 28 mbgl. The mesh Standard Penetration Test (SPT) is conducted to obtain the
215
information of subsoil for geotechnical design and analysis purpose. calibration by experience from previous study on similar ground
However, the empirical correlation between stiffness of soil and condition. Table 4 summarizes the soil parameters of Case Waler.
number of SPT (SPT-N) are used for prediction the ground HS model is used for upper alluvium soil, Mohr-Coulomb (MC)
movement induced by excavation. model is used for rock simulation. The secant Young’s modulus of
The constitutive soil model of HS model used for case history is soil is also taken with STP-N as below:
studied by Law et al. (2016), the performance between numerical
analysis and field observation reading are similarly, the empirical E50ref = 3000N (4)
correlation is suitable for simulating in the formation. Triaxial
secant modulus E50ref with SPT-N is taken as below: For this study, 6 of STP-N is selected.
E50ref = 2000N (1) Table 2 Diaphragm wall elastic properties (Plate element)
The ratios between E50ref, Eoedref and Eurref are as follows: Parameter Name Value Unit
Compressive strength of concrete fc’ 40 MPa
Thickness d 0.6 M
E50ref = Eoedref (2)
Young’s modulus E 28 x 106 kPa
Young’s modulus x 70% 70% E 19.6 x 106 kPa
E50ref = Eurref (3) Unit weight w 9 kN/m3
Poisson’s ratio v 0.2
The simple assumption as suggested by Schanz et al. (1999) and
Brinkgreve et al. (2012), the effective stress strength (c’ and ф’) and Table 3 Steel struts properties (Anchor element)
stiffness parameters adopted in the 2D numerical study are
summarized in Table 1. Based on Tan (2010), the ground properties Level Strut size Section EA 60%EA
of excavation in residual soils is fully drained condition. Therefore, Area (m2) (kN) (kN)
effective stress analysis (drained material) is performed in the 3D Level 1 H400 x 400 0.0219 4.483 x 106 2.6898 x 106
numerical back analysis.
Table 4 Input soil parameters of case CO
Table 1 Input soil parameters of case KLCC
(a) Sandy soil
Symbol Unit S1 S2 S3 S4 S5 S6
c’ kPa 1 5 5 8 15 50 Layer Depth Drainage Unit weight ϕ' E50ref
ϕ' o 28 30 31 33 35 35 (m) type (kN/m3) (kPa) (o )
(kPa)
E50ref MPa 15 30 45 75 150 225
Eoedref MPa 15 30 45 75 150 225 1 0.0-25 Drained 19 1 29 18 x 103
Eurref MPa 45 90 135 225 450 675
m - 0.5 0.5 0.5 0.5 0.5 0.5 (b) Limestone (Mohr- Coulomb model)
vur - 0.2 0.2 0.2 0.2 0.2 0.2
pref kPa 100 100 100 100 100 100 Depth Soil  c' ϕ' E'
Layer
Rinter - 0.67 0.67 0.67 0.67 0.67 0.67 (m) type (kN/m3) (kPa) (o ) (kPa)
2 25-104.5 Limestone 24 400 32 1 x 106
Linear elastic plate element is used to simulate the diaphragm
with 6-node. The Young’s modulus of plate element is assumed 19.6
GPa which 70% of the concrete compression stiffness of 28GPa. Three types of plate element are adopted in this study and the
Assuming the diaphragm wall is “wished-in-place”, which means input parameters of plate are summarized in Table 5. There are
this work does not consider the stress and ground movement during primary piles (a), secondary piles (b) and average piles (c). Equation
the construction diaphragm such as trench excavation and (x) for secondary piles and average piles, equation (x) for primary
concreting. Table 2 is the input parameters of plate element wall caused of no steel rebar in the pile. Corresponding to the
(diaphragm wall). A required input parameter is unit weight, due to British Standard Institute (BS 8110), Young’s modulus of concrete
the overlapping of unit weight and volume between soil element and can be obtained by:
plate element, so need to subtract the surrounding unit weight of soil
for the unit weight of plate element (diaphragm wall). The node to E = 20 + 0.2 x fc’ (MPa) (5)
node anchor is used for simulating the steel struts to support
diaphragm wall. The input parameter of anchor is EA, where E is 70% of Young’s modulus is used for input parameters. The primary
the Young’s modulus of steel struts, and A is the cross-sectional piles and secondary piles are modeled at the 17m spacing of struts.
area of steel struts. 60% of Young’s modulus of steel is adopted for Universal beam (UB) and universal column (UC) are selected
the input parameters. The parameters of node to node anchor are for strutting system in Case CO, the steel grade of S 355 for UB and
summarized in Table 3. As the plate element is assumed to be fully UC. Struts are also simulated by node to node anchor. Table 6 is the
connected to each other in the model, waler is thus not simulated in input parameters required by anchor element (strut).
this research.
According to Ang et al. (2017), only two layers of soils and Table 5 Various retaining wall properties
rocks are categorized of Case Waler, which is karst bedrock
(limestone) underlying by loose sand. For simulating the behavior of (a) Primary piles
soil and bedrock, two constitutive soil models were selected, HS Parameter Name Value Unit
model for the upper loose sand with drained and MC model for the Compressive strength of fc’ 40 MPa
limestone with undrained A. Advanced model is not eligible to be concrete
adopted due to limit of available and reliable site investigation data. Thickness d 0.88 M
The modeled soil profile is based on available borehole logs and site Young’s modulus E 28 x 106 kPa
survey information. Due to limited results of in-situ tests and the Young’s modulus x 70% 70% E 19.6 x 106 kPa
condition of soil during the time of testing and sampling quality may Unit weight w 9 kN/m3
be highly disturbed and the result by using empirical method to Poisson’s ratio v 0.2 -
216
(b) Secondary piles Similarly, bottom-up construction method is also adopted for the
Case Waler. Number of total excavation stages is 11 and the purpose
Parameter Name Value Unit excavation depth is 31.5 mbgl. There are 9-layers strut for
Compressive strength fc’ 40 MPa supporting the retaining wall. For strutting excavation, every depth
of concrete of excavation stage is 1 m below strut level and the excavation
Thickness d 1.48 M without any supporting system when the excavation starts from top
Young’s modulus E 28 x 106 kPa of bedrock.
Young’s modulus x 70% E 19.6 x 106 kPa The actual prestress of the struts applied is around 20% to 30%
70% of design strut load. Point load is used to simulate prestress of the
Unit weight w 9 kN/m3 struts at both ends of struts. Table 9 is the construction phase
Poisson’s ratio v 0.2 - description for Case Waler.
(c) Average piles Table 4.9 Construction phases of case CO
Parameter Name Value Unit Phase Construction
Compressive strength fc’ 40 MPa 0 Initial phase (stress regeneration)
of concrete 1 Installation of second bored piles wall
Thickness d 1.072 M 2 1st stage excavation to 3.5 mbgl, dewatering to 3.5 mbgl
Young’s modulus E 28 x 106 kPa 3 Installation of 1st layer of struts at 2.5 mgbl
Young’s modulus x 70% E 19.6 x 106 kPa 4 2nd stage excavation to 7 mbgl, dewatering to 7 mbgl
70% 5 Installation of 2nd layer of struts at 6 mgbl
Unit weight w 9 kN/m3 6 3rd stage excavation to 9 mbgl, dewatering to 9 mbgl
Poisson’s ratio v 0.2 - 7 Installation of 3rd layer of struts at 8 mgbl
8 4th stage excavation to 11 mbgl, dewatering to 11 mbgl
Table 6 Steel strut properties 9 Installation of 4th layer of struts at 10 mgbl
Level Strut size Section EA x 60%EA x 10 5th stage excavation to 14 mbgl, dewatering to 14 mbgl
Area 106 106 11 Installation of 5th layer of struts at 13 mgbl
(m2) (kN) (kN) 12 6th stage excavation to 16 mbgl, dewatering to 16 mbgl
13 Installation of 6th layer of struts at 15 mgbl
1 2-UB 610x324 0.106 21.73 13.038
14 7th stage excavation to 18 mbgl, dewatering to 18 mbgl
2 2-UB 610x324 0.106 21.73 13.038
15 Installation of 7th layer of struts at 17 mgbl
3 3-UB 610x324 0.159 32.56 19.557
16 8th stage excavation to 21.5 mbgl, dewatering to 21.5
4 3-UB 610x324 0.159 32.56 19.557
mbgl
5 3-UB 610x324 0.159 32.56 19.557
17 Installation of 8th layer of struts at 20.5 mgbl
6 3-UB 610x324 0.159 32.566 19.557
18 9th stage excavation to 23.5 mbgl, dewatering to 23.5
7 3-UB 610x324 0.159 32.56 19.557
mbgl
8 2-UB 610x324 0.106 21.73 13.038
19 Installation of 9th layer of struts at 22.5 mgbl
9 2-UB 610x324 0.106 21.73 13.038
20 10th stage excavation to 25 mbgl (bedrock), dewatering
to 25 mbgl
3.2 Computational sequences
21 Excavation to purpose depth (31.5 mbgl), dewatering to
The construction method of Case KLCC is bottom-up 31.5 mbgl
construction. Since the site only has one-layer of steel strut to
support the retaining wall, so the total of construction phases in 4 RESULTS AND DISCUSSIONS
PLAXIS 3D are 5 phases, the depth of first excavation stage is 0.5
From the Figure 9 is the comparison between the observation and
meter below the first struts layer, the simulation of phases is
numerical analysis of Case KLCC. The inclinometer name of IN-3
described at Table 7.
and IN-8 are installed at the approximately center of long walls and
the lateral wall movement is no significant difference between
Table 7 Construction phase for case KLCC
inclinometers readings and result of numerical model analysis due to
Phase Construction the cross-section might be in the plain strain condition. As
0 Initial phase (stress regeneration) illustrated in figure 9 (IN-1 and IN-10), modelling the wall
1 Installation of diaphragm wall underestimates the lateral wall movement possibly due to
2 1st stage excavation to 4.5 mbgl, dewatering to 5.5mbgl comparatively worse quality of vertical joint between the
3 Installation of 1st layer of struts rectangular panel and L-shaped panel.
4 Excavation to purpose depth (13.2mbgl), dewatering to The location of IN-5 and IN-6 are stood along the
14.2 mbgl complementary wall, the wall movement decreases while decreasing
the distance from the corner. Although IN-5 is at the centre of
For the preloading of steel struts is already created in the complementary wall, but the wall movement at IN-5 is smaller than
structure section as a point load instead of the prestress function of IN-3 and IN-8. The reason is IN-5 is located on a relatively short
PLAXIS 3D by using point element. In calculation analysis, the wall. Refer to the Figure 9, majority of inclinometers reading are in
calculation type of K0 procedure for initial phase and plastic good agreement with the prediction expect the ones near the corner
analysis for the rest of construction phases, the calculation of pore by reasons stated above.
pressure in all the construction phases are used phreatic. The For Case Waler, this is the Class B analysis which means to do
calculation types are described as below (PLAXIS): the analysis at the same time during occurrence of the event.
1. K0: Direct generation of initial effective stresses, pore Currently, the construction progress of the site is going to second
pressures and state parameters; excavation stage where excavate to 7 mbgl. This study only shows
2. Plastic: Elastoplastic drained or undrained analysis. the comparison of first excavation stage between inclinometer
Consolidation is not considered readings and numerical analysis (Figure 10). There is significant
3. Phreatic: Direct generation of steady-state pore pressures difference between the readings and result. The difference between
from phreatic level and cluster-related conditions. numerical analysis and inclinometer readings around 10 mm, the
217
reasons may be caused by the capping beam at the top of SBPW is

not simulated in the model or underestimation of the stiffness of
upper alluvium soil.
Figure 10 Comparison between observation and numerical analysis

of Case Waler
The total displacement shading is shown as Figure 11 (a), shear

force on wall indicated as Figure 11 (b) and bending moment on
wall are illustrated as Figure 11 (c). Due to the axis of plate
modelling is difference between PLAXIS in this study, the result of
bending moment from PLAXIS is difference. From the result of
PLAXIS, maximum shear force and bending moment of wall are
3069 kN/m and 4427 kNm/m respectively and both are located at
the center of the 17 m spacing. The waler sizes which can take the
shear force and bending moment is 2UC 400x400x283 kg/m, this
waler size is selected for this condition. The shear capacity and
bending moment capacity of 2UC 400x400x283 kg/m are 4520 kN
and 5554kNm. The cross-section area of 2UC 400x400x283 kg/m is
0.0722 m2.
(a)
Figure 9 Comparison between observation and numerical analysis of
Case KLCC
218
c. Strut type of case A;

d. Retaining wall type of case KLCC
As refer to Figure 12, the impacts on PSR from factors stated
above are presented. Table 8 is the summarized of the difference in
percentage with the factor used to analysis in this study. The
baseline of the difference is Kaohsiung sand which provided by
Hsiung et al. (2016). In the Table 5.2, name of KHH stand for Case
A. From the summary of the result, higher value of difference when
increasing the value of B/L due to the larger width stiffness for the
corner is low. Meanwhile, when excavated in the narrow excavation,
strut sizes and strut spacing is the key impact for the wall
deformation; impact of horizontal spacing for the PSR is significant.
Overall of the result, the most impact for PSR is ground condition
and arrangement of strutting system.
(b)
(a)
(c)
Figure 11 Result of numerical analysis: (a) total lateral wall

displacement on wall; (b) shear force on wall; (c) bending moment
on wall
To further evaluate the impacts from several factors which may

influence PSR such as type of wall, horizontal spacing of waler,
sizes of waler, parametric studies based on the same excavation
model were conducted. By using the analytical results of case study
of deep excavation in similar high permeability ground condition
provided by Hsiung et al. (2016), named as Case A to compare with
Case KLCC. The followings cases are simulated in this study: (b)
1. Fully follow Case A:
a. Horizontal and vertical strut spacing of Case A;
b. Strut type of Case A;
c. Retaining wall type of Case A;
2. Follow Case A strut and spacing, Case KLCC wall:
b. Strut type of Case A;
c. Retaining wall type of Case KLCC;
3. Follow Case A strut spacing and wall, Case KLCC strut:
b. Strut type of Case KLCC;
c. Retaining wall type of Case A;
4. Follow Case A strut and vertical spacing and wall, Case
KLCC horizontal spacing:
a. Vertical strut spacing of Case A;
b. Horizontal strut spacing of Case KLCC;
c. Strut type of Case A;
(c)
d. Retaining wall type of Case A;
5. Follow Case A strut and vertical spacing, Case KLCC
Figure 12 Variation of PSR with d for various value of ratio B and L:
horizontal spacing and wall:
(a) B/L=0.5; (b) B/L=1; (c) B/L=2
a. Vertical strut spacing of Case A;
b. Horizontal strut spacing of Case KLCC;
219
Table 8 Summarized the difference of impact factors

(a) Difference of B/L =0.5 (B = 40 m, L = 80 m)
B/L = 0.5
Name of cases Difference (%)
KKH -
Fully follow case A 5.178
Case A strut & spacing & wall, KLCC wall 5.39
Case A strut & spacing & wall, KLCC strut 6.82
Case A strut &vertical spacing & wall, KLCC 12.16
horizontal spacing
Case A strut & vertical spacing, KLCC horizontal 5.35
spacing & wall
(b) Difference of B/L =1 (B = 80 m, L = 80 m)

B/L = 1
KKH -
Fully follow case A 40.28
Case A strut & spacing & wall, KLCC wall 39.81
Case A strut & spacing & wall, KLCC strut 39.68 Figure 13 Relationship between maximum bending moment on wall
Case A strut &vertical spacing & wall, KLCC 6.5 and depth of bedrock
horizontal spacing
Case A strut & vertical spacing, KLCC 41.088
horizontal spacing & wall
(e) Difference of B/L =2 (B = 80 m, L = 40 m)

B/L = 2
KKH -
Fully follow case A 112
Case A strut & spacing & wall, KLCC wall 120
Case A strut & spacing & wall, KLCC strut 103
Case A strut &vertical spacing & wall, KLCC 62.28
horizontal spacing
Case A strut & vertical spacing, KLCC 119
horizontal spacing & wall
For the evaluation of waler sizes, various depths of bedrock are

given which are 7 mbgl, 11 mbgl, 14 mbgl, 18 mbgl, 21.5 mbgl and
25 mbgl. There is no additional support system when the excavation
is beyond the bedrock level as the bedrock is recognized to be stiff
enough to stabilize the ground. The depth of the lowest strut is only
at above 1 m of the bedrock. The retaining wall is socketed 3 m into
bedrock in every model and rock bolt is ignored in this study. Once
the bedrock becomes shallower, the layer of strut where deeper than
Figure 14 Envelope of suitable waler sizes for connect 17 m of
the bedrock will be removed.
spacing wall.
From the numerical result of stresses on the wall, once the
bedrock becomes shallower and retaining wall become shorter, the
value of shear force and bending moment are lower, which mean the 5 CONCLUSIONS
load from the wall to waler is lower, the sizes of waler can be
smaller. Figure 13 is the relationship between stresses and depth of Based on this study, the following conclusions can be drawn by two
the retaining wall. The maximum shear force and bending moment parts:
on wall are located at the center of 17 m strut spacing. For Case KLCC
Based on the numerical results and the calculation of the shear 1. It is aware that very limited information about site
force and bending moment capacity of waler, the envelopes for the investigation and strut details are available so certain
suitable sizes of waler for the strut spacing of 17 m has been reasonable assumptions have to be made for analyses.
established in this study and presented in Figure 14, the cross- Based on the result of the numerical analysis, majority of
section of the suitable strut size for x-axis of the chart, depth of the observed field data clearly in good agreement with the
bedrock for y-axis of the chart. Regarding the envelopes of Figure prediction.
14, the equation of the line regression is: 2. The wall where near the corner at the section of
inclinometers (IN-1 and IN-10) are underestimated which
might be due to comparatively worse quality of the
y = 288x + 4.8642 (6)
vertical joint between the rectangular panel and L-shape
panel of the wall and it thus induces a larger displacement
where y is depth and cross-section area of waler is x in unit of m2.
from the observations.
3. Larger difference with increasing B/L is aware which is
220
due to the wider excavation has a low stiffness at the in Soil Mechanics and Geotechnical Engineering,
corner. Proceedings of the 5th International Young Geotechnical
For Case Waler: Engineers’ Conference, pp559-563.
Law K. H., Ismail Z. and Roslan H. (2016) “3D Finite Element
1. From the comparison between results of the numerical
Analysis of a Deep Excavation Considering the Effect of
analysis and observations, it is likely the stiffness of soil is
Anisotropic Wall Stiffness” 19th Southeast Asian
underestimated. As the result, the analysis might
Geotechnical Conference & 2nd AGSSEA Conference
overestimate the lateral wall deflection.
(19SEAGC & 2AGSSEA) Kuala Lumpur, pp659-664.
2. Due to 3- dimensional characteristics of the model and
Law K. H., Hashim, R. and Ismail Z. (2013) “Performance of
software, the result of shear force and bending moment on
Multipropped Deep Excavation in Kenny Hill Formation”,
wall from numerical analysis of PLAXIS3D is possible to
Proc. 18th Southeast Asian Geotechnical Conference,
be adopted for the evaluation of waler sizes.
Singapore, pp705-712.
3. The envelope of suitable waler size is suitable for prelim
Ou, C. Y., (2006) “Deep Excavation: Theory and Practice” Taylor &
selection of the situation with 17 m of strut spacing in
Francis, Netherlands
various depths of bedrock with upper alluvium soil.
Ou, C. Y., Chiou, D.C., and Wu, T. S. (1996) “Three-dimensional
4. The reason why having the 17 m of strut spacing is caused
Finite Element Analysis of Deep Excavations”, Journal of
by the need to put a whole tunnel bored machine (TBM)
Geotechnical and Geoenvironment Engineering, pp337-345
from ground surface to purpose excavation level. The
Schanz, T., Vermeer, P. A., and Bonnier, P. G. (1999) “The
envelope of suitable waler size is a reference for selecting
hardening soil model: formulation and verification”, Beyond
waler sizes in this kind of situation.
2000 in Computational Geotechnics, Balkema, Rotterdam,
pp281-296.
6 REFERENCES
Skempton, A. W. (1986) “Standard Penetration Test Procedures”
Ang J. S., Hsiung B. C. B. and Wu, M. R. (2017) “Performance of Geotechnique, vol. 36, no. 3, pp.425-557.
Deep Excavation in Kuala Lumpur” Proceedings of the 17th Tan, S. A. (2010) “One North Station Excavation in 30m of Jurong
Conference on Current Researches in Geotechnical Residual Soils in Singapore”, Earth Retention Conference,
Engineering in Taiwan. Washington, pp732-739.
Bishop, A. W. (1966) “The Strength of Soil as Engineering Tan Siow Meng and Simon (2011) “Karstic Features of Kuala
Materials” Geotechnique, pp91-128 Lumpur Limestone”
Brinkgreve, R.B.J. et al., Editors (2012) “Plaxis 3D 2012 user’s Waltham, A. C. and Fookes, O.G. (2003) “Engineering
manual”, Netherlands, Plaxis bv. classification of karst ground conditions”. Quarterly Journal
BS 5950: Part 1 (2000), Structure use of Steelwork in Building, Part of Engineering Geology and Hydrogeology, pp101-118
1: Code of Practice for Design and Construction, British Zdravkovic, L., Potts, D. M., and St Jogn, H. D. (2005) “Modelling
Standards Institution of a 3D Excavation in Finite Element Analysis”,
Chiew, S. P. and Leow, B. H. (2006) “Common Mistakes in Steel Geotechnique, pp497-513.
Design” International Symposium on Worldwide Trend and
Development in Codified Design of Steel Structures. Kuala
Lumpur, Malaysia.
Chogueur A., Abdeldjalil Z. and Reiffsteck P. (2018) “Parametric
and Comparative Study of a Flexible Retaining wall”
Periodica Polytechnica Civil Engineering, pp295-307
Eurocode 3 Part 1-1 (2005), Design of Steel Structures: Part 1-1
General Rules for Building, British Standards Institution
Ganno J. (2016) “Primary Firm Secant Pile Concrete Specification”
Proceedings of the institution of Civil Engineers, pp110-120
H. Zabidi, M. Termizi, S. Aliman, K. S. Ariffin and N. L. Khalil
(2015) “Geological Structure and Geomorphological Aspects
in Karstified Susceptibility Mapping of Limestone
Formations” 5th International Conference on Recent
Advances in Materials, Minerals and Environment & 2nd
International Postgraduate Conference on Materials, Mineral
and Polymer, pp659-665
Hsiung, B. C. B. (2009) “A Case Study on the Behavior of a Deep
Excavation in Sand. Comput. Geotechn. Pp.665-675
Hsiung, B. C. B., Hwang, R.N. (2009) “Evaluating Performance of
Diaphragm walls by Wall Deflection Path. SEAGS. Special
Issue on Excavation and Tunneling in Geotechnical
Engineering, pp.81-90
Hsiung, B. C. B., Wang, C. L., Lin, H. T. and Chen C. H. (2013)
“Design and Performance of a large-scale Excavation
Adjacent to Sensitive Structures in Urban area” Proceedings
of the 2nd International Conference on Geotechnics for
Sustainable Development-Geotec. Hanoi, Vietnam
Hsiung, B. C. B., Yang, K. H., Aila, W., Hung, C., 2016. Three-
dimensional Effects of a Deep Excavation on Wall
Deflections in Loose to Medium Dense Sands. Comput.
Geotechn.
Koo Kuan-Seng (2013), “Design and Construction of Excavation
Works for Klang Valley Mass Rapid Transit Underground
Station at Cochorance, Kuala Lumpur, Malaysia” Advances
221
Assessment of mechanical behavior of granular soils adopting various plasticity models

Deepa Patil1, S. D. Anitha Kumari1 and Nimmy Mariam Abraham1
1
Department of Civil Engineering, M. S. Ramaiah University of Applied Sciences, Bangalore, India
E-mail: anitha.vipin@gmail.com
ABSTRACT: The mechanical behavior of soils is highly complex due to the heterogeneous behavior of these deposits. There are various
experimental and empirical methods followed to assess the behavior of these granular masses. However, field tests, laboratory tests and empirical
relationships have inherent uncertainties associated with them. The empirical correlations developed are based on the data collected from various
field tests. The quality of the data collected from various field tests like Standard Penetration Test (SPT) and Cone Penetration Test (CPT)
depends on the field practices which may be inconsistent due to different test methods followed in the field. Also the quantity of data available for
adopting these correlations are not sufficient in some cases. Hence for a sophisticated evaluation of various properties, analytical methods are a
must. In addition to this, the behavior of saturated soil during sudden loading is highly influenced by the development of excess pore water
pressure. The sudden development of excess pore pressure in soils is directly related to the field condition of that soil deposit. Analytical methods
help to understand the gradual accumulation of permanent strain. In this research, various models based on theory of plasticity is adopted for
assessing the behavior under drained static conditions. The data obtained from the reported experimental results are collected for various soils and
are utilized to obtain the elastic and hardening parameters required for the numerical and analytical models. Numerical modeling of the sample
and various loading conditions are simulated using a finite element software. The stress strain relationships which accounts for the accumulation
of plastic strain and the subsequent plastic flow is defined and a comparison of the different models considered is reported.
Keywords: Drained Triaxial test, Cam Clay Model, Drucker Prager Model, Shear stress, Strain
1. INTRODUCTION
2. BACKGROUND STUDY
Granular material is made up of discrete particles. The shape and
Jefferies (1993), developed constitutive model for sand based on
size of particles influence the mechanical behavior of granular critical state theory. The study indicated that soil moves to critical
materials. Also it depends upon particle orientation, friction state regardless of initial conditions. Khalili et. al (2005),
among particles, pore spaces and saturation level of particles. formulated Bounding surface plasticity model for granular soils
When external forces such as static and cyclic loads act on these
subjected to static and cyclic loads. They developed the model
materials, deformation takes place by particle sliding leading to using the available experimental data from literature. It is reported
changes in mechanical behavior. Therefore, understanding the that the developed model predicts stress softening and dilatancy
mechanical behavior is important in designing the structures. during drained loading of dense sand. Arvelo (2005) evaluated the
Because any changes in behavior of soil affects the structures
behavior of dense granular soil using Modified Cam Clay Model
directly and may cause failure of structure (Brenda, 2003). (MCCM). The results of the analytical model are compared to the
When granular material such as sand or silty sand (saturated laboratory drained triaxial test data. The study suggests that
cohesion less soil) is subjected to rapid rate of loading, positive MCCM with some modifications, is applicable to dense sands.
excess pore pressure is developed. This causes decrease in
Thomas Oommen et. al (2010) compared the predictive
effective stress leading to the reduction on shear strength of soil performance of empirical liquefaction models. Deterministic and
and subsequently soil behaves like a liquid. This phenomenon is probabilistic empirical liquefaction models are developed using
known as liquefaction which is usually associated with earthquake SPT and CPT data. But it is found that identifying gaps in data
loads. But liquefaction can occur due to static sudden loads also.
sets is extremely important for improving empirical models. Xilin
This phenomenon is known as static liquefaction. Static & Maosong (2014) proposed a model to analyze the stress strain
liquefaction causes damage to the infrastructure from landslides behavior of soil when it is subjected to static liquefaction. Mohr
and bearing capacity failure. So it is important to study the Coulomb elasto-plastic hardening model is proposed to predict
behavior of soil before infrastructure construction. (Ellison & static liquefaction. It is concluded that static liquefaction is
Andrade, 2009). Triaxial tests are generally conducted to initiated in loose sands when the undrained stress path occurred
understand the the behavior of granular soil. But it is difficult to along with potentially unstable stress path. If sand is dense, it fails
collect the undisturbed sample of loose sand to conduct triaxial when state of sand becomes potentially unstable. Results show
test. To overcome this difficulty, empirical methods are developed that static liquefaction occurred for selected sample at the
based on SPT, CPT and Borehole Penetrometer Test (BPT) hardening stage only before it reaches plastic limit failure. Rani. et
(García et.al, 2012). But empirical method depends upon material al. (2014) studied the prediction accuracy of Mohr-Coulomb and
conditions and is limited to specific topographic conditions. Drucker Prager models in evaluating the behavior of clayey soil.
Accuracy of prediction of behavior of soil adopting empirical It is reported that the Mohr- Coulomb model showed lower
methods depends on knowledge of soil properties such as predictions compared to Drucker-Prager model.
strength, stiffness and sampling method (Beaty and Byrne, 2000).
These all limitations affect the accuracy of evaluation of shear 3. CAM CLAY MODEL
failure of granular soil under various loading conditions. To Cam Clay model is developed to represent the soil behavior based
overcome the limitations of empirical method, analytical methods on critical state. This model considers strength, compression and
are adopted which are simpler, easier and more accurate than critical state aspects while formulating the behavior under various
traditional empirical methods (Lee, 2009). In this study, Cam Clay loading conditions. The basic parameters which describe this
model and Drucker Prager models are used to analyze the model include the effective mean stress, deviatoric stress and
behavior of granular soil subjected to drained static loading specific volume. The initial state of stress is also considered
conditions.
222
significant in this model which can be considered in terms of over developed using SOLID 185 element in ANSYS (Figure 2). This
consolidation ratio or pre-consolidation stress. element is suitable for 3-D modeling of solid structure which is
3.1 Collection of experimental data defined by 8 nodes. The material properties given in Table 1 are
The various input parameters required for the model has to be used for the modeling. To replicate the experimental conditions,
obtained from experimental data. For this, drained triaxial test fixed base is given at the bottom of sample by selecting the
data reported by Arvelo (2005) is adopted for this study. From this bottom surface as shown in Figure 3. In this study, the gradual
test, material properties and parameters for Cam Clay model are application of loads is done by providing 600 load steps. These
selected to evaluate the stress and strain behavior of the granular load steps were sufficient to capture the gradual development of
material. This data can be used as the input for both numerical and the non-linear behavior of sand.
analytical modeling. The properties collected from the triaxial test
are given in Table 1. The sand sample is subjected to a confining An isotropic loading condition is imparted to the sample by
pressure of 100kPa. Figure 1 shows the variation of shear stress applying a confining pressure of 100kPa in all directions and is
with shear strain for the drained triaxial laboratory test. shown in Figure 4(a). After isotropic stage, shear stage is
developed by giving vertical displacement at the rate of 2mm per
Table 1 Material Properties of Dense Sand Obtained from minute throughout the last load step. i.e. up to 600 seconds as
Triaxial Drained Test shown in Figure 4(b).
Internal friction angle (Φ in degree) 34.96
Initial voids ratio (eo) 0.28
Swelling index (κ) 0.0021
Slope of critical straight line (M) 1.416
Poisson’s ratio (μ) 0.35
Young’s modulus (E) 48 MPa
Density of sand (γ) 2700 kg/m3
Specific gravity (Gs) 2.7
Figure 2 Sand sample for numerical simulation in ANSYS
Figure 1 Shear stress vs Shear strain: Experimental results

(Arvelo, 2005)
3.2 Analytical modeling adopting Cam Clay Model

The analytical model to predict the mechanical response of
granular sand under drained conditions based on Cam Clay theory
is developed in MATLAB. The relevant material properties to
apply in the constitutive model are obtained from drained triaxial Figure 3 Fixed support at the bottom of sand sample
test results (Arvelo, 2005).
3.3 Numerical modeling adopting Cam Clay model

The sand sample is numerically modeled in ANSYS workbench
for drained triaxial test adopting Cam Clay model. The cylindrical
sand sample having diameter 36mm and height 72 mm is
223
Figure 4 (a) Isotropic loading stage of sand in ANSYS

Figure 5 Stress path for drained condition
(a) Deformation
Figure 4 (b) Shear loading stage of sand
3.4 Results and Discussions of Cam Clay model

The loading path followed for the shear test in both numerical and
analytical studies is shown in Figure 5. The deformed shape and
strain in the sand sample when subjected to drained triaxial test
condition in numerical analysis is shown in Figure 6(a) and (b).
The formation of shear band is clearly visible in these figures.
Figure 7 shows a comparison of the constitutive behavior of the

experimental, numerical and analytical studies. The results
indicate the Cam Clay model prediction is fairly accurate in low
strains during the strain hardening stage. It can be observed that
the initial stiffness is very high and is similar to experimental
results in both the numerical and analytical studies. However the
peak stress attained by both the models is less than that observed
(b) Strain
in the experiment. More over the strain softening behavior
predicted by Cam Clay model shows much lower residual strength
values when compared to experiments. This clearly shows that Figure 6 Shearing stage in sand sample
Cam Clay model can accurately predicts the initial hardening
behavior whereas the strain softening on residual strength is not
accurately predicted. A comparative study of the maximum shear stress and the
corresponding shear strain is shown in Table 2. This table
indicates that the strain at which peak shear stress is observed in
224
numerical and analytical solutions is less than that of experimental Table 3 Material properties of Sand
results. This also points to the fact that this model underestimates
the residual value or fails to predict the strain softening behavior. Peak State Internal friction angle (Φ in degree) 39.5
Cohesion (kPa) 2.9
Residual state Internal friction angle (Φ in degree) 36.3
Cohesion (kPa) 1.7
Soil Properties Density of the soil (kg/m3) 1767

Poisson’s ratio 0.214
Figure 7 Comparison of shear stress vs. shear strain: Experiment,

numerical and analytical methods
Table 2 Comparison of maximum shear stress and corresponding

shear strain
Type Shear stress Shear strain
(MPa) (%)
Experimental results 0.46 2.5
Analytical results 0.44 0.83 Figure 8 Shear stress vs. axial strain: Experimental results
(Arvelo, 2005)
Numerical modeling 0.38 0.8
4.3 Results and Discussion of Drucker Prager model
4. DRUCKER PRAGER MODEL
The deformation behavior of the sand sample when subjected to
Drucker Prager criterion is a generalization of Mohr-Coulomb shear loading is shown in Figure 9(a) and (b). The formation of
criterion for soils. Failure in this model is estimated by assuming the shear bands in this case also is obvious from these figures. A
that the ultimate shear stress is dependent on octahedral normal comparison of the variation of shear stress with axial strain for the
stress. However the relation can be used for the estimation of the numerical and experimental studies is shown in Figure 10. The
ultimate strength through a number of material constants. These initial stiffness values predicted by Drucker Prager model are
material constants can be arrived from the various experimental similar to the experimental results. However, even at small strains,
results. the prediction of the stresses varies from that the experiment.
4.1 Collection of experimental data Also, the maximum stress predicted by the numerical model is
The various material parameters required for the modeling of the higher than the experimental results and at a much smaller strain.
mechanical behavior using Drucker Prager model is collected The strain softening behavior indicates that the predicted values
from Abe et. al (2012). The idea of the paper is to understand the are much lesser than the experimental values observed. This is
failure patterns and condition of a slope after failure. To evaluate similar to the observations made when Cam Clay model is
the properties of the soil constituting the slope, drained triaxial adopted. However the percentage differences in the predicted
test is carried out on the soil sample subjected to a confining residual values are lesser in Drucker Prager model. A comparative
pressure of 50kPa. The material properties for modeling the soil behavior of the two models with the corresponding experimental
sample as collected from the experimental test results are given in data is given in Table 5 and Table 6.
Table 3. Figure 8 shows the variation of shear stress with axial
strain for the experiment (Abe et. al, 2012).
4.2 Numerical modeling adopting Drucker Prager Model
After the collection of the material data, the sample is modeled
similar to that described in the previous section. The cylindrical
sample modeled is subjected to an all round confining pressure of
50kPa. This sample is then subjected to shearing.
225
Drucker Prager Model, the strain softening behavior is not

captured properly. It also overestimates the strength at lower
strain by 49.46%. However, this model captures softening
behavior better than Cam Clay Model. The percentage difference
in the observed shear stress at around 5% strain is 29.9% for
Drucker Prager model whereas it is around 63% for the Cam clay
model.
(a) Deformation
Figure 10 Comparison of experimental and numerical results for

Drucker Prager model
Table 5 Comparison of behavior of sand obtained using Cam Clay

Model
Behavior Experiment Numerical %
(Shear (Shear Difference
Stress kPa) Stress, kPa)
Hardening(0.6% 302.54 277.51 8.26 %
strain)
Softening(5% 447.7 165 63.14%
strain)
Table 6 Comparison of behavior of sand obtained using Drucker

(b) Strain Prager Model
Figure 9 Shearing stage in sand sample (Drucker Prager model) Behavior Experiment Numerical %
(Shear (Shear Difference
6. CONCLUSION Stress kPa) Stress kPa)
The behavior of granular soil under static drained condition is Hardening(0.3% 76 150.4 49.46 %
analyzed numerically and analytically. Numerical results and strain)
analytical results are compared with experimental values. Results Softening(5% 185 129 43.41 %
show that Cam Clay Model does not predict the shear behavior of strain)
sand sample adequately. It fails to predict the softening behavior
of the sample. It is also observed that the predicted maximum ACKNOWLEDGEMENTS
shear stress and the corresponding shear strain is lower than the The authors would like to extend their gratitude to M.S. Ramaiah
experimental values. This model is suitable for normally University of Applied Sciences for providing all the facilities for
consolidated and lightly over consolidated soil (O-C ratio<= 2). conducting the research work.
But the selected sample is over consolidated sand. So it fails to
predict strain softening part. Strain hardening or small strain REFERENCES
behavior is captured well in Cam Clay model. But at higher Arvelo, J., (2005) “Adapting the Modified Cam Clay Constitutive
strains it fails to do so. At lower strains, the difference of stress in Model to the Computational Analysis of Dense Granular Soils”,
experiment and numerical results is 8%, whereas at higher strains Ph.D Thesis, University of Central Florida.
the percentage difference is high. A comparison of the numerical Brenda Novoa-Martinez., (2003) “Strength properties of granular
and analytical studies using Cam clay model shows a difference of materials”, Louisiana State University LSU Digital Commons.
only 0.03% in the strains at maximum shear stress. In the case of
226
Beaty, M. H., and Byrne, P. M., (2000) “A Synthesized Approach Ellison, K. C., and Andrade, J. E., (2009) “Liquefaction Mapping
For Predicting Liquefaction”, WCEE, pp1589-1597. in Finite-Element Simulations”. Journal of Geotechnical and
García, S., Ovando-Shelley, E., Gutierrez, J., and Garcia, J., Geoenvironmental Engineering, November, 135, pp1693-1701.
(2012) “Liquefaction Assessment through Machine Learning”, Lee, Shuh-Gi Chern., and Ching-Yinn., (2009) “CPT-Based
WCEE. Simplified Liquefaction Assessment By Using Fuzzy-Neural
Xilin Lu., and Maosong H., (2014) “Static Liquefaction of Sands Network”. Journal of Marine Science and Technology, 17(4),
under Isotropically and K0-Consolidated Undrained Triaxial pp326-331.
Conditions”. Journal of Geotechnical and Geoenvironmental Sandhya Rani, R., Nagendra Prasad, K., Sai Krishna., (2014).
Engineering, 141(1), pp. 04014087(1-9). “Applicability of Mohr-coulomb & Drucker Prager Models for
Jefferies, M. G., (1993) “Nor-Sand: A simple critical state model Assessment of Undrained Shear Behaviour of Clayey Soils”.
for sand”. Géotechnique, 43(1), pp. 91-103. International Journal of Civil Engineering and Technology,
Khalili, N., Habte, M. A., and Valliappan, S., (2005) “A bounding 05(10), pp. 104-123.
surface plasticity model for cyclic loading of granular soils”. Thomas Oommen, Laurie, G. B., and Richard, V., (2010)
International Journal For Numerical Methods In Engineering, 63, “Validation and Application of Empirical Liquefaction Models”.
pp. 1939-1960. Journal Of Geotechnical And Geoenvironmental Engineering,
136(12), pp1618-1633.
227
Remediation of Oil Tank using PLAXIS 3D

A. Gunawan1
1
Department of Civil Engineering, Bina Nusantara University, Jakarta, Indonesia
E-mail: agunawan@connect.ust.hk
ABSTRACT: Good and accurate soil investigation is necessary to calculate pile performance with considerable accuracy. Inaccurate or
insufficient soil investigation can lead to underestimation or overestimation of pile capacity. For the former case, the design becomes
uneconomical, while for the latter case, it can lead to unacceptable settlement, and in worst case scenario, even failure. Recently, there was
one such case of an oil tank in East Java. Initial investigation found very dense layer 8 metres below the ground surface until end of
investigation depth (30 m depth). Over 300 driven piles were installed to support the tank. Ten piles were tested with Pile Driving Analyzer,
and 8 out of 10 did not meet the design capacity. Additional soil investigation later revealed that the very dense layer was only 3 m thick.
Below the very dense layer, loose sand was found until 45 m depth. The design was re-evaluated with PLAXIS 3D, and installation of
additional piles were required to meet the settlement criterion of 100 mm. However, since it was possible to jack up the oil tank after its
settlement, it was decided not to do any remediation.
Keywords: pile, serviceability, soil investigation, numerical analysis.
1. INTRODUCTION 2.2 Plan View of the Oil Tank

Soil investigation plays a vital role in optimizing a geotechnical Figure 1 shows the plan view of oil tank showing the pile layout, as
design. The number of tests and their depths depends on a lot of well as the old and new soil investigation location. The 304 piles
factors, namely the load and sensitivity of structure to be were spaced 2.5 m apart from each other. For the piles, open circles
constructed, location of hard layers, variability of soil in the area. denote the piles that reached the target depth; red solid circles (36 of
The location of the tests should be determined by experienced them) denote piles with broken pile cap; and blue solid circles (23 of
geotechnical engineer in order to achieve the most valuable them) denote piles which did not reach their target depth. For the
information with the least tests required. Unfortunately, many soil investigation, solid square (1 only) denotes the previous
project owners, who usually does not come from engineering borehole location, while the open squares (3 of them) denote the
background, underestimate the importance of soil investigation. additional soil investigations.
They treat them as unnecessary expenses, used only to satisfy the It can be seen that the damaged and short piles (the piles that did
requirement of national standards. However, the cost saving from not reach target depth) are scattered all around the oil tank, but with
reducing the number of investigations is far less than the majority on the right side of the tank.
consequences that may occur. Since the old soil investigation was carried out in the middle of
Insufficient or inaccurate soil investigation will require the tank, additional soil investigation (BH1) was conducted in its
designers to take a more conservative approach, such as taking proximity to verify the information provided from existing soil
lower strength and stiffness parameters or higher factor of safety. investigation. Another 2 soil investigations were conducted to
This makes design inefficient and expensive, often costing more confirm the uniformity of soil profile under the oil tank.
than the money saved from soil investigations. A more dire
consequence is when there are unforeseen soil conditions, resulting 2.3 Soil Profile (Existing and New Information)
in excessive settlement, damages or even failures. The
repair/remediation cost is usually more expensive than building a Previous soil investigation showed that the soil in the area consist of
new one from scratch. sand with varying density. From the ground surface to 3 m depth,
One recent case was of an oil tank in East Java. Due to medium dense sand was found, followed by dense sand to 7 m depth.
inaccurate soil investigation, design capacity of the piles installed From 8 m depth onwards, very dense sand with SPT value over 50
was not met. Concern was raised by the project owner, and the was found. The SPT profile of the previous test is shown in Figure 2.
design was re-evaluated. The water table was found 7 m underneath the ground surface.
To verify this data, 1 new borehole was drilled beside the
existing borehole, and another 2 at the north east corner and south
2. PROJECT DETAILS AND PROBLEMS
west corner of the tank. The new investigations revealed that only
ENCOUNTERED
the top 6 meters of sand were in a dense to very dense state. Below
2.1 Project Information this depth, loose sand was encountered up to 45 m depth, where the
sand starts to increase in density. The SPT profile of the new tests is
A 50 m diameter oil tank with 20 Ml (mega litre) capacity was to be also shown in Figure 2 for comparison. The water table was found at
built in Tanjung Wangi, East Java, Indonesia. The tank concerned 6 m underneath the ground surface, similar to the previous
was one of the five tanks to be constructed on site. Three boreholes investigation.
with SPT (standard penetration test) were carried out to design the The new investigations unfolded the reason as to why the piles
five tanks, in which one of the borehole was located in the middle of tested did not meet their design capacity. The piles that were meant
concerned tank. Three hundred and four driven spun piles (0.4 m to be end-bearing piles were in fact floating piles. The short piles
diameter) were installed to support this oil tank. Problems were were caused by the very dense layer, and the piles were not able to
encountered during installation of the piles. Out of the 304 piles be driven past this layer. The damaged piles were caused by
installed, 23 of the piles were damaged due to overdriving, while 36 overdriving when attempting to penetrate this very dense layer. Re-
others did not reach their target depth of 10 m. Further increasing evaluation of the oil tank was conducted based on the new
the project owner’s concern, the pile capacity test (with Pile Driver information.
Analyzer) reported that 8 out of the 10 piles tested did not meet their
design capacity. Thus, the project owner request re-evaluation of the
design, and remediation if needed.
228
Figure 1 Layout of installed piles

3. RE-EVALUATION OF OIL TANK PERFORMANCE 3.2 Liquifaction Analysis
Since the oil tank is to be founded on sand, liquefaction analysis to
3.1 Bearing Capacity
be analysed. The liquefaction potential is analysed using correlation
Firstly, bearing capacity of the oil tank was evaluated ignoring any of cyclic stress ratio versus SPT values by Seed et al. (1985). First
of the existing piles (i.e. raft only). The calculation was done using the cyclic stress ratio (CSR) has to be calculated using the formula:
Terzaghi’s bearing capacity formula for circular foundation CSR = 0.65 x amax/g x 0/0’ x rd (2)
(Terzaghi, 1943).
where amax is the peak ground acceleration, g is gravity, 0 is total
qu = 1.3 cNc + qNq + 0.3 BN (1) stress, 0’ is effective stress and rd is reduction factor that ranges
from 1 at the ground surface, to 0.9 at a depth of 9.6 m. The peak
where qu is the unit bearing capacity, c is soil’s cohesion, q is ground acceleration for 10% exceedence in 50 years at Tanjung
surcharge above the foot of foundation,  is unit weight of soil Wangi is between 0.2 to 0.25 g (Irsyam et al., 2017). By knowing
below the foundation, B is the width/diameter of foundation and Nc, the location of water table, and soil density (Table 1), the CSR can
Nq and N are the bearing capacity factors. be calculated at any depth. Then, minimum SPT values that does not
Since the soil was sand and the foundation was to be founded on cause liquefaction under a given CSR can be derived from the chart
the ground surface, c and q = 0, making the first 2 terms in the shown in Figure 3. Before comparing the derived SPT values with
equation equal to zero. Taking friction angle of sand as 30o, and field data, the SPT values from the field have to be corrected to 60%
effective unit weight of sand = 8 kPa, the bearing capacity can be energy transferred at 1 atmospheric pressure of overburden. The
calculated as qu = 2300 kPa. The weight of the 21 Ml oil tank is method to correct the SPT values is given in Peck et al. (1974).
210,000 kN, giving a pressure underneath the oil tank as 107 kPa. From the evaluation, it was found that the top 6 meters of sand
Adding the weight of 0.6 m of raft (pile cap) supporting the oil tank does not liquefy, while the next 25 meters liquefy under 0.25 g
(0.6 x 24 = 14.4 kPa), the total pressure underneath the oil tank is acceleration. However, this does not immediately mean that the oil
121.4 kPa. This gives a factor of safety of almost 20, hence bearing tank is susceptible to liquefaction damages. Ishihara (1985) state
capacity is not a problem. that liquefaction damages is also dependent on the thickness of non-
229
liquefied layer. If there is sufficient thickness of non-liquefied soil Table 1 Soil Profile and Parameters Used for Analysis
above the liquefied layer, there will not be liquefaction-induced Young’s
damages. Based on the chart proposed by Ishihara (1985) shown in Depth SPT Unit weight Friction
3 o Modulus
(m) Values (kN/m ) angle ( )
figure 4, for 0.25 g acceleration, as long as the non-liquefied layer is (MPa)
thicker than 4.5 m, there is no liquefaction-induced damages. Hence, 0.0-2.5 15 18 30 18000
the oil tank is deemed safe from liquefaction problems. 2.5-5.5 50 19 27 60000
5.5-47.0 8 16.5 28 9600
3.2 Settlement Analysis 47.0-70.0 60 20 40 72000
To evaluate the settlement of the oil tank, a finite element software,
PLAXIS 3D was used. The soil profile used was idealised using the
3 new SPT data, shown by the dashed line in Figure 2.
3.2.1 Constitutive Model and Soil Parameters

The soil model used was Mohr Coulomb, and soil parameters were
based on the SPT values. The friction angle and unit weight were
estimated using a table by Bowles (1977), while the Young’s
modulus was estimated by a correlation given by Stroud (1989).
Stroud (1989) gives the following correlation:
E’/N = 1 MPa (3)
where E’ is the Young’s modulus, and N is the SPT values. The

correlation given was for foundation with factor of safety of 3. For
foundation with higher factor of safety, higher ratio of E’/N can be
used. For this case, E’/N = 1.2 was chosen. The soil parameters are
summarized in Table 1.
Figure 3 Relationship between stress ratio causing liquefaction

and N-values for silty sands (Seed et al. 1985)
Figure 2 Existing and new SPT N values Figure 4 Boundary curves for site identification of liquefaction-
induced damage (Ishihara, 1985)
230
3.2.2 Numerical Model For the plate element, the required inputs are thickness, unit
weight, Young’s modulus and Poisson’s ratio. The thickness used
The numerical model used is shown in Figure 5. The model had a was 0.6 m, the same as that planned. Unit weight and Poisson’s ratio
dimension of 150 x 150 x 70 m (length x width x height), with the are 28 GPa and 0.2 respectively, typical values for reinforced
oil tank located in the center of the model. The piles were modelled concrete.
as embedded beam elements (Brinkgreve et al., 2017). The broken
piles were not modelled, while the short piles were modelled as they
3.2.4 Modelling Procedure
were on site. The raft was modelled as plate elements, and the
weight of the oil tank was modelled as a distributed load. The modelling of the oil tank followed the following procedure:
1. Generation of initial stress by K0 procedure
3.2.3 Properties of Embedded Beam and Plate 2. Installation/activation of piles
3. Installation/activation of plate
In PLAXIS, the required inputs for embedded beam are Young’s 4. Activate surcharge load
modulus, unit weight, diameter, maximum axial skin resistance and
base resistance. The young’s modulus and unit weight are 28 GPa,
3.2.5 Evaluated Settlement
and 24 kN/m3 respectively, typical values for reinforced concrete.
The diameter of pile was 0.4 m as previously mentioned. In Figure 7 shows the settlement contour of oil tank if the oil tank was
PLAXIS, the axial skin resistance is automatically calculated, but to be constructed as it was. The maximum settlement is located in
not the base resistance. The base resistance has to be calculated the middle of the tank and minimum settlement is located in the
manually and inserted as input. For the axial skin resistance, only edge of the oil tank. The maximum and minimum settlement are 330
the limiting value is required. mm and 230 mm respectively. The settlement was higher than the
Base resistance of pile was calculated using correlations tolerable settlement of 100 mm. Therefore, improvement was
tabulated by Poulos (1989). Decourt (1982) gave a correlation for required.
driven pile in sand as:
fb = 400 x N kPa (4)
where fb is the unit base resistance and N is the average SPT values
in the local failure zone.
Figure 7 Evaluated settlement distribution without improvement
4. PROPOSED REMEDIATION
In order to reduce the settlement to the acceptable settlement, it was
Figure 5 Finite element mesh used proposed to install additional piles with longer length in between the
existing piles. The additional piles were to be driven pile with pre-
drilling to 6 metres depth. This was planned to ensure that no piles
were damaged while penetrating the very dense layer from 2.5 to 5.5
m depth. Several configurations were attempted to optimize the
design, while meeting the settlement criterion. Only the final
configuration is presented in this paper.
Figure 8 shows the final proposed layout of additional piles. One
center pile and four ‘rings’ of additional piles were designed. The
additional piles were designed to be 0.6 m diameter. The center pile
and innermost ring of piles (1st ring: shown by blue diamond in
Figure 6) had a depth of 47 m, i.e. founded on the very dense sand
layer. The innermost ring of piles consisted of 12 piles and were
located 5 m from the center of the foundation. The second innermost
ring of piles (2nd ring: shown by green squares) consisted of 24, 40-
m long piles to be located 10 m from the center of the foundation.
The second outermost ring of piles (3rd ring: shown by blue
diamond) had the same depth as innermost ring of piles, i.e. 47 m.
Figure 6 Zoomed in view of foundation piles and pile cap
The second outermost ring of piles consisted of 36 piles, located 15
m from the center of the foundation. The outermost ring of piles (4th
The limiting shaft resistance was also calculated based on
ring: shown by orange triangle) had a depth of 30 m and consisted
Decourt (1982), whereby he gave the correlation
of 48 piles. This gave a total of 121 additional piles.
The embedded beam properties of the additional piles were
fs = 3.3 x N kPa (5)
evaluated using the methods laid out in section 3.2.3. The numerical
procedure was identical to those written in section 3.2.4.
where fs is the unit shaft resistance and N is the SPT values at
certain pile depth.
231
Figure 8 Plan view of the oil tank with additional piles installed
Figure 9 shows the settlement contour of oil tank with the their design capacity. Additional soil investigations revealed that the
additional piles. The maximum and minimum settlement were 90 previous soil investigation was not accurate. PLAXIS3D was used
mm and 80 mm respectively, with the maximum settlement (shown to re-evaluate the oil tank design, and remediation was suggested to
in red) located at the corners of the tank and minimum settlement meet the initial design criterion.
(shown in blue) located at the second outermost ring of piles. With From the analysis conducted, in order to satisfy the settlement
the added piles, the settlement was reduced from 330 mm to 90 mm. criterion of 100 mm, another 121 piles would be needed, in addition
The differential settlement was also reduced from 100 mm to 10 to the existing 304 piles. Further analysis shows that by using only
mm. 121 piles, the settlement criterion can be met. In fact, without the
existing piles, the 121 piles would perform better and settle less than
5. EVALUATING EFFECTIVENESS OF ADDITIONAL 425 piles. This shows that more piles do not necessarily means
PILES ONLY better performance, optimization is required.
This case also highlights the importance of good and accurate
To evaluate the effectiveness of ‘additional’ piles, additional run soil investigations to an effectiveness of geotechnical design. A
was conducted without the existing piles (i.e. the oil tank is to be good practice is to have the design consultant to supervise the driller
founded only on the 121 piles). to countercheck the borelog. This practice is common in countries
Figure 10 shows the settlement contour of the oil tank, if it were with established geotechnical practice, e.g. Australia. Unfortunately,
to be founded only on the ‘additional’ 121 piles. It can be seen that such practice is rarely practiced in Indonesia. Another important
the settlement magnitude is similar, in fact the oil tank founded on thing to note about soil investigation are the numbers of soil tests.
the 121 piles without the existing 304 piles had slightly lower Many project owners belittle the importance of soil investigation
settlement than the one with existing 304 piles (refer to Figure 9). and try to save cost by limiting the number of soil tests. However,
This is due to pile group efficiency (Poulos, 1989). For the analysis the cost saved by reducing the soil tests is often insignificant
with existing piles included, the center to center spacing between 2 compared to the potential cost saving from design optimization. In
adjacent piles were as close as 1.25 m, which is only twice the worst case scenario, the lack of information can lead to failures, and
diameter of ‘additional’ piles. Whereas, if the existing piles did not lead to heavy losses in terms of money and time. It is an engineer’s
exist, the piles would have 2.5 m spacing, which is more than 4 duty to inform project owners the importance of soil investigation,
times the diameter. Therefore, the ‘additional’ piles would be more as it is the very basis of a geotechnical design.
efficient without the existing piles, and hence less settlement. Due to the urgency and heavy cost to remediate the oil tank, the
project owner decided to build the tank as it is. Another reason is
6. CONCLUSIONS AND DISCUSSION that the oil tank can be built to accommodate the settlement. The
In this paper, an oil tank to be founded on 304 piles encountered accommodation can be done by jacking up the oil tank after its
problem during pile installation. Problems encountered include piles settlement, and the foundation can be releveled (Wit, 1991).
not reaching target depth, damaged piles, and piles not reaching Furthermore, before the oil tank operation, hydrostatic test (Morera
232
and Makar, 2009) would be conducted, whereby the oil tank would 4. REFERENCES
be filled to its full capacity with water. Fortunately, the oil tank is
founded on sand, hence the settlement will occur very quickly, and Bowles, J. E. (1977). Foundation Analysis and Design. McGraw-
the releveling can be conducted after the hydrostatic test. Little or no Hill Companies, United States.
settlement is expected after the hydrostatic test, as oil has lower Brinkgreve, R. B. J., Kumarswamy, S., Swolfs, W. M., and Foria, F.
density than water, i.e. the operational weight of oil tank is lower (2017). PLAXIS 2017 Manual. Plaxis bv, Netherlands.
than the hydrostatic test. Decourt, L. (1982). Prediction of the bearing capacity of piles based
In conclusion, a good and accurate soil investigation is required exclusively on N values of the SPT. Proceedings of ESOPT2,
for any construction. Without a good soil investigation, the Amsterdam, 1, 29-34.
foundation design will not be optimum, and cost more than a design Ishihara, K. (1985). Stability of Natural Deposits During
with proper soil investigation. From this case study, one can also Earthquakes. Proceedings of the Eleventh International
learn that remediation may not be necessary, even though the initial Conference on Soil Mechanics and Foundation Engineering,
design criteria is not met. It is dependent on the structure, as well as 1, San Francisco, 321–376.
soil condition. However, careful evaluation need to be conducted Irsryam, M., Widiyantoro, S., Natawidjaja, D. H., Meilano, I.,
before making the decision. Rudyanto, A., Hidayati, S., Triyoso, W., Hanifa, N. R.,
Djarwadi, D., Faizal, L. and Sunarjito (2017). Peta Sumber
Bahaya Gempa Indonesia Tahun 2017. Pusat Penelitian dan
Pengembangan Perumahan dan Permukiman.
Morera, J. and Makar, G. (2009). Specification for Tank Hydrotest.
TransCanada.
Peck, R. B., Hanson, W. E. and Thornburn, T. H. (1974).
Foundation Engineering. 2nd edition, Wiley, New York, 113-
115.
Poulos, H. G. (1989). Pile behaviour – theory and application.
Géotechnique, 39(3), 365–415.
Seed H.B, K. Tokimatsu, L.F. Harder, and Riley M. Chung (1985).
Influence of SPT Procedures in Soil Liquefaction Resistance
Evaluations. Journal of Geotechnical Engineering, ASCE,
111(12), 1425-1445.
Stroud, M. (1989). The SPT – Its applications and interpretation,
Penetration testing in the UK. Thomas Telford, London.
Figure 9 Evaluated settlement distribution with additional piles Terzaghi, K. (1943). Theoretical Soil Mechanics. Wiley, New York.
Wit, J. (1991). Safe jack up method permits repair of tank bottoms
and foundations. Oil & Gas Journal, 89(44).
Figure 10 Evaluated settlement distribution with only 121 piles
233
Dynamic Analysis of Underwater Tunnels

Akhila Babu, S.D. Anitha Kumari and Nimmy Mariam Abraham
Department of Civil Engineering, M. S. Ramaiah University of Applied Sciences, Bangalore, India
E-mail: nimmy555@gmail.com
ABSTRACT: Tunnels have become an important mode of transportation due to increase in population and density of traffic all over the
world. An underwater tunnel is a tunnel which is partly or fully constructed under a body of water. Underwater structures are not absolutely
immune against dynamic waves and are prone to damages. Thus it is significant to perform the dynamic analysis of underwater tunnels. In
the present study, the dynamic behaviour of underwater single and twin tunnels are studied numerically with different tunnel materials. Study
includes static and dynamic analysis of twin tunnel models with varying overburden, varying position along depth and varying spacing. The
dynamic analysis is done adopting the load data of Kutch earthquake. It is observed that twin tunnels with varying overburden developed
maximum stresses when compared with uniform overburden.
Keywords: Underwater Twin tunnels, Dynamic analysis, Numerical modelling.
1. INTRODUCTION literature review indicates that the studies reported on dynamic

analysis of underwater tunnels especially twin tunnels are limited.
Tunnels are underground passageways, which is dug through the The study is focused to analyze the response of underwater single
surrounding soil/earth/rock. It can be for foot or vehicular road and twin tunnels subjected to seismic loading followed by a
traffic, rail traffic or a canal. An underwater tunnel is a tunnel which parametric study under different conditions including depth from
is partly or fully constructed under a body of water. They are often soil bed, spacing between tunnels, location of tunnels and materials
used where building a bridge or operating a ferry link is impossible, used for tunnel.
or to provide competition (or relief) for existing bridges or ferry
links. The underwater tunnels are of two types: (i) underwater tunnel
which is immersed under the water and is influenced only by 2. NUMERICAL MODELLING OF UNDERWATER
buoyancy and hydrostatic pressure (ii) underwater tunnel which TUNNELS
passes below the bed of the water body and is influenced by De et al. (2017) studied the response of a tunnel subjected to blast
hydrostatic pressure and overburden pressure. The underground loading through a combination of physical model tests utilizing a
systems help to reduce the cost and project completion time when geotechnical centrifuge and numerical model analyses. The model
the acquisition of land is nearly impossible or moving surface tunnel was 760 mm long with an outer diameter of 76 mm and wall
utilities are expensive. Generally underground structures are safer thickness of 2.5 mm. In all centrifuge tests the soil cover above the
than surface structures against dynamic waves. This is because tunnel crown was kept constant at 38 mm. The centrifuge model
surface structures are only connected to the ground from the lower consisted of a copper tunnel built to a 1:70 scale and it was tested at
surface and vibrate free whereas underground structures are 70g acceleration. This represents a prototype with a length of 53 m,
completely connected to the surrounding environment and thus are an outside diameter of 5.3 m, wall thickness of 0.133 m. The
more resistant to vibration. In spite of this, there are reports of flexural stiffness (EI) of the tunnel was 13 × 106 kNm2. The
damage to such structures due to dynamic waves. The long-term dimensions and the properties of water, soil and tunnel material for
stability of tunnels is one of the most important factors in the design the current study are taken from this reported study. The dimensions
and implementation of underground spaces. Since tunnels and of the tunnel model is shown in Table 1. Nevada soil properties is
underground spaces can be subjected to dynamic loads like used for the study and the same is tabulated in Table 2. Since the
earthquakes and explosions during construction and operation tunnel is below the soil bed, the overburden pressure is calculated
stages, dynamic stability analysis of such structures is of great with the saturated density of the soil bed. The relative density of the
importance. soil is reported as 60%.
Collapse of a number of underground structures has
occurred during various earthquakes (1995 Kobe, Japan earthquake, Table 1 Dimensions of the tunnel model (De et.al, 2017)
1999 Chi-Chi, Taiwan earthquake and 1999 Duzce, Turkey
Variable Value (m)
earthquake). During the 1999 Duzce earthquake in Turkey, Bolu
highway twin tunnels experienced a wide range of damage. Based Length 53
on these collected details, it is understood that the study on the Outer diameter 5.3
behavior of underground/underwater tunnels is of high significance. Wall thickness 0.133
Moreover, the recently inaugurated Mumbai-Ahmedabad Soil cover 2.7
Shinkansen high speed rail route has around 7km undersea tunnel. Water cover 2.7
Hence this study is focused on the dynamic behavior of twin tunnels
subjected to earthquake loading. Table 2 Properties of Nevada sand
Boldini et.al (2010) conducted one dimensional (1D) and
two dimensional (2D) fully coupled finite element simulation to Variable Value
investigate the dynamic behaviour of circular tunnels in the Maximum dry density 17.33 kN/m3
transverse direction. Tiwari et.al (2016) reported parametric Maximum dry density 13.87 kN/m3
sensitivity studies adopting coupled Eulerian-Lagrangian (CEL) Maximum void ratio 0.887
analysis tool in finite element software. Navarro (1992) reported Minimum void ratio 0.511
seismic design of tunnels adotping FEM analysis by using numerical Specific gravity 2.67
tool FLUSH. Anastasopoulos et.al (2007) used finite-element
method to perform nonlinear dynamic transient analysis of the After fixing the dimensions, numerical modelling was done by using
tunnel. Anitha Kumari et.al (2014) modeled twin tunnels with the CATIA software. The 3-D model developed in CATIA software
circular and horse-shoe cross sections to understand the effect of the was imported into Hyper mesh software for meshing. Models were
shape of the tunnel and varying shallow overburden on the developed for single and twin tunnels. The model is divided into a
mechanical response when subjected to seismic loading. The number of elements and nodes. 3-D hexahedral meshing was done
234
using the software. Hexahedral or brick elements are solid elements

which are extracted from 2D Quad elements. Hyper mesh builds
around 8 to 20 noded hexa elements. The 3-D view of the twin
tunnels modeled is shown in Figure 1. For the general cases, the
optimum spacing between the tunnels is taken as 4 times diameter
(4d) of the single tunnel. Soil is modeled using an eight noded
element with three degrees of freedom at each node including
translations in nodal x, y and z. This element type is referred to
SOLID 185. There are two forms of SOLID 185 elements which
are classified as homogeneous structural solid and layered structural
solid. Since soil is considered as homogeneous, the element Figure 3 Stresses in concrete twin tunnels
considered is homogeneous structural solid. 3.2 Dynamic analysis of underwater tunnels
Kutch earthquake data, shown in Figure 4 was considered as the
load for dynamic analysis.
Figure 4 Kutch earthquake data (M. Shrikhande et.al, 2001)
In order to analyse the twin tunnel for seismic load,

Figure 1 3-D view of the twin tunnel model
random vibration analysis was adopted. Through a Fourier analysis,
3. ANALYSIS OF UNDERWATER TUNNELS the non-periodic acceleration data can be converted into a series of
many overlapping sine waves, with each curve cycling at its own
The CATIA model was imported into ANSYS WORKBENCH for frequency and amplitude. Even though, acceleration amplitude
static and dynamic analysis. The model is confined from all the five constantly changes in the time-history data, the average value of all
directions and the pressures due to soil and water is applied onto the the amplitudes within a given frequency range can be determined to
model corresponding to the depth of soil cover and water obtain the power spectral density (PSD) function. The PSD function
respectively. was given as input to finite element software and the stress values
3.1 Static analysis of underwater tunnels on the twin tunnel were obtained with its probability of occurrence
in a linear elastic framework.
A static structural analysis determines the displacements, stresses, Figure 5 shows the stresses developed in the twin tunnels
strains, and forces in structures or components caused by loads that when subjected to the given earthquake load. The probability of
do not induce significant inertia and damping effects. Steady loading occurrence is 68.3%. The results of analysis of different cases with
and response conditions are assumed. After applying the overburden various materials indicate that the maximum deformation occurs
pressure, static analysis is performed on the underwater tunnel along the spring line whereas the maximum stresses are observed at
model made of copper, steel and concrete. In this case, the spacing the crown of the tunnel.
between the tunnels is maintained as 4 times the diameter of the
tunnel and the soil overburden is kept uniform. Figure 2 and 3 show
the deformation and stress in concrete twin tunnels respectively
when subjected to static loading. The material property of the tunnel
model was changed to copper and steel and the analysis was done
for the same boundary conditions and loading conditions.
Figure 5 Stresses along the twin tunnels during dynamic loading

3.3 Parametric study on the behaviour of twin underground
Figure 2 Deformation in concrete twin tunnels tunnels
In real situations, the height of soil cover or water depth need not be
uniform. To understand the effect of these variations, different cases
are studied. The effect of free water depth and the thickness of soil
cover on the seismic behaviour of tunnels is studied. In addition to
235
this, the effect of spacing between the tunnels in the lateral direction
as well as in the vertical direction is also analysed.
Figures 6 and 7 represent the models for different overburden cases.

It can be seen that the depth of soil cover and the depth of water
above the tunnels are different. In twin tunnel model with varying
overburden, the spacing between the tunnels is maintained as 4
times the diameter of the tunnel. The stresses around the tunnels are
calculated by considering the varying depth of the soil as well as
water.
Figure 8 Stress in twin tunnel with constant and varying overburden:

Static analysis
Figure 6 Twin tunnels subjected to varying overburden
Figure 9 Stress in twin tunnel with constant and varying overburden:

Dynamic analysis
Figure 7 Twin tunnels located at different vertical depths

The results of the static and dynamic analyses of twin tunnels with
constant and varying overburden are shown in Figures 8 and 9. The
maximum stress values in the twin tunnel model with varied
overburden increased by about 7.3% when compared with tunnels
subjected to uniform overburden. Deformation in copper twin tunnel Figure 10 Stress in twin tunnels with 2d and 4d spacing: Dynamic
analysis
model with varied overburden tunnel is 19.6% less than that of
concrete twin model with uniform overburden. The dynamic
response was studied for three different materials and the maximum
stress values were seen in copper twin tunnels. When compared the
behaviour between concrete and steel, stresses developed in tunnels
made of concrete is found to be marginally less compared to steel
when uniform overburden is considered.
The influence of lateral spacing between the tunnels was studied by
applying earthquake load. Figure 10 shows the variation of stress
when the spacing between the tunnels is 2d and 4d. There is an
increase in stress by 4.2% for the copper twin tunnel model with 2d
spacing in comparison with the twin tunnel model with 4d spacing.
When the tunnels are located at two different depth levels from the
uniform soil bed, it was observed that the increase in stress was
Figure 11. Stress in twin tunnels when located at same depth
around 40.89% for the copper twin tunnel model as shown in Figure
from uniform soil bed and at different depths: Dynamic analysis
11. The stress increase was found to be in the range of 40% for the
tunnels with 2d spacing compared to 4d spacing.
5. CONCLUSION
In this study, the behaviour of single and twin tunnels buried below
the water bed is analysed under static and dynamic loading
236
conditions. Parametric studies involving the effect of varying Anirban De, Anthony Niemiec, S.M., and Thomas F. Zimmie.
overburden, spacing in lateral direction and depth of tunnels on the (2017), “Physical and Numerical modelling to study
seismic response of twin tunnel models are carried out. Twin tunnels effects of an underwater explosion on a buried tunnel”,
with varying overburden indicated that the maximum stresses Journal of Geotechnical and Geoenvironmental
developed was more for the varying overburden situation when Engineering, 143(5), https://doi.org/10.1061/(ASCE)
compared with uniform overburden and can be quantified to an
GT.1943-5606.0001638.
increase of 88% for the case considered. The results of dynamic
analysis indicate that the twin tunnels located in varying overburden Boldini,D., Amorosi, A., and Misano, F.P. (2010), “Analysis of
terrain showed an increase of 7.3% in the stress values when Tunnel Behaviour Under Seismic Loads by Means of
compared with the tunnels located below uniform overburden. Simple and Advanced Numerical Approaches”,
When the tunnel models are subjected to dynamic loading with 2D International Conferences on Recent Advances in
and 4D spacing, concrete showed significant increase in stresses Geotechnical Earthquake Engineering and Soil Dynamics,
compared to the other two models. (6), pp.1-12.
Bazaz, J. B, and Besharat, V. (2008), “An Investigation on Seismic
6. REFERENCES Analysis of Shallow Tunnels in Soil Medium”, 14th
World Conference on Earthquake Engineering, 50, pp.
Anastasopoulos, I., Gerolymos, N., Drosos, V., Kourkoulis, R.,
368-382.
Georgarakos, T. and Gazetas, G. (2007), “Nonlinear
Response of Deep Immersed Tunnel to Strong Seismic Navarro, C. (1992), “Seismic analysis of underground structures”,
Shaking”, Journal of Geotechnical and Geoenvironmental Earthquake Engineering Tenth World Conference
Engineering, 133(9), pp.1067-1090. Balkerna, Rotterdam, 60, pp. 480-496.
An, J., Tuan, C.Y., Cheeseman, B.A., and Gazonas, G.A. (2011), Shrikhande, M., Basu, S., Kumar, A., Das, J., Thakkar, S.K. and
“Simulation of Soil Behaviour under Blast Loading”, Chandra,B. (2001), “Analysis of Strong Motion Data of
International Journal of Geomechanics, 142(9), pp.1053- Bhuj Earthquake of January 26”, Journal of applied
1064. Geophysics, pp.324-338.
Anitha Kumari, S. D., Vipin, K. S., and Sitharam, T. G. (2014), Tiwari, R., Chakraborty, T., and Matsagar, V. (2016), “Dynamic
“Effect of shape of twin tunnels during seismic loading”, Analysis of Underground Tunnels subjected to Internal
Journal of Rock Mechanics and Tunnelling Technology, Blast Loading”, 11th World Congress on Computational
20(1), pp 49-59. Mechanics, pp.1-12.
237
3D finite-element analysis on behaviours of pile group and piled raft foundation

models subjected to cyclic horizontal loading
T. V. Anh1, M. Tatsunori2
1
Le Quy Don Technical University, Vietnam
2
Kanazawa University, Japan
E-mail: matsumoto@se.kanazawa-u.ac.jp, vuanhtuan@mta.edu.vn
ABSTRACT: In this research, the authors carry out a numerical study through three-dimensional finite element method to investigate the
behaviours of pile foundations subjected to cyclic horizontal loading. Three-pile pile foundation models (with or without batter piles) and
six-pile pile foundation models (with or without batter piles), which were used in the experiments by the authors, are considered in the
numerical analyses. The foundations work as pile group foundations if the raft base is not in contact with the ground surface, while they are
piled rafts if the raft base is in contact with the ground surface. The foundations are modelled as linear elastic. Interface elements are
employed to simulate the slippage between the foundations and the soil. In this study, the hypoplastic model, an incrementally non-linear
constitutive model, is used to model the ground. The analysed results indicate that the piled rafts have higher horizontal resistance than the
corresponding pile groups and the horizontal resistance of the foundations are improved by inclusion of batter piles. The results also show
that the resistance of six-pile pile foundations is not equal two times the resistance of the corresponding three-pile pile foundations due to the
influence of interaction.
Keywords: Piled raft, pile group, numerical analysis, cyclic loading
1. INTRODUCTION
40 80 80 40 40 80 80 40
Pile foundations including pile group and piled raft foundations are
usually applied to structures subjected to large horizontal load, such
40 40
40 40
as bridges, wind-turbine towers or offshore structures. Horizontal 1 2 3 1 2 3
loads acting on these structures are wind load and/or water wave
load and can be considered as cyclic load.
Piled raft has been increasingly used as an effective foundation 240 240
type to reduce average and/or differential settlement, e.g. Poulos et
30
30
al. (2011), and Yamashita et al. (2011). Experimental studies as well
as numerical analyses on piled raft foundations having vertical piles 80 80 80 80
alone have been conducted, e.g. Randolph (1994), Horikoshi et al.
(2003), Matsumoto et al. (2004), Reul (2004), Unsever et al. (2014),
1.1 1.1
255
Hamada et al. (2015), Vu et al. (2017), and Vu et al. (2018).

It seems that the behaviours of pile foundations under cyclic 15
horizontal loading have not been fully understood. Hence, the
authors carry out a numerical study through three-dimensional finite
20
(in mm) 20
element method to investigate the behaviours of pile foundations
subjected to cyclic horizontal loading.
P1 P2 P3 P1 P2 P3
2. DESCRIPTION OF EXPERIMENTS (a) 3PG and 3PR (b) 3BPG and 3BPR
The experimental study has been presented in Vu et al. (2017). 40 80 80 40 40 80 80 40

However, it is appropriate to describe the outline of the experiments
40
40
here again for comparisons with the numerical analyses.

4 5 6 4 5 6
80
80
2.1 Pile foundation models

Figure 1 shows the foundation models used in the experiments. The 1 2 3 1 2 3
40
40
foundation models consist of 3 piles or 6 piles (with or without

batter piles). They are pile groups (3PG, 3BPG, 6PG and 6BPG) if
240 240
the raft base is not in contact with the ground surface, while they are
piled rafts (3PR, 3BPR, 6PR and 6BPR) if the raft base is in contact
30
30
with the ground surface.

Close-ended aluminium pipes having a total length of 285 mm, 80 80 80 80
an outer diameter of 20 mm and a wall thickness of 1.1 mm were
used for the model piles. The upper 30 mm of the pile is embedded
1.1 1.1
255
in the raft, resulting in the effective length of 255 mm. Centre-to- 15

centre pile spacing, s, is 80 mm, 4 times the pile diameter. The
inclination angle of the batter piles is 15 degrees. Young's modulus (in mm)
of the piles, Ep, was estimated from bending tests of the piles. The 20 20
geometrical and mechanical properties of the model pile are
summarised in Table 1. In order to obtain axial forces, bending P1 P2 P3 P1 P2 P3
moments and shear forces in the model piles during load tests, strain P4 P5 P6 P4 P5 P6
gauges were arranged on the pile shafts (Figure 2). The piles were (c) 6PG and 6PR (d) 6BPG and 6BPR
covered with the silica sand particles in order to increase the shaft
resistance. Figure 1 Dimensions of the foundation models
238
static horizontal load was applied at the raft in longitudinal direction

of the raft by means of winches and pulling wires. The horizontal
30
30
50
50
load was measured by 2 load cells (LC-R and LC-L) arranged in the
level 1 right (positive) direction and in the left (negative) direction. Both
the horizontal and vertical displacements of the foundations were
35 40 40 40 40 40
level 2 measured by horizontal and vertical dial gauges.
120
level 3
285
VDG-L VDG-R
level 4
80
level 5 Winch
Winch
LC-R
level 6 LC-L
35
HDG
20 (in mm) 20
(a) P1, P2 and P3 (b) P4, P5 and P6 Figure 3 Experiment setup with measuring instruments
Figure 2 Model piles with strain gauge instrumentation
3. FEM MODELLING
The rafts were made of duralumin with the dimensions as shown
in Figure 1 and can be regarded as rigid. The sand particles were 3.1 FEM simulations of the triaxial tests
adhered on the raft base surface to increase the friction between the A series of triaxial CD tests of the sand having a relative density Dr
raft and the ground during horizontal loading. = 82% were conducted under different confining pressures, p0 = 7,
Table 1 Geometrical and mechanical properties of the model pile 17, 27 and 50 kPa, in order to obtain the mechanical properties and
and the raft to investigate the behaviour of the sand. To select an appropriate soil
model and to evaluate the soil parameters, simulations of the triaxial
Property Value tests were carried out, prior to the analyses of the load tests. The
Outer diameter, D (mm) 20.00 experimental and simulation results of the triaxial tests are shown in
Wall thickness, t (mm) 1.1 Figure 4 (Vu et al., 2018).
Length from raft base, L (mm) 255 250
Cross section area, A (mm2) 65.31
Deviatoric stress, q (kPa)
Solid: experiment
Moment of Inertia, I (mm4) 2926.2 Open: simulation
200
Young’s modulus of the pile, Ep (N/mm2) 70267 p0= 50 kPa
Young’s modulus of the raft, Er (N/mm2) 68670
Poison’s ratio, v 0.31 150
p0= 27 kPa
2.2 Model ground 100 p0= 17 kPa
The soil used for model ground in this study is a dry silica sand
50 p0= 7 kPa
having the properties shown in Table 2. The model ground with a
relative density, Dr, of about 82% (d = 1.533 t/m3) was prepared in
a soil box having dimensions of 800 mm in length, 500 mm in width 0
0 2 4 6 8 10 12 14 16 18 20
and 530 mm in depth. In order to control the density of the model
Axial strain, a (%)
ground, the model ground was prepared by 11 layers (10 layers of
50 mm and 1 layer of 30 mm). In each layer, the sand was poured (a) Deviatoric stress q versus axial strain a
and compacted by tapping so that the target relative density of 82% -12
Solid: experiment
was achieved.
Volumetric strain, vol (%)
-10 Open: simulation

Table 2 Physical properties of silica sand #6 p0= 7 kPa
-8
Property Value -6
p0= 17 kPa p0= 27 kPa
Density of soil particle, s (t/m3) 2.668 p0= 50 kPa
Maximum dry density, dmax (t/m3) 1.604 -4
Minimum dry density, dmin (t/m3) 1.269 -2
Maximum void ratio, emax 1.103
Minimum void ratio, emin 0.663 0
Relative density, Dr (%) 82.0 2
Dry density, d (t/m3) 1.533 0 2 4 6 8 10 12 14 16 18 20
Axial strain, a (%)
2.3 Loading method (b) Volumetric strain vol versus axial strain a
Figure 4 Experimental and simulation results of triaxial CD tests of
Figure 3 shows a photo of the experiment setup with measuring
the sand
instruments. Vertical load was applied by placing lead plates of
about 600 N and 1200 N on the raft in the cases of 3-pile In this research, the hypoplastic model (Wolffersdorff, 1996)
foundations and 6-pile foundations, respectively, in order to having the parameters shown in Table 3 was employed to model the
simulate the dead weight of the super structure. After that, cyclic sand.
239
Table 3 Parameters of the hypoplastic model 4. RESULTS AND DISCUSSIONS

Property Value Figure 6 shows the relationships of horizontal load, H, and
c (deg.) 31 normalised horizontal displacement, u/D, in the cases of the 3-pile
hs (kN/m2) 2x106 foundations. Both the experimental and FEM results indicate clearly
n 0.28 that the piled rafts have much higher horizontal resistances than the
ed0 0.663 corresponding pile groups. It is also seen that the resistances of the
ec0 1.100 foundations are effectively improved by inclusion of batter piles in
ei0 1.200 both cases of piled raft (BPR) and pile group (BPG).
 0.12 1000
 1.2 800 3PG Exp.
mR
Horizontal load, H (N)

5 3BPG
600 3PR
mT 2
Rmax 0.5×10-4 400 3BPR
r 0.5 200
 0.5 0
pt (kN/m2) 3 -200
e 0.739 -400
-600
3.2 FEM modelling of loading tests -800
Numerical analyses were carried out using a three-dimensional FEM -1000
-1.2 -0.8 -0.4 0.0 0.4 0.8 1.2
program, PLAXIS 3D. Only a half of the foundation and the ground
was modelled owing to symmetric conditions. Figure 5 shows an Normalised horizontal displacement, u/D
example of the finite element mesh of the modelling. (6a) Experimental results
1000
3PG FEM
800
3BPG
600 3PR
800 250
400 3BPR
200
530
(mm) 0
-200
-400
-600
-800
Figure 5 Finite element mesh -1000
-0.4 -0.2 0.0 0.2 0.4
Boundary conditions are applied as follows: Normalised horizontal displacement, u/D
- Vertical model boundaries with their normal in x-direction (i.e. (6b) FEM results
parallel to yz-plane) are fixed in x-direction (ux = 0) and free in y- Figure 6 Horizontal load vs. normalised horizontal displacement for
and z-directions. 3-pile pile foundations
- Vertical model boundaries with their normal in y-direction (i.e.
parallel to xz-plane) are fixed in y-direction (uy = 0) and free in x- Similar results are also obtained in the cases of the 6-pile pile
and z- directions. foundations, in which the piled rafts have much higher horizontal
- The model bottom boundary is fixed in all directions (ux = uy = resistances than the corresponding pile groups and the resistances of
uz= 0). the foundations are enhanced by inclusion of batter piles, as shown
- The ground surface is free in all directions. in Figure 7. It is seen from the above results that the FEM
calculations simulate the experimental results very well.
The raft and the piles were considered as linear elastic materials.
2000
In order to model the pile, a hybrid model in which beam element 6PG Exp.
surrounded by solid elements was used, according to Kimura and 1600 6BPG
Zhang (2000). The raft was modelled by using solid elements. 1200 6PR
Interface elements of Mohr-Coulomb type were assigned at the 800 6BPR
raft base (in the cases of the piled rafts) and along the pile shafts. 400
Interface cohesion was set as 0, and the interface friction angle was 0
set as 40.2˚ following Unsever et al. (2015). The analysis procedure
is as follows: -400
Step 1: Self-weight analysis of the model ground alone, where -800
K0= 1-sin ( is internal friction angle of the soil) was -1200
assumed. -1600
Step 2: Setting the foundation in the ground, and self-weight -2000
analysis including the foundation. -1.2 -0.8 -0.4 0.0 0.4 0.8 1.2
Step 3: Analysis of loading process using displacement control Normalised horizontal displacement, u/D
manner. (7a) Experimental results
240
2000 7
Inclination of the raft,  (deg.)

6PG FEM 6
1600 Positive (+)
6BPG 5 6PG
1200 6PR 4
800 6BPR 3
2 6BPG
400 1
0 0
-400 -1
-2
-800 -3
-1200 -4
-5
-1600 -6
-2000 -7
-0.4 -0.2 0.0 0.2 0.4 -1.2 -0.8 -0.4 0.0 0.4 0.8 1.2
Normalised horizontal displacement, u/D Normalised horizontal displacement, u/D
(7b) FEM results (10a) by experiments
Figure 7 Horizontal load-normalised horizontal displacement for 6-
pile pile foundations 1.2

1.0 6PG FEM
Figure 8 shows comparisons of horizontal load vs. normalised 0.8 6BPG
horizontal displacement between 6PG and 2×3PG, and between 6PR 0.6
and 2×3PR at the initial loading stage according to the experimental 0.4
results. Similarly, the FEM results are shown in Figure 9. It is seen 0.2
from both experimental and FEM results that the horizontal 0.0
resistances of the 6-pile foundations (6PG and 6PR) are smaller than -0.2
two times the resistances of the 3-pile foundations (2×3PG and -0.4
-0.6
2×3PR), in which the difference of resistance between 6PR and
-0.8
2×3PR is more prominent than that between 6PG and 2×3PG.
-1.0
Obviously, the influence of interaction between the raft and the piles -1.2
through the ground is indicated from the results. -0.4 -0.2 0.0 0.2 0.4
Normalised horizontal displacement, u/D
1600
2 x 3PG Exp. (10b) by FEM
1400 2 x 3PR 2 x 3PR
1200
6PG Figure 10 Inclination of the raft of 6PG and 6BPG during cyclic
6PR horizontal load the experiments and FEM
1000
6PR
800 7
2 x 3PG
600 6
Positive (+) 6PR
6PG 5
400 4
200
3
2
0 1 6BPR
0.00 0.05 0.10 0.15 0.20 0.25 0
Normalised horizontal displacement, u/D -1
Figure 8 Horizontal load vs. normalised horizontal disp. curves for -2
-3
6PG, 6PR, 2×3PG and 2×3PR (Experimental results) -4
-5
1600
-6
FEM -7
1400 2 x 3PG 2 x 3PR -1.2 -0.8 -0.4 0.0 0.4 0.8 1.2
1200 2 x 3PR
6PG Normalised horizontal displacement, u/D
1000 6PR 2 x 3PG (11a) by experiments
800 6PR
1.2
600 6PR
6PG 1.0 FEM
6BPR
400 0.8
200
0.6
0.4
0 0.2
0.00 0.05 0.10 0.15 0.20 0.25
Normalised horizontal displacement, u/D 0.0
-0.2
Figure 9 Horizontal load vs. normalised horizontal disp. curves for -0.4
6PG, 6PR, 2×3PG and 2×3PR (FEM results) -0.6
-0.8
Figure 10 shows comparisons of inclination of the raft during -1.0
cyclic horizontal loading between 6PG and 6BPG by the -1.2
experiments (Figure 10a) and FEM (Figure 10b). Figure 11 shows -0.4 -0.2 0.0 0.2 0.4
the corresponding results of 6PR and 6BPR. It is indicated from Normalised horizontal displacement, u/D
both experimental and FEM results that the inclination of raft (11b) by FEM
increases almost linearly with the increase of normalised horizontal
displacement in all cases, and the inclination is suppressed by Figure 11 Inclination of the raft of 6PR and 6BPR during cyclic
inclusion of the batter piles. horizontal load by the experiments and FEM
241
Figure 12 and Figure 13 show the relationship between the 15

inclination of the raft and horizontal load during the initial loading FEM
Bending moment, M (N.m)

stage for 6PG, 6BPG, 6PR and 6BPR according the experiments 10 P1 in 3PR
(Figure 12) and FEM results (Figure 13). The numerical results are
in a good agreement with the experimental results, indicating that 5
the inclinations of the piled rafts are smaller than those of the
corresponding pile groups at any given horizontal load, and the 0
inclination of the foundations is effectively reduced by the inclusion Level 1
-5 Level 2
of batter piles. It is worth to notice that the piled raft with batter
Level 3
piles is the most favourable foundation type to minimize the -10 Level 4
inclination induced by horizontal loading. Level 5
Level 6
1600 -15
-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3
6PG Exp.
6BPG Normalised horizontal displacement, u/D

1200 6PR 6BPR (a)
6BPR
6PR 15
FEM

800 6BPG
10 P2 in 3PR
6PG
400 5
0
0 Level 1
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 -5
Level 2
Inclination of the raft,  (deg.) Level 3
-10 Level 4
Figure 12 Inclination of the raft vs. horizontal load during the initial Level 5
loading stage for 6-pile foundations (Experimental results) -15
Level 6
-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3
1600
6PG FEM 6BPR
(b)
6BPG
1200 6PR
6BPR 6PR 15
Level 1
Level 2
6BPG 10 Level 3
800
Level 4
5 Level 5
6PG
Level 6
400 0
-5
0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 FEM
-10
P3 in 3PR
Figure 13 Inclination of the raft vs. horizontal load during the initial -15
loading stage for 6-pile foundations (FEM results) -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3
Figures 14 and 15 show the numerical results for 3PR and (c)
3BPR, respectively, in which changes of bending moments with
normalised horizontal displacement, u/D, at different levels (see Figure 14 Bending moments of piles for 3PR (FEM)
Figure 2) of each pile during horizontal loading are given. Note that
P3 is the front pile and P1 is the rear pile for positive loading, and 15
vice versa for negative loading. These numerical results are in good FEM
agreement with the experimental results presented in Figures 16 and 10 P1 in 3BPR

17 (Vu et al., 2017) as follows:
As for the piled rafts without batter piles (3PR), the largest 5
magnitudes of bending moments in the front piles and in the centre
piles are similar, and higher than those in the rear piles. The 0
magnitudes of bending moments in the centre piles are similar Level 1
-5 Level 2
between positive loading and negative loading. The maximum Level 3
bending moments occur at the top of the piles (level 1). Level 4
-10 Level 5
It is obvious to see from the result of 3BPR that significantly
Level 6
larger bending moments are generated in the vertical centre piles
-15
(P2) compared with the other piles (P1 and P3). The bending -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3
moments in P2 of 3BPR are also considerably larger than those in Normalised horizontal displacement, u/D
P2 of 3PR.
(a)
242
15 35
FEM 30 Level 1

25 Level 2
10 P2 in 3BPR
20 Level 3
15 Level 4
5 10 Level 5
5 Level 6
0 0
-5
-10
-5 -15
-20
-10 Level 1 Level 4 -25 P3 in 3PR
Level 2 Level 5 -30
Level 3 Level 6 -35
-15 -1.2 -0.8 -0.4 0.0 0.4 0.8 1.2
-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3
Normalised horizontal displacement, u/D Normalised horizontal displacement, u/D
(b) (c)
Figure 16 Bending moments of piles for 3PR (Experiment)
15
35
Level 1 Level 4 FEM 30

10 Level 2 Level 5
Level 3 Level 6 P3 in 3BPR 25 P1 in 3BPR
20
5 15
10
0 5
0
-5 Level 1
-5 -10 Level 2
-15 Level 3
-10 -20 Level 4
-25 Level 5
-30 Level 6
-15 -35
-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 -1.2 -0.8 -0.4 0.0 0.4 0.8 1.2
(c) (a)
Figure 15 Bending moments of piles for 3BPR (FEM)
35
30
25 P2 in 3BPR
35 20
30
P1 in 3PR 15
25 10
20 5
15 0
10 -5
5 -10 Level 1
0 -15 Level 2
-5 Level 1 -20 Level 4
-20 Level 4 -35
-25 Level 5
-1.2 -0.8 -0.4 0.0 0.4 0.8 1.2
-30 Level 6 Normalised horizontal displacement, u/D
-35
-1.2 -0.8 -0.4 0.0 0.4 0.8 1.2 (b)
(a) 35
30 Level 1
25 Level 2
35 20 Level 3
30 15 Level 4
25 P2 in 3PR 10 Level 5
20 5 Level 6
15 0
10 -5
5 -10
0 -15
-5 -20
Level 1 -25
-10 Level 2 P3 in 3BPR
-15 -30
Level 3 -35
-20 Level 4 -1.2 -0.8 -0.4 0.0 0.4 0.8 1.2
-25 Level 5
-30 Level 6 Normalised horizontal displacement, u/D
-35
-1.2 -0.8 -0.4 0.0 0.4 0.8 1.2 (c)
Normalised horizontal displacement, u/D Figure 17 Bending moments of piles for 3BPR (Experiment)
(b)
243
5. CONCLUSION
A numerical study through three-dimensional finite element method
was carried out to investigate the behaviours of pile foundations
subjected to cyclic horizontal loading and the results were compared
with the corresponding experimental study.
The analysed results indicate that the piled rafts have higher
horizontal resistance than the corresponding pile groups and the
horizontal resistance of the foundations are improved by inclusion
of batter piles. It is worth to notice that the piled raft with batter
piles is the most favourable foundation type to minimize the
inclination induced by horizontal loading.
The results also show that the resistance of six-pile pile
foundations is not equal two times the resistance of the
corresponding three-pile pile foundations due to the influence of
interaction.
6. REFERENCES
Hamada, J., Tsuchiya, T., Tanikawa, T., and Yamashita, K. (2015)
“Lateral loading tests on piled rafts and simplified method to
evaluate sectional forces of piles”. Geotechnical Engineering
Journal SEAGS & AGSSEA, 46(2), pp29-42.
Horikoshi, K., Matsumoto, T., Hashizume, Y., Watanabe, T., and
Fukuyama, H. (2003) “Performance of piled raft foundations
subjected to static horizontal loads”. Int. Jour. of Physical
Modelling in Geotechnics, 3(2), pp37-50.
Kimura, M., and Zhang F. (2000) “Seismic evaluations of pile
foundations with three different methods based on three-
dimensional elasto-plastic finite element analysis”. Soil and
Foundation, 40(5), pp113-132.
Matsumoto, T., Fukumura, K., Pastsakorn, K., Horikoshi, K., and
Oki, A. (2004) “Experimental and analytical study on behaviour
of model piled rafts in sand subjected to horizontal and moment
loading”. International Journal of Physical Modelling in
Geotechnics, 4(3), pp1-19.
Poulos, H.G., and Small, J.C., Chow, H. (2011) “Piled raft
foundations for tall buildings”. Geotechnical Engineering
Journal SEAGS & AGSSEA, 46(2), pp78-84.
Randolph, M. F. (1994) “Design methods for pile groups and piled
rafts”. Proc. 13th ICSMGE, Vol. 5, New Delhi, pp. 61-82.
Reul, O. (2004) “Numerical study of the bearing behaviour of piled
rafts”. International Journal of Geomechanics, 4(2), pp59-68.
Unsever, Y.S., Matsumoto. T., Shimono, S., and Ozkan, M.Y.
(2014) “Static cyclic load tests on model foundations in dry
sand”. Geotechnical Engineering Journal SEAGS & AGSSEA,
45(2), pp40-51.
Unsever, Y.S., Matsumoto, T., and Ozkan, M.Y. (2015) “Numerical
analyses of load tests on model foundations in dry sand”.
Computers and Geotechnics, 63, pp40-51.
Vu Anh-Tuan, Matsumoto, T., Kobayashi, S., and Shimono, S.
(2017) "Experimental study on pile foundations having batter
piles subjected to combination of vertical and horizontal loading
at 1-g field", Geotechnical Engineering Journal of the SEAGS &
AGSSEA, Vol. 48(3), pp12-24.
Vu Anh-Tuan, Matsumoto, T., Kobayashi, S., and Nguyen Tuong-
Lai (2018) "Model load tests on battered pile foundations and
finite-element analysis", Int. Journal of Physical Modelling in
Geotechnics, 18(1), pp33-54.
Wolffersdorff, P.A. (1996) “A hypoplastic relation for granular
materials with a predefined limit state surface”, Mechanics of
cohesive-frictional materials, 1, pp251-271.
Yamashita, K., Yamada, T., and Hamada, J. (2011) “Investigation of
settlement and load sharing on piled rafts by monitoring full-
scale structures”. Soils and Foundations, 51(3), pp513-532.
244
Finite Difference Analysis of Raft Foundations under Vertically Static Loads

Der-Wen Chang1 and Hsin-Wei Lien2
1
Professor, Department of Civil Engineering, Tamkang University, New Taipei City, Taiwan
2
Graduate Research Assistant, Department of Civil Engineering, Tamkang University, New Taipei City, Taiwan
E-mail: dwchang@mail.tku.edu.tw
ABSTRACT: This paper introduces the Finite Difference formulas derived from the thin-plate theory on raft foundation with the influences
from boundary conditions. The proposed solution was established for a raft located at the ground surface, where the bending moments and
shears along the edges of the foundation were presumed vanished. The solutions were examined with the three-dimensional Finite Element
analysis on numerical foundation subjected to uniformly load. Computer program WERAFT-S was developed for raft foundation on elastic
soils under vertically static load. Effectiveness of a couple of soil spring models was discussed with the validation of WERAFT-S to three
dimensional Finite Element solutions.
Keywords: Finite difference analysis, raft foundation, plate theory, boundary conditions, soil springs
1. INTRODUCTION into account of the effects of boundary conditions when the

foundation is relatively large and flexibility of raft became
The load-displacement relations of raft foundations can be modeled significant. In contrast with the one suggested by Bowles (1977),
by two-dimensional or three-dimensional analyses. Both modeling FDA based on FLAC program (Itasca 2017) was established in
can be done by the Finite Element Analysis (FEA). For two- examining the stress continuities of the material particles. Applying
dimensional analysis, the foundation can be treated as a one- them to the nodes of the foundation requires considerable amount of
dimensional beam on a set of soil springs. The analysis is termed as iterations to ensure the equilibriums.
Beam on Elastic Foundation or Winkler foundation where the soil This study proposed a three-dimensional FDA for a surface raft
springs can be elastic or inelastic. This type of solution is applicable foundation subjected to vertically static loads with boundary effects.
when the length-to-width ratio (L/W) of the foundation (where L is The governing equations were adopted from the thin plate theory.
the length, W is the width) exceeds 10. The two dimensional Computer program WERAFT-S was developed and verified with
analysis has been discussed for decades (Biot 1937; Mathews 1958; three-dimensional FEM analysis.
Bowles 1977; Ting and Mockry 1984; Jones 1997; Chen 1998;
Tomlinson and Boorman 2001; Dinev 2012; Chiou et al. 2016;
2. METHODOLOGIES
Chang et al. 2016). Among the available solutions, the Finite-
Difference Analysis (FDA) of the foundation can be found by Theory of Plate can be categorized for thin plate and thick plate. In
Bowles (1977), Tomlinson and Boorman (2001) on settlements of general if the thickness of the plate (D) is less than a tenth t of the
the foundation under static loads. Chang et al. (2016) on the other width (W) of plate, it can be regarded as thin-plate. The Kirchhoff-
hand has demonstrated that the one-dimensional beam on soil-pile Love classical plate theory was suggested on thin plates, whereas
elements can be used to simulate the seismic behaviors of the the Mindlin-Reissner plate theory is applicable to the thick plates.
foundation subjected to horizontal ground motions. A recent report The thick plate theory considers the in-plane shear strains whereas
comparing various solutions of the beam analysis can be found in the thin plate theory does not.
Omer and Arbabi (2015).
The three-dimensional analysis can be conducted by treating the 2.1 Governing Equation
raft foundation as a two-dimensional plate (or mat) resting upon the
ground soils. Complexities and versatility are involved in deriving According to Timoshenko and Woinowsky-Krieger (1959),
the governing equations of foundation’s loading behaviors. governing equation of the vertical displacements of a thin plate
Analytical solutions in this regard have been presented (Timoshenko subjected to vertically uniform load (q) and point load (P) can be
and Krieger 1959; Vlasov and Leontev 1966; Kukreti and Ko 1992). written as follows,
However owing to the complexities involved in the analytical  4 w 2 4 w  4 w 12q (1   2 ) 12 P(1   2 ) (1)
expressions, these solutions were rarely used in engineering practice.    
x 4 x 2 y 2 y 4 ED 3 ED 3 (xy)
A numerical solution based on a series of strip footings was then
suggested by Poulos (1991). This solution was later improved by where w is the vertical displacement of the raft,  and E are the
Poulos (1994) with the approximate numerical solutions for plate on Poisson’s ratio and Young’s Modulus of raft, D is thickness of the
soil continuums from boundary integrals. At the present time,
modeling of the 2D raft can be found in many research studies on raft, and x and y are the spatial variables.
piled raft foundation (Randolph 1983; Clancy and Randolph 1993;
Horikoshi and Randolph 1996; Kobayashi et al. 2009; Kitiyodom 2.2 Boundary Conditions
and Matsumoto 2002; Kitiyodom et al. 2005). In general, discrete
foundation can be established using two approaches, 1. Equation of For a raft foundation located at the ground surface as shown in Fig.
Motion, and 2. Wave Equation. The former solves the foundation 1, the moments and shear forces are assumed vanished at edge of the
displacements based on stiffness matrix (or via the mass/damping foundation. By looking at the foundation with plan view, the top and
matrices for dynamic problems) of the foundation. Raft was bottom edges of the raft where y=constant, Mx (bending moment
modeled using two-dimensional finite elements and/or simplified rotating at the x-direction) and Vy (vertical shear force at the surface
grid (beam-column) elements underlain by a set of soil springs. normal to the y-direction) can be written as follows,
The latter requires a governing differential equation of the Mx   B(y 2   2 w / x 2 )  0 (2)
deformational characteristics of the foundation. The representative
solution of such approach can be found in Bowles (1977) where the   3 w ( 2   ) 3 w 
Vy   B  3  0 (3)
yx 2 
two-dimensional Finite Difference Analysis was introduced to solve
the problem. Comprehensive summaries regarding the design and  y
analysis of the raft foundations can be found in Gupta (1997) and
where B is the expression of ED3/(12(1-2)). Similarly, at the left
Hemsley (1998). The FD analysis needs further attentions to take
and right edges of the raft where x=constant, the boundary
245
conditions My and Vx are:

2.4 Finite Difference Formulas
My   B( 2 w / x 2   2 w / y 2 )  0 (4)
Using the central-difference formulas, the resulting formulations for
(4)
the nodes at a surface foundation can be derived. The orientation
  3 w ( 2   ) 3 w  (5) (5) of
and categories of these nodal points are shown in Fig. 2. Details
Vx   B  3  0
 x xy 2  the derivations can be found in Lien (2018). Note that the spacing
distance between the nodes in x- and y- directions are kept the same
(i.e., x=y=s) for simplicity of the expressions. Moreover, the
point load P applied at arbitrary nodes of the raft can be taken as an
extra uniform load applied to that node within the area which is
equal to xy. Fig. 3 shows the nodal points used and the fictitious
points encountered in the derivations.
Figure 1 Layout and discrete nodes of a vertically loaded raft

foundation
Figure 3 Nodes for derivations of the FD formulas, (a)
2.3 Soil Springs General node (b) Edge node
For governing equation of the raft foundation with the influences of

soil reactions, a number of models can be used. For simplicity, the With the discrete equations derived, one can easily establish a
compressional spring model was adopted in this study, in which the set of dependent equations for the nodes allocated at the raft. Matrix
soil spring constant Ks is simply calculated by EsAs/l where Es is the analysis is required to solve for the foundation displacements. When
Young’s Modulus of the soil, As is the effective area of the soils calculating the soil reactions at the nodes along the edges, the area
underneath the nodes of the raft, and l is the length of the soil spring ratio (Ask/Ark) of the soil springs can be represented by a value, n
(or effective thickness of the soils). For a surface foundation, the defined by dividing the length of the area with the standard width
influences of the soil reactions on the raft can be taken into account (which is equal to 1.0 m) of the soils underneath the general nodal
assuming that their influences are similar to the superstructure loads. points. For the nodes at the corners, the area ratio of the soil spring
Therefore in Eq. (1), q can be replaced by q* where q*=q-ΣKswk /Ar would be expanded to n×n/1.0 as n2. The spacing distance between
= q–(Es/l)ΣAskwk /Ar; wk is foundation settlement at the kth node, two adjacent nodes is kept as 1 m for simplicity. Fig. 4 depicts the
Ask is the area of soil spring under the kth node, and Ar is the total area of soil springs for nodes at the edges and the corners. As a
area of the raft which equals to ΣArk, where Ark stands for the area of result, a computer program WERAFT-S was suggested (Lien 2018).
raft at the kth node. Now defining qk* as the modified load allocated
at the kth node, for simplicity, qk* can be approximated as q-
(Es/l)wk(Ask/Ark), where (Ask/Ark) is called as the area ratio at the kth
node.
Figure 4 Schematic layout of the area ratio (As/Ar) in

calculating soil reactions at the nodes
3. VALIDATION
Validation of the proposed analysis is presented next. Assuming that
a square raft has width (W) of 26 m and thickness (D) of 1 m.
Figure 2 Allocations of the corresponding equations of the proposed Young’s modulus (E), Poisson’s ratio (), and unit weight () of the
analysis concrete raft foundation are 3×104 MPa, 0.15, and 24 kN/m3,
This assumption matches closely with the flexible foundations respectively The soils underneath the foundation are assumed to
where the raft size is relatively large. For smaller raft foundation have a shear wave velocity (Vs) of 150 m/s, Poisson’s ratio (s) of
that behaves more rigidly, this assumption should be not applicable. 0.4, and unit weight (s) of 19 kN/m3. Note that Vs2×s/g=Gs, where
One can adopt other types of soil spring model. For example, the g is acceleration of the gravity, Gs is Shear modulus of the soil. Note
Lysmer Analog soil spring can also be used. The total soil spring that Es=2Gs(1+s), therefore Es can be calculated. The uniform load
constant of the whole raft foundation could be calculated as applied on top of the foundation is 100 kPa. Three dimensional FEA
4Gsro/(1s), where Gs is the shear modulus of the soil, s is the solutions of the numerical model were first obtained from Midas-
Poisson’s ratio of the soil, and ro is the equivalent radius of the raft GTS (Midas 2017) varying the dimensions of the FE zone. Essential
foundation. boundaries (i.e., roller and hinge) were adopted in the modeling.
246
Eight-node brick elements were mainly used to generate the 4. EFFECTS OF SOIL SPRINGS
structural system.
Fig. 5 shows that the foundation settlements obtained from FE The influences of effective thickness of the soil spring were studied.
analysis will gradually converge by increasing the FE zone at It was found that the shear wave velocity of the soil is the most
60~200 m (note that the length, width and thickness of the zone important parameter if the compressional spring was adopted. Fig. 7
were kept the same for the analytical zone). The settlements became depicted the results. More details of the discussions can be found in
stable when the analytical zone was increased to 200 m x 200 m x Chang et al. (2018). Fig. 8 shows the matching of the displacements
200 m. Therefore the FEA of the study was conducted using the at centre of the foundation with the FEM ones on numerical model
optimized core. It can be seen that the foundation settlements are varying the shear wave velocity of soil. Corresponding
different at various locations. In general, the center exerts the largest interpretation of the appropriate thickness of the soil springs with
settlement whereas the corner has the smallest settlement. The soil respect to shear wave velocity of the soil can be obtained. Note that
mass involved in the FE analysis seems to have significant influence when Vs is at 120~180 m/s, the effective thickness of the soil spring
on foundation settlements. Validation of the WERAFT-S analysis will be around 32~17 m. The effective thickness of the soil spring
was then conducted to compare with the FE solutions. The seems to be much smaller than what was expected (for square
influences of the area ratio (Ask/Ark=n) of the edge node, the shear foundation, effective depth of the foundation is around 2B). The
reason is because that the force acting along the spring is the same
wave velocity (Vs) and the Poisson’s ratio (s) of soil, the thickness
while the real stress magnitudes of the foundation will gradually
of raft (D), and the load intensity (q) were studied to see the
decrease with the depth.
appropriateness of the length (l) of soil spring.
Figure 5 Influence of FE analytical zone on the foundation

settlements
Figure 7 The effects of shear wave velocity of soil and length of

underneath soil spring on foundation settlements, (a) Vs = 120m/s (b)
Vs = 150m/s (c) Vs = 180m/s
Figure 6 Effect of area ratio at edge nodes on foundation

settlements
Fig. 6 shows the effects of area ratio (n) at the edge nodes and Figure 8 Matching and interpretation of the optimized effective
length of the soil springs (l) from the WERAFT-S analysis on thickness of soil springs underneath the raft foundation
normalized settlements (w/D) appeared at center, edge, and corner
of the foundation. Note that x and y are kept as one meter in this
study. Therefore the raft will have 729 (27×27) nodal points. The
horizontal dash lines represent the settlements from FE analysis.
Note that when n=2.5 and the length of soil spring (l) approximates
21 m seems to provide compatible solutions with the FE analysis.
The matches were found very sensitive for the settlement occurred
at the corner. As the area ratio increases from 2 to 3, the appropriate
length of the soil spring for the corner settlement is increased
rapidly from 17 m to 25 m, where the corresponding ones at the
edge and the centre are varying insignificantly around 18~21 m and Figure 9 Comparisons of displacement contours from Midas-GTS
21 m, respectively. analysis and WERAFT-S analysis
247
Comparisons of the displacement contours of the standard Chang, D.W., Lien, H.W. and Wang Tzuyu (2018) “Finite
model foundation from both WERAFT-S and FEM can be found in difference analysis of vertically loaded raft foundation based
Fig. 9. To show the possible influences of the soil springs using on the plate theory with boundary concern,” J. of
Lysmer Analog model, the results were listed in Table 1. It can be GeoEngineering (under review)
found that the deviations between the WERAFT-S and FEM are Chen, C.N. (1998) Solution of beam on elastic foundation by
smaller when finding the optimized solutions of simplified soils DQEM, Journal of Engineering Mechanics, ASCE, 124(12),
springs. The predictions using Lysmer Analog springs need to pp1381-1384.
further examinations.
Chiou, J.S., Lin, H.S., Yeh, F.Y., and Sung, Y.C. (2016) “Plastic
Table 1 Comparisons on Foundation Displacements Obtained from settlement evaluation of embedded railroads under repeated
FEA and WERAFT-S using Different Soil Spring Models train loading”, Journal of GeoEngineering, Taiwan
Shear w/ Lysmer Geotechnical Society, 11(2) pp97-107.
w/ EA/l Clancy, P. and Randolph, M.F. (1993) ″Simple design tests for piled
wave Analysis Midas Analog
model raft foundations.″ Geotechnique, 36(2), pp169-203.
velocity model
(m/s) location foundation displacement (mm) Dinev, D. (2012) “Analytical solution of beam on elastic Foundation
by singularity functions”, Engineering Mechanics, 19(6)
center 42.3 40.3 28.7 pp381-392.
120 edge 30.0 31.8 21.7 Gupta, S.C. (1997) Raft Foundation Design and Analysis with a
corner 22.5 22.7 13.2 Practical Approach, New Age International Publisher.
center 18.8 18.6 18.6 Hemsley, J.A. (1998) Elastic Analysis of Raft Foundations, ICE
150 edge 13.1 13.7 13.6 pub.
corner 9.7 9.2 7.9 Horikoshi, K. and Randolph, M.F. (1996) ″Estimation of overall
center 10.6 11.1 13.0 settlement of piled rafts.″ Soils and Foundations, 39(2), pp59-
180 edge 7.5 7.9 9.2 68.
corner 5.6 5.1 5.2 Itasca (2017) FLAC Version 8.0, Itasca Consulting Group. Inc.
Jones, M. (1997) Analysis of Beams on Elastic Foundations: Using
Finite Difference Theory, ICE pub.
5. CONCLUDING REMARKS
Kitiyodom, P. and Matsumoto, T. (2002) ″A simplified analysis
This paper presents a newly proposed finite difference analysis method for piled raft and pile group foundations with batter
WERAFT-S for the surface raft foundation under vertical loads piles.″ Int. Journal for Numerical and Analytical Methods in
using the Thin-Plate Theory. The effects of boundary conditions Geomechanics, 26, pp1349-1369.
were considered to derive the formulas of the raft foundation Kitiyodom, P., Matsumoto, T. and Kawaguchi, K. (2005) ″A
settlements. Elastic compressional soil springs were used to simulate simplified analysis method for piled raft foundations subjected
the soil resistances underneath the raft. Solutions of the proposed to ground movements induced by tunneling.″ Int. Journal for
analysis were verified on a numerical raft foundation located at the Numerical and Analytical Methods in Geomechanics, 29,
surface of a ground consisting of elastic soils (where Vs=150 m/s pp1485-1507.
and s=0.4). Vertical uniform load of 100 kPa was mounted on the Kobayashi, H., Nishio, H., Nagao, T. Watanabe, T., Horikoshi, K.,
foundation. Validation of the analysis was made with those from Matsumoto, T. (2009) ″Design and construction practices of
three-dimensional finite element analysis (FEA) using Midas-GTS. piled raft foundations in Japan.″ Proc., Int. Conf. on Deep
The effective thickness of the compressible soil springs underneath Foundations - CPRF and Energy Piles, pp101-135.
the raft was studied. It was found that the effective thickness of the Kukreti, A.R. and Ko, M.G. (1992) “Analysis of rectangular plate
soil at 17~32 m (for Vs at 180~120 m/s) can provide agreeable resting on an elastic half space using an energy approach”,
solutions with the FEA. Although the compressible soil spring Applied Mathematical Modeling, 16(7), pp338-356.
model has some defects in the real applications, it has been shown Lien, H.W. (2018) Finite Difference Analysis of Piled Raft
that the adequate solutions can be still achieved. On the other hand, Foundations under Vertically Loads, Master Thesis, Dept. of
the Lysmer Analog springs seem to provide much smaller Civil Engineering, Tamkang University, Taiwan (in Chinese).
foundation displacements on stiffer site. The results need further Mathews, P.M. (1958) “Vibrations of a beam on elastic foundation”,
studies with the concerns of uneven soil stiffness distributed Journal of Applied Mathematics and Mechanics, 38(3-4),
underneath the foundation. If the raft foundation is rigid enough pp105-115.
where the boundary effects can be neglected, then the solution Midas (2017) Midas GTS NX User Manual, Midas IT Co.
proposed by Bowles (1977) can be used assuming uniform Omer, J.R. and Arbabi, A. (2015) “Evaluation of finite element,
settlement. Vertical loads such as uniform, non-uniform, and point finite difference and elasticity methods for hypothetical raft
loads need to be taken into account with pre-calculations. foundations installed on layered strata”, Geotechnical and
Geological Engineering, 33(4), pp1129-1140.
6. ACKNOWLEDGEMENT Poulos, H.G. (1991) “Analysis of piled strip foundation”, Computer
Methods and Advances in Geomechanics, ed. Beer et al.,
The paper presents the partial results of the research grant MOST Rotterdam, 1, pp183-191.
106-2221-E-032-025-MY2. Finance support is greatly appreciated. Poulos, H.G. (1994) “An approximate numerical analysis of pile-raft
7. REFERENCES Interactions”, Int. J. for Numerical and Analytical Methods in
Geomechanics, 18, pp73-92.
Biot, M.A (1937) “Bending of an infinite beam on an elastic Randolph, M.F. (1983) “Design of Piled Foundations”, Cambridge
foundation”, Journal of Applied Physics, 12, pp155-164. Univ. Eng. Dept., Res. Rep. Soils TR143.
Bowles, J.E. (1977) Foundation Analysis and Design, 2nd ed., Timoshenko, S. and Woinowsky-Krieger, S. (1959) Theory of Plates
McGraw-Hill Companies, Inc. and Shells, 2nd ed., McGraw-Hill. New York.
Chang, D.W., Lee, M.R., Hong, M.Y. and Wang, Y.C. (2016) “A Ting, B.Y. and Mockry, E.F. (1984) “Beam on elastic foundation
simplified modeling for seismic responses of rectangular finite element”, Journal of Structural Engineering, ASCE,
foundation on piles subjected to horizontal earthquakes”, J. of 110(10), pp2324-2339.
GeoEngineering, Taiwan Geotechnical Society, 11(3) pp109- Tomlnison, M.J. and Boorman, R. (2001) Foundation Design and
121. Construction, 7th ed. Addison-Wesley Longman Ltd.
248
Vlasov, V.Z. and Leontev, U.N. (1966) Beams, Plates and Shells on
Elastic Foundation. (Translated from Russian), Israel Program
for Scientific Translation Jerusllem, Israel.
249
250
Determining Unsaturated Soil Properties through Parameter Estimation

Ibrahim1, D.N. Arif2, and N. Gofar3
Department of Civil Engineering, Politeknik Sriwijaya, Palembang, Indonesia
1
2
PT. Adaro Indonesia. Wara Tabalong, Kalimantan Selatan, Indonesia
3
School of Civil & Environmental Engineering, Nanyang Technological University, Singapore
E-mail: ibrahim@polsri.ac.id
ABSTRACT: The stability of slope is known to be affected by rainfall infiltration, especially when groundwater table is relatively deep. Thus,
careful analysis of slope stability requires the determination of unsaturated properties of the soil. Laboratory tests for the determination of
hydraulic properties and unsaturated shear strength are tedious and time consuming. Thus, numerous fitting equations were proposed by
researchers for determination of soil water characteristic curve (SWCC), permeability function and shear strength parameter of the unsaturated
soil. This paper presents derivation of unsaturated soil properties from simple laboratory tests using relevant parameter estimation equations.
The laboratory data include the suction measured using filter paper, and the shear strength parameters obtained from Triaxial UU test on
compacted clayey soil under different gravimetric water contents. Results show that hydraulic properties and unsaturated shear strength
parameters could be derived from available data and the slope stability analysis could be carried out using the estimated properties.
Keywords: Compacted soil; Suction; SWCC; permeability function; rate of strength increase with respect to suction.
1. INTRODUCTION Unsaturated permeability of soil could be measured using

modified Triaxial cell or flexible wall permeameter test (ASTM
Soil near ground surface are naturally found in unsaturated condition, D7664). The coefficient of permeability with respect to water for an
thus the stability of slope is affected by changes in climatic condition, unsaturated soil is a non-linear function of the volumetric water
especially rainfall infiltration (Ching et al. 1984; Toll et al. 1990; content of the soil. When the soil approaches saturation, the
Gasmo et al. 2000; Tsaparas et al. 2002; Rahardjo et al. 2008). Field permeability becomes constant and equal to the saturated coefficient
monitoring of suction response to rainfall infiltration (e.g. Gasmo et of permeability with respect to water, ks. Thus, in the absence of
al. 2000; Tsaparas et al. 2002; Kassim et al. 2012) supported that the laboratory data, the permeability function can be obtained using
suctions could be lost during wet season. Previous study showed that statistical method proposed by Childs & Collis-George (1950) based
the critical rainfall pattern causing slope failure was influenced by the on the saturated coefficient of permeability and the SWCC curve. The
coefficient of saturated permeability of soil and initial moisture procedure for the prediction can be found in Fredlund & Rahardjo
condition. Gofar et al. (2008) showed that the ratio of rainfall (1993).
intensity to the coefficient of saturated permeability of soil Mostly adopted measuring devices for unsaturated shear strength
determines the critical rainfall pattern causing slope instability. of soil are direct shear box (Gan et al. 1988; Schnellmann et al. 2013),
Rahardjo et al. (2008) showed that soil slopes with lower permeability and triaxial cell (Bischop & Donald, 1961; Ho et al. 1982, Rahardjo
requires more antecedent rainfall than residual soil slopes with higher et al. 1995), with several modifications to implement the effect of
permeability to lead to slope failure. In general, five days antecedent suction on soil samples. These procedures were also tedious and time
rainfall could be required to produce the worst pore-water pressure consuming, thus fitting equation was also developed for estimation of
profiles in a slope. un-saturated soil shear strength based on SWCC curve (e.g. Vanapalli
The preceding discussion inferred that soil water characteristic et al. 1997; Goh et al. 2010).
curve (SWCC) and permeability function of unsaturated soil are This paper presents the estimations of SWCC, permeability
required to obtain the variation of negative pore water pressure function and the shear strength of un-saturated soil by employing
(suction) in the soil while the unsaturated shear strength parameter is limited data obtained in laboratory on compacted clayey soil by Arif
needed for slope stability analysis. Field study by Gofar et al (2008) (2005). The data was then analysed using relevant parameter
showed that the range of suction variation is reasonably consistent estimation model to establish the unsaturated soil properties required
with the soil water characteristic curve (SWCC). The shear strength for further analysis for example the stability of a slope subjected to
of the soil is usually high due to the apparent cohesion contributed by rainfall infiltration.
negative pore-water pressure or suction (Fredlund & Rahardjo, 1993).
Once the soil becomes wet due to rainfall infiltration, the suction
2. THEORETICAL BACKGROUND
component of the shear strength was minimized, and the shear
strength decreased significantly. Thus, a detailed analysis of rainfall- A fundamental property that relates the suction and water holding
induced slope stability requires the establishment of hydraulic ability of soil is characterized by the water retention curve, which is
properties (SWCC and permeability function) and unsaturated shear also known as the soil water characteristic curve (SWCC). The
strengths parameters of the soil. SWCC curve can be divided into three regions as shown in Figure 1
Experimental works, despite imposing extra time and expenses, i.e. boundary effect zone, transition zone and residual zone (Fredlund
evidently provide the most appropriate means for measuring the & Rahardjo, 1993). The definition of transition zone, which occurs
properties. Several methods such as hanging column, pressure plate, between the air entry point (a) and residual suction (r) is very
Tempe cell, chilled mirror hygrometer and centrifuge were developed important. Hence; the shape of SWCC curve is characterized by the
to establish the SWCC of the soil. The procedures are standardized in saturated volumetric water content (s) the air entry value (a), and
ASTM D6836. Combination of Tempe cell and Pressure plate test are the residual suction (r) at the corresponding residual water content
the most common method used for determination of SWCC. Besides, (r).
filter paper method (Likos & Lu, 2002) is commonly used to develop Several SWCC equations were reviewed by Sillers & Fredlund
SWCC curve through suction measurements. The procedure is (2001) using a database of laboratory measurements. They concluded
standardized in ASTM D5298. The procedures for defining the points that Fredlund & Xing (1994) equation provided the closest fit to the
to develop SWCC curve were time consuming, thus numerous fitting data sets. The equation is as follows:
equations were developed to form the SWCC curve based on limited
number of laboratory measurements of suction and volumetric water
content (e.g. van Genuchten, 1980; Fredlund & Xing, 1994).
251
s  f  c'   ' tan  '  tan  b (5)

 w  C ( )

m
(1) Theb for undrained condition can be expressed as follows:
     
n

ln e    

   a   

 ( 1   3 )  (   3 ) ( 1   3 )  
 cos  'c' 1  sin  '  tan  '
 2  2 2   (6)
where: w is the volumetric water content (VWC), s is the VWC at tan  
b

saturation, e is the natural logarithmic base number, 2.71828, a, n, m
are fitting parameters related to the air-entry value of the soil in kPa By extending the classical Mohr-Coulomb envelope, the three-
(a), the slope at the inflection point of the SWCC; and residual water dimensional representation of Equation 4 can be visualized in Figure
content (r) respectively;  = (ua–uw) is the matric suction (kPa); and 2.
C() is a correction factor given as follows:
�
ln(1+ � )
�(�) = 1 − �
(2)
ln (1+1000000
� )
�
In which Cr is a constant related to matric suction corresponding to

residual water content. Fredlund & Rahardjo (1993) suggested that Cr
= 1500. Alternatively when all laboratory data were in the low range
of suction value (<1500 kPa), Leong & Rahardjo (1997a) suggested
to use C() =1 in Equation (1). The characteristic values of the
SWCC curve (i.e. air entry value, residual suction and the slope of
SWCC curve at the inflection point) can be calculated using Zhai &
Rahardjo (2012) equations.
Figure 1 Extended Mohr-Coulomb failure envelope for unsaturated
soil (after Fredlund et al, 2012).
Unsaturated shear strength from direct shear tests (Gan et al., 1988)
indicated that the relationship between the shear strength versus and
suction is not linear. However, Rahardjo et al. (1997) suggested it is
possible that the relationship over a selected range of suction is linear.
This is applicable for results of Undrained Unconsolidated (UU)
Triaxial test performed on different water contents (Fredlund et al.,
1978). As the undrained compressive strength (qu) obtained from this
test can be defined as half of the maximum deviator stress (i.e.
qu=(1–3)/2).
Many shear strength equations were developed to predict the rate
of strength increase with respect to suction (b) (Goh et al’ 2010;
Abdullah et al. 2013), among them is the equations proposed by
Vanapalli et al (1997) is as follows:
 f  c'  ( n  uw ) f tan  '  (ua  uw )  K tan  ' (7)
Figure 1 Typical soil water characteristic curve showing zones of
desaturation (Fredlund et al, 2012).
where  is normalized volumetric water content (w/s) and  is an
unknown fitting parameter. This equation is a fitting type of equation
Permeability function can be plotted based on saturated
developed for sand and fine-grained soil. For sand, is equal to one,
coefficient of permeability (ks) and a few data on water coefficient of
but the value increases with soil’s plasticity. The equation was found
permeability (kw) obtained from triaxial test with fitting parameters of
to fit results of UU test by (Gofar & Lee, 2014) for both coarse and
the SWCC curve (a, n, and m). In the absence of data on the water
fine-grained soil for suction less than 100 kPa.
coefficient of permeability, statistical method proposed by Childs &
Goh et al (2010) equation is basically the improvement of
Collis-George (1950) and presented in Fredlund & Rahardjo (1993)
Equation (7). Based on their study on eleven sets of published data as
could be used. Alternatively, equation (3) which took the same form
well as laboratory tests, they found that the relationship between shear
as Equation (1) could be used to predict the permeability function
strength and suction changed at suction equal to air entry value AEV
(Leong and Rahardjo, 1997b).
= (ua–uw)b. therefore:
ks
k w  C ( )  f  c'  ( n  uw ) f tan  '  (ua  uw ) tan  b (8)
n mp (3)
      if (ua–uw)< (ua-uw)b or <AEV
ln e    
   a    f  c'  [( n  u w ) ua  u w b ] tan  ' 
(9)
with p value ranges from 2.4 to 25 depending on the type of soil. The [(ua  u w )  (ua  u w )b ]b  K tan  '
estimation of p value could be made based on the similarity of the
if (ua–uw)≥ (ua-uw)b or ≥ AEV
permeability function with the SWCC.
where
A practical formulation for shear strength of unsaturated soil was
proposed by Fredlund et al. (1978) as follows:  [log (un  uw )  log( ua  uw ) b] y (10)
 f  c'  ( n  u w ) f tan  '  (u a  u w ) f tan  b (4) where b and y are fitting parameters.
This equation was validated with published laboratory data which
where c’ is the effective cohesion; ’ is the effective angle of shearing mostly fine grained soil (classified as CH, CL, MH, ML and SM
resistance; (n–uw)f is the effective normal stress at failure; b is the according to USCS).
rate of change in shear strength relative to change in suction. For ua
equal to atmospheric, Equation (4) can be simplified as:
252
3. METHODOLOGY
The study presented in this paper made use of the available data in
Arif (2005) for the evaluation of unsaturated soil properties required
for slope stability analysis. The soil for his study was retrieved from
Babarsari, Yogyakarta. The data were originally used for evaluation
of bearing capacity of the soil. Index properties of the soil comprising
grain size distribution, Atterberg limits and falling head test for
determining the coefficient of saturated permeability of the soil were
tested in laboratory. Compaction test was carried out to define the
maximum dry density and the optimum water content of the soil.
Advanced laboratory tests were performed on the compacted soil
including suction measurements and shear strength tests.
The SWCC was obtained experimentally by measuring suction of
the soil samples using Filter Paper method following the procedure
outlined in ASTM D 5298. The calibration of the Whatman filter
paper No. 42 follows the following equation: Figure 3 Grain size distribution of the soil evaluated in this study
�= 105.327−0.0779 �� for wf < 45.26 (11a)
3.412−0.0135 ��
Filter paper test was performed on two sets of samples. Each set
�= 10 for wf ≥ 45.26 (11b) comprises four (4) samples were prepared at gravimetric water of
where  is the suction (kPa) and wf is the water content of the filter about 22, 19, 16, and 14% (denoted as sample A, B, C and D). In other
paper. However, direct correlation between the gravimetric water words, two samples were prepared below the optimum water content
content of sample with suction should be derived empirically based while the other two were prepared above the optimum water content.
on a number of tests performed in the study. The corresponding suction for the samples are calculated based on
The shear strength parameters were obtained using conventional empirical correlation developed in this study is as follows:
Triaxial testing apparatus under Unconsolidated Undrained (UU)
condition. Frelund & Rahardjo (1993) suggested that the procedures �= 107 × �0.6519 �� (12)
outlined for conventional Triaxial test on saturated soil can be applied The corresponding suctions for the gravimetric water content for soil
for undrained loading condition of unsaturated soil. In this case, the preparation is shown in Table 1.
suction stress can be linearly correlated with the undrained strength
of soil (Chae et al., 2010). The unsaturated soil specimen is tested at Table 1. Suction corresponding to the water content of soil at
its initial water content or suction. In this case, the initial matric preparation
suction is not relaxed or changed prior to commencing the test by
Sample Gravimetric Volumetric Suction
replacing the porous stone with metal or plastic discs on top and
water water (kPa)
bottom of specimen and by closing the specimen with rubber
content content
membrane during the test. It is assumed that the variation is negligible
A 22 33.26 5.9
despite the suction may decrease due to the increase of pore-water
pressure during shearing stage of the soil sample. Soil specimens (38 B 19 28.73 41.8
mm dia. and 76 mm height) were prepared at the same initial C 16 24.18 295.2
gravimetric water content as for the suction measurement tests. The D 14 21.17 1087.3
shearing rate was controlled at 1.50 mm/min or equivalent to a strain
rate of 0.033% per second for the specimen height of 76 mm. The resulting points from filter paper test were plotted in Figure 4 in
Fredlund & Xing (1994) equation was used to develop SWCC terms of volumetric water content. The laboratory data points were
curve based on the suction measurement data using filter paper test. fitted using Fredlund & Xing (1994) equation to form the SWCC as
The permeability function was plotted using the coefficient of shown in Figure 4. The fitting parameters of the equation a, n, m are
saturated permeability data and the SWCC curve. Subsequently, the 23 kPa, 0.8 and 0.5 respectively. The air entry value (ψa) and residual
rate of shear strength in-crease with respect to suction (b) was suction (ψr) are 8 and 3700 kPa respectively. The saturated volumetric
obtained by applying non-linear failure envelope equation (Vanapalli water content (θs) and the residual volumetric water content (θr) were
et al., 1996 and Goh et al, 2010) on the experimental test results. The 0.37 and 0.15 respectively while the slope of SWCC curve m2w was
performance of both predicted types equation was evaluated for this 0.083. The SWCC was the typical of drying SWCC of silty soil
soil by comparing with the results of Triaxial UU test. (Fredlund and Rahardjo, 1993). The permeability function was
calculated based on ks and the SWCC using statistical method (Childs
4. RESULTS AND ANALYSIS & Collis-George, 1950) and Leong and Rahardjo’s fitting equation is
Following ASTM D2487, the soil used in this study was classified as shown in Figure 5.
low plasticity silt (ML). Specific gravity of the soil was 2.67. The
grain size distribution of the soil is presented in Figure 3. The soil
contains more than 60% fines with 53.5% silt and 10.1% clay
fractions. Liquid limit was 47.8% while plastic limit was 29.2%, thus
the plasticity index of the soil was 11.6. The coefficient of saturated
permeability of the soil is quite low (i.e. 5.6×10-6). Compaction test
results in a maximum dry density of 1.68 Mg/m3 and optimum water
content of 18.4%. Soil samples were prepared at initial relative
density of 90% or dry density of 1.51 Mg/m3.
Figure 4 SWCC curve and the characteristic values.
253
Figure 5 Permeability function.
As for the SWCC test, four samples were prepared at gravimetric

water of about 14, 16, 19, and 22% (denoted as sample A, B, C and
D) for shear strength test under Unconsolidated Undrained (UU)
condition. The results were analyzed and plotted as Mohr circles to
obtain the effective cohesion and effective angle of friction for each
sample. Mohr circle derived from Triaxial UU test on sample B (16%
gravimetric water content) is shown in Figure 6. The figure shows
that the soil prepared at gravimetric water content of 16% had an
apparent cohesion of 68 kPa and effective friction angle of 34o. Figure 7 Rate of increase in shear strength with respect to suction
5. CONCLUSIONS
Fitting and predictive type equations were used in this study to

determine unsaturated soil properties based suction measurements
using filter paper method and shear strength measurements using
simple Triaxial UU test. The samples were prepared under four
different gravimetric water contents. Equations for Whatman No.42
filter paper was used to calculate suction for each sample. Fredlund
and Xing (1994) fitting equation was used to plot the SWCC curve.
Then, permeability curve was obtained using statistical method and
fitting equation by Leong & Rahardjo (1997b) based on the
coefficient of saturated permeability and the SWCC curve. The
characteristic values from SWCC curve are s = 0.37, a= 8 kPa, r
= 0.15 and r = 3700 kPa. The effective strength parameters of the
Figure 6 Mohr circle derived from Triaxial UU test on sample B. soil are given by c’ = 38 kPa and ’= 34o. The shear strength data in
terms of apparent cohesion obtained from Triaxial UU tests were used
to predict the rate of strength increase due to increasing suction ( b).
Results of all tests (sample A, B, C, and D) in terms of apparent
The contribution of suction on the shear strength of the soil decreases
cohesion was plotted in Figure 7a with suction in linear scale and
Figure 7b with suction in logarithmic scale. The unsaturated soil with increasing suction. It varies from b = ’= 34o to 0 at suction 100
shear strength properties was predicted using equation (7) as well as kPa. This means that, for this soil, suction greater than 100 kPa is no
longer a state of stress. Since the range of field suction is less than
equations (8) and (9). Note that equation (8) is used for  < a while
100 kPa, both Goh et al (2010) and Vanapalli et al. (1997) equation
equation (9) was used for  ≥ a. In this case a = 8 kPa. Observation
could be used to estimate the rate of shear strength increase with
of Figure 7b indicated that both equations are in good agreement with
respect to suction.
laboratory data obtained for samples A, B, and C or for suction below
100 kPa. However, when data obtained from sample D was included,
4. REFERENCES
the prediction using Goh et al (2010) equation showed a better
agreement as compared to Vanapalli et al. (1996) equation. Extension Abdullah, NHH. Gofar, N. Rahman, NAA. & Roslan, SM. (2013)
of the data to zero suction indicated that the soil has an effective “Empirical Correlation for Estimation of Unsaturated Soil
cohesion c ’ =38 kPa. The internal friction angle of the soil was Shear Strength”. Proc. IEEE Business Engineering and
Industrial Applications. pp. 690 – 694
estimated based on the results of the shear strength test on samples A
Arif, D.N. (2005) Deformation stress behavior of Unsaturated soil
and B i.e. ’=34o. Thus, based on Figure 7a, the rate of increase of due to Static Load (Perilaku Tegangan Deformasi Tanah Jenuh
shear strength with suction b = ’= 34o for suction less or equal to Sebagian Akibat Beban Statis) (in Bahasa Indonesia). Master
Thesis. Universitas Gadjah Mada.
the air entry value. For suction between 8 and 100 kPa, b = 0.5 ’= Bishop, A. W., and Donald, I. B. 1961. The Experimental Study of
17o while for suction greater than 100 kPa, b = o. Previous Partly Saturated Soil in the Triaxial Apparatus. Proc. 5th Conf.
publications revealed that, in tropical environments, field suction Soil Mechanics and Foundation
rarely increased to more than 100 kPa. Chae, J., Kim, B., Park, S. W. and Kato, S. 2010. Effect of Suction
on Unconfined Compressive Strength in Partly Saturated Soils.
KSCE Journal of Civil Engineering. 14(3):281-290.
254
Ching, R.K.H., Sweeney, D.J., & Fredlund, D.G. (1984) “Increase in Vanapalli S.K., Fredlund D.G., Pufahl D.E, & Clifton A.W. (1996).
factor of safety due to soil suction in two Hong Kong slopes”. Model for the Prediction of Shear Strength with Respect to Soil
Proc. 4th Intl. Symp. Landslides. pp.617-623 Suction. Canadian Geotechnical Journal, (33) pp. 379-392.
Childs E.C., Collis-George G.N. (1950) “The permeability of porous Yunusa, GH., Kassim, A., and Gofar, N. (2014) Effect of Surface
materials”. Proc. Royal Society of London 210A pp. 392-405. Flux Boundary Conditions on Transient Suction Distribution
Fredlund, D.G. & Xing, A. (1994) Equations for the Soil-Water in Homogeneous Slope. Indian Journal of Science and
Characteristic Curve. Canadian Geotechnical Journal, 31 Technology, 7(12): 2064–2075
pp.521-532. Zhai, Q. and Rahardjo, H. 2012. Determination of Soil-Water
Fredlund, D.G., Xing, A. & Huang, S. (1994) Predicting the Characteristic Curve Variables. Computer and Geotechnics,
Permeability Function for Unsaturated Soils Using the Soil- 42: 37-43.
Water Character Curve. Canadian Geotechnical Journal. 31(3)
pp. 533-546
Fredlund, D.G. & Rahardjo, H. (1993). Soil Mechanics for
Unsaturated Soils. New York: John Wiley & Sons, Inc.
Fredlund, D.G., Morgenstern, N.R., & Widger, R.A. (1978) “The
Shear Strength of Unsaturated Soil”. Canadian Geotechical
Journal. 15 pp. 313–321.
Gan, J.K.M., Fredlund, D.G., & Rahardjo, H. (1988) “Determination
of the Shear Strength Parameters of Unsaturated Soil Using
Direct Shear Test”. Canadian Geotech. Journal. 25 pp.500-510
Gasmo, J.M., Rahardjo, H., & Leong, E.C. (2000) “Infiltration effects
on stability of a residual soil slope”. Computers and
Geotechnics, 26 pp. 145–165.
Gofar, N., Lee M.L., & Rahardjo, H. (2015) “Field Monitoring of
Suction Distribution in Coarse-Grained and Fine-Grained
Residual Soil”. Proc. 6th AP-UNSAT2015 pp. 679-683.
Gofar, N. & Lee, ML. (2014) “Unsaturated Shear Strength of Coarse
Grained and Fine Grained Tropical Residual Soil”. Proc. ISGE
Annual Science Meeting PITHATTI XX: 257-262
Gofar N. Lee, M.L. & Kassim, A. (2008) “Response of Suction
Distribution to Rainfall Infiltration in Soil Slope”. Electronic
Journal of Geotechnical Engineering. 13(E) pp.1-13
Goh, S. G., Rahardjo, H., & Leong, E.C. (2010) “Shear Strength
Equations for Unsaturated Soil Under Drying and Wetting”.
ASCE Journal of Geotechnical and Geoenvironmental
Engineering. 136(4) pp.594-606.
Ho, D.Y.F., & Fredlund, D.G. (1982) “A Multi-Stage Triaxial Test
for Unsaturated Soils”. ASTM Geotechnical Testing Journal.
5(1-2)pp.18-25.
Kassim, A., Gofar, N., Lee, M.L., & Rahardjo, H. (2012) “Modeling
of Suction Distribution in an Unsaturated Heterogeneous
Residual Soil Slope”. Engineering Geology 131-132 pp.70-82.
Leong, E.C. & Rahardjo, H. (1997a) “A Review on Soil-Water
Characteristic Curve Equations". ASCE Journal Geotechnical
& Geoenvironmental Eng. 123(12) pp.1106-1117.
Leong, E.C. and Rahardjo, H. (1997b) “Permeability Functions for
Unsaturated Soils”. ASCE Journal Geotechnical &
Geoenvironmental Eng. 123(12) pp.1118-1126.
Likos, W.J., & Lu, N. (2002) “Filter Paper Technique for Measuring
Total Soil Suction”. Transportation Research Record 1786.
Paper No. 02-2140 pp. 120-128
Liu, C. & Evett, J.B. 2005. Soils and Foundation SI Edition by Nurly
Gofar, Prentice Hall.
Rahardjo, H., Leong, E. C., & Rezaur, R. B. (2008) “Effect of
antecedent rainfall on pore-water pressure distribution
characteristics in residual soil slopes under tropical rainfall”.
Hydrological Processes, 22(4) pp.506–523.
Rahardjo, H., Lim, T.T., Chang, M.F., and Fredlund, D.G. 1995.
“Shear Strength Characteristics of a Residual Soil in
Singapore”. Canadian Geotechnical Journal, 32 pp.60 - 77.
Sillers, W.S. and Fredlund, D.G. (2001) “Statistical assessment of
soil-water characteristic curve models for Geotechnical
Engineering”. Canadian Geotechnical Journal. 38(6) pp.1297-
1313
Toll, D.G., Rahardjo, H., & Leong, E.C. (1999) “Landslides in
Singapore”. Proc. 2nd Intl. Conf. Landslides, Slope Stability &
the Safety of Infrastructures. pp.269–276.
Tsaparas, I., Rahardjo, H., Toll, D.G., & Leong, E.C. (2002)
“Controlling parameters for rainfall-induced land-slides”.
Computers and Geotechnics, 29 pp. 1–27.
255
Stochastic estimation of consolidation settlement of upper Pleistocene clay, Ma12,

layer in Osaka Bay with a particle filter method
Shotaro Kubota1 and Kazuhiro Oda2
1
Graduated student, Osaka Univercity
2
Associated Professor, Osaka Univercity
E-mail: skubota@civil.eng.osaka-u.ac.jp
ABSTRACT: There are a number of man-made island in Osaka Bay. In order to maintain and manage the man-made island, we need to
predict consolidation behavior. The Pleistocene clay layer of Osaka Bay has few conservation of ground survey and has characteristic
consolidation characteristics of pseudo overconsolidation. Therefore, it is difficult to predict its settlement behavior only from indoor test
results. Incidentally, in recent years, data assimilation methods have been spreading in various fields. Data assimilation is a method of obtaining
a highly accurate predicted value by correcting a simulation by observation value. In this study, data assimilation was carried out using the
measured value of the pore water pressure for the subsidence amount for the upper Pleistocene clay (Ma12) layer. And we derived Ma12’s
future settlement as a probability distribution.
Keywords: Data assimilation, Particle Filter, Consolidation settlement, Pleistocene clay
1. INTRODUCTION was constructed in the northern part of Osaka Bay, is predicted

through a particle filter method (hereafter, PF), one of the most
The seabed ground of Osaka Bay, Japan, is composed by thick clay popular data assimilation techniques. Moreover, the applicability of
layers and sandy gravel layers, which are accumulated alternately. the particle filter method is discussed from stochastic viewpoints.
Therefore, the significant consolidation settlement of the clay layers
occurs in the construction of man-made islands in Osaka Bay. The
appropriate estimation of consolidation settlement of the man-made 2. DATA ASSIMILATION ANALYSIS
islands is necessary to lay the management plan for facilities on the
man-made islands. PF is a data assimilation method that approximates the probability
distribution on the state of the system with a large number of
The Pleistocene clays in Osaka Bay have been called as "quasi- realization value sets called particles. Each particle has information
overconslidated clays". Their consolidation behaviors are quite on a numerical analysis model and information on physical quantities
distinguished from those of over-consolidated clays due to loading (subsidence amount, pore water pressure, etc.) at each time calculated
and unloading histories. Moreover, the in-situ consolidation by simulation in each model. PF also has various algorithms.
behaviors are different from those given by consolidation tests in the Sequential Importance Sampling (hereinafter referred to as SIS) 4)
laboratory, because of some factors which cannot be eliminated, for which is one of representative ones will be described. In SIS, we
example, the difference of strain rate in the consolidation process. repeat the two calculation steps of one-period prediction and filtering.
That is, it is remarkably difficult to estimate the consolidation Then, we estimate a model with a high degree of conformity with the
settlement of the man-made islands in Osaka Bay based on the measured values. First, in the first-period prediction, we perform
consolidation properties given through consolidation tests and so on. simulations from time t-1 to t. Next, in filtering, we give weight to
each particle by Bayes' theorem according to the degree of conformity
By the way, the data assimilation technique has been developed in between simulation result and measured value. We repeat the two
the weather forecasting, the maritime weather prediction, and so on. steps, taking the sum of the weights calculated at each time step. This
Recently, its application to geotechnical problems has been widely gives more weight to the particles that fit the measured values. As a
spreading. In the data assimilation, firstly, a large number of result, we can estimate models with a high degree of conformity to
numerical simulations is carried out. Then, the numerical simulations actual measurements.
of which the results match with the field observation results should
be selected. After repeating these processes, the analytical results In this research, the simulation model we used is one - dimensional
which could match the field observation results are given specialized soil - water coupling analysis. Figure 1 shows a
stochastically. conceptual diagram of the compression curve used in this study. This
model assumes a one-dimensional consolidation state. Therefore, the
In this study, the long-term consolidation settlement behaviors of compression curve is a general e-log p 'relationship. Therefore, we
the upper Pleistocene clay (Ma12) layer in the Kobe Airport, which need to identify the compression index Cc and over compaction ratio
OCR. We assimilate the three parameters of these parameters and
p0 pc logp the permeability coefficient k which determines the consolidation
rate. Table 1 shows the range of each parameter set for analysis. We
set these ranges considering variations in the soil test results. In the
analysis, we created 700 particles with parameters randomly
1
Cc
2756cm=68.9cm×40 element
e ：settlement amount
：pore water pressure
Figure 1 the conceptual diagram of the compression curve
：non drainage boundary
Table 1 the range of each parameter set for analysis ：free drainrage bounfdary
k (cm /s) Cc OCR
M axim um m agnification 1.50 1.40 1.04
M inim um m agnification 0.67 0.67 0.96 Figure 2 the one-dimensional analytical model
256
determined within this range. Also, we assumed that the error of the Kuroku clay layer (Ma12). And we derived the probability
measured values follows the normal distribution. In addition, we distribution of future sinking amount of Ma12. As a result, the
decided that standard deviation is 1 (cm) for settlement amount and settlement amount of Ma12 at 2051 can be represented by the
0.3 (kgf/cm2) for pore water pressure. In addition, we decided that the probability distribution with an average of 115.38 cm and a standard
error covariance of subsidence and pore water pressure is 0. deviation of 16.07 cm.
6. REFERENCES
3. MEASURED VALUE USED FOR DATA ASSIMILATION 1) T. Higuchi,
In this study, we used the data of settlement and pore pressure 2) S. Ito, K. Oda, K. Koizumi, Y. Usuki,
measured at Ma12 (layer thickness of about 27m) of Osaka Bay
Miocene clay from 2002 to 2014 as actual measurement values used 3) T. Shuku, A. Murakami, S. Nishimura, K. Fujisawa, K. Nakamura,
for data assimilation. Figure 2 shows the one-dimensional analytical Data assimilation of the Settlement Behavior of Kobe Airport
model used in this study. The red circle on the analysis model shows Constructed on Reclaimed Land Using the Particle Filter, Journal of
the measurement position of the settlement amount (1 depth). The applied mechanics, Vol.72, No.4, pp.354-367, 2016.
blue circle shows the measurement position of pore water pressure (9 4) Doucet, A., Godsill, S. and Andrieu, C.: On sequential Monte Carlo
depth, 0, 2, 4, 6, 13, 20, 22, 24, 26 m from the top). sampling methods for Bayesian filtering, Statistics and Computing,
Vol. 10, pp.197-208, 2000 .
4. DATA ASSIMILATION RESULT AND PREDICTION OF 1999/12/6 2006/10/10 2013/8/14

FUTURE SETTLEMENT AMOUNT
0
Settlement amount（cm）
Figure 3 shows the temporal change of measured values of settlement date
amount and settlement amount identified by data assimilation. In
addition, we decided the settlement amount that weighed average was 20
identified from the weight distribution obtained after data
assimilation. Also, Figure 4 shows the probability distribution of the
subsidence amount after data assimilation at 2014. And, Table 2 40
shows weighted average and standard deviation of the probability
distribution. The probability distribution of the settlement amount,
which we have derived is a weighted average 67.15cm, the 60 Actual m easurem ent value
probability distribution of the standard deviation of 4.2cm. W eighted average
W eighted average±σ
Using this result, we derive the probability distribution of the 80
subsidence amount of Ma12 layer about 40 years later. Figure 5
Figure 3 Change with time of settlement amount
shows the temporal change of the predicted value of the settlement
amount up to 2051 and the actual measurement value of the settlement 0.4
amount. Figure 6 shows the probability distribution of the settlement
amount at 2051. Table 3 shows the average of the predicted values of
the settlement amount at 2051 and its standard deviation. The 0.3
Probability
probability distribution of the amount of settlement of the Ma12 layer

at 2051 we found is a probability distribution with an average of 0.2
115.38 cm and a standard deviation of 16.07 cm.
0.1
5. CONCLUSION
0
In this study, we performed data assimilation based on measured
100
40
45
50
55
60
65
70
75
80
85
90
95
values of settlement amount and pore water pressure for Osaka Bay
Settlement amount (cm )
Figure 3 Weighted average and Standard deviation at 2051 Figure 4 probability distribution of the subsidence amount
after data assimilation at 2014
2051 W eighted A verage S tandard D eviation
Table 2 weighted average and standard deviation at 2014
S ettlem entA m ount 115.38 16.07
W eughted A verage S tandard D eviation
(cm )
A fter D ata A ssim ilation 67.15 4.20
0.2 1 A ctual M esurem ent V alue 70.90
Cumulative distribution
(cm )
Probability distribution 0.8
0.15 1999/12/6 2013/8/14 2027/4/23 2040/12/30 2054/9/8
Probability
0
0.6 date
20 Actual measurement value
0.1 Weighted average
0.4 40 Weighted average±σ
0.05 60
0.2
80
0 0 100
120
124
134
105
115
144
153
163
66
76
85
95
140
Figure 6 Probability distribution of settleatment amount 2051 Figure 5 Change with time of settlement amount to 2051
257
Volcanic Cohesive Soil Behaviour under Static and Cyclic Loading

W. O. Sumartini1, H. Hazarika1, T. Kokusho2 and S. Ishibashi3
1
Department of Civil Engineering, Kyushu University, Fukuoka, Japan
2
Chuo University, Tokyo, Japan
3
Department of Survey and Analysis, Nihon Chiken Co. Ltd., Fukuoka, Japan
E-mail: waode_sumartini@yahoo.co.id
ABSTRACT: The objective of this study is to evaluate the behavior of reconstituted samples of a volcanic soil under static and cyclic
loading using series of undrained static and cyclic test. The static test results show that at a low confining pressure, the soil exceeds
contractive behavior followed by dilative behavior. This behavior indicates that the pore water pressure development is not higher than the
soil shear strength. Otherwise, the soil shows contractive behavior. The cyclic test results show cyclic mobility behavior under an
investigated cyclic stress ratio. In one hand, at low cyclic stress ratio, the shear strain increased slowly, and after a certain number of cyclic, it
significantly increases. In another hand, the shear strain increases gradually at high cyclic stress ratio. These results indicate a contradictive
behavior of the soil under different confining stress and cyclic stress ratio.
Keywords: Static triaxial test, cyclic triaxial test, volcanic soil, reconstituted samples, soil behavior.
1. INTRODUCTION in this paper). This Orange soil deposit, which acted as the slip
surface of the landslide, is located on the top of Pre Takanoobane
Historically, in Japan, some damages triggered by an earthquake on Lava pumice deposit (referred as Blackish soil). In one hand, from
volcanic soil areas has been reported (Hazarika et al., 2018; Song et the map of resistivity distribution of the slope (Fig. 6) that has been
al., 2017; Kazama et al., 2012; Miyagi et al., 2011; Sassa, 2005). drawn by Kochi et al. (2018), it can be concluded that the Orange
One of those area is Aso caldera in Kumamoto prefecture. The area soil which is located on the ground waterbed, is in saturated
suffered from widespread landslides which were triggered by the condition. In another hand, the resistivity map also shows that the
2016 Kumamoto earthquake (Fig. 1). The landslides were Blackish soil is an impermeable deposit. Based on these fact, the
concentrated in the Mount Aso area, within a 64 km radius of the authors presume that the Orange soil deposit was liquefied during
epicenter. The 2016 Kumamoto earthquake is series of earthquakes the earthquake and becomes the main reason of the occurrence of
which struck Kumamoto Prefecture of Kyushu Island on 14–16 the landslide.
April 2016. The foreshock earthquake occurred at 21.26 JST on 14
April 2016 at an epicentral depth of ~11 km at a magnitude (Mw) of
6.5 and the main shock struck at 01.25 JST on 16 April at an
epicentral depth of ~10 km at an Mw of 7 (USGS). The source of
the earthquakes was the activity of the Hinagu and Futagawa faults
(GSI). These two faults experienced more than 2 m of strike-slip
displacement at shallow depth. One of the landslides occurred near
Aso Volcanological Laboratory of Kyoto University (Fig. 2). This
landslide brought damage to houses (Fig. 3), public spaces and roads
(Fig. 4). The inclination of the slope is about 10–15° Kochi et al.
(2018) which is consistent with the value of 12° found by Song et al.
(2017). In those landslides, several volcanic soil deposits have been
found scattered on those slope. Sumartini et al. (2017) reported that
the slope is composed of volcaniclastic deposits (Fig. 5) which
visually have different colors and characteristics. The volcanic soil
deposits came from different places as listed in Table 1. Song et al.
(2017) and Kochi et al. (2018) stated that the landslide is composed
of Kusasenrigahama pumice tephra beds (referred to as Orange soil
Figure 2 A massive landslide near Aso Volcanological Laboratory
of Kyoto University
Figure 1 Map showing the landslide distribution by size within

a 64 km radius of the Aso caldera (Sourced from GSI) Figure 3 Swept away houses
258
Several studies related to the Orange soil were done and reported
by Sumartini et al., (2017-2018). They studied chemical, mineral
and microstructure characteristics as well as the behaviour of the
Orange soil under static and cyclic loading. According to those
studies, the landslide occured because the earthquake breakout the
soil structure of the Orange soil deposit and led to liquefaction on
that deposit. Several researchers have been studied the behavior of
those volcanic soils due to earthquake by conducting triaxial cyclic
tests (Ishikawa et al. in 2011, Suzuki and Yamamoto in 2004,
Hatanaka et al. in 1985 and Sumartini et al. from 2017 to 2018).
However, the behaviour of the Orange soil under cyclic loading in
disturbed condition has not been done yet. Even though, the
landslide occurred during the main shock, the deposit is not
completely in undisturbed condition due to the foreshock. Thus, it is
necessary to understand the behaviour of the Orange soil in
disturbed condition. For that reason, series of undrained cyclic
triaxial tests were performed to evaluate the behaviour of disturbed
samples under cyclic loading. Finally, the results were compared
with previous study and presented in this paper.
Figure 5 Schematic profile of the slope in Aso caldera (Modified
from Sumartini et al. 2017)
Table 1 Origin of Volcanic Soil in Aso Saldera (Kochi et al., 2017).

Deposit Origin Age (Cal ka)
Black soil Organic (OL) 10-present
Aso Central Cone Pumice
Brown soil 7.3-10
(AC)
Kikai Akahoya Ash
Dark brown soil 7.3
(K-Ah)
Light brown and
Otogase Lava Pumice
grayish soil with 29-7.3
(Otp)
sand
Light brown soil Aira Tn (Atn) 29
Kusasenrigahama Pumice
Orange soil 31
(Kpfa)
Takanoobane Lava
Blackish soil 51±5
Figure 4 Damage to roads Pumice (Tp)
Figure 6 Resistivity distribution of the slope in Aso Volcanological Laboratory (Kochi et al., 2018)
259
2. MATERIAL PROPERTIES
The Orange soil is taken from the scarp of the slope near Aso
Volcanological Laboratory. This soil contains about 60 percent of
fine particles (Fig. 7) and based on its properties (Table 2) it can be
classified as volcanic cohesive soil type II. It also contains 97 % by
weight of feldspar mineral (Table 3) and has vesicular fabric. The (a) Petal-type (b) Flower type
fabric is composed of crystal flakes (Fig. 8). Sumartini et al. (2018b) Figure 9 Idealized crystal flake structures found in the deposits of
idealized the flakes of the volcaniclastic deposit of the slope into the Aso Volcanological Laboratory Landslide. (a) Petal-type
flower type (Fig. 9a) and petal-type (Fig. 9b). structure and (b) Flower type structure (Sumartini et al., 2018b)
3. TESTING METHOD
Undrained cyclic triaxial tests were conducted on the Orange soil
with 60 kPa of confining pressure. The disturbed samples were
reconstituted from the undisturbed samples which have been tested.
About 100 mm in height and 50 mm in diameter of the specimen
have been produced using a steel mould. Double negative pressure
and appropriate back pressure were applied to the samples and
isotropically consolidated at the target effective pressure. B-values >
0.95 was ensured for all samples before shearing. The frequency of
the cyclic axial load was 0.1 Hz for the undrained triaxial cyclic
tests. To decide whether liquefaction occurred in this study or not,
the pore water pressure ratio (ru), define as the ratio of the pore
water pressure to the normal stress, was used. When ru ≥ 0.95, the
specimen was considered to have liquefied.

Figure 7 Grain Size distribution of the Orange soil (Sumartini et al.,
4.1 Soil behavior under static loading
2017)
Figures 10 and 11 show respectively the stress-strain relation of
Table 2 Physical Properties of Orange Soil (Sumartini et al., 2017) disturbed and undisturbed samples in undrained static triaxial test.
For disturbed samples, the deviatoric stress is rising gradually with
Physical Properties Orange Soil the progress of axial strain for each confining pressure (Fig. 10).
Specific Gravity 2.24-2.38 However, at low confining pressure, the peak is reached at large
Dry Density, g/cm3 0.51-0.58 strain while at high confining pressure the peak is reached at small
Wet Density, g/cm3 1.23-1.30 strain. For undisturbed samples, the deviatoric stress also is
Water Content, % 54.62-58.36 increasing gradually with the progress of axial strain, but when it
Liquid Limit, % 113.40 reaches the peak with a specific strain, the deviatoric stress
Plastic Limit, % 88.25 decreases (Fig. 11).
Plasticity Index 25.15
Table 3. Physical Properties of Orange Soil (Sumartini et al., 2018a)

Orange soil (Percent by
Contents
weight)
Albite 57
Bytownite 40
Sodium hydrogen sulfide 2.0
Calcium copper germanium oxide 1.4
Figure 10 Stress versus axial strain of disturbed samples (Sumartini

Figure 8 A vesicular structure of Orange Soil fabric (Sumartini et al., et al., 2018a)
2017)
260
Figure 11 Stress versus axial strain of undisturbed samples Figure 13 Pore water pressure versus axial strain of undisturbed
(Sumartini et al., 2018a) samples (Sumartini et al., 2018a)
Figures 12 and 13 display respectively the pore water pressure- The behavior of the disturbed and undisturbed samples can be
strain relation of disturbed and undisturbed samples in the undrained clearer by looking at the stress path of the soil as respectively shown
static triaxial test. In low confining pressure, the pore water pressure in Figures 14 and 15. In one hand, under a low confining pressure,
is lower than the deviatoric stress and in high confining stress, the the disturbed sample is contracted and then dilated with no sign of
pore water pressure is higher compared to deviatoric stress. temporary liquefaction. In contrary, under low confining pressure,
Consequently, when the confining pressure is low, the soil the undisturbed sample is dilated, and when reaching the peak of the
susceptibility to liquefaction is low. Otherwise, under high confining soil strength, the soil contracted. In another hand, under high
pressure, the soil has a high susceptibility to liquefaction. confining pressure, disturbed and undisturbed samples show a
contraction which means that they have a high susceptibility to
liquefaction.
Figure 12 Pore water pressure versus axial strain of disturbed Figure 14 Stress path of the disturbed Orange soil samples
samples (Sumartini et al., 2018a) (Sumartini et al., 2018a)
261
Table 4 Strength Parameter of the Orange Soil
Strength parameter Disturbed Undisturbed

Ccu (kPa) 17 60
φ (°) 11 7
c’ (kPa) 11.5 27.5
φ’ (°) 29 30
4.2 Soil behavior under cyclic loading

Figures 18 and 19 display the response of disturbed samples while
Figures 20 and 21 show the response of undisturbed samples in the
undrained cyclic triaxial test. The stress path trend of disturbed
samples is similar to undisturbed sample although the disturbed
sample appears faster in liquefying. The effective stress path of both
samples indicates that the effective stress in undisturbed samples
tends to decrease at a lower rate than in disturbed samples. The
Figure 15 Stress path of the undisturbed Orange soil samples generation of excess pore water pressure for each cyclic load
(Sumartini et al., 2018a) application is much faster for the disturbed samples compared to
undisturbed samples. Therefore, it can be concluded that the
Figures 16 and 17 show the Mohr’s stress circle of disturbed and deformation of the soil structure due to reconstituting process has a
undisturbed samples separately. The cohesion and the angle of the significant effect on the pore water pressure development. Thus, the
shear strength of total stress and effective stress are listed in Table 4. disturbed samples appear to be more susceptible to liquefaction. The
For disturbed samples, the cohesion of effective stress is about 1.48 liquefaction susceptibility of disturbed samples and undisturbed
times the total stress, and the angle is about 0.34 times the total samples are presented in Figure 22.
stress. For undisturbed samples, the cohesion of effective stress is
about 2.18 times the total stress while the angle is about 0.23 times a b
the total stress. Finally, by comparing the cohesion and the angle of
both type of samples, it can be concluded that the reconstitution
process is reducing the cohesion by about 3.53 times for total stress
and about 2.40 times for effective stress while the angle increases
about 1.57 for total stress and about similar at effective stress.
c d
Figure 16 Mohr’s Circle of disturbed samples
Figure 18 Soil response for CSR = 0.402: (a) effective stress

path, (b) shear stress versus shear strain, (c) shear strain versus a
number of cycles, and (d) pore water pressure ratio versus a number
Figure 17 Mohr’s Circle of undisturbed samples of cycles
262
a b
c d

Figure 19 Soil response for CSR = 0.402: (a) effective stress path, path, (b) shear stress versus shear strain, (c) shear strain versus a
(b) shear stress versus shear strain, (c) shear strain versus a number number of cycles, and (d) pore water pressure ratio versus a number
of cycles, and (d) pore water pressure ratio versus a number of of cycles (Sumartini et al. 2018a)
cycles
Figure 22 Liquefaction susceptibility of Orange soil: (a) undisturbed

samples and (b) disturbed samples
4.3 Effect of cyclic loading to soil fabric

Figs. 23 and 24 respectively show the SEM analysis results of the
Orange soil structure before and after the liquefaction tests
(Sumartini et al., 2018). Fig. 13 shows that the soil structure is
composed of a stack of the crystal flakes and is highly porous. In
path, (b) shear stress versus shear strain, (c) shear strain versus a comparison, Fig. 14 shows that the soil structure is visibly broken,
number of cycles, and (d) pore water pressure ratio versus a number and a reduced crystal flake size. Consequently, the small flakes
of cycles (Sumartini et al. 2018a) content of the fabric increases.
263
4. ACKNOWLEDGMENT
The financial grant for this research under the J-RAPID program
(Principal Investigator: Hemanta Hazarika) of Japan Science and
Technology Agency (JST) is gratefully acknowledged. The first
author is grateful to the Indonesia Endowment Fund for Education
(LPDP) for financially supporting her study.
5. REFERENCES
Hatanaka, M., Sugimoto, M. and Suzuki, Y. (1985) "Liquefaction

resistance of two alluvial volcanic soils sampled by in situ
freezing". Soils and Foundations, 25, pp49-63.
Hazarika, H., Kokusho, T., Kayen, R.E., Dashti, S., Fukuoka, H.,
Ishizawa, T., Kochi, Y., Matsumoto, D., Hirose, T., Furuichi,
H., Fujishiro, T., Okamoto, K., Tajiri, M. and Fukuda, M.
Figure 23 The fabric of Orange Soil before cyclic loading (Sumartini et al.
(2017) "Geotechnical Damage due to the 2016 Kumamoto
Earthquake and Future Challenges". Lowland Technology
2018a) International, Special Issue on Kumamoto Earthquake and
Disasters, 19, pp189-204.
Ishikawa, T., and Miura, S. (2011) "Influence of freeze-thaw action
on deformation-strength characteristics and particle
crushability of volcanic coarse-grained soils". Soils and
Foundations, 51, pp785-799.
Kazama, M., Kataoka, S., and Uzuoka, R. (2012) "Volcanic
Mountain Area Disaster Caused by the Iwate-Miyagi Nairiku
Earthquake of 2008". Japan. Soil and Foundations, 52, pp168-
184.
Kochi, Y., Kariya, T., Matsumoto, D., Hirose, T., and Hazarika, H.
(2018) "Investigation of Slopes on The Takanoobane Lava
Dome Using Resistivity Imaging Method". Lowland
Technology International, Special Issue on Kumamoto
Earthquake and Disasters, 19, pp 261-266.
Miyagi, T., Higaki, D., Yagi, H., Yoshida, S., Chiba, N., Umemura,
J., and Satoh, G. (2011) "Reconnaissance Report on Landslide
Disaster in Northeast following the M 9 Tohoku Earthquake".
Landslides, 8, pp339-342.
Sassa. K. (2005) "Landslide disasters triggered by the 2004 Mid-
Figure 24 The fabric of Orange Soil before cyclic loading (Sumartini et al. Niigata Prefecture earthquake in Japan". Landslides, 4, pp113-
2018a) 122.
Song, K., Wang, F., Dai, Z., Iio, A., Osaka, O., Sakata, S. (2017)
3. CONCLUSION "Geological Characteristics of Landslides Triggered by the
2016 Kumamoto Earthquake in Mt. Aso Volcano, Japan".
From the results of the investigation, the conclusions are made as Bulletin of Engineering Geology and the Environment,
following: Springer-Verlag, pp1-10.
1. The Orange soil is a cohesive soil. The cohesion of the Orange a
Sumartini, W. O., Hazarika, H., Kokusho, T., Ishibashi, S.,
soil is reduced after reconstituted. Matsumoto, D., and Chaudhary, B. (2018) "Deformation and
2. The behavior of the disturbed samples under low confining Failure Characteristics of Volcanic Soil at Landslide Sites due
pressure is contradictory with undisturbed samples. However, to the 2016 Kumamoto Earthquake". Lowland Technology
for high confining pressure, the behavior is similar. As the International, Special Issue on Kumamoto Earthquake and
results, it can be concluded that the Orange soil deposit in small Disasters, 19, March 2018, pp237-244.
b
depth is less susceptible to liquefaction. Sumartini, W. O., Hazarika, H., Kokusho, T., Ishibashi, S.,
3. The slip surface is confirmed on the Orange soil layer deposit. Matsumoto, D., and Chaudhary, B., (2018) "Microstructural
Under cyclic loading, both samples show cyclic mobility Characteristics of Volcanic Soil in Aso Caldera related to the
behaviors under the CSR investigated. Although the CSR Landslide Triggered by the 2016 Kumamoto Earthquake",
Proceedings of 16th European Conference on Earthquake
applied to disturbed samples quite lower than undisturbed Engineering, Thessaloniki.
samples, the disturbed samples appear significantly more
Sumartini, W. O., Hazarika, H., Kokusho, T., Ishibashi, S.,
susceptible to liquefaction compared to undisturbed samples. Matsumoto, D. and Chaudhary, B., (2017) "Liquefaction
The results successfully describe the behavior of the Orange Susceptibility of Volcanic Soil in Aso Caldera due to the 2016
soil under cyclic loading. Also, the paper explains that the Kumamoto Earthquake", Proceedings of the 19th International
landslide occurred due to the liquefying of the Orange soil Summer Symposium, Fukuoka, pp13-14.
deposit triggered by the 2016 Kumamoto earthquake. Suzuki, M., and Yamamoto, T., (2004) "Liquefaction characteristic
4. The cyclic loading affects the deformation of the soil fabric as of undisturbed volcanic soil in cyclic triaxial test".
confirmed by the reducing of the crystal flakes of the fabric in Proceedings of 13th World Conference on Earthquake
the SEM analysis. Engineering, Vancouver, Paper No. 465.
264
Screw Driving Sounding Test for Soil Identification and Classification

A. Marto1*, G. Sakai2, N. Suemasa3, N.Z. Mohd Yunus4, S.N. Jusoh4, N. Jamaludin4, M.M. Ponimin4 and M.F. Mohd Tahir4
1
Malaysia-Japan Int. Institute of Technology, Universiti Teknologi Malaysia, Kuala Lumpur, Malaysia
2
Japan Home Shield Corporation, Tokyo, Japan
3
Department of Urban and Civil Engineering, Tokyo City University, Tokyo, Japan
4
School of Civil Engineering, Faculty of Engineering, Universiti Teknologi Malaysia, Johor, Malaysia
*
E-mail: aminaton@utm.my
ABSTRACT: Screw Driving Sounding (SDS) test had been developed about a decade ago and has now been widely used in Japan together
with the Swedish Weight Sounding (SWS) test in site investigation work, particularly for housing development project and investigating the
performance levels of housing lots. Unlike SWS, studies in Japan and New Zealand show that the results from SDS tests can be used to
identify and classify soil types and therefore it is able to give the soil profile without the needs of taking any soil samples. Recently this test
has been introduced to researchers in Universiti Teknologi Malaysia to characterize various soil types. This paper discusses the results of the
SDS tests in identifying and classifying various soil types in Malaysia. Test sites around Kuala Lumpur, Selangor and Johor had been chosen
in performing the SDS tests. Results from three of the sites are compared with the borelogs obtained from the study area comprising of soil
profile, Standard Penetration Test and laboratory tests results. Analysis shows that SDS is able to classify soil types in Malaysia besides its
ability to identify soil layers in a more comprehensive manner. SDS data also correlates well with SPT-N values.
Keywords: Screw Driving Sounding, Standard Penetration Test, site investigation, soil profile, soil classification
1. INTRODUCTION rate of 25 rpm. The 7 load steps for SDS are 0.25, 0.38, 0.50, 0.63,
0.75, 0.88 and 1 kN. The load is increased for every revolution of
In Malaysia, deep boring with Standard Penetration Tests, SPT the rod. For each 250 mm penetration, the rod will move up about
(with disturbed sampling) and undisturbed sampling for laboratory 10 to 20 mm and rotate back down to calculate the rod friction.
tests have been the common method for determining the subsurface
soil profile and geotechnical engineering properties for foundation
design purposes. SPT measures the resistance to penetration offered
by soil at various depths (Wazoh and Mallo, 2014). SPT in Malaysia
is carried out in accordance with MS1056:2005, using a hammer
weight of 65 kg and a drop height of 750 mm. Total penetration is
450 mm and the number of blows for the last 300 mm is the SPT-N
value.
Since performing deep boring either through wash boring, rotary
drilling or percussion drilling are expensive, other tests such as (a) SDS equipment (b) Data can be sent (c) Graphs of collected
Mackintosh or JKR probe is also carried out to complement the soil to office through data can be viewed from
exploration work. In Japan, Swedish Weight Sounding, SWS has cloud system mobile phone
been widely used instead of Mackintosh or JKR probe. Recently, the Figure 1 Screw driving sounding test
Screw Driving Sounding, SDS test had been developed in Japan for (https://www.j-shield.co.jp/english/sds.html)
value added to SWS. SDS is able to classify soil types as already
proven for soils in Japan (Tanaka et al. 2012 and 2015, and Suemasa Measurement items during SDS test includes torque of rod,
et al. 2018) as well as in New Zealand (Mirjafari, 2016 and Mirjafari applied load, penetration velocity and rotation speed at every
et al. 2016). An attempt to implement the test in Malaysia has also revolution of the rod. Both load and torque are applied to the rod at
been reported by Marto et al. (2018a) and (2018b). the same time during the SDS test which is done automatically
A comparison made by Orense et al. (2013) between SDS and (Suemasa et al. 2018).
SWS, SPT as well as Cone Penetration Test, CPT indicated that
SDS test has many advantages including simpler system, faster 2.2 Theoretical Assumptions and Analysis Method
procedure, lighter reaction weight and better cost efficiency than
other sounding tests. SDS also needs only small working area. Not Suemasa et al. (2005) proposed a plasticity model for the SDS test
just that, the data from the SDS machine could be sent to the ‘cloud’ from the results of SDS miniature tests. This plasticity model has
and would be later downloaded from the G-web system for analysis. been illustrated in Figure 2 by Tanaka et al. (2007). According to
Once results are sent to the ‘cloud’, anyone with the given password Tanaka et al. (2015), the combination of torque and vertical load
could access to the analysed data. While the test crew is still at the measured in the SDS test forms a yield locus and the corresponding
site, one could instruct them to do more tests if upon checking incremental components of a rotation rate and a settlement rate obey
through G-web system it shows questionable or contradictory results. the plastic potential rule. The interactive relationship between the
This is the technology in line with the Industrial Revolution 4.0 that combined loads and the corresponding displacement of the soil
is by using automation in work and making use of the cloud element had been described as a constitutive equation.
computing technology (Marto et al., 2018a and 2018b).
2. SCREW DRIVING SOUNDING EQUIPMENT, TEST

AND ANALYSIS METHODS
2.1 Equipment and Test Method
SDS equipment is shown in Figure 1 while the test method has been
widely explained by previous researchers such as Tanaka et al.
(2015), Mirjafari (2016), Mirjafari et al. (2016), Tanaka et al. (2016)
and Marto et al. (2018b). Basically SDS test uses 7 number of load
steps and the rod would penetrate the soil layer continuously at the Figure 2 Concept of plasticity model for SDS (Tanaka et. al., 2007)
265
SDS takes into consideration on rod friction during the test. The The Soil Classification Chart is a plot of the coefficient of
concept of estimating the rod friction has been explained by Tanaka plastic potential, Cp’ versus the slope of corrected torque divided by
et al. (2012) and shown in Figure 3. When the rod penetrates into screw point diameter and corrected load, dT/dWD graph. Cp’, which
the ground while being rotated during SDS test, two components of represents the difficulty of penetration, is obtained from
rod friction occurred, which are vertical component, Wf and approximate slope of the graph between the number of normalised
horizontal component, Tf. The frictions are measured after each 250 half turns, NSDD and normalised torque ratio, πT/WD (Tanaka et al.,
mm penetration when the rod is lifted about 10 to 20 mm and then 2015). For New Zealand’s Soil Classification Chart, it is the plot of
rotated back to the previous position. Through the Eq. (1) and Eq. Cp’ against average change of torque, Ave (T) with the equation as
(2) below, the corrected torque, T and corrected load, W at the follows (Mirjafari, 2016):
screw point are calculated for each 250 mm penetration. It is
necessary to deduct the friction to obtain the actual force applied to
the rod. 1 n 1
Ave(T )   (Ti  Ti )
n  1 i 1 1
(3)
Wa = Wf + W (1)
Ta = Tf +T (2) where T is the change in torque, T at each step of loading, i and

n (=7) is the number of loading.
The chart developed by Tanaka et al. (2015) is divided into three
areas; A, B and C based on the Cp’ values of greater than 1.0 for
sand and loam, between 0.4 to 1.0 for silt and clay, and below 0.4
for peat and organic soils. From SDS tests at 164 sites in
Christchurch, Auckland and Wellington, New Zealand, Mirjafari et
al. (2016) developed the Soil Classification Chart which is quite
different with the Japan’s chart. It is divided into five areas using
four boundary lines to group the soil types; A is for sand, B for stiff
peat, C for silty sand and sandy silt, D for stiff plastic clay and E for
clayey silt, silty clay, clay and silt soils.
Some correlations between the data obtained from SDS tests had
Figure 3 Concept of rod friction (Tanaka et al., 2016) been carried out with SPT-N values by previous researchers in order
to determine the relationship from both tests. This is necessary if
Through SDS data from 25 sites, the soils in Japan has been SDS test is aimed at replacing SPT tests in the future because the
classified by Tanaka et. al (2015) in a so called “ Soil Classification design of shallow foundation in particular is generally using SPT-N
Chart” shown in Figure 4. The soil types are Clay, Silt, Sandy Clay, value besides the strength parameters obtained from laboratory tests.
Loam and Peat. Similarly for soils in New Zealand, almost the same Attempt on the correlation is through determining the relationship
chart has been produced by Mirjafiri et al. (2016), as shown in between the SPT-N value and the E0.25, the data acquired from
Figure 5. The type of soils in New Zealand are Clay, Stiff Peat, SDS test. E0.25 is the penetration energy required for every 250 mm
Plastic Clay, Silty Clay, Clayey Silt, Sand, Sandy Silt, Silty Sand penetration of screw point (Tanaka et al., 2012). It has been reported
and Silt. by Tanaka et al. (2012) that the correlation between E0.25 and SPT-
N value for Japan soils is as follows:
E0.25 = 0.268NSPT (4)
In New Zealand, the correlation between E0.25 and SPT N value

for sandy soil has been reported by Mirjafiri et al. (2016) as follows:
E0.25 = 0.34NSPT (5)
3. SDS FIELD TESTS

In the first attempt of using SDS in Malaysia, the SDS tests had
Figure 4 Soil Classification Chart for soils in Japan been conducted at a total of 10 sites; four in around Kuala Lumpur
(Tanaka et al., 2015) and Selangor area and six in Johor area. The general soil types at the
sites includes peat, residual soils, marine clay and sandy soils. In
this paper only the test results from three sites are analysed and
presented to represent preliminary results of the research.
Borehole records from deep boring using rotary drilling,
comprising of soil profile, Standard Penetration Test as well as the
sieve and Atterberg limit laboratory tests results are used to compare
with SDS test data. These borehole records had been obtained from
Public Works Department, PWD and the project owner. SDS tests
were carried out close to the existing boreholes, which were within 1
to 2 m away in radius from the boreholes. Table 1 shows the site
locations with respective SDS and Borehole number while Figure 6
shows the SDS test being carried out at one of the sites.
Figure 5 Soil Classification Chart for soils in New Zealand
(Mirjafari et al., 2016)
266
Table 1. Site Locations with Respective SDS and Borehole Number Setapak and Figure 8(c) for Batu Pahat area. Results from the
borelogs are checked with the laboratory tests from the disturbed
Borehole soil samples obtained through SPT tests. However, there is a
No Site location SDS No.
No. Obtained from disadvantage of soil profile plotted from borelogs such that the SPT-
1 Cheras SDS28 BHJ1 PWD N values and soil laboratory results are obtained only every 1.5 m
2 Setapak SDS37 BH2 PWD depth, unlike the SDS test which is a continuous result throughout
3 Batu Pahat SDS13 BH2 Project owner the soil depth. In this case a more comprehensive results will be
Note: PWD – Public Works Department, Malaysia obtained from SDS tests.
Figure 6 Screw Driving Sounding test at Setapak site
3.1 Test Sites Location

Figure 7(a) shows the test site location around Kuala Lumpur and (a) Cheras
Selangor area while Figure 7(b) shows the sites around Johor area.
The three sites in which results are presented in this papers are
Cheras in Selangor, Setapak in Kuala Lumpur and Batu Pahat in
Johor.
Setapak
Cheras
(b) Setapak
(a) Kuala Lumpur and Selangor area
Batu Pahat
(b) Johor area

Figure 7 Test site locations
3.2 Soil Profile of Test Sites (c) Batu Pahat

The soil profile from the nearby borelog and the plots of SPT-N Figure 8 Soil profile at three selected sites from borehole record and
value with depths as well as the SDS test results in terms of screw driving sounding test
corrected torque are shown in Figure 8(a) for Cheras, Figure 8(b) for
267
From borehole data, it can be seen that topsoil of 1.5 m thickness

overlaid the soils at all sites. In Cheras the topsoil is underlain by 9
m of Sandy clay layer with SPT-N values ranges from 5 to 8
followed by silty SAND with traces of gravel between 10.5 m to 12
m depth and very dense Silty sand with SPT-N values between 10
and 50 below 10.5 m.
For Setapak, limestone is found at 15 m depth while the soil
profile below topsoil are mixtures of sandy CLAY (1.5 to 3 m and 9 (c) 8.25 to 8.5 m (firm Sandy CLAY) (d) 9.0 to 9.25 m (Silty SAND)
to 12 m depths), Sandy SILT (3 to 4.5 m and 12 to 15 m depths), Figure 9 Relationship between normalised corrected torque and
stiff CLAY (4.5 to 6.7 m depth) and stiff SILT (6.7 to 9 m depth). corrected load (at selected 0.25 m section) from Cheras Site
The SPT-N values at Setapak site is very low except at 3 m depth
(N=15), probably due to the existence of gravel within the Sandy According to Tanaka et al. (2015), the slope of the plot between
SILT layer. Soil profile at Batu Pahat site generally consists of soft normalised corrected torque and corrected load tends to have a
Clayey SILT up to 9 m depth with SPT-N value of 1, underlain by 6 positive value for frictional soil like sand or loam, and negative
m thick of stiff Sandy SILT and a very stiff gravelly sandy SILT value or zero for clay and silt. Also, through the plots of corrected
thereafter. SPT-N values increased up to 20 below 7.5 m depth. torque versus corrected load, Tanaka et al. (2016) reported that the
slope of the graphs shows positive value for sand or pumice while a
4. RESULTS AND DISCUSSION negative or small value for silt or clay. The results obtained for the
4.1 Identification of Soil Layer soil in Cheras are in agreement with what had been found by
previous researchers using SDS test. As an example, for stiff clay,
Through SDS test, different layer and different types of soils are the normalised torque does not increase that much as the corrected
recognised from the range and pattern of corrected torque. load increases since the soil is fine-grained and frictionless under
Comparing the soil profile obtained from the borelog and SDS test, undrained condition. However for sand, the normalised torque
it seems that through SDS tests, the boundary of soil layers are increases as the corrected load increases. This is due to the
mostly found earlier or shallower than the borelog. This is the resistance through the friction given by the coarse-grained materials
advantage of SDS since the test is carried out continuously thus the of sand to the penetration of the SDS screw point.
boundary of each soil layers are detected more precisely. Examples of Cp’ obtained for Cheras site is shown in Figure10.
From the plot of corrected torque with depth, layers of soil such A linear regression analysis from the plots of NSDD versus πT/WD
as sand can be detected. From Figs. 8(a) and 8(c), thin layers of sand identified the slope of the linear lines, also known as Cp’. For stiff
are detected at Cheras site (from 6.75 to 8 m depth) and Batu Pahat sandy clay, the small value of Cp’ obtained (0.4) is in line with the
site (from 10 to 10.5 m depth). Sand layers are identified through the theory since the difficulty of penetration through cohesive soil is
shape of corrected torque developed at each 250 mm penetration. At low. For silty sand with Cp’ of 2.6, it is in agreement with the
this layer, the value of corrected torque increased with penetration properties of frictional material, in which difficulty of penetration is
depth at each seven steps load increment due to high friction as a shown by high Cp’ value. However, the Cp’ results for sand and
result of the screw point going through the frictional soils. This firm sandy clay are incompatible with the soil types. By omitting
phenomena, recognised in SDS test, has also been explained by some of the points, the expected lines to give the Cp’ suitable to the
Mirjafari et al. (2016). Since this is a very thin layer, investigation soil types are shown in Figure 10(b) for sand and Figure 10(c) for
through deep boring using SPT was not able to identify its existence. firm sandy clay. The outliers probably transpired due to the ‘noise’
On the other hand, torque is consistent with depths when SDS as a result of gravels that present within the soils.
penetrates through clayey or silty soils due to none or limited
frictional materials within the layer besides due to undrained
situation resulted with zero friction angle. This can be seen as an
example from 6 to 10 m depth at Setapak site.
Expected line
4.2 Soil Classification
Previous studies on SDS data show that the data can be used to
classify the soil type by determining the values of dT/dWD as well (a) 1.75 to 2.0 m (stiff Sandy CLAY) (b) 7.0 m to 7.25 m (SAND)
as Cp’. From the results of SDS test in Cheras, the graphs of
normalised corrected torque, T/D versus corrected load, W at four
selected 0.25 m section of penetration are shown in Figure 9. The
slopes (dT/dWD) of the graphs are obtained from the equation of
Expected line
linear regression line shown in respective graphs. For stiff sandy
clay in Figure 9(a) (depth between 1.75 m and 2.0 m) and firm
sandy clay in Figure 9(c) (depth between 8.25 m and 8.5 m), the
obtained slopes of 0.2 and 0.4, respectively are very small. (c) 8.25 to 8.5 m (firm Sandy CLAY) (d) 9.0 to 9.25 m (Silty SAND)
Meanwhile for sand layer in Figure 9(b) (depth between 7 and 7.25
m, the slope is 1.7 and for silty sand in Figure 9(d) (depth between 9 Figure 10 Relationship between NSDD and πT/WD (at selected
to 9.25 m depth), the slope is nearly 1. 0.25 m section) from Cheras Site
The variations of dT/dWD and Cp’ for the soils with depth are
shown in Figs.11(a), (b) and (c) for the sites of Cheras, Setapak and
Batu Pahat, respectively. From these, the results for respective soil
types are plotted in the Soil Classification Charts in which some are
shown in Figure12. From the graphs in Figure12(a) and (b), it can be
seen that generally the Silty SAND from Cheras occupied the right
hand side of the chart with positive values of dT/dWD while for
(a) 1.75 to 2.0 m (stiff sandy CLAY) (b) 7.0 to 7.25 m (SAND) Cp’, majority of the points spans from 0.9 to 7. For stiff Clayey
SILT in Figure12(c), the points scattered on the left side of the chart
with negative values of dT/dWD and Cp’ of between 0.5 and 2.
268
(c) Clayey SILT (d) Stiff CLAY

(a) Cheras
Figure 12 Relationship between coefficient of plastic potential, Cp’
and dT/dWD for Silty SAND, Clayey SILT and Stiff CLAY
A preliminary Soil Classification Chart for soils in Malaysia

based on the results from Cheras, Setapak and Batu Pahat is shown
in Figure13. Although it has been developed from a very limited
data, the trend for each soil types could be recognised. The most
obvious one is for SANDY soils (silty sand, sandy silt and sandy
clay) that occupied the top right hand side of the graph and also
clayey SILT on the left hand side. This chart shows some similarity
with soils in Japan and New Zealand.
(b) Setapak
Figure 13 Preliminary Soil Classification Chart for soils in

Malaysia
(c) Batu Pahat
4.3 Correlation between SDS and SPT Test
Figure 11 Variation of corrected torque, dT/dWD and coefficient of
plastic potential with depths Data from SDS tests are correlated with SPT-N values obtained
from nearby borehole record. Figure 14 shows a comparison
between E0.25 and SPT-N values with depth for all sites. In general
the trend between E0.25 and SPT-N values with depth is similar for
all sites
Figure 15 shows the plots of E0.25 with SPT-N or NSPT
values.Based on the statistical analysis using linear regression
method, it is found that the coefficient of determination, R2 for the
correlation between SPT-N value and E0.25 are all above 0.71
(coefficient of correlation, R greater than 0.842) which shows more
than 71% of E0.25 is dependent on SPT-N. Hence this is considered
as a very good correlation since SDS test result (E0.25) is highly
correlated with SPT-N values despite the limited data obtained.
Below are the summary of correlation equations between SDS data
and SPT-N values:
(a) Silty SAND (b) Silty SAND with traces of gravel Cheras, E0.25 = 0.2NSPT (6)
Setapak, E0.25 = 0.16NSPT (7)
Batu Pahat, E0.25 = 0.27NSPT (8)
269
data. This is also required to develop a valid correlations between

SDS data and SPT-N value. Hence, the current research will be
E0.25
E0.25 SPT-N extended to other test sites throughout Malaysia in the near future.
SPT-N
6. ACKNOWLEDGEMENT
SPT-N The authors gratefully acknowledge the financial support and
contribution given by the Japan Home Shield Corporation through
E0.25
the Research Contract No. 4C134, in undertaking this research.
Acknowledgements are also conveyed to respective researchers
from the School of Civil Eng., Universiti Teknologi Malaysia
(UTM) for their assistance in this research. Lastly, the first author
would like to give a special thanks to the Tokyo City University for
appointing her as a Visiting Professor when this paper was written.
(a) Cheras (b) Setapak (c) Batu Pahat 7. REFERENCES

Figure 14. Comparison between E0.25 and SPT-N values with depth
Malaysian Standard MS 1056 (2005) “Method of Test for Soils for
Civil Engineering Purposes”, SIRIM, Malaysia.
Marto, A., Sakai, G., Suemasa, N. and Jamaludin, N. (2018a) “An
Attempt of Screw Driving Sounding Test in Malaysia”,
International Conference on Civil, Offshore & Environmental
Engineering (ICCOEE), 13-15 August, Kuala Lumpur.
Marto, A., Sakai, G., Suemasa, N, Mohd Yunus N.Z., Jusoh S.N.,
Kasim, F., Ahmad, K. and Mohd Satar, M.H. (2018b) “Screw
Driving Sounding Test; A New Technology for Soil
(a) Cheras (b) Setapak (c) Batu Pahat Investigation Work Particularly for Soft Soil”, Int. Conf. on
Advances in Civil and Environmental Eng. (ICANCEE), 24-25
Figure 15. Correlation between total energy required for 0.25 m Oct., Bali, Indonesia.
penetration (E0.25) with SPT-N value (NSPT) Mirjafari Miandeh, S.Y. (2016) “Soil Characterisation using Screw
Driving Sounding (SDS) data”, PhD Thesis, University of
Comparing the results obtained by soils in Japan (Tanaka et al., Auckland, New Zealand.
2016) shown in Eq. (4) and New Zealand, (Mirjafari, 2016) shown Mirjafari, S.Y., Orense, R.P. and Suemasa, N. (2013) “Comparison
in Eq. (5), it seems that the results are comparable. From the results between CPT and SDS Data for Soil Classification”, Proc.
obtained, the variation of SPT-N with depth is compared through the 10th Int. Conf. on Urban Earthquake Eng., pp. 561–566.
plots in Figure 16 for all the three sites. The variation is in Mirjafari, S.Y., Orense, R.P. and Suemasa, N. (2016) “Soil
agreement except for the site in Cheras. Beyond 7 m depth the SPT- classification and liquefaction evaluation using Screw
N values seem to be consistent while for SDS-N, it increases with Driving Sounding,” Proc. 5th Int. Conf. Geotechnical
depth. The results of SDS is in line with the type of soil which is Geophysical Site Characterisation, ISC 2016, 1, no. 2001, pp.
firm Sandy CLAY followed by dense Silty SAND as shown earlier 284–286.
in Figure 8(a). Logically, the N value will increase with depth in this Orense, R., Mirjafari, Y. and Suemasa, N. (2013) “Geotechnical
type of soils as resistance to the hammer penetration increases. Site Characterisation using Screw Driving Sounding
Method”, New Zealand-Japan Workshop on Soil Liquefaction
during Recent Large-Scale Earthquakes, New Zealand.
SDS-N SDS-N SDS-N Suemasa, N., Shinkai, K., Suzawa, T. and Tamura, M. (2005) “A
SPT-N plasticity Model for Swedish Weight Sounding Test”. Proc.
SPT-N SPT-N 4th Japan-Philippines Workshop on Safety and Stability of
Infrastructure, University of Philippines, 169-177.
Suemasa, N., Tanaka, T., Yamato, S. and Sakai, G. (2018)
“Application of SDS and SD-Sampler to Survey on
Embankment Improved by Floating Columns”, Int. Sym. on
Lowland Technology, Sept. 26-28, Hanoi, Vietnam.
Tanaka, T., Suemasa, N., Yamato, S. and Katada, T. (2007) “New
Operation System for Swedish Weight Sounding Test”, Proc.
Of the 7th International Offshore and polar Engineering
(a) Cheras (b) Setapak (c) Batu Pahat
Conference, 1-6 July, Portugal.
Figure 16. Comparison between SPT-N value and estimated SDS-N Tanaka, T., Suemasa, N., Ikegame, A. and Yamato, S. (2012)
value throughout soil depths “Classification of Strata Using Screwdriver Sounding Test,”
Proc. 22nd Int. Offshore and Polar Eng. Conference, pp. 851–
5. CONCLUSION 856, 17-22 June, Rhodes, Greece.
Tanaka, T., Suemasa, N. and Tamato, S. (2015) “The Method of
Screw Driving Sounding Tests, SDS had been performed Soil Classification using Screw-Drive Sounding Test and Its
successfully at various sites in Selangor, Kuala Lumpur and Johor Application”, Int. Conference on Soft Ground Engineering, 3-
area. The results analysed from three sites indicate that SDS data 4 December, Singapore.
have a very good correlation with SPT-N values. SDS data are also Tanaka, T., Suemasa, N. and Yamato, S. (2016) “Assessment of Soil
capable to be used in identifying and classifying different type of Characteristics by Screw Driving Sounding,” 19th Southeast
soils. Through the development of a preliminary Soil Classification Asian Geotechnical Conf., 2nd AGSSEA Conf. (19SEAGC
Chart, Sandy Clay, Sandy Silt, Clayey Silt, Silty Sand, Stiff Clay 2AGSSEA), Kuala Lumpur
and Soft Silt, found from Cheras, Setapak and Batu Pahat sites could Wazoh, H. and Mallo, J. (2014) “Standard Penetration Test in
be identified. However, for developing the chart that will represents Engineering Geological Site Investigations – A Review,” Int.
validly the soils in Malaysia, it is necessary to have more SDS test J. Eng. Science, pp. 1813–2319.
270
ARGILLACEOUS ROCK PROPERTIES CHANGES DUE

TO THE WEATHERING PROCESS
Idrus M. Alatas
Civil Engineering Department of ISTN Jakarta, Indonesia
Masyhur Irsyam
Civil Engineering Department of ITB, Bandung, Indonesia
Ramli Nazir
Centre of Tropical Geoengineering UTM, Johor Bahru, Malaysia
Pintor T. Simatupang
Civil Engineering Department of Mercubuana University, Jakarta, Indonesia.
ABSTRACT : Argillaceous rock is a clay rock in fresh condition has a very high shear and crushing
strength. This is because the index properties of argillaceous rock are included in the soft rock category.
However, if weathering occurs due to contact with atmosphere and hydrosfere, then argillaceous rock will
quickly reduce its strength. In this paper, wetting-drying cycle process and drying process were done on
argillaceous rock in laboratory were reported. Changes in the indes and physical properties of argillaceous
rock show a complete adverse behavior due to wetting-drying cycle process and drying process. Due to
wetting-drying cycles, from disintegration ratio parameter (DR) of Semarang-Bawen argillaceous rock
appears to have completely weathered on 24th days and completely non-durable. While due to the drying
process until the 80th day, the physical properties of argillaceous rock seems to be reliable and completely
durable.
Keywords : argillaceous rock, index properties, physical properties, wetting drying cycle, drying process
changes in water content, bulk density and void

I. INTRODUCTION ratio. The physical properties observed are
changes in disintegration ratio and also
Argillaceous rock is a formation of shrinkaging-swelling behavior. The behavioral
sedimentary rocks from clay deposits changes of these properties are illustrated from
("https://www. thefreedictionary.com " 2008), drying time and wet-dry cycle times (Alatas
or in other terms referred to as clay stone and 2015 c, 2015 d; Irsyam M 2007).
more specifically with the term clay shale. By
adding the word "shale" which describes the
specific behavior of this type of clay rock that is II. PROPERTIES OF HAMBALANG AND
easily weathered. This is due to contact with SEMARANG BAWEN
atmosfere or hydrosfere. During the argillaceous ARGILLACEOUS ROCK
rock weathering process through the drying
process and wetting drying cycles there is a Preliminary testing on the field and
changes in the physical and index properties, laboratory was also conducted to obtain the
which eventually changes in the index and index, physical and mechanical properties of
physical properties. argillaceous rock. Test pit was performed to
These properties have been observed during ensure that argillaceous rock to be tested in the
the weathering process of up to 80 days. The laboratory is fresh and has not been weathered.
incdex properties of the observed indices are Samples of undisturbed and disturbed
The 20th SEAGC – 3rd AGSSEA and 22th IGC (PIT) HATTI, Jakarta, November 2018
271
argillaceous rock were brought to the laboratory

for index, physical and mechanical properties. In
this study, an undistubed argillaceous rock
samples in sufficient numbers with initial
properties of the overall sample were relatively
the same.
In Figure 1, more than 130 argillaceous
rock undisturbed samples from Hambalang West
Java and Semarang-Bawen Central Java are
placed in the drying room to be tested for shear
strength on a predetermined schedule up to the
80th day. In Figure 2 shows a measurement of
undistubed sample change during the weathering
Figure 2. Three dimension (3D) of undisturbed
process due to the drying process . While in sample volume changes device in drying process
Figure 3 shows the wetting drying cycle process (Alatas I. M. 2017).
using the same device. However, it can be seen
that its swelling behavior though filling the
chamber with water, and drying it in order to
obrserve the shrinkage in the sample in 3
dimension. In Figure 4 shown that schematic for
drying room process and 3D volume change
device.
In Figure 5, Figure 6, Figure 7 and Figure 8
shown the initial index properties parameter of
argillaceous rock from Semarang-Bawen and
Hambalang before weathering process been
done.
Figure 3. Three dimension (3D) of undisturbed

sample volume changes device in wetting-drying
process (Alatas I. M. 2017).
Figure 1. More than 130 argillaceous rock

undisturbed sample fro two different location in P.
Jawa, have been prepared for this reseach (Alatas I.
M. 2017).
Figure 4. Schematic for drying room and 3D volume

change device
272
The research program has been prepared in

accordance with the objectives of the study
illustrated in Figure 9. This study requires a
considerable amount of undisturbed samples.
From each observation location it takes
approximately 70 undistubed argillaceous rock
samples having the same initial properties.
Therefore, sampling is not possible through
depth boring. One of the solution is to obtain the
sample through Test Pit. Generally in a soil
investigation, the sample is obtained from a thin
walled sample through drilling. Sample from test
pit is usually onbtained from the cor box sample.
Samples from thin-walled samples or cast boxes
were brought to the laboratory and undisturbed
sampling was undertaken with standard sizes
determined (Alatas I. M. 2017).
Preliminary investigation was conducted to
develop an argillaceous rock undisturbed
sampling methodology. Methodology wa
developed undistubed samples in large Figure 9. Flow chart for reaserch programe
quantities, this is because argillaceous rock is
rapidly weathered when reacted with Drilling Core Machine
hydrosphere and atmosfere. In this case dry
drilling becomes the right solution. Rotational Ground Surface
speed drilling machine becomes an important

equipment to minimize disturbance effect. The
solution of using cast drilling machine which
usually used for concerete sampling was
modified. Single core barrels have been Test Pit for
developed as a split single core barrel to argiilaceous rock
facilitate operators to obtained an argillaceous Sampling methods
rock sample from inside the core barrel. The
Split single core barrel
for Triaxial Sample
sample preparation for laboratory testing is done Testing
on site after the sample is successfully removed
from the split ingle core barrel (Alatas I. M. Figure 10. An argillaceous rock sampling
2017). Schematic sampling of argillaceous rock methodology developed.
samples on the pit test as illustrated in Figure 10
below. By using new sampling methodology, it Teble 1. Comparison of standard sampling with the
has been successfully obtain undisturbed sample developed methodology for taking argillaceous rock
or argillaceous rock. The comparison of standard
STANDARD UNDISTURBED SAMPLING METHODS
sampling methods compared with the new SITE INVESTIGATION LABORATORY TEST
developed methodology can be seen in Table 1 From BORING Triaxial UD
TEST or Sampling
below. TEST PIT as Thin Mobilization Using
Weathering Process
Triaxial
Walled Tube to Moulding,
Testing
Sampling Core/ Laboratory Extruder,
Barrel Sampling/ Triming
Box Sampling
ARGILLACEOUS SAMPLING DEVELOPMENT
METHODS
Undistubed
Mobilization
Sampling using Triaxial
to
Split Single Core Testing
Laboratory
Barrel
273
With this new improvised developed method, process. The change in volume and moisture
then undisturbed sample in the laboratory can be content of the argillaceous rock samples results
eliminated and only done in the field. also changes in other properties index (Alatas I.
M. 2017).
IV. CHANGES ARGILLACEOUS ROCK Initial Hambalang argillaceous rock water
PROPERTIES DUE TO content between 4.1% and 5.3%, during the 80
WEATHERING PROCESS days drying process it reduces to 2.1%. While
the initial water content in Semarang-Bawen
Changes in the properties of argillaceous between 15.9% to 17.6% in 80 days drying time
rock due to drying process as shown in Figure it reduces to 8.9%. Both samples have shown
11 for wetting-drying process include changes in similar behavior to water-level reduction, where
index properties such as moisture content, bulk significant water content reduction occurred
density, dry density, saturation density, void early in the drying process (Alatas I. M. 2017).
ratio. Changes in soil properties of argillaceous
rock due to wetting-drying cycles show a very
Figure 11. Changes in water content on argillaceous rock where (a) Hambalang (b) Semarang-Bawen due to drying process
up to 80 days.
Changes in physical properties include changes significant difference when compared to

in GSD are measured by determining the weathering by drying process. This can be seen
disintegration ratio (Dr). In general, due to in Figure 12, where argillaceous rock
weathering process decreases from Hambalang and Semarang-Bawen day on 40th
disintegration ratio (Alatas 2015 c; Erguler Z A day already full weathered (completelly non
2009; Shakoor A 2011; Shakoor A 2015). durable). Changes in the index properties due to
Changes in sample volume such as swelling and the swelling process due to wetting and
shrinkage occur due to wetting and drying
274
shrinkage due to drying. If wetting-drying chloride. While Semarang-Bawen is

cycles occur more dominated by smectite.
Figure 12. Swelling and shrinkage behavior of argillaceous rock due to wetting-drying cycle process
(a) Hambalang, (b) (Semarang-Bawen)
frequently within a certain time, it will cause a The smectite mineral from the montmorillonite
faster weathering process. Wetting process is mineral family has a greater swelling effect
done 1 soaking, 2 soaking and 3 soaking every 8 when compared with other clay minerals
days and only 5 minutes soaking (Pineda J.A (Mitchell J.K 2005).
2014; Alatas I. M. 2017). Measurements of
vertical and horizontal deformations of the Table 2. Mineralogy content for two types argillaceous
Hambalang and Semarang-Bawen argillaceous rock (LEMIGAS 2015)
TOTAL
rocks were performed with free swells, without Location
CLAY MINERAL (%)
(%)
pressure. . Smg-Bwn
Smectite
50
Illite
3
Kaolinite
2
Chlorite
1 56
From the results of the above properties Hambalang - 12 30 20 62
change research, argillaceous rock from Calsite
CARBONATE MINERAL (%)
Dolomite Siderite -
TOTAL
(%)
Semarang-Bawen has a slightly larger vertical Smg-Bwn 30 - 2 - 32
strain and horizontal strain compared to the Hambalang - - 4
OTHERS MINERAL (%)
- 4
TOTAL
argillaceous rock of Hambalang. This is due to a Quarts Feldspar Plagioclase Pyrite (%)
dominant clay mineral element that is different Smg-Bwn 8 - 2 1 12
Hambalang 30 - 4 - 34
from the two argillaceous rocks. Argillaceous
rock Hambalang contained kaolinite and
275
Disintegration ratio is represented as an When due to drying process until day 80, DR
argillaceous rock weathering parameter though equal to 0.9162 show both sample still
Grained Size Distribution testing with sieve completelly durable.
shaker. By plotting between the cumulative 1,00
percentage against the grained size as in Figure 0,9162
13. The next process is to determines the
disintegration ratio (DR) by utilizing the graph in 0,90 Hambalang- Drying Process
Figure 13. Changes in the physical properties of

the argillaceous rock weather due to the wetting- 0,80
Hambalang- Wetting-Drying
Process with 1 soaking in 8
drying cycle process is as shown in Figure 14 days
below.
0,70
100 days
0 day
(%)

0,60 days
Ratio ( DR )
8 days Smg-Bwn- Drying Process
80
Cumulate Retained
16 days
0,50 Smg-Bwn- Wetting-Drying
60 Process with 1 soaking in 8
24 days
days
Disintegration 0,40 Smg-Bwn- Wetting-Drying
32 days Process with 2 soaking in 8
days
40
40 days Smg- Bwn- Wetting-Drying
0,30 Process with 3 soaking in 8
48 days
days
20
64 days 0,20
0,1467
80 days 0,1368
0
100 10 1 0,10 0,0965
Diameter ( mm )
Figure 13. GSD Curve Hambalang argillaceous rock due to -
wetting-drying proses with 1 soaking every 8 days in 5 0 8 16 24 32 40 48 56 64 72 80
munites (Alatas I. M. 2017). Time (days)
Figure 14. Changes disintegration ratio (DR) Hambalang
From Figure 14 it can be seen that and Semarang-Bawen argillaceous rock. (Alatas I. M.
Hambalang argillaceous rock is more resistant to 2017).
weathering than Semarang-Bawen argillaceous
rock. On the 24nd day it appears that argillaceous
rock Semarang-Bawen has been completelly non V. CONLUSSION
durable with DR approaching the number 0.00
with a frequent of 2 and 3 soaking in every 8 From the discussion about the behavior of
days until the undisturbed sample completely argillaceous rock properties due to the
non durable on day 24th . However, for weathering process, it can be concluded as
Hambalang argillaceous rock samples on day follows:
24th , DR was 0.67 on 1 soaking , 0.55 on 2 1. The process of wetting and drying is more
soaking and and 0.38 on 3 soaking. Soaking influential to accelerate weathering rather
period 1, 2, 3 times every 8 days, and every than the drying process on argillaceous
soaking period only takes 5 minutes. (Shakoor A rock.
2011; Alatas 2015 c). Until the end of day 80, 2. When wetting-drying cycles occur more
Hambalang argillaceous rock has a DR frequently in certain periods, it is more
respectively of 0.15, 0.14 and 0.10 on the same likely to accelerate the occurrence of
soaking day as above. The difference between weathering in argillaceous rock.
the argillaceous rocks from two samples is due 3. Acceleration of properties change during
to the different dominance of clay minerals. the weathering process is also highly
276
dependent on the main dominant clay LEMIGAS, Laboratory. 2015. "SEM and XRD Report for
minerals from argillaceous rock. Semarang-Bawen and Hambalang Clay Shale."
In. Jakarta: LEMIGAS.
4. From shrinkage with a frequency of 2 Mitchell J.K, Soga K. 2005. Fundamentals of Soil
soaking in 5 minutes per day for every 8 Behaviour 3rd edition (John Wiley & Sons, Inc.).
days on undisturbed sample, both samples Pineda J.A, Romero E, De Gracia M, Sheng D, . 2014.
from Hambalang and Semarang bawen 'Shear strength degradation in claystones due to
environmental effects', ' Geotechnique, 64 493-
argillaceous rock can still be measured by 501 (2014).
shrinkage until day 40. After 40 days both Shakoor A, Gautam T P. 2015. 'Influence of Geologic and
samples have collapsed, and the test is not Index Properties on Disintegration Behavior of
can proceed. Clay-Bearing Rocks', Environmental &
5. from disintegration ratio testing using Engineering Geoscience, Vol. XXI, No. 3, August
2015, pp. 197–209.
distubed sample, Semarang-Bawen Shakoor A, Tej P Gultom 2011. 'Assessing The Slaking
argillacesous rock completelly non durable Behaviour of Clay- Bearing Rock', 10th Annual
on day 24 with DR approaching 0.00. While Tecnical Forum Geohazards Impacting
on Hambalang on the same day its DR Transportation In The Appalachian Region,
Columbus Ohio USA.
ranges from 0.38 to 0.67. So it can be
concluded that Hambalang argillaceous
rock is more resistant to weathering than
Semarang-Bawen argillaceus rock.
REFERENCES
Alatas I. M. 2017. 'Kesan Luluhawa Terhadap Kekuatan

Ricih Syal Lempung Dalam Penentuan Parameter
Kestabilan Cerun', Universiti Teknologi
Malaysia.
Alatas, I. M., Samira, A. K., Ramli, N. and Irsyam, M. .
2015 d. "Effect Of The Weathering On
Disintegration And Shear Strength Reduction Of
Clay Shale " In The Manuscript Paper For :
"Geotropika 2016", The 10th International
Conference of Geotechnical and Transportation
Engineering. Kuala Lumpur, Malaysia, 16 - 17
February 2016: Penerbit UTM Press.
Alatas, I. M., Samira, A. K., Ramli, N., Irsyam, M. and
Himawan, A. 2015 c. 'Shear Strength
Degradation of Semarang-Bawen Clayshale Due
To Weathering Process.', Jurnal Teknologi
(Sciences and Engineering ) Special Edition ,
acceptance September 2015.
Erguler Z A, Shakoor A. 2009. 'Quantification of Fragment
Size Distribution of Clay-Bearing Rocks after
Slake Durability Testing', Environmental &
Engineering Geoscience, Vol. XV, No. 2, May
2009, pp. 81–89.
"https://www. thefreedictionary.com " In. 2008. WorldNet
3.0. farlex clipart collection.
Irsyam M, Endra S, Himawan A. 2007. 'Slope Failure of an
Embankment on Clay Shale KM 97+500 Of The
Cipularang Toll Road and The Selected Solution
: A Case Of Slope Failure Due To Strength
Degradation Of Clay Shale.', Proceeding of The
Geotechnical International Symposium in
Bangkok 6-7 Decemebr 2007.
277
Investigation of Aqueous Phase Liquids Migration in Double-Porosity Soil under

Isothermal and Non-Isothermal Effect
K.F. Loke1, R. Nazir2, and H. Moayedi3
1,2,3
Center of Tropical Geoengineering, Faculty of Civil Engineering, Universiti Teknologi Malaysia,
81310 Johor Bahru, Johor, Malaysia
E-mail: 1edwinloke84@yahoo.com; 2ramlinazir@utm.my; and 3hossein.moayedi@gmail.com
ABSTRACT: As in civil engineering field, the intricate problem emerges when subsurface system undergoes the vibration, isothermal and
non-isothermal effect in double-porosity soil. This will influence the migration of aqueous phase liquid (APLs) in the subsurface system. A
series of laboratory experimental were carried out to monitor and observe the characteristics of the soil structure and APLs migration in
deformable double-porosity soil under vary temperature. In order to monitor the changes, a unique digital image processing technique was
used. The contour pattern of APLs migration was plotted based on the generated Hue Saturation Intensity (HSI), image analysis using Matlab
and Surfer software. The results of the experiments show that both samples do not have a uniform migration pattern. In addition, the
migration of fluid in soil sample 2 (isothermal condition) was faster than sample 1 (non-isothermal condition). This is mainly because the
condition of heat imposed into the soil could influence the duration required for the migration. It is found that the heat will slow down the
speed of migration. It can be concluded that factors that significantly influence the APLs migration (under vibration, isothermal and non-
isothermal effect) were the soil structure, soil fractured pattern, and soil temperature.
Keywords: Non-isothermal porous media, isothermal porous media, APLs migration, digital image analysis technique, geo-environmental.
1. INTRODUCTION on double-porosity soil, which have contributed to the knowledge.

However, the studies were limited to common double-porosity soil
Both development and industrial projects have played a key part in and reaction such as vibration and non-isothermal effect was never
contribution to climate changes, global warming, pollution and imposed to the double-porosity soil.
natural disaster, which led to negative impact on the health problem Therefore, the main objective of this study was to apply the
and geo-environment hazard. These problems have called attention model concept of the soil that overlaps the three continuums of
to the vibration, non-isothermal and isothermal effect in double- fracture porosity, intra-aggregate porosity and inter-aggregate
porosity soil resulting influence the migration of APLs into porosity features as developed by Loke et al., (2018) with
groundwater resources. Therefore, the main problem that need to be considering the condition of isothermal and non-isothermal. A series
focused is ensuring the geo-environment safety issues. Vibration of laboratory experiment model were carried out to study the
could cause the soil structure rearrangement, soil structure unstable, characteristics of APLs migration into a vibrated double-porosity
cracked soil, and soil structure volumetric deformation, which could soil. The experiments were conducted under the isothermal and non-
affect the characteristics of pore sizes in future (Loke et al., 2017). isothermal condition by using digital image processing technique.
Double-porosity media was identified as a two specific sub-region Thus, this study aimed (i) to determine the characteristics of APLs
scale with the soil condition of intra-aggregate and inter-aggregate migration in fractured double-porosity soil using digital image
pores, which can be found in both compacted soil and agriculture analysis, (ii) to distinguish the phenomena of APLs migration with
tops-soils (Li and Zhang, 2009). The structure of double-porosity the condition of isothermal and non-isothermal in fracture double-
soil affects the pattern and speed rate of liquids migration. porosity soil and (iii) to identify the APLs migration speed rate for
Generally, the fractured porosity formations are characterized by the specific soil column circumference zone.
water-bearing formations. It is well established that the groundwater
flows along fractured solid rocks. Indeed, the fractured solid rocks is
2. MATERIALS AND METHODS
the result of the break in the rock mass normally caused by tectonic
force (Masciopinto et al., 2001). According to Krisnanto et al., 2.1 Fracture Double-Porosity Soil Sample Preparation
(2014), the fractured soil plays an essential role in the liquids flow
through the complicated soil structure. To create a double-porosity characteristic conditions, the soil sample
The liquid migration included gravity infiltration and transport. materials used in this study was a commercially available kaolin
It is influenced by factors such as moisture content and temperature soil. The kaolin soil properties were tested based on BS 1377-2,
(i.g. non-isothermal and isothermal condition) (Grifoll et al., 2005). (1990) and BS 1377-5, (1990) for the purpose of acquiring
Fredlund et al., (2010), recognized the mechanical properties and Atterberg limits, particles size distribution, soil particle density, and
hydrological characteristics of the fractured soil could significantly saturated permeability of the kaolin soil. The properties of kaolin
change. Besides, the dual-continuum method is more reliable with soil sample are shown in Table 1.
fracture matrix compared to discrete-fractured model (Fredlund et Table 1 Kaolin soil properties
al., 2010).
The image analysis method has been used in many research Property Value
areas. This technique can be applied to investigate the liquid Liquid Limit (%) 41
contaminants and determine the saturation rate of liquid (Luciano et Plastic Limit (%) 27.50
al., 2010). The common ways to observe and monitor liquid Plasticity Index (%) 13.50
migration in soils consisting many measuring apparatuses. This Particle Density (Mg/M3) 2.65
interfere with the sample original setup and not economic. The Sand: 2mm to 0.06mm (%) 3
porosity and cracked soil are difficult to monitor by naked eyes, and Fine: Less than 0.06mm (%) 97
for this reason, digital image processing was suitable and acceptable USCS Classification Clay with low plasticity (CL)
for use in this study. According to Ngien et al., (2012), the problem Saturated Permeability (m/s) 5.42 x 10-9
of very hard to gathering data concerning liquid migration
characteristics and the physical experiments will go a long way in
the effort to comprehensively monitor, observe, understand and The method used to prepare the aggregated kaolin soil was
evaluate the polluted liquid migration into the groundwater based on the method developed by Bagherieh et al., 2009. As to
resources. Besides, a number of researchers such as Alazaiza et al., prepare the soil sample specimens, the dried kaolin powder mixed
(2017) and Sitthiphat and Siam (2016), have conducted experiments with the water to make a 30% moisture content for the soil samples.
278
The dried kaolin powder was constantly poured into the distilled
water to control the moisture content. The reason for chosen 30%
moisture content for this study was because the optimum moisture
content for the used kaolin soil was 28%. Next, as in the sample
curing period, the moisture kaolin was kept in a cool condition for a
minimum of 24 hours. In order to keep the moisture constant during
the cooling the samples were placed in a re-sealable plastic bag.
Thereafter, the cured mixtures were broken simply by hand and
passed through a 2.36 mm sieve (sieve No. 8) to obtain kaolin
granules to create the double-porosity soil structure. The kaolin
granules were placed in an acrylic soil column and compressed to a
height of 100 mm using a compression machine. The 100 mm height
of soil sample was chosen to ensure uniformity through the sample
depth. The prepared double-porosity kaolin soil sample is given in
Figure 1. Figure 3 Soil sample after vibration process with fracture
2.2 Non-Isothermal Experiment Setup

The adjustable heater machine (for the purpose of the double-
porosity soils) that is made in this study can impose heat to the
whole volume of soil with the specific temperature. The invented
adjustable heater machine design connected to two heater rods in
order to induce and transfer the heat into the soil samples. As stated
earlier one thermocouple rod was also used for the purpose of
checking the actual initial soil temperature inside the tested soil. The
cable of the rods was connected and fixed at the control panel on the
heater machine. After connecting the cable to the heater machine,
the heater induced the heat to the soil sample as in non-isothermal
condition. However, in the isothermal condition only the samples
Figure 1 The prepared double-porosity soil were placed inside the room temperature and in these steps were
carried out without using the adjustable heater machine. In each
An acrylic soil column with dimension of 150 mm high, 100 sample, the experimental measurement and setup of the V shape
mm outer diameter and 94 mm inner diameter was used for the mirror and soil column position as well as the non-isothermal heater
experiment. In order to induce the heat to the soil, two heater rods machine was arranged as shown in Figure 4 for APLs migration
and 1 thermocouple rods were installed into the centre of the soil image acquisition.
(Figure 2). The acrylic soil column with installed rods into the soil
was perfectly fixed and bolted on the vibration table in order to
prevent any movement of the soil column during vibration process.
The setup of the vibration table to vibrate the double-porosity soil
sample was developed by the author at the centre of geotropic in
Universiti Teknologi Malaysia (UTM).
Figure 4 Digital image acquisition setup with non-isothermal heater

machine
A Nikon D90 DSLR digital camera was used for APL migration
image acquisition at specific time frequency interval. The setup of
digital camera during the experiments was set at minimum shutter
speed of 1/640 second and ISO speed set at ISO-2500, which has
Figure 2: Heater and thermocouple setup been implemented for all the soil sample experiments. To soften the
problem of insufficient image acquisition, this experiment used a
The vibration frequency for the vibration table was set on the single digital camera, and two mirrors for reflection positioned
control panel with 0.98 Hz frequency with 60 seconds duration for behind the circular soil column were adjusted to enable a clear
vibration process based on the previous literature. The result of image facing the digital camera. Thus, the image of the APL
fractured double-porosity soil sample with 30% moisture content migration throughout the whole area of soil column circumference
after the vibration process is displayed in Figure 3. The concept of can be initially captured by just a single click on the DSLR digital
fracture double-porosity soil was to prove that the vibration effect camera. The light source for soil sample come from linear
had caused the double-porosity soil to fracture. The fractures on the fluorescent lamp-40 watt that was placed slightly above the circular
top of soil sample can be clearly seen. soil column.
279
Both soil samples were first sheathed in white paper with pre- 3. RESULTS AND DISCUSSION
drawn gridline (20 mm x 20 mm) onto the soil column as a control
point on the reference image. Once the reference image was taken, After the APL migration process, the top view of the fractured soil
the pre-drawn gridline paper can be removed from the circular soil samples 1 and 2 as shown in Figure 6. The actual size measurement
column. The adjustable heater machine starts to heat the soil with 30 for the column circumference zone was divided to aid visualize the
Degrees Celsius and verified by the thermocouple to confirm the crack position for samples 1 and 2.
temperature of 30o C for soil sample 1 with the non-isothermal
condition. Meanwhile, sample 2 for isothermal condition was
implemented in the room temperature of 23o C without used the
adjustable heater machine. Both experiments began by pouring the
dyed water instantaneously onto the top centre of fractured
aggregated soil sample in acrylic soil column. The quantity of 70 ml
dyed water was used in samples 1 and 2. After the dyed water had
covered the whole surface area of the fractured soil sample, the first
digital image of aqueous phase liquid migration was taken. The
dyed water migration pattern at a specific time frequency interval
was captured for the rest of the subsequent digital images. A total of
115 images were subsequently recorded in 60 minutes for soil a) Soil sample 1 (Non-isothermal) b) Soil sample 2 (Isothermal)
sample 1, while 118 images were captured in 90 minutes for soil Figure 6 APL migrated soil surface with measurement of
sample 2. actual column circumference zone for samples 1and 2
2.3 Digital Image Processing Analysis Setup The downward migration pattern of HSI contour plot of APLs in
The captured colour digital images were saved in JPEG format and fractured double-porosity soil sample with 30% moisture content for
transferred from Nikon D90 digital camera to computer for further sample 1 (non-isothermal) and sample 2 (isothermal), respectively,
image processing using Matlab routine and Surfer Software. A are shown in Figures 7 and 8. The hole appear on sample 1 was the
developed Matlab routine for digital image processing was used to effect of the rods removed from the sample. When the HIS plot in
extract the area of interest from captured image and to transform the curve joint the right and left boundary has formed the circular shape
area of interest from distorted image to a scale image via affine was clearly apparent in two-dimensional shape; however, the actual
transformation method, which involves converting the JPEG scale aqueous phase liquid migration in the acrylic soil column was one-
images to Red Green Blue (RGB) and HSI images; extracting HSI dimensional.
digital value from HSI image and saving the HSI value in a text file
using American Standard Code for Information Interchange (ASCII)
format.
First, the surfer software was used to digitize the control point
from reference image to extract actual image coordinates of control
points to serve as control coordinate. While, area of interest refers to
pre-determined migration boundary area (front image and V shape
refection image) for sample 1 and 2 that contained the APL as
shown in Figure 5. Matlab routine was then used to convert area of
interest into RGB and HSI image format, where the values were
extracted and saved in ASCII format. Matlab routine was used to (0.5 Minute)
loop three times for the subsequent digital image to extract and save
the intensity values for all three section area of interest of the acrylic
soil column. Finally, the contour plot pattern of the APL migration
in fracture porous media based on the generated HSI value by digital
image processing analysis was plotted. The HSI contour plot of APL
migration behaviour in the condition of non-isothermal and
isothermal condition can provide detailed information to facilitate
researchers to understand the pattern of APL.
(6 Minutes)
(40 Minutes)
Figure 5 Digitization of the control point for area of interest
280
Based on the observation soil sample 1, the soil condition was dried,
and process of liquid evaporation occurred due to the heat induced
in the soil sample. Both samples were not fully migrated due to
water has higher viscosity compared to previous research by Sa’ari
et al., (2015) was fully toluene migrated to the bottom of soil sample
with the lower toluene viscosity. Besides, this could be because the
physical bonding between toluene and soil weaker than water and
soil. The physical bonding between toluene and soil was attributed
to Van Der Waals Force, which are weaker than hydrogen bonding,
(90 Minutes) which has stronger physical bonding between water and soil.
Figure 7 HSI plots of downward migration in fractured
double-porosity soil for non-isothermal condition
Both samples used 70 ml dyed distilled water was poured

instantaneously on top of the soil samples surface by using glass
funnel to ensure that the APL penetrated in one-dimension. In both
experiments, the flow of the APL migration was not uniformly
downward at the front boundary horizontal line due to the non-
homogeneity of the fractured double-porosity soil structure. In soil
sample 1, the selected HSI plots of dyed distilled water migration at
intervals of 0.5, 6, 40, and 90 minutes, respectively, can been seen in
Figure 7. Based on the observation and the HSI intensity contour
plot results, the APL was not consistently downward the front x-axis (0.5 Minute)
horizontal line as the temperature and non-homogenous fractured
affect the double-porosity soil sample. Further investigation, the
faster migration occurred at the cracked soil surface condition
compared to other locations on the soil surface that were not cracked
in soil sample 1 as shown in Figure 6a. Monitor during migration
process, the dyed water completely migrated on the whole top soil
surface area into the heated fracture soil sample of the test, which
took about 25 minutes. Meanwhile, the duration for dyed water
migration from the top surface to the stop point was 90 minutes and
further observation at 120 minutes indicated no further changes in
migration pattern. 30 seconds after initiation of the experiment, the (6 Minutes)
dyed water had penetrated just 10% of overall height of soil sample
0
as shown in Figure 7. The deepest APL migration depth along the
soil column was 38 mm out of 100 mm soil column at
circumference length between 230 mm to 250 mm along the x-axis.
The APL migration stopped at 90 minutes and did not fully migrate -50
in this study because the water viscosity was higher compared to
toluene viscosity (Assael et al., 2001). Thus, characteristics of water
caused the high resistance and friction to gradual migration.
In soil sample 2, the interval of 0.5, 6, 20, and 60 minutes, -100
0 50 100 150 200 250 300
respectively, was selected for dyed APL migration HSI plot as
shown in Figure 8. The APL migration was similar to the result (20 Minutes)
found in soil sample 1 in term of the APL migration no uniformly
downward at the front boundary horizontal line. At 30 seconds after
the initiation of the experiment, the dyed APL migration reached
halfway of the test sample at the fractured location 90 mm to 140
mm along the column circumference length. For the dyed water to
completely recede from view on the whole top soil surface area into
the fractured soil sample of the test took about 20 minutes.
Meanwhile, the overall duration for dyed APL migration from the
top surface to the stop point was 60 minutes and further observation
at 80 minutes showed no changes in migration pattern where the
APL migration fully stopped where the dyed water migration (60 Minutes)
between 60 to 190 mm along the x-axis have been reached Figure 8 HSI plots of downward migration in fractured
approximately 85% downward depth of the soil column. The dyed double-porosity soil for isothermal condition
water migrated fastest between 60 to 190 mm and 250 to 280 mm
along the column circumference length. The deepest downward Figures 9 and 10 present the measured values of dyed water
migration was 85 mm out of 100 mm of the soil sample height. The saturation depth as a function of time for every 30 mm column
APL migration was not fully migrated because of the same circumference in soil samples 1 and 2, respectively. Based on the
condition like soil sample 1 due to the viscosity of the water have result in Figure 9, it was found that the cumulative saturation depth
higher friction to gradual migration. of APL migration at 240 mm column circumference displayed the
The phenomena of APL migration for non-isothermal and most critical migration downward within 30 seconds shown by the
isothermal effect was difference because of the distinctive steepest gradient of the graph lines with that duration and continue
temperature. The non-isothermal soil sample took a longer time for to gradually incline horizontally until the end of the experiment. The
the process of APL migration compared to the isothermal soil second and third critical migration was at 270 mm and 0 mm
sample faster APL migration process. This could be because of the column circumference, respectively. It within 6 minutes as shown
heat imposed in soil sample 1 has slow down the migration speed. by the sharp gradient of the graph lines within that duration and
281
continue after 6 minutes to gradually incline horizontally until the the other high speed rates occurred at 270 mm and 0 mm column
end of the experiment. Meanwhile, the rest of the column circumference with the value of 0.322 mm/s and 0.308 mm/s,
circumference positions continue a slow, decreased migration from respectively. Thus, the overall average dyed water migration speed
start until the end of the experiment. rate for soil sample 1 is 0.043 mm/s. The highest average migration
speed rate was at 240 mm column circumference zone at 0.061
mm/s. This could be because the larger fractured soil structure after
the process of vibration was at the position of 240 mm.
The soil sample 2 displayed the higher migration speed rate
from initial to 30 seconds at 90 mm column circumference zone
with the speed rate of 2.234 mm/s. Meanwhile, the column
circumference zone at 150 mm and 120 mm also have a high speed
rate with the values of 2.231 mm/s and 2.224 mm/s, respectively.
Therefore, the overall average dyed water migration speed rate for
sample 2 is 0.097 mm/s. This scenario occurred because the soil
sample 2 with isothermal condition without the temperature effect,
could cause the penetration to be faster than soil sample 1.
Table 2 Migration speed rate for every 30 mm column

circumference
Migration Flow Rate (mm/s)
Soil sample 1 Soil sample 2
(Non-Isothermal) (Isothermal)
Column Higher Average Higher Average
Circumference Flow Flow for Flow Flow for
Figure 9 Measured values of dyed water saturation depth as a Zone (mm) Between All the Between All the
function of time for every 30 mm column circumference in Initial to Time Initial to Time
sample 1 30 Interval 30 Interval
Seconds Seconds
0 0.308 0.048 0.495 0.067
Based on Figure 10, the fastest and most critical migration 30 0.262 0.045 0.193 0.034
downward to a cumulative saturation depth of APL migration 60 0.296 0.047 0.332 0.046
occurred at 90 mm column circumference within 30 seconds as 90 0.304 0.045 2.234 0.227
demonstrated by the steepest gradient of the graph lines within that 120 0.270 0.042 2.224 0.218
duration, and after 30 seconds showed a gradual downward decline 150 0.034 0.036 2.231 0.224
180 0.211 0.035 0.140 0.033
until the end of the experiment. 150 mm and 120 mm column
210 0.143 0.023 0.211 0.031
circumference showed the second and third fastest critical 240 0.339 0.061 0.151 0.031
penetration, respectively, within 60 seconds as demonstrated by the 270 0.322 0.054 0.736 0.109
sharp gradient of the graph line within that duration and continue 300 0.219 0.039 0.418 0.050
after 60 seconds to gradually incline horizontally until end of the
experiment. Meanwhile, the remaining column circumference 4. CONCLUSION
positions displayed slight decreased migration from start until the
end of the experiment. A series of physical laboratory experiment model on non-aqueous
phase liquid migration, in fractured double-porosity soil, with 30%
moisture content in the condition of isothermal and non-isothermal
condition was conducted. This laboratory experiment was
intentionally designed to investigate and differentiate the APL
migration characteristic and pattern in the fractured double-porosity
soil under the condition of different soil temperature. The digital
image processing technique using Matlab routine and Surfer
software was applied to analyse and extract the APL migration data
acquired from captured digital image. From the results observed,
both experiments indicated that in comparison to APL migration, the
isothermal soil sample was migrated faster from the surface to the
stopped point compared to the non-isothermal soil sample. The
significant finding from this study was that the APL migration in
non-isothermal soil sample undergoes the process of liquid
evaporation and could dried the soil samples after the heat induce
into the soil sample. Noteworthy, in both samples the APL
migration was not fully migrated to the bottom of soil column. This
is because the physical bonding between the soil and water
attributed to Van Der Waals Force due to stronger hydrogen
bonding.
The overall average of the fluid migration speed rates for
Figure 10 Measured values of dyed water saturation depth as a experiment 1 and 2 were 0.043 mm/s and 0.097 mm/s, respectively,
function of time for every 30 mm column circumference in sample 2 where the liquid migration speed rate was faster than 0.04 mm/s in
the previous research by Sa’ari et al., (2015). This is because this
The calculated migration speed rate for the higher and overall study applied the effect of vibration on the double-porosity soil with
average speed for every 30 mm column circumference zone was the end product of fractured double-porosity soil. Therefore, it can
demonstrated in Table 2. Soil sample 1 shows the higher migration be concluded that the factors that significantly influenced the APL
speed rate from initial to 30 seconds was at 240 mm column migration in samples 1 and 2, respectively, was the structure of the
circumference zone with the speed rate of 0.339 mm/s. Meanwhile,
282
soil sample, fractured pattern of the soil sample, physical interaction

between the liquid and soil sample, viscosity of liquid and the British Standard Institute (BS 1377-5). (1990) Methods of test for
temperature of the soil sample. The HSI value and contour plot of soils for civil engineering purposes - Part 5: Compressibility,
dyed water migration could produce detailed particulars to permeability and durability test, 389 Chiswick High Road
professionals to understand and simulate the behaviour of APL London W4 4AL, UK.
migration that could be used to identify the remediation method Fredlund, D.G., Houston, S.L., Nguyen, Q. and Fredlund, M.D.
most suitable to sustainable groundwater utilization. (2010) "Moisture Movement Through Cracked Clay Soil
Profiles", Geotechnical and Geological Engineering, 28,
5. ACKNOWLEDGEMENT Issues 6, pp865–888.
Grifoll, J., Gastó, J.M. and Cohen, Y. (2005) "Non-isothermal soil
This study was supported by the Research Management Centre water transport and evaporation", Advances in Water
(RMC), Universiti Teknologi Malaysia under Research University Resources, 28, Issues 11, pp1254–1266.
Grant – Tier 1 (PY/2016/06547) from the Ministry of Higher Krisnanto, S., Rahardjo, H., Fredlund, D.G. and Leong, E.C. (2014)
Education Malaysia. The authors would also like to thank their "Mapping of cracked soils and lateral water flow
respective University, Public Service Department Malaysia, characteristics through a network of cracks", Engineering
Geotechnical Laboratory, Hydraulic and Hydrology Laboratory, Geology, 172, pp12–25.
Engineering Seismology and Earthquake Engineering Research Li, X. and Zhang, L.M. (2009) "Characterization of dual-structure
Group (eSEER), and Survey Unit, Faculty of Civil Engineering, pore-size distribution of soil", Canadian Geotechnical
Universiti Teknologi Malaysia for kind assistance lent to this Journal, 46, pp129–141.
research. Special thanks to beloved late Professor Dr. Norhan Abd Loke, K.F., Rahman, N.A. and Nazir, R. (2017) "Experimental
Rahman for his great contribution to this research, which has been study on unsaturated double- porosity soil phenomena under
passed away in 2018The first author was supported through the vibration effect", Jurnal Teknologi, 4, Issues 79, pp65–72.
federal training award by the Public Service Department under Loke, K.F., Rahman, N.A., Nazir, R. and Lewis, R.W. (2018)
Prime Minister’s Department, Malaysia. "Study of aqueous amd non-aqueous phase liquid in fractured
double-porosity soil using digital image processing",
6. REFERENCES Geologia Croatica, 71, Issue 2, pp55–64.
Luciano, A., Viotti, P. and Papini, M.P. (2010) "Laboratory
Alazaiza, M.Y.D., Kong, S., Bob, M.M., Kamaruddin, S.A., Mohd, investigation of DNAPL migration in porous media", Journal
W. and Ishak, F. (2017) "Influence of Macro-pores on of hazardous materials, 176, pp1006–1017.
DNAPL Migration in Double-Porosity Soil Using Light Masciopinto, C., Benedini, M., Troisi, S. and Straface, S. (2001)
Transmission Visualization Method", Transport in Porous Conceptual Models and Field Test Results in Porous and
Media, 117, Issues 1, pp103–123. Fractured Media, In Groundwater Pollution Control,
Assael, M.J., Avelino, H.M.T., Dalaouti, N.K., Fareleira, J.M.N.A. Southampton, UK: WIT Press.
and Harris, K.R. (2001) "Reference correlation for the Ngien, S.K., Rahman, N.A., Ahmad, K. and Lewis, R.W. (2012) "A
viscosity of liquid toluene from 213 to 373k at pressures to review of experimental studies on double-porosity soils",
250Mpa", International Journal of Thermophysics, 22, Issues Scientific Research and Essays, 7, issues 38, pp3243–3250.
3, pp789–799. Sa’ari, R., Rahman, N.A., Latif Abdul, N.H., Yusof, Z.M., Ngien,
Bagherieh, A.R., Khalili, N., Habibagahi, G. and Ghahramani, A. S.K., Kamaruddin, S.A., Mustaffar, M. and Hezmi, M.A.
(2009) "Drying response and effective stress in a double (2015) "Application of digital image processing technique in
porosity aggregated soil", Engineering Geology, 105, Issues monitoring LNAPL migration in double porosity soil
1-2, pp44–50. column". Jurnal Teknologi, 3(72), 23–29.
British Standard Institute (BS 1377-2). (1990) Methods of test for Sitthiphat, E. and Siam, Y. (2016) "Investigation of average optical
soils for civil engineering purposes - Part 2: Classification density and degree of liquids saturation in sand by image
Tests, 389 Chiswick High Road London W4 4AL, UK. analysis method", KKU Engineering Journal, 43, Issues SI,
pp147–151.
283
Lessons Learned from Pressuremeter Tests on Stone Columns

Richard Ong1
1
Regional Technical Director, Menard Geosystems Sdn Bhd, Subang Jaya, Malaysia
E-mail: richard@menard-asia.com
ABSTRACT: The Menard pressuremeter was developed by Louis Menard in the 1950s. Since its availability as a tool to the engineers for
the last 60 years, pressuremeter has been used extensively for ground improvement projects. It is proven to be a valuable tool throughout the
life cycle of a project, from the geotechnical investigation for design works to acceptance test conducted to verify the construction completed
for a ground improvement project. This paper presents the application of Menard pressuremeter in the construction of an LNG tank in South-
East Asia using stone columns as the foundation system. Some of the important lessons learned from the pressuremeter tests conducted for
this project are presented and discussed in this paper.
Keywords: Menard pressuremeter, ground improvement, stone column, LNG tank.
1. INTRODUCTION tank on the project site. This former tank applied some 150 kPa on
the ground. Thus, the area in Layer 2 that was under the old tank
Stone columns are often used as reinforcing and stiffening elements footprint is over-consolidated, whereas it is normally consolidated
in ground improvement works to increase the bearing capacity and everywhere else. Groundwater level is at about 1.0 m below the
reduce the settlement of the treated ground. Numerous methods existing ground level.
(e.g., Priebe (1995), Goughnour and Bayuk (1979), Balaam and
Booker (1981)) are available to calculate the settlement of ground Table 1 Soil characteristics
improved by stone columns. These methods rely on the modulus of
the stone column, Ec and that of the soil, Es to determine the SPT CPT
Layer
settlement reduction factor following ground treatment with stone N-value qc (MPa)
columns. In recent years, numerical modelling such as finite element 1b: Medium dense to dense sand 23 14.5
analysis has also been used in the design of stone column projects. 1c: Medium dense sand 16 8.2
These types of modelling also require reliable soil and stone column 2: Very soft to soft clay 2 0.5-0.9
moduli to be utilised in the analysis. One of the in-situ tests that can Note: The SPT N-value and qc shown are average values.
provide a direct measurement of the soil modulus is the Menard
pressuremeter test (PMT). This paper demonstrates the use of PMT 2.2 Description of the Ground Improvement Works
in the construction of stone columns for the foundation of a large
LNG tank in South-East Asia. The difficulties encountered to The overall ground improvement works include the removal and
conduct the PMT inside stone columns for the project and the replacement of Layer 1a (i.e., loose sand layer) with well compacted
solution adopted are highlighted. Finally, some lessons learned from granular material and the wet top feed stone column method. Under
examining the results of the PMT by considering the nature and the LNG tank footprint, Layer 1a will be excavated and replaced
characteristics of the soils, the intensity of the stone column with well compacted granular fill (with minimum oedometric
compaction and the measured Menard pressuremeter modulus, EM at modulus, Eoed of 10 MPa) from +1.2 m EL to +2.2 m EL. From +2.2
different depths are presented. m EL to the bottom of the tank slab (+2.9 m EL), the tank pad will
be constructed with well compacted granular material with Eoed of
2. LNG TANK PROJECT 40 MPa. The scope of the stone column works includes penetration
of the vibroflot down to 10.7 m depth from working platform level
2.1 General Description of the Project (i.e., from +2.7 m EL to -8.0 m EL). This is followed with the
The purpose of the ground improvement works is to reinforce the installation of stone columns with average diameter of 1,200 mm
soft compressible cohesive soil (i.e., Layer 2 in Figure 1) with within the clay layer and 600 mm in the sand layer as shown in
stones and to improve the sand material on top of it (i.e., Layers 1b Figure 1.
and 1c in Figure 1) to construct a Liquefied Natural Gas (LNG) The tank dead load is shown in Figure 2.
storage tank of 77.9 m diameter on top of the treated ground.
Figure 2 Tank geometry with dead load at the end of construction
Figure 3 shows the live load during hydrotest, which is the most
critical stage of the live load applied to the tank. The ground
improvement was therefore designed for it based on the live load
Figure 1 Cross section of ground improvement with stone column during hydrotest.
The characteristics of the layers of soil to be treated are

summarised in Table 1. The sand layers were quite homogeneous
according to the standard penetration test (SPT) and cone
penetration test (CPT) results. For the very soft to soft clay layer,
two distinct zones were identified due to the existence of a former Figure 3 Live load during hydrotest
284
The ground improvement scheme for the LNG Tank foundation has ensure the probe is within the stone column at the soft clay layer.
been designed with the aim to minimise two differential settlements: Thus, the idea of constructing “test” stone column within the clay
a. Settlement between centre and edge of slab must be lower than layer was adopted. The “test” stone column was constructed by
1/300 (dishing criterion); installing stone column only within the clay layer with stone
b. Settlement between two diametrically opposed edge points consumption and compaction energy similar to those applied for the
must be lower than 1/500 (tilting criterion). construction of normal stone columns. The difference was that
The final design is for the stone columns to be installed with a 2 m beyond the clay layer, the hole created for the construction of the
square grid spacing pattern, giving the area replacement ratio, ar of “test” stone column was backfilled using sand with little
7% within the sand layer and 28% within the clay layer. compaction. This ensured the pressuremeter probe can penetrate to
the “test” stone column within the clay layer and meaningful test
2.3 Testing Programme can be conducted.
Apart from the digital recordings of the construction parameters
0 0
(primarily depth and compaction effort versus time) for every stone
column using a data logger during construction of stone columns, -1 -1
plate load test (PLT) and pressuremeter test (PMT) were specified as 2.3 15
acceptance tests. The PLT was carried out on top of some stone -2 -2
2.6 18
columns using 600 mm diameter steel plate to check on the -3 -3
deformation modulus based on the settlement versus load graphs 2.6 15
obtained. The PMT was carried out within some stone columns at 1 -4 -4
m interval to obtain the limit pressure, PL and Menard pressuremeter 2.6 16
-5 -5
modulus, EM. The acceptance criteria for PMT were to achieve
Depth (m)
2.4
Depth (m)
13
minimum PL of 1.5 MPa and EM of 15 MPa at each test location. -6 -6
These values were derived based on French recommendations NFP 1.9 11
11-212 (2005), which recommends the Young’s modulus of the -7 -7
stone column (average value over the stone column), Ec to be 60 2.4 12
-8 -8
MPa. 1 6
-9 -9
0.2 2
-10 -10
3. PRESSUREMETER TEST RESULTS 0.2 2
-11 -11
3.1 Menard Pressuremeter Test (PMT)
The Menard pressuremeter was developed by Louis Menard for -12 -12
0 1 2 3 4 0 10 20 30 40
commercial application through his first Menard company known as PL (MPa) EM (MPa)
“Centre d’Etudes Menard” in 1957. The original pressuremeter was
developed in an attempt to solve the problem of sampling Figure 4 Initial pressuremeter test result
disturbance and to ensure that the macro-fabric of the soil can be
adequately represented from an in-situ test. To date, the Menard 3.3 PMT Result for Stone Column within Clay Layer
pressuremeter has evolved as part of an international standard
procedure to give design parameters directly. This procedure covers Figure 5 shows the typical PMT result obtained from a “test” stone
the type of probe, installation, testing and interpretation as provided column formed within the clay layer. It is shown that the PL is about
in the publication “Interpretation and Application of Pressuremeter 1.1 MPa and the EM is around 8.5 MPa. This indicates that the PL
Test Results to Foundation Design” published in Sol-Soils (Menard, and EM are below the specified values despite being much higher
1975). The pressuremeter test provides two main soil parameters, than the natural soil’s PL and EM. The EM of 8.5 MPa translates to
namely Menard pressuremeter modulus, EM and limit pressure, PL, Young’s modulus within stone column, Ec of 34 MPa (i.e., Ec 
which can be used directly for settlement and bearing capacity EM/; where   0.25 for gravel). This is about 11 times higher than
calculation, respectively. the Young’s modulus of the natural clay (i.e., Es of about 3 MPa)
and it is well above the typical modular ratio (i.e., Ec/Es) for the
3.2 Initial PMT Result application of stone columns in soft soil. In fact, for the design
approaches using elastic methods that consider vertical deformation
Figure 4 shows the plot of PL and EM versus depth from the initial only, Barksdale and Bachus (1983) have suggested that Ec/Es in
pressuremeter test (PMT) conducted from the working platform practice to be ranging from 3 to 10, depending on the area
level at +2.7 m EL. The result suggests that the upper 5 m of the replacement ratio, ar adopted. The reasons why such a high stiffness
stone column was well constructed. However, below that level it ratio was obtained for this project are likely to be due to the fact that
appears as though the stone column was not properly compacted high area replacement ratio, ar was adopted with column diameter of
within the medium dense sand layer (i.e., Layer 1c) and the test 1.2 m and high compaction energy was applied during construction.
result within the clay layer is even worse where the PL and EM
obtained are not any better than the PL and EM of the natural soil.
-8 -8
This kind of PMT test result was not expected in the project as 1.2 8
D e p th (m )
-9 -9
D e p th (m )
the site recordings during construction showed that stone

consumption was much higher than the theoretical volume required 1 9
and very high compaction energy (i.e., not less than 180 A) was -10 -10
applied during the construction of the stone column. Hence, it is 1.2 8
suspected that the pressuremeter probe might have walked out of the -11 -11
stone column as it went deeper from the working platform level. 0 1 2 3 4 0 10 20 30 40
This is not an uncommon phenomenon when conducting tests inside PL (MPa) EM (MPa)
a stone column. Figure 5 PMT result of stone column within clay layer
A few other attempts had been made to ensure verticality of the
pressuremeter probe during penetration but were not fruitful to
285
3.3 PMT Result for Stone Column within Sand Layer 20 (i.e., Ec/Es  60/3). However, as noted from the PMT results, the
modular ratio achieved is only about 11. Clearly, a constant Ec
Figure 6 shows the typical PMT result for the test conducted on a
should not be specified and a limit on the modular ratio should be
stone column within the sand layer when the pressuremeter probe
imposed in the design of stone columns. The following limits can be
did not walk out from the centre of the column. The PL is about 2.8
considered:
MPa and the EM is around 22 MPa, which are higher than the
required values in the specification. The Young’s modulus of the  In sand, despite the fact that relatively high Ec is achievable, the
stone column, Ec is calculated to be 88 MPa, which is 3 to 4.5 times modular ratio is likely to be around 3 to 5 due to the relatively
higher than the Young’s modulus of the original sand layer (i.e., Es high Es for sand.
of 20 MPa to 30 MPa).  In clay, although relatively low Ec is the norm, the modular ratio
is likely to be around 5 to 12 due to the relatively low Es for soft
0 0 clay.
The above limits are similar to the findings by Barksdale and
-1 -1 Bachus (1983), which suggested Ec/Es in practice is ranging from 3
1.8 11
-2 -2
to 10 and Han (2012) which noted that the modular ratio is from 1 to
2.8 16 5 for flexible columns and 5 to 10 for semi-rigid columns.
-3 -3
2.4 13 4.3 Settlement Calculation
Depth (m)
Depth (m)
-4 -4
3.1 26 Based on the experience in this project, it appears that it is better to
-5 -5 consider the minimum modular ratio between stone column and soil
2.9 22
-6 -6 (i.e., Ec/Es) in the design of stone column than specifying a constant
3 25 stone column modulus, Es to be achieved. In fact, designing and
-7 -7 specifying stone column projects using the modular ratio should
3.4 34 come quite naturally as most of the analytical methods to calculate
-8 -8
28
settlement of the soil treated with stone columns are using this ratio
3.3
-9 -9 to determine the settlement reduction factor for the settlement
0 1 2 3 4 0 10 20 30 40 calculation. Settlement reduction factor, n is the ratio of the
PL (MPa) EM (MPa) settlement of the untreated ground to the settlement of the ground
improved with stone column.
Figure 6 PMT result of stone column within sand layer
Zooming into the Priebe’s (1995) method, which is one of the
most popular methods used for calculating the settlement of ground
4. DISCUSSION ON THE PMT RESULTS treated with stone columns, the maximum allowable settlement
4.1 Stone Column Modulus reduction factor is given by the equation below:
In this project, the modulus of the stone columns, Ec had been
specified to be 60 MPa throughout the length of the column. The
PMT results show that the Ec in the clay layer and sand layer is 34
MPa and 88 MPa, respectively. Clearly, the value of Ec is not where:
constant throughout the length of a stone column. From various nmax  maximum allowable settlement reduction factor
literature, the value of Ec is shown to range from 25 MPa to 100 ar  area replacement ratio  Ac/A
MPa and it is a function of the confining pressure (Castro, 2017). In Ac  cross section area of the stone column
fact, the value of Ec is likely to be interdependent on factors such as A  tributary soil area of the stone column (or unit cell area)
the stiffness of the natural soil, the compaction energy provided, the Ec/Es  modular ratio between stone column and soil
area replacement ratio (i.e., spacing and diameter of stone columns),
etc. Therefore, it may not be suitable to specify or design a stone It should be noted that the above equation (1) is not normally the
column project using a constant Ec throughout the treatment depth, settlement reduction factor to be adopted in the calculation of
especially when the subsoil is heterogeneous. settlement after stone columns installation using Priebe’s (1995)
method, but it is the cap of the settlement reduction factor that can
4.2 Modular Ratio be adopted in the analysis. To illustrate quickly the effect of natural
Table 2 shows the modular ratio between stone column and natural soil and stone column moduli in settlement calculation, the above
soil (i.e., Ec/Es) obtained from the pressuremeter test conducted in equation is adopted to calculate the maximum settlement reduction
this project. factor for this project as tabulated in Table 3.
Table 2 Modular ratio Table 3 suggests that by using a constant Ec of 60 MPa to

calculate the settlement of the soft clay treated with stone columns,
Stone Natural soil Stone column Modular the settlement calculated might be about 38% lower than the reality
column EM Es EM Es ratio (i.e., the settlement calculated based on the actual Ec of 34 MPa).
in (MPa) (MPa) (MPa) (MPa) (Ec/Es) Indeed, this is quite an alarming discrepancy. For the stone columns
installed in the sand layer, the issue of using the constant Ec of 60
Sand 7 to 10 20 to 30 22 88 3 to 4.5
MPa might not be that severe with the calculated settlement to be
Clay 2 3 8.5 34 11 about 6% higher than the reality. This implies that the Ec to be
adopted in the design, especially for soils with low Es (e.g., the soft
The modular ratio is not constant and ranges from 3 in dense clay layer in this project) needs to be selected carefully.
sand to 11 in soft clay. This suggests that the modular ratio is likely
to be higher in soils with low Young’s modulus than those with high To help the selection of the Ec to be adopted in the settlement
Young’s modulus. Having said so, there should be a limit on the analysis, it might be useful to rely on the proposed limits of modular
maximum modular ratio to be adopted in the design. Take for instant ratio. It can be seen from Table 3 that the nmax for the clay treated
the targeted Ec for this project, which is 60 MPa. If this targeted Ec with stone column is within the nmax calculated using the proposed
is indeed achievable in the soft clay layer, the modular ratio will be range of modular ratio for soft clay (i.e., 5  Ec/Es  12). For the
sand layer treated with stone column, the nmax measured at site is
286
also within the proposed modular ratio for sand (i.e., 3  Ec/Es  5). 5. CONCLUSION
Indeed, the proposed modular ratio seems to be a good tool to
This paper has presented the application of Menard pressuremeter
estimate the range of settlement reduction factor following the
test as part of the quality control measures for the application of
installation of stone columns to treat the soils in a project.
stone columns as the foundation system for a large diameter LNG
tank. The challenges associated with conducting pressuremeter test
Table 3 Calculation of settlement reduction factor
inside the stone column is presented and the solution to overcome
Natural Ec Es these challenges is provided. The pressuremeter test results from this
ar Ec/Es nmax Note project are presented showing the limit pressure, PL and
soil (MPa) (MPa)
pressuremeter modulus, EM obtained inside stone columns within
34 3 11.3 3.92 (a) the sand and clay layers. It is shown that the PL and EM inside stone
60 3 20 6.37 (b) columns are interdependent of the nature and characteristics of the
Clay 28%
- - 5 2.13 (c) natural soils treated with stone columns, the area replacement ratio
- - 12 4.11 (d) (i.e., spacing and diameter of the stone columns) and the compaction
energy adopted during construction of the stone columns. The
88 30 2.9 1.14 (a) pressuremeter test results also suggest that specifying or designing
Dense 60 30 2 1.07 (b) using a constant Ec throughout a stone column might not be
7%
Sand - - 3 1.14 (e) appropriate, especially for heterogeneous soil. It is best to utilise the
- - 5 1.28 (f) modular ratio between stone column and soil (i.e., Ec/Es) for the
settlement analysis of soil treated with stone columns.
Note:
(a) Actual data from project
6. REFERENCES
(b) The Ec is specified to be 60 MPa
(c) Using lower bound of modular ratio (i.e., Ec/Es  5) Balaam, N. P., and Booker, J. R. (1981). “Analysis of rigid rafts
(d) Using upper bound of modular ratio (i.e., Ec/Es  12) supported by granular piles”. International Journal for
(e) Using lower bound of modular ratio (i.e., Ec/Es  3) Numerical and Analytical Methods in Geomechanics, Vol. 5,
Issue 4, pp379-403.
(f) Using upper bound of modular ratio (i.e., Ec/Es  5)
Barksdale R.D. and Bachus R.C. (1983). “Design and Construction
of Stone Columns Volume I”. Federal Highway
4.4 Recommendations Administration Report FHWA/RD-83/026, National
The above section shows that the important parameters for the Technical Information Service, Springfield, Virginia.
calculation of settlement of a ground treated with stone columns are Castro, J. (2017). “Modeling Stone Columns”. Materials, Vol. 10
the area replacement ratio, ar and the modular ratio, Ec/Es. The area (7), Article 782 (https://doi.org/10.3390/ma10070782).
replacement ratio, ar is related to the diameter and spacing of stone Goughnour, R. R., and Bayuk, A. A. (1979). “Analysis of Stone
columns which can be designed or determined with relatively high Column – Soil Matrix Interaction Under Vertical Load”,
precision. The difficult part in the design or settlement calculation of Proceedings of International Conference on Soil
ground treated with stone columns is determination of the modulus Reinforcement, Vol. 1, Paris, pp279-285.
of stone column, Ec to be adopted. It has been shown that the Han, J. (2012). “Recent Advances in Column Technologies to
approach of using a constant Ec in the settlement calculation of Improve Soft Foundations”, Proceedings of International
ground treated with stone columns might not be a good idea, Conference on Ground Improvement and Ground Control,
especially for heterogeneous soils (with mixture of sand and clay or Wollongong, Australia, pp99-113.
silt). As such, the modulus of stone column, Ec should not be Menard, L. (1975). “The Menard Pressuremeter: Interpretation and
specified with a constant value, especially with a high value that Application of Pressuremeter Test Results to Foundation
might not be achievable for stone columns constructed in soft soils Design”. General Memorandum D.60.AN. Sols Soils, N° 26,
due to the low confining pressure from the soft soils. Thus, it is pp5-43.
recommended for the modular ratio, Ec/Es with the limits proposed Normes Françaises NFP 11–212. (2005). “Recommandations sur la
on the above to be used in the settlement analysis to get the idea of conception, le calcul, l’exécution et le contrôle des colonnes
the likely improvement factor of installing stone columns for a ballastées sous bâtiments et ouvrages sensibles au tassement”.
project. Following this, a verification at site can be done by Revue Française de Géotechnique, N° 111, pp3-16.
conducting in-situ test such as the pressuremeter test. Finally, if Priebe, H. J. (1995). “The design of vibro replacement”. Ground
necessary, improvements or changes in the design can be conducted Engineering, December, pp31-37.
before full production works is carried out.
287
Correlations between Gradation, Physical and Mechanical Parameters

for Material Embankments Reclamation
Herman Wahyudi1 , Yudhi Lastiasih2
1,2
Department of Civil Engineering, Institute of Technology Sepuluh November, Surabaya, Indonesia
E-mail: herman_its@yahoo.com; yudhi.lastiasih@gmail.com
ABSTRACT: A structural reclamation embankment material generally requires minimum sand content of 50%, maximum gravel of 30%
and silt & clay of 20%. Related to quality control material in the field, it is desirable to only do sieve analysis with some gradation coefficient
and water content test to know the correlation some parameters of soil density such as dry density, void ratio, effective cohesion, internal
angle friction & CBR. Thus it can be determined at that time in the field whether the embankment of reclamation has been qualified or not.
To study the correlation between gradation and CBR with physical and mechanical parameters of embankment material then performed
laboratory tests, among others CBR, sieve analysis, volumetric gravimetric, proctor and direct shear tests. Research through laboratory tests
to study the correlation between the CBR with the physical parameters of the soil embankment uses reclaimed embankment material
consisting of sand (pass of sieve No.2) and clay-silt (pass of sieve No. 200). Samples are made of 5 types with composition sand and silt-clay
as follows 100% : 0% , 95% : 5% , 90% : 10% , 85% : 15% , 80% : 20%. The next step is the compaction of modified proctor and CBR test
with saturation-unsaturated condition and then volumetric-gravimetric and specific gravity test. Other tests such as sieve analysis and
hydrometer test are performed to determine Cu (coefficient of uniformity) and Cc (coefficient of curvature). This test is done using
reclamation embankment material from river, beach and hill quarry, while the material from the selected quarry is material containing
maximum 20% clay and sand 80% minimum. The next step is compacted by Modified Proctor and Direct Shear testing with saturated water
content (on Zero Air Void line) and unsaturated (on wopt). The results of the tests indicate that the increasing value of Cc/Cu then the value of
the maximum dry unit weight (dmax), effective cohesion and effective internal friction angle decreases and increases the optimum water
content. In addition the test results show that the increasing value of d then the CBR value increase and the void ratio (e) and porosity (n)
decrease. The correlation between sieve analysis and CBR with soil physical and mechanical parameters is applicable for reclamation
material with minimum sand composition 80% and clay maximum 20% and the limit of Cu is 1<Cu <50.
Keywords: embankment, reclamation, density, shear strength, CBR.
1. INTRODUCTION - What is the effect of water content (w) on dry density (d) and
soil shear strength ( and C).
Implementation of coastal reclamation requires very large
embankment material. The general requirements of embankment
This research is expected to be a solution in the selection of
material must be fulfil maximum 30% gravels, minimum 50% sand
reclamation material. The expected solution is related to the material
and maximum 20% clay and density must be comply with
composition and parameters based on the required CBR, in addition
requirements. Generally the field practitioners want to know quickly
to estimate the required density based on the composition of
and easily when the embankment material arrives in the field by
embankment material grain size.
only observe at the composition of sand and clay-silt. Based on the
composition can be determined whether the material involves the
requirements or not as a material embankment. Furthermore, the 2. LITERATURE STUDY
required density can be determined by performing one test only. The 2.1 Compaction
test can be a CBR test or a sieve analysis test.
Determination of the density level of a soil can be seen from Lee and Suedkamp (1972) concluded that the soil compaction curve
three parameters: relative density (Dr), dry unit weight (d) and void can be divided into four types. Type A compaction curve has bell-
ratio (e). Relative density is only used for granular soil type, while shaped. Type A is generally present in most clay soils with a liquid
dry unit weight (d) and void ratio (e) for all types of fine and limit value (LL) between 30 - 70. Type B curve is one-half peak,
generally present in the sand with LL<30 (type B curve is a more
coarse-grained soils. Therefore the dry unit weight (d) and void
suitable result with condition of our predominant sand test sample).
ratio are more suitable for use in reclamation work because the soil
Type C has double peak curve, found on soil with LL<30 or LL>70.
type of embankment consists of fine and coarse grained soils. The
The D-type curve is odd-shaped, generally present in soils having
determination of grain size distribution is very important in order to
LL>70.
determine including into well graded, poorly graded or gap graded.
These gradations are determined from the value of uniformity
The relationship between moisture content (w) and dry unit
coefficient (Cu) and the curvature coefficient (Cc), which greatly
weight (γd) and weight of zero air void volume (γZAV) can be
affects the density of the embankment. In addition, the density is
formulated as follows:
also influenced by water content. Water content in the field is
influenced by the rainy season and reclamation work is always  sat
associated with tidal conditions. This submerged and non-
d  (1)
1
submerged condition affects the compaction. Such conditions affect
the compaction process and the stability of embankment.  G
S  100%    W s
(2)
The existing problems based on above condition are as follows : r ZAV 1  G  
s
- How does the composition of sand and silt-clay affect the dry
unit weight (d), and the correlation of soil composition with the
where :
CBR value.
- How is the influence of water content and the relationship sat = saturated unit weight of soil
between void ratio and CBR value. w = unit weight of water
- How is the relation between Cu, Cc, gradation of grain size with Sr = degree of saturated (%)
dry unit weight (d), internal friction angle () and cohesion (C), Gs = spesific gravity
for saturated and unsaturated initial conditions. w = water content
288
Day (1997) says that the requirement factors decrease of void the contrary it is said to be negative if the increase (decrease) x is
ratio (e) at the compaction are the distribution of grain size, the ratio generally followed by a decrease (increase) y.
between d100/d0 (ratio between the largest and smallest diameter) The strength of the relationship between x and y is measured by
clay (with low activity) to fill the smallest void and compaction the correlation coefficient (denoted r). The correlation coefficient
process, to compress soil particles into more dense arrangement. value is -1 ≤ r ≤ 1, which means:
If r = 1, the relationship of x and y is perfect and positive (close to 1,
2.2 Swelling ie very strong and positive relationship).
R = -1, the relationship of x and y is perfect and negative (close to -
Swelling calculation used the equation below: 1, ie very strong and negative relationship).
H R = 0, the relationship of x and y is very weak and there is no
1   100% (3)
relationship.
H0
It says x affects y if the change in value of x will cause a change
where : in the value of y; that means the rise and fall of x will make the y
ε1 = axial strain (%) value also up and down, so the y value will vary, either to the y
ΔH = change of sample height (mm) average or to the linear line representing the scatter diagram.
Ho = initial of sample height (mm) However, the rise and fall of y is such that the y value varies, not
only due to x, because there are other factors that cause it.
Previous studies have been conducted to predict the magnitude The magnitude of the contribution of x to the rise and fall of the
of swelling. One of which is the research conducted by Seed et al. y value is obtained from the coefficient of determination, denoted by
1962, it discuses the correlation between plasticity index (IP) with R2. If r = 0.92 for example, then the value of R2 = (0.92)2 = 0.85
swelling potential (Table 1). (85%), ie the amount of factor x contribution to the rise and fall of y
is 85%, while 15% by the other factor. In addition to the value of r
Table 1 Correlation between swelling (ε1) and Plasticity Index (IP) and R2 is also required standard deviation value (s) to determine
(Seed et al., 1962) whether valid or not valid data from the distribution of data
obtained. The smaller standard deviation value, the result of the
IP (%) εs (%) Swelling potential
existing analysis is closer to mean of linear regression curve, which
0 – 10 0 – 1.5 Low
means the existing data is more valid. Correlation regression theory
10 – 20 1.5 – 5 Mean
is used to prove that the relationship between parameters sought is
20 – 30 5 – 25 High valid.
>35 >25 Very high
2.5 Correlation of CBR with Physical Parameter of Soil
2.3 Gradation
Several studies related to CBR correlation and other soil physical
The distribution of soil grain size is known as gradation (Bowles, parameters have also been performed by several researchers. Shirur
J.E. Hainim, Johan K, 1984). Gradation is usually known on coarse et al. (2014) with submerged soil having LL between 20 - 70 gives
grained soils. The gradient indication can be calculated numerically CBR relationship and parameters as follows:
from the grain size curve by using the uniformity coefficient (Cu)
defined as : CBR = 5.09477-0.09323 (LL) + 0.10939 (SL) + 0.022566 (SI)
D CBR = 5.813 – 0.007826 (LL) + 0.12097 (PL)
Cu  60 (4) CBR = -4,8353 – 1,56856(OCMC) + 4.6351 (MDD)
D10 CBR = -3.2353 – 0.06939 (PI) +2.8(MDD)
The shape of the curve defined as the coefficient of curvature (Cc), CBR = 6.5452 – 0.07703(OMC) – 0.10395(PI)
Cc 
D30 2 (5) where :
D60 D10 CBR = california bearing ratio
LL = liquid limit
where :
PL = plastic limit
D10 = grain diameter at 10% through sieve SL = shrinkage limit
D30 = grain diameter at 30% through sieve PI = plasticity index
D60 = grain diameter at 60% through sieve OMC = optimum moisture content
MDD = maximum dry density
When the Cu value gets larger (≥4), the soil grain size
distribution becomes heterogeneous, indicating that the gradation of Zohib et al. (2016) with soil type SW-SM (well graded sand
the soil is well graded. And the other hand, if the value of Cu is containing silt) submerged condition has the following relation:
getting smaller then the grain size is getting the same (uniform). CBR = 0.142 (%fines) + 0.0262 (LL) +0.0283 (OMC) + 1.043
The curvature coefficient (Cc) is used as a reference of the grain (MDD) -17.029
size distribution curve shape between D60 and D10. The ideal grain The parameter relation has been valid since it has R2 = 0,7 - 0,836
size distribution curve shape of the Cc value lies between 1 and 3. If
the value of Cc <1 then the shape of the grain size distribution curve 3. RESULTS
shows the curved shape above the ideal curve and if Cc >3 then the
shape of the soil grain size distribution curve shows the curved Based on previous research, the authors conducted laboratory
shape below the ideal curve. experiments on two types of soil samples i.e soil samples made so as
to have the composition of sand and clay 100%: 0%, 95%: 5%,
2.4 Regression Correlation Theory 90%: 10%, 85%: 15%, 80%: 20% and soil samples taken from
various locations as in Table 1. In all the compositions of sand and
Correlation is the relationship degree between several variables that clay, the greater the moisture content (wc) the greater the dry
aim to know how well a linear equation or other equations. density (γd) as shown in Figure 1. At certain water content γd will
Relationships x and y are said to be positive if the increase decrease, because the percentage of water that fills the pores
(decrease) x is generally followed by an increase (decrease) y. On between the grains enlarged. Therefore the percentage of incoming
solid granules is not maximal. The greater the percentage of clay-silt
289
in the mixture, the optimum water content will be greater, because Figure 2 shows for all the composition of sand and clay tend to
the content of large clay (SiO2) can absorb water (H2O). In terms of greater the dry unit weight (γd) the greater the CBR value.
density (γdmax), more greater the percentage of clay-silt then more This is due to the greater unit weight of soil thus the more dense soil
greater γdmax will be but in the mixture clay-silt greater than 10%, and the soil bearing capacity increase, which is seen with the value
the γdmax decreases. It can cause instability such as low bearing of CBR increase.
capacity and large settlement.
Table 2 Initial Condition of Soil Embankment Material

Sieve-
Hydrometer
Analysis
Location Gs Cu Cc Cc/Cu
Silt-
Sand
clay
(%)
(%)
Kedungombo
2.87 86.57 13.43 43.67 2.94 0.07
River
Lengkong Figure 3 Relationship CBR vs void ratio (e) for unsoaked condition
2.67 86.80 13.20 50.00 5.56 0.11
River
Lanang Figure 3 shows that the CBR value increases hence the void
2.87 99.82 0.18 4.68 0.72 0.15
River ratio (e) decreases. This is due to the increasing value of CBR thus
Ramania Hill 2.56 97.04 2.96 4.13 0.90 0.22 the soil occur granules reposition. Therefore, the void ratio is
Brantas smaller.
2.84 96.26 3.74 2.76 0.87 0.32
River
Complong
2.68 99.79 0.21 2.36 0.82 0.35
Beach
Kenjeran
2.61 99.17 0.83 2.20 0.89 0.40
Beach
Prigi Beach 2.92 99.87 0.13 2.07 0.90 0.44
Talambung
2.61 94.13 5.87 2.68 1.25 0.46
River
Lombang
2.63 99.46 0.54 1.90 0.90 0.47
Beach
From the sieve and hydrometer test results (Table 1) shows that
the embankment material from rivers and hills content of the fine
Figure 4 Relationship CBR vs porosity (n) for unsoaked condition
fraction (silt-clay) is greater than 2%, except in Lanang River
Ngrambe Ngawi containing only the fine fraction 0.18%. When
Figure 4 shows that the CBR value increases hence the porosity
compared with the location of embankment material from the beach
(n) decreases. This is due to the increasing value of CBR thus the
fine fraction has ranged between 0.13% - 0.83%.
soil occur granules reposition. Therefore, the porosity is smaller.
Based on Figure 2 to 4 can be obtained the relationship between
CBR with γd, e, n for unsoaked condition as shown in the following
equation :
a. Sand 100% dan Silt-Clay 0%
CBR = 29.583(γd) – 35.98 (R2 = 0.9800)

CBR = -34.871(e) + 40.175 (R2 = 0.9821)
CBR = -0.953(n) + 54.791 (R2 = 0.9917)
b. Sand 95% dan Silt-Clay 5%

Figure 1 Relationship dry unit weight (d-gr/cc) vs water content
(Wc-%)
CBR = 37.386(γd) – 36.258 (R2 = 0,9829)
CBR = - 44.089(e) + 59.569 (R2 = 0,9173)
The correlation between γd, e, n with CBR for unsoaked
CBR = -1.1371(n) + 75.442 (R2 = 0,9311)
condition for sample type 1.
c. Sand 90% dan Silt-Clay 10%
CBR = 31.989(γd) - 20.096 (R2 = 0,9373)

CBR = -36.562(e) + 60.134 (R2 = 0,9961)
CBR = -0,8238(n) + 69.165 (R2 = 0,9986)
d. Sand 85% dan Silt-Clay 15%
CBR = 44.071(γd) – 49.913 (R2 = 0,8760)

CBR = -52.397(e) + 59.786 (R2 = 0,9072)
CBR = -1.3045(n) + 77.028 (R2 = 0,9279)
Figure 2 Relationship CBR vs unit weight (d-gr/cc) for unsoaked
condition
290
e. Sand 80% dan Silt-Clay 20% 4. CONCLUSION
CBR = 38.524(γd) – 46.992 (R2 = 0.9991) 1. From the results of this study concluded that optimal
CBR = -29.128(e) + 39.851 (R2 = 0.9977) composition of reclamation embankment material to density is
CBR = -0.709(n) + 48.836 (R2 = 0.9992) 80% sand and 20% clay. If the content of clay is multiplied (>
10%) or increases slightly (<10%), resulting in decreased
The correlation between γd, e, n with CBR for soaked condition density.
can be seen in the following equation : 2. The results of this study indicate that the value of CBR, physical
parameters and material density in unsaturated conditions is
a. Sand 100% dan Silt-Clay 0% more optimal than with saturated conditions.
3. The most dominant soil parameter of the effect on the CBR value
CBR = 134.95(γd) – 214.94 (R2 = 0.9956) is the dry unit weight parameter (γd), this is because the CBR
CBR = -75.451(e) + 67.337 (R2 = 0.9788) value increase if the soil density increases.
CBR = -2.2052(n) + 105 (R2 = 0.9699) 4. The larger the Cc/Cu ratio the smaller dry unit weight maximum
(dmax), the shear strength of the soil (effective cohesion, C' and
b. Sand 95% dan Silt-Clay 5% effective shear angle, ') while the larger optimum water content
(wopt). This is due to the greater Cc/Cu ratio, the embankment
CBR = 55.967(γd) – 90.89 (R2 = 0.9453) material tends to be uniform and the smaller the grain diameter
CBR = - 66.287(e) + 51.681 (R2 = 0.9699) of the soil. Therefore, there is void resulting in density, bonding
CBR = -1.6763(n) + 74.39 (R2 = 0.9829) and interlocking between granules getting lower. At the same Cc
/Cu value, the shear strength of the soil (effective cohesion and
c. Sand 90% dan Silt-Clay 10% shear angle in effective) in the unsaturated initial state is greater
CBR = 58.014(γd) – 92.613 (R2 = 0.9969) than in the saturation initial state. The presence of water-
CBR = -70.884(e) + 56.679 (R2 = 0.9998) saturated conditions will make the distance between the bonding
CBR = -1.7141(n) + 78.045 (R2 = 0.9991) granules getting away and consequently reduced cohesion and
can block interlocking or friction between grains.
d. Sand 85% dan Silt-Clay 15% 5. Materials from the ocean tend to have greater optimum water
content to reach their density compared to materials from rivers
CBR = 37.995(γd) – 54.207 (R2 = 0.9685) because the embankment material on the coastal site is
CBR = -42.771(e) + 40.517 (R2 = 0.9884) dominated by fine sand with a grain diameter smaller than 0.425
CBR = -1.1253(n) + 56.692 (R2 = 0.9760) mm, requiring larger water to be solid compared with the
embankment material of the river with an average content of fine
e. Sand 80% dan Silt-Clay 20% sand is less than 40%.
CBR = 61.895(γd) – 102.47 (R2 = 0.9995) 5. ACKNOWLEDGMENT

CBR = -88.835(e) + 61.256 (R2 = 0.9994) Thanks to Kukuh and Baurada for the help of research on
CBR = -2.2069(n) + 90.511 (R2 = 1) Correlation Between Gradation, Physical And Mechanical
Parameters For Material Embankments Reclamation.
The experimental results are described by the relationship
between γd and CBR, where the linear regression for unsaturated 6. REFERENCES
soil is γd = 0.0293(CBR) + 0.7088 and for saturated soils is
represented by γd = 0.0171(CBR) + 1.6356. The equation between Day, Robert W., (1997) "Discussion Grain Size Distribution for
these parameters is only valid for reclamation materials with a Smallest Possible Void Ratio", Journal of Geotechnical and
minimum of 80% sand composition and 20% maximum clays. Geoenvironmental Engineering, ASCE, Volume 123, No.1,
Based on the results of Proctor, CBR and volumetric & pp 78.
gravimetric tests conducted on samples type 2 with different sand Lee,P.Y. and Suedkamp, R.J., (1972) "Characteristic of Irregularly
and clay compositions are obtained correlation relationship between Shaped Compaction Curves of Soil", Highway Research
Cu/Cc and soil parameters such as dry unit weight and water content. Record No 381, National Academic of Sciences, Washington,
The results of the research obtained as follows. D.C, pp 1-9.
Correation between Cc/Cu with maximum dry unit weight (dmax) Shirur ,Naveen B and Hiremath, Santosh G.., (2014) "Establishing
and optimum water content (wopt) : Relationship between CBR Value and Physical Properties of
Soil", IOSR Journal of Mechanical and Civil Engineering,
a. dmax = 2,061 – 0,808.Cc/Cu with r = 0,927 Volume 11, Issue 5, pp 26-39.
b. wopt = 9,492 + 7,512.Cc/Cu with r = 0,874 Janjua, Zohib Shahzad and Chand, Jagdish., (2016) "Correlation of
CBR With Index Properties of Soil", International Journal of
Correlation between Cc/Cu with effective cohesion (C’) and internal Civil Engineering and Technology, Volume 7, Issue 5, pp 57-
friction angle (’) at saturated initial condition (degree of saturation, 62.
SR = 1) dan unsaturated initial condition (SR < 1) :
a. Effective Cohesion (C’)
- SR < 1 (unsaturated)  C’ = 0.087 – 0.088(Cc/Cu) with
r = -0.963
- SR = 1 (saturated)  C’ = 0.072 – 0.057(Cc/Cu) with
r = -0.920
b. Effective internal friction angle (’)
- SR < 1 (unsaturated)  ’ = 53.309 – 29.886(Cc/Cu)
with r = -0.722
- SR = 1 (saturated)  ’ = 48.761 – 33.052(Cc/Cu) with r
= -0.821
291
Characterization of leachate distributions at Ngipik Municipal Solid Waste Disposal

site
Ria Asih Aryani Soemitro1, Dwa Desa Warnana2 and Nila Sutra3
1
Department of Civil Engineering, Institut Teknologi Sepuluh Nopember, Surabaya, Indonesia
2
Department of Geophysical engineering, Institut Teknologi Sepuluh Nopember, Surabaya, Indonesia
3
Departments of Civil Engineering, Institut Teknologi Sepuluh Nopember, Surabaya, Indonesia
E-mail: ria@ce.its.ac.id
ABSTRACT: The Municipal Solid Waste Disposal in Ngipik – Gresik (near Surabaya, in East Java Province, Indonesia) has become an
overflowing landfill due to the indiscriminate dumping of solid waste at the site. Leachate migration away from the landfill boundaries
creates serious environmental problems which include, and are not limited to vegetation damage, groundwater pollution and soil pollution.
This present paper is meant to study the possible impact of the leachate percolation to the adjacent soil and groundwater. The soil samples
and leachate samples were collected from this landfill-site and its surrounding area to study the possible impact of leachate percolation on
soil characteristics. The geophysical (Electrical Resistivity Tomography – ERT) investigations were carried out on this landfill-site and its
adjacent area, to better understand the sub-soil characteristics. The water quality data were collected to study the possible leachate
penetration on the ground water. Concentration of various physicochemical parameters and engineering properties were then be determined.
The study indicated that leachate percolation quantity has strongly affected the groundwater properties. There is a general change in
groundwater properties which are attributed to the chemistry of leachate.
Keywords: leachate percolation, ERT method, groundwater pollution, Landfill.
1. INTRODUCTION
Ngipik Landfill is one of the landfills in East Java Province,
Indonesia, which still applies the open dumping method for waste
disposal (Figure 1). Although Ngipik landfill was designed for
sanitary landfill in 2002, this landfill still applies the open dumping
method until 2016. Recently, Ngipik landfill is shifting to be a
controlled landfill. The amount of waste carried and dumped in
Ngipik Landfill was around 624,72 m3 per day or 187,42 tons per
day with waste density 300 kg/m3 (Agustina, 2012).
The common problem of landfills is the leachate deployment.
Leachate means any liquid percolation through the deposited waste
and emitted from or contained within a landfill. The leachate
consists of many different organic and inorganic compounds that
may be either dissolved or suspended. They will bring potential
pollution issues for groundwater and surface waters in nature. The
landfill leachate is a secondary contamination related to landfills
(Monroe, 2001). Landfill deployment problems have been studied
by several researchers such as Suparmanto et al (2006) in Benowo
Landfill Surabaya and Endah and Pudjiastuti (1997) in Keputih Figure 1. Location map of the Ngipik landfill site
Landfill Surabaya. They found out that the disposal method was the
main cause of the groundwater contamination. Based on geological setting, the geological formation at
Sutra et al.(2016) found that water from the monitoring wells Ngipik Landfill is in the Lidah Formation (Figure 2) with Quarter
(Figure 1) has been influenced by the leachate from Ngipik Landfill. age, composed of clay, black and blackish, hard when dry, little
It means that the clay soil has also been contaminated due to the fossil content and sandstone clay. Characteristics of the Lidah
contaminant migration by the water. The increasing use and Formation are to have fairly thick clay content on the layer (Sukardi,
expanding of this site will seriously affect the quality of 1992). This can be proved by drilling results, up to 30 m indicating
groundwater in the region besides the major influence on the air that the research area is dominated by silt and clay. Detailed soil
pollution. characteristics, will be discussed further in this paper.
292
Figure 3. Locations of bore holes and resistivity measurements

The application of 2-D resistivity is aimed to describe resistivity
below the surface in 2 dimentional figures so that both vertical and
Tpl : Lidah formation Qal : Alluvium horizontal distributions of resistivity values may be determined. The
Tmpm : Madura formation Qpk : Kabuh formation field resistivity measurement used equipment’s, such as Resistivity-
QTp : Pucangan formation : study area meter, 4 rolls of connecting cables, 4 rods of electrodes (consisting
of 2 current electrodes and 2 potential electrodes), hammer, and
Figure 2. Geological map of the study area (Sukardi, 1992) some supporting equipment (GPS, Meter Gauge, Geological
Compass, Communication Tools, and Data Sheets). In this study the
method of 2-D resistivity measurement is employed using Wenner-
Generally, this paper is aimed to investigate the effect of Ngipik
Schlumberger electrode configuration. As explained by Loke
landfill site on groundwater by employing an integrated use
(1999), the Wenner-Schlumberger configuration allows a relatively
geophysical method (Electrical Resistivity Tomography – ERT), soil
good vertical and horizontal resolution and the configuration must
samples and leachates samples.
be used when we are not sure with the local geological condition.
The ERT is able to efficiently define the electrical properties of
The flow net map is a basic tool for groundwater flow dynamics
shallow layers. In environmental pollution investigations, ERT can
investigation. A flow net is a 2D diagram of equipotential and flow
survey pipeline or tank leak incidents, verify sewage leakage and
lines (Braja, 2013). The flow net map can be used to estimate the
define the scope and depth of landfill (Ayolabi et al, 2013; Awni
quantity of recharge across site feeding front boundary, and
T.B., 2005). The extensive application ERT method for mapping
contaminant load can be estimated from flow net map (Domenico
leachate movement is based on strong response of conductivity
and Schwartz, 1990). Flow net can show the topographic control of
variations for leachate distribution (E.Al-Tarazi et al.,2008).
groundwater flow (Fetter, 1988). To construct a flow net map at this
Relatively, adding of conductive or resistive contaminants into
study site, water levels are measured in 3 boreholes and 2
groundwater through soil porosity, will increase in the concentration
monitoring wells and their surfaces are interpolated between
of inorganic and/or organic constituents. Eventually, the
measuring points.
contaminants will change the electrical properties of surrounding
soil and groundwater.
2.2. Laboratory Investigation

2. FIELD AND LABORATORY INVESTIGATIONS
The soil samples from the drilling result are then taken to the
2.1. Field Investigation
laboratory for physical and mechanical properties tested. The
There were three field investigations conducted in this research, properties of soil examined are physical properties including the
Boring test and Standard Penetration Test (SPT), 2D-resistivity physical properties: volumetry & gravimetry (ASTM D 2216-71;
measurements and groundwater level testing. SPT purposed to ASTM D 854-72), Atterberg Limit (ASTM D4318-00), and grain
obtain the field density of the soil and to see the physical condition size distribution including the hydrometer (ASTM D2487; D422);
of the in situ soil. The SPT showed how many blows the landfill soil the mechanical properties: Standard Compaction Proctor (ASTM
can bear for 30 centimetres lowering (ASTM D1586). It represented D698); and soil strength properties: Unconfined Compressive
the compactness and related to the bearing capacity of the soil. The Strength (ASTM D 3080-72).
SPT was taken up to 30 meters depth in three Bore Holes (BH). BH- Finding the characteristic of soil resistivity functioning as water
1 took place next to IPAL (Instalasi Pengolahan Air Lindi/ Leachate content or leachate content in the research location requires a
Process Installation) as well as the Monitoring Well 1 (MW-1), BH- laboratory resistivity measurement over undisturbed sample of soil.
2 was taken next to the office, Monitoring Well 2 (MW-2) and the The soil samples were collected from BH-3. Samples from the depth
waste mound, and one bore hole taken a bit far from landfill, to above 2 m were used in the laboratory experiment. The resistivity of
assess the initial uncontaminated soil. BH-3 is 250 meter away from the soil samples was measured using the four-electrode cell.
the landfill. Locations of the three bore holes and lines of 2D- In an effort to study the extent of the groundwater contamination
resistivity measurement shown in Figure 3. 3 groundwater sampling was taken from boreholes. Since the
landfill site was not equipped with a leachate collector, the leachate
collected at the base of the landfill. The tests was held in two
laboratories, Laboratory of Environmental Quality, Environmental
Engineering, ITS and TAKI Laboratory in Chemical Engineering,
ITS.
3. RESULTS OF MEASUREMENTS
3.1. Soil characteristics
293
The in-site geotechnical measurements that have been conducted

include boring and measuring Standard Penetration Test (SPT) for
30 m depth. From SPT, until 5 meters soil still classified as very soft
clay, and the next depth varied from medium to stiff clay in 10 to 30
meters depth. The results of sieve analysis of soil samples resulting
from sieve drilling indicate that in general the soil at the surveyed
site consists of loam soil (SC) dominated by brown color. The
following Figure 4 show parameter of soil in the surveyed location
respectively, based on water content test and gravimetry-volumetric
test, degree of saturation (Sr), specific gravity (Gs), unit weight of
soil (t), and void ratio (e).
Figure 5. Local flow net map of unconfined aquifer (water table)

based on wells in the study area
3.3. Effect of Leachate movement through aquifer

The leachate composition affects the chemistry of groundwater, as it
percolates through the soil horizon reaching the subsurface
groundwater aquifers. Therefore, three wells (two wells from
boreholes and one well from monitoring well) have been selected to
show the effect of the leachate contamination on the quality of
groundwater.
The target of the quality of groundwater testing was through the
basic laboratory experiment to analyze and realize the characteristic
of leachates. In the laboratory analyses total suspended solids (TSS),
pH, total dissolved solids (TDS), biochemical oxygen demand
(BOD), and chemical oxygen demand (COD) of leachate were
analyzed. Also the chemical composition of leachate sample such as
Figure 4. Soil parameter from surveyed area nitrite, nitrate, ammonia, N-total and cadmium were determined.
According to the analysis experiment of selected parameters to
make a table to concentrated present the whole results (Table 1).
The liquid limit (LL) values for soil samples from BH-1, BH-2 Based on Indonesian Ministry of environmental and Forestry
and BH-3 are 77,69%, 69,36% and 86,05%, with the Plasticity Regulation No. 59 (2016) for effluent standard for leachate, the
Index (PI) 55,69%, 46,68% and 57,22%, respectively. These values threshold of physical and chemical parameters required for leachate
are chategorized as high plasicity clay. The higher plasticity index is presented in column 2 (Table 1). From comparison of
gives a lower value of hydraulic conductivity. (Daniel, 1987; measurement results of landfill leachate with values of effluent
Arifin, 2001). The void ratio (e) for BH-1, BH-2 and BH-3 are standard, and then obtained all parameters (TSS, COD, BOD and N
1.299, 1.148 and 1.635, respectively. These values correspond to total) exceeds the threshold, up to 3 – 17 times.
the density value of each BH. BH-3 has the lowest density among The pH of landfill leachate is 8,05, which is present weak
three boreholes, 1.62 tons/m3. alkaline. The high of COD and BOD in landfill are producing from
the organic matter of domestic biological waste. The relatively high
value of TDS indicated the presence of inorganic material. The high
3.2. The Flow net concentration of ammonia nitrogen (N –Total NH3) in landfill
indicates that it is still active in the decomposition of organic
Groundwater level measurements also have been conducted from
material.
same shallow wells that distributes at surveyed area. The depth of
The concentration of Cd (one of heavy metal ionic) is low. The
the groundwater in the study site is very shallow at less than 2 m. It
amount of heavy metals is related to industrial level of local urban
is obvious that the groundwater flow within the study area is
and how much industrial waste will be land-filling. The domestic
controlled by topographic (Figure 5). The direction of groundwater
waste only contains heavy metal is low. Generally, based on
flow or the flow net in the Ngipik Landfill is mainly to the North.
physical and chemical parameters of landfill leachate then leachate
generated from Ngipik landfill is the leachate of young landfill
(above 3 – 5 years) (Alloway & Ayres, 1997).
Based on Indonesian Government Regulation No. 82 (2001) for
quality standard for drinking water/clean water, the maximum
physical and chemical parameters required for clean water is
presented in column 1 (Table 1). From comparison of measurement
results of wells BH-1, MW2 and BH-3 with values of quality
standard, then obtained almost all parameters exceeds the quality
standard threshold. These high values obtained for the underground
water near the landfill is an indication of its effect on the water
quality.
TDS indicates the general nature of water quality or salinity.
The range of TDS at all sites higher than 860 mg/L. As per the
classification of Rabinove et al. (1958) based on TDS, one sample
(BH 3) was non-saline and other samples were slightly saline. The
294
high value of TDS may be due to the leaching of various pollutants worth less than 2.4 ohm m. This data will be used to interpret
into the groundwater (Suman Mor, et al, 2006). resistivity measurement results in the field
COD is a measure of oxygen equivalent to the organic matter
content of the water susceptible to oxidation by a strong chemical
oxidant and thus is an index of organic pollution. The COD level in
the groundwater samples higher than 24 mg/L, indicating the
presence of organic contaminants in the water and can be used as
organic indicators to assess the groundwater pollution caused by
landfill (Suman Mor, et al, 2006).
The N total (NH3) and Cd concentration was also found to be
remarkably high at all samples. The Cd is one of heavy metal that
undesirable metal or toxic in drinking water. Thus, it can be said that
the groundwater around the landfill could not be used for drinking
water purposes and likely indicate that groundwater quality is being
significantly affected by leachate percolation. Strictly speaking one
should avoid using groundwater drawn from the wells located in
proximity of the landfill sites.
Table 1. Physical and Chemical parameters of groundwater

No Parameters Unit BH-1 MW2 BH-3 Leachate 1 2
1 pH - 7,15 7,35 7,75 8,05 6-9 6-9 Figure 6. Curve of resistivity correlation () vs water content (wc)
2 TSS mg/L 20 202 90 280 50 100
3 TDS mg/L 3040 1380 860 3750 1000 -
4 COD mg/L 44 44 24 2480 10 300
5 BOD mg/L 25 26 14 1512 2 150 4.4. 2D-resistivity measurement
6 Nitrite (NO2) mg/L 0,921 6,241 0,135 nd 0.06 -
7 Nitrate (NO3) mg/L 8,8 6,15 0,86 1058 10 - The resistivity measurement was conducted on five lines with
8 Ammonia (NH3) mg/L 1,53 127,25 26,85 1020,44 - - varieties of seven length in the surveyed location (see Figure 3).
9 N Total (NH3) mg/L 11,25 140,35 27,84 1031,02 0,5 60
Measurement in each line was conducted in the dry season
10 Cadmium (Cd) mg/L 0,04 0,04 0,04 0,08 0,01 0.1
(August 2017).
1 Indonesian Government Regulation No.82, 2001 The result of resistivity measurement is obtained through data
2 Indonesian Ministry of Environment and Forestry No.59, 2016 processing and topographical correction using the Res2Dinv ver.
nd Not defined
BH Bore Hole 3.56 Software (Figure 7). Based on the data, the Root-Mean-
MW Monitoring Well Squared (RMS) error is less than 10% with 10 iterations.
However, the addition of iterations on the data processing does not
result in significant change against reduction of RMS errors.
3.4. Laboratory resistivity measurement Hence, the resistivity model to choose is the one on the 10th
A laboratory-scaled resistivity measurement is carried out on some iteration. Distribution of resistivity value (2-D resistivity
samples of undisturbed soil to find out the characteristics of soil tomography) measured from the surface down to the depth of 20 m
resistivity as a function of water content (fresh water and leachate). for all lines ranges from 0.1 m to 29.2 m.
The resistivity measurement on some soil samples is adjusted with In the cross section of the inversion of the line 1 to the line 7
the targeted depth of the 2D resistivity measurement in the field. there is a zone with a low resistivity value (Figure 7). Low
The soil sample used is the soil samples from BH-3 near the 6th resistivity values range from 0.1 - 1.7 Ωm. Based on the results of
2D-resistivity path. The parameters used in making the sample are laboratory scale resistivity measurements, the value is indicated as
listed in Table 2. the accumulation of leachate (which is characterized by blue).
From all measurement lines, it was found that leachate water
leakage was at a depth of 5 meters to 10 meters.
a. Line 1
Table 2. Parameters of resistivity sample preparation
γd Ww Ws wc opt Wt d t
(gr/cm3) (g) (g) (%) (g) (cm) (cm)
1.32 31.91 104.62 30.50 136.53 3.72 7.27

b. Line 2
where γd is the dry unit weight that obtained from the test proctor of
the soil sample used. The sample volume made has a diameter of
3.72 cm and a height of 7.27 cm. From the dry weight unit and the
known volume of the sample, it can be determined the weight of the
soil to be made. The soil weight, Wt consists of the sample weight
Ws and the water weight Ww. The weight of water is obtained from
the optimum water content value to the optimum dry weigh unit of c. Line 3
sample.
Soil sample resistivity () can be calculated by multiplying
resistance value (R) with geometric factor (k = 2a) where a = 1 cm.
A curve of correlation between resistivity () to water content (wc)
on each fluid (clean water and leachate) can be seen on Figure 6.
From the figure below (Figure 6), soil resistivity can be clearly
distinguished between the contaminated soil and the soil that has not d. Line 4
been contaminated with leachate. The soil contaminated leachate is
295
percolation in the subsoil. The leachate plume occurs at a maximum

depth of 10 m on the 2-D inverse models of resistivity with an
average depth of infiltration being 5 m in the study area.
It can be obviously seen that physical and chemical parameters
of groundwater are greatly and strongly affected by the leachate. All
e. Line 5 groundwater physical and chemical parameters exceed the standard
value of Indonesian Ministry of Environment and Forestry No. 59 ,
2016.
Furthermore, a multidisciplinary technique of
geotechnics, hydrogeological, geochemical and geophysical
techniques should be adopted on the dumpsite to be able to
holistically assessed and interpreted.
f. Line 6
6. ACKNOWLEDGEMENTS
The authors would like to thank PMDSU Program from Indonesian
Ministry of Research and Higher Education for the scholarship and
research grant.
g. Line 7
7. REFERENCES
Agustina, Ummy Arofah. (2013). Effect of Leachate Recirculation
on Solid Waste Degradation Rate In Ngipik Landfill Gresik. ITS
Undergraduate Thesis of Environment Engineering, RSL 628.42
Umm p.
Alloway, B.J; Ayres, D.C. (1997) Chemical principles of
environmental pollution (Second edition). Waste and other
multipollutant situation. P 357. Blackie A &P. ISBN
0751403806.
Figure 7. The 2D resistivity tomography on line 1 – line 7 Ayolabi, E.A., Folorunso, A.F., Idem, S.S. (2013) “Application of
electrical resistivity tomography in mapping subsurface
Furthermore, from the inversion 2D resistivity results, 3D cross-
hydrocarbon contamination”, Earth Sci. Res. 2 (1), 93-104.
section is made by displaying only the target resistivity value which
Awni T. Batayneh. (2005) “2D Electrical Imaging of an LNAPL
is a low anomaly that is the presence of considerable water content
Contamination, Al Amiriyya Fuel Station, Jordan”, Journal of
in the soil (leachate water flow). The 3D result (as illustrated in
Applied Sciences, 5: 52-59.
figure. 8) combined with the direction of the aquifer stream (figure
Braja M, (2013) “Principles of geotechnical engineering”, 7th edn.
3), shows that the leachate plume follows the aquifer stream.
Domenico PA, Schwartz FW, (1990) “Physical and Chemical
This indicates that the leachate distribution pattern that occurs
Hydrogeology”, Wiley, New York.
from the overloaded area of eastern garbage, the leachate spreads to
E. Al-Tarazi, J. Abu Rajab, A. Al-Naqa, M. El-Waheidi. (2008)
the lower groundwater level that is towards the Southeast -
“Detecting leachate plumes and groundwater pollution at
Northwest.
Ruseifa municipal landfill utilizing VLF-EM method”, Journal
of Applied Geophysics, 65. 121-131.
Endah, N. and L. Pudjiastuti. (1995) “The effect of Leachate on
Surabaya MSW in Sukolilo Landfill to Water Quality of
Resident’s Wells Around”, Research Institution, Institut
Teknologi Sepuluh Nopember.
Fetter CW. (1988) “Applied hydrogeology”, Nerrill Pub Co., A well
and Howell information Co., Columbia, p 529
Loke, M.H,. (1999) “Electrical imaging surveys for environmental
and engineering studies”: A practical guide to 2-D and 3-D
surveys.
Monroe, Michele. (2001) “Landfill Leachate Treatment: VSEP
Offers a Revolutionary Solution Introduction”, New logic
Research, Inc. Company News.
Rabinove C. J., Long Ford, R. H. and BrookHart, J. W. (1958)
‘Saline water resource of North Dakota U.S.’, Geol. Sur. Water
Supply Paper 1428, 72.
Suman Mor, Khaiwal Ravindra, R. P. Dahiya, A. Chandra. (2006)
“Leachate Characterization and Assessment of Groundwater
Pollution Near Municipal Solid Waste Landfill Site”,
Environmental Monitoring and Assessment (2006) 118: 435–
Figure 8. The 3D perspective view of leachate plumes 456.
Sukardi. (1992) “Peta Geologi Lembar Surabaya & Sapulu, Jawa”.
Pusat Penelitian dan Pengembangan Geologi, Bandung.
4. DISCUSSION AND CONCLUSION Sutra, N., Soemitro, R.A.A., Warnana, D.D. (2016) “Preliminary
Assessment to Contaminated Soil Characteristics on Ngipik
From the groundwater testing and resistivity measurement, it is Open Dumping Waste In Gresik Residence”, Proceedings of 2nd
clearly seen that the leachate generated from the landfill site is International Conference of Problematic Soil, Bandung,
affecting the groundwater quality in the adjacent areas through Indonesia.
296
Evaluation of Remedial Works for a Spillway on Landslide-dammed Lakes by

an Earthquake, a Case Study in the Jiufengershan Landslide
I-Hui Chen1, Shei-Chen Ho 1, Yu-Shu Lin 1, Jun-Yang Chen 1, Miau-Bin Su 1
1
Department of Civil Engineering, National Chung Hsing University, Taichung City 402, Taiwan
E-mail: ian.cih82@gmail.com
ABSTRACT: This paper applied digital elevation models of different periods and long-term field monitoring data to compare elevation
changes of a spillway for remedial works of landslide-dammed lakes in the Jiufengershan landslide. Digital aerial photogrammetry and
LiDAR techniques were used to rebuild orthophotos and digital elevation models in the area from 1998 to 2015 in order to review the
changes of deposit and erosion in topographies and profiles of the spillway. Through many torrential rainfalls in the past, it was particularly
effective for related spillway remedial works to suppress the production and flow of collapse debris from the bed and bank of the spillway
channel. It is the most important for spillway remedial works to remain the stabilization of two dammed lakes.
Keywords: Landslide-dammed lake, spillway, remedial works, digital elevation model.
1. INTRODUCTION 2. JIUFENGERSHAN LANDSLIDE

This paper employed digital elevation models (DEMs) of different A case study is located at the Jiufengershan landslide in Taiwan by
periods and long-term field monitoring data to compare elevation the Chichi earthquake in September 1999, as shown in Figure 1. The
changes of a spillway and water-level variations of landslide- area of the landslide is 195 ha and the amount of the collapse is 35
dammed lakes in the Jiufengershan landslide by the 1999 Chichi million cubic meters that blocked two creeks, namely Jiutsaihu and
earthquake. Schuster (1986) argued that many landslide-dammed Sezaikeng streams, and caused two dammed lakes. The area of the
lakes were often transitory to keep their integrity because the flow Jiutsaihu and Sezaikeng dammed lakes is 4.4 ha and 6.4 ha,
of water in a dammed lake could easily spill over the top of a respectively. Since 2003, there has been a monitoring landslide
landslide dam in which unconsolidated debris may rapidly be eroded project in the Jiufengershan landslide with some monitoring
and cut; then, it could cause the dam to break down in a few hours equipment, including extensometers, inclinometers, groundwater
so that occurred a catastrophic flood disaster. The author indicated lever gauges, etc.
91% of landslide dams failed within 1 year of formation based on 63
cases from literature and past experience. Thus, it is necessary for
the safety of residents in the downstream to remediate the dammed
lakes against failure disasters.
For emergency treatments of landslide-dammed lakes, excavated
spillway is a fast and simple method which is used for the overflow
discharge of a dammed-lake and the mitigation of potential disasters
caused by the breakage of landslide dams (Schuster, 1986; Li et al.
1986; Chen, et al., 2011). Other methods include drainage by intake
structure, siphon pipes, pump systems, tunnel excavation and
diversions (Sager and Chambers 1986; Code and Sirhindi, 1986;
Hansen and Morgan, 1986).
The case study of the research is in the Jiufengershan landslide
by the Chichi earthquake in September 1999 in Taiwan, which
caused two landslide-dammed lakes. There are two main emergency
measures for the landslide and two dammed lakes. Firstly, one of
emergency mitigative measures was an innovative method, namely
temporary geotextile dams of debris-filled shipping containers, in
the end part of deposit area in order to increase strength of the slope Figure 1 Geographic Map of the Jiufengershan Landslide
stable and avoid headward erosion. The other emergency measure
was an excavated spillway of dammed lakes that was an emergency The geology of the landslide is underlain mainly by Miocene
° °
drainage channel so as to discharge the overflow of dammed lakes sedimentary rocks where the strike is N36 E and the dip is 21 SE
and decrease the storage volume of two lakes. After emergency with an N-S trending synclinal axis of Daanshan syncline in the
mitigative measures for dammed lakes, Remedial works of a eastern part; the geologic formation of the syncline axis is the
permanent spillway needed to be undertaken so as to retain the two Kueichulin Formation which unconformably contacted with the
dammed lakes. There were a slit-type sabo dam, several columnar- underlain Changhukeng Shale; the lower formations include
type submerged dams, groundsill works and some check dams in the Miocene strata of Shihmen Formation (Shou & Wang, 2003), as
spillway from 2000 to 2008. shown in Figure 2.
For the performance evaluation of the spillway, digital aerial With regards of two dammed lakes, the surface elevation of the
photogrammetry and LiDAR techniques were used to rebuild Jiutsaihu and Sezaikeng lake was 585 m and 577.5, respectively, as
orthophotos and DEMs in the area from 1998 to 2015 in order to the earthquake occurred. The detailed data of two lakes is shown in
review the changes of deposit and erosion by profiles of the spillway Table 1. Furthermore, the greatest depth of the Jiutsaihu and
and to evaluate lake-surface variations of dammed lakes by long- Sezaikeng lake was estimated about 29 m and 37.5 m, respectively,
term monitoring data. before completed spillway.
297
Figure 3 Relationship between rainfall and elevation of lake level in

the Jiutsaihu dammed lake
Figure 2 3D Geological Structure in the Jiufengershan landslide Figure 4 Relationship between rainfall and elevation of lake level in
the Sezaikeng dammed lake
Table 1 Relative elevations, depths and volumes of two landslide-
dammed lakes 3. REMEDIAL WORKS OF THE LANDSLIDE AND
Jiutsaihu Sezaikeng DAMMED LAKES
dammed lake dammed lake
3.1 Emergency treatment
Area (ha) 4.43 6.48
Dammed lakes were vulnerable to catastrophic failure by
Before completed spillway overtopping and breach, landslide emergency treatments involved in
the potential failure of these dams need to be carried out for
Elevation (m) 585 577.5 protection of relative people’s living and property. Thus, there were
two items of remedial works for the landslide and two dammed
lakes.
Max lake depth (m) 29 37.5 Firstly, one of emergency mitigative measures was an
innovative method, namely temporary geotextile dams of debris-
Storage volume (m3) 678,000 1,089,700 filled shipping containers, in the end part of accumulation area in
order to increase strength of slope stable and avoid headward
After completed spillway erosion. The measure of debris-filled shipping containers was the
first for the slope toe loading of a landslide in Taiwan whose height,
width and height were 122.7 m, 8.2 m and 7.5m, respectively, and
Elevation (m) 575.89 570.05
the aims of the measure were for four functions as marked in Figure
5(1) (they are: ①slope stable, ②slope toe loading, ③headward
Max lake depth (m) 15 29 erosion control, ④ protection of stream bank).
Max lake-level rise as a

1.2 4
heavy rainfall (m)
For long-time monitoring the lake surfaces of the Jiutsaihu and

Sezaikeng lakes in Figure 3 and Figure 4, the elevations of lake
surfaces dropped to 575.89 m and 570.05 m, respectively, after
completed spillway. From rainfall and lake-surface monitoring data
of typhoon events in September 2015, the highest water levels of the
Jiutsaihu and Sezaikeng lakes was 577.1 m and 574 m which went
up approximately 1.2 m and 4 m, respectively. As a result, it is
successful to control the water surfaces of two dammed lakes while
the spillway completed. Figure 5 Locations and photos of emergency treatments in the
landslide
298
The other emergency measure is a spillway of dammed lakes 4.1. Landform changes with photogrammetry and LiDAR
that is an emergency drainage channel with 1450 m in length and 8 survey
m in depth in order to discharge the overtopping of dammed lakes
and decease the storage volume of two lakes in October 1999, as Digital aerial photogrammetry and analysis tools of ERDAS
shown in Figure 5(2). IMAGINE were used to rebuild orthophotos and DEMs for different
periods before and after the 1999 Chichi earthquake. The result is
shown in Figure 7. According to the topographic analysis (see Table
3.2 Soil and water conservation treatment
2), the average depth of the landslide is 34 m with a collapse volume
After emergency mitigative measures of dammed lakes, it is of 34.92 million cubic meters and a deposit volume of 36.58 million
important for remedial works of a permanent spillway so as to avoid cubic meters.
severely incised channel, losing a lot of debris and dammed-lake
failure. One of the remedial works was a slit-type sabo dam that was
located at 1K+230 m from the overflow outlet of the spillway in
2000. The aim of the dam was for a direct barrier to debris flows
which could occur. The location is shown in Figure 6.
Figure 7 Landform change between 1998 and 2002 (collapse is in

brown color; deposit is in blue color)
Table 2 Relative areas and volumes of collapse and deposit in the

landslide
Maximum
Debris volume
Type Area (ha) elevation
(m3)
change (m)
Collapse 102.5 34,923,400 -85
Deposit 92.5 36,585,000 +108
Sum 195
Figure 6 Location of spillway works in the landslide
There was LiDAR data to make DEM for this area after 2004.
The slit-type sabo dam divided two constructions including a The data of DEMs in 2004, 2011 and 2015 for the landslide can be
main dam and auxiliary dam, of which the lengths and heights were compared with ground surface changes in three dimensions and help
131 m and 15.5 m, and 84 m and 8 m, respectively. The sabo dam to understand area and volume changes.
that categorized as a permeable structure is one of popular methods The data of DEMs was built in 2004, 2011 and 2015. The
currently used for debris flow prevention. The dam can effectively results of elevation changes illustrate the differences in elevations
block huge boulder as well as debris from upstream while allow from 2004 to 2015 in the landslide, as shown in Figure 8(1). The
smaller-grain debris and stream flow to be released into downstream figure shows that erosion area is in blue color and deposit area is in
channel. red color between 2004 and 2015 in which the landslide volume is
The other remedial works were 20 columnar-type submerged estimated about 267,000m3 with a maximum erosion depth of 27 m
dams on the spillway so as to increase the stabilization of channel and maximum deposit depth of 24 m. However, the elevation
bed and decrease fast headward erosion of the spillway in 2003 (see variation was smaller from 2011 to 2015 in Figure 8(2).
Figure 6). The size of a single submerged dam is 42 m in length and
5 in height. The column diameter is 1 m and the spacing in 3 rows is
all 1m. In contrast with traditional dams, this dam was more
permeable from its bottom to all other parts which was suitable for
the special geology in the Jiufengershan landslide, so 20 columnar-
type submerged dams were immediately designed and implemented
by a local government. After that, remedial works of check dams
was performed to enhance the bed stability of the spillway from
2008 to 2013, as can be seen in Figure 6.
4. PERFORMANCE EVALUATION OF SPILLWAY Figure 8 (1) Landslide change between 2004 and 2015; (2)
WORKS Landslide change between 2011 and 2015. (Erosion area is in green
color and deposit area is in orange color.)
The landslide blocked two streams which caused two landslide-
dammed lakes so a spillway was built to discharge water volume of 4.2 Profile changes of the spillway with photogrammetry and
the two lakes. Thus, it is important to evaluate remedial works of the LiDAR data
spillway. The paper applied DEMs for different periods and long-
term monitoring data to compare elevation changes of the spillway DEMs of the area were built by the data of photogrammetry in 1998,
and water surfaces of dammed lakes. Digital aerial photogrammetry 2002 and 2005. Then, profiles of the Jiutsaihu stream before and
and Light Detection and Ranging (LiDAR) techniques were used to after the 1999 Chichi earthquake were drawn in Figure 9 by using
rebuild orthophotos and DEMs in the area from 1998 to 2015 in the data of DEMs. The graph of profiles in the figure shows
order to review the change of accumulation and erosion in profiles elevation changes of the stream bed before and after the earthquake
of the spillway. in which the maximum landslide depth is 87 m at 2K+250 of the
profile line. It is clear for the profiles that there was an erosion
299
phenomenon at 3K+500 of the upstream and a deposit area occurred Through some heavy rainfall events from 2004 to 2015, the highest
from 0K+775 to 1K+365 at the downstream that was possibly lake surface went up approximately 1~4 meters, which means that it
caused by excavated spillway or collapsed debris of the upstream. is successful for the spillway to control the changes of lake surfaces
in two dammed lakes. According to Schuster (1986), 91% of 63
global barrier lakes broke down within a year after having formed.
As a result, it is valuable for the spillway in the landslide that two
dammed lakes can be maintained and there was no severe disaster in
the past.
Compared to elevation changes of the spillway by DEMs of
different periods with photogrammetry and LiDAR techniques, there
are some results of the analysis. Firstly, the average depth of the
landslide is 34 m with a collapse volume of 34.92 million cubic
meters and a deposit volume of 36.58 million cubic meters while the
Chichi earthquake occurred. Secondly, the data of DEMs in 2004,
2011 and 2015 shows that landform changes in the landslide were
nonstop in which the landslide volume is estimated about 267,000
m3 with a maximum erosion depth of 27 m and maximum deposit
depth of 24 m. Then, there was a significant downcutting in the
elevation change by 20 m between 0K+870m and 1K+80m in the
spillway from 2004 to 2011, which caused some submerged dams
damaged but the changes gradually decreased from 2011 to 2015
after some check dams were completed from 2008. Finally, the most
Figure 9 Location and profile of the Jiutsaihu stream before and
important performance is to remain the stabilization of two dammed
after the Chichi earthquake
lakes after emergency treatment and a series of spillway remedial
works.
Other DEMs of the area were built by the data of LiDAR in
2004, 2011 and 2015 with the grid precise of 5 m. Then, profiles of
6. REFERENCES
the spillway were drawn by using the data of DEMs in the three
periods, as shown in Figure 10. Meanwhile, the locations of Schuster, R. L. (1986). Landslide dams: processes, risk, and
remedial works, including a sabo dam, 20 columnar-type submerged mitigation. ASCE.
dams and check dams, are plotted in the figure. Shou, K. J., & Wang, C. F. (2003). Analysis of the Chiufengershan
landslide triggered by the 1999 Chi-Chi earthquake in
Taiwan. Engineering Geology, 68(3), 237-250.
Chen, X. Q., Cui, P., Li, Y., & Zhao, W. Y. (2011). Emergency
response to the Tangjiashan landslide-dammed lake resulting
from the 2008 Wenchuan Earthquake, China. Landslides,
8(1), 91-98.
Delaney, K. B., & Evans, S. G. (2017). The evolution (2010–2015)
and engineering mitigation of a rockslide-dammed lake
(Hunza River, Pakistan); characterisation by analytical
remote sensing. Engineering Geology, 220, 65-75.
Xu, Q., Fan, X. M., Huang, R. Q., & Van Westen, C. (2009).
Landslide dams triggered by the Wenchuan Earthquake,
Sichuan Province, south west China. Bulletin of engineering
geology and the environment, 68(3), 373-386.
Cedergren, H., 1989 “Seepage, drainage & flow nets”, McGraw
Figure 10 Elevation changes of the spillway for different periods Hill.
Li, T., Schuster, R. L., & Wu, J. (1986). Landslide dams in south-
Figure 10 illustrates that there was a significant downcutting in central China. In Landslide dams: processes, risk, and
the elevation change by 20 meters between 0K+870 m and 1K+80 m mitigation (pp. 146-162). ASCE.
in the spillway from 2004 to 2011, which caused falling and damage Sager, J. W., & Chambers, D. R. (1986). Design and construction of
of some columnar-type submerged dams. Nevertheless, the the Spirit Lake outlet tunnel, Mount St. Helens, Washington.
elevation changes gradually remained stable from 2011 to 2015 after In Landslide Dams: Processes, Risk, and Mitigation (pp. 42-
remedial check dams and groundsills from 2008. The other point is 58). ASCE.
that the sabo dam effectively blocked the massive debris against its Code, J. A., & Sirhindi, S. (1986). Engineering implications of
flowing to downstream. All spillway works could have a capacity of impoundment of the Indus River by an earthquake-induced
debris pool by 6 million cubic meters. Dammed-lake failure and landslide. In Landslide Dams: Processes, Risk, and
debris flow did not occur in the landslide area while all spillway Mitigation (pp. 97-110). ASCE.
works completed. Cameron, C. P., Cato, K. D., McAneny, C. C., & May, J. H. (1986).
Overall, through many torrential rainfalls and typhoons from Geotechnical Aspects of Rock Erosion in Emergency
2004 to 2015, related remedial works which have cost about 12 Spillway Channels. Department of the Army, Waterways
million USD were particularly effective suppress the production and Experiment Station, Corps of Engineers, Geotechnical
flow of collapse debris in the bed and bank of the spillway channel. Laboratory.
The most important performance is to remain the stabilization of Hansen, D. C., & Morgan, R. L. (1986). Control of Thistle Lake,
two dammed lakes. Utah. In Landslide Dams: Processes, Risk, and Mitigation
(pp. 84-96). ASCE.
5. CONCLUSION
From the lake-surface monitoring data of two dammed lakes, an
average drop of the lake surface in the Jiutsaihu and Sezaikeng
dammed lakes was 8.28 m after completed spillway works in 2000.
300
A Comparison Between VS30 Based- and Natural Frequency Based-

Site Amplification Factor for Three Different Types of Soil Classification
B.A. Yogatama1 and B.A. Tirta1
PT. Witteveen+Bos Indonesia, Jakarta, Indonesia
1
E-mail: bonifacius.yogatama@witteveenbos.com, budiwan.tirta@witteveenbos.com
ABSTRACT: In this paper, a comparison is shown between site factors derived from the conventional VS30 parameter based on SNI 1726-
2012 and from the soil profile natural frequency (f0) referred to Cadet et al (2012). Three series of stochastic site response analyses (SRA) as
a benchmark for the former two approaches have been performed considering linear viscoelastic condition from three large data sets. The
data sets were stochastically generated to represent different site classification uniformly: soft, medium, and hard soil (SE, SD, and SC soil
class, respectively). For each stochastic model set, 1000 soil profiles were generated and its linear responses were computed to derive the
spectral ratio at 1 Hz (FV – SA 1 s), 5 Hz (FA – SA 0.2 s), and 25 Hz (PGA). Correlation scatter charts were then developed to observe and
to check the performance of each approach. The charts showed correlation of site factors based on the VS30 and f0 parameters with the SRA
modeling. Based on the correlation, the best site amplification factor estimation approach for each spectral acceleration (FPGA, FA, FV) and
for each soil class (SE, SD, SC) is provided.
Keywords: site factor, site amplification, local site effect, natural frequency, VS30
1. INTRODUCTION developed by Cadet et al (2012) (herein denoted as SAPE: Site

Amplification Prediction Equation).
In a seismological point of view, earthquake wave propagation can The comparison is performed for different range of soil profile,
be classified into three different phases: (1) earthquake rupture or covering from very soft sediment (low frequency, low V S30) up to
source effect, (2) wave propagation through crustal rock medium, very hard rock / outcropping rock (high frequency, high V S30). In
and (3) local site effect. All above three aspects may introduce either order to do that, a numerical method is implemented to create
amplification or de-amplification over a certain frequency range. thousands of stochastic artificial soil profile, which then the natural
Although its depth scale is the smallest compared to the other two frequency and the VS30 of each profile will be calculated.
(easier to be characterized by an engineer), the local site effect Afterwards, a linear site response is performed for each soil profile
contributes a large uncertainty to produce the recorded ground to derive the computed site factor for three different spectral
motion at surface. Local site effect has many forms, the most acceleration levels: FPGA, FA (0.2 s), and FV (1.0 s). The annotation
common is the wave amplification at soft sediment overlying a rigid of these site factors is made similar to those parameters used in
bedrock, which is the main topic of this paper. NEHRP provision and SNI 1726-2012.
The consideration of local site effect in a modern seismic code The goal of this paper is to compare the computed site factors
has been implemented after the occurrence of Loma Prieta with the estimated site factors from the two methods (SAPE and
earthquake, California in 1989. Borcherdt (1994) collected many SNI) and to provide an explanation of the possible discrepancies
accelerometer recordings from different stations of the event in between them. The validity of this linear site response analysis is
California region, studied the relation between site characteristics considered sufficient for this purpose, since SAPE was also
and the corresponding amplification, and proposed a correlation developed only for linear elastic condition, and the output variability
between VS30 (the average shear velocity of the top 30 m) and the due to non-linear behavior as described by Yogatama and Lengkeek
amplification factor for PGA, short period and long period spectral (2017) is not expected.
acceleration (0.2 s and 1.0 s). Such correlation became popular since
VS30 is a parameter which can be obtained with less effort. This
2. ANALYSIS PROCEDURE
study is then implemented in many seismic codes around the world,
starting from NEHRP provision 1997 (implemented in UBC 1997) 2.1 Frequency-based site factor (SAPE approach)
and eventually adopted in the latest Indonesian seismic code SNI
1726-2012. In their publication, Cadet et al (2012) developed a different set of
Despite its high popularity, scientific community put a high site factor of ground motion at a certain frequency based on the
doubt on the use of such simple parameter to characterize the natural frequency of the soil profile from KiK-NET ground motion
amplification of a soil profile, which contains a lot of uncertainty database. Thousands of weak motion surface and downhole
along the whole depth. VS30 site factor fails to take into account the accelerometer recordings from approximately 400 sites were used to
effect of soil layer deeper than 30 m, e.g. for region with a thick derive the amplification factor (spectral ratio) for each site. Before
alluvial deposit (for example Jakarta). Castellaro et al (2008) the downhole recording was used as the reference motion to derive
criticized the way Borcherdt (1994) draw the regression line of such the spectral ratio, the signal was corrected against depth and
correlation, and showed that actually there is very little correlation normalized against impedance contrast in order to uniformly
between VS30 and site factor. Cadet et al (2012) used a lot of KiK- represent a signal recorded at a generic outcropping rock station
NET data from Japan to develop a relation between site with a shear wave velocity of 800 m/s. This procedure is necessary
amplification factor at a certain frequency with the natural to be performed since KiK-NET downhole sensors are positioned at
frequency (f0) of a soil profile. Their study suggested that f0 is a varying bedrock depth and velocity (different for each site) and such
better site proxy than VS30 as it has smaller misfit value at all issue will create a bias if not treated properly.
frequency range. Zhao and Xu (2013) also studied the same thing After the spectral ratio has been derived, then all spectral ratios
but using a different set of KiK-NET data, and they stated that were normalized against the natural frequency of each site so that
natural period (T) of a soil profile is a better proxy than V S30 for the spectrum peak is positioned at unity, with the x-axis as a
sites with high natural period (very soft sediment or very thick dimensionless frequency. The dimensionless frequency itself is the
alluvial deposit). ratio between the considered frequency of spectral ratio value and
In this paper, a performance comparison is made between the the natural frequency of the respective amplification spectrum. A
site factors derived from VS30 (herein denoted as SNI approach) with regression analysis is then made to correlate the amplification factor
the site factors derived from the soil profile natural frequency (f0) as (spectral ratio) and the site proxy (soil natural frequency) at different
301
value of dimensionless frequency. The equation of their SAPE  Model 3 (soil class SC)  High range site frequency:
method is given by the following formula: approximately f0 > 5 Hz
The purpose of this division is to have a clear verification on the
reliability of FA (0.2 s – 5 Hz) and FV (1 s – 1 Hz) for sites with
(1)
broadband f0 and VS30.
The artificial soil profiles are modeled by assuming three-
where Gx is the estimated site amplification factor, X is the soil layered profile overlying a halfspace (bedrock, soil class SB). The
natural frequency, Xref and β are the coefficients derived from linear thickness and the velocity of each layer in each model set are
regression at a considered dimensionless frequency, v. The determined stochastically using truncated Gaussian distribution with
coefficient table of Xref and β is given as an appendix in their the parameter set as presented in Table 2.
publication.
Table 2 Statistical parameters for developing artificial soil profiles
2.2 VS30-based site factor (SNI approach)
VS (m/s) thickness (m)
In this paper, the other site proxy for comparison is the 30 meters mean std mean std
averaged shear wave velocity (VS30) which is determined by: layer 1 100 10 10 2
model 1;
layer 2 120 30 10 4
class SE
(2) layer 3 150 30 10 5
layer 1 160 20 10 2
model 2;
layer 2 350 40 10 4
where di is the layer thickness up to 30 meters depth and VSi is class SD
layer 3 500 60 10 5
the shear wave velocity of layer i (in m/s). The calculated VS30 then layer 1 550 20 10 2
corresponds to five soil classifications (Table 1), each has its own model 3;
layer 2 650 40 10 4
site amplification factor which was determined based on Borcherdt class SC
layer 3 700 50 10 5
(1994). The linear site amplification factors as applied in SNI 1726-
2012 are represented in the following table. Aside from layer thickness and shear velocity, unit masses and
material damping are also required for linear viscoelastic analysis.
Table 1 Site classification and its corresponding linear site In this study the unit masses and material damping of each layer is
amplification factor (SNI 1726-2012) determined by using an approach proposed by Cadet et al (2012).
Soil VS30 (m/s) FPGA FA FV The unit masses of each layer are assumed to be increasing with
Class shear wave velocity, as follow:
(PGA<0.1g) (SS<0.25g) (S1<0.1g)  ρ = 1.7 g/cm3 for VS < 180 m/s
SA > 1500 0.8 0.8 0.8  ρ = 2.0 g/cm3 for 180 < VS < 360 m/s
 ρ = 2.2 g/cm3 for 360 < VS < 1500 m/s
SB 750 – 1500 1.0 1.0 1.0  ρ = 2.5 g/cm3 for VS > 1500 m/s
The damping ratio of each layer is represented as quality factor
SC 350 – 750 1.2 1.2 1.7
(Q) and is derived using the following formula, with VS is the shear
SD 175 – 350 1.6 1.6 2.4 velocity in m/s.
SE < 175 2.5 2.5 3.5 Q = VS/10
2.3 Stochastic analysis Based on the statistical parameter set and assumptions as
described above, 1000 artificial soil profiles are created for each
The simulation framework of this study is to divide the stochastic model set, and the histogram of the site frequency (f0) and VS30 are
model of soil profile into three different VS30 ranges: presented in the figure below. VS30 is determined using Equation (2),
 Model 1 (soil class SE)  Low range site frequency: while f0 is determined from the transfer function of each profile.
approximately f0 < 1 Hz
 Model 2 (soil class SD)  Medium range site frequency:
approximately 1 Hz < f0 < 5 Hz
302
Figure 1 Histogram of natural frequency and VS30 for all model sets
The above histograms show that not all generated soil profiles regression coefficient tables as in Cadet et al (2012) is used for
comply within the desired frequency/VS30 range, for instance not SAPE approach.
all profiles of model 1 have f0 lower than 1 Hz, as there is still about Linear viscoelastic site response analysis is then performed for
50% which has larger than 1 Hz. However, large portion of the each generated soil profile using a recorded outcropping rock
profiles have VS30 value lower than 175 m/s which put them to soil ground motion during Kobe (1995) earthquake, scaled down to PGA
class SE. Most of soil profiles in model 2 lie within the desired of 0.05g to ensure a linear condition. The response spectrum of the
range, for both f0 and VS30. Similar to model 1, part of soil profiles calculated motion at ground surface is calculated, and compared
in model 3 have f0 lower than 5 Hz while the rest has f0 exceeding with the response spectrum of the input motion at three specific
beyond 5-10 Hz. However, most of the profiles have VS30 between frequencies, correspond to the site factor value of FV, FA, and
350-750 m/s (soil class SC). It is not clearly depicted in the figure FPGA. The calculated site factors from site response analysis are
but there is one soil profile in model 3 which lies in soil class SB then compared with the estimated site factors from both SAPE and
classification and two soil profiles lie in soil class SD. SNI approaches. The analysis procedure in this study is better
After VS30 and f0 have been determined for each generated soil illustrated by the following flowchart as below.
profile, the next step is to determine the corresponding site
amplification factor (FPGA, FA, FV) based on the two approaches: 3. RESULTS
SNI (using VS30) and SAPE (using f0). Table 1 is used for SNI
approach, only considering the site factor for the smallest spectral The correlation charts between both estimation approaches and the
acceleration (linear condition). Equation (1) along with several site response result are given in Figure 3 and Figure 4. The
observation result from the charts is summarized in this section.
Figure 2 Flowchart of analysis procedure
303
Figure 3 SAPE approach correlation scatter chart for all soil classes
Figure 4 SNI approach correlation scatter chart for all soil classes
304
3.1 Soil class SE For SNI approach, FPGA has a good correlation while FA
underestimate the site response factors. FV-SNI does not give a
With the exception of FV-SAPE, it is clear from the figures above good result for this soil class since the estimated site factors over-
that both approaches do not give a satisfactory correlation with site predict the value if compared with site response result.
factors from site response analysis as the scatters are deviating from
the correlation line. FPGA-SAPE tends to have a narrow value while
4. DISCUSSION
the value distribution from the site response is broad. F A-SAPE has
a broader distribution compared to FPGA, but the deviation is still Figure 5 below shows an illustration of how SAPE was derived by
high. One interesting observation is for FV-SAPE, as the trend and Cadet et al (2012). As explained above, hundreds of spectral ratio
the distribution of the scatter is good for profiles with site frequency from KIK-Net database was collected and the frequency of each
(f0) higher than 1.0 Hz (FV is in the range of 2.2 - 3.9). For profiles curve is normalized with the natural frequency of the corresponding
with f0 lower than 1.0 Hz, the scatter becomes worse and deviating soil profile, leading to a set of “bell-shaped” curves with maximum
from the correlation line (FV-SAPE range is narrow, from 3.5-4.0 amplitude located at abscissa of unity. Then, a correlation is made
while FV-site response has very broad distribution). for natural frequency and its amplification at a specific
FPGA, FA, FV determined from SNI all show bad correlation with dimensionless frequency value.
site response site factor. Having that in mind and by observing Figure 5, by logic it is
safe to say that the correlation produced for dimensionless
3.2 Soil class SD frequency less than one (left side of black dashed line) will be better
than for dimensionless frequency beyond unity (right side of dashed
For this class, FPGA-SAPE has a bad correlation similar to soil class line). The left side correlation will be better because each curve is
SE. SAPE gives a significantly narrow distribution of FPGA values always asymptotic at the beginning and eventually reach its maxima
while site response gives a broader distribution. FA-SAPE has a at the dashed line, thus the scatter distribution is more regular. On
good correlation for profiles with f0 larger than 5 Hz, while the the contrary, for dimensionless frequency larger than one, the scatter
distribution becomes worse for profiles with f0 lower than 5 Hz. FV- distribution is more irregular due to the presence of amplification at
SAPE in general has a satisfactory correlation, although the scatter higher modes. This will have implication on the quality of the
slightly deviates from the correlation line, but the trend is good and correlation between amplification value and natural frequency.
correlates well with site response FV. It can be observed that all soil The explanation above explains Figure 3 on how SAPE
profiles in this model set have f0 larger than 1 Hz. performs at a certain frequency (FA / FV / FPGA) in each soil class.
Site factors from SNI approach still have bad correlations with For soil class SE, SAPE gives good result for FV only, for sites with
site response site factors, however the distribution of site response f0 larger than 1 Hz. Recall that FV is a site factor at 1 Hz frequency
site factors are now become narrower if compared with the (period 1 sec) and dimensionless frequency is a ratio between the
distribution from soil class SE. considered frequency of spectral ratio value and the natural
frequency of a soil profile (f0). Therefore, the calculated
3.3 Soil class SC dimensionless frequency for sites with f0 larger than 1 Hz is less
FPGA-SAPE has a good correlation (values varying between 1.0 - than one, thus lies on the left side of the dashed line in Figure 5.
1.5). A bad correlation is observed for FA-SAPE, since the Meanwhile, sites with f0 lower than 1 Hz will have dimensionless
distribution of FA by the approach is very broad ranging from 1.0 - frequency value which is larger than one, thus plotted in the right
3.0 while the site response factor have values ranging from 1.0 - 2.0. side of the dashed line and it explains the bad correlation observed
Similar to FA, FV-SAPE also seems to overestimate the site for these soil profiles.
response result.
Figure 5 Illustration of normalized transfer function
For soil class SD, SAPE gives good result for FA (frequency 5 performance might be due to the difficulty in obtaining a good
Hz; period 0.2 sec) for profiles with f0 larger than 5 Hz. SAPE also transfer function with a clear amplification by using field
performs well for FV (frequency 1 Hz; period 1 sec) since generally instrumentation at rocky sites with high VS30 (soil class SC, SB).
all soil profiles in this model set has f0 larger than 1 Hz. For FPGA It is confirmed that for “hard soil” sites, a good quality
(frequency 25 Hz, period 0.04 sec), SAPE leads to bad correlation amplification spectrum is really difficult to get either by using
since at this frequency (25 Hz), all soil profiles have dimensionless geophysical tool (H/V microtremor) or by deriving the spectrum
frequency value greater than one. from surface/borehole accelerometer since the impedance contrast
By using the same logic as described above, for soil class SC, between the soil profile and the reference rock site is relatively
SAPE should give good result for FA and FV and bad result for weak (Pierre-Yves Bard - personal communication). However,
FPGA since almost all of the soil profiles have f0 larger than 1 Hz despite the bad data scatter of FA-SAPE for this soil class, in fact
and 5 Hz, but lower than 25 Hz. However the result is the contrary; the FA correlation is satisfactory enough for sites with f0 larger than
FPGA has a satisfactory correlation while bad correlation is 15 Hz.
achieved for FA. The main reason behind this anomaly in SAPE
305
Although the trend of FV-SAPE correlation for soil class SC such method since site estimation method always comes with
seems to be not good enough, the fact that: uncertainty (epistemic and aleatory).
1. FV correlation achieve good result for soil class SE and SD for 2. It appears that SAPE approach is not fully reliable, as it has bad
sites with f0 larger than 1 Hz, and correlation for soft soil sites and for specific spectral
2. All soil profiles within soil class SC has f0 larger than 1 Hz acceleration. However, SAPE result is better than SNI approach
lead to an argument that in fact FV -SAPE also performs well for in several conditions.
class SC. 3. In-line with the finding of Zhao and Xu (2013) and Cadet et al
The bad correlation of SNI approach as observed for FPGA, FA, (2012), natural frequency of a soil profile (f0) turns out to be a
FV for SE sites explains very well the complexity of a site response better site proxy compared to VS30. The very root of the reason is
at soft soils, which has broad variability and cannot be simplified because VS30 cannot include the effect of deeper soil stratum
into a single scalar site factor value (this is also the case for SAPE below 30 m. Furthermore, amplification is also a function of
result). However, it seems that for soil class SD, the correlation is impedance contrast of the soil profile with the underlying
not becoming any better. The SNI site factors achieve satisfactory halfspace, which makes an even worse estimation for VS30
result for hard soil sites (class SC) for PGA. For F A, good result is approach for deep-bedrock profiles.
only valid for sites with VS30 larger than approximately 450 m/s. 4. The study presented in this paper is only performed for linear
Compared to SAPE site factors for class SC, SNI factors tend to condition (low PGA, small strain), therefore non-linearity is not
have low value which can be explained by two reasons. First, the taken into account. The inclusion of such variable will make it
reference stations within this soil class, as used by Borcherdt (1994) more challenging since the uncertainty is increased.
in his study, is basically consist of hard soils and soft rocks, which Based on the finding and discussion, a summary table is
eventually will lead to less amplification. Second, as mentioned prepared which consist of suggested approach for different site
above, the difficulty of obtaining f0 in hard soil sites leads to conditions and different spectral accelerations. The authors hope
overestimation of site factors. In this case, class SD and SE are the that this paper can give an insight for practicing engineers and
governing soil profiles and lead to correlation bias. stakeholders, that there is much into it behind the site factors of SNI
1726-2012 as adopted from NEHRP provision 1997. In the end, the
5. CONCLUSION AND RECOMMENDATION selection of any method to estimate site amplification will largely
depend on many aspects, e.g. project stage, project budget, project
Through stochastic site response analyses performed on varying risk, etc.
artificial soil profiles, the application of site factor estimation It is important to note that this table is made solely based on the
procedure by VS30 (as adopted in SNI 1726-2012) and by f0 (Cadet finding of this study. Improved aspects of this study e.g. more
et al, 2012) has been compared and several conclusions can be taken uniform distribution of artificial soil profile, deeper bedrock depth,
as follow: different method to estimate unit weight and material quality factor,
1. There is no perfect site factor estimation approach. Site proxy is etc may lead to different recommendation as presented herein.
just a proxy of a soil profile, a simple representation of a
complex presence. Although it seems that one method
outperforms the other, it does not guarantee the reliability of
Table 3. Recommended site factor estimation approach based on this study

soft soil medium soil hard soil
SNI soil class SE SD SC
VS30 (m/s) < 175 175 - 350 350 - 750
f0 (Hz) < 1.8 1.8 - 5.6 >3
FPGA site response site response SNI / SAPE
FA site response site response / SAPE* SNI / SAPE**
FV site response / SAPE*** SAPE SAPE
*) SAPE is applicable for sites with f0 > 5 Hz
**) SNI is applicable for sites with VS30 > 450 m/s; SAPE is applicable for sites with f0 > 15 Hz
***) SAPE is applicable for sites with f0 > 1 Hz
6. REFERENCES Castellaro, S., Mulargia, F., Luigi Rossi, P., (2008) “Vs30: Proxy
for Seismic Amplification?”, Seismological Research Letters,
Borcherdt, R.D., (1994) “Estimates of Site-Dependent Response Vol. 79, pp. 540-543.
Spectra for Design (Methodology and Justification)”, Yogatama, B.A. and Lengkeek, H.J., (2017) “Comparison of several
Earthquake Spectra, Vol. 10, No. 4, pp. 617-653. non-linear soil models in a ground response analysis for a
Cadet, H., Bard, P-Y., Rodriguez-Marek, A., (2012) “Site effect typical sand and clay profile”, Proceedings of the 3rd
assessment using KiK-net data: Part 1 - A simple correction Performance Based Design in Earthquake Geotechnical
procedure for surface/downhole spectral ratios”, Bulletin of Engineering Conference, Vancouver, Canada.
Earthquake Engineering, Vol. 10, pp. 421-448. Zhao, J.X. and Xu, H., (2013) “A Comparison of VS30 and Site
Cadet, H., Bard, P-Y., Duval, A-M., Bertrand, E., (2012) “Site Period as Site-Effect Parameters in Response Spectral
effect assessment using KiK-net data: Part 2 - site Ground-Motion Prediction Equations”, Bulletin of the
amplification prediction equation based on f0 and VSZ”, Seismological Society of America, Vol. 103, No. 1, pp. 1-18.
Bulletin of Earthquake Engineering, Vol. 10, pp. 451-489.
306
MAPS OF CORNER PERIOD (Tc) OF RESPONSE SPECTRA

IN CITY OF JAKARTA
Delfebriyadi1, I. Masyhur1, M. H. Bigman1, I. Iswandi1 & M. Asrurifak2
1
Faculty of Civil and Environmental Engineering, ITB, Indonesia
2
Research Center for Disaster Mitigation, ITB, Indonesia
Email: delfebri@yahoo.co.id
ABSTRACT: The seismic response of deep soil surface layers is of concern in developing the seismic microzonation in Jakarta. The present
studies show that the thick sedimentary layer in Jakarta may locally up to 750 m, which means the effect of bedrock depth on site
characterization for this typical of site is importance. To develop design response spectra for structures, the present building codes are now
based on short-period (0.2 sec) and long-period (1.0 sec) spectral accelerations instead of basing design spectra on the effective peak
acceleration (EPA) and effective peak velocity (EPV). For differences in this basing design spectra, the corner period (Tc) of the response
spectra provides the most stable information on the frequency contents of the ground motion. From our study, for 2475 years design response
spectra, the general tendency appears to consist in larger values of Tc (Tc > 1.0 sec) in the several parts of Jakarta, in particular in the eastern
region. These results are based on 1-D site response analysis performed at 383 borehole sites. The modified earthquake acceleration time
histories were adopted in this study.
Keywords: Deep soil layer; site response analysis; design response spectra; corner period.
1. INTRODUCTION 2. METODOLOGY
To construct the smooth response spectra for design purposes, the For estimating ground acceleration at ground surface in Jakarta, the
use of a constant Sa for short periods of response, constant Sv for site response analysis is performed. One of the most important steps
the mid range, and constant Sd for long period response had been in the analysis is site characterization. The average of standard
compensated for the peaks, valleys, and shape variations in actual penetrate blow count numbers (N) value up to 30 metres is a
response spectra (Newmark and Hall, 1982). In earlier, ATC-3-06 common parameter for classifying sites. The classifications
proposed the effective peak acceleration (EPA) and effective peak according to the Indonesian seismic building codes (SNI-1726-
velocity (EPV) to account the differences in the high and low 2012) show that most of location in Jakarta (based on 383
geotechnical boring sites) is classified as the soft soil site (SE) and
frequency characteristics of ground motions to be used in
medium soil (SD) with average N value less than 15 and ranging
developing of the design response spectra for structures. In present
from 15 to 30 respectively (Figure 2).
time, the building codes are now based on short-period (0.2 sec) and
long-period (1.0 sec) spectral accelerations. For example, SD1 and
SDS are the spectral response design values adopted in FEMA 302.
The value of corner period (Tc) and the spectral acceleration value
(SA) in mid range of design spectrum shall be as given by the
following equation:
Tc = SD1/SDs (1)
Sa(T>Tc) = SD1/T (2)
106.7 106.75 106.8 106.85 106.9 106.95
(a)
-6.1
-6.15
-6.2
Tc
-6.25
Figure 1 Design response spectrum (FEMA, 1997) -6.3
-6.35
Then, the curve will be constructed from the parameter above.
The equation force the Sd1 value to be lower than Sds value. So, Tc 106.7 106.75 106.8 106.85 106.9 106.95
value will not greater than 1.0. From our study, for 2475 years (b)
design response spectra, the general tendency appears to consist in Figure 2 The distribution of borehole points in Jakarta and the
larger values of Tc (Tc > 1.0 sec) in the several parts of Jakarta. contour of site classification based on N values (SE site = triangle
These results are based on 1-D site response analysis performed at symbol or grey colour; SD site = square symbol or black colour)
the borehole sites.
307
The shear wave velocity (Vs) distribution with depth determined

by utilizing available N in boreholes is established using the g
empirical relationship proposed by Ohta & Goto (1978), Imai &
0.6
Tonouchi (1982). These empirical relationships have been selected
0.55
since they involve all soil types. The variations of Vs values with
0.5
depth for formations in 383 borehole sites are then given in Figure 0.45
3. 0.4
0.35
Vs Profile for All Borehole Sites 0.3
Vs (m/s) 0.25
0.2
0 200 400 600 800
0.15
0
690000 695000 700000 705000 710000 715000
Figure 4 The contour of EPA value at ground surface in Jakarta city

with 2% probability of exceedance in 50 years
20
cm/sec
40
Dep th (m)
65
55
52.5
60 50
47.5
45
42.5
80 40
690000 695000 700000 705000 710000 715000
Figure 5 The contour of EPV value at ground surface in Jakarta city

100 with 2% probability of exceedance in 50 years
Figure 3 Shear wave velocity profile across Jakarta from all
borehole sites
9325000
sec
Site response analyses were performed at 383 borehole sites.
Representative shear wave velocities were determined for each layer 9320000
2.4
2.2
using estimated Vs along boreholes. All of the boreholes failed to
2
reach the engineering bedrock. Therefore a layer representing 9315000
1.8
engineering bedrock deposits was included into each shear wave
1.6
velocity profile, with Vs of 760 m/s. In this case, extrapolation of Vs 9310000 1.4
values from end of boreholes to layer representing engineering 1.2
bedrock has to be done. The base of the soil columns constructed 9305000 1
down between 350 to 750 m depth as information from the depth to 0.8
engineering bedrock beneath Jakarta. Borehole sites were modelled 9300000

0.6
by NERA (Bardet & Tobita, 2001). As input data the interval Vs as

well as the mass density of soil (ρ) and thickness of each layer were
690000 695000 700000 705000 710000 715000
used. For each soil types, the shear modulus reduction and damping
Figure 6 The contour of Tc value at ground surface in Jakarta city
ratio curves that default by NERA were used in this study. The
with 2% probability of exceedance in 50 years
strong motion was considered to be the the rock outcropping for
modelling purposes.
4. CONCLUSION
3. RESULT The input PGAs of rock outcrop motions are varied from 0.1g to
0.4g and the output EPAs of surface motions are varied from 0.15g
The map contour of the effective peak acceleration (EPA) and the to 0.5g. For 2475 years design response spectra, the general
effective peak velocity (EPV) at the ground surface are obtained tendency appears to consist in larger values of Tc (Tc > 1.0 sec) in
from plots of result of site response analysis on 383 borehole sites. the several parts of Jakarta, in particular in the eastern region.
The EPA and EPV are calculated from 0.4 value of the maximum Which mean, the larger values of Tc would give the SD1 value
values of the response spectra averaged on a 0.4 s period mobile greater than SDS value that affect the effectiveness of the
window. The Tc value then estimates from the following equation: combination of 2 equations (equation 1, equation 2) adopted by the
code for constructing the smooth response spectra for design
2 .EPV (3)
Tc  purposes.
EPA
308
5. REFERENCES
Applied Technology Council (1978) “Tentative Provisions for the
Development of Seismic Regulations for Buildings,” ATC 3-
06, NSF 78-8, Applied Technology Council, Redwood City.
Bardet, J.P. & Tobita, T., (2001) NERA, A Computer Program for
Nonlinear Earthquake site Response Analyses of Layered
Soil Deposits, Departement of Civil Engineering USC, Los
Angeles.
Federal Energy Management Agency (1997) "FEMA 302- NEHRP
Recommended Provisions for Seismic Regulations for New
Buildings and Other Structures.
Imai, T. & Tonouchi, K., (1982) Correlation of N-value with S-
wave Velocity and Shear Modulus, Proceeding of The 2nd
European Symposium of Penetration Testing, Amsterdam,
57-72.
Newmark, N.M., and Hall, W.J., (1982) “Earthquake Spectra and
Design”, Earthquake Engineering Research Institute,
Oakland, California, U.S.A.
Ohta, Y. & Goto, N., (1978) Empirical Shear Wave Velocity
Equations In Term Of Characteristic Soil Indexs, Earthquake
Engineering Structural Dyanamics, 6(2),167-187.SNI-1726-
2012, (2012) Tata Cara Perencanaan Ketahanan Gempa untuk
Struktur Bangunan Gedung dan Non Gedung, Badan
Standarisasi Nasional, Jakarta.
309
Analysis of Bedrock Synthetic Ground Motion on Bandung City using PSHA Method
Arifan Jaya Syahbana1, Anggun Mayang Sari2, Eko Soebowo2, Masyhur Irsyam3, Muhammad Asrurifak4, Hendriyawan5
1
Student of Doctoral Program of Civil Engineering, FTSL ITB, Bandung, Indonesia
2
Researchers LIPI, Bandung, Indonesia
3
Professor of Civil Engineering, FTSL ITB, Bandung, Indonesia
4
Researcher of PPMB ITB, Bandung, Indonesia
5
Marine Engineering Lecturer, FTSL ITB, Bandung, Indonesia
E-mail: arifanjaya@s.itb.ac.id
ABSTRACT: It is a necessity to calculate ground motion in large and dense urban areas such as Bandung City. In order to calculate the
threat of earthquake hazard and risk both primary analysis (for soil or structural dynamic stability) and secondary analysis (for liquefaction)
requires bedrock ground motion. The synthetic ground motion could be an alternative method if the recording data is not attained yet. This
study used Probabilistic Seismic Hazard Analysis (PSHA) method to generate the synthetic ground motion on the bedrock. This study used
earthquake sources in the radius area of 500 km from the survey location, using the latest earthquake data in order to obtain results closer to
real condition. These ground motions were useful for identifying site soil response and mitigation. The results of peak ground acceleration at
the bedrock are 0,502g; 0,505g; 0,510g and 0,539g for BH-1, BH-2, BH-6, and BH-7 site, which is similar to the analysis result of
Indonesian Earthquake Source and Hazard Map 2017 at 2475 years return period.
Keywords: bedrock synthetic ground motion, PSHA, Bandung City
1. INTRODUCTION Seismic sources used in this study are included all mechanism
(shallow crustal and megathrust) within 500 km from studied
1.1 Hazard Analysis location (Table 1 and Table 2). These sources are obtained from
Analysis of ground motion for large and dense urban areas such many catalogue (USGS, Engdahl, BMKG, and PusGen).
as Bandung City is very needed. The purpose of calculating the Table 1. Seismic Source for Megathrust
earthquake hazard is to determine the primary hazard (earthquake
shock) and secondary hazards (phenomenon). In seismic analysis for W M
structural building, sometimes ground motion is required in the L b- a-
Structure Segment
dynamic analysis (Argyroudis & Pitilakis, 2012; Azzam, 2014; (km) (km) PSHA val val
Houshmandan, Rasouli, & Hajiamiri, 2014; Sugianti, Sari, & Sumatran
Syahbana, 2015; Syahbana & Sugianti, 2013; Xiaodong, Weiyuan, Selat Sunda 290 100 8.7 1.15 5.99
Megathrust
& Ruoqiong, 2001). Problems arise if there is no data recorded
Sunda West-Central
ground motion at that location. Thus, an approach is needed to 700 150 8.7 1.08 5.55
Megathrust Java'
obtain the ground motion, which is referred to synthetics ground
Sunda
motion. East Java 280 150 8.7 1.08 5.63
Megathrust
In this study, earthquake hazard in bedrock will be analyzed
using PSHA approach. It is compulsory to gain earthquake data with
a radius of 500 km from the study location to get spectral curves Table 2. Seismic Source for Shallow Crustal
from various sources (Uniform Hazard Spectra/UHS). The next
L W M
analysis is deaggregation to determine the dominant earthquake Segment Type Dip
(km) (km) Max
source to obtain the response spectra. The next step is generating
spectral matching from UHS and RS, and the last is running spectral Kumering-
SS ~90 111 20 7.5
matching using ground motion database. The result is synthetics North
ground motion. Kumering-
SS ~90 60 20 7.1
South
1.2 EZ-FRISK Semangko
SS ~90 90 20 7.4
EZ-FRISK ™ is one of earthquake analysis software that contains Barat-A
several key capabilities which are Seismic Hazard Analysis, Semangko
SS ~90 80 20 7.3
Spectral Matching, and Site Response Analysis. This capability Barat-B
allows various seismic hazard issues to be solved with direct input Semangko
SS ~90 12 20 6.5
specifications using a graphical user interface. EZ-FRISK ™ is Timur-A
designed to be a user-friendly tool for beginners who often interact Semangko
SS ~90 35 20 6.9
with seismic issues. This allows seismic risk analysts focused on Timur-B
identifying key inputs and making decisions that affect earthquake Semangko
Normal 60 50 20 6.5
hazard evaluation, rather than in routine preparing input files, Graben
running command line programs, and plotting out the calculated Ujung
SS ~90 80 20 7.3
results. EZ-FRISK ™ helps analysts make a better risk mitigation Kulon A
planning and decisions in dealing with earthquake threats (Risk Ujung
Engineering, 2011). SS ~90 150 20 7.6
Kulon B
Cimandiri R E-W 23 23 6.7
2. PROBABILISTIC SEISMIC HAZARD ANALYSIS
(PSHA) Nyalindung-
R 30 15 6.5
Cibeber
2.1 Seismic Sources Rajamandala SS 45 22.5 6.6
310
Lembang SS E-W 29.5 29.5 6.8 and unbranched sandstone with Holocene age. This formation has a
fascinating relationship with the upper Cibeureum Formation.
Subang R NW 33 16.5 6.6 Cikapundung Formation is the oldest rock unit exposed in the study
Cirebon-1 R NW 15 7.5 6.5 area (Koesoemadinata & Hartono, 1981), and consists of a
conglomerate and compact breccia, tuff, and andesite lava. The age
Cirebon--2 R NW 18 9 6.5 of this formation is thought to be early Plistocene. The compactness
Karang of the lithology of this formation can be used as one of the
R EW 22 22 6.7
Malang distinctions with Cibeureum Formation, as well as the basis for
Brebes R 22 11 6.5 determining this formation as bedrock in this area. According to
(Silitonga, 1973), this formation is equivalent to Qvu. In addition to
Tegal R ENE 15 15 6.5 this formation, based on its lithological properties, Qvl, Qvb, Qob,
Pekalongan R NE 16 16 6.6 and Qyl can be included as bedrock. Other units that form the
bedrock are Quaternary volcanic rocks (except Cibeureum
Weleri R 17 17 6.6 Formation and Cikapundung Formation), Tertiary volcanic rocks,
Semarang R EW 34 17 6.6 Tertiary sedimentary rocks, and breakthrough rocks encompassed in
geologic maps composed by Alzwar et al. (1992) and Sudjatmiko
Rawapening R NW 18 9 6.5 (1972). Based on the gravity data on the basin of Bandung reflect
Demak R EW 31 15.5 6.6 basement / basement configuration, it is seen that graben structure
pattern of low and high that describe has happened deformation
Purwodadi R EW 38 19 6.7 process which is controlled by the appearance of structure of
Cepu R ESE 100 50 7.1 alignment of northwest trending structure -southeast and northeast-
southwest. Lower Zone such as in Dayeukolot, Cimahi, Rancaekeh
and altitude in Gedebage, Banjaran, Cililin. Water drilling data and
geotechnics drilling by Soebowo et al. (2016) and Tohari et al.
2.2 Attenuation (2015) show that bedrocks are found at depths -50 to -80 m in the
eastern part of Gedebage, apparently in some locations may be more
Attenuation used to calculate fault/shallow crustal mechanism are than 90 meters. The seismicity recorded in the Bandung Basin in
Bore-Atkinson (2008) NGA USGS 2008, Campbell-Bozorgnia 2011 by Sulaeman & Hidayati (2011) is 3.4 SR. In addition, the
(2008) NGA USGS 2008 and Chiou-Youngs (2007) NGA USGS history of seismicity in the Bandung Basin was also observed in
2008. Then, atttenuation for subduction mechanism is Zao et all 1999, 2000, 2003, 2004, 2005 (Marjiyono, Soehaimi, & Kamawan,
(2006) USGS 2008, Atkinson-Bore (2003) Worldwide Subduction 2008).
USGS 2008 and Young (1997) Subduction Rock All.
The study area consists of four locations in Bandung area. This
2.3 Site Location location is a point that belongs to the eastern area of Bandung and
still lies in the basin of Bandung. The point coordinates to be
Bandung basin plains represent depression in West Java that shows reviewed can be observed in Table 3 and can be seen in Figure 1.
the basin between the mountains (van Bemmelen, 1949 in Dam,
1994). The basin is surrounded by Quarter volcano cones and lava Table 3. Coordinates of Site Location
domes that are characterized by active volcanic activity. Basin
Location Latitude Longitude
Bandung has a length of approximately 60 km starting from
Padalarang-Nagrek and width 40 km from Lembang-south Soreang. BH-1 -6.981917 107.700945
This height is between 650 m to 2200 m. The Bandung Basin is
composed of lake sediments, lake fans, swamps, alluvial fans, BH-2 -7.01091 107.725492
floodplains, paleosoils of the Quaternary, in the middle of the basin BH-6 -6.995883 107.693052
there are four runways of lake clay with layers of volcaniclastic and
fluvial facies (Dam, 1994). These sediment conditions are soft and BH-7 -6.981313 107.65068
consolidated. The Bandung basin plain is a historical record of
environmental change as it grows in continuous physical changes
such as changes in river flow, flooding, degradation and
neotectonics on different time scales at the time to illustrate some
types of sedimentary facies, where the condition of geological
structures is a determinant of the sedimentation process (Dam, 1994).
This process is characterized by the looping of the silt-sand unit
(flood plains), the clay of lakes and gravel (alluvial fans) with
varying thickness and geometric shapes in several locations
mutually paying and slightly decreasing and rising and cutting.
Based on the geological history of the Bandung Basin area in the
Quaternary (Dam, 1994), there are several volcanic fans sourced
from Mount Malabar and Wayang Mountain Complex in the south
and Manglayang Mountain in the north, there are also 4 phases of
material that fill the ancient lake basin of Bandung. Formation
Cibeureum lithology compilers of volcanic fans with lithology tufa
sand, lapili, breccia, lava and aglomerate (Qyu) and the results of
old volcano collapse (Qc) mainly found in the north and south
(Silitonga, 1973). Kosambi Formation proposed by Koesoemadinata
& Hartono (1981) to substitute the name of the Lake Deposit used
by Silitonga (1973). Kosambi Formation is also used for the name of
Lake Deposits used by Alzwar, Akbar, & Bachri (1992) and
Sudjatmiko (1972). The distribution of this formation on the surface
is in the middle. Its lithology consists mainly of claystone, limestone,
311
Figure 1. Site Location and schematic cross section geological sub

surface Bandung basin (modified from Hutasoit (2009)).
3. DEAGGREGATION
Deagregation is the process of determining magnitude (M) and
distance (R) which is dominantly based on the result of total
probability which gives the biggest hazard contribution to the
location reviewed in the time of earthquake request and period of
building structure. The distance and magnitude are determined
based on the relationship between magnitude, distance and rate
(incidence/year over the acceleration being reviewed) to then
searched for emphasis on the 3D deaggregation curve (Asrurifak,
2010; Irsyam, 2012). The equations used can be seen in Equations 1
and 2 (Sunardi, 2015).
(1)
(2)
The pair of one magnitude (M) and the distance from the site to
the dominant source (R) and the hazard due to the earthquake can be
expressed in one function. This concept obtains on seismic
deaggregation (McGuire, 2001). Deaggregation can provide an
overview of earthquake magnitude and distance for a particular
earthquake source, which is likely to have a major effect on the site.
One deaggregation purpose is in the dynamic analysis of a building
with a large number of floors. In the dynamic analysis of such high-
rise buildings, it is not possible to use PSHA's ground motion input,
for which the PSHA result mechanism is a combination of various
earthquakes mechanism. For that, it is needed separation of
earthquakes that represent each mechanism. Referring to the
Earthquake Indonesian National Standard 2012, it is required to
simulate the earthquake at least 5 pieces of ground motion input
representing all existing mechanisms (Anonim, 2012).
In deaggregation analysis, the selected point is PGA period
(period close to 0 sec) and return period of 2475 years with damping
5%. Analysis result using Ez Frisk software can be observed in Figure 2. The result of deaggregation at BH-1, BH-2, BH-6
Figure 2. and BH-7 site.
Once the result of deaggregation at certain location obtained,
then it is necessary to define what earthquake mechanisms affected
the hazard predominantly. Here the mechanism could be determined
by observing the graph relationship between hazard and PGA
curved (often referred to as hazard curve). The function of this graph
is to obtain earthquake spectra with modification of Uniform Hazard
Curve. This graph of hazard and PGA relationships can be observed
in Figure 3.
312
become to be point of interest is structure period, where in some

study there is PGA period. In this study, the period to be used is
PGA.
As mentioned before, the reference is the hazard curve of the
Uniform Hazard Spectra (involving all seismic mechanisms around
the site), so that the deagregated (RS) spectral response follows the
bonding point of the period by multiplying by a certain coefficient.
This coefficient is then applied to the entire period. In this study, the
RS was calculated based on the Young 1997 attenuation for the
subduction mechanism. Spectral Matching results can be observed
in Figure 4.
Figure 3. Hazard Curve of various earthquake sources at BH-1,

BH-2, BH-6 and BH-7 site.
As shown in Figure 3 the dominant mechanisms obtained in the
four review sites are Sunda Megathrust in the central-western part of
Java based the PGA dan hazard result.
4. SPECTRAL MATCHING
4.1 Spectral Matching
Spectral matching is conducted to restrain the period that become
concerned to be key of this process. In some work, period that
313
Figure 4. Spectral matching at at BH-1, BH-2, BH-6 and BH-7 site.
4.2 Ground Motion Selection

By comparing the spectral response to the existing ground motion
data base, the results obtained as shown in Figure 5. The peak
acceleration of the ground motion based on the ground motion
synthetic formed is 0.502g; 0,505g; 0.510g and 0.539g for the
location of BH-1, BH-2, BH-6 and BH-7 respectively.
Figure 6. Peak acceleration (PGA) at 2475 year old basalt rock (2%
exceeded in 50 years) from PusGen (2017)
5. CONCLUSION AND RECOMMENDATION

Based on this study, it can be concluded that the peak acceleration
on bedrock based on synthetic ground motion on the return period
2475 years is 0.502g; 0,505g; 0.510g and 0.539g for the location of
BH-1, BH-2, BH-6 and BH-7 sites. This is result has the same value
with the Map of Earthquake Source and Hazard Indonesia in 2017
(PusGen, 2017) for the Bandung area during the PGA period. In
addition, ground motion synthetic analysis is highly sensitive to 1)
the result of deagregation which will affect the attenuation to be
used in determining deaggregation spectral response and 2)
adjustment of the ground motion database affected by the
completeness of the database. The result of ground motion can be
different between the researchers with each other depending on the
data base owned.
In order to calculate ground motion on the surface and its effect
on amplification in the Bandung Basin area, the basin effect should
be taken into account
6. ACKNOWLEDGMENTS
The authors would like to thank the Indonesian Endowment Funds
of Education (LPDP) for the scholarship being granted to carry out
doctoral studies in ITB. Acknowledgments are also addressed to the
LIPI research team of the Center for Geotechnology Research on
data and cooperation in this paper. Last and not least, the authors say
many thanks to the promoter for the science and guidance that has
been given and to others who helped this project whom cannot be
mentioned one by one.
7. REFERENCES
Alzwar, M., Akbar, N., & Bachri, S. (1992). Peta Geologi Lembar
Garut dan Pameungpeuk, skala 1: 100.000. Pusat Penelitian
Dan Pengembangan Geologi, ESDM, Bandung.
Anonim, A. (2012). SNI 1726:2012 Tata Cara Perencanaan
Ketahanan Gempa untuk Struktur Bangunan Gedung dan Non
Gedung, (1), 149.
Argyroudis, S. A., & Pitilakis, K. D. (2012). Seismic fragility curves
of shallow tunnels in alluvial deposits. Soil Dynamics and
Earthquake Engineering, 35, 1–12.
Asrurifak, M. (2010). Peta Respon Spektra Indonesia Untuk
Perencanaan Struktur Bangunan Tahan Gempa Dengan
Model Sumber Gempa Tiga Dimensi Dalam Analisis
Figure 5. Ground motion at locations at BH-1, BH-2, BH-6 Probabilitas. Disertasi Doktor Teknik Sipil ITB, Indonesia.
and BH-7 site. Azzam, W. R. (2014). Seismic Response of bucket Foundation and
The value of PGA produced in this study is valid compare to the structure under earthquake loading. Electronic Journal of
result PusGen Research Team on Indonesia Earthquake Source and Geotechnical Engineering, 19, 1477–1489.
Hazard Book of 2017 for the area around Bandung City which is Dam, M. A. C. (1994). The Late Quaternary Evolution of the
worth between 0,5-0,6g as presented in Figure 6. Thus, PGA value Bandung Basin, West-Java, Indonesia. Thesis Vrije
into the top of the spectra in each synthetic ground motion can be Universiteit, Amsterdam, 252.
used as input on the analysis. Houshmandan, H., Rasouli, R., & Hajiamiri, M. (2014). The Effect
of Near Fault and Far Fault Earthquake on Dynamic Analysis
of Concrete Face Rockfill Dam with Case Study on the Neka,
Gelevard CFRD. Journal of American Science, 10(1s).
Hutasoit, L. M. (2009). Kondisi Permukaan Air Tanah dengan dan
314
tanpa peresapan buatan di daerah Bandung: Hasil Simulasi

Numerik. Indonesian Journal on Geoscience, 4(3), 177–188.
Irsyam, M. (2012). Pengantar Dinamika Tanah dan Rekayasa
Gempa. Catatan Kuliah Jurusan Teknik Sipil ITB.
Koesoemadinata, R. P., & Hartono, D. (1981). Stratigrafi dan
sedimentasi daerah Bandung. PIT X Ikatan Ahli Geologi
Indonesia, Bandung.
Marjiyono, M., Soehaimi, A., & Kamawan, K. (2008).
IDENTIFIKASI SESAR AKTIF DAERAH CEKUNGAN
BANDUNG DENGAN DATA CITRA LANDSAT DAN
KEGEMPAAN. Jurnal Geologi Dan Sumberdaya Mineral,
18(2), 81–88.
McGuire, R. K. (2001). Deterministic vs. probabilistic earthquake
hazards and risks. Soil Dynamics and Earthquake
Engineering, 21(2001), 8.
PusGen. (2017). Peta Sumber dan Bahaya Gempa Indonesia Tahun
2017 (1st ed.). Bandung: Puslitbang Perumahan dan
Pemukiman.
Risk Engineering, I. (2011). EZ-FRISK.
Silitonga, P. H. (1973). Peta Geologi Lembar Bandung, skala 1:
100.000. Bandung: Pusat Penelitian Dan Pengembangan
Geologi.
Soebowo, E., Tohari, A., Wibowo, S., Sugianti, K., Sari, A. M.,
Satriyo, N. A., … Lestiana, H. (2016). Kajian Ancaman
Seismik dan Gerakan Tanah di Wilayah Kawasan Strategis
Cekungan Bandung. Bandung.
Sudjatmiko. (1972). skala 1: 100.000 (Pusat Penelitian dan
Pengembangan Geologi, 1972). Google Scholar.
Sugianti, K., Sari, A. M., & Syahbana, A. J. (2015). DMT method
approach for liquefaction Hazard Vulnerability Mapping in
Bantul Regency, Yogyakarta Province, Indonesia. Indonesian
Journal on Geoscience, 2(1).
https://doi.org/10.17014/ijog.2.1.53-61
Sulaeman, C., & Hidayati, S. (2011). Gempa Bumi Bandung 22 Juli
2011. Jurnal Lingkungan Dan Bencana Geologi, 2(3), 185–
190.
Sunardi, B. (2015). Percepatan Tanah Sintetis Kota Yogyakarta
Berdasarkan Deagregasi Bahaya Gempa. Journal of
Environment and Geological Hazards, Vol. 6 No.(Desember
2015), 211–228.
Syahbana, A. J., & Sugianti, K. (2013). Pemodelan 2D perubahan
tekanan air pori hubungannya dengan Likuifaksi: Studi kasus
Kabupaten Bantul, Daerah Istimewa Yogyakarta. Jurnal
Lingkungan Dan Bencana Geologi, 4(3), 163–177.
Tohari, A., Soebowo, E., Wibowo, S., Syahbana, A. J., Sari, A. M.,
Sugianti, K., … Lestiana, H. (2015). Kajian Ancaman Seismik
dan Gerakan Tanah di Wilayah Kawasan Strategis Cekungan
Bandung. Bandung.
Xiaodong, K., Weiyuan, Z., & Ruoqiong, Y. (2001). STABILITY
ANALYSIS ON THE HIGH SLOPES OF THREE-GORGES
SHIPLOCK USING FLAC-3D [J]. Chinese Journal of Rock
Mechanics and Engineering, 1, 6–10.
315
Influence of Cyclic Behaviour of Vibratory Pile Driving and Surging on Pile

Performance Observed in Model Load Tests in Dry and Saturated Sand Grounds
S. Moriyasu1, M. Aizawa2, T. Matsumoto3, S. Kobayashi3,and S. Shimono3
1
Steel Structures Research Lab, NIPPON STEEL & SUMITOMO METAL CORPORATION, Chiba, Japan
2
Toyama Prefecture, Japan (former student of Kanazawa University)
3
Graduate School of Natural Science and Technology, Kanazawa University, Ishikawa, Japan
E-mail: moriyasu.45e.shunsuke@jp.nssmc.com1, matsumoto@se.kanazawa-u.ac.jp2, koba@se.kanazawa-u.ac.jp3
ABSTRACT: This study focuses on the effect of “cyclic” behaviour of pile installation methods on the penetration resistance and bearing
capacity. A series of laboratory model test was conducted to investigate the cyclic effect by comparing three kinds of piling methods;
monotonic jack-in, pseudo-dynamic push-in and pull-out, i.e. “surging” and vibratory driving in dry or saturated sand grounds. It was found
that surging or vibratory pile driving decreased the pile penetration resistance due to the negative dilation caused by the cyclic shearing of the
soil surrounding the pile. The static load test showed that surging and vibratory pile driving provided the same or larger pile head load than
jack-in method. Furthermore, the fluctuation of the pore water pressure strongly indicated the change of dilation. It was seen that surging and
vibratory pile driving prevent the positive dilation than jack-in method due to the difference of cyclic shearing and monotonic loading.
Keywords: Pile installation method, jack-in, surging, vibratory pile driving, pore water pressure.
1. INTRODUCTION
70 SG1
Recently, it is rather difficult to employ the impact hammer method
in urban areas and residential area, because it causes noises and 175
ground vibrations. Vibratory pile driving and rotary pile installation
(jack-in) methods have been developed as the alternative methods,
SG2
which can install the pile with less noises and ground vibrations.
Previous studies about these methods mainly focused on the soil 100
resistance during pile penetration. In contrast, only a few studies SG3 595mm
focused on the bearing capacity of the pile installed by these 100
methods. If the bearing capacity is clarified, these methods will SG4
become more available.
100
The purpose of this study is to investigate influences of the
different piling methods on the pile performance, such as SG5
penetration resistance, and bearing capacity and load-settlement 50
relation during the pile load test. Significant factor is “cyclic”
movement during pile penetration. Moriyasu et al. (2016) compared strain gauge
the pile behaviour between the monotonic jack-in and surging in dry (a) location of strain gauges (b) photo of the pile
model sand. Surging refers to the cyclic push-in and pull-out Figure 1 Model pile
movement of a pile during installation. Cyclic shearing by surging
caused negative dilation around the pile shaft and soil compaction 2.2 Model ground
below the pile tip.
Thus, this paper focused on the influence of the dynamic surging The material of the model ground was Silica sand #6. Table 1 lists
on the pile penetration resistance and bearing capacity by comparing the physical properties of the sand. The model ground was prepared
to with the influence the pseudo-dynamic surging. A series of in a rigid cylindrical soil container having a diameter of 566 mm and
laboratory tests were conducted in this paper. The laboratory tests a height of 580 mm. The model ground was prepared with 11 soil
included three kinds of piling method (monotonic jack-in, surging layers of 50 mm thick and one layer of 30 mm thick. The sand for
and vibratory pile driving) in two sand conditions (dry or saturated each soil layer was put into the soil box, and the sand layer was
ground). compacted by hand tamping to have a designated relative density, Dr
= 80%. The procedure was repeated to complete the model ground
2. EXPERIMENT DESCRIPTION of 530 mm high.
For the preparation of saturated ground, the soil box was filled
2.1 Model pile with water. Then the sand was poured into the soil box and
compacted, similarly to the case of the dry ground. Although Dr =
An aluminium pipe pile with an outer diameter of 32 mm, a wall
80% was intended also for the saturated ground, Dr = 70% was
thickness of 1.3 mm and a length of 595 mm was used for model
achieved in the saturated ground.
piles. Regarding the pile tip condition, six cases were open-ended
and one case was close-ended. In order to obtain the distribution of
Table 1 Physical properties of the sand
axial forces, strain gauges were attached to the pile shaft at 6
different levels, two strain gauges in the opposite faces at each level, Soil particle density, s (t/m3) 2.679
as shown in Figure 1(a). Minimum dry density, dmin (t/m3) 1.366
The pile surface was coated with an acrylic adhesive to protect Maximum dry density, dmax (t/m3) 1.629
the strain gauges, and glued with silica sand to increase the pile Maximum void ratio, emax 0.962
shaft resistance (Figure 1(b)). The sand was the same material as the Minimum void ratio, emin 0.645
model ground. Mean particle size, D50 (mm) 0.52
316
2.3 Experimental procedure and cases For each ground condition, three kinds of penetration methods
were employed. Jack-in method in Case 1 and Case 5 is the
Figures 2 and 3 show the experimental apparatuses. During pile monotonic pile installation at a penetration rate of 0.2 mm/s.
penetration tests (PPT), the pile was installed by means of a motor Surging method in Case 2 and Case 6 means the cyclic installation
jacks or a vibratory hammer. When the pile was penetrated to a with repetition of 2 mm jack-in and 1 mm pull-out. In these methods,
depth of 400 mm, PPT was finished. Subsequently, during static the pile was installed by using the jack without pile rotation. In the
load test (SLT), the pile head was pushed in two or three loading vibration method in Cases 3, 4 and 7, a vibratory hammer model
cycles. At the end of SLT, the pile extraction force was measured by was employed. As shown in Figure 3, the hammer has a tandem
extracting the pile slightly. Finally, Cone penetration test (CPT) was rotating eccentric weights which generates vertical vibrations and
carried out to investigate the soil strength profile of the model cancels horizontal vibrations. The vibratory hammer has a weight of
ground. 300 N and a maximum frequency of 60 Hz. The close-ended pile
Table 2 shows the test cases. Case 1 to Case 4 were carried out was used in Case 4, while the open-ended pile was employed in the
in the dry ground, while Case 5 to Case 7 were carried out in the other cases.
saturated ground. SLT was carried out subsequent to PPT. In all cases, the pile
penetration rate was 0.1 mm/s.
Screw jack
Vibratory
hammer
Pile
Sand box
Figure 2 Loading apparatus Figure 3 Vibratory hammer model
Table 2 Experimental cases and conditions

Case No. 1 2 3 4 5 6 7
Model ground Dry Dry Dry Dry Saturated Saturated Saturated
Relative density, Dr (%) 79.9 79.9 80 80 69.5 64.3 69.5
Dry density, d (t/m3) 1.568 1.568 1.568 1.568 1.538 1.524 1.538
Pile tip condition Open Open Open Close Open Open Open
Penetration method Jack-in Surging Vibration Vibration Jack-in Surging Vibration
Penetration speed during PPT (mm/s) 0.2 0.2 - - 0.2 0.2 -
Penetration speed during SLT (mm/s) 0.1 0.1 0.1 0.1 0.1 0.1 0.1
Vibration frequency (Hz) - - 20 to 35Hz 20 to 35Hz - - 15 to 20Hz
Static Static
Static loading
loading loading
by V.H. *4
by V.H. *4 by V.H. *4
weight
PPT*1 PPT*1 weight weight PPT*1 PPT*1
↓
↓ ↓ ↓ ↓ ↓ ↓
Test Vibratory
SLT*2 SLT*2 Vibratory Vibratory SLT*2 SLT*2
sequence penetration
↓ ↓ penetration penetration ↓ ↓
↓
CPT*3 CPT*3 ↓ ↓ CPT*3 CPT*3
SLT*2
SLT*2 SLT*2
↓
↓ ↓
CPT*3
CPT*3 CPT*3
*1 PPT: Pile Penetration Test, *2 SLT: Static Loading Test, *3 CPT: Cone Penetration Test, *4 V.H.: Vibratory Hammer
2.4 Measurements axial force was estimated from the strain measured near the pile
head (i.e., SG level 1 in Figure 1(a)). Furthermore, a pair of
During PPT and SLT, the pile head load, pile head displacement and acceleration transducers was attached near the pile head. In the
pile strains were measured. In the cases of jack-in and surging, the saturated sand cases, pore water pressures were measured by the
pile head load was measured by a load cell set between the pile head transducers located as shown in Figure 4.
and motor jack. On the other hand, in the cases of vibration, the
317
566 mm 566 mm 566 mm
PWP3 PWP3
PWP1, PWP3, PWP7 PWP1, PWP6 50 PWP1, PWP6 50

50 10 50 50
10 50 50
PWP4 PWP5 PWP4 PWP7 PWP4 PWP7
50 50 50
PWP2, PWP6 PWP2, PWP5 PWP2, PWP5
140 145 140

PWP1 PWP1 PWP1
100 100 100
PWP2 PWP2 PWP2
540 mm
545 mm
540 mm
50 50 50
PWP4 PWP5 PWP4 PWP4
PWP3 PWP3 PWP3
100 100 100
PWP6 PWP5 PWP5
100 100 100
Coupler PWP7 Coupler PWP7,PWP6 Coupler PWP7,PWP6
50 50 50
Figure 4 Locations of 7 pore water pressure transducers in Cases 5 to 7
CU test (Silica #6) CU test (Silica #6)

800 100
Deviatorio stress, q (kPa)
80 Monotonic
600 Cyclic
60
400 r' = 30 kPa
r' = 30 kPa 40
200 Monotonic
20
Cyclic
0 0
0 2 4 6 8 10 12 14 16 18 20 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
Axial strain, a (%) Axial strain, a (%)
(a) General view (b) Enlarged view
Figure 5 Relationship between the axial strain, a,, and the devlatoric stress, q
CU test (Sillica #6) CU test (Sllica #6)

50 25
Excess pore water press., ue (kPa)
Excess pore water press., ue (kPa)
r' = 30 kPa Cyclic

0 20 Monotonic
r' = 30 kPa
-50 15
Monotonic
Cyclic
-100 10
-150 5
-200 0
0 1 2 3 4 5 6 7 8 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
Axial strain, a (%) Axial strain, a (%)
(a) General view (b) Enlarged view
Figure 6 Relationship between the axial strain, a, and the excess pore water pressure, ue
2.5 Monotonic and cyclic triaxial shear tests of the sand stress paths. The confining pressure of these tests was about 30 kPa.
The internal frictional angle was 38 to 40 degrees, and the shift
In order to investigate the mechanical property of the sand, a phase angle was 28 degrees.
monotonic and a cyclic consolidated undrained (CU) shear tests of
During cyclic loading, it is seen from Figure 5 that a increased
the sand were carried out.
with accumulation of the number of cycle loading with each
Figure 5 shows the relationship between the axial strain, a, and constant q. At the same time, cyclic loading generated excess pore
the deviatoric stress, q. Figure 6 shows the relationship between a water pressure higher than that in monotonic loading. At the
and the excess pore water pressure, ue. Figure 7 shows the effective monotonic loading stage after the cyclic loading, the increment of ue
318
turned from positive to negative, because the negative dilation

changed to positive one. The change point in the cyclic loading (a = Effective vertical stress,  v' (kPa)
1.2 %) was delayed behind the case of monotonic loading (a = 0
0 1 2 3 4 5 6 7 8 9
0.4 %). Therefore, the cyclic loading may have an action to prevent
the positive dilation. The soil behaviour is referred to discuss the (Dry)
Depth form G.L, z (mm)

experiment results later. 100 (Saturated)
CU test (Silica #6) 200 Cases 1 to 3

80
Cyclic 300
70
Monotonic
400 Cases
60
5&7
50 500
Case 6
40
F.L.
600
Figure 8 Vertical effective stress
30
CPT tip resist, tip (kPa)
20 r'=30kPa 0 1000 2000 3000 4000
0
10 (Dry)
100 (Saturated)
Depth from G.L., z (mm)

0
0 10 20 30 40 50 60
Mean effective normal stress, p' (kPa) 200
Figure 7 Effective stress channel paths Case 2
300
3. EXPERIMENT RESULTS
400 Case 6
3.1 Results of Cone Penetration Tests in the model ground
Figure 8 shows the vertical effective stresses, v’, with depth in the 500
cases of dry and saturated grounds. It is noted that ground water
levels in Cases 5 and 7 were 100 mm below from the ground surface. 600
Figure 9 shows the CPT tip resistance, qc, measured in Case 2 (dry
Figure 9 CPT results
ground) and Case 6 (saturated ground). The CPTs were carried out
after SLT. CPT profiles of the other cases were similar to Figure 9.
T im e , t (s )
The qc in the saturated ground (Case 6) was smaller than that of the 0 50 100 150 200
dry ground (Case 2) due to smaller v’ in the saturated ground.
Furthermore, qc may be influenced by the soil behaviour during PPT 10 P ile h e a d
G ro u n d s u rfa c e
and SLT, because CPT was carried out after the PPT and SLT. V ib ra to ry h a m m e r
Acceleration,  (m/s )
2
5
3.2 Results of the pile penetration tests
Figure 10 shows the time history of the accelerations of the pile 0
head, the ground surface and the vibratory hammer during the
vibratory pile penetration in Case 4. The downward acceleration is
taken as positive. As shown in the figure, the acceleration of pile -5
20H z 25H z 30H z
head was much larger than those of the ground surface and the
vibratory hammer. The frequency of the vibratory hammer was -1 0 35H z
changed from 20 Hz to 35 Hz to enhance the pile penetration ability
of the vibratory hammer during PPT. Figure 10 Acceleration during vibratory penetration
Figure 11 shows the relationship between the pile head load, Ph,
and pile head displacement, wh, in Cases 1 to 4 (dry ground). It can Figure 13 shows the relationship between the pore water
be seen that Ph in Case 3 and Case 4 during PPT was almost similar. pressure, p, and pile head displacement, wh, during PPT. In Case 7, p
In Cases 3 and 4, at first, the pile was installed by the self-weight of increased with increases in wh, and p increased dramatically when f
the vibratory hammer from the ground surface. When the pile could was increased to 18.3 Hz at wh = 400 mm. It is seen that soil
not be penetrated by the self-weight at wh = 75 mm, the operation of liquefaction was caused, because the excess pore water pressure
the vibratory hammer was started. Until the pile reached wh = 400
corresponded to the effective vertical stress, v, shown in Figure 8.
mm, the frequency of the hammer was increased if the pile
Therefore, Ph decreased with the reduction of v related to the soil
penetration was degraded. Ph in Case 3 and Case 4 (vibration) was
liquefaction. As mentioned earlier, the CU test shows that cyclic
smaller than that of Case 1 and Case 2 after wh exceeded 200 mm.
shearing generated higher excess pore water pressure than
Figure 12 shows the relationship between Ph and wh in Cases 5
monotonic loading, as shown in Figure 6. If the degree of the excess
to 7 (saturated ground). During PPT, Ph in Case 6 (surging) was
pore water pressure depends on the number of cyclic shearing, the
slightly smaller than that in Case 5 (jack-in). In Case 7 (vibration),
vibration provided a large number of cyclic shearing (3800 times)
Ph decreased significantly, when f was increased to 18.3Hz at wh =
than surging (350 times).
400 mm.
319
Pile head load, Ph (N) Pile head load, Ph (N)

-1000 0 1000 2000 3000 4000 5000 -1000 0 1000 2000 3000 4000 5000
0 0
Pile head disp., wh (mm) PPT PPT
SLT
Pile head disp, wh (mm)

SLT
100 100
200 200
300 300
400 400
500 500
(a) Case 1 (Dry, Push-in) (a) Case 5 (Saturated, Push-in)
Pile head load, Ph (N)

-1000 0 1000 2000 3000 4000 5000 Pile head load, Ph (N)
0 -1000 0 1000 2000 3000 4000 5000
0
PPT
PPT
SLT
100

SLT
100
200
200
300
300
400
400
500
(b) Case 2 (Dry, Surging) 500
(b) Case 6 (Saturated, Surging)
Pile head load, Ph (N)
-1000 0 1000 2000 3000 4000 5000
0 Pile head load, Ph (N)
Static penetration
-1000 0 1000 2000 3000 4000 5000
Vibration 0
Pile head disp., wh (mm)
SLT Static penetration

100 20 Hz (n = 520)
Vibration
29 Hz (n = 435) SLT
100
29.6 Hz (n = 858)
200
33.5 Hz
(n = 720) 200
300
34.5 Hz 15.5 Hz (n = 1603)
(n = 1397) 17.1 Hz (n = 982)
35.3 Hz
300
400 (n = 918)
17.7 Hz (n = 1047)
n : number of vibration cycle 400

18.3 Hz (n = 140)
500
(c) Case 3 (Dry, Vibration) n : number of vibration cycle
500
(c) Case 7 (Saturated, Vibratory)
Pile head load, P h (N)
-1000 0 1000 2000 3000 4000 5000 Figure 12 Relationship between the pile head load, Ph, and the pile
0
Static penetration head displacement, wh, in Cases 5 to 7
Vibration
SLT 20Hz (n=600)

100 3.3 Results of the static load test
25Hz (n=975)
The static load test was carried out subsequent to the pile
200
penetration test. As shown in Figure 11 (a) (b), the pile head load, Ph,
30Hz (n=930)
during SLT was smaller than that during PPT. A possible reason is
300 the difference of the pile penetration rate. Watanabe et al. (2012)
investigated that the failure strength and deformation modulus of
400 35Hz (n=1155)
sand increases with increasing the loading rate. The similar
behaviour may appeared. Conversely, in the Case 3 and Case 4, Ph
n : number of vibration cycle during SLT was higher than that during PPT. A possible reason is
500
the generation of excess pore air pressure during PPT. Watanabe et
(d) Case 4 (Dry, Vibration, close-ended pile) al. (2013) investigated that excess pore air pressure is generated in
Figure 11 Relationship between the pile head load, Ph, and the pile dry sand when it is sheared very rapidly in the triaxial compression
head displacement, wh, in Cases 1 to 4 testing. In order to clarify the behaviour, more study is needed.
320
Figure 14 (a) shows the comparison of load-settlement relations As Bolton et al. (2013) pointed out, the cavity created below the
between Case 1 (jack-in) and Case 2 (surging) in dry sand condition. pile tip collapses during the upward movement of the pile, and the
It should be noted that the origin of vertical axis of Figure 14 (i.e, soil below the pile tip is compacted during the following downward
pile head displacement during SLT, wh_SLT) is the end of wh during movement. This cyclic compaction enhanced the pile base resistance
PPT due to adjust the difference end of wh among cases. #1 in each largely.
figure means the yield point in the 1st loading cycle, and #2 is the Figure 14 (b) shows the comparison of load-settlement relations
end point of increasing load. As shown in Figure 14 (a), the initial among Case 1 (jack-in), Case 3 and Case 4 (vibration). The yield
pile head stiffness in Case 2 (surging) is higher than that in Case 1 loads (#1) in Case 3 and Case 4 (vibration) are larger than that in
(jack-in), and the yield point (#1) in Case 2 is larger than that in Case 1. However, a softening behaviour occurred: i.e., Ph increased
Case 1. Figure 15 shows the change of pile shaft resistance and pile to 3878 N at #1, and dropped immediately to Ph = 3604 N with a
base resistance from #1 to #2. The pile base resistance, Pb, is the little increasing of wh_SLT. It can be seen from Figure 15 (a) that the
axial force obtained from strain SG5, and the pile shaft resistance is pile shaft resistance in Case 4 decreased with increasing wh_SLT
the difference between the pile head load, Ph, and the pile base (from #1 to #2), while the pile base resistance increased. This is not
resistance. The pile base resistance in Case 2 (surging) is higher than related with soil plugging because the behaviour was occurred both
that in Case 1. This is similar to Moriyasu et al. (2016). condition of open-ended pile (Case 3) and close-ended pile (Case 4).
P.W.P, p (kPa) P.W.P, p (kPa) P.W.P, p (kPa)

0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0
200
0

0
250
100 100
300
200
200
350
300
300 400
400
400 450
500
PWP1 PWP2 PWP3 PWP1 PWP2 PWP3 PWP4 PWP1 PWP2 PWP3 PWP4
PWP4 PWP5 PWP7 PWP5 PWP6 PWP7 PWP5 PWP6 PWP7
(a) Case 5(Jack-in) (b) Case 6 (Surging) (c) Case 7 (Vibration)
Figure 13 Relationship between the pile head displacement, wh, and the water pressure (P.W.P.), p, during PPT
Pile head load, Ph (N) Pile head load, Ph (N) Pile head load, Ph (N)
Pile head disp. during SLT, wh_SLT (mm)
-1000 0 1000 2000 3000 4000 5000 -1000 0 1000 2000 3000 4000 5000 -500 0 500 1000 1500
0 #1 0 #1 0 #1
#1
#1 #2 #1 #1 #2
#1
10 10 #2 10
#3
20 #2 20 20 #2 #2
30 30 30
#2 #2
40 40 40
50 50 50
60 60 60
case1(Push-in)
case1(Push-in) case3(Vibration) case5(Push-in)
70 70 70
case2(Surging) case4(Vibration_C.P.) case6(Surging)
case7(Vibration)
80 80 80
(a) Cases 1 and 2 (dry ground) (b) Cases 1, 3 and 4 (dry ground) (c) Cases 5, 6 and 7 (saturated ground)
Figure 14 Relationship between the pile head load, Ph, and the pile head displacement, wh_SLT
5000 1500
base shaft base shaft
4000
Load (kN)
Load (kN)
1000
3000
2000
500
1000
0 0
#1 #2 #1 #2 #1 #2 #1 #2 #1 #2 #1 #2 #1 #2
Case 1 Case 2 Case 3 Case 4 Case 5 Case 6 Case 7
(Jack-in) (Surging) (Vibration) (Vib._C.P.) (Jack-in) (Surging) (Vibration)
(a) dry ground (b) saturated ground

Figure 15 Change of pile shaft resistance and pile base resistance from #1 to #2 marked in Figure 14
321
P.W.P, p (kPa) P.W.P, p (kPa) P.W.P, p (kPa)

0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0
410 390 430
420 435

400
430 440
410
440 445
450 420 450
460 455
430
470 460
440
480 465
490 450 470
PWP1 PWP2 PWP3 PWP4 PWP1 PWP2 PWP3 PWP4 PWP1 PWP2 PWP3 PWP4
PWP5 PWP6 PWP7 PWP5 PWP6 PWP7 PWP5 PWP6 PWP7
(a) Case 5 (Jack-in) (b) Case 6 (Surging) (c) Case 7 (Vibration)
Fig. 16 Relationship between the pile head displacement, wh, and the water pressure (P.W.P.), p, in SLT
Figure 14 (c) shows the load-displacement curves in Cases 5 to 7 of the soil surrounding the pile, judging from the results of the
(saturated ground). The initial pile head stiffness in Case 6 (surging) cyclic shearing in the triaxial CU test. The vibratory pile
was comparable to that in Case 5 (jack-in), and that in Case 7 driving may have caused negative dilation higher than surging
(vibration) was higher than that in Case 5 and Case 6. Conversely to due to a great number of loading cycles in the vibratory pile
the result in Cases 3 and 4 (dry ground), the decrease of Ph in the 1st driving.
loading cycle did not occur in Case 7. As shown in Figure 15 (b), 3) In the static load tests in the dry and saturated grounds, the
the pile base resistance in Case 6 (surging) and Case 7 (vibration) pile head load of the pile installed by surging or vibratory pile
was larger than that in Case 5 (jack-in). And, the pile shaft driving was the same or larger than that of jack-in.
resistance increased with increase in wh_SLT (from #1 to #2). Furthermore, the pile base resistance of the pile installed by
Additionally, when the pile was extracted at the final stage of SLT, surging or vibratory pile driving was relatively large,
the extraction resistance was obtained. It can be seen from Figure 14 especially in the saturated ground. A reason for this result is
(c) that the extraction resistance in Case 6 and Case 7 was that the cyclic compaction increased the soil resistance below
comparable to that in Case 5. the pile tip.
Figure 16 shows the pore water pressure, p, during SLTs. It can 4) During the SLT in the saturated ground, the pore water
be seen from Figure 16 (a) that p in Case 5 (jack-in) is closely pressures were rarely changed in the cases of surging and
related to the dilation of the ground. The negative p (i.e., positive vibratory pile driving, while the negative pore water pressures
dilation of the ground) increments were generated when the pile was were observed clearly in the case of jack-in piling method.
loaded into the compressed direction, while positive p (i.e., negative Since the negative pore water pressure means positive dilation,
dilation of the ground) increments were generated when the pile was the pile installed by jack-in caused the positive dilation of the
unloaded to zero. Although the magnitudes of the pore water ground during SLT. In contrast, surging and vibratory pile
pressures in Case 6 (surging) and Case 7 (vibration) are smaller than driving prevented the dilation of the ground during SLT due to
those in Case 5 (jack-in), they have similar trend to those in Case 5. the cyclic movement during PPT.
Furthermore, p in Case 7 was comparable to the other cases, From these findings, it can be said that surging and vibratory
although the soil liquefaction occurred at the final stage of PPT. pile driving install the pile with causing the negative dilation of the
Hence, it is thought that excess pore water pressure dissipated and ground. And, the bearing capacity of the vibratory driven pile is
the ground was reconsolidated during the preparation work for SLT. comparable to that by jack-in piling method, because the cyclic pile
As mentioned in 2.5., it can be seen in the CU test that cyclic movement enhances the pile base resistance.
shearing has the effect to prevent the positive dilation. If the similar
situation occurs in the laboratory pile load test, the cyclic pile 5. ACKNOWLEDGEMENT
movement during PPTs in Case 6 (surging) and Case 7 (vibration)
may prevent the positive dilation during SLTs. The authors would like to express their thanks to Ms. Syafinaz
Saadon (former student of Kanazawa University), Dr. Vu Ahn Tuan
(former PhD student of Kanazawa University) and Mr. Mayumi
4. CONCLUSIONS
(former student of Kanazawa University) for their kind supports in
In order to investigate the influence of cyclic pile movement on the this study.
penetration resistance and bearing capacity, a series of laboratory
model tests were conducted with the different piling methods, i.e., 6. REFERENCES
monotonic jack-in, surging and vibratory pile driving, in dry or
saturated sand grounds. Furthermore, triaxial CU tests of the sand Bolton, M.D., Haigh, S.K., Shepley, P., and D’Arezzo, F.B. (2013)
used for the model ground were carried out to compare the soil "Identifying ground interaction mechanisms for press-in
behaviours under monotonic or cyclic shearing. The findings are as piles". Press-in Engineering, Proc. 4th IPA International
follows. Workshop in Singapore, pp84-95.
Moriyasu, S., Meguro, H., Matsumoto, T., Kobayashi, S., and
1) In the pile penetration tests in the dry ground, the pile head Shimono, S. (2016) "Influence of Surging and Jack-in Pile
load (penetration resistance) in the surging piling method was Installation Methods on Pile Performance Observed in Model
comparable to that in the jack-in piling method. The pile Load Tests in Dry Sand Ground", 19th Southeast Asian
penetration resistance in the vibratory driving was smaller than Geotechnical Conference & 2nd AGSSEA Conference
those in the jack-in and the surging piling methods. A possible (19SEAGC & 2AGSSEA), Kuala Lumpur.
reason for this is the generation of pore air pressure in the dry Watanabe, K., Sahara, M. (2012) "Effect of Loading Rate on
ground, although more investigation is needed for this aspect. Bearing Capacity and Soil Spring of Pile Foundations", report
2) In the pile penetration tests in the saturated ground, the pile of Obayashi Corporation Technical Research institute, No.
head load (penetration resistance) in the surging piling method 76.
was smaller than that in the jack-in piling method. The pile Watanabe, K. and Kusakabe, O. (2013) "Reappraisal of Loading
penetration resistance in the vibratory piling method was Rate Effects on Sand Behaviour in View of Seismic Design
significantly reduced due to the occurrence of soil liquefaction. for Pile Foundations" Soils and Foundations, 53(2), pp.215-
The cyclic pile movements may have caused negative dilation 231.
322
Dynamic Compaction at New Yogyakarta International Airport for

Liquefaction Mitigation
R.R. Setiaji1, A.M. Hakim2, F.H. Adinda3, and K.M. Abuhuroyroh4
1
BUT Menard Geosystems, Jakarta, Indonesia
2
3
3
Independent Technical Expert for Ground Improvement, Bandung, Indonesia
E-mail: ryan.setiaji@menard-asia.com, abi.hakim@menard-asia.com, febrini.hartianty@menard-asia.com,
roy.yusfi@gmail.com
ABSTRACT: In high seismic areas, liquefaction is a major issue when a designer dealing with loose sand below groud water level and low
fines content. To solve this problem, densification by means of compaction is the most simple approach to be accomplished. Dynamic
compaction technique by repeated impacts of a heavy weight on to the soil can successfully densify the soil and thus, dismiss the liquefaction
issue. The project case described in this paper is the development of New Yogyakarta International Airport, which is located in Kulon Progo,
Indonesia. It is located in high seismic area with Peak Ground Acceleration (PGA) of 0.4g. Sand with fines content ranging from 0.25% to
17.23% was found all over the project site. Furthermore, the design criteria was stated to mitigate the liquefaction during the designed
earthquake. The evaluation of project achievement to mitigate the liquefaction was performed based on In-Situ Testing Parameter; in this
case is Cone Penetration Test. This paper presents a brief discussion of the entire project case which is using Dynamic Compaction method
to mitigate the liquefaction issue.
Keywords: Ground improvement, dynamic compaction, liquefaction mitigation, high seismic area, cone penetration test, new airport.
1. INTRODUCTION 2. DYNAMIC COMPACTION

Liquefaction has been a major issue in geotechnical engineering of The basic principle of the dynamic compaction is transmission of
the countries within seismic belt. Physically, loose sand with low high energy impacts to loose soils which has originally high
fines content below groud water level tends to densify subjected to compressibility, low bearing capacity, and liquefaction potentials
loading; the excess pore water pressure would barely build up due to with the purpose of substantially enhance their characteristic with
its eminent permeability. Nevertheless, when the loading is occured depths. The mechanism of dynamic compaction is repeatedly
immensely and repeatedly in very quick fashion e.g. earthquake dropping a heaving pounder (15-20 tons) from a height (5 to 20 m)
loading, the excess pore water pressure is developed much faster on to soil surface. The compaction is normally carried out in few
than it takes to dissipate untill an extent that the sand loose of all its phases, the first phase is generally higher energy to compact the
strength, liquefaction. When it happens, the upper structure is no deepest treatment zone, and subsequent phase will be carried out
longer supported and building collapse is inevitable. Evidence of with lesser energy and finally ironing phase that consists of
liquefaction has been prevalent in historic earthquakes (Obermeier, overlapping pounding with lower drop height to treat the upper 2-3
1998; Tuttle et al., 2000; Schneider et al., 2001). The most m. It is common that the top 0.5 m is not well compacted due to
susceptible to liquefaction is sediments, including saturated leveling activities after ironing phase. This layer shall be compacted
Holocene to late Pleistocene age deposits, river channel and flood by roller compactor to compact the surface layer. The high energy
plain alluvium, aeolian deposits, and poorly compacted fills (Youd, impacts will generate compression wave that build up porewater
1991; Krinitzsky and Hynes, 2002). pressure and dislocate the soil particles. Shear wave and Rayleigh
Liquefaction can be prevented by densification the existing sand wave would rearrange the soil particles into denser form as shown in
deposits or reclaimed sand using Dynamic Compaction method. By Figure 1.
performing repeated impact of a heavy mass on to the soil, Dynamic
Compaction produces sands with denser state. The denser sands
become, the less liquefaction potential would be. As one of the
ground improvement technique, this method has been recognized as
very effective in densification of loose soil within a variety of depth
(Menard and Broise 1975, Lukas 1980, Chow et al. 1992). In
landfill areas, it has also been successfully utilized for numerous
purposes (Mayne et al. 1984).
This paper presents experiences gained from case history where
Dynamic Compaction technique was adopted for the densification of
existing natural sand deposits. The project case history was a coastal Figure 1 Waves generate by Dynamic Compaction
where the soil investigation carried out before the work shows that
the site was composed of natural sand deposits with low fines The degree of improvement is a function of the applied energy
content. Based on earthquake map of Indonesia, the project was where it depends on tamper weight, drop heights, number of blows
located in high seismic zone. The liquefaction sucseptibility of the and grid pattern. In term of depth of improvement, it depends on
site was obvious, hence the liquefaction criteria was developed as a tamper weight and drop height. The empirical correlation below
main objective of ground improvement works. The calibration test could be used for preliminary design.
was performed prior to the densification works to determine the
optimum energy required. The production and post-production D = n (W.H)0.5 (1)
monitoring and evaluation was also performed. It shows that the D = depth of Improvement
Dynamic Compaction is successfully met the required design n = empirical coefficient
criteria. The works was completed with high productivity resulting a W = weight of pounder
short and compact working period. H = drop height
323
The emprical coefficient n is the range of 0.3 to 1.0 depending

on soil types, efficiency of the drop mechanism and existance of
hard layer or energy absorbing layer. Table 1 shows the recomended
values for n extracted from the past experience and record.
According to FHWA-SA-95-037 (1995), the applied energy is
generally given as the average energy applied over the entire area.
W ∙H ∙N ∙P
AE =
G2
(2)
AE = applied energy
N = number of drops at specific drop point
W = mass of tamper in tons
H = drop height in meters
P = number of passes
G = grid spacing in meters
Figure 2 Project location
More recently, Varaksin and Racinais (2009) have proposed Eq.
(3) which considers the degree of improvement as a function of
depth.
݂2 - ݂1
f(z) = (‫ݖ‬− ܰ‫)ܮܩ‬2 + ݂1
‫ܦ‬2
(3)
where,
f(z) = improvement ratio at elevation (z)
z = elevation in meters
NGL = natural ground level
D = depth of influence of dynamic compaction
f1 = maximum improvement ratio at surface Figure 3 Project ground improvement area
f2 = improvement ratio at maximum depth of influence
Table 1 Recommended Values for n 3.2 Design Criteria – Liquefaction Mitigation
Source n-values Soil Type The project was located within high seismic area as shown in
Menard & Broise Indonesia seismic map, Figure 4, with Peak Ground Acceleration
1.0 All soils (PGA) value of 0.4g. The design criteria was to mitigate the
(1975)
Leonards, Cutter & liquefaction, with moment magnitude, Mw, of 7.8 and safety factor
0.5 - against liquefaction of 1.2.
Holtz (1980)
Soil with unstable
0.5
structure
Smoltczyk (1983)
0.67 Silts and sands
1.0 Pure frictional soils
Lukas (1980) 0.65-0.8 -
Mayne, Jones, &
0.3-0.8 -
Dumas (1984)
Gambin (1985) 0.5-1.0 -
0.65 Fine sand
Qian (1987) 0.66 Soft clay
0.55 Loess
0.65 Silty sand
Vsn Impe (1989) 0.35 Municipal waste
0.5 Clayey sand
Yee, Setiawan & Calcareous sand/coral
0.5
Baxter (1998) sand
Faisal, Yee &
0.33-0.59 Municipal waste
Varaksin (1997)
3. CASE STUDY: NEW YOGYAKARTA

INTERNATIONAL AIRPORT
Figure 4 Indonesia seismic map 2012
3.1 Project Description
The project was located in Kulon Progo Regency, Yogyakarta In regard to the seismic data, the quick acceptance criteria was
Province. The exact location can refer to Figure 2. then developed to evaluate the compaction results using simplified
The total of treatment area, which is composed of one runway, procedure of 1996 NCEER and 1998 NCEER/NSF (Figure 5). It
two rapid exit taxiways, two holding bay, one paralel taxiway, three was presented in qc value versus depth, as the post-in-situ-test was
exit taxiways, one taxiway apron, and one apron, was approximately using Cone Penetration Test. The quick acceptance criteria is
900,000 m2. Figure 3 shows the ground improvement works area. presented in Figure 6.
324
Figure 6 Quick acceptance criteria
Based on the laboratory result, it was found that the sand

throughout the project site having fines content ranging from 0.25%
to 17.23%. Hence, it can be simplified that the sand was
compounded of clean sand to silty sand.
Figure 5 Evaluation risk of liquefaction procedure based on

simplified NCEER 1998
3.3 Geotechnical Condition

To understand comprehensively about the soil condition, it is better
to identify about the history of soil formation specifically at project
site. The site was located within the watershed of rivers originally
coming from the Merapi Volcano. The nature of sand in project sand
is predominantly black colored. This kind of sand is well-known in
Figure 7 Water flow in river system and sedimentation process
Southern Java as Iron Sand, which is abounding of iron mineral.
Iron sand is commonly found along the coast, formed by the process
of razing the origin rocks from Merapi Volcano by weather and
surface water, which are then transported and deposited along the
coast. Waves of sea with a certain energy sorted and accumulated
the sediment into iron sand deposit. Figure 7 presents the illustration
of the sedimentation process. Figure 8 shows the site condition prior
to dynamic compaction work.
A robust geotechnical investigation was performed by the
consultant prior to project design and execution. The in-situ test
consisting of 84 boreholes were combined with laboratory test. The
plot of N-SPT data from boreholes is presented in Figure 9.
Generally, the soil condition was sand. Based on boring logs and
N-SPT data, it was found that the top four meters is composed of
loose to medium dense sand (N-SPT: 3 – 25 blows/30 cm). Then,
the next two meters below the top layer the sand condition was in
denser state (N-SPT: 15 – 50). At eight-meter depth and below, the
sand was found in very dense form (N-SPT >50). Figure 8 Site condition prior to ground improvement work
325
3.4.1 Field Calibration Test

A full scale of calibration test was performed on site. An area of 2 x
20 m x 20 m (total area of = 800m2) was selected for field
calibration test. This test incorporated heave and penetration test,
pre- and post- cone penetration test (CPT). The objective of this
calibration test was to determine the optimal compaction energy,
which is composing of:
 Number of blows
 Drop height
 Spacing and grid pattern
 Magnitude of heaving
The purpose of heave and penetration test (HPT) was to
determine the optimal energy (number of blows) required for the
compaction works. The test involved calculating the penetration
depth, heave volume, print volume and the effective volume (i.e.,
print volume minus heave volume). The data was recorded and
plotted into graph as shown in Figure 11 and Figure 12.
Figure 11 Typical cross section due to pounder compaction
Figure 9 N-SPT summary graph
3.4 Dynamic Compaction Works

Dynamic compaction (DC) method was chosen based on the
required treatment depth that is clearly suitable, as the maximum
treatment depth for DC is 10 meters. The fines content was also
properly fit below the maximum fines content for DC work which is
20%. Environmentally, the site was located in green area and almost
no critical structure nearby the site so that the vibration induced by
the works would not be an issue. Figure 11 Typical graph to analysed HPT result
As automatic free fall method was used (Figure 10), the
properties of pounder used for the DC work was as follows: The drop height was selected 15 m. Based on heave and
 Shape : Circular penetration test, the result was analysed and evaluate to obtain the
 Diameter : 2.20 m optimum production parameter (Figure 12). As shown in Figure 13,
 Height : 1.25 m the effective volume is started to change in shape at blows 4 and 5,
 Weight : 17 ton indicating that the effective blows to be use is either 4 or 5.
Accordingly, the calibration test was performed using 4 blows at
Zone A and 5 blows at Zone B (Figure 14).
After the calibration test was carried out, several post-CPT test
was performed as shown in the Figure 14. The result was showing
that both zones achieved the design criteria. As conclusion, DC
work was comprised of one phase with 5 m spacing, square grid
pattern, and 4 number of blows.
Figure 10 Dynamic compaction rig and pounder on site Figure 12 Ground settlement and heave with distance
326
Figure 13 Volume vs number of blows
Figure 16 Work acceptance flowchart
all over the site. The work acceptance flow chart was also
estabilished as presented in Figure 16. The first-rate CPT rig with
capacity of 20 ton and auto-generated CPT result was used to carry
out the tests (Figure 17).
Figure 14 Schematic plan of field calibration test
3.4.2 Construction Performance

Menard has maintained the working capacity and quality through
the time since the beginning. The highest production capacity at the
site was 13,000 m2/day and the average was about 8,000 m2/day.
The total area of 40 Ha was finished completely within 3 months.
Total of volume sand treated by the dynamic compaction works was
approximately 2,500,000 m3. Figure 15 presents the documentation
during compaction work on site.
Figure 17 CPT rig with 20 ton capacity and auto-generated result
3.5 Result and Evaluation

Figure 15 Dynamic compaction work on site
As an evaluation, the pre-test CPT and post-test CPT result are
compared. The post tests result are showing that the dynamic
3.4.3. Quality Assessment and Quality Control
compaction work is resulting the post CPT qc value above the qc
As part of quality assessment and quality control, cone penetration value of quick acceptance criteria. The sample of comparison
test was carried out for every 50 m x 50 m (2,500 m2) of treatment between pre-CPT and post-CPT are presented in Figure 18.
area throughout the project site for both pre- and post- ground The improvement ratio is also analysed to obtain more big
improvement work. By using this scheme, the evaluation of the picture of the result of dynamic compaction work at the project site.
work was more comprehensive and the data was more representative The graph is showing that the improvement ratio is vary to 6.5 depth
327
(Figure 19). The 0.5 m thick layer on top is loosened certainly due 4. CONCLUSION
to the mechanism of heaving. The improvement ratio increase until
reach the peak at 3.2 m, so-called optimal depth, then decrease as Dynamic compaction has demonstrated to be a successful technique
the depth getting deepened and the existing qc value getting higher. to densify and mitigate the liquefaction issue in high seismic area.
The improvement ratio is insignificant after the existing qc value This paper has highlighted the effective implementation of dynamic
higher than 25 MPa. The average improvement ratio is compaction to treat liquefiable ground condition for new
approximately 1.8. international airport in Indonesia with compaction area of
approximately 900,000 m2. The calibration test is required to be
carried out prior to dynamic compaction work. Proper quality
assessment and quality control is essential to maintain the accurate
result and evaluation.
5. ACKNOWLEDGEMENTS
Special acknowledgement is made to PT. Pembangunan Perumahan
(Persero) as our client and to PT. Angkasa Pura I (Persero) as our
project owner for their positive response towards the use of non-
conventional solution using dynamic compaction for the new
international airport to mitigate liquefaction problem – first time in
Indonesia.
For the unforgettable work, acknowledgement is also made to
all Menard project team of New Yogyakarta International Airport
which has delivered outstanding efforts to be able to achieve this
work smoothly executed.
6. REFERENCES
Chow, Y.K., Yong, D.M. and Yong, K.Y. (1992) “Dynamic
Compaction of Loose Sand Deposits”, JSSMFE, Soils and
Foundations, Vol.32, No.4, p.93-106.
Figure 18 Comparison of pre-test and post-test CPT Krinitzsky, E.L., Hynes, M.E., (2002) “The Bhuj, India, earthquake:
lessons learned for earthquake safety of dams on alluvium”,
Engineering Geology 66 (3– 4), 163– 196.
Lukas, R.G. (1980) “Densification of Loose Deposits by Pounding”,
ASCE, Journal of Geotechnical Engineering, GT4, p.435-
446.
Mayne, P.W., Jones, J.S. and Dumas, J.C. (1984) “Ground Response
to Dynamic Compaction”, ASCE, Journal of Geotechnical
Engineering, Vol. 110, No.6, p.757-774.
Menard, L. and Broise, Y. (1975) “Theoretical and Practical Aspects
of Dynamic Compaction”, Geotechnique 25, No.1, p.3-17.
Obermeier, S.F., (1998) “Liquefaction evidence for strong
earthquakes of Holocene and latest Pleistocene ages in the
states of Indiana and Illinois, USA”, Engineering Geology 50
(3 – 4), 227– 254.
Schneider, J.A., Mayne, P.W., Rix, G.J., (2001) “Geotechnical site
characterization in the greater Memphis area using cone
penetration tests”, Engineering Geology 62 (1– 3), 169–184.
Tuttle, M.P., Sims, J.D., Dyer-Williams, K., Lafferty, III, R.H.,
Schweig, III, E.S., (2000) “Dating of liquefaction features in
the New Madrid seismic zone”, Technical Report, US
Nuclear Regulatory Commission NUREG/GR-0018, 78 pp.
Youd, T.L., (1991) “Mapping of earthquake-induced liquefaction
for seismic zonation”, Proc., Fourth International Conference
on Seismic Zonation, vol. 1, pp. 111 –147.
Youd, T.L., Idriss, I.M., Andrus, R.D., Arango, I., Castro, G.,
Christian, J.T., Dobry R., Finn, W.D.L, Harder, L.F., Hynes,
M.E., Ishihara, K., Koester, J.P., Liao S.S.C., Marcuson,
W.F., Martin, G.R., Mitchell, J.K., Moriwaki, Y., Power,
M.S., Robertson, P.K., Seed, R.B. and Stokoe, K.H., (2001)
“Liquefaction Resistance of Soils: Summary Report from the
1996 NCEER and 1998 NCEER/NSF Workshops on
Figure 19 Improvement ratio of dynamic compaction works Evaluation of Liquefaction Resistance of Soils.”
328
Field Identification of Active Fault nearby the Footprint of the Dam

D. Djarwadi1, D.H. Natawidjaja2 and M.R. Daryono2
1
PT Pamapersada Nusantara, , Jakarta, Indonesia
2
Research Center for Geotechnology, Indonesia Institute of Science, Bandung, Indonesia
E-mail: didiek.djarwadi@pamapersada.com
ABSTRACT: Tigadihaji dam located in the mountain range and have only 18 km from the Komering section of the Sumatra Fault Zone,
which known have a high seismic activity, and has already known as an earthquake sources. The seismic hazard assessments reveal that 2
faults located on the downstream of the dam as secondary faults which parallel to the Sumatra Fault Zone. The Pauh and Sukarena faults
named after the nearby villages have a distance of 1.85 and 8 km from the dam site respectively. Since the faults were closed to the dam, a
further detail investigation was carried to reveal the level of activity of the Pauh fault. An electrical resistivity tomography (geo-electric)
method and paleoseismic trenching were carried out to confirm the level of activity. The carbon dating test was also carried out from the
samples taken from the fault zone at the depth of 2,00 m from the ground surface. The result indicated that the Pauh fault considered as an
active fault with low sliprate, and shall be account in the seismic hazard analysis of the dam.
Keywords: Active fault, electrical resistivity tomography, paleoseismic trenching, carbon dating test.
1. INTRODUCTION The substantial portion of the dextral component of the Sumatran

oblique convergence is accommodated by SFZ. At its northern
Tigadihaji dam was a rock fill dam which will be built on Selabung terminus, the SFZ transforms into the spreading centers of the
river, in the South Sumatra province of Indonesia. The dam was Andaman Sea (Curray, 2005). At its southern end, around the Sunda
located only 18 km from the Komering segment of the Sumatra Strait, the fault curves southward toward and possibly intersects the
Fault Zone, which known have a high seismic avtivity, and has Sunda trench (Hutchon & Pichon, 1984., Diament et al, 1992).
already known as an earthquake sources. The seismic hazard Figure 1 shows the complexity of Sumatra fault zone.
assessment reveal that 2 faults located on the downstream of the
dam as secondary faults which parallel to the Sumatra Fault Zone.
The Pauh and Sukarena faults named after the nearby villages have
a distance of 1.85 and 8 km from the dam site respectively.
A study on Sumatra Fault Zone as a major source of earthquake
as well as the slip rate were carried out in order to indentify the
possibility of earthquake magnitude. Also the Komering segment of
Sumatra Fault Zone as the nearest earthquake source were studied.
Two major earthquake on 1933 and 1994 were recorded occured at
Komering segment of Sumatra Fault Zone.
Earthquakes have an effects to the dam such as vibration to the
dam body, appurtenant structures, dam foundation and rock fall on
the reservoir slopes. Fault movement in the dam foundation may
causing structural deformation and distortion, while fault movement Figure 1. Neotectonics of the Sumatran oblique convergent showing
in the reservoir may caused seiches in the reservoir and loss of major active tectonic elements
freeboard.
Dams shall be designed to resist the maximum credible Convergence along Sumatra is highly oblique where tectonics
earthquake (MCE) which defined as the largest reasonably strain is strongly partitioned into a dip-slip on the subduction
conceiable earthquake magnitude that is considered possible along a interface or the megathrust, and a dextral slip component on the
recognized fault systems or within a geograpically defined tectonic Sumatran fault zone that bisect the Sumatra island (Fitch, 1972.,
province, under the presenly known or presumed tectonic McCaffrey, 1991). With this high relative plate motions and all
framework (ICOLD, 2016). active fault arrays accomodating the movements, Sumatra ranks is
Seismic hazards assessments is compulsory for dam considered as one of the most active seismic region on Earth. In the
construction, particularly if the proposed dam is located on and past 12 years seismic activities in Sumatra is unussually highten,
around the known active fault zones. Detailed investigations of both on the megathrust and along the Sumatran fault zone.
active faults occurrences incuding their exact locations, level of The Sumatran fault zone (SFZ) has been mapped using mostly
activities, earthquake history shall be conducted carrefully since the the 1:100,000-scale aerial photographs and the 1:50,000-scale
active faults pose three kind of hazards to the dam constructions, topographic maps (Sieh & Natawidjaja, 2000). The results of their
there are: strong-high frequency ground shaking, direct ground works are digitized and put into a GIS database. Sieh and
deformations related to fault movements during earthquake events Natawidjaja’s Sumatran fault map has a large-enough scale to
and earthquake triggered landslide and liquefaction. enable discrimination of fault segmentations, so it can be used for
In the vicinity of Tigadihaji dam, detailed geology and active seismic hazard evaluation. The SFZ is highly segmented, and
fault studies have been conducted. This includes mapping of the consists of 20 major geometrically defined segments, which varies
lithology and suspected active fault strands using high-resolution in length from about 35 to 200 km, separated by major fault
digital topography and latest GIS techniques, documenting discontinuities, mostly dilational jogs, which are expressed as
geological outcrop data, shallow geophysical imaging using valleys and lakes. These segment lengths influenced seismic source
resistivity method, and conducting paleoseismological trenching on dimensions and have limited the magnitudes of large historical fault
suspected active fault strand. ruptures to between Mw 6.5 and about Mw 7.7.
In the past two years, based on current availability of modern
2. SUMATRA FAULT ZONE digital elevation maps of SRTM-30m and TERRASAR World-30m
data, coupled with high-resolution satellite images, Natawidjaja
The 1900-km long Sumatran fault zone (SFZ) traverses the back- relocated previously identified active fault strands and identified
bone of Sumatra, within or near the active volcanic arc (Sieh & new fault strands to updated the Sieh and Natawidjaja’s map. New
Natawidjaja, 2000., Katili & Hehuwat, 1967., Bellier et al, 1997). fault strands are particularly mapped in the northern and southern
329
parts of SFZ where it runs toward the Andaman spreading center elevation map (DEM) and then processing using GIS (Geographic
and the Sunda Strait grabens respectively. Information System) software to obtain detail topographic map
The SFZ map that has been significantly revised is In the which may shows the geologic structures and suspected fault
northern and the southern parts (Natawidjaja, 2017). In the northern location. Tigadihaji dam was built on the Gumai and Air Benakat
part of Sumatra, Renun – Tripa segments of SFZ runs along the west geologic formation which covered by thick Ranau tuff as product of
side of Lake Toba toward north and make a compressive bend mega eruption of ancient Ranau volcano, which at present left
before it merges with the Batee fault that ascending from the west Ranau lake as their huge crater. Figure 2 shows the distribution of
coast line. Previously, in Sieh and Natawidjaja (2000) as well as in Ranau tuff as product of mega eruption of ancient Ranau volcano
the 2010 Indonesian PSHA map, the Batee fault was not taken into (Natawidjaja, 2017).
account since it was previously considered not active. Recent data
and analysis indicates vice versa (Natawidjaja et al., 2017).
2.1 Slip Rate along the Sumatra Fault Zone

The geological sliprates of the Sumatran fault at three locations have
been measured (Sieh & Natawidjaja, 2000). From the southern part,
at about 3.6 S on the western flank of Kaba Volcano, they measured
~ 600 meters offset of the Musi River that flows on the ~60,000
years-old lava. The geological sliprates at this latitude, then, is ~10
mm/year. Near the Equator, Sianok Rivers was offset about 720
meters along the fault line that cuts the thick, ~60,000-year-old
Maninjau tuff deposits. The geological sliprates here then is about
11 mm/year. On about 2N latitude, Several streams of the Renun
River are offset dextrally about 2 km along the Sumatran fault
segment that cuts the 70,000 years-old Toba tuff on the western side
of the Toba Lake. They use this evidence to determine sliprate at
this location, which is about 27 mm/year. These geological sliprates
have been published in (Sieh et al, 1991., Sieh et al, 1994., and
Natawidjaja & Triyoso, 2007).
Bellier and Sebrier (1995) also estimated sliprates at few Figure 2. Distribution of Ranau tuff surrounding the present Ranau
locations using the principle that offset stream's age is linearly lake (Natawidjaja, 2017)
related to its length upstream from the fault. Then they calibrate this
relationship with one location (i.e. Toba) that has good offsets and The Gumai and Air Benakat formations were tertiary sediment
known age of underlying geological unit, but these values are less rocks which composed of massive of shale-limestone interbedded
reliable. with tuff-sandstone and sandy limestone. Outcrop of this tertiary
Genrich et al (2000) and Prawirodirdjo et al (2000) determined rocks found along the Selabung riverbed. Pauh fault characterized
the fault sliprates at several locations from their GPS survey-mode by the straightness of the hills that extends towards the northwest
study. Their GPS sliprate values at 2.7°N, 2.2°N, 1.3°N, 0.6°N, southeast and becomes the limit of surface texture morphology
0.4°S, 0.8°S are 26 ± 2 mm/yr, 24 ± 1 mm/yr, 24 ± 2 mm/yr, 23 ± 2 changes that express two different rock formations. Ranau tuff laid
mm/yr, 23 ± 3 mm/yr, 23 ± 5 mm/yr. The GPS sliprate at latitude unconformity above the tertiary rock formation of Gumai and Air
~2°N is about similar to the geological slip; but at latitude ~0.8°S in Benakat. Figure 3 shows the rock formation in the surroundings the
Bukit Tinggi area, the slip rate deducted from the GPS Tigadihaji dam site. At the northeast of Ranau lake, two secondary
measurements is about twice as much as that of geological sliprate. fault lines runs parallel to the Kumering fault we obseved, called
Further study is required to resolve this discrepancy. here as the Sukarena and Pauh faults. These fault line were visible
Bradley et al (2016) reevaluate slip-rates along SFZ based on as straight morphological lineaments by inspecting SRTM 30.
multiple offsets of isochronous streams incising Toba tuffs in North
Sumatra and Maninjau Tuffs in central/west Sumatra. The results of
their measurements yielded that geological slip-rates near Lake
Toba and Lake Maninjau are about similar values, 14 – 15 mm/yr.
In other word, slip-rate at Sianok valley increase a bit higher, and
slip-rate near Toba becomes only half as much as that of the
previous study. In south Sumatra, Natawidjaja (2017) conducted a
new measurement on geological sliprate of the Kumering segment
south of Ranau Lake. Their analysis yields about 8 to 12 mm/year,
thus twice as much as that of the previous measurement (Bellier and
Sebrier (1995).
Bradley et al (2016) also conducted robust geodetic modelling
and sliprate analysis for Sumatra region based on continued GPS
data from SuGAr and all available campaign-GPS data as well as
coral uplift rate data. Their detail sliprate analysis at Toba and
Maninjau area yields geodetic sliprates of ~15 mm/yr. Hence, these Figure 3. Unconformity on the boundary between Gumai rock
new geodetic sliprates are similar with new geologic sliprates. formation and Ranau tuff
3. SEISMICITY CONDITION OF TIGADIHAJI DAM During field survey, we confirmed that these faults are indeed
presence, marked by topographical breaks, fault-water falls and
Geological regional survey was carried out within radius of 50 km, lineaments of scarps and streams. We also conducted GPR surveys
to include the extent of the Pauh and Sukarena faults, while detail two ascertain the locations of the faults. Similar to the Ranau fault,
geological survey was carried out surroundings the Tigadihaji dam however, the geomorphological evidences of fault movements have
site in order to identify the fault strands and the activity of the been subdued and unclear. Luckily, on the main road between
Sukarena and Pauh faults. The basic data in this survey was digital Ranau lake and Muaradua town, we found a fresh cut of road cliffs
330
exposing the southern end of Sukarena fault outcrop as shown in Southeast of Ranau lake, Kumering segment of Sumatra Fault
Figure 4 (Natawidjaja et al, 2017). Zone straightly runs through the thick, flat-top Ranau-tuff filling
Liwa basin, marked by linear alignment of Way Robok deep
canyons and major deflections of major river canyons crossing the
fault line. Five major river canyons are right-laterally offset about
the same amount along the single major traces of Kumering fault
indicating they are formed isochronously or nearly so. Three of the
displaced canyons are across the fault at right angle, thus give
excellent measures for fault offset of about 320 to 380 meters, as
shown in Figure 5. Uncertainty of the amount offset are
appoximated by the width of displaced river canyons. We do not
include the other two offsets since the river cut at low angle to the
fault, so they yield less accurate measure of the offsets. The average
of these three river offsets is 350 ± 50 meter, thus yields a sliprate of
10.4 ± 2.4 mm/year for the Kumering fault segment.
Figure 4. Road cut exposure between Muaradua town to Ranau Lake

exposed the Sukarena fault (Natawidjaja et al, 2017)
The outcrop shows the dipping of the tertiary sedimentary rocks

that is unconformably overlied by the Ranau tuff. The fault clearly
cut the underlying sedimentary rocks, but ends at the Ranau tuff’s
basal contact. Hence, it clearly proves that the fault has been
inactive prior to the Ranau tuff emplacement, 31,000 years. We also
observed thin paleosoils on the upperpart of the tertiary rocks,
burried by the tuffs, but we do not conduct radiometric dating Figure 5. Consistent right lateral offsets of isochronous river
analysis here since the the paleosoils do not seem to have rich in channels deeply incised into Ranau tuff near Liwa. The total offset
organics and there are modern plants and roots contaminate it. is estimated from three channels that across the fault at high angle
Figure 4 shows that Sukarena fault which found on the Gumai
Formation did not cross the Ranau tuff layer which found on the 4. INVESTIGATION ACTIVE FAULTS
above the Gumai formation. Since the Ranau tuff was deposited
In this study, the geoelectric survey using SuperSting R8
around 31.000 years ago, it means that there was no movement of
multichannel equiped with 112 electrodes at 5.00 meter spacing,
the Sukarena fault for the last 31.000 years. It can be concluded that
thus every measurement have around 555 meter in length. Total
the Sukarena fault was inactive at present time.
eight (8) lines, one line across Pauh fault, and seven (7) lines on the
The remnant of the center of the paroximal eruptions of Ranau
dam area were carried out in order to investigate the presence of the
tuff is the huge caldera, known as the a rectangular-shaped Ranau
faults in dam foundations.
lake, 14 km long and 7 km wide. This rectangular-shape was
Geoelectric survey on line 1 primarily intended to scan
thought to be controlled by inactive SFZ strands that were operating
underground structures of the Pauh fault. The result of this scan will
during the development of a hyphothetical Ranau pull-apart basin,
be used to pinpoint the location of the trench excavation for
and was also suspected to have significant role in facilitating the rise
palaeoseismology study. Figure 6 shows the geoelectric result of
of magma forming giant volcano similar to Toba. The pull-apart
line 1 which across the Pauh fault.
bounding faults serve like ‘ring structures’ in ordinary volcanoes,
thus forming rectangular-shape caldera after explosion as seen
today.
Ranau tuff blanketed a wide area about 70-km radius around its
center of eruption. Clearly the voluminous Ranau ash flows reset the
previous landscapes by effectively buried the pre-existing
topography upon its emplacement. Then, it was followed by the fast
creation of a new network of drainages. In the proximal area, south
of the lake, the tuffs filled the Liwa Basin, which then has been Figure 6. Scanning result of line 1 with indication of Pauh fault
incised 40 to 80 meter deep by wide river canyons forming a
plateau-like landscape. Away from the center, the Ranau tuff bed Scanning results indicated that the decomposed top soil have a
becomes thinner and then dissapears. thickness more than 5 meter, interpreted from the resistivity less
On the cliff exposures along the main road near Muaradua town , than 200 Ohm.m. This soil layer probably young alluvial deposit
80 km northeast of Ranau Lake, 4-5 meter-thick Ranau tuff bed lies and decomposed of Ranau tuff. Below the decomposed soils, the
on the moderately dipping tertiary sedimentary rocks, nicely Ranau tuff layer with high resistivity found up to 80 meters except
showing an angular unconformity contact. Along the boundary a at the suspected Pauh fault. Below the Ranau tuff, the tertiary
few tens of cm organic-rich paleosoils had been developed on the sedimentary rock was found indicated from low resistivity which
upper surfaces of the tertiary rocks. We carefully digged and range from 2 to 100 Ohm.m. The most important from line 1 scan
selected part of this organic sample for radiocarbon dating to avoid result was the capture of the suspected Pauh fault.
possibilities of any modern-carbon contaminations from nearby Palaeoseismic trench ussually carried out to prove the existence
modern plants and roots. The AMS analysis of this sample (RAN- of a fault structure that has been identified from morphological
014) yields a conventional carbon age of 26,830 ±140 BP, or the analyses or geoelectric scanning. It can be use also to evaluate the
corrected calendar-year age of Cal BP 31085 to 30805 (95% fault activity and look for traces of past seismicity recorded in the
probability)(Natawidjaja et al, 2017). soil layer (Yeats, 1996). Palaeoseismic trench was a method that has
widely used in the earthquake research. This method became
331
popular following the successfully study of palaeoseismology in Geological observation, geo-electric survey and data evaluation
Pallet Creek, California (Sieh et al, 1989., Sieh, K , 1977). on the paleoseismic trenching as shown on Figure 8, indicated that
Palaeoseismic trench was carried out on the approximate the Pauh fault was still active, even there was not clearly exposed on the
Pauh fault line. Trenching was made over the line 1 of the ground surface. The Ranau tuff layer as a product of mega eruption
geoelectric survey, cut the Pauh fault perpendicularly. The above the Gumai formation indicated a shifting in vertical direction
dimension of the trench are; 20 meter in length, 1.50 in width and (left lateral normal fault). Carbon dating results on the soil samples
2.00 m in depth. Logging on the both (left and right) walls were taken from the soils on the fault location indicated that the age of the
made using different colours to identify the soil layers and fault soils was 114 to 249 yaears. This indicated as the last earthquake
structures. Logging were made to identify colour, texture, type or resulting from the Pauh fault, even there was no historical record
nomenclatures of soils. Trenching in Pauh faults have five (5) was found. The calculation on the Mmax based on the length of the
different soil layers and described as follows: fault based on several GMPE (ground motion prediction equation)
Layer 100 was the first soil layer, identified by black colour was resulting the possibility Mmax 6,4 earthquake may occured from the
top soil consists of sandy silt soils, dark brown in colour, many roots Pauh fault. Based on this result, the earthquake which may occured
found. from Pauh fault shall be included in the Seismic Hazard Analysis for
Layer 200 was the second layer, gravelly sandy silt with some Tigadihaji dam.
lithic fragment up to 4 cm, brown in colour. This founding will change the Seismic Hazard Analysis which
Layer 300 was the third layer, coarse sand from Ranau tuff, with has been made previously. In the previous Seismic Hazard Analysis,
some lithic fragment up to 2 cm, yellowish brown in colour. the Pauh fault was considered as an inactive fault, and the Seismic
Layer 400 was the fourth layer, coarse sand from Ranau tuff with Hazard Analysis were made based on the Probabilistic Seismic
some sand layers interbedded, greyish brown. Hazard Analysis (PSHA). But considering that Pauh fault was an
Layer 500 was the bottom layer, coarse sand from Ranau tuff active fault and the distance to the proposed Tigadihaji dam was
with some sand layers interbedded, grey colour. very closed, and also the length of Pauh fault has already known, the
On the trench walls there is an indication of fracture and fault new Seismic Hazard Analysis was made based on the Deterministic
plane that shear the soil layers with the parallel direction of the Pauh Seismic Hazard Analysis (DSHA).
fault. Although the fault structure did not appear on the surface,
indication of Pauh fault found at least on the soil layer 400 that 5. CONCLUSION
shear the soil layer, so that the existence of possible ttectonic
fracture structures deforming this soil layer is classified as an active The identification of active fault nearby the foot print of the
fault. Tigadihaji has been presented. The presence of the Sumatra Fault
Figure 7 shows the logging of the east and west walls of the Zone, the sliprates along the fault and paleoseismic on the Pauh fault
palaeoseismic trenching, while Figure 8 indicated the Pauh fault in has been carried out to study the condition of the fault. The
the palaeoseismic trenching wall. conclusions of the investigation are summarized as follows;
i. The Tigadihaji dam location was covered by Ranau tuff as a
product of mega eruption of Ranau volcano which have an
age of around 31.000 years,
ii. Pauh fault which located at 1.85 km in the downstream of
the dam with 14 km in length shall be accounted as an
active fault, based on the finding from paleoseismic
trenching
Logging of the left wall iii. The Seismic Hazard Analysis for Tigadihaji dam shall
included the Pauh fault in the analysis.
iv. There was a change in Seismic Hazard analysis for
proposed Tigadihaji dam, where before Pauh fault was
consideren as inactive fault, the Seismic Hazard Analysis
was made base on the probabilistic (PSHA), while after the
Pauh fault was considered as an active fault, the Seismic
Logging of the right wall Hazard Analysis was re-calculated based on Deterministic
(DSHA).
Figure 7. Logging of the palaeoseismic trenching walls
6. REFERENCES
Bellier, O. and Sebrier, M. (1995) Is the slip rate variation on the
Great Sumatran fault accommodated by fore-arc stretching?
Geophys.Res.Lett. Vol.22, No.15. pp1969-1972.
Bellier, O., Sebrier, M., Pramumijoyo, S and Forni, O. (1997)
Paleoseismicity and seismic hazard along the Great Sumatran
fault (Indonesia). Journal of Geodynamics. Vol.24, Nos 1-4,
pp169-183.
Curray, J. (2005) Tectonics and history of the Andaman Sea region.
Journal of Asian Earth Sciences.
Diament, M., Harjono, H., Karta, K., Deplus, C., Dahrin, D., Zen,
M. T., Gerard, M., Lassal, O., Martini, A., and MAlod, J.
(1992) Mentawai fault zone off sumatra - a new key to the
geodynamics of western Indonesia. Geology. Vol.20, No3.
pp259-262.
Fitch, T. (1972) Plate convergence, transcurrent faults, and internal
deformation adjacent to southeast Asia and the western
Pacific. Journal of Geophysical Research. No.77. pp4432-
Figure 8. Indication of Pauh fault on the palaeoseismic trench wall 4462.
332
Genrich, J.F., Bock, Y., McCaffrey, R., Prawirodirdjo, L., Steven, Natawidjaja, D.H., Bradley, K., Daryono, M.R., Aribowo, S. and
C.W., Puntodewo, S.S.O., Subarya, C and Wdowinski, S. Herrin, J. (2017) Late Quarternary eruption of the Ranau
(2000) Distribution of slip at the northern Sumatran fault Caldera and new geological slip rates of the Sumatran Fault
system. Journal of Geophsical Research. Vol.105(B12): p. Zone in Southern Sumatra, Indonesia. Geoscience Letter. pp
28,327-28,341. 4-21.
Huchon, P., and Pichon, X.L. (1984) Sunda Strait and Central Prawirodirdjo, L., Bock, Y., Genrich, J.F., Puntodewo, S.S.O., Rais,
Sumatra fault. Geology. Vol.12. pp668-672. J., Subarya, C., and Sutisna, S. (2000) One century of tectonic
International Commission on Large Dams (ICOLD). (2016) deformation along the Sumatran fault from trianggulation and
Selecting Seismic Parameters for Large Dams. Bulletin no. Global Positioning System survey. Journal of Geophysical
148. 2016. p75. Research. Vol.105(B12): p. 28,295-326.
Katili, J.A. and F. Hehuwat. (1967) On the occurence of large Sieh, K., (1977) Late Holocene displacement history along the
transcurrent faults in Sumatra, Indonesia. Journal of south-central reach of the San Andreas fault.. Stanford
Geoscience, Osaka City University. Vol.10, pp 5-17. University, Stanford, California.
McCaffrey, R. (1991) Slip vectors and stretching of the Sumatran Sieh, K., Stuiver, M and Brillinger, D. (1989) A more precise
fore arc. Geology. Vol.19. pp881-884. chronology of earthquakes produced by the San Andreas fault
Natawidjaja, D.H. and Triyoso, W. (2007) The Sumatran Fault in southern California. Journal of Geophsical Research.
Zone: From source to hazards. Journal of Earthquake and Vol.94. pp603-623.
Tsunami, Vol.1. Sieh, K., Bock, Y and Rais, J. (1991) Neotectonic and paleoseismic
Natawidjaja, D.H. (2017) Updating active fault map and slip rates studies in West and North Sumatra. AGU 1991 Fall Meeting
along the Sumatran Fault Zone, Indonesia. in Global Program & Abstacts. 72(44). p460.
Qolloqium on Geoscience and Engineering LIPI 2017. Sieh, K., Zachariasen, J., Bock, Y., Edwards, L., Taylor, F and
Jayakarta Hotel, Bandung: Puslit Geoteknologi LIPI. Gans, P. (1994) Active tectonics of Sumatra. Geo. Soc. of
Natawidjaja, D.H., Sapiie, B., Daryono, M.R., Gayatri, I.M., Astyka, Amer. Bull. Vol.26. p A-382.
P., Mukti, M.M., Supartoyo, Hidayati, S and Solikhin, A. Sieh, K. and D. Natawidjaja. (2000) Neotectonics of the Sumatran
(2017) Geologi Gempa Indonesia, in Peta Sumber dan fault, Indonesia. Journal of Geophysical Research, 105(B12).
Bahaya Gempa Indonesia tahun 2017, M.Irsyam, et al., pp295-326.
(Editors). Kementrian PUPR: Bandung. p400. Yeats, R.S. (1996) Introduction to special section: Paleoseismology.
Journal of Geophsical Research. Vol.101. pp5847-5833.
333
Seismic Analysis of El–Agrem Concrete Face Rockfill Dam

M. Abdellaoui1 and B. Moussai2
1
Faculty of Civil Engineering, University of Science and Technology Houari Boumedienne,
Algiers – Algeria
E-mail: bmoussai@yahoo.ca
ABSTRACT: This paper presents the dynamic analysis of El–Agrem concrete face rockfill dam using the finite element software Plaxis-2D.
A parametric study is performed to identify the degree of influence of some parameters such as peak ground acceleration, rockfill shear
modulus and constitutive model on the dam behavior. The results indicated that these parameters have an important effect on the response of
the dam under seismic loading
Keywords: Behaviour of concrete face rock fill dam, stress - strain analysis, earthquake behaviour, performance of CFR dams.
1. INTRODUCTION The concrete slab on the upstream side provides the

watertightness of the dam above the ground together with the
The design of concrete face rockfill dams for earthquake is mainly perimetric joints and the plinth. The concrete slab has a total area of
based on experience and engineering judgement (Cooke, 1984; 31000 m2 with a variable thickness of 0.5 m at the base and 0.35 m
Cooke, 1997; Núñez, 2007b). For concrete face rockfill dams, at the top of the dam. The dam foundation consists of granite gneiss
performance rather than safety is the main concern, as it is widely rock on the right bank and of marly rock on the left bank. The dam
accepted that the effect of seismic loading on this type of dams is construction was completed in 2000.
plastic deformation and settlement but not a slope failure in its
classical sense (Cooke, 1984; Makdisi and Seed, 1978; Newmark,
1965). For a given CFRD and earthquake, total settlement of the
dam with empty reservoir is larger than the settlement of the dam
with full reservoir (Uddin and Gazetas, 1995; Núñez, 2007).
Methods used to estimate dam deformation induced by
earthquake range from simple analytical tools (Newmark,1965;
Makdisi and Seed, 1978; Núñez, 2007) to three dimensional
numerical models. Analytical tools are simple to use but they cannot
take into account special features of dam design, like zonification,
berms or non uniform slopes. However, the reliability of numerical
methods depends on the choice of the constitutive models and the
selection of input parameters (Sfriso, 2008).
Materón and Fernandez (2011) reported two major earthquakes
occurred in 2008 close to Zipingpu concrete face rockfill dam in
China (156 m high, shaken by a magnitude 8 earthquake) and F igu r e 1 a : G en er a l v ie w o f E l - Ag r e m c on cr e te fa c e r oc kf ill da m
Ishibuchi concrete face rockfill dam in Japan (53 m high, shaken by Figure 1a General view of El-Agrem concrete face rockfill dam
a magnitude 7.2 earthquake). The severe shaking triggered
displacement and cracking at the crest as well as loosening of
boulders in the downstream slope, but the performance of the dams
from the viewpoint of stability were adequate.
The objective of this paper is to investigate the behavior of El-
Agrem concrete face rockfill dam during seismic loading using the
finite element code Plaxis 2D (Brinkgreve et al., 2010). The
analyses were performed using the linear elastic perfectly plastic
model (Mohr Coulomb model). The effect of some parameters on
the response of the dam to seismic loading such as seismic
acceleration, material shear modulus and constitutive models are Figure 1b Typical cross-section of El-Agrem concrete face
also presented in this paper. rockfill dam
2. MAIN FEATURES OF EL-AGREM DAM 3. DYNAMIC ANALYSIS

The El–Agrem concrete face rockfill dam as shown in figure 1a, is The dynamic analysis was conducted on the maximum cross –
located 15 km south - east of Jijel city (Algeria). The dam reservoir section of the dam resting on rock foundation with empty reservoir
can store approximately 35 million cubic meters of water and using Plaxis 2D software. The dam was subjected to the seismic
provides drinking water to the surrounding region. excitation that was recorded during the 2003 Boumerdes earthquake
The dam is 64 m high and 395 m long with a crest width of 10 (Algeria) with a peak ground acceleration of 0.34 g in the east-west
m. The embankment volume is 1.6 x 106 m3 and consists of two direction (Fig. 2).
zones of granitic rockfill materials as shown in figure 1b. Zone E1 The embankment was modelled using fifteen noded plane strain
comprises the majority of the embankment section with a maximum elements. Figure 3 shows the finite element model having 2228
particle size of 0.6 m and compacted in layers of 0.8 m thick. Zone elements and 18217 nodes.
E2 is a 1 m thick transition zone consists of smaller-size rock, The deformation of the dam foundation is considered negligible;
provided between the concrete face slab and the rockfill therefore the model sets fixed displacements along the dam base in
embankment. The rockfill is compacted to a high density in order to contact with the rock foundation. Furthermore, the dam is located in
minimize deformations, concrete slab cracks and leakage. The a wide valley (crest length / dam height equals to 6), thus three
upstream and downstream slopes of the dam were established at dimensional effect is negligible.
1.7H:1V (Fig. 1b). The concrete slab was modelled as linear elastic with Young’s
modulus equals to 30000 MPa and Poisson ratio equals to 0.15.
334
Rockfill materials were modelled with Mohr Coulomb model. Table 4.1 Effect of constitutive model
1 lists the parameters of the dam materials used in the numerical
The effect of rockfill constitutive model on the dam response to
analysis. The dilation angle was estimated by the relation = - 30
seismic loading was investigated using Mohr Coulomb model and
(Bolton, 1986).
elastic model.
The elastic model indicated a crest settlement (2.6 mm) less than
200
that obtained with Mohr Coulomb model (6.8 mm). Moreover, the
A c c e le r a t io n [ c m / s e c 2 ]
100
acceleration at the crest obtained with elastic model (0.92g) is about
0
1.4 times the acceleration obtained with Mohr Coulomb model
-100
(0.66g). Thus the elastic model resulted in a decrease of the crest
-200
settlement by about 62% and an increase of the acceleration at the
-300
crest by about 39% compared to Mohr Coulomb model.
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
Time [sec] 4.2 Effect of shear modulus
Figure 3.21 : Accélérations du séisme de Boumerdes composante de la direction Est-Ouest. The effect of the rockfill shear modulus on the response of the dam
Figure 2 Seismic excitation record in the east-west direction
(Boumerdes 2003 earthquake, Algeria) to seismic loading was investigated using shear modulus values of
400 MPa, 450 MPa and 500 MPa. The results shown in Figure 5
indicate that the increase of the shear modulus leads to the decrease
of the settlement and increase of the acceleration at the dam crest.
Figure 3 Two dimensional finite element model
Table 1 Rockfill parameters

Layer Zone E1 Zone E2
Unit weight:  (kN/m3) 20.5 21
Cohesion: c’ (kPa) 0 0 Figure 5 Crest settlement and acceleration versus rockfill shear
Friction angle: 'o 45 42
modulus
Dilatancy angle o 15 12
Poisson’s ratio:  0.2 0.2 4.3 Effect of peak ground acceleration
Shear modulus: G (MPa) 450 450
Acceleration time histories for east-west and north-south directions
4. RESULTS AND DISCUSSION recorded during Boumerdes 2003 earthquake (Algeria) were used to
investigate the effect of peak ground acceleration on the dynamic
The results indicate that the horizontal displacement of the dam crest response of the dam. The peak ground accelerations in the east-west
is about 320 mm in the upstream-downstream direction and the and north-south directions are 0.34 g and 0.25 g respectively.
maximum settlement of the dam crest is about 7mm located on the The analysis results indicate that the increase of the ground
upstream side. The peak acceleration obtained at the crest (0.66 g) is acceleration led to the increase of the crest settlement (from 3.6 mm
about 1.94 times the acceleration at the dam base (0.34 g) (Fig. 4). to 6.8 mm) as well as the amplification of acceleration between the
These results are in accord with those obtained by many researchers base and crest of the dam. The acceleration obtained at the dam crest
such as Haciefendiglu (2009); Bayraktar et al. (2011) and Kartal et is about 1.4 times the ground acceleration in the north-south
al. (2011). direction and about 1.94 times the ground acceleration in the east-
west direction.
5 CONCLUSIONS
The dynamic analysis of El–Agrem concrete face rockfill dam using
Plaxis-2D with Mohr Coulomb model indicated that the
amplification factor of the acceleration at the dam crest is about 1.94
and this amplification increases with the increase of the peak ground
acceleration.
The elastic model resulted in a decrease of the crest settlement
and an increase of the acceleration at the dam crest compared to
Mohr Coulomb model.
The increase of the rockfill shear modulus led to the decrease of
Figure 4 Horizontal accelerations at the base and crest of the dam settlement and increase of acceleration at the dam crest.
335
REFERENCES Makdisi F., Seed B. 1978. Simplified procedure for estimating dam
Bayraktar, A., Kartal, M.E. and Adanur, S. 2011. The effect of and embankment earthquakes induced deformation.
concrete slab –rockfill interface behaviour on the earthquake ASCEJGE, 104(7), 849-867.
performance of a CFR dam. International Journal of Non- Materón, B. and Fernandez, G. 2011. Consederations on the seismic
linear Mechanics, 46, 35-46. design of high concrete face rockfill dams. The second
Bolton, M.D. 1986. The strength and dilatancy of sands. International Symposium on Rockfill Dams, Rio-de-Janeiro,
Geotechnique, vol. 36, pp. 65-78. Brazil
Brinkgreve, R.B.J.; Swolfs, W.M. & Engin, E. 2010. Plaxis 2D Newmark N. 1965. Effects of earthquakes on dams and
User’s Manual. Plaxis 2010, Netherlands, 199 p. embankments. Geotechnique 15(2), 139-160.
Bureau, G. et al. 1985. Seismic analysis of concrete facerock fill Núñez E. 2007. Behavior of coarse alluvium slopes subjected to
dams”, Proc. Of Symp. On rockfill concrete face dams – earthquakes. Proc. XIII PCSMGE, Margarita (Venezuela),
design, construction and performance. ASCE, New York., 862-867.
pp. 479 – 508 Sfriso, A. O., 2008. Numerical assessment of the deformation of
Cooke J. 1984. Progress in rockfill dams. ASCE, JGE, 110(10), CFR dams during earthquakes. The 12th International
1381-1414. conference of international association for computer methods
Cooke J. 1997. The concrete face rockfill dam. Proc. 17 USCOLD and advances in geomechanics. Goa, India., pp. 4054 – 4061.
Lect., San Diego (USA), 117-132. Sherard, J. L., and Cooke, J. B., 1987. Concrete face rockfill dam:
Haciefendiglu , K. 2009. Stochastic response of concrete faced strong seismic excitation. Journal of Geotechnical
rockfill dams including partially ice-covered reservoir – Engineering, ASCE, 121(2). 185 – 197
foundation interaction undr spatially varying seismic waves. Uddin, N. and Gazetas, G. 1995. Dynamic response of concrete
Cold regions Science and Technology, 58, 57-67. faced rockfill dams to strong seismic excitation. Journal of
Kartal, M.E., Bassaga, H.B. and Bayraktar, A. 2010. Probabilistic Geotechnical Engineering, ASCE, 121(2). 185 – 197.
nonlinear analysis of CFR dams by MCS using Response
Surface Method. Applied Mathematical Modelling, 35, pp.
2752-2770.
336
Effects of Vibrating Frequency of a Plate Compactor on Soil Density

Y. S. Fang1, W. H. Lin2 and C. Liu1
1
Department of Civil Engineering, National Chiao Tung University, Hsinchu, Taiwan
2
Facility and Engineering Division II, InnoLux Corporation, Tainan, Taiwan
E-mail: ysfang@mail.nctu.edu.tw
ABSTRACT: This paper presents experimental data on the change of density in a cohesionless fill due to vibratory compaction at different
vibration frequencies. A variable-frequency plate compactor with a frequency range of 10 ~ 100 Hz was designed and constructed. Air-dry
Ottawa sand was used as fill material. The initial relative density of the uncompacted fill was about 33 %. Vibratory compaction was applied
on the surface of a 1.5 m-thick lift. The cyclic normal stress applied on the surface of the fill was kept a constant (cyc = 30 kN/m2) under the
compaction frequencies of 45, 58, 64, 74, 83, 89 and 100 Hz. Test results indicated that within the depth of 0.3 m, which was the width of the
compaction plate, the density increase was most significant. The density increase effect reduced with increasing depth. The effective
compaction depth varied from 0.43 to 0.54 m, which was deeper than the maximum permissible lift thickness 0.15 ~ 0.30 m suggested by
NAVFAC DM-7.2 (1982). Under the same amplitude of cyclic normal stress, both the peak relative density of the fill and the effective
compaction depth were not significantly influenced by the applied compaction frequency.
Keywords: Compaction, Frequency, Relative density, Plate compactor, Sand.
1. INTRODUCTION
Model Wall
In the construction of highway embankments, earth dams, and many
0
other engineering structures, engineers are often required to compact 150
45
loose soils to increase their densities. The purpose of compaction
operation is to improve the engineering properties of soil such as Sidewall
increasing shear strength, reducing permeability and 15
Fo 00
compressibility. Various soil improvement techniques had been used End
otb
oar
d
in the past. Wall
Vibration is especially effective to compact granular soils.
1600
Typical vibrational methods included hand temper compaction,
20
roller compaction, vibroflotation, dynamic compaction, and
compaction sand pile. However, in the literature, there were few
research regarding the relationship between vibration frequency and
soil density change. Limited laboratory and field studies were made,
and no explicit conclusion was drawn. The objective of this study
was to investigate how does the vibratory frequency affects the Steel Column
Steel Base Plate
compaction results.
Baidya and Krishna (2001) conducted experiments to investigate
the effects of loading frequency and amplitude of a vibrating footing
on a layered soil system. It was concluded that, the frequency Unit : mm
corresponding to maximum surface displacement amplitude from Figure 1 NCTU non-yielding model retaining wall and soil bin
the response curve was considered as resonant frequency. For the (after Chen and Fang, 2008)
layered soil system, the observed natural frequencies varied from
8.3 to 31.3 Hz. Wersäll and Larsson (2013) conducted small-scale
tests to investigate the influence of loading frequency on
compaction effects. For the test with large-amplitude the resonant
frequency was about 48 Hz, and for small-amplitude tests the
resonant frequency was about 68 Hz.
Handle
2. EXPERIMENTAL FACILITY
2.1 Soil Bin
The soil bin was designed to minimize the lateral deformation of
soil specimen during testing. In Figure 1, the soil bin was fabricated
of steel plates with inside dimensions of 1,500 mm × 1,500 mm ×
1,600 mm. In Figure 1 the model wall is 45 mm-thick, and the
sidewalls are 35 mm-thick. To achieve an at-rest condition, the wall
material should be nearly rigid.
In Figure 1, twenty-four 20 mm-thick steel columns were
welded to the four sidewalls to reduce any lateral deformation Eccentric
during loading. In addition, twelve C-shaped steel beams were also Motor
welded horizontally around the box to further increase the stiffness

Steel Plate
of the box. For more information regarding the soil bin, the reader is
referred to Chen and Fang (2008).
2.2 Variable-Frequency Plate Compactor

This section introduces the design of the NCTU variable-frequency Figure 2 NCTU vibrating plate compactor (after Lin, 2015)
hand tamper used in this study (Lin, 2015). To design the
337
Q(t) was produced by the eccentric motor with rotating mass. The
force amplitude applied was:
Qo = me e 2 (2)
where me was the eccentric mass, e was the eccentricity, and  was
the angular frequency. The amplitude of cyclic normal stress applied
from the plate compactor to the fill was cyc = Qo/A, where A was
the area of the loading plate. To keep the cyclic normal stress a
constant with increasing frequency, the rotating mass was decreased
by remove the rotating eccentric plate attached to the central axis of
the motor shown in Figure 4.
(a) Frequency of vibratory compaction vs. static normal stress
Figure 4 Measurement of revolution speed of rotor plate with photo

tachometer
(b) Frequency of vibratory compaction vs. cyclic normal stress

Figure 3 Distribution of vibration frequency for commercial
vibratory tampers
experimental compactor, 83 commercial compactors on the market

were collected and investigated. Four main factors considered for
the design of the experimental hand tamper were: (1) compacting (a)
plate area; (2) applied static pressure; (3) cyclic normal stress; and
(4) vibration frequency. Figure 2 shows the vibrating plate
compactor used for all experiments in this paper. The distribution of
vibration frequency as a function of static and cyclic normal stresses
for the 83 commercial hand tampers are illustrated in Figure 3 (a)
and (b).
For this study, since the area of the soil bin for testing was only
1.5 m × 1.5 m, therefore a small plate size 0.30 m × 0.30 m was
selected. The area of the compaction plate was 0.09 m2. The mass of
the compactor was 47 kg, and the applied static stress was 5.12
kN/m2.
2.3 Constant cyclic normal Stress

The amplitude of cyclic normal stress was kept to be σcyc = 30
kN/m2 for all experiments in this investigation. The cyclic normal
load applied could be described as follows:
(b)
Q(t) = Qo sin t (1) Figure 5 Soil density molds buried at different elevations
338
3. BACKFILL CHARACTERISTICS 4.2 Density Distributions at Different Vibration Frequencies

3.1 Specimen Preparation Figure 9 summarized the relative density distributions after 2 passes
with the compaction frequencies of 45, 58, 64, 74, 83, 89 and 100
Air-dry Ottawa sand (ASTM C-778) was used throughout this Hz. Within the depth of 0.3 m, which was the width of the
investigation. Physical properties of the soil include Gs = 2.65, emax compaction plate, the density increase was most significant. The
= 0.76, emin = 0.50, D60 = 0.39 mm, and D10 = 0.26 mm. To observe density increase effect was reduced with increasing depth.
the distribution of soil density in the bin, soil density molds were Figure 10 shows the peak relative densities Dr,peak induced after
made. During the preparation of the 1.5 m x 1.5 m x 1.5 m soil 2 passes of the tamper at different compaction frequencies. The
specimen, density cups were buried in the soil mass at different relative density for the uncompacted fill was 33.3 1.8 %. After
elevations and different locations in the fill as indicated in Figure 5. compaction with the cyclic normal stress cyc = 30 kN/m2, the peak
After the soil had been filled up to 1.5 m from the bottom of the soil relative densities increased to 114 ~ 119 %. However, test data
bin by air-pluviation, density molds were dug out from the soil mass indicated that the peak density of the fill was not significantly
and soil density was measured carefully. affected by compaction frequency.
3.2 Testing Procedure 4.3 Effective Compaction Depth
Figure 5 shows the soil density cups were placed at spread locations Figure 11 illustrates the effective compaction depths obtained at
to avoid hindering the propagation of vibratory waves to reach the different compaction frequencies. In the figure, the effective
lower-elevation locations. For a 1.5 m-thick air-pluviated Ottawa compaction depths varied from 0.43 to 0.54 m, which was deeper
sand layer, the distribution of soil density with depth is shown in than the maximum permissible lift thickness 0.15 ~ 0.30 m (6 ~ 12
Figure 6. For the sand specimen, the mean unit weight is 15.5 inch) suggested by the US Navy Design Manual 7.2 (1982). From
kN/m3, the mean relative density is Dr = 33.3% with the standard these experimental data, it may be concluded that the effective
deviation of 1.8%. Das (2010) suggested that for the granular soil compaction depth was not significantly influenced by the
deposit with a relative density between 15% and 50% is defined as compaction frequency.
loose sand. Figure 6 shows the relative density Dr = 33.3 %
achieved by the air-pluviation method was quite loose and uniform N
with depth.
The fill surface was divided into five lanes as indicated in Figure
7. Each 0.3 m-wide 1.5 m-long lane was compacted with the
vibratory compactor for 35 seconds for each pass. For the travel
distance of 1.2 m, the speed of the compactor movement was 34 Compaction Lane 1
mm/s. Path
Lane 2
4. DENSITY INCREASE DUE TO VIBRATORY
COMPACTION
Lane 3
4.1 Density Distributions at Low Vibration Frequency
Figure 8 shows the density distribution with depth after two passes Lane 4
of the vibratory plate compactor at the loading frequency of 45 Hz.
In the figure, for the uncompacted fill, the relative density was 33.3 Lane 5
 1.8 %. After 2 passes of the compactor with the rotating rotor
frequency of 45 Hz, the relative density increased up to the peak
relative density of about 116 %. For the upper 0.53 m of fill, the
relative density of the compacted fill was greater than the Dr = 75 % Steel Base Plate Steel Beam Steel Column
which was required by the US Navy Design Manual 7.2 (1982).
Top-View
Figure 7 Compaction path on fill surface (after Lin, 2015)
Figure 6 Distribution of relative density of soil with depth

Figure 8 Distribution of relative density with depth at frr = 45 Hz
339
5. CONCLUSIONS
For the cohesionless fill improved with the compaction frequencies
of 45, 58, 64, 74, 83, 89 and 100 Hz, the density increase was most
significant within the depth of 0.3 m. The density increase effect
was reduced with increasing depth.
The measured peak relative densities varied from 114 to 119 %.
The peak relative density of the fill was not significantly affected by
the compaction frequency.
The effective compaction depths varied from 0.43 to 0.54 m,
which was deeper than the maximum permissible lift thickness 0.15
~ 0.30 m suggested by NAVFAC DM-7.2 (1982).
Under the same amplitude of cyclic normal stress, both the peak
relative density of the fill, and the effective compaction depth were
not significantly influenced by the applied compaction frequency.
6. REFERENCES
Baidya, D. K. and Krishna, G. M. (2001). “Investigation of resonant
frequency and amplitude of vibrating footing resting on a
layered soil system.” Geotechnical Testing Journal, 24(4),
Figure 9 Distribution of relative density with depth at different 409-417.
compaction frequencies Chen, T. J. and Fang, Y. S. (2008). “Earth pressure due to vibratory
compaction.” Journal of Geotechnical and Geoenvironmental
Engineering, 134 (4), 437-444.
Das, B. M. (2010). “Principles of geotechnical engineering.” 7th Ed.,
Cengage Learning, Stamford, CT, USA.
Lin, W. H., (2015). “Effects of Compactor Vibration Frequency on
Ground Settlement and Density Change of Sandy Soil,”
Peak Relative Density, Dr, peak (%)
Master of Engineering Thesis. National Chiao Tung

University, Hsinchu, Taiwan.
US NAVY. (1982). “Foundations and earth structures.” NAVFAC
Design Manual DM-7.2. Naval Facilities Engineering
Command, U.S. Government Printing Office, Washington, D.
C.
Wersäll, C. and Larsson, S. (2013). “Small-scale testing of
frequency-dependent compaction of using a vertically
vibrating plate.” Geotechnical Testing Journal, 36(3), 1-10.
Figure 10 Peak relative densities at different compaction frequencies
Figure 11 Variation of effective compaction depth with compaction

frequency
340
Using Time Domain Reflectometry for Monitoring Slope Movement

in the Jiufenershan Landslide
Shei-Chen Ho 1, I-Hui Chen1, Yu-Shu Lin 1, Jun-Yang Chen 1, Miau-Bin Su 1
1
ABSTRACT: This paper applied time domain reflectometry (TDR) technology to monitor slope movement of a landslide. The TDR system
was installed from 2012 to 2017 in order to monitor slope movement in the Jiufenershan landslide. TDR waveform signature showed a cable
ruptured by extension at 7.51 m and 35.72 m in depth of TDR_N2 station between 2012 and 2013. Compared to the inclinometer data of the
area only appears some sliding phenomena in a range of 0 m and 8 m in depth. Overall, the TDR technology shows that the locations of
failure planes and the magnitude of the displacements can be determined by the changes and quantification of TDR reflected waveforms in
the landslide.
Keywords: Time Domain Reflectometry, slope movement, landslide monitoring.
1. INTRODUCTION and the magnitude of a cable deformation (Su and Chen, 2000).
Furthermore, whether shearing deformations are big or small, the
This paper uses Time Domain Reflectometry (TDR) technology to magnitude of TDR reflected waveforms is significant correlation
monitor slope movement of a landslide and compares other on-site with the deformation (Drusa and Bulko, 2016). They point out that
measures such as inclinometers to TDR monitoring data. there is a correlation between TDR reflection coefficient and shear
It is the most important for monitoring slope deformation to deformation in a laboratory test so that a regression equation of
determine the location of the sliding surface and the magnitude of them can be determined.
displacements in a landslide. However, traditional monitoring The paper analyzes the waveforms interrogated by TDR
equipment, such as an inclinometer, is time-consuming and difficult monitoring stations in a landslide area and compares to the data of
to interpret an accurate location of sliding surface (Dowding and on-site inclinometers in order to understand the location, type and
Pierce, 1994; Dowding et al., 2003). Thus, TDR technology is a new deformation magnitude of slope movement in the field.
method of slope movement monitoring which is a continuous sensor
to detect any deformation along its cable length when sliding
2. JIUFENERSHAN LANDSLIDE
deformation occurs (O’Connor and Dowding, 1999). They argue
that TDR has been used for slope deformation by comparing to A case study is located in the Jiufenershan landslide in Taiwan with
mechanical inclinometers and the differences of TDR reflected TDR technique where occurred severe landslides by the 1999
waveforms can monitor the locations of cable deformation. Su et al. earthquake. The area of the landslide is 195 ha and the amount of
(2009) demonstrate that TDR would be applied to long-term the collapse is 35 million cubic meters that blocked two creeks,
monitoring of underground sliding planes in alpine landslide areas namely Jiucaihu and Sezikeng creeks, and caused two dammed
effectively, fast and economically. lakes. The area of the Jiucaihu and Sezikeng dammed lakes is 4.4 ha
However, the magnitude of the cable deformation cannot be and 6.4 ha, respectively. Since 2003, there has been a monitoring
directly determined in the field although the instrumentation of TDR landslide project in the Jiufenershan landslide with some monitoring
would monitor suitably deep underground movement (Su et al., equipment, including extensometers, inclinometers, groundwater
2009; Osasan and Afeni, 2010). The paper explains the data of TDR lever gauges, etc. TDR was installed in the area from 2012 in order
monitoring in a case study compared to the data of inclinometers; to monitor the slope movement.
moreover, the paper measured the magnitude of rock displacement The geology of the landslide is underlain mainly by Miocene
corresponding to TDR reflected waveforms in laboratory sedimentary rocks where the strike is N36°E and the dip is 21°SE
experiments. Generally, drill-log reports can determine in situ with an N–S trending synclinal axis of Daanshan syncline in the
geological conditions. According to Su, et al. (2009), they indicate eastern part; the geologic formation of the syncline axis is the
that the TDR system of coaxial cables grouted inside a drill hole can Kueichulin Formation which unconformably contacted with the
detect sliding surfaces and their movements as the function of a underlain Changhukeng Shale; the lower formations include
traditional in-place inclinometer, which showed sliding planes Miocene strata of Shihmen Formation (Shou & Wang, 2003).
between layers of colluvium and strongly weathered slate compared
with the reports of rock cores. TDR has been used for rock and soil
movements in slope monitoring (Federico et al., 2012; Lin, 2014;
Drusa and Bulko, 2016).
In terms of the qualification of TDR monitoring slope
deformation, the technique has been employed to identify zones of
rock mass deformation and blasting performance (Dowding et al.,
1988; Dowding et al., 1989). Then, much research has been
completed with good results in relation to TDR indoor experiments,
including the effect of TDR cable length on reflection signal and
different types of deformation of TDR waveforms (Su, 1990).
Furthermore, the methodology to quantify TDR waveform change
and correlate to ground deformation has made the TDR monitoring
system used in landslides more useful and advanced (Su and Chen,
1998). Su and Chen (1998) declare that an integrated area method is
proposed to quantify the magnitude of a TDR reflected waveform on
the deformation of rock masses and concrete structures through
TDR theories and bench tests. By this method, the study applies
laboratory experiments of shear tests to determine equations in the Figure 1 Location map of the monitoring stations of the
relation between the integrated area of TDR reflection coefficient Jiufenershan landslide
341
For the shear test, a test specimen was installed on the bench test
shown in Figure 3 and was design as two shearing units, S1 and S3.
While the amount of shear displacement at S1 location was fixed to
0, 10, 20, 30 and 40 mm, the amount of shear displacement at S3
location incremented by every 2 mm until the cable ruptured,
respectively. Meanwhile, a TDR tester interrogated the reflected
waveforms of the cable displacements at the two locations. The
results of TDR interrogated waveforms are plotted in Figure 5. The
TDR cable signature (reflection coefficient, unit: millirhos)
represents the amplitude of reflected waveform at a given distance
(unit: meter) of grouted cable. The zoom-in graph of the reflected
waveforms is shown in Figure 8 that illustrates the increase of the
shear displacement on S3 plane until the cable ruptured by 54 mm.
All shear tests are the same tendencies as the graph of Figure 5
at S3 plane when the amount of S1 displacement is 10, 20, 30 and
40 mm, respectively. Above all shear tests, the average maximum
Figure 2 Diagram of on-site TDR instrumentation magnitude of cable displacement by shear is about 60 mm.
According to the results of the shear tests, there is a significant
There are two TDR monitoring stations in the area where coaxial relationship between shear displacements and the magnitude of
cables were grouted. Also, there are two inclinometers, as shown in reflection coefficients. As can be seen in Figure 6, a quadratic
Figure 1. A diagram of on-site TDR instrumentation is shown in regression equation is built up in the relations between the reflection
Figure 2. A TDR cable tester (HL 1101) sends a voltage pulse coefficient and shearing deformation (The R-Square of each
waveform which travels along a metallic coaxial cable through a equation is more than 0.99). As a result, it is applied to determine
connector. The cable was grouted into a borehole in a landslide with the magnitude of shear slips in field corresponding to the magnitude
cement. When the waveform of TDR encounters a change in the of reflection coefficients interrogated by on-site TDR system.
characteristic impedance of the cable, it is partially reflected by the
deformation of the cable in a potential sliding surface. The change
could be determined as the location of the shear or extension zone in
a landslide.
3. MAGNITUDE OF CABLE DEFORMATIONS IN

LABORATORY EXPERIMENTS
A semi-rigid coaxial cable was grouted into a cement test specimen
which was one-meter in length and connected a TDR cable tester
(HL 1101). Then, the test specimen was installed on the bench test.
The purpose of the test setup was twofold. They are: (1) to
determine the relation between the magnitude of cable deformations
and TDR reflected waveforms by shear displacement (Figure 3). (2)
to measure a change length of the cable with TDR reflected
waveforms by extension displacement (Figure 4). The properties of Figure 5 Graph of reflected waveform changes at S3 shear plane
the cable and the grout in tests are shown in Table 1.
Figure 3 Instrumentation of shear bench test
Figure 6 Graph of the relationship between the shear displacement

(millimeters) and reflection coefficient (mρ) in shear tests
Figure 4 Instrumentation of extension bench test
For the extension test, a test specimen was installed on the
Table 1 TDR coaxial cable and grout properties bench test shown in Figure 4 and the amount of extension
Coaxial cable Grout displacement at T1 location incremented by every 5 mm until the
Grout Compressive cable ruptured. Again, the TDR tester interrogated the reflected
Diameter Cement mixed waveforms of the cable displacements at the location. The results of
Type strength
(mm) type ratio TDR interrogated waveforms are plotted in Figure 7. The magnitude
(kgf/cm2)
(water to of extension displacement was 35 mm as the cable ruptured.
CTLLCX φ25.0 Portland O’Connor and Dowding (1999) suggest that it has also been possible
1:3 146.5 to quantify the extension deformation by changes in distance
7/8” CFC mm Cement I
between the locations of crimping cable made prior to installation in
a borehole. As the result of the test, the change in distance between
342
first waveform and the waveform of cable ruptured in the extension

test was measured to 0.0347 m that was almost the same as the
magnitude of extension displacement at 35 mm. Therefore, the
result can be applied to measure the magnitude of extended cable in
field by comparing to the initial length of the grouted cable with on-
site TDR monitoring.
Figure 9. Recorded waveform at the N2 TDR monitoring station
5. COMPARISON OF INCLINOMETER DATA

Two TDR monitoring stations have been set up in this landslide area
since 2012. Each station is also equipped with facilities including
Figure 7 Graph of reflected waveform changes in extension tests the piezometer for measuring the groundwater level, the in-place
inclinometer near two TDR monitoring stations. The inclinometer
4. RESULTS OF TDR MONITORING IN FIELD TESTS sensors detect the angles of slope deformations in a borehole which
are measured to determine the location, displacement, and slide
As mentioned above, there are two TDR monitoring stations in the
direction (Su, et al., 2009). An inclinometer probe is sensitive to
area and each coaxial cable of the station was grouted into a
gradual changes in the inclinometer casing where is localized at
borehole. The TDR monitoring waveforms are plotted in Figure 8
shear zone. On the other hand, a metallic coaxial cable of TDR
and Figure 9 that show the changes of the reflection coefficients at
deforms easily when subjected to rock mass deformations
N1 and N2 TDR monitoring stations. The TDR cable signature
(O’Connor and Dowding, 1999). The difference between the
shows the amplitude of reflected waveform at a given distance
inclinometer and TDR is that the former can detect the accumulation
corresponding to a depth of grouted cable at TDR monitoring
of displacement in the inclinometer casing, while the TDR cable can
stations. All recorded waveforms of different date are plotted in the
detect deformation at any point along its length.
same graph every station so that it is easy to interpret a change of
Two inclinometers were installed near the TDR monitoring
waveforms at a depth where a shear failure or extension failure
stations in the landslide area between 2012 and 2017. Table 2 shows
occurred.
the comparison between the inclinometer and the TDR
characteristics in the research. There are some results of two TDR
monitoring stations comparing to the bore logs of core drilling and
inclinometer data.
Table 2. Overview and characteristics of slope monitoring methods

in the field
Method Specification Tester type Price
Accuracy:0.01° KOWA,GIC- $600 per
Inclinometer
Range:-15°~ +30° 45S sensor
Diameter : 12.7 mm
Gain: 20 mrho $6-10 per
TDR HL1101
Ddiv: 0.1 meter meter
Vp:0.85
5.1 N2 monitoring station

Figure 8. Recorded waveform at the N1 TDR monitoring station
There was an extended displacement at a depth of 35.72 m at
TDR_N2 station (see Figure 9) in which the extended length of the
With regards to the changes of TDR waveforms in Figure 8,
end cable was 0.31 m from July to November in 2012. Also, the
there was not a significant change of the waveforms at N1 station
TDR waveform signature showed the cable ruptured by extension at
from 11/2014 to 11/2017. However, some TDR waveforms were
a depth of 7.51 m in April 2013. In terms of inclinometer
formed as the type of extension failure at 7.51 m and 35.72 m in
monitoring, the inclinometer data only shows some sliding
depth of N2 station because the type of the waveforms was similar
phenomena in a range of 0 m and 8 m in depth, which cannot
to the type of extension experiment in the laboratory. Furthermore,
determine accurate location of sliding plane from 06/2012 to
the length of the TDR cables was extended at the end of them. As
10/2017, as shown in Fig 10. Also, a failure plane cannot be
shown by the Nov/2012 TDR record in Figure 9, there was an
interpreted at the range of upper 30 m in depth in despite of a slight
extended displacement at a depth of approximately 50 m of N2
signature of displacement.
station by 0.31 m.
343
Meanwhile, TDR did not detect any displacement at N1 TDR

monitoring station from 2012 to 2017.
Overall, the results of the TDR monitoring system show that it
is useful and accurate to detect the locations of sliding surfaces in
the landslide area. The locations of failure planes and the magnitude
of the displacements can be determined by the changes and
quantification of TDR reflected waveforms. Thus, TDR is an
efficient and economical method of locating the depth and
measuring the magnitude of slope movement in landslides. In the
future, it could be a real-time monitoring and pre-warning system.
7. REFERENCES
Blackburn, J. T. and Dowding, C. H. (2004). Finite-element analysis
of time domain reflectometry cable-grout-soil
interaction. Journal of geotechnical and geoenvironmental
engineering, 130(3), 231-239.
Dowding, C. H. and Huang, F. C. (1994). Early detection of rock
movement with time domain reflectometry. Journal of
Geotechnical Engineering, 120(8), 1413-1427.
Figure 10 Results of BH8 inclinometer measures Dowding, C. H., Dussud, M. L., Kane, W. F. and O’Connor, K. M.
(2003). “Monitor deformation in rock and soil with TDR
5.2 N1 monitoring station sensor cables.” Geotechnical Instrumentation News, June,
51–59.
There was not a significant change of the waveforms at TDR_N1 Dowding, C. H. and Pierce, C. E. (1994). “Use of Time Domain
station from 2012 to 2017, which means no slope movement Reflectometry to Detect Bridge Scour and Monitor Pier
occurred. Meanwhile, the inclinometer data did not detect any Movement,” Proceedings of the Symposium on Time Domain
sliding phenomenon in Figure 11. The station could be located at the Reflectometry in Environmental, Infrastructure, and Mining
outside of the landslide area but the monitoring data still needs to be Applications, Evanston, Illinois, Sept 7-9, U.S. Bureau of
detected for long-term observation. Mines.
Dowding, C. H., Su, M. B. and O'Connor, K. M. (1988). Principles
of time domain reflectometometry applied to measurement of
rock mass deformation. International Journal of Rock
Mechanics and Mining Sciences & Geomechanics
Abstracts, 25 (5), 287-297.
Dowding, C. H., Su, M. B. and O'Connor, K M. (1989).
Measurement of rock mass deformation with grouted coaxial
antenna cables. Rock Mechanics and Rock
Engineering, 22(1), 1-23.
Drusa, M. and Bulko, R. (2016). Rock Slide Monitoring by Using
TDR Inclinometers. Civil and Environmental Engineering,
12(2), 137-144.
Federico, A., Popescu, M., Elia, G., Fidelibus, C., Internò, G. and
Murianni, A. (2012). Prediction of time to slope failure: a
general framework. Environmental Earth Sciences, 66(1),
245-256.
Kane, W. F. (2000). "Monitoring Slope Movement with Time
Figure 11 Results of BH7 inclinometer measures
Domain Reflectometry (TDR)", Geotechnical Field
Instrumentation: Applications for Engineers and Geologists.
To summary, there were some extended displacements at
different depths over time at N2 TDR monitoring station in the Sponsored by: ASCE Seattle Section Geotechnical Group
Spring Seminar and the University of Washington
landslide area, which was the same as the interpretation of rock core
Department of Civil Engineering, Seattle.
data. Meanwhile, the locations of failure planes and the magnitude
Kane, W. F. and Beck, T. J. (1996). Rapid slope monitoring. Civil
of the displacements can be determined by the changes and
quantification of TDR reflected waveforms. However, the data of Engineering, 66(6), 56.
Lin, C. (2014). Slope’s automatic monitoring and alarm system
inclinometers was not significant result of monitoring. Inclinometers
based on TDR technology. Civil Engineering and Urban
may not detect smaller changes in the displacement of slope
Planning III, 371.
movement. It is obvious that TDR is an efficient and economical
O’Connor, K. M. and Dowding, C. H. (1999). GeoMeasurements by
method of locating the depth and measuring the magnitude of slope
Pulsing TDR Cables and Probes. CRC Press, Boca Raton,
movement in the landslide.
Florida.
Osasan, K. S. and Afeni, T. B. (2010). Review of surface mine slope
6. CONCLUSIONS
monitoring techniques. Journal of mining science, 46(2), 177-
The research applied TDR to detect the displacement of slope 186.
movement and to measure the type and magnitude of the Shou, K. J., & Wang, C. F. (2003). Analysis of the Chiufengershan
deformation in the Jiufenershan landslide. Compared to the data of landslide triggered by the 1999 Chi-Chi earthquake in
on-site inclinometers, it proved that there was an extended Taiwan. Engineering Geology, 68(3), 237-250.
displacement at a depth of 35.72 m at N2 TDR monitoring station Su, M. B. (1990). Fracture Monitoring within Concrete Structure by
by 0.31 meters and TDR waveform signature showed the cable Time Domain Reflectometry. Engineering Fracture
ruptured by extension at a depth of 7.51 m from 2012 and 2013. Mechanics, 35 (1/2/3), 313-320.
344
Su, M. B. and Chen, Y. J. (1998). Multiple Reflection of Metallic Su, M. B. and Chen, Y. J. (2000). TDR Monitoring for Integrity of
Time Domain Reflectometry. Experimental Techniques, Structural Systems. Journal of Infrastructure Systems, 6(2),
22(1), 26-29. 67-72.
Su, M. B. and Chen, Y. J. (1999). MTDR monitoring systems for Su, M. B., Chen, I. H. and Liao, C. H. (2009). Using TDR Cables
the integrity of infrastructures. Journal of intelligent material and GPS for Landslide Monitoring in High Mountain Area.
systems and structures, 10(3), 242-24 Journal of Geotechnical and Geoenvironmental Engineering,
ASCE, 135(8), 1113-1121.
345
Application of data mining technique to complement photogrammetric roughness

data
Dong-Hyun Kim1, Chul-Ho Lee2, Arumugam Balasubramaniam3 and Ivan Gratchev4
1, 2
Institute of research and development, Hulogue Inc., Seongnam-si, Gyeonggi-do, Republic of Korea
3, 4
School of Engineering, Griffith University, Gold Coast, Australia
E-mail: d.kim@hulogue.com
ABSTRACT: This paper presents the applicability of data mining techniques to improve the accuracy of photogrammetric roughness data.
Close range photogrammetry (CRP) has employed to create competent 3D slope models and can be used to highlight geological features on
rock slopes. However, when a high degree of accuracy is required, this technology has also shown inconsistent values with respect to its
accuracy because of various influencing factors induced by camera structures and survey environment. Thus, the uncertainty of this
technology demands complex post image processing frequently. Data mining (DM) techniques are increasingly being used to identify valid
and potentially useful correlations and patterns in existing data. In this study, a data mining technique is applied to reduce the noise of 3D
images and to analyse multiple images effectively. Conclusions showed that a better performance was achieved using DM technique and this
data mining approach was able to improve the accuracy of photogrammetric roughness data.
Keywords: Data mining, close range photogrammetry, image mining.
1. INTRODUCTION al., 2003; Thompson et al., 2005). For example, 3D images are used
in a medical image analysis area, sets of 3D images of human brains
Data mining (DM), which is also known as knowledge discovery obtained from MRI scanning have been analysed using a fully
from data (KDD) is the process of discovering interesting patterns
automatic method using a data mining program (Uher et al., 2013).
and knowledge from large amount of data. As a fast-growing field,
DM has been applied effectively to solve multiple problems This study presents applicability of data mining techniques to
connected to risk management, including error detection and predict the accuracy of photogrammetric roughness data. Data sets
quantification, detecting fraud etc. in our data-rich world. Higher and attributes of joint roughness coefficient (JRC) and asperity
level of this technique can be to build learning models to heights obtained from a set of photogrammetric laboratory tests of a
automatically extract knowledge from big and complex data. This previous study are employed as input data. Also, the results of
technique has been progressively introduced to various industry image mining approaches using 3D images corresponding to the
sectors like marketing, manufacturing and construction. data sets are discussed.
In geotechnical engineering area, since early 1990s’ Artificial 2. OVERVIEW OF DATA MINING TECHNIQUES
Neural Network (ANN) has been increasingly employed to analyse
various engineering issues such as constitutive modelling, bearing Data mining applies statistical and logical methods to large data sets
capacity of pile and slope stability. This technique was regarded as and allows people to locate and interpret data patterns, helping them
the most appropriate and effective tool among other data mining better informed decisions. The analysis process, called Knowledge
techniques as it is known as the most relevant to engineering Discovery in Database (KDD), defines the procedure for
disciplines. Recently, Tinoco et al., (2010) approached to predict transforming raw data into useful knowledge. The method can be
compressive strength of jet grouting columns using a data mining used to categorize the data, or this can be used to create predictive
technique. A data mining approach for the volcanic rock models. As shown in Fig. 1, a general KDD analysis process
classification has been reported by Miranda et al. (2018). consists of the following multiple steps: [1] Data selection, [2] Data
cleaning, [3] Data transformation, [4] Pattern extraction and
Close distance photogrammetry has been increasingly adopted to interpretation.
create 3D models to investigate detailed aspects of rock slopes, as
well as large-scale surveys in rock engineering and mining. Recent
high-end equipment in the field of digital photographing can create
high resolution images and this technology has encouraged the
spread of close range photogrammetry to obtain surface features
such as undulation and roughness which require higher degree of
accuracy of 3D models. However, there are several known and
unknown factors that can affect the accuracy of the results and the
accuracy of photogrammetric models has insufficiently discussed.
The authors have been studied on the accuracy of photogrammetric
roughness data for many years (Kim et al., 2015; 2016). However, Figure 1 Process of data mining
the relevant data have still been insufficiently explored and analysed In general process of data mining, firstly, we need to identify the
by advanced data analysis techniques. objects of interests in the data and secondly, the quality of the data
As the pixels of digital photographs embody large number of can be improved by techniques such as de-noising and cleaning.
features, images represent a significant assembly of measurements. Third, the multivariate data can be converted into suitable forms for
As a part of data mining, image mining as a specific term has been pattern recognition by pre-processing. Then, the patterns in the data
used for analysing various images of DNA microarrays, can be identified using techniques such as classification, clustering
astronomical observations, satellite maps, medical images etc (Van and association rules. The overall data mining is interactive process.
der Walt et al., 2014). With regard to image data, data mining is Data analysts involved in every step starting with an initial
associated with four broad problems: finding association, formation of data sets in the data collection process to validating the
classification, sequential patterns and time series patterns (Ordonez results obtained at each step. The process of this study is similar to
and Omiecinski, 1998). The applications of image mining the basic procedure of data mining shown in Figure 1.
techniques have been reported in the literature (Soh, 1999; Uher et
346
3. METHODOLOGY In the program ˋRapidminerˊ, there are many operators for image
transforms and manipulation. As a feature extraction algorithm, this
This study analyses rock surface roughness data obtained from a set study uses “local-level” feature extraction. This is suitable for
of photogrammetry laboratory tests. The image data used in this extracting information such as pixel brightness integer values after
study is parts of image sections extracted from the 3d surface image enhancement from each point in the image. Then, the image
models. JRC values and the corresponding measurements such as data can be transformed from image into a structured table form.
camera-to-object distances, pixel sizes and focal length of lenses are The first operator used as a median filtering with a size of 3 pixels.
also employed in this study. To analyse the data in each process, this In this stage, multiple image operator was also employed. This
study uses a data mining program, RapidMiner studio (ver. 8.1) operator iterates the next analyses over all image files in a folder.
which is an open-source and flexible platform implemented in Java This provided convenience for the reiterative procedure for dozens
(RapidMiner Inc.). Also, in the pre-processing stage, the 3D images of images in this study.
are filtered by smoothing to reduce the influence of the noise of raw
images.
A digital image can be regarded as a group of discrete pixels, each
of which has various ranges of colour and brightness information.
Colour images with various geological features can be more
complex than greyscale images. In this study, pixels converted from
colour to 8-bit grey-scale are used as the input data. Thus, the
processed values are in the range from 0 to 255. Common approach
in data mining is to mine information from structured data with
tabular form. However, image is an example of the unstructured data
form. Thus, a feature extraction process which transforms an image
into the tabular data is needed. In this study, Image mining
extension (IMMI) operator is employed in RapidMiner for the
feature extraction process. Figure 3 Concept of image mining using image processing operators
with multiple images
3.1 Data cleaning
There are several operators which can enhance the quality of images Then, a learning process can be applied to reduce noise, recognize
in the IMMI operator. As a representative tool, median filtering is and identify a pattern of pixel information for multiple images.
one of effective solutions to reduce the influences of the noise from There are several learning algorithms in IMMI’s operators such as
the obtained photogrammetric roughness data. Poropat (2008) support vector machines (SVM), decision tree and random forest. In
showed a clear effect of median filters from some simulations using this study, SVM with a Kernel method was selected for training
roughness profiles (Fig. 2). To enhance the accuracy of the stage based on the results of trials with other algorithms. It has been
photogrammetric roughness data, the raw 3D images are processed reported by Burget et al. (2012) that SVM with Kernel method was
with a median filter in this study. also successfully used for analysing a biomedical image. The
designed process of Rapidminer in this study is shown in Figure 4.
The results obtained from the learning process can be presented in
tabular forms for further analysis steps. These structured data were
used for statistical analyses using the differences of integer numbers
between neighbouring points of interest. This data collection method
was applied to the pixels at the given points of interest at the same
locations on several sets of different resolution images.
Figure 2 Details of the surface profile, corrupted with noise a),

filtered with a 5 pixel median filter b) (Poropat, 2008)
Figure 4 Designed process of image mining using IMMI
3.2 Data training and extraction

To measure the discrepancy of the bright integers, the root-mean-
In image mining, it is possible to connect each pixel intensity of an squared-error (RMSE) is employed in this study (See Eq. (1)).
image to each input of process. In the case of 3D images, the pixels RMSE is defined to be the square root of the average of the squared
of the image can be given some meaning with their corresponding discrepancies. For photogrammetry applications, RMSE has been
coordinates. This study attempts to correlate between the data of widely used to identify the accuracy of data due to its high
pixels and the photogrammetric roughness measurements on the correlation between the predicted values and the observed values
corresponding locations. The first stage of this study is the learning (ASPRS, 2014).
stage. Pixel information of points along measured profiles on the
different resolution images are compared by a learning procedure of RMSEInt=SQRT {Ʃni=1(Into,i-Intp,i)2 / N} (1)
the image mining and the results.
347
where Into is the brightness integers obtained from the highest values between two consecutive pixels along the profiles of interest
resolution image; and Intp is the brightness integers obtained from obtained from different resolution images obviously indicate
compared images. downward trends as the camera moves farther from the rock sample,
corresponding to the decrease of image resolution. This is directly
dependent on the sizes of pixels of the created images.
4. CASE STUDY
4.1 Photogrammetry laboratory data
The data used in this paper were collected from a previous
laboratory work on accuracy of photogrammetry roughness data.
The data were images of a rock sample with different resolutions as
shown in Table 1. The photographs were taken in the range of 1.0 ~
7.0 m, using three different focal length of lenses: F=24mm, 50mm
and 85mm. The length and width of the targeted area on the rock
surface were 20 cm and 8 cm, respectively. Details of the tests are
presented in the previous literature by the authors (Kim et al., 2015).
The spatial data of the 3D surface models were created using a
photogrammetry code, Sirovision (CAE). JRC values were
Figure 6 Schema of image profile
estimated along 8 measurement profiles on the rock sample.
This study selected high resolution images with a range of pixel
sizes less than 1.0 mm as summarized in Table 1. The selected area
of the sandstone surface was appeared to be reasonable in this study
because in observations it was found that the range of brightness on
the surface was quite simple as an indicator of the change of
roughness based on its simple rock-forming. Figure 5 compares the
greyscale images of the sample area with different resolutions.
Using the targets on the images, 8 profiles were selected at the same
locations. The data of pixels with 1 mm interval along the profiles
were extracted by using an operator of IMMI. A total of 28,000
integer values were collected from 14 images and employed for the
learning process.
Table 1 Summary of image data used in this study
FL c-to-o Pixel size No. No.
(mm) distance (m) (mm) images data
24 1~3 0.5 ~ 1.0 3 6,000
50 1~5 0.1 ~ 1.0 5 10,000
85 1~7 0.1 ~ 1.0 6 12,000
Figure 5 Comparison of images of the rock sample of pixel size = Figure 7 Data distributions of integer values
1.5 mm a), 1.0 mm b) and 0.2 mm c)
4.2 Integer values of pixels The histograms of all training showed an interesting trend. Using all
resolution images, Figure 7 indicates the differences of integer
The designed learning process produced integer values of the values obtained from comparison between highest resolution (pixel
smoothed images by a median filter in a tabular form. This image size = 0.1 mm, FL= 85mm) and target images. The data formed an
mining process enabled us to collect tens of thousands of data in a exponential distribution for each profile. Mean values of the integer
short period of time from hundreds of the selected points on differences was ranged from 13.7 to 15.7 for all image profiles. Also,
multiple images. As presented in Figure 6, for the continuous integer a high range of standard deviation which ranges from 11.6 to 21.2
of a profile on an image, absolute values of the difference between was calculated from the set of data values. The ranges of statistics
two consecutive integer values along the profile is regarded as a were quite similar to each other in the different resolution images.
roughness parameter in this study. Sets of the differences of integer
348
In Figure 7, the histograms and the cumulative distribution curves - Using RMSEINT, a range of JRC values can be estimated and
distributed by their corresponding JRC values also indicate this method can complement when photogrammetry is
interesting trends to evaluate the accuracy of photogrammetric JRC employed to quantify JRC values.
values. Overall, bigger JRC values formed wider dispersion of data.
The range of integer values for smaller JRC values is composed of 6. REFERENCES
smaller ｜Into-Intp｜values. The threshold values which composes
American Society for Photogrammetry & Remote Sensing (2014)
50% of data are 10 (JRC= 9.9), and 14 (JRC= 11.5) and 17
"ASPRS positional accuracy standards for digital geospatial
(JRC=13.5). This may be attributed to the fact that the standard
data". PE & RS, 81, 3, ppA1-A26.
deviations increase with the JRC values. The inspection of
Burget, R., Vaclav, U. and Masek, J. (2012) "Trainable
cumulative distribution plots indicates that the integer values
segmentation based on local-level and segment-level feature
obtained from different resolution images can be used to verify
extraction". IEEE International symposium on biomedical
photogrammetric JRC values.
imaging: from nano to macro, pp1-8.
4.3 Integer values and JRCs Han, J., Kamber, M., Pei, J. (2012) "Data mining concepts and
techniques". 3rd edition, Elsevier Inc., p.703.
Using the RMSEINT values obtained as described in Section 3.2 and Hofmann, M. and Klinkenberg, R. (2014) "RapidMiner dataming
using the equation (1), statistics analysed a relationship between use cases and business analytics applications". Taylor &
RMSEINT and JRC values as presented in Figure 8. The ranges of Francis Group, LLC, p.518.
RMSEINT obviously increases as the corresponding JRC values are Kim, D-H., Poropat, G., Gratchev, I., Balasubramaniam, A. S.
raised. This is an interesting finding because the integer values are (2015) "Improvement of photogrammetric JRC data
obtained from 2D images which exclude any information of distributions based on parabolic error models". International
coordinates of the rock surface. Journal of Rock Mechanics & Mining Sciences, 80, pp19-30.
In the authors’ previous study, the data distributions of normalized Kim, D-H., Poropat, G., Gratchev, I., Balasubramaniam, A. S.
JRC values comparing photogrammetric JRC values and manual (2016) "Assessment of the accuracy of close distance
measurement were fairly scattered showing the ranges of the photogrammetric JRC data". Rock Mechanics and Rock
coefficient of determination, R2 from 32% to 48%. The uncertainty Engineering, 49, pp4285-4301.
of this technology to estimate JRC values need assistants to improve Miranda, T., Sousa, L. R., Gomes, A. T., Tinoco, J. and Ferreira, C.
its accuracy to quantify roughness data of rock surface. The integer (2018) "Geomechanical characterization of volcanic rocks
values obtained from different resolution greyscale images can using empirical systems and data mining techniques". Journal
complement the shortcoming of photogrammetry. of rock mechanics and geotechnical engineering, 10, pp138-
150.
Ordonez, C. and Omiecinski, E. (1998) "Image mining: a new
approach for data mining". Georgia Institute of Technology
technical report, GIT-CC-98-12.
Poropat, G. V. (2008) "Remote characterisation of surface
roughness of rock discontinuities". In: Proceedings of the 1st
Sourthern hemisphere international rock mechanics
symposium, Perth, Australia, pp 447-458.
Soh, L-K. (1999) "Segmentation of satellite imagery of natural
scenes using data mining". IEEE Transactions on geoscience
and remote sensing, 37, 2, pp1086-1099.
Thompson, D. R., Smith, T. and Wettergreen, D. (2005) "Data
mining during rover traverse: from images to geologic
signatures". In: Proceedings of 8th International Symposium
on Artificial Intelligence, Robotics and Automation in Space,
Germany.
Uher, V., Burget, R., Masek, J. and Dutta M. K. (2003) "3D brain
tissue selection and segmentation from MRI". In: Proceedings
of 36th International conference on telecommunications and
signal processing (TSP), Rome, Italy.
Figure 8 Correlation between RMSEInt and JRC values
5. CONCLUSION
This paper presented an applicability of image mining technique to
analyse multiple images of rock surface for rock surface
investigation. A learning process has been designed to perform the
image mining process using a data mining program, Rapidminer.
Using an image data set collected from a previous photogrammetry
laboratory tests, following conclusions can be drawn:
- Image mining techniques can be effectively used for
analysing multiple images including a large volume of image
data. In this study, IMMI operators of Rapidminer was
employed.
- Differences of integer values of pixels are used to identify
rock joint roughness coefficient (JRC) in two different types
in this study; [1] ｜Inti+1-Inti｜and [2] ｜Into-Intp｜. The
cumulative distribution of｜Into-Intp｜indicated a relationship
between JRC values and the integer values.
349
Case Study of a Partially Collapsed RS Wall at a Building Site

EG Balakrishnan1, Allan Y. L. Chwee2 and A. Mohd Redzuan3
1
Managing Director, GCU Consultants Sdn Bhd, Malaysia
2
Assistant Engineering Manager, GCU Consultants Sdn Bhd, Malaysia
3
Director, GCU Consultants Sdn Bhd, Malaysia
E-mail: bala@gcu.com.my
ABSTRACT: It has been a common practice in Malaysia to adopt reinforced soil walls as one of the common wall types particularly at
housing earthworks platforming. This paper will discuss one case study of the failed reinforced soil walls used in the housing project. The
paper will discuss on the background on the construction of the site development works leading to the construction of the RS wall on a deep
fill ground. The paper will highlight on the wall failure and the postulations on the likely reasons for the failure. The subsoil and groundwater
details from extensive ground investigation together with the short term and long term remedial solutions to strengthened the fill slopes and
wall will be discussed. The paper will also present all the instrument monitoring results validating the remedial works performance. Lessons
learned from these experiences will be highlighted and discussed in detail.
Keywords: Failed reinforced soil wall, remedial works with soil nail, case study.
1. INTRODUCTION
The project site where the wall collapsed is located about 15km
from Kuala Lumpur City within a refutable suburb housing project.
Existing bungalow houses are located on the south side of the site
with a road dividing the collapsed wall and the houses. The north
side is predominantly undeveloped and comprise of flat ground after
the slopes reserved for future development of Muslim cemetery and
school. The forest reserve is located on the eastern side. A stream is
located at the toe of the slope flowing from the forest reserve to the
west.
Based on the topographic map obtained from the relevant
authority of Malaysia, the site was located at the eastern boundary of
an estate on a north trending valley along the eastern slope of a
ridge. The map also indicates a stream together several ephemeral
streams originally passed through the site. Three streams or rivulets
coming from the north and east directions converged into a bigger
stream north of the site. As the three streams shown in the map
initiated just outside or within the development site, this may not
contribute to any underground seepage. The stream from the north
side originating within forest serve may contribute to the
underground seepage as this stream was not diverted during the site
development stage in year 2000 to 2001.
During the site development (earthworks), the western and
south-eastern parts of the site were being cut whilst the north part
being filled with the cut materials which comprise of earth and rock
fragments. Between 2001 and 2004, the southeastern hill was further
cut and the north valley was filled to form the building platform
level. The fill slopes were formed with slope drainage and
vegetation based on the satellite imagery. In year 2005, a 120 m
long RS wall with heights varying from 4m to 10m was constructed
on the northern most slope to create space for the construction of the
access road at RL 63.20 m to the residential units.
Part of the wall about 11m stretch collapsed on 20 May 2011.
The collapsed wall is 9m high with exposed wall height of 25 m.
The top 6 m of the RS wall with 4 panels vertical and 7 panels
horizontal collapsed. The location and photograph of the collapsed
wall is shown in Figure 1 and Figure 2. Figure 1 Location of collapse wall
350
remedial works carried out, the geotechnical instrumentation

monitored during and after the works and finally outline the lessons
learned from this case study on the failure of a RS wall.
2. SUBSOIL CONDITIONS AND PROPERTIES
Based on published Geological Map of Kuala Lumpur, the site is

underlain by Kenny Hill Formation of Permian Carboniferous age.
Kenny Hill formation consists of monotonous sequence of
interbedded sedimentary rocks such as sandstone, siltstone, shale
and mudstone. The damp tropical climate with average annual
rainfall of more than 2500 mm has resulted in extensive chemical
Figure 2 Partial collapsed of reinforced soil wall on 20th May 2011 weathering of these rocks to produce weathered profile above the
rocks. Variations in the weathering profile are mainly due to
The collapsed wall is founded on a fill ground of maximum 19 differences in the parent material and site morphology.
m high with 4 fill slopes below the wall. Based on the survey Ground investigation comprising of boreholes (7 Nos), trial pits,
drawings and detail site inspection, variations in the slope height seismic refraction and electrical resistivity survey were carried out
and gradients were observed. The slope height was not consistent under different agencies involved in this project. Standpipe
with heights varying from 3.5 m to 7 m. The slope gradient also was piezometers installed in all boreholes to monitor the groundwater
inconsistent with varying values of 1(V):1.2(H) to 1(V): 2(H). The levels over a period. Appropriate laboratory tests were carried out
slope berms also vary from 1 m to 2 m. The slopes were generally on disturbed, undisturbed and trial pit samples.
covered with vegetation with shrubs and small tress have grown on Based on the soil investigation and laboratory testing results, the
the slopes. The fill slopes have shown signs of erosion over the subsoil profile is identified to be 3 distinct layers. The first layer
years with some locations only resulting in three (3) slopes with the consist of fill material with thickness varies between 1m and 25m.
bottom slope has been eroded badly. Water seepage was observed This layer consists of a mixture of soft to stiff sandy SILT and loose
coming out from behind the wall after collapse and continuous flow to dense silty SAND with high content of granular materials. The
at the toe of the thick embankment fill as shown in Figure 3. The next layer consists of residual soil of medium stiff to hard sandy
topography and drainage pattern of the study area are shown in SILT with SPT-N value less than 50 blows. The thickness varies
Figure 4. from 4 m to 12 m. The final layer consists of hard layer with SPT-N
more than 50 blows and generally made up of hard sandy SILT. The
SPT-N versus depth plot is shown in Figure 5. The properties of
these layers are summarized in Table 1.
Figure 3 Copious flow of water seeping out from the toe of

embankment
Figure 4 Original topography and drainage pattern of the site
This paper will address the subsoil conditions together with the
ground water conditions, the topographical changes over the period
at this site, subsoil design parameters interpreted from the laboratory Figure 5 SPT-N versus Depth plot
tests, likely reasons for the wall collapse, the short term & long term
351
Table 1 Typical Properties of Subsoil Layers
Layer Description SPT – N Thickness Moisture Bulk Atterberg Limits Grain Size (%)
Blows (m) Content Density (%)
(%) (kN/m3) LL PL PI Coarse Fine
1 Fill 2 – 31 1 - 25 10 - 25 19.7 – 20.8 20 - 15 - 5– 45 - 90 10 - 55
(varies) 40 25 15
2 Residual Soil < 50 4 - 12 10 - 35 15.8 – 21.5 25 - 15 - 10 - 20 - 75 25 - 80
50 35 15
3 Weathered Soil > 50 > 10 15 - 25 - 20 - 15 - 10 - 50 - 80 30 - 50
40 20 25
The typical t-s plots for the fill material and the residual Block samples from the trail pits were also obtained to validate
weathered formation to obtain the effective strength parameters are the effective strength parameters consolidated undrained (CIU) and
shown in Figure 6a and 6b. drained (CD) triaxial tests. The summary of these results shown in
Table 2 validated the values used in the analysis.
Table 2 Design Parameters of Subsoil Layers

Lower bound Median Upper bound
c' φ' c' φ' c' φ' Subsoil Description Thickness Bulk Effective
(kPa) (˚) (kPa) (˚) (kPa) (˚)
Layer (M) Density Strength
6.4 29.3 7.0 32.1 7.4 36.9
(kN/M3) Parameters
C’ Φ’
(kPa) (°)
1 Fill 1.0 - 25.0 19.0 5 30
2 Residual Soil 4.0 – 12.0 18.0 5 32
(SPT – N < 50)
Adopt:
3 Weathered Soil > 10 20.0 7 34
c' = 5.0 kPa
(SPT – N > 50)
φ’ = 30˚
The groundwater levels vary from 3 m to 21 m below existing
ground level with closer to the surface at the toe of the slopes. The
typical subsoil profile at the distress area is shown in Figure 7.
Figure 6a t-s Plot for fill

γ = 19kN/m³
c' = 5kPa
φ’ = 30˚
Lower bound Median Upper bound
c' φ' (˚) c' φ' (˚) c' φ' (˚)

(kPa) (kPa) (kPa)
4.8 27.0 5.1 33.3 9.0 39.1
RESIDUAL SOIL
(SPT-N<50)
γ = 18kN/m³
c' = 5kPa
φ’ = 32˚
RESIDUAL SOIL
(SPT-N<50)
Adopt: γ = 20kN/m³
c' = 5.0 kPa c' = 7kPa
φ’ = 34˚
φ’ = 32˚
Figure 7 Typical subsoil profile at distress area
Seismic refraction survey was carried out along the wall line as
shown in Figure 8 and the results in shown in Figure 9. The seismic
refraction survey indicates two distinct layers based on the contrast
of the velocities. The upper layer is characterized by low p-wave
velocity representing the fill materials of sandy SILT or silty SAND.
Underlying this fill layer is the highly weathered meta-sedimentary
profile with higher p-wave velocity. These results conform closely
Figure 6b t-s Plot for cut (SPT-N < 50) to the findings of the boreholes.
352
 Regular water ponding after each heavy downpour on the road

next to the collapsed wall area. Due to the water ponding and
delayed dissipation of surface runoff, water could have infiltrate
into the ground and cause an increase in hydrostatic pressure
behind the wall especially over a prolonged period of time.
 From the site visit, it is observed that a lot of shrubs and some
trees growing between the RS Wall panels. This could be another
contribution factor to the failure as the roots growing from the
shrubs and trees may cause the wall panel to be displaced and
instill additional stresses in the reinforcing tendons.
The RS wall has been designed in accordance to the code of

Figure 8 Location of seismic profiles practice BS 8006 (1995). Flooding and leakages of the water
carrying services near to the wall has caused an increase of pore
water pressure. This has caused the overstressing of the tendons. A
detail analysis simulating the buildup water pressure show that
reinforced wall tendons in layers 5 to 8 were overstressed and the
final collapse took place due to the failure of the tendon T12 at layer
8 which has the highest stress. The settlement of the fill under the
wall could have further aggravated the situation. The likely
postulation of the failure is shown in Figure 11.
Figure 9 Seismic profile along seismic lines S1 and S2
Three lines of resistivity survey were carried out at the first berm
from the base of slope, base of the RS wall and at the roadside near
to the houses. Typical resistivity profile is shown in Figure 10.
These profiles show the presence of partial saturated soil layers
indicating presence of groundwater seepage.
Figure 11 Flow Chart of possible factors for the wall collapse
4. REMEDIAL WORKS
The remedial works were carried out in two stages viz short term
and long term. The details of the remedial works are elaborated in
the following sections.
4.1 Short Term Remedial Works
Immediately after the wall failure, a short term remedial of

reconstructing the wall using cast insitu RC wall was carried out.
Figure 10 Typical resistivity profile along the affected area After removal of the collapsed RS panels, an RS wall was built with
the existing tendons were extended into the RC wall. Damaged or
3. POSTULATION ON THE WALL FAILURE bend tendons were replaced with new tendons. This allows the road
to be constructed immediately for the residents to use. This is the
The partial collapse of the RS wall can be attributed to the following only road for the residents to access their homes. One (1) row of
factors: horizontal drains of 12m length were also installed at about 4m from
 Leakages of fluids from certain sources which could indirectly the top of wall to mitigate any buildup of water and allow flow of
increase the hydrostatic pressure behind the RS Wall. These any trapped water behind the wall. Figure 12 illustrated the short
leakages may not be from recent incidents and may have been term remedial works using RC wall.
occurring over a long period. The potential sources would be
from concealed drains, water supply pipes and sewerage pipes. 2
out of 3 sources have been confirmed by the respective owners
that leakages have been detected in the nearby pipes. These pipes
could have been damaged over the time due to the post
construction settlement of the fill materials not properly placed
and compacted.
353
Without strengthening works, the current embankment slope

shows FOS of about 1.06 to 1.14. The embankment slope has been
strengthened using soil nails, horizontal drains and rock fill buttress
to enhance the FOS to the required levels. The analysis output for
slope without and with strengthening are shown in Figure 14a and
14b.
FOS: 1.06
Figure 12 Short term remedial work using RC wall
Another short-term measure proposed and constructed at the site

is the permanent drain channel along the forest reserve to allow
water from the forest reserve to flow to the toe drain. This will FOS: 1.405
reduce the ground water seepage due to surface runoff from the
Forest Reserve. Figure 13 indicated the drainage path.
Figure 14 Typical slope analysis output (a) without strengthening

(b) with strengthening
The typical strengthening details is shown in Figure 15. The

details of the strengthening works are:
 Significant rise in the ground water table may adversely affect

the stability of the slopes and the wall and cause potential
failures. In order to mitigate the ground water fluctuations,
horizontal drains have been installed. One row of drains at a
spacing of 4 m centers and 12 m length installed along the
bottom berm. A total of forty horizontal drains of 12 m length
and four (4) no’s of 7 m to 8 m length due to rock were installed.
 Additional soil nails of 100 mm dimeter and 12 m length at
horizontal and vertical spacing of 1.5 m centres at the bottom
Figure 13 Drainage path for short term measure
slope to enhance the FOS of deep seated failure. Galvanized T25
reinforcement bar with working load of 100kN and cement grout
4.2 Long Term Remedial Works
of minimum compressive strength of 30 N/mm2 was used. The
nails were inclined at 30 degree from horizontal. A total of 249
The entire slope system with the existing wall need to be
strengthened to achieve higher factor of safety of minimum 1.4 as no’s of soil nails were installed with some socketed into rock.
recommended in the Hong Kong Slope Manual for high risk slope. Pull out tests were performed on 4 soil nails up to 150 kN, i.e.
An extract of the GEO 84 which provides some guidelines on the 1.5 times the working load. The values are within the acceptable
Factor of Safety (FOS) based on risks to the economy loss and lives limits of 12.5 mm.
is shown in Table 3.  A rock fill buttress of compacted stones were placed on the
existing slopes after the soil nails and horizontal drain works.
Table 3 Guidelines on the Factor of Safety (FOS) based on risks to The stone used consist of hard durable inert rocks of granite
the economy loss and lives (GEO 1984) origin. The rock fill was well graded from 150 mm to 400 mm.
The rock fill was placed in layers of 500 mm and compacted by
Economic Risk to Lives
Risk at least 12 passes of vibrating roller. The gradient of the rock
Negligible Low High
>1.0 1.2 1.4 slopes was initially fixed at 1; (V): 1(H) but later amended to
Negligible
Low 1.2 1.2 1.4 1(V): 1.5 (H) upon review of the independent consultant to avoid
High 1.4 1.4 1.4 any localized failure of the rock fill.
354
 To mitigate erosion problems on slopes, close turfing and Instrumentation were installed during the remedial works to
mortared stone pitching were used on slopes not covered by the monitor the performance of the slopes stabilization measures during
rockfill. and upon completion of construction works. The instrumentation
 As failure of the reinforced soil wall involved leakages of the plan is shown in Figure 17.
water carrying services and soaking of the cohesive soil behind
the wall, free draining material was used to backfill behind the
wall. By doing so, water will be easily drained out from behind
the wall and the probability of another failure would be much
reduced. The existing precast drainage system along the wall and
part of the site road, which is currently leaking was replaced with
cast in situ drains to reduce or avoid leakages. The water supply
pipes along the wall was totally exposed and installed inside a
RC channel with measure to drain the water from the pipe if any
leakage. Sewage pipes along the wall was also redesigned and
replaced with a more robust system that could withstand large
differential settlement.
Figure 17 Instrumentation Layout
The instrumentation readings showed that the constructed

remedial works have performed as expected and all the readings are
well within the acceptable limits. Typical results of the
instrumentation readings are shown in Figure 18a and 18b.
Figure 15 Typical strengthening details
The damaged water and sewer pipes were repaired with special
allowance for damage. The final completed remedial work is shown
in Figure 16.
Figure 16 Final view of completed rectification work

Geotechnical Instrumentation
355
6. LESSONS LEARNED
From this case study of a RS wall failure at a residential housing site

occurring about 7 years after the installation of the wall, the
following lessons learned can be summarized so that such failure
will not occur in other similar projects.
1. During any site development (earthworks), adequate study shall
be carried out on the existing drainage patterns and proper
drain diversion shall be provided for stream flowing from the
area outside the development area and into the development
area. This will minimize future seepage of groundwater due to
the infiltration of surface runoff into the ground.
2. Proper selection of the cut materials shall be used for the filling
works. Rocks bigger than the sizes allowed in the specification
shall be broken into smaller sizes and used with soil fill. This
will ensure adequate compaction of the fill soils. All fill shall
be placed and compacted as per specification to minimize long
term consolidation settlement of the fill due to self-weight
particularly for high fill.
Figure 18 Typical instrumentation results (a) Inclinometer reading 3. Proper cut-off drains shall be provided at the top of fill to divert
(b) Ground Settlement Marker reading the water from the fill slopes. Preferably use open drains at hill
site to facilitate easy maintenance in the event of any clog gage
5. MAINTENANCE OF SLOPES in the drains.
4. RS wall can be constructed on thick fill ground provided all the
It is crucial that a routine slope inspection be conducted on the measures as highlighted above are strictly adhered.
embankment and its slope to ensure that the remedial works are 5. The combination of soil nails, horizontal drains and rock fill
performing according to design and to detect any signs of
mattress can be successfully implemented on the distress areas
instabilities. For a start, the inspection should be conducted every 6
of thick fill of unconsolidated fill with large rock fragments and
months. Its frequency could be revised later by the consultant,
depending on findings from the routine inspection. The slope underground seepage. This system allows the free flow of the
maintenance guidelines issued by the Public Works Department seepage water whilst maintaining the required stability at any
could be referred to for this purpose. However, a few important slip failures.
maintenance items specific to the study site shall also be carried out, 6. Adequate ground investigation is very important in such high
such as measuring the total dissolved solid and flow rate of water fill area to determine the subsoil conditions and properties.
seepage from the horizontal drains and springs at the toe of 7. Adequate routine maintenance of high slopes must be carried
embankment, checking on the buried water carrying services, taking out covering the slope drains, slope vegetation, retaining wall,
the inclinometer readings at the reinforced soil wall and checking on horizontal drains, utilities on slopes, etc.
the functionality of the monitoring and relief wells. The
maintenance report should be submitted to the consultant and 7. REFFERENCES
relevant authority for reference and further action, if required.
Routine maintenance works, such as clearing and sealing the
Atkinson, J. H. (2007). The Mechanics of Soils and Foundation. 2nd
drainage system, repairing or replacing the pump or sensor and
Edition, Taylor and Francis (Publishers).
replacing the horizontal drains, if necessary, shall be conducted
Bittelli et al. (2012). Monitoring soil – water and displacement
immediately after the inspection.
conditions leading to landslide occurrence in partially
saturated clay. Geomorphology 173- 173, 161 – 173.
Geotechnical Engineering Office, The Government of the Hong
Kong Special Administrative Region. (1984).
Geotechnical Manual for Slopes.
IKRAM & Geo Solution Resources. (2012). Geophysical
Assessment Study Report.
Krahn, J. Geoslope software manual
Mineral and Geoscience Department, Malaysia. (1993). Geological
Map of Kuala Lumpur
356
Numerical Simulation of Some Debris Flow Events in

Central Java for Predicting Run-out Distributions
A.S. Imam1, 2, D.K. Rendy1,and A.D. Indra1
1
Applied Geology Research Group, Faculty of Earth Sciences and Technology, Institut Teknologi Bandung
2
Research Center for Disaster Mitigation, Institut Teknologi Bandung
E-mail: imam@gc.itb.ac.id
ABSTRACT: The occurrence of several landslides followed by the debris flow, such as Sijeruk Landslide on January 4, 2006, Jemblung
Landslide on December 12, 2014 and Pasir Panjang Landslide on Februari 22, 2018 has make a lot of casualties and terrible destruction in
some parts of Central Java. In general, debris flow can be divided into source area, flow track and depositional area. This study will focus on
modelling the run-out of debris materials, starting from the source area to the distribution of debris materials at the depositional area.
In this study, debris flow simulations were used to better understand with regard of predictions of volume, run-out mechanisms, flow rate of
materials and the distribution of debris materials. In general, the results of this study will be very useful to predict the run-out of any potential
similar debris flows in some areas of Indonesia which might experience the similar events as well as the determination of mitigation
measures to minimize the negative impact of these events.
Keywords: landslide, debris flow, run-out, modeling.
1. INTRODUCTION Debris flows always occur in an area which has hill morphology
with high intensity of rainfall. Slope angle and river stream are
The occurrence of several landslides followed by the debris flow, factors that determine the process of debris flow. Debris flow will
such as Sijeruk Landslide on January 4, 2006, Jemblung Landslide happen if a slope has huge amount of debris material with huge
on December 12, 2014 and Pasir Panjang Landslide on Februari 22, amount of water supply as its stream. Debris flows always occur
2018 has make a lot of casualties and terrible destruction in some after or when heavy rain. Debris flows has high specify gravity
parts of Central Java. which make a boulder flow through the stream to the toe of
From Stiny (1910) in Jacob and Hungr (2005), debris flow depositional area.
begins with the description of a flood in a mountain torrent, carrying
suspended load and transporting quantities of bedload. As the
2. METHODS
amount of sediment carried by the flow increases, at certain limit it
has change into a viscous mass consisting of water, soil, sand, This study will focus on run-out modeling of debris materials, since
gravel, rocks and wood mixed together, which flows like a lava into the flow of debris materials started to distribution of debris materials
the valley. Meanwhile, Varnes (1978) in Blijenberg (2007) describe in the depositional area, which is usually found in the plain areas of
that debris flows are mass movements consisting of granular solids, foot slope. Some of debris flow events in Central Java are selected
water and air moving as a viscous flow. to be modeled in this research by mean of numerical simulation
In general, debris flow zonation can be divided into three parts, technique with Graphical User Interface (GUI). The software that
which are source area, flow track and depositional area of debris used in this simulation is kanako ver.2.01. From Takahashi and
materials. The Geometri of debris flow can be seen in Figure 1. The Nakagawa (1991) in Nakatani, et al. (2008), the modeling concept in
first part is source area which an area where the accumulation of 2D based from the equations for momentum, continuation, riverbed
debris material occurs. The accumulation of debris materials might deformation, erosion/deposition and riverbed shearing stress.
be consist of materials of slope itself or material that make natural Input data parameters obtained from field conditions and the
dam. The second part is flow track which an area where debris results of physical properties analysis of materials, geomorphology
materials flow through stream to the valley. The third part is conditions and hydrogeological conditions. These parameters are
depositional area which a plain for sedimentation of debris used to modeling source area of debris flow. Shape of river was
materials. The Distribution of debris flow in depositional area used as analogy of flow track. These parameters which used as input
depends on topography. If in the depositional area still has a data can be seen in Table 1. Based on input data parameters, a series
difference in slope angle with slope below, secondary flow track of numerical simulation techniques performed by several stages.
maybe occurs which will make secondary depositional area First stage is conversion contour elevation into the form of a grid
(multiple depositional area). system with a scale adapted to the area affected. Second stage is
define engineering properties of rocks as the model’s variable input
and flow tracks. Third stage is analysis of supply hydrograph for
indicating the influence of water that would affect flow behavior.
Last stage is simulate to predict the volume, run-out mechanisms,
the flow rate, as well as the distribution of debris materials that
formed.
Table 1 Input Parameters
Parameters Unit
Simulation Duration second
Calculation time interval second
Diameter of materials m
Mass density of bed materials kg/m3
Mass density of fluid phase (water, mud, and silt) kg/m3
Consentration of movable bed unitless
Gravity acceleration m/s2
Figure 1 Block diagram of debris flow geometry Coefficient of erosion rate unitless
(Modified from Nettleton et al., 2005) Coefficient of accumulation rate unitless
357
Table 1 Input Parameters (Cont.)

Internal friction angle °
Minimum depth at the front of debris flow m
Minimum flow depth m
Manning’s roughness coefficient unitless
π (Phi) 22/7
Consentration of sediments in the slope when debris flow occurs

determine by using equation 1 (Takahashi et al., 2001 in nakatani, et
al., 2011):
Cd=(ρ tan θ)/((σ- ρ)(tan φ-tanθ)) (1)
where: Figure 2 Topographic area around Pawinihan Hill

σ = Mass density of bed material
ρ = Mass density of liquid
φ = Internal friction angle
θ = Slope angle of river
Cd = Consentration of debris flow
If the consentration of sediments have been determine, the value

of peak of sediments supply in debris flow can be calculated using
equation 2 from “Sabo Master Plan for Debris Flow” (NILIM Japan,
2007 in Nakatani, et al., 2011):
Qsp =0.01 x ΣQ (2)
ΣQ = (Vdq C*)/Cd (3)
where:
Qsp = Peak of the sediment supply per second (m3/second)
ΣQ = Total amount of moveable material that define in Figure 3 The distribution of debris flow with viscosity 0.45 at 891
equation 1 (m3) seconds that seen from southeast
Vdq = Volume of the sediment (m3)
C* = Consentration of moveable bed
3. CASE STUDY
3.1 Sijeruk Village, Banjarnegara, Central Java, January 4
2006
Debris flow occurred in Pawinihan Hill, Sijeruk Village,
Banjarnegara Regency at coordinates S 07° 19 '16.1"and E 109° 42'
08.3'' with elevation 994 m. Over the incident, as many as 77 people
were killed, 8 people missing. Pawinihan Hill composite rock
consists of flow breccia flow, pyroclastic, lava, and alluvium
(Condon W.H., et al., 1996). The slopes of Mount Pawinihan have a
slope of between 20° to 60°. The direction of landslide at the crown
on the slopes of Pawinihan Hill is towards the southeast (N 145° E).
In the former landslide area, remnants of breccias that experienced
landslide still visible.
Debris flow in the Sijeruk Village has uniqueness in the flow
path, which is encounter a sharp deflection. It is alleged that the
deflection is caused by a ridge that blocks the flow track of the
flowing material that move southeast. The ridge causes the direction
of the flow that originally toward the southeast transformed to the
east. The depositional area tends to the southeast following the slope Figure 4 The distribution of debris flow with viscosity 0.45 at 891
of the valley of Pawinihan Hill. The topographic of the Pawinihan seconds that seen from top
Hill can be seen in Figure 2 with 1 grid representing the 25x25
meter area. The length of debris flow in Sijeruk Village reached 3.2 Jemblung Village, Banjarnegara, Central Java, December
about 1.1 km. 12 2014
The modeling of the debris flow using scenario of the viscosity The location is on Jemblung Village, Banjarnegara Regency which
value 0.45. The process of direction deflect in flow track and the has steep topography that has angle reached 75°. Landslide occurs in
distribution of sediment in the deposition area can be seen in Figure the southern slope of Jemblung Village with coordinate 7° 16' 51.6"
3 and 4. When the material enters the flow track, material moves at S and 109° 43' 12.6" E. The victims died reached 95, 24 people
speed 21.17 km/hour. The flow of materials is slowly moving in the missing, and dozens more injured. From December 9 to 11,
deposition area at speed 5.4 km/hour. The time it takes the debris Banjarnegara has high intensity of rainfall reached 100 mm.
flow material to reach the depositional area is 891 seconds. The Material compilers in the slopes of Jemblung Village are Jembangan
volume of debris flow material which flow reaches 127,735 m3.
358
Volcanics (Qj) which consists of Andesitic lava and volcaniclastic 3.3 Pasir Panjang Village, Brebes, Central Java, February 22
(Condo et al., 1996) and have largely been strongly weathered. 2018
These materials became source of the debris material. Figure 5
shows the debris flow area and surrounding area with 1 grid Landslide incident occurred in Pasir Panjang Village, Brebes
representing an area of 14.55 (west-east) x 15.31 (north-south) Regency with coordinates 108° 47' 19.68" E and 07° 7' 6.59" S.
meters which seen from northeast. Victims that recorded are 18 peoples who work as farmer in the
southeast landslide location. Landslide occurs on hills morphology
with steep slope. The hill consists of Tapak Formation which
composed of greenish sandstone in the lower part that gradually
grades upward into greenish sandstone with some intercalation of
grey to yellowish sandy marl (Kastowo, 1975). The Landslide that
flows toward southeast in slopes that have angle between 35° - 45°.
Soil cover in the landslide location is light-reddish brown sandy clay
with 5 m depth on top of dark grey marl. Figure 8 shows the
geometry of the debris flow from southwest direction and
topography around Pasir Panjang Village with 1 grid representing
area 29.57 (east-west) x 25.35 (north-south) meters. The length of
debris flow in Pasir Panjang Village reached about 2.3 km.
Figure 5 Topographic area northern Jemblung Village
The viscosity value of 0.5 became the closest scenario to

Jemblung debris flow in 2014. The higher viscosity of the fluid is
due to the volume of water flowing much more than the sedimentary
material carried by the water stream. The sediment material carried
by the debris flow has a relatively small size. Figure 6 and 7 show
the distribution of debris flow material seen from the northeast and
top. In the flow track, material speed is estimated to reach 66.14
km/hour. The volume of material flowing into the deposition area is
estimated to reach 111,962 m3 with an average speed of 16.53
km/hour. The time it takes the material to reach the deposition area
is estimated for 420 seconds. The length of debris flow in
JemblungVillage reached about 1 km.
Figure 8 Topographic area northern Pasir Panjang Village
The scenario viscosity value of 0.38 shows the most closely of

debris flow conditions. Materials flow with an average flow rate
8.68 km/hour on deposition area. Meanwhile in flow track, the flow
rate reaching 25.58 km/hour. The volume of debris materials that
flow from the slope is estimated about 275,295 m3. In the deflection
area, strong lateral erosion increase supply of debris materials.
Figure 9 and 10 show the distribution of debris flow from the
southwest and from the top. Depositional area which located
southeast form the landslide location used as rice fields. The time
Figure 6 The distribution of debris flow with viscosity 0.5 at 420 needed for debris materials to reach the depositional area is
seconds that seen from northeast estimated for 828 seconds.
Figure 7 The distribution of debris flow with viscosity 0.5 at 420

seconds that seen from top which have been overlayed image after Figure 9 The distribution of debris flow with viscosity 0.38 at 828
landslide from Pleiades (2004) seconds that seen from southwest
359
In Jemblung landslide, debris flow has higher viscosity of the

fluid is due to the volume of water flowing much more than the
sedimentary material carried by the water stream. The sediment
material carried by the debris flow has a relatively small size.
In Pasir Panjang landslide case, it has the longest length of
flowing material. It occurs because the volume material which
flowing is very huge supported with relatively straight flow track in
35° - 45° of slope angle. Although it has smallest viscosity of all
cases.
In general, the results of this study will be very useful to predict
the run-out of any potential similar debris flows in some areas of
Indonesia, especially in Central Java which might experience the
similar events as well as the determination of mitigation measures to
minimize the negative impact of these events
Figure 10 The distribution of debris flow with viscosity 0.38 at 828 5. REFERENCES
seconds that seen from top which have been overlayed with Centre
of Volcanology and Geological Hazard Mitigation (CVGHM) and Blijenberg, H.M. (2007) "Application of physical modelling of
Geospatial Information Agency (GIA) aerial mapping (2018) debris flow triggering to field conditions: Limitations posed
by boundary conditions". Engineering Geology, 91, pp 25-33.
4. CONCLUSION Centre of Volcanology and Geological Hazard Mitigation
(CVGHM) and Geospatial Information Agency (GIA) (2018)
All of the debris flow simulations were used to better understand "Mosaic Aerial Image, Recording dated February 26, 2018".
with regard of predictions of volume, run-out mechanisms, flow rate Condon, W. H., Pardyanto, L., Kentner, K. B., Amin, T. C., Gafoer,
of materials and the distribution of debris materials. Table 2 shows S., and Samodra, H. (1996) "Geologic Map of The
the viscosity, flow rate, volume of material and the length of Banjarnegara and Pekalongan Quadrangle, Java, Scale 1 :
flowing material closest to debris flow conditions of all cases. 100,000", Geological Research and Development Centre.
Table 2 Summary of simulations were closest Jacob, M., and Hungr, O. (2005) Debris-flow Hazard and Related
to debris flow conditions Phenomena. Praxis Publishing Ltd., Chichester.
Kastowo (1975) "Geologic Map of the Majenang Quadrangle, Java,
Sijeruk Jemblung Pasir Panjang Scale 1 : 100,000", Geological Research and Development
Results Centre.
Landslide Landslide Landslide
Viscosity 0.45 0.5 0.38 Nakatani, K., Wada, T., Mizuyama, T., and Satofuka, Y. (2008)
Flow rate in "Development of “Kanako” a wide use 1-D&2-D debris flow
flow tract 21.17 66.14 25.58 simulator equipped with GUI", 2nd Intl. Conf. on Debris
(km/hour) Flow, The Wessex Institute, UK.
Flow rate in Nakatani, K., Wada, T., Matsumoto, N., Satofuka, Y., and
depositional 5.4 16.53 8.68 Mizuyama, T. (2011) "Development and Application of GUI
area (km/hour) Equipped 1-D and 2-D Debris Flow Simulator, Applied to
Volume of Mixed-Size Grains", 5th International Conference on Debris-
127,735 111,962 275,295 Flow Hazards Mitigation: Mechanics, Prediction and
material (m3)
Length of Assessment, Itali, pp 735-744.
flowing material 1.1 1 2.3 Nettleton, I. M., Martin, S., Hencher, S. and Moore, R. (2005)
(km) Debris Flow Type and Mechanism in Winter, M. G.,
Macgregor, F. and Shackman, L (editor) "Scottish Road
Debris flow in the Sijeruk Village has uniqueness in the flow Network Landslide Study", Scottish Executive, Edinburg.
path, which is encounter a sharp deflection. It is alleged that the Pleiades (2004) "Pleiades Image, Date: December 16, 2014 (After
deflection is caused by a ridge that blocks the flow track of the Landslide)", Centre National d'Etudes Spatiales.
flowing material that move southeast. The ridge causes the direction
of the flow that originally toward the southeast transformed to the
east.
360
Channeled Landslide Protection Using Flexible Barriers -

Learning from more than 10 Years of Experience
T. Hangartner1, D. Matthias2, R. Prosida3, and B. Christophe1
1
Geobrugg AG, Geohazard Solutions, Switzerland
2
EMDES Environmental Engineering, Switzerland
3
Geobrugg AG, Geohazard Solutions, Indonesia
E-mail: Thomas.hangartner@geobrugg.com
ABSTRACT: Several flexible ring net barriers against debris flow have been installed worldwide during the last 10 years mainly for
retention and a few for erosion control in the way by reducing the inclination of the river bed. The increasing number of projects showed the
economic benefit of this solution. Special applications to retain big volumes with single barriers have been successfully installed in
Switzerland and Japan. This paper gives examples of different special applications like one big barrier providing big retention capacity,
several barriers in line or filled up barriers for erosion control. Advantages and challenges for the use of flexible ring net barriers are
discussed on a technical and economic level. Furthermore, needs for maintenance and replacement works are addressed.
Keywords: landslides, debris flow, protection, barriers, experience, maintenance
1. INTRODUCTION
Since 2005, over 250 flexible debris flow barriers have been
installed in more than 25 countries. Between 2005 and 2008, full
scale experiments at the test site Illgraben in Switzerland proved the
feasibility of retaining debris flows. The efficiency of some of the
first reference projects, mostly installed in Switzerland, was
analysed and a load design was then established together with the
Forest, Snow and Landscape Federal Institute (WSL). Standard
systems were then developed with the simulation software FARO.
Data from real-scale testing were used to verify and calibrate the
software outputs. Following this development, the flexible ring nets
became increasingly an alternative to classical debris flow
protections in Europe, USA and South America. In large scale
projects, where nets were installed in a row in the same channel, the
efficiency of retaining large volumes and the feasibility of this type
of installation in a row were proven as well. The nets are
appreciated, by designers and engineers, as a practical and
economical addition or alternative to existing classical debris flow
protections. Ten years of experience with flexible ring net barriers
signify that their advantages have been recognised and their
efficiency in the field have been established. The increasing
knowledge of single barriers, barriers in a row and large-scale
barriers have allowed to understand the advantages but also the Figure 1 Testing of debris flow retention system with ring net in the
limits of such a netting system for debris flow retention. This Illgraben channel, 2006. Retention volume approx. 1000 m3
acquired knowledge is presented in the following paper,
accompanied by case studies. On the dimensioning side, the weight acting on a debris flow net
during an event were better understood, thanks to an extensive
2. REAL-SCALE TESTING AT ILLGRABEN, measuring concept on and around the system (Wendeler, 2006),
DEVELOPMENT OF STANDARD BARRIERS AND which lead to the final dimensioning concept (Wendeler, 2008).
CE-MARKINGS
2.1.1 Remarks about 1:1 field tests results
2.1 Real-scale testing at Illgraben
Based on the described test site and recording facilities, real impact
Between 2005 and 2008, real-scale testing was conducted in the
forces on flexible barrier systems could be derived and used for the
Illgraben debris flow channel in Wallis, Switzerland (Wendeler,
development of a load model for the interaction between debris
2008). Prior testing it was observed that rockfall protection nets
flows and flexible, permeable barriers. Herein, the measured rope
were retaining some slides but the dimensioning concept was
forces during impact and overflow of debris flows present the most
missing to prove that flexible ring nets could retain larger debris
important results of the test barriers. An example is given in Figure
flows in a channel without sustaining damage. In Illgraben, a middle
2. Clearly, one can follow the filling process with the increasing
to large debris flow is occurring at least once a year naturally and
rope forces. Sudden load reductions show the activation of the brake
therefore a flexible ring net could be tested yearly (Figure 1).
elements. Even large blocks could be measured by higher weight at
Two key characteristics were defined and analysed with testing.
the balance and later by single peaks in the rope measurements. An
On one hand, a single barrier could, depending on the channel
example of a large block can be seen in Figure 3.
geometry, retain over 1000 m3. On the other hand, it was observed
If a barrier stays in use fully filled e.g. for river bed stabilization
that over 10’000 m3 were flowing over the barrier without damage.
one should consider that the remaining barrier height in the middle
This led to planning and constructing a debris flow retention system
of the barrier is only ¾ of the original barrier height.
with several nets in a row to retain successfully most of the material.
361
Figure 4 UX debris flow barrier, with posts for wider stream

channels application. Example of the Trachtbach in Switzerland.
Additional kolk protection, rip-rap and lean concrete were placed
along the stream bed
2.3 CE-marking
The real scale testing was also basis for certifying all standard
barriers. Certification was achieved in 2017 (EAD document No.
340020-00-1062). The CE marking is based on a “European
Assessment Document” which defines precisely the suitability, the
type classification and yearly quality controls necessary to
Figure 2 Rope forces and the filling height over time for the filling correspond to a certain standard. This states that the products with
event in 2006 CE marking fulfil the European guidelines for product quality and
field appropriateness (ETA 17/0268-17/0276 and ETA 17/0439).
Figure 3 demonstrates also that a debris flow can overflow a
filled flexible barrier. The overflow material is the remaining 3. DIMENSIONING
volume of a surge exceeding the retention capacity of a single
barrier which passes the barrier without causing any damages. The 3.1 Results of laboratory tests
barrier shown in Figure 3 was overflown by several 10’000 m2 In general, it is difficult to compare such laboratory tests with field
without any damage to the barrier. situation. Usually the physical modelling only provides an
informative basis on typical tendencies. After a detailed analysis of
the performed test results with a dimensional analysis the most
important result is that the stopping process of the front behaves like
an impulse of a pressure surge. Slow and friction dominated fronts
hitting the barrier showed a reflected surge. These observations are
in support of the physical behaviour of a stopping pressure surge.
The maximum dynamic force is dependent upon the velocity
squared and the flow consistency. For muddy flows, the pressure
surge needs a lower pressure coefficient Cw than for granular and
viscous flows. Additionally, flow density influences the pressure
Figure 3 Large block (left) flowing over the balance and afterwards experienced by the barrier linearly. Faster traveling fronts are
over the barrier (right) deviated vertically when they interact with the barrier and result in
material overtopping the barrier without filling it. The same effect
2.2 Development of standardized barriers also occurs for rigid barriers resulting in less retention capacity. This
observation could be confirmed because the rigid barriers were
The dimensioning concept as well as the distribution of the loading greater affected by smaller impact forces than the more flexible ones.
on the net were integrated in the finite element software FARO It therefore can be concluded that the flexible barriers are able to
(Volkwein, 2004) and first projects, mostly in Switzerland, were stop the debris flow whereas the rigid barriers can only deviate it.
dimensioned with it. Tests comparing the retention capacity of different mesh sizes in
Following the first projects, standard barriers were designed relation to the maximum particle size were conducted (Figure 5).
with a given load capacity in kN/m2. VX barriers are conceived for This enabled the determination of a good retention behaviour with a
channels up to 15m in width und barrier height of up to 6m, taking mesh size as big as the d90-grain size (90% of the grains are smaller
loads up to 160 kN/m2. UX-barriers find their application in larger than d90).
channels, are installed with additional posts, a barrier height up to 6
m and taking up loads of 180 kN/m2 (Geobrugg, 2016, Figure 4).
The dimensioning concept for debris flows is now state of the art
and freely accessible through the software DEBFLOW on Geobrugg
website. After registration on the website, everybody can use this
software and produce a first estimate for the dimensioning of a
barrier.
Figure 5 Influences of different mesh sizes on the retaining volume
362
3.2 Load model

In nature, a debris flow fills a barrier continuously. As a
simplification, the developed load model for the flexible barriers
uses a time discretization and considers the debris flow impact as a
series of several surges which can be said to move on top of any
previous surge stopped by the barrier system. The height of the
single surges corresponds to the calculated flow height of the debris
flow. Hence, the barrier is filled after many surges that result from
the barrier height divided by the debris flow height. The drainage
process of stopped material is enhanced through the added weight of
additional surges. The load model and design concept given by
Wendeler et al, 2006, consider usability aspects, actual valid codes
and standards, debris flow intensities, annularity and fail
probabilities and corresponding consequences. Figure 8 Overflowing process of an overtopping surge after the
filling of the barrier
3.2.1 First impact
Two components act upon the barrier during the stopping process of 3.2.4 Rope and anchor loads
the first wave: a hydrostatic and a dynamic pressure. These depend The loads acting in the ropes and transmitted to the anchorage
on the velocity of the flow squared, the flow consistency and its finally result from the theory of the rope equation. For that the
density (Figure 6). acting pressures should be divided over the single support rope
bundles. The parameters that can be used to solve the rope equation
iteratively with Newton iteration are for each rope bundle the rope
length and its section area times the rope modulus, the decisive rope
load in kN/m and the assumed sag of the rope alone together with
the elongation of the integrated brake rings. After a rope force has
been obtained the according elongation of the brake elements must
be re-checked with the brake ring characteristics. If the difference to
the initially assumed elongation is too large the rope equation must
be solved again using an adjusted elongation.
Figure 6 First impact of a debris flow and its loading components of
dynamic pressure (ΔP) and hydrostatic pressure (Phyd) 3.3 Special load case scenario such as snowslide and rockfall
In certain cases, mostly very steep slopes (>35°) and at high altitude,
3.2.2 Filling process snow slides, small avalanches or rockfall will be encountered which
After the first impact, the assumed following surges spill over any additionally impact the debris flow nets. An example of this
previous now stopped material. Therefore, the dynamic component situation is the multiple barrier setup in Hasliberg in Switzerland.
only interacts with the barrier area above the stopped material. The Some of the barriers are situated above 2000 m in elevation. Since
hydrostatic pressure, on the other hand, acts over the entire debris flexible net barriers are also used as a protection against avalanches
material caught in the net (Figure 7). and rockfall, a certain degree of combined loading can be
guaranteed. The combined loading can be calculated and a barrier
dimensioned for every special case with the use of FARO simulation
software (Volkwein, 2004). Specific components of the debris flow
barrier can be individually reinforced depending on the simulation
results (Wendeler, 2014). Figure 9 illustrates the simulated load case
for barrier number 2 in Hasliberg in a situation of a lateral avalanche
impact, with an angle of 10° and a load of 120 kN/m2.
Figure 7 Second surge hiding the barrier while the filling process.
The dynamic component acts one surge higher
Figure 9 FARO simulation software output when barrier number 2
3.2.3 Overflowing is impacted by an avalanche in Hasliberg, Switzerland
If a barrier is filled completely, any following surges spill over the
top of the barrier adding an additional load to the system by both In this special case, the upslope guy wires are loaded up to 70%
their weight (σ) and the acting shear forces (т). The shear force is of their capacity. Figure 10 shows the snow load on the barrier in
winter.
usually ten times smaller than the normal force and is neglected in
An easy predetermination of the dimensioning of a standard
this loading approach for flexible barrier design. The retained
material behind the barrier changes now from the hydrostatic state to barrier up to 6 m in height can easily be performed with DEBFLOW
an active earth pressure for wet material. The time it takes to drain software. A more complicated scenario can still be dimensioned by
Geobrugg or WSL with FARO simulation software. A few special
the material depends on the debris flow composition and the water
cases regarding construction are described in section 6.
content. Figure 8 illustrates the loads that must be applied.
363
4.2 Reuse of the anchoring after a debris flow event

Without additional flank stabilisation, a certain degree of washing
out can be observed along these stream banks, especially in loose
soil (figure 12).
When exchanging the net, the anchoring can technically be
reused when the top of the anchor is cut off, a loading test is
performed and a new flexible anchor head is mounted. If the anchor
length was drilled the first time with a safety factor and possesses a
certain length in reserve. In the case of frequent filling of the net it is
recommend designing the anchors with sufficient length or to
prevent the washing out of the banks with structural counter-
measures.
4.3 Structural countermeasures: protection of the banks in

stream beds
Especially in bends along the stream, the washing out of the outer
bank and its erosion are prevalent when a debris flow occurs. The
amount of erosion is dictated by the volume and the velocity of the
Figure 10 Barrier partially snowed in during winter. The snow load
flow. Depending on the project a reinforcement of the outer bank
must be considered for designing
should be considered (rock blocks, wall, gabions or additional flank
stabilization by netting with or without erosion control mats. It is
4. CONSTRUCTION ASPECTS
important to consider that the shearing forces of a debris flow are
4.1 Subsurface and anchoring much higher than of water and this must be incorporated in the
design calculation for the protection measures.
While the netting itself is easy to model and to dimension, safe
anchoring is more complicated (figure 11). Ideally, a detailed
geological profile of the section to be protected is available as well
as the geotechnical parameters of the subsurface. Having the
possibility to perform pulling tests on the soil nails to assess the
friction between the subsurface and the grout is another advantage.
Debris flow deposits are heterogeneous in nature and deposited
along the sides of the channel affecting the subsurface quality for
anchoring. The dimensioning of anchor forces need to be
determined by experts in those cases. It is as well recommended to
use self-drilling anchors with a flexible anchor head. The barrier
when loaded is largely deformed and the forces of the ropes on the
anchors can change up to 30° in angle. This eccentricity without
flexible anchor head is often not bearable for a normal threaded
Figure 12 Slope stabilisation with TECCO® for flank stabilisation
anchor since the pushing resistance is much smaller than the pulling
when installing a debris flow barrier in loose material
component.
4.4 Kolk protection
When barriers are filled or partially filled, the material downstream
must be directed back into the original stream bed. This is especially
important for barriers retaining a debris flow in an open field rather
than in the stream bed itself. When planning, it must be considered
whether a field needs an artificial channel back to the stream. The
necessity for a protection against kolk must be established, for
example with anchored rock blocks. When choosing this solution,
the dimensioning of the lower nets should consider the possibility of
these rock blocks getting torn away. The additional load being
potentially fatal to the barrier.
5. PLANING ASPECTS
Often debris flow barriers are installed close to the source zone of
the debris flows while greater structural measures such as a retaining
basin or deviation measures are constructed further down.
Net barriers and large steel and concrete construction can
therefore be perfectly combined. The advantages of both methods
can be specifically used together. Examples of this combination are
the streams Trachtbach in Brienz and Milibach in Hasliberg, both in
Switzerland. In both projects, the combination of the nets upstream
and the larger construction measures downstream allowed to
increase the retained mass upstream and diminish the erosion in the
stream bed. Therefore, the capacity of the concrete protection
Figure 11 Washed out anchoring of the debris flow barrier number measures could be lowered and constructed at smaller scale and
25 in the Illgraben channel. Anchoring partially in loose material existing protection structures were easily and cost effectively
and partially in disused concrete debris flow barrier renovated and added to the protection measures series.
364
5.1 Protection nets as an immediate solution 5.3 Passages for small animals and greening
Protection nets installed in the source zones of debris flows, slow The relatively large openings of the ring nets allow for passage of
these down, which allows for longer warning and evacuation time in small animals, when the barrier is not filled, even fishes when the
the endangered areas. This is especially of importance in small barrier is immersed in water, in contrast to a concrete structure
catchment zones where debris flows are rapid and travel along short (Wendeler, 2008). There are examples where this was an expressed
distances only. The easily installed protection nets are therefore wish of the developer. Ring nets are as well appropriate for greening
practical for an immediate protection solution. They increase the and blend perfectly into the landscape.
safety of the infrastructure downstream and even allow for the
protection of the construction crew building a retaining basin for 6. DIFFERENT TYPES OF DEBRIS FLOW BARRIERS
example. These protection nets can be equipped as well with a
warning system (more details are given in section 7). 6.1 Single barriers
Most barriers installed are single barriers along roads and railway
5.2 Visual and landscape protection aspects tracks or above settlements (Figure 15).
Debris flow protection nets instead of concrete dams are more and
more an alternative regarding landscape protection and visual
aesthetics. The filigree design is almost invisible from far away and
a primary argument for protection measure construction in
landscape protection zones. An example is the UNESCO World
Cultural Heritage along the Rhine close to Koblenz (Figure 13). At
the back of the village debris flow nets are installed and even with
one barrier partially filled in 2017, the nets are still barely visible
but fulfilling their purpose (Figure 14).
Figure 15 Debris flow barrier in Isenflue above a settlement. The

outer bank of the stream was reinforced with a rock wall
6.2 Barriers in a row (multi-level barriers)

Debris flow nets can be installed in a row, to increase the retained
volume. The first multi-level barriers were installed in Merdenson in
Figure 13 Almost invisible debris flow barrier close to Koblenz Switzerland for observational purposes by the WSL (Denk et al.,
along the Rhine above an UNESCO World Culture Heritage 2008). Subsequent laboratory tests to analyse the overflow
protected village behaviour, and more specifically the overflow velocity evolution
during a flow, confirmed the developed load design for multi-level
Additionally, environment friendly building and sustainability is barriers (Wendeler et al., 2010). Examples for this setup are the
more and more an important argument for construction. For example, multi-level barriers in Hasliberg and Menderson (Wendeler et al.,
a debris flow barrier (ten by 4 meters) is 30 times lighter than a 2014) in Switzerland but also in Portainé in Spain (Luis et al., 2010)
concrete barrier of the same dimensions, making it the ‘greener as well as Chosica in Peru. Most of the multi-level barriers have
solution’. On top of that with less weight, less carbon dioxide is already been successfully filled during events (Figure 16). Chosica
emitted during transport to site (Wendeler, 2008). is the most recent example in 2017, protecting efficiently several
cities built downstream (Figure 17).
Figure 14 partially filled debris flow barrier above the German Figure 16 11 debris flow protection barriers, successfully filled in
Railway close to Koblenz Hasliberg in 2011
365
Figure 19 Another special construction acting as a debris flow

barrier in Sitäbach consisting of concrete slices piled up and netting
mounted in between
Figure 17 Filled debris flow barrier in 2017 in Chosica Peru, 8. MAINTENANCE AND CLEANING OF BARRIERS
protecting successfully a large city downstream As any protection structure, debris flow barriers require
maintenance from time to time. It is recommended to undertake
6.3 Large debris flow retention with single barrier (special regular, for example yearly, checks of the protection system if no
construction) event (debris flow, slides, …) occurred during that time span.
In special cases, an adapted design higher than 10m and larger than Working with a checklist and a maintenance scheme, such as for any
40m can be constructed. A typical example is the debris flow barrier other protection structures, should facilitate regular controls. After
in Hüpach, next to Oberwil in the canton Berne in Switzerland an event, the barrier needs emptying and replacement of certain
(Berger et al., 2016). This barrier has a retaining capacity of more components. A filled barrier can for example be cleaned from
than 12’000 m3. Such a construction necessitates strong abutments behind with an excavator. It is essential, when planning for the
of steel reinforced concrete, long anchors and needs special ropes system, to consider what happens to the material of the debris flow
used for cable cars which need precise adjustment (Figure 18). and to organise a deposit area. Budget wise, it should be considered
Special calculations for the netting and the ropes, adjustment to the that after a fully filled barrier, parts should be replaced, whereas the
anchoring and special foundation engineering in exposed terrain was anchoring can often be reused, as explained earlier. A net can be
necessary to complete the project. emptied from the front when certain conditions are fulfilled. The
material of the debris flow should be dry and stable and the netting
must be stabilized upslope and safety aspects for the working crew
should be respected.
9. ADVANTAGES AND LIMITS OF FLEXIBLE

NETTING FOR DEBRIS FLOW PROTECTION
The main advantages of these systems are their relative low weight
and rapid installation. Especially in steep and in terrain difficult of
access. The materials can be transported by hand or with helicopters
wherever construction machines cannot reach the site or where it
would not be economical. Ring nets can be used for immediate
protection in endangered zones to safeguard the construction of a
permanent structure below. These practices are common for
example in Japan. Ring nets can therefore be incorporated in an
overall protection concept for an entire catchment area. At the same
time, it has been proven over time that ring net barriers are fully
Figure 18 Special construction of a debris flow barrier in Hüpach, equivalent to large concrete structures when properly planned, with
in Switzerland, with a width of 40 m and a netting height of 10 m an erosion control concept and an established maintenance plan.
The decision to install a large retaining structure with netting 10. CONCLUSION
was based on the topography, the difficulty of access and lack of
Since the publication of the load design of flexible protection nets
alternatives to protect the village below. The debris flow barrier has
and their appropriateness tests in the Illgraben in Switzerland, many
not been filled yet. Another special construction is situated in
projects have been successfully installed in the last 10 years. Several
Sitäbach along the stream Lenk, in Switzerland. The construction is
construction details have been revised and improved. When
based on concrete slices and netting in between (Figure 19).
considering the hydrological processes affecting the stability of the
stream banks and planning for reinforcement, the flexible ring net
7. SURVEILLANCE
systems can be considered as equivalent to classical large concrete
Protection nets can be monitored with sensors (Sentinel System). In protection structures. The lighter conception of the barriers makes it
larger systems, some components can be monitored such as the ring an unavoidable solution when easy handling, environmental
brakes and when a loading threshold is reached, an alarm is requirements and landscape protection are key issues of a project.
triggered. An example is the debris flow net, installed as an The dimensioning concept developed at the WSL, in use worldwide,
immediate protection solution, in Magnacun in Switzerland. The has been verified by several fillings and successfully retaining
railway tracks of the Rhaetian Railway are perfectly protected since events. A further adaptation and refining of the dimensioning
2009, with the surveillance system working faultlessly, according to concept could be achieved with more testing, but is hampered by
the developer. lack of funding.
366
REFERENCES Volkwein A. (2004): Numerische Simulation von flexiblen

Steinschlagschutzsystemen, Dissertation ETHZ, Schweiz.
Berger C., C. Wendeler, L. Stieglitz and G. Lauber (2016): Wendeler C., B.W. McArdell, D. Rickenmann, A. Volkwein, A.
Examples of debris retention basins combining concrete and Roth and M. Denk (2006): Testing and numerical modeling
net structures, Interpraevent Luzern, Switzerland. of flexible debris flow barriers. In Zhang, M. and H. Wang
Denk M., A. Roth, C. Wendeler und A. Volkwein (2008): 1:1 (eds.): Proc. Of the sixth International Conference on
Feldversuche für flexible Schutznetze gegen Murgang – Physical Modeling in Geotechnics, pp. 1573-1578. Balkema.
Versuche, Bemessung, Anwendung, Publikation für die Wendeler C. (2008): Murgangrückhalt in Wildbächen – Grundlagen
Technische Akademie Esslingen, Deutschland. zu Planung und Berechnung von flexiblen Barrieren,
Geobrugg (2016): Ringnetzbarrieren aus hochfestem Stahldraht: Die Dissertation ETHZ, Schweiz.
ökonomische Lösung gegen Murgänge, Schweiz. Wendeler C., J. Glover (2014): Multiple load case on flexible
Luis-Fonseca R.., C. Raimat, J. Albalate and J. Fernandez (2010): shallow landslide barriers – mudslide and rockfall, IAEG
Proteción contra Corrientes de derrubios en areas del Pirineo. Conference Turin, Italy.
Obras Urbanas, Julio/Agosto 2010 número 22, Spain. Wendeler C., A. Volkwein, A. Roth and N. Nishimura (2014):
Speerli J., R. Hersperger, A. Roth and C. Wendeler (2010): Physical Successful hazard prevention using flexible multi-level
modeling of debris flow over flexible ring net barriers, barriers, Interpraevent in Nara, Japan.
Conference on Physical Modeling in Geotechnics ETHZ,
Switzerland.
367
Research on Failure of Aeolian Sand Roadbed Slope Through Laboratory

Static Load Test
Xiukun Dong1 , Liying Liu1
1
School of Civil Engineering and Architecture , Chong Qing University of Science and Technology, Chong Qing, China
E-mail:zhangyiluul@126.com
ABSTRACT: There is little study on failure of aeolian sand roadbed slope. Field test is more difficult, so the author prepares a
representative sample of aeolian sand, and makes the failure of aeolian sand roadbed slope through static load test simulation in laboratory to
analyze the aeolian sand roadbed slope stability. Test results indicate that the roadbed slope is stable when the roadbed slope ratio is 1:2 and
test load is under critical load, the aeolian sand is compacted by water dumping and mechanical compaction or mechanical compaction. The
allowable bearing capacity of roadbed increases significantly when the slope ratio is 1:2 and especially the roadbed is compacted by the
water dumping and mechanical compaction.
Keywords: aeolian sand, roadbed slope, static load test, slope stability.
1. INTRODUCTION (3) According to the position shown in Figure 2 to install

displacement meter, including 1 # measured Point test bearing plate
There are many desert area in China, which distributed mainly in the vertical displacement; 2 #, 3 # measuring points were tested side line
north-western region arid area, the Qinghai-Tibet Plateau area, Such vertical and lateral displacement, 2 # measuring point reading is
as Xinjiang, Inner Mongolia, Gansu, Ningxia, Shaanxi, Qinghai, positive, said Measuring point displacement upward, on the contrary
Liaoning, Ji Lin and Heilongjiang provinces and regions. With the displacement down; 4 #, 5 # measuring points respectively Measure
western development there will be a large number of projects are the vertical displacement 50 cm from the top of subgrade and the
being built in these areas, especially road construction projects. bottom of subgrade.
The stability of aeolian sand embankment slope is to ensure the (4) Making the reaction frame is at a predetermined position. Place
aeolian sand overall stability of subgrade slope to prevent the slope the load plate in the position shown in Figure 2 and to keep the
slipped and the foundation instability as well as local subgrade baring plate is level.
depression or push deformation. The second aim is to ensure that the
road base slope, shoulder and both sides of the environment stability, uncompaction
to prevent the subgrade sand buried, shoulder and slope were wind
50cm
roadbed filling
erosion, rain erosion and man-made damage. The design of desert
highway subgrade slope is necessary to design a reasonable slope condition 1
ratio, also to adopt reasonable and reliable protective measures. It is
difficulty to do destruction test of aeolian sand subgrade slope (Liu compaction
Dapeng, Li Zhiping), and the information of destruction test
50cm
roadbed filling
currently is lack. Therefore it is necessary to carry out simulation
tests indoors. The author choose a representative of aeolian sand condition 2
specimens, studying the sand subgrade slope under different Figure 1 Subgrade construction of two conditions
loadings and different compaction conditions in indoors. The
study’s purpose is to analyse the wind sand roadbed slope stability,
then to guide the design and application of roadbed slope Work. longitudinal center line
That is to ensure the stability of subgrade slope, to extend the useful
life of roads. The research have some economic and social benefits.
2. INDOOR TEST ON FAILURE OF AEOLIAN SAND

toe of slope
slope top
SUBGRADE SLOPE UNDER STATIC LOAD

In order to facilitate comparison and analysis of test data, this
simulation test use a circular bearing plate with a diameter of 30 cm
bearing plate
at a distance of 50 cm from the edge of the embankment. Load test
(simulation of roadbed test components shown in Figure 1 and 1
50cm 50cm
Figure 2) set the provisions when the bearing plate settlement reach
2
500 mm Load test stops. Simulation test on basis of subgrade filling
slope top 3
methods are divided into two different conditions: (1) The subgrade
4
is layered buried, and each layer thickness is 50 cm. The subgrade
have not used mechanical compaction, The simulated subgrade
slope ratios are 1:1, 1:1.5, 1:2.0 and 1:2.5 respectively; (2) The
simulation subgrade is divided into four layers, each layer of 50 cm 5
50cm
thick virtual filling. Using compaction machinery to rolling

subgrade and the slope ratio are 1:1, 1:1.5, 1:2.0 and 1: 2.5.
toe of slope
2.1 Experimental Procedure Figure 2 The monitoring instrument layout
(1) Leveling the test site and compacting the ground surface with a (5) Placing in the bearing plate jack and putting the ball seat at the
flat-panel rammer, the author put iron nail as a subsidence by the top of the load beam contact. Dynamometer should be placed in the
interval of 0.15 m into the soil, and measure its height. middle of the jack and the beam. The jack must be kept vertical.
(2) Filling the embankment stratified according to different working Then experimenters start to load with a jack and then relax the jack
conditions. Measuring subgrade compactness. Then continuing with throttle unloading, regulator 1 min. The displacement meter is
the layered packing until the desired elevation. initialized to zero.
368
(6) Jack load, using step by step loading method, is to measure force
design control load, is less than 0.1 MPa, each level increases 0.02
MPa, after each increase of about 0.04 MPa. In the experiments,
experimenter should pay attention to the top of the subgrade and
slope cracks. Then begin to the next scheduled load, and continue to
read displacement.
2.2 Experimental data

Simulating subgrade static load failure test and geotechnical test
specification. There are some differences in the plate load test:
Simulate the static load on the subgrade destruction test load time of
1 min, while the plate load test each level load is generally 2 h.
According to the compression test, it can be seen that aeolian sand is
made at load compression deformation with a very short period of
time, the latter part of the small deformation creep. Simulation At
the beginning of subgrade test, first read the 1 # meter reading
indicator at 1 min. number, reading again after 6 h, dial indicator
only changes 0.02 mm. The test loading time of 1 minute to meet
the plate load test the requirements of the test. The flat load test
generally refers to the role of rigid platens. Half-space surface or
near surface, and according to a certain theoretical system
foundation bearing capacity, the role of the test load located at the
roadbed from the edge Line 50 cm department. (Li Zhiyong) The
test data processing only consider the load plate Test data before
significant cracks appear around.
The test results show that when working condition 1 simulates
subgrade test, it is carried plate settlement larger, p-s curve no
obvious straight line and inflection point, mining The allowable
bearing capacity is obtained by the method of three-point method or
step-by-step regression fitting.
Are less than 110 kPa, the test results and analysis of the
operating conditions 2 p-s curve finishing. Method is as follows:
Draw the final p-s curve based on p and the corresponding s value
line. According to China's building foundation design specifications
(GB50007-2002): For low compressibility soil and sand, take s/b =
0.01 ~ 0.015 corresponding to the load of the basic bearing capacity
value.
2.3 Analysis of test results

2.3.1 The first condition
Figure 3 shows the displacement and load pressure of different slope
ratios in condition 1 of roadbed curve.
As can be seen from Figure 3, the displacement of each
measuring point increases as the load increases Large, the
displacement of the same measuring point decreases as the slope
ratio increases. 1 # measuring point Large amount of displacement,
the displacement of 500 mm when the corresponding external load
0.42 ~ 0.46 MPa; 2 #, 3 #, 4 #, 5 # measurement point displacement
smaller, With the increase of load, the displacements of 2 #, 3 #, 4 #
and 5 # points also increase obviously.
Figure 3 Measuring point displacement and load curve
When the slope ratio is 1:1, the displacement of each measuring

point increases sharply point, indicating that subgrade deformation
will occur in subgrade under this load shaped. With the load
continues to increase, the overall slope of subgrade slope will occur
moving, this and measuring point 2 and measuring point 3 appear to
increase with the load the corresponding displacement bounce; when
369
the load increases to 0.30 MPa, measured the displacement of point

5 increases from zero and the curve shows that the load continues
increase, the overall slope of subgrade slope occurs. After the end of
the experiment, simulation subgrade slope overall sliding surface
shape is not obvious, there are obvious uplift. Slope ratio of 1:1.5,
the initial load, 4 #, 5 # measuring point displacement close to zero,
with the load increased to 0.40 MPa, 5 # measuring point only then
began to increase, which shows that the simulated roadbed under the
action of load. First, local depression deformation occurs, with the
load continues to increase, the roadbed slope will slide overall. After
the test, simulate subgrade slope the overall shape of the sliding
surface changes are not obvious, the top of the subgrade obvious
long Start.
When the slope ratio is 1:2.0, the displacement of the top and the
top of the slope is vertical and lateral obviously. The horizontal
displacement of measuring point 50 cm away from the top of the
slope is smaller. 50 cm away from the bottom of the test point in the
upper load level increased to a certain value. The displacement
increased sharply, indicating that the simulated subgrade slope is on
the part of the bearing plate under the action of stress, there is less
possibility of overall slump, and the main subgrade local depression
deformation occurred; after the test, the subgrade slope rupture
surface not obvious, no obvious uplift slope.
Slope ratio of 1:2.5, the top of the slope at a straight upward and
lateral. The horizontal displacement is larger, the displacement at
measuring point 50 cm away from the top of the slope is smaller,
50 cm away from the bottom of the test point displacement remains
unchanged, indicating that the slope bearing lateral extrusion
deformation occurred under the action of the carrier, resulting in the
overall collapse of the test, the rupture surface of subgrade slope is
not obvious, only a small amount of cracks, no obvious uplift slope.
2.3.2 The second condition

Figure 4 shows the displacement and pressure of different slope
ratios in condition 2 of roadbed curve.
Figure 4 shows that the load plate initial displacement relative to
the conditions 1 significantly reduced, the displacement of the
measuring point with the pressure change is small. The overall
stability of the simulated roadbed slope is good; when the load is
over critical load (0.32 ~ 0.42 MPa), 3 #, 4 # measuring points The
displacement obviously increases, indicating that the top of the slope
soil slides as a whole with the increase of slope ratio, the failure of
slope is from strong to weak. The slope ratio is 1:1.0, the bearing
plate slides integrally with the slope, with a slope ratio of 1: 1.5,
cracks began to appear on the subgrade slope, and then the overall
slip occurred dynamic. Slope ratio 1:2.0, the slope of the overall
destruction of cracks. When the ratio is 1:2.5, the top of the slope
can be seen to bulge and the subgrade is partially depressed
deformation. Slip surface is not through the foot of the slope of the
plane (after the end of the experiment to verify).
Figure 4 Measuring point displacement and load curve
3. COMPREHENSIVE ANALYSIS
In condition 1, 1# measuring point displacement under load larger.
When the settlement is 500 mm, the corresponding external load
varies with slope ratio Increase the trend. Load ratio is 1:2.0 and
1:2.5 when the corresponding load containing the basic approach,
mainly due to the subgrade slope ratio increased to a certain extent.
370
When the top of the roadbed to improve the carrying capacity less. Based on the analysis of the stratified settlement of subgrade
Slope ratio 1:1 and 1:1.5, the subgrade is first sent under the stress under two conditions, the subgrade settlement of the simulated
of the carrier board deformation of local depression, with the embankment is larger when the conditions are filled. When the
increasing of external load, the slope occurred slip and slippery vibratory compaction the subsidence of the subgrade filler in the
surface through the foot. Slope ratio of 1:2.0 and 1:2.5, the surface simulated subgrade is small. These are caused for static load on the
of the slope is 50cm away from the top of the roadbed and the aeolian sand subgrade compaction less impact, vibration load on the
bottom of the roadbed. The measuring point of the horizontal subgrade packing density had a significant impact.
displacement relative to the first two when the slope ratio is smaller,
subgrade mainly under the action of the bearing plate partial 5. REFERENCES
depression deformation (Liu Cheng, Wang Xiaodong).
In condition 2, the initial load carrying capacity of the settlement Liu Dapeng, Yang Xiaohua, Wang Jing and Li Yuan. (2016)
plate relative to the work. Condition 1 significantly reduced. This “Experimental Study on Dynamic Stress-strain Relation of
shows the compacted subgrade under the action of less than its Aeolian Sand in Xinjiang Desert Region”, Journal Of Wuhan
critical load the overall stability is good, when the upper part University Of Technology, 38, Issues 3, pp71 ~ 374.
pressure is greater than the critical load, the top of the slope of the Li Zhiping. (2003) “Experimental research on the strength of lime-
overall collapse occurred. Slippery does not pass slope foot, load is flyash stabilized aeolian sand” Transportation and
different, the shape of the sliding surface is not the same. Second Technology, Issues 2, pp30 ~ 31.
subgrade slope than 1:1 when the load-bearing plate allowed to Li Zhiyong, Cao Yuanwen, Liang Naixing and Mei Yingjun.(2006)
carry force of 81 ~ 121 kPa, slope bearing capacity of 1:1.5 “Compaction mechanism of aeolian sand” , Journal of China
allowable 64 ~115 kPa. Slope is than 1:2.0 allowable bearing Highway Journal, 19, Issues5, pp 6 ~ 11.
capacity of 75 ~ 112 kPa, slope ratio 1:2.5 allowable bearing Liu Cheng ， Wu Guanling ， Zhang Yanzhao ， Yang Youhai.
capacity of 73 ~ 129 kPa. When the slope ratio is 1:2.5, the top of (2016)“Research on Features and Compaction Coefficient of
subgrade is under additional load sink deformation occurs, the slope Subgrade Aeolian Sand Filling for Heavy Haul Railway in
has obvious uplift (Zhang Yang). Mu Us Desert”, Subgrade Engineering, Issues 6, pp138 ~
141.
Wang Xiaodong.(2016) “Influence of The Perfomance of aeolian
4. CONCLUSIONS AND RECOMMENDATIONS Arenaceous Cemented Filling Materials on Response of
Water-Solid Ratio”, Coal Geology And Exploration, 44,
In summary, the static load on the slope of aeolian sand subgrade,
Issues 3, pp106 ~ 112.
the results of simulating indoor tests show that aeolian sand is in the
machinery Compaction or water fall plus mechanical compaction Zhang Yang ， Zhang Yanzhao ， Wu Guanling ， Yang Youhai.
state, slope ratio of about 1:2.0, the stability of subgrade slope is less (2017)“Compaction Characteristics of Aeolian Sand and
than the critical load. When slope ratio is 1:2.0 the subgrade slope Compaction Technology for Filling Subgrade of Heavy Haul
stability is not only good, but also allow the roadbed allowing Railway”, Railway Engineering, Issues 3, pp77 ~ 80.
bearing capacity increased significantly.
371
Stability Analysis of an Overall Failure Excavation Case in Hang Zhou

T. N. Do1
1
Department of Civil Engineering, Thuyloi University, Hanoi, Vietnam
E-mail: dotuannghia@tlu.edu.vn
ABSTRACT: In this paper, stability of an overall failure excavation case in Hang Zhou, China was analyzed using the finite element method
(FEM). The retaining system of the excavation was fully modelled, including walls, horizontal struts, and vertical center posts. For
comparison, the structural elements were simulated using plates with both elastic and elastoplastic behaviors. The soil response nesar failure
was assumed to follow the Morh-Coulomb model. Results showed that the FEM using the elastoplastic retaining system gave a more
reasonable estimate of stability of the excavation than that using the elastic support system. With the elastoplastic retaining system, yielding
firstly occurred on the wall and then on the struts, which caused large movement of surrounding soil toward the excavation. On the other
hand, with the elastic retaining system, failure of the excavation was only due to the great plastic heave of soil at the excavation bottom. The
predicted movement of soil and wall was nearly one meter as using the ealstoplatic support system but several meters as using the elastic
one.
Keywords: Deep excavations, stability analysis, finite element method, failure mechanism.
1. INTRODUCTION horizontal spacing of 2.9 m. Center posts were embedded into 0.8-
m-diameter bored piles. The site stratigraphy consists of eight
Stability of deep excavations is one of the main concerns of subsoil layers, at which most of them are soft clay, ranging from GL
practical engineers. Failure of excavations is often characterized by -2.0 m to -50.0 m. The undrained shear strength of soil obtained
collapse of the support system and the large inward movement of from field Vane shear test at two boring holes (V01 and V07) (Chen
surrounding soil. Therefore, occurrence of failure would result in et al., 2013) are also included in the figure. The ground water table
not only economic losses (e.g., destruction of neighboring facilities) is located at GL -2.0 m. As shown in Fig. 2, when collapse of
but also casualties. Most recently, a 15.7-m-deep excavation in excavation occurred, the wall had a deep inward deformation and
Hangzhou, China, collapsed because the lowest strut level was not was broken into two parts. Connections between struts and walls
installed timely, causing twenty one casualties. Based on an were failed and center posts were tilted toward the right wall. The
intensive investigation, Chen et al. (2013) predicted the failure soil heave at the excavation bottom was 2.5 m and the subsidence of
surface of soil from the disturbance of surrounding soil. the adjacent road was 7 m (Chen et al., 2013).
Stability problem of excavations has been studied by many
researchers using the FEM with reduced shear strength. A B = 21.2 m
11.9 m 9.3 m Right wall su (kPa)
parametric study by Goh (1990) showed that the factor of safety 0 20 40 60 80 100
Left wall
would be increased with the embedded depth and the stiffness of the 1 f 609, d 12 GL -0.0m 0
FILL
wall but reduced with the thickness of the clay layer below the 2 f 609, d 16 CL
3 f 609, d 16
excavation bottom. Faheem et al. (2003) also performed a OL
4 f 609, d 16
parametric study and gave consistent results with those by Goh OL 10
(1990). Furthermore, closed form equations considering the wall GL -15.7m
Depth, m
embedded depth were developed to estimate the bearing capacity Not OL

Center post 20
factor (Nc). Based on case studies, Do et al. (2013) showed that the 2x(280x82x installed
strength reduction ratio (SR) of soil, corresponding to the onset of 7.5x12.5) OL
rapid development of the nodal displacement, was able to represent
the factor of safety of excavations. For convenience, most of GL -33.0 m 30
Bored pile
previous studies employed the elastic support system and did not Diaphragm (f 800) OL
model the existence of center posts used to support the horizontal wall (0.8 m 40
Field Vane
struts, so that the finite element model might not simulate thickness) GL -46.0m CL
(V01 undisturbed)
sufficiently behaviors of excavations in the site. In addition, few Field Vane
researchers address to failure mechanism of excavations in soft clay, (V07 undisturbed)
which is strongly affected by the subsoil profile (e.g. existence of Chen et al. (2015)
hard stratum) and stiffness of the support system, so that this Figure 1 Construction sequence and subsurface conditions
problem remains to be resolved.
In this study, the FEM with reduced shear strength was used to
investigate failure mechanism of the Hangzhou case. Details of the
strength reduction technique can be seen elsewhere (e.g., Do et al.,
2013). Center posts were considered in the finite element model. For
comparison, both the elastic and elastoplastic support systems were
employed.
2. HANGZHOU CASE
2.1 Ground Conditions and Construction Sequence
The Hangzhou case was a 21.2- × 107.8-m internally braced
excavation, which was a part of an underground subway station. The
construction sequence and subsurface conditions of the case are
plotted in Fig. 1. The excavation had a maximum depth of 15.7 m
and was performed in five stages. The support system was
composed of a 0.8-m-thick and 33-m-deep diaphragm wall and four
steel pipe strut layers at GL -0.5 m, -4.2 m, -8.5 m, and -12.4 m with Figure 2 Failure of excavation
372
2.2 FE Analysis The maximum wall deflection developed near the final excavation
grade (GL -15.7 m). The maximum soil heave occurred at 4 m away
Since most of the subsoil is saturated clay, it was assumed that from the left wall.
the behavior of soil is elastoplastic and well-described by the Mohr-
Coulomb model. The model requires five input parameters,
including Young’s modulus (E), Poisson’s ratio (), friction angle
(), cohesion (c), and dilatancy angle (). Clayey soils were
simulated as undrained materials, at which u = 0.495, u = 00, cu =
su, and Eu = 450su. su was taken from the average results of Vane
shear tests.
The structural elements (struts, walls, and center posts) were
modeled using plate elements, which required four input parameters
including axial stiffness (EA), flexural rigidity (EI), maximum
(plastic) bending moment (Mp), and maximum (plastic) axial force
(Np). The input parameters of the elastoplastic structural elements
used for analysis were listed in Table 1. When the elastic structural
elements were employed, the Mp and Np values would be very high,
e.g., Mp = 1015 kNm and Np = 1015 kN (per meter width). For the
reinforced concrete wall at this case, the compressive strength of
concrete (fc’) was assumed to be 21 MPa. The Young’s modulus (E)
was equal to 15000(fc’)0.5. The Mp value was calculated using a
cross section analysis program, namely XTRACT, from wall
thickness, area of steel reinforcement, and properties of concrete and
steel. The Np value was estimated according to ACI 318-11. Since
the wall may have cracks due to bending during excavation and the
quality of concrete of the wall cast in stabilizing fluid properly does
not meet the design requirements, a reduction factor of 0.8 was
applied to the input parameters (EA, EI, Mp, and Np) of the wall. On Figure 3 Wall deflection and soil heave as using the elastoplastic (a)
the other hand, EA, EI, Mp, and Np of struts were calculated based and elastic (b) support systems
on the assumption that the Young’s modulus (E) and the yield
stress (y) of struts were 2.04x108 kN/m2 and 250 MPa, respectively. When the elastic support system was employed, the final
Due to the improper installation of struts in the field (i.e., they are excavation stage was computed completely and the SRmax value
not totally straight as splicing H steel together), their parameters was 3.07975. The wall deflection and the soil heave corresponding
were reduced by 10%. EA and Np of center posts were determined to 98%, 99%, and 100% of SRmax are plotted in Fig. 3b. As shown
similarly to those of struts. EI and Mp of center posts were not in the figure, when SR was increased, the wall deflection remained
considered to avoid any restriction on soil movement below the constant and the maximum wall deflection, about 3×104 mm,
excavation bottom. happened at the wall toe. On the other hand, the soil heave
During the strength reduction procedure, the soil strength at the developed gradually with the maximum soil heave occurring at the
final excavation stage was reduced successively by increasing the center of the excavation, up to 4×104 mm. It is observed that the
SR ratio. The divergence of numerical solutions was defined as the constant wall deflection is due to its elastic behavior whereas the
failure of the excavation. The maximum ratio, namely SRmax, at increasing large soil heave indicates the plastic behavior of soil,
which numerical solutions still converge, was treated as the factor of which then leads to the divergence of numerical solutions.
safety of the excavation. For understanding the percentage of load Fig. 4 shows the interaction diagrams of internal forces (M, N)
being applied at the final stage, Mstage is defined as the ratio of of struts and walls in the numerical analysis as using the
the load applied successfully in calculation to that caused by elastoplastic support system. As shown in the figure, when
excavation at the final stage. excavation was performed from the 1st stage to the 5th stage (final
stage), bending moment had a greater impact than axial load on the
Table 1 Input parameters (per meter width) of structural elements behaviors of struts at the 1st and 2nd layers and the wall because
Strut y f c’ EA EI Np Mp N, kN/m
1 -26000
layer No. (MPa) (MPa) (kN) (kNm2) (kN) (kNm) 2
1st strut layer
-18000 2nd strut layer
1 250 1.4x106 6.4x104 1750 256 3
4 3rd strut layer
5 -2000 wall
2 and 3 250 1.9x106 8.3x104 2310 334
5
Center post 250 2.4x105 0 288 0 4
-1000
3
Wall 40 1.9x107 106 24260 1186 4 5
1 2 2 34 5
0 M, kNm/m
-1200 -800 -400 -200 0 34 5200 400 8001200
When the elastoplastic support system was employed, the final
excavation stage was not calculated successfully and the Mstage 1000
max value was 0.6652. Fig. 3a plots the wall deflection and the soil BL of 1st strut layer
heave as Mstage is increased to 98%, 99%, and 100% of Mstage 2000 BL of 2nd and 3rd
max. It is noted that since center posts stayed closer to the left wall strut layers
than the right wall, as shown in Fig. 1, the deformation of the left 18000 BL of wall
wall was greater than that of the right wall in this analysis. Note:
26000
Therefore, only the left wall deflection was presented in Fig. 3a. As BL represents boundary lines.
shown in Fig. 3a, when Mstage increased from 98% to 99% of At the 5th stage, internal forces corresponding to SMstage
Mstage max, the wall deflection and the soil heave did not develop max are plotted.
significantly. But when Mstage reached the maximum value, the Figure 4 Interaction diagram of internal forces of elastoplastic
deformations of soil and wall increased rapidly, up to 800 mm each. structural elements
373
their internal force curves developed along the M axis. On the other Fig. 5b is the plastic point plot of soil at Mstage max as using
hand, since the internal force curve of struts at the 3rd layer grew the elastoplastic support system. It is observed that plastic points of
closer to the N axis than the M axis, their behaviors were strongly soil distribute in front and back of the wall but do not spread below
affected by axial load. The wall started to yield at the 4th excavation the wall toe, which is consistent with the soil movement mentioned
stage, which was earlier than the strut system. At the final stage previously. Also, the distribution of plastic points is in good
(Mstage max), both the wall and struts at the 3rd layer yielded agreement with the failures surface predicted by Chen et al. (2013)
whereas struts at the 1st and 2nd layers remained elastic behaviors. based on the soil investigation after failure of the excavation.
Fig. 5a shows the incremental displacements of soil and
elastoplastic structural elements at the last calculation step of the 3. CONCLUSION
final excavation stage corresponding to the Mstage max value. As
The following conclusions can be drawn on the basis of the work
shown in this figure, because of the soil weight, the soil behind the
presented herein:
wall moved downward and toward the wall but did not pass below
i. With the numerical analysis of the Hangzhou case, it is found
the wall toe. The wall was pushed to deform and yield at the final
that when the elastoplastic support system is employed, the
excavation grade, as marked with a rectangle and a square in the
yielding of struts and walls firstly causes a sudden increase in
figure. Due to the asymmetric characteristics of the excavation, the
the wall deflection and the soil heave, and then the failure of
left wall yielded more seriously than the right one and had an
the excavation. On the other hand, as using the elastic support
additional plastic hinge at the 3rd strut level. The inward deformation
system, the failure of the excavation is caused by the large
of the wall mainly caused the soil heave at the excavation bottom.
soil plastic heave at the excavation bottom. The amount of the
The downward sinking of the wall bent the horizontal struts, as
wall deflection and the soil heave obtained in the elastic case,
presented in the enlarged area, but the bending effect was not
i.e., soil plastic heave, was much higher than that in the
enough to cause yielding of struts at the 1st and 2nd layers. Yielding
elastoplastic case, i.e., failure of structural system, by several
of struts at the 3rd layer, as marked with a circle, on the other hand,
orders from the point of view of numerical solutions.
was a result of the large axial load transferred from the wall. It is
ii. For the Hangzhou case (with an insufficiently installed strut
noted that since center posts were supported by bored piles, which
system), the soil behind the wall moved directly toward the
were embedded deeply into subsoil, the upward movement of center
wall, pushing the wall to have a bulging deflection and yield.
posts under the effect of the soil heave was not significant. These
Then, struts yield due to the large axial load transferred from
observed deformations are in good agreement with those in the site
the wall. Yielding of struts and walls leads to the movement
(Fig. 2), at which the wall was broken, struts were destroyed, and
of surrounding soil toward the excavation zone.
center posts were tilted toward the right wall.
4. REFERENCES
American Concrete Institute (ACI), (2011) “Building code
(a) requirements for structural concrete (ACI 318-11) and
Commentary (ACI 318R-11)”. Detroit.
Left wallRight wall Chen R.P., Li Z.C., Chen Y.M., Ou C.Y., Hu Q., and Rao M.,
Legend: (2013) “Failure investigation at a collapsed deep excavation
Plastic hinge of strut in very sensitive organic soft clay”. Perform. Constr. Facil.
Plastic hinge of wall
04014078.
Spread of plastic hinge of wall
Do T.N., Ou C.Y., and Lim A., (2013) “Evaluation of factors of
Left wall Right wall safety against basal heave for deep excavations in soft clay
using the finite element method”. Geotech Geoenviron Eng
2013, 139, pp2125–2135.
Faheem H., Cai F., Ugai K., and Hagiwara T., (2003) “Two-
(b) dimensional base stability of excavations in soft soils using
Failure surface (Chen et al., 2015) FEM”. Computer and Geotechnics, 30, Issue 2, pp141–163.
Goh A.T.C., (1990) “Assessment of basal stability for braced
Figure 5 Incremental displacement plot (a) and plastic point plot (b) excavation systems using the finite element method”.
at Mstage max as using the elastoplastic support system Computer and Geotechnics, 10, Issue 4, pp325–338.
374
Effect of the Initial Suction Boundary on the Slope Failure of Volcanic Residual Soil
Agus Setyo Muntohar1
1
Department of Civil Engineering, Universitas Muhammadiyah Yogyakarta, D.I. Yogyakarta, Indonesia
E-mail: muntohar@umy.ac.id
ABSTRACT: Rainfall-induces slope failure has been intensively investigated to obtain the failure mechanism and its affecting and controlling
factors. This paper presents a numerical analysis of the rainfall and limiting pore water pressure on the slope instability of volcanic residual
soil. The slope was natural slope where was located in Kedungrong, Kulonprogo Regency. The slope was modeled using 2D finite element
method to anlayze the rainfall infiltration and slope stability. In this study, the initial suction at the surface was limited from 1 to50 kPa with
initial groundwater depth up to 20 m below the slope surface. In general, a higher initial suction induces a delayed rainwater infiltration and
result in delayed slope failure. When the initial suction was generated from ground water table, limiting the initial suction in the numerical
analysis can provide a reasonable result to evaluate the slope failure
Keywords: initial suction, rainfall, slope stability, infiltration, pore water pressure, volcanic soil.
1. INTRODUCTION
Many landslide cases in Indonesia occurred in rainy session during
November to February. At this period, the rainfall continuously
precipitated and triggered the slope failures. Indonesia receives
significant rainfall year-round but experiences a wet season that peaks
in January and a dry season that peaks in August (Hendon, 2003; Lee,
2015). Evaluation of the rainfall triggering slope instability shows
complexity in the numerical model. Several studies indicated that
slope stability was affected by the boundary on the slope surface and
initial condition such as flux, initial suction, rainfall intensity and
duration (Tsaparas et al., 2002; Rahardjo et al., 2013; Muntohar et al.,
2013). The researcher also investigated the controlling parameter on
the rainfall-induced slope failures. Rainfall characteristics and
hydraulic properties of the soil were the most controlling parameter
that should be paid attention to the numerical model (Tsaparas et al.,
2002; Rahardjo et al., 2007).
In the 2D numerical analysis for infiltration or seepage requires the
initial conditions of pore-water pressure at the soil surface. Through
this initial condition, then pore pressures are calculated for each
change in volumetric water content during infiltration. Many
Figure 2 Location of Kedungrong landslide
researchers (e.g. Ng and Shi, 1998; Tsaparas et al., 2002; Lee et al.,
2009; Rahardjo et al., 2010) suggest that initial negative pore water
2. RESEARCH METHOD
pressure at the soil surface needs to be restricted, especially when
very deep groundwater is encountered. In a very deep groundwater 2.1 Soil and Slope Properties
table, when the initial conditions of pore water pressure are based on
hydrostatic pressure, they tend to produce unrealistic pore pressures. Both field and laboratory tests were carried out to obtain the soil
layers and the geotechnical properties. Figure 3 shows the slope
A landslide case in Kedungrong village occurred in November 2001 profile which was interpreted from hand-auger work. The slope was
after five days of intense rainfall (see Figure 1). This landslide is covered with residual soil of tuffs, weathered-breccia, and breccia-
interested to be studied since the area is populated. Figure 2 shows rock at the lowest layer. The geotechnical properties of each layer are
the location of landslides in District of Kulonprogo, Yogyakarta. This presented in Table 1. The soil-water characteristic curve and
paper is addressed to investigate the effect of limiting of initial pore hydraulic conductivity – suction relationship are shown in Figure 4.
water pressure on the rainfall-induced slope instability.
Figure 1 Rainfall record at the landslide area

Figure 3 The sectional area of the slope and soil layers
375
Table 1 Geotechnical properties of the soil layers

Parameter Residual Soil Weathered Breccia Massive Breccia
(Layer A) (Layer B) (Layer C)
Natural moisture content, wN(%) 33.2 39.4 40.2
Bulk unit weight, b (kN/m3) 17.7 15.1 14.8
Unit weight above water table, d (kN/m3) 13.4 12.1 11.7
Degree of saturation, Sr (%) 90.1 64.8 41.9
Saturated permeability coefficient, ksat (m/s) 1.19x10-4 1.74x10-8 -
Cohesion at failure (peak), c’ (kPa) 16 48 -
Residual cohesion, c’r (kPa) 12 36 -
Internal friction angle at peak, ’ (_o) 24 10 -
Internal friction angle at residual, ’r (_o) 18 9 -
Figure 4 (a) The soil-water characteristics curve, (b) Hydraulic conductivity – suction curve, (c) numerical modelling condition
Pressure (kPa) Pressure (kPa) Pressure (kPa) Pressure (kPa)
-100 0 100 200 300 -100 0 100 200 300 -100 0 100 200 300 -100 0 100 200 300
320 Top Top (b) Top (c) Top (d)

(a)
300
280
Elevation, Y (m)
260
Mid Mid Mid Mid
240
220
200
Toe Toe Toe Toe
180
160
Figure 5 Initial condition of suction at the top, mid, and toe slope (a) ua-uw = 50 kPa, (b) ua-uw = 10 kPa, (c) ua-uw = 5 kPa,
and (d) ua-uw = 1 kPa
2.2 Numerical Analysis these analyses, a finite element software SEEP/W was used
(GeoStudio, 2004a). The slope stability analyses were used to study
Figure 4c illustrates the idealization of the slope for numerical the effect of seepage conditions on the factor of safety of the slope.
analysis. The finite element used quadrilateral meshes of 3 m size. The analysis was conducted by SLOPE/W (GeoStudio, 2004b).
This size was suitable to obtain numerical convergence result. For
376
In this study, the initial suction at the surface was limited from 1, 5, Bishop method. This analysis allowed the factor of safety varied with
10, and 50 kPa with initial groundwater depth up to 20 m below the the elapsed time as the pore water pressure changed.
slope surface (see Figure 5). The rainfall was applied on the slope
surface as unit flux (q) and the seepage review was allowed on the
surface. The pore water pressure are reviewed at the top (top), middle 3. RESULTS AND DISCUSSION
(mid) and toe slopes (toe). The pore water pressure obtained from The typical pore pressure profile of the numerical simulation is
SEEP/W analysis was then directly linked to SLOPE/W analysis. The presented in Figure 6 and 7 for initial suction 1 kPa and 50 kPa
factor safety (FS) was calculated according to limit equilibrium respectively. The pore water pressure profile clearly shows the
Figure 6 Pore water profiles for limiting suction 1 kPa (a) at top, (b) mid, and (c) toe slope
Figure 7 Pore water profiles for limiting suction 50 kPa (a) at top, (b) mid, and (c) toe slope
60
(a)
Rainfall (mm/day)
50
Infiltration (mm/day)
40
30
Infiltration rate
20
10
0
2.0 Initial Suction:
(b) 1 kPa
5 kPa
Factor of Safety
10 kPa
1.5 50 kPa
1.0
0.5
0 1 2 3 4 5 6 7 8 9
Elapsed Time, t (day)

Figure 8 (a) Variation of infiltration rate, and (b) Variation of the factor of safety with elapsed time.
377
change in pore pressures against the initial conditions of suction on was generated from ground water table, limiting the initial suction in
the slope surface. After a day of rainfall (t = 1 day), it was observed the numerical analysis can provide a reasonable result to evaluate the
that the surface was saturated for all initial suction condition. The slope failure.
saturation advanced to a deeper depth with the elapsed time of
rainfall. Positive pore water pressure was build up on the interface of 5. REFERENCES
the Layer B and Layer C, because of the Layer C is almost
impervious. It allows the rainwater to flow along the interface and GeoStudio, (2004a). "Seepage modeling with SEEP/W." GEO-
induced seepage at the toe of slope (see Figure 6c and 7c). This SLOPE/W International Ltd., Canada
mechanism was introduced in Muntohar and Soebowo (2013). GeoStudio, (2004b). "Stability modeling with SLOPE/W." GEO-
SLOPE/W International Ltd., Canada
At low initial suction (Figure 5), the wetting front zone is rapidly Hendon, H. H. (2003) “Indonesian Rainfall Variability: Impacts of
achieved at low rainfall intensity and shorter duration. In contrast to ENSO and Local Air–Sea Interaction”, Journal of Climate,
a higher initial suction (Figure 6), it takes longer to form a wetting 16, pp. 1775-1790.
zone. During infiltration, rainwater displaced the air in unsaturated Lee, H.S. (2015) “General Rainfall Patterns in Indonesia and the
zone and delaying infiltration was induced. In contrast at a lower Potential Impacts of Local Season Rainfall Intensity”,
initial suction, the soil is nearly saturated. Thus, the rainfall Water, 7, pp. 1751-1768
infiltration rate is slower. At the observed point at the mid slope, Lee, L. M., Gofar, N. and Rahardjo, H. (2009), "A simple model for
infiltration rate tends to increase with increasing initial suction preliminary evaluation of rainfall-induced slope
(Figure 8a). This condition generally explains that slope surfaces with instability." Engineering Geology 108(3-4), pp: 272-285
a higher suction values tend to be near dry or dry. Thus, rainwater can Muntohar, A. S. and Soebowo, E. (2013), "Mechanism of rainfall
rapidly flow into the slope subsurface. However, the depth of the triggering landslides in Kulonprogo, Indonesia." Geo-
wetting zone achieved is relatively shallow depending the rainfall Congress 2013. Ed. C. L. Meehan, D. Pradel, M. A. Pando
intensities. and J. F. Labuzs. Reston, Virginia: American Society of
Civil Engineers, pp: 452-461.
It can been seen in Figure 8b that the factor of safety decreases as the Muntohar, A. S., Ikhsan, J. and Liao, H. J. (2013), "Influence of
rainfall intensity increases. The decrease of the factor of safety was Rainfall Patterns on the Instability of Slopes." Civil
attributed to the increase in pore water pressure (see Figure 6 and 7). Engineering Dimension 15(2), pp: 120-128
Ng and Shi (1998) explained that increase in pore water pressure Ng, C. W. W. and Shi, Q. (1998), "A Numerical Investigation of the
reduce the shear strength of the soil according to the Mohr-Coulomb Stability of Unsaturated Soil Slopes Subjected to Transient
failure criterion for unsaturated soils. Slope with a higher initial Seepage." Computers and Geotechnics 22(1), pp: 1-28
suction result in delayed slope failure because of delayed infiltration Rahardjo, H., Ong, T. H., Rezaur, R. B. and Leong, E. C. (2007),
and build up pore water pressure. The result was also explained by "Factors controlling instability of homogeneous soil slopes
Rahardjo et al. (2010). A shallower initial groundwater table under rainfall." Journal of Geotechnical and
generated a lower initial suction. As a result, the combination Geoenvironmental Engineering 133(12), pp: 1532-1543
constituted to a worst factor of safety of slope (Rahardjo et al., 2007). Rahardjo, H., Satyanaga, A. and Leong, E.-C. (2013), "Effects of flux
Thus, the determination of the initial conditions of suction greatly boundary conditions on pore-water pressure distribution in
affects the stability of the evaluated slopes. slope." Engineering Geology 165, pp: 133-142
Rahardjo, H., Satyanaga, A., Leong, E. C. and Song, N. Y. (2010),
"Effects of Groundwater Table Position and Soil Properties
4. CONCLUSION on Stability of Slope during Rainfall." Journal of
A numerical analysis has been successfully performed to investigate Geotechnical and Geoenvironmental Engineering 136(11),
the effect of limiting initial suction on the rainfall-induced slope pp: 1555–1564
failure. A remarkable results and discussion conclude that the initial Tsaparas, I., Rahardjo, H., Toll, D. G. and Leong, E. C. (2002),
conditions of suction greatly affect the stability of the evaluated "Controlling parameters for rainfall-induced landslides."
slopes. In general, a higher initial suction induces a delayed rainwater Computers and Geotechnics 29, pp: 1-27
infiltration and result in delayed slope failure. When the initial suction
378
Applying In Situ Debris-cement Mixtures to the Ground Improvement of Bank in

Wild Creeks
Jun-Yang Chen1, I-Hui Chen1, Yu-Shu Lin1, Shei-Chen Ho1, Miau-Bin Su1
1
ABSTRACT: This paper explores a new method of in situ debris-cement mixtures for laboratory experiments in order for the reuse of
dredging debris for the ground improvement of bank in wild creeks. The paper undertook laboratory testing to determine the optimum ratios
of water, cement and in situ debris by weight which can be suitable for the field work and compressive strength of ground improvement in
wild creeks. Taking into account an economical and efficient mixing method of recycling debris in wild creeks, the optimal cement-
aggregate ratio is 1:12, slump test is 15 ± 3.8 cm, sand content is less than 50%, and unit weight of test specimen is greater than 2.07 t/m3 as
compressive strength of the admixture can be as high as 400 psi, which is suitable for erosion control and bank protection of ground
improvement in wild creeks.
Keywords: Debris-cement Mixtures, ground improvement, wild-creek remediation.
1. INTRODUCTION specimens can be determined by laboratory experiments so as to be

suitable for field work and reach desired strength of ground
This paper presents a new method of in situ debris-cement mixtures improvement in wild creeks.
(ISDCM) for the reuse of desilting debris in wild creeks so that it
can be used for the ground improvement of the creek bank. In recent
2. METHOD
years, large amounts of landslides in the watershed occurred in
Taiwan after extreme torrential rain, which caused debris blockage There are four processes of laboratory tests for the ISDCM method
on the drainage way or debris flow. In order to solve the problem, to determine the optimum mixing ratios of cement, water, and
dredging debris of rivers or creeks is a method of remediation aggregate, which affects the workability, strength and cost of
(Kantoush, & Sumi, 2010; Kondolf et al., 2014; Knighton, 2014). admixtures. Firstly, tests of particle-size characteristics for field
However, debris disposal of creeks is a difficult subject for the sampling are carried out so as to understand the condition of in-situ
environmental management and rehabilitation work in alpine debris in case study areas. According to ASTM D2487 (2011), the
watersheds. In term of the processes on the dredging work in wild standard of the Unified Soil Classification System (USCS) provides
creeks, debris could be used as useful resources by improving its a useful first step in any field or laboratory investigation for
engineering properties with cement for the ground improvement in geotechnical engineering purposes which classifies soils from any
the field. geographic location into categories representing the results of
By comparison of engineering methods of the ground prescribed laboratory tests so that the particle-size characteristics
improvement, the first kind of methods is a concept of soil cement and other engineering properties can be determined. The paper uses
which is used as a stabilizing material for soil in the construction of the standard to analyse fine and coarse aggregates and to classify
highways or earth dams (Bell, 1993; Das, 2014). The standard gravel and sand with No.4 sieve in seven different field samplings.
process of soil cement is that the aggregate particle size should be The sand content and soil classification of every field site is shown
no more than 75 mm and the least 55% of the particles should be in Table 1. The results of article-size distribution curves for field
able to pass through a 4.75 mm sieve (American Concrete Institute sites are plotted in Figure 1.
(ACI), 1990). Secondly, Controlled low-strength material (CLSM)
is a self-compacted, cementitious material used primarily as Table 1 Soil classification of field sites
backfills, structural fills, pavement bases, erosion control, etc. (ACI,
Site Sand Content Symbol of Soil
1999). CLSM can be produced by mixing clean aggregates, in-situ Code
County Region
(%) classification
excavation or recycled materials which be delivered to a ready-
mixed plant (Adaska, 1997; Gabr, & Bowders, 2000; Chang & and A Nantou Hewang 25.2 GW
Chen, 2006). Finally, an in situ mixing method was developed in River
Japan, which uses in-site soil, sand and gravel with cement to fill the
foundations and wings of Sabo dam, groundsill, riverbed girdle, B Kaohsiung Taoyuan 34.7 GP
front apron and so on (Watanabe, et al., 1999; ACTEC & ISM District
method Association Office, 2017).
C Kaohsiung Namasia 17.6 GW
About all, some methods are not suitable for alpine field sites in District
the research, for example CLSM and in situ mixing method both
need a big mixing machine for production and transport of D Nantou Chichi weir 46.0 GP
admixtures but it cannot be installed in wild creeks of alpine sites
easily and economically. Thus, the paper integrates these concepts
to innovate a method of on-site excavation, in-situ mixing and real- E Pingtung Laiyi 34.0 GW
time fill, namely ISDCM, for the remediation work and dredging of Township
wild creeks, which can solve the problem of debris disposal in wild
creeks and make them become the materials of ground improvement F Taitung Dazhu River 28.5 GW
for the protection of bank and embankment of creeks.
In the paper, there are 7 case study areas where the
characteristics of the debris are significant various in the field. After G Taitung Taimali 60.7 SP
Township
field investigation and debris sampling are carried out, some
laboratory experiments were undertaken in the research, such as
water-cement ratio, cement-aggregate ratio, sand content, slump test
and unit weight, as well as compressive strength tests. Finally, the
optimum mixing ratio of water, cement and in-situ debris of test
379
improvement, respectively (Watanabe, et al., 1999). Above all,

design strengths of ISDCM method can be determined by all tests in
the research corresponding to the optimum slump (water-cement
ratio), sand content and cement-aggregate ratio.
3. RESULT AND DISCUSSION

In terms of sand content, water-cement ratio and slump tests, the
strength tests of sand contents include three ratios (30%, 40% and
50%), corresponding to three water-cement ratios, namely 0.75, 1.25,
1.75, respectively. The results of compression strengths, sand
contents and water-cement ratios related to slump tests are shown in
Table 2. Figure 2 illustrates the relationships of compression
strengths, sand contents and water-cement ratios, which shows the
strength is strongest as water-cement ratio is 1.25 in tests. However,
the testing shows the results of slump tests are different although the
water-cement ratio is the same. That is because the sand contents of
Figure 1 Soil particle-size distribution curves of field sites test specimens are different in the laboratory, which is similar to the
cases in the field. The sand content of each case study area is
Secondly, the paper explores the influence of water-cement ratio different (see Table 1) and the water amount control of water-
and sand content on the compressive strength of in-situ admixture. cement ratio by weight is difficult in alpine wild creeks so the paper
The ratio of water and cement by weight, namely workability, can suggests that optimum water-cement ratio of ISDEM method can be
be regarded as the slump of a specimen test. Although workability is controlled to approximately 1.25 by slump test in the field, at 15 cm
better if the slump is larger, the strength and durability could reduce roughly.
(Su & Miao, 2003). Meanwhile, the strength of admixtures is
affected by in-situ sand content as well as the amount of cement Table 2 Results of sand content and water-cement ratio tests
(Tomomatsu, 1998; Watanabe,1999; Katayama, 2008). Thus, this
study used three sand-aggregate ratios of 30%, 50% and 70% by Water-
Unit Compressive Sand
weight in the upstream of Jiji Weir to undertake tests of compressive Region Code cement Slump
weight strength content
strength, corresponding to three water-cement ratios of 0.75, 1.25 ratio
and 1.75, respectively. There are 27 test specimens of which J1 2.10 21 50 10.2
cement-aggregate ratios are the same as 1:12 and curing ages are 28
days in the experiments in order to determine the optimum ratios of J2 2.13 37 40 0.75 10.5
sand to aggregate and water to cement for the ISDCM method. In
terms of the ratio of cement to aggregate 1:12, According to Bell J3 2.17 59 30 10.4
(1993), the author suggests that the amount of cement requirement J4 2.37 99 50 15.0
for the soil types of GM, GP, SM, or SP is 5%~8% by weight on the Chichi
soil improvement of soil-cement method. As a result, if the in-situ J5 2.36 77 40 1.25 15.5
weir
soil or debris belongs to Class G of the USCS classification, the
J6 2.34 63 30 15.8
ratio of cement to aggregate 1:12 (the cement usage is 7.7%) by
weight is recommended. J7 2.32 42 50 18.5
Third step of the study methods is concerned about the ratio of
cement to aggregate for the ISDCM method. Generally, the concrete J8 2.31 33 40 1.75 18.3
mixing ratios are taken into account cement, fine and coarse J9 2.27 29 30 18.0
aggregates that are commonly applied the ratios of 1:2:4, 1:3:6 and
1:4:8 to construction sites or soil improvement (Bell, 1993; Das,
2016). However, the research used five cement-aggregate ratios of
1:8, 1:9, 1:10, 1:11 and 1:12 by weight in seven case study areas,
respectively, to undertake tests of compressive strength. The water-
cement ratio of these test specimens is 1.25 and curing ages are 28
days in the experiments. After tests, the optimum cement-aggregate
ratio for the ground improvement of ISDCM method can be
determined.
Finally, the paper establishes a desired compressive strength for
the ground improvement of the bank in wild creeks by using
ISDCM method practicably and economically. In terms of the
compressive strength for the ground improvement, According to
Bell (1993), the author points out that while the amount of cement is
5%~8% by weight used for the soil improvement of soil-cement
method, their values of compression strength are approximately Figure 2 Relationships of compression strengths, sand contents and
6.5MN/mm2 (66 kgf/cm2) and 1.2 MN/mm2 (12 kgf/cm2) for soil water-cement ratios
types of 'GW, GP, GM, GC, SW' and 'SP, ML, CL', respectively.
Based on CLSM method, compressive strengths for the ground With regards to the tests of cement-aggregate ratios by weight,
improvement of pavement bases and erosion control of can be the results of 28-day compression strengths in mixing ratio 1:12 are
estimated to range from 400 to 1200 psi (28 to 84 kg/cm2) (ACI, shown in Table 3. Then, using all results of compression strengths
2005; Siddique, 2009; Yan et l., 2014). With regards to strength from cement-aggregate ratios 1:8 to 1:12, the relationships of
tests of ISM method, it uses in-situ dredging debris efficiently for compressive strengths, sand contents and cement- aggragate ratios
the foundation of construct and the ground improvement by mixing for the test specimens of in-situ debris in field sites are shown in
cement; the design strengths are between 18 to 24 N/mm2 and Figure 3 where x axial is the sand content of each site, y axial is the
between 5 to 10 N/mm2 for foundation structure and ground compression strength of each test specimen in different cement-
380
aggragate ratios. The results illustrate that compression strengths of ISDCM ground improvement. It is economically and practicably
and cement-aggregate ratios are positive correlations which means considerable for the ground improvement that the admixtures of a
the more cement contents, the stronger compressive strength of test ratio of 1:12 cement to aggregate, slump test of 15 ± 3.8 cm and
specimens are. As can be see in Figure 3, the compressive strenght sand content less than 50% can be designed to achieve 28-day
dramatically drops when sand content is approximately 60% and the compressive strength as high as 28 kgf/cm2 (400 psi) as a standard
soil class of the USCS classification is ‘SP’. It means that if sand of ISDCM method.
contents are greater than 60% and soil class is poorer than ‘SP’ in
field sites, there will be in need of increasing cememt amount to
cement-aggregate ratio of 1:8 or 1:9, for desired strength of ISDCM
method.
Table 3 Results of the strength tests of cement- aggragate ratio 1:12

for the test specimens of in-situ debris in field sites
84 kgf/cm2
Sand
Site Compressive Unit
Code
County Region
strength (kgf/cm²)
content
weight 28 kgf/cm2
(%)
A Nantou Hewang River 44.1 25.2 2.10
B Kaohsiung Taoyuan 64.5 34.7 2.25

District
C Kaohsiung Namasia 74.9 17.6 2.29
District
D Nantou Chichi weir 64.6 46.0 2.21
Figure 4 Relationship between unit weight and compressive strength
E Pingtung Laiyi 31.4 34.0 2.10 of each test specimen
Township
F Taitung Dazhu River 41.3 28.5 2.16 Finally, the paper divides the ground improvement of ISDCM
method into two types and establishes their optimum ratios of
G Taitung Taimali 16.9 60.7 2.09 related materials, compressive strengths and engineering properties,
Township as shown in Table 4. In terms of type I, the design strength is up to
84 kgf /cm2 (1200 psi) which is suitable for the permanent structural
fill of embankments, berms or dams in wild creeks. The optimal
design is a ratio of 1:9 cement to aggregate , slump test of 15 ± 3.8
cm (corresponding to a water-cement ratio of 1.25), sand content of
20% to 40% and unit weight of greater than 2.15 t/m2. With regards
to type II, the design strength is between 28 kgf /cm2 (400 psi) to 84
kgf /cm2 (1200 psi) which is used for erosion control or bank
protection in wild creeks. The optimal design is a ratio of 1:12
cement to aggregate, slump test of 15 ± 3.8 cm and sand content of
less than 50% and unit weight of greater than 2.07 t/m2.
Table 4 Design strength and engineering properoties for the ground

improvement of ISDCM in wild creeks
Cement-
Figure 3 Relationships of compressive strengths, sand contents and to-
Compressive Slump Sand Unit
cement- aggragate ratios for each test region Type Aim aggregate
strength test content weight
ratio by
Furthermore, using other data of Table 3, the relationship weight
between unit weight and compressive strength of each test specimen
can be plotted, corresponding to different cement-aggregate ratios; Up to 84 Greater
Permanent
then the maximum envelope curve of an equation (y=486.67x-977.9) 15 ± 3.8 20% to than
I structural kgf /cm2 1:9
cm 40% 2.15
is built up, where y means unit weight and x means compressive fill
(1200 psi) t/m3
strength, as shown in Figure 4. As a result, the desired strength of an
in-situ admixture in the field can be determined by the equation as
known unit weight of the admixture. Erosion 28 kgf /cm2 Greater
Above all, the design strengths of the ISDCM method can be control or (400 psi) to 84 15 ± 3.8 less than than
II 1:12
determined as two types by the results of laboratory tests and the bank kgf /cm2 cm 50% 2.07
standards of CLSM method. One is type I that the design strength is protection (1200 psi) t/m3
up to 84 kgf/cm2 (1200 psi); the other is type II that the design
strength is between 28 kgf/cm2 (400 psi) to 84 kgf/cm2 (1200 psi).
The design strengths of type I and II are in line of the permanent 4. CONCLUSION
structural fill and erosion control, respectively, for the ground
The paper presents a new method of ISDCM for the ground
improvement by CLSM suggestion (ACI, 2005; Yan et al., 2014).
improvement of band protection and erosion control in wild creeks.
The compression strengths of two types are plotted in the Figure 4
In terms of the cement-aggregate ratio of laboratory experiments,
that shows almost all test specimens of 1:12 cement to aggregate
using ratios of 1:8, 1:9, 1:10, 1:11 and 1:12, respectively, there are
belong to type II and all test specimens of 1:9 cement to aggregate
significant correlations between the cement contents and
are over the strength of type II except some test specimens of sand
compressive strength as well as the unit weight of test specimens.
content roughly or over 50%. As a result, the design strengths of two
The optimal ratios of ISDCM for up to 84 kgf/cm2 (1200 psi) of 28-
type is suitable for the aim of creek remediation and erosion control
381
day test specimens which is in line of the permanent structural fill System). ASTM International, West Conshohocken, PA
designed by CLSM are a ratio of 1:9 cement to aggregate, a water- www. ASTM. org.
cement ratio of 1.25. Meanwhile, it is economically considerable for Bell, F. G. (1993). Engineering Treatment of Soils. London: CRC
the aim of river erosion control and bank protection of ISDCM Press.
ground improvement that the admixtures of a ratio of 1:12 cement to Chang, C. F. and Chen, J. W., 2006, “Development and Production
aggregate can be designed to achieve 28-day compressive strength of Ready-Mixed Soil Material”, Journal of Materials in Civil
as high as 28 kgf/cm2 (400 psi) of CLSM standards; however, it is Engineering, Vol. 18, No. 6, pp. 792-799.
not suitable for soil class ‘SP’ of the USCS classification because Das, B. M. (2016). Principles of foundation engineering (Eighth
there is a significant drop in the strength of the soil. The paper edition). Cengage learning.
suggests that if the in-situ soil or debris belongs to Class S of the Gabr, M. A., and Bowders, J. J., 2000, “Controlled Low-strength
USCS classification or sand content greater than 50%, the cement- material using fly ash and AMD sludge ＂ , Journal of
aggregate ratio of ISDCM should be increased to 1:9 or more for the Hazardous Materials, Vol. 76, No. 2-3, pp. 251-263.
desired strength of the method. Kantoush, S. A., & Sumi, T. (2010). River morphology and
Overall, two types of ISDCM designed standards are: 1) type I sediment management strategies for sustainable reservoir in
is the compressive strength of up to 84 kgf /cm2 (1200 psi) which is Japan and European Alps.
suitable for the permanent structural fill of embankments, berms or Knighton, D. (2014). Fluvial forms and processes: a new
dams in wild creeks. The optimal design is a ratio of 1:9 cement to perspective. Routledge. London and New York.
aggregate , slump test of 15 ± 3.8 cm (corresponding to a water- Kondolf, G. M., Gao, Y., Annandale, G. W., Morris, G. L., Jiang,
cement ratio of 1.25), sand content of 20% to 40% and unit weight E., Zhang, J., ... & Hotchkiss, R. (2014). Sustainable sediment
of greater than 2.15 t/m3. 2) Type II is the compressive strength management in reservoirs and regulated rivers: Experiences
between 28 kgf /cm2 (400 psi) to 84 kgf /cm2 (1200 psi) is used for from five continents. Earth's Future, 2(5), 256-280.
erosion control or bank protection in wild creeks. The optimal Krell, W. C. (1989). Flowable fly ash. Concrete International,
design is a ratio of 1:12 cement to aggregate, slump test of 15 ± 3.8 11(11), 54-58.
cm and sand content of less than 50% and unit weight of higher than Siddique, R. (2009). Utilization of waste materials and by-products
2.07 t/m3. In conclusion, type II of ISDM is an economical, efficient in producing controlled low-strength materials. Resources,
and useful method of recycling debris and ground improvement in conservation and recycling, 54(1), 1-8.
wild creeks. In the future, the method could be applied to filed Su, N., & Miao, B. (2003). A new method for the mix design of
experiments in order to understand the practices of the optimal medium strength flowing concrete with low cement content.
relative ratio of ISDM. Cement and Concrete Composites, 25(2), 215-222.
Watanabe, F., Fukumoto, K. and Horiguchi, T., 1999, Landslide
5. REFERENCES Disaster Countermeasures Aiming at Twenty-first Century.
Construction of Debris Barrier by ISM (in-situ Mixing
ACTEC(Advanced Construction Technology Center) & ISM
Solidification) Method. Cost Reduction by Foundation
method Association Office (2017), ISM (in situ mixing)
Excavation Labor Saving, Civil Engineering Journal, 40(6):9-
Method. Retrieved from http://www.ism-method.jp/
15.
Adaska, W. S., 1997, “Controlled Low-Strength Materials”, A
Yan, D. Y., Tang, I. Y., & Lo, I. M. (2014). Development of
Report From ACI Committee 229, Concrete International,
controlled low-strength material derived from beneficial reuse
Vol. 19, No. 4, pp. 41-43.
of bottom ash and sediment for green construction.
ASTM D2487 (2011). Standard practice for classification of soils
Construction and Building Materials, 64, 201-207.
for engineering purposes (Unified Soil Classification
382
Dynamic Compaction of Lateritic Fill for Property Development

Richard Ong1 and Marini Mardi2
1
Regional Technical Director, Menard Geosystems Sdn Bhd, Subang Jaya, Malaysia
2
Senior Technical Engineer, Menard Geosystems Sdn Bhd, Subang Jaya, Malaysia
E-mail: richard@menard-asia.com
ABSTRACT: Dynamic compaction is a ground improvement technique to densify soils and fill materials by dropping a falling weight called
pounder typically made of steel via a crawler crane. The method is to densify the compacted material so that its bearing capacity can be
increased, and the settlement can be reduced when additional loads are introduced over the compacted material. Its application has been
associated mainly to the compaction of sand deposit or sand fill, construction debris and even sanitary landfill. In this paper, the compaction
of lateritic fill using dynamic compaction is presented via two case studies. The first project is the filling of a 23 m deep valley using lateritic
soil compacted with dynamic compaction for a heavy industry factory. The second project is the application of dynamic compaction to
recompact up to 6 m deep of a filled ground where proper compaction was not done during filling for a housing development project.
Keywords: Dynamic compaction, Menard pressuremeter, lateritic soil, ground improvement.
1. INTRODUCTION geologically consisting of lateritic soil and topographically a valley

area. The height between the lowest part and the highest part of the
Dynamic compaction (DC) is a ground improvement method to valley is about 23 to 28 m within the factory footprint. The valley is
compact soils by heavy tamping which involves repeated dropping also a water catchment area. Therefore, about 1 to 2 m deep of water
of heavy weight called pounder typically made of steel onto the ponding exists on the lowest part of the valley. From the soil
ground surface. This method is also known as dynamic investigation results conducted at different locations within the
consolidation, heavy tamping, or pounding. The technique at its valley, the subsoil is generally 2 to 4 m thick of medium stiff silty
current form for improvement of large areas was pioneered by soil with SPT N-value varying from 4 to 9. This layer is underlain
Menard (Menard & Broise, 1975) and was extensively tested and by a sandy silt layer with SPT N-value of 50 and above. Figure 1
optimised hence its safe and economic application today. Since the shows the cross-section and elevation of the valley to be filled for
late 1960s, DC has been developed on the numerous sites all over the factory construction.
the world for various soil conditions and for a variety of applications
such as roads, airports, seaports and more.
The basic principle behind dynamic compaction (DC) technique
consists in the transmission of high energy waves in order to
improve weak subsoil. As a result of the impact, the soil is
compacted depending on its condition, structure and depth. The
energy is transferred to the subsoil by multiple impacts with
properly shaped weight (normally steel pounder) with a weight
ranging from 10 up to 40 tonnes free falling from a height ranging
from 5 up to 40 m. The densification effect when DC is applied on
partly saturated soils is analogous to impact (Proctor) compaction in
the laboratory while the application of DC on saturated cohesionless Figure 1 Cross-section of the valley to be filled
soils induces liquefaction and the densification process is like
combining blasting and vibro-compaction. It is worthy to note that The factory is proposed to be built across the valley on EL
the effectiveness of DC in saturated fine-grained soils is uncertain +65.00 m. The filling of the valley is carried out with the in-situ
(Mitchel, 1981). soil. Thus, cut and fill operations will be conducted at the project
Dynamic compaction (DC) has been used successfully for site with the cutting of the lateritic soil from the higher areas of the
compaction of sand for large scale development or reclamation (e.g., valley being filled to the lower areas of the valley to bring the valley
Hamidi et al., 2011 and Bo et al., 2009). This paper presents the to the final elevation of EL +65.00 m. The initial planning is to
application of DC to compact unsaturated lateritic soil for real estate compact the cut material using the conventional layer-by-layer
developments on a smaller scale, which is an application less compaction method (CCM). For this method, the fill materials shall
discussed. In this paper, the term “lateritic soil” is used to describe be placed in layers, and uniformly compacted to achieve dry density
the reddish highly weathered residual soils typically found in of not less than 95% of the maximum dry density at optimum
tropical or sub-tropical regions, which consists of particles ranging moisture content as determined in the standard Proctor laboratory
from gravel to clay, usually coated with sesquioxide rich crusts tests before the next layer is placed and compacted. Loose thickness
(Oyelami and Van Rooy, 2016). The two projects discussed in this of each layer shall not be greater than 300 to 400 mm depending on
paper are from Malaysia and the lateritic soil is silty in nature. The the type of compaction machinery.
objective of the paper is to demonstrate that DC can be used Since this project involved large amount of earthmoving, with
successfully to compact lateritic soil for property developments cutting and filling of a 23 m deep valley with approximately
provided proper design and construction control are adopted. The 684,000 tonnes of fill materials, alternative solutions are sought by
extensive use of Menard pressuremeter in the design and the developer during the construction stage due to time constraint
construction control for both projects is presented in this paper. and concerns over the construction activities to be carried out during
monsoon season. The backfilling of the valley is determined as the
2. DYNAMIC COMPACTION OF A DEEP VALLEY critical activity and must be completed on time. Time allocated for
the backfilling works is 3 months, which means at least one layer of
2.1 Project Background compaction needs to be completed per day if the conventional layer-
A factory for the manufacturing of heavy engineering products is by-layer compaction method (CCM) is adopted. The developer and
planned to be constructed over a deep valley in Selangor, a state in earthwork contractor are not confident this can be achieved,
central area of Peninsular Malaysia west coast. This development is especially during monsoon season. Hence, alternative solution using
on a piece of land with area of approximately 35,000 m². The land is dynamic compaction (DC) is proposed and adopted. A comparison
of CCM and DC for this project is shown in Table 1.
383
Table 1 Comparison between CCM and DC
CCM DC
Thickness of
0.3 m thick 3 to 5.15 m thick
compaction layer
Number of
> 70 layers 5 layers
compaction layer
Duration of
> 3 months 2 months
compaction
In this project, dynamic compaction (DC) is applied by dividing

the backfilling of the deep valley into five layers. The fill material is
placed with thickness ranging from 3 to 5.15 m per layer before DC
is carried out. Figure 2 shows the cross-section of the five layers of
DC works. It should be noted that the water ponding area is
backfilled with crusher run instead of lateritic soil in order for the
compaction using DC to be effective. In addition, a layer of crusher
run up to 50 cm thick is placed at the top of each layer to be
compacted to ensure the compaction works using DC can be carried
out without being adversely affected by the heavy rain that occurs
during the monsoon season.
Figure 3 DC rig setup for deep valley compaction
2.3 Pressuremeter Test Results

More than 30 Menard pressuremeter tests (PMT) have been carried
out during the execution of the dynamic compaction (DC) works for
this project as part of the quality control and quality assurance
requirements. Figure 4 shows the typical result from the
Figure 2 Cross-section of the DC works pressuremeter test comparing the pressuremeter modulus, EM and
limit pressure, PL before DC works (i.e., Pre-PMT-02) and after
2.2 Design and Execution compaction with DC (i.e., Post-PMT-B (outside)). This test is
carried out at level RL +50.15 m (i.e., after the second layer of fill
The Engineer for this project has specified for 50 kPa of safe material, specifically the lateritic fill is compacted using DC). In
bearing capacity and maximum allowable settlement of 150 mm, addition, it should be noted that the post-treatment PMT is
considering raft foundation as the foundation system to be adopted conducted in between DC prints.
for the factory. Thus, the design and performance criteria at RL
+65.00 m after dynamic compaction (DC) works at the fill area are Pressuremeter modulus, EM (bar) Limit pressure, PL (bar)
as follows:
 The safe bearing capacity shall be 50 kPa; 0 100 200 300 400 500 600 0 10 20 30 40 50 60
0 0
 The total post construction settlement shall not exceed 150 mm
under the maximum load of 50 kPa;
 The maximum differential settlement shall not be more than
1:500. 1 1
The bearing capacity and settlement calculations are carried out
using the methods outlined in D.60.AN General Memorandum on
the “Interpretation and Application of Pressuremeter Test Results to 2 2
Foundation Design” published in Sols-Soils No.26 (Menard, 1975).
Depth (m)
Depth (m)
These methods rely very much on the accuracy of the soil modulus
and limit pressure obtained from in-situ soil test. Therefore, Menard
pressuremeter has been specified as the pre-treatment and post- 3 3
treatment tests for this project. Based on the calculations conducted,
to comply with the above performance criteria, the compacted
laterite fill for each layer should have the following properties: 4 4
 Harmonic mean pressuremeter modulus, EM ≥ 6,000 kPa;
 Geometric mean limit pressure, PL ≥ 600 kPa
In this project, the spacing of the dynamic compaction (DC)
5 5
prints is 5 to 6 m. The DC works is carried out in 2 phases, the main
phase and the ironing or overlapping phase. The steel pounder
Pre-PMT-02 Pre-PMT-02
weight is 15 tonnes and the drop height is 20 m. The DC works has
been carried out using two lattice-boom cranes with capacity to lift Post-PMT-B (outside) Post-PMT-B (outside)
the 15 tonnes steel pounder up to 20 m. Figure 3 shows the DC rig
Figure 4 Pressuremeter test results at RL +50.15 m
setup for the compaction of the valley. This photograph is taken
during the DC works at RL +45.00 m, where the water ponding area
has been filled with 3 m thick of crusher-run and to be improved
with DC.
384
The PMT test results clearly show that the DC works has  Change the foundation system from pad footings to piles
significantly increase the EM and PL of the lateritic soil. The EM and foundation;
PL after DC compaction are as follows:  Excavate the fill material and recompact;
 Harmonic mean pressuremeter modulus, EM = 27,180 kPa;  Deep compaction using dynamic compaction (DC).
 Geometric mean limit pressure, PL = 3,150 kPa Eventually, DC is selected as it is the cheapest and fastest solution
These values exceed the required values (i.e., acceptance criteria) by among the options available. The deep compaction up to 6 m thick
more than 4.5 times. This means the DC has been successfully using DC is possible for this project because the filled ground is
implemented on the lateritic fill. unsaturated.
3. DYNAMIC COMPACTION OF A FILLED GROUND 3.2 Design and Execution

3.1 Project Background The foundations for the buildings are originally designed as pad
footings with size ranges from 900 mm × 900 mm to 2850 mm ×
This project is a housing development consists of double storey link 2850 mm and minimum embedment of 1.2 m. The net allowable
houses and shop houses. It is situated in Kedah, a state located in bearing capacity of 106 kPa is required throughout the compacted
the north-western part of Peninsular Malaysia. The project is to be ground. The allowable settlement criteria for the pad footings is
built on a land area of approximately 34,000 m². The existing land shown in Table 2. Following the dynamic compaction (DC) works
elevation varies from RL +20.00 m to RL +30.00 m. Cut and to recompact the non-uniformly compacted lateritic soil, the
backfilling works is performed to achieve the final building platform Engineer has asked for the same pad footings design and acceptance
of RL +26.00 m. criteria to be maintained. The maximum treatment depth is limited
The backfilling works using surrounding lateritic soil from to 6 m because this is almost the maximum thickness of fill at the
cutting of hill is carried out using the conventional layer-by-layer project site.
compaction method (CCM). The lateritic soil is mainly silty. The
houses are to be built mainly on the filled ground. In order to meet Table 2 Allowable settlement criteria
the required design criteria in terms of bearing capacity and
settlement for the buildings to be built on pad footings, compaction Criteria Isolated footings
of the backfilling soil must be at least 95% of the maximum dry Angular distortion 1:300
density at optimum moisture content of the soil. The thickness of the
Maximum settlement (i.e., long term
fill ranges from 2 to 6 m. 75 mm
settlement after completion of building)
After the backfilling has completed, Mackintosh probe (MP) test
is carried out on the ground surface up to 6 m depth or refusal.
Figure 5 shows some of the MP test results conducted on site. The The design and calculation for the bearing capacity and
test results show that the compaction completed at certain areas settlement are carried out in accordance to the following section in
within the project site does not achieve the required degree of the D.60.AN General Memorandum (Menard, 1975):
compaction. This is confirmed by the existence of soft layers (i.e.,  Section 3 Calculation of the bearing capacity
MP blows count less than 30 blows per 300 mm) within the  Section 4.2 General formula for settlement of an isolated
compacted ground at varying depths up to 4.5 m. It should be noted foundation (T-0 rule)
that for the areas where the refusal is reached before 6 m, the probe Based on the design conducted, in order to meet the bearing capacity
is likely to have hit the granite bedrock. and settlement requirements, the acceptance criteria for the dynamic
compaction (DC) works are specified as follows:
No. of blow per 300 mm  Geometric mean limit pressure, PL, over the improvement depth
0 50 100 150 200 250 shall not be less than 500 kPa to achieve the net allowable
0 bearing capacity.
 Harmonic mean pressuremeter modulus, EM, of the compacted
fill shall not be less than 5,000 kPa for the upper 6 m to ensure
1 the allowable settlement criteria can be met.
In order to verify the performance criteria are achieved, acceptance
test using pressuremeter test (PMT) is specified. Plate load test
2 using 1 m diameter steel plate is also conducted after the DC works
to cross-check the PMT results.
The dynamic compaction (DC) works is only proposed at the
3 building footprint. The requirement is to treat the building footprint
Depth (m)
area so that the pad footings for the double storey houses and shop
houses can be constructed at any location within the building
4 footprint. The typical layout of DC prints adopted for the project is
shown in Figure 6. The spacing adopted for the main phase DC
compaction is 5.2 m centre-to-centre on a square grid and this is
5 followed by the ironing phase. A lattice-boom crane capable of
lifting 15 tonnes pounder with 20 m drop height is used for this
project to give maximum compaction energy of 300 tonnem per
6 blow.
A3 B14 C11
3.3 Pressuremeter Test Results
E62a E64
Thirteen Menard pressuremeter tests (PMT) have been carried out in
Figure 5 Mackintosh probe test results this project. The tests are carried up to 6 m or refusal depth. Figure 7
shows the 3 pre-treatment PMT conducted before dynamic
From the Mackintosh probe (MP) test results, it is concluded compaction (DC) works. The results show that the thickness of the
that the ground is non-uniformly compacted. Consequently, the fill is not uniform and ranging from about 2 to 6 m throughout the
construction of the buildings cannot be started. The developer has development area. In terms of degree of compaction, it is shown that
considered a few options to solve this issue as follows: there are localised layers within the lateritic fill which are not
385
properly compacted (i.e., pressuremeter modulus, EM less than 5,000 Pressuremeter modulus, EM (bar) Limit pressure, PL (bar)
kPa). For instance, at depth of 2 m for Pre-PMT-01, at depth of 1 m 0 100 200 300 400 0 10 20 30 40
0 0
for Pre-PMT-02, and at depth of 3 to 4 m for Pre-PMT-03.
1 1
2 2
Depth (m)
Depth (m)
3 3
4 4
5 5
6 6
Post-PMT-1 Post-PMT-2 Post-PMT-1 Post-PMT-2
Post-PMT-3 Post-PMT-3
Figure 8 Post-treatment pressuremeter test (PMT) results
4. DISCUSSION
Figure 6 Layout for dynamic compaction (DC) works Notwithstanding the uncertainness of dynamic compaction (DC) to
treat soil with high fines content, this paper has presented two case
studies of the application of DC to compact lateritic fill for property
Pressuremeter modulus, EM (bar) Limit pressure, PL (bar)
development projects. The main reason for the two projects to be
0 100 200 300 400 0 10 20 30 40
0 0 successfully implemented, despite the fill material is mainly silty
soil with fines content more than 50%, is because the lateritic fill is
unsaturated or partially saturated. Furthermore, for the case of the
1 1 deep valley compaction, additional construction measure has been
taken to minimise the effect of rainwater from wetting the lateritic
fill by placing a layer of crusher run on top of each layer of fill to be
2 2
compacted. This increases the effectiveness of the DC works
tremendously. Indeed, DC will not be the go-to solution for the
Depth (m)
Depth (m)
3 3 compaction of lateritic fill in property development projects.

Nonetheless, it could be an attractive solution under certain
conditions as highlighted in the two case studies in this paper.
4 4 Therefore, property developers and their designers should keep in
mind the potential of using DC for their projects, especially when a
5 5
thick layer of fill is to be compacted.
This paper also discusses on the application of pressuremeter
test (PMT) in the design and construction quality control including
6 6 verification of the design conducted for dynamic compaction (DC)
Pre-PMT-01 Pre-PMT-02 Pre-PMT-01 Pre-PMT-02 works. The pressuremeter modulus, EM and limit pressure, PL from
the PMT are the two soil parameters adopted in the settlement and
Pre-PMT-03 Pre-PMT-03
bearing capacity calculations. As such, verification at the project site
Figure 7 Pre-treatment pressuremeter test (PMT) results after compaction works can be done almost instantaneously upon
completion of the PMT. If the required EM and PL are not achieved,
Figure 8 shows the post-treatment pressuremeter test (PMT) additional compaction works can be conducted. If the specified EM
conducted at the same location with the pre-treatment PMT after and PL are met, the compacted ground can be handed over for next
dynamic compaction (DC) works. It is shown that the fill material phase of construction. Indeed, PMT is a valuable tool in DC works
has been well compacted as demonstrated with the overall increase as it provides the possibility for the designer and constructor to
of the pressuremeter modulus, EM and limit pressure, PL. The check on the deep compaction works at every metre interval and
variability of the lateritic fill is also reduced. Table 3 summarises the verification of the design can be done swiftly following the test.
harmonic mean EM and geometric mean PL of the lateritic fill
recompacted using DC method. Clearly, the values of EM and PL 5. CONCLUSION
obtained after DC works exceeded the specified acceptance criteria.
This paper has presented two case studies of the application of
Thus, the pad footings can be built confidently within the building
dynamic compaction (DC) for the compaction of unsaturated
footprint area to support the double storey houses and shop houses.
lateritic fill with high fines content in property development
It is worthy to note that the plate load test conducted up to 2 times
projects. The first case study is the compaction of a deep valley up
the net allowable bearing capacity (i.e., 212 kPa) recorded a
to 23 m using DC. The 23 m deep valley is subdivided into 5 layers
maximum settlement of less than 21 mm.
of 3 to 5.15 m to ensure effective compaction using DC. The second
Table 3 Harmonic mean EM and geometric mean PL values case study is the recompaction of a non-uniformly compacted filled
ground up to 6 m. In both projects, pressuremeter test (PMT) has
Harmonic mean Geometric mean been used in the design and construction control. The PMT results
EM (kPa) PL (kPa) show the effectiveness of DC to compact the lateritic fill to the
Post-PMT-1 11,100 760 required level in both projects. Indeed, unsaturated lateritic fill with
Post-PMT-2 26,100 1,190 high fines content can be compacted successfully using DC if proper
design and construction tools are adopted.
Post-PMT-3 11,000 1,000
386
6. REFERENCES Menard, L. and Broise, Y. (1975) “Theoretical and practical aspects

of dynamic consolidation”. Geotechnique, 15(1), pp3-18.
Bo, M.W., Na, Y.M., Arulrajah, A. and Chang, M.F. (2009) Mitchell, J.K. (1981) “Soil improvement: state of the art report”,
“Densification of granular soil by dynamic compaction”, Proceedings of the 10th International Conference on Soil
Ground Improvement, 162(3), pp121-132. Mechanics and Foundation Engineering, Stockholm, Vol 4,
Hamidi, B., Varaksin, S. and Nikraz, H. (2011) “Dynamic pp509-565.
compaction for treating millions of square meters of sand”, in Oyelami, C.A. and Van Rooy, J.L. (2016) “A review of the use of
Shahin, M. and Nikraz, H. (ed), International Conference on lateritic soils in the construction/development of sustainable
Advances in Geotechnical Engineering (ICAGE 2011), Perth, housing in Africa: A geological perspective”, Journal of
pp475-480. African Earth Sciences, Vol. 119, pp226-237.
Menard, L. (1975) “The Menard pressuremeter: interpretation and

application of pressuremeter test results to foundation
design”. General Memorandum D.60.AN. Sols Soils, N° 26,
pp5-43.
387
Nilai Settlement Ratio Pada Pekerjaan Perbaikan Tanah

Dengan Teknik Vacuum Preloading
(Studi Kasus Proyek Jalan Tol Palindra Seksi 1)
D. Herwan1, I. Masyhur2, M. H. Bigman2, S. Endra2, S. Rizal3, dan S. Idwan3
1
Program PascaSarjana Institut Teknologi Bandung, Bandung, Indonesia
2
Institut Teknologi Bandung, Bandung, Indonesia
3
PT Hutama Karya, Jakarta, Indonesia
E-mail: herwand@yahoo.com
ABSTRAK: Analisis untuk memprediksi penurunan konsolidasi umumnya dilakukan dengan menggunakan metode Terzaghi 1 dimensi.
Berdasarkan beberapa data pengamatan lapangan (settlement plate) yang dilakukan, prediksi penurunan konsolidasi dengan metode tersebut
memberikan nilai terlalu konservatif. Penyebabnya kemungkinan besar adalah masalah konsepnya yang terlalu sederhana dan nilainya
cukup jauh dengan yang terjadi di lapangan. Penurunan berdasarkan analisis 2D memberikan hasil yang lebih baik, hal itu dapat
dilihat pada perbandingan prediksi nilai penurunan berdasarkan analisis 2D elemen hingga dengan data hasil instrumentasi lapangan.
Penggunaan analisis 2D untuk sebagian para praktisi juga merupakan masalah tersendiri, selain perlu pemahaman yang cukup mendalam
juga membutuhkan software yang harganya tentu saja tidak murah. Dalam penelitian ini secara sederhana disampaikan suatu nilai rasio
penurunan, yaitu perbandingan antara nilai penurunan yang diperoleh berdasarkan analisis 1D dan nilai penurunan yang diperoleh dari hasil
pengamatan lapangan. Penelitian sebelumnya berdasarkan data laboratorium, secara tidak langsung menyebutkan nilai rasio ini antara 0.81-
1.1, untuk kondisi tanah di lokasi studi dan pendekatan yang berbeda tentu saja nilainya dapat berbeda. Dengan demikian seharusnya untuk
memperoleh nilai penurunan yang lebih sesuai dengan data pengamatan lapangan, maka para praktisi dapat menggunakan analisis 1D yang
lebih sederhana kemudian tinggal mengkalikannya dengan rasio penurunan berdasarkan penelitian ini untuk memperoleh nilai prediksi
penurunan yang lebih mendekati data pengukuran lapangan, dengan batasan-batasan tertentu. Data penyelidikan tanah (uji lapangan dan
laboratorium) dalam penelitian ini tersedia cukup baik, demikian juga data hasil pengamatan lapangan seperti settlement plate dan
piezzometer. Area yang menjadi studi kasus untuk penelitian ini adalah area Proyek Jalan Tol Palindra seksi 1. Nilai rasio penurunan juga
telah dibandingkan dengan penelitian sebelumnya untuk meningkatkan kualitas penelitian yang dilakukan.
Kata Kunci: Penurunan konsolidasi, settlement rasio, settlement plate
1. PENDAHULUAN
Mekanisme konsolidasi secara sederhana telah jelaskan oleh
Analisis dan pemodelan vacuum, terutama masalah tekanan vacuum Terzaghi (1940), dimana dalam teorinya diasumsikan struktur atau
dan distribusi tekanan vacuum, merupakan tantangan tersendiri partikel tanah saling terikat satu sama lain dengan tegangan
dalam perencanaan perbaikan tanah dengan metode vacuum molekuler tertentu dan merupakan material yang porous dengan
preloading. Pendekatan konvensional dalam analisis penurunan properties yang elastik. Pori tanah dianggap dipenuhi oleh air.
timbunan adalah dengan menggunakan analisis penurunan Beban yang diberikan kepada tanah akan mengakibatkan penurunan
konsolidasi 1D Terzaghi, yang dapat saja diterapkan dalam bertahap yang tergantung kepada rate keluarnya air dari pori tanah.
perencanaan vacuum preloading, yaitu dengan mengalikannya Terzaghi menerapkan konsep ini dalam penurunan tanah pada suatu
dengan suatu faktor yang disebut settlement ratio. kolom tanah di bawah beban konstan dan dijaga dari deformasi
lateral. Persamaan aliran air transient pada tanah butir halus adalah
sebagai berikut:
 2u  2u  2u e S 0  w 0  u a v  u  
kx  ky 2  kz 2  0 C 2     (1)
x 2
y z 1  e0 100  t e0  t t 
Untuk kondisi sistim:

 1 dimensi, arah vertikal saja
 Tidak ada perubahan tegangan terhadap waktu
 Properties tanah dianggap tetap terhadap waktu (uncoupled)
 Tersaturasi penuh
Maka formula diatas dapat diubah menjadi:
k z (1  e 0 )  2 u u 
  (2)
 w0 av z 2 t t
Dengan    0 , dan cv = koefisien konsolidasi = k z (1  e 0 )

t  w0av
Sehingga persamaan konsolidasi 1 dimensi Terzaghi;
 2u u
cv  (3)
z 2 t
Figure 1 Proyek Tol Palembang Indralaya
388
Teori ini umum digunakan dalam analisis penurunan Area Tol Palindra merupakan area genangan dan rawa. Pada
mengingat metode yang diusulkan mudah dan sangat sederhana. musim hujan, kondisi muka air dapat naik hingga 1 m di atas
Namun demikian dari asumsi-asumsi yang digunakan dan permukaaan tanah. Pada kondisi banjir, area yang tergenang dapat
penyederhanaan yang dilakukan, terdapat kondisi tertentu dimana mencapai 80% dari total rencana panjang jalan tol.
metode ini tidak bisa digunakan. Seringkali terdapat perbedaan yang
cukup besar antara hasil analisis dengan metode ini dengan hasil 3. KONDISI TANAH SEKSI 1 TOL PALINDRA
pengukuran di lapangan.
Untuk settlement ratio, dalam Stapelfeldt, T (2006) disajikan Untuk Seksi 1 (STA 0+000 s.d STA 7+100) terdapat 9 titik
hubungan antara settlement ratio terhadap stress ratio, diketahui pemboran teknik. Nilai NSPT & sampel inti menunjukkan ketebalan
bahwa untuk area sekitar permukaan tanah maka nilai settlement tanah lempung lunak berkisar antara 15.5 m hingga 33.5 m.
ratio adalah sebesar 0.81, lalu nilai settlement ratio ini meningkat
seiring dengan penurunan nilai stress ratio, hingga dicapai nilai
maksimum settlement ratio sebesar 1.1 pada stress ratio 0.5.
Figure 4 Stratifikasi Tanah Seksi 1 Tol Palindra
Serangkaian pengujian laboratorium pada beberapa titik bor

yang telah disesuaikan dengan posisi rencana lokasi vacuum
dilakukan untuk mengetahui engineering properties dari tanah di
lokasi ini.
Figure 2 Settlement Ratio vs Stress Ratio (Stapelfeldt, 2006)
Sedikit berbeda yang dilakukan dalam penelitian ini, settlement

ratio yang cari adalah nilai settlement rasio dari total penurunan
badan timbunan yang terbaca pada instrumentasi dan pada elevasi
settlement plate.
2. PROYEK TOL PALINDRA

Serangkaian pengujian dan pengumpulan data untuk menentukan
nilai settlement rasio ini telah dilakukan, yaitu dari Proyek Tol
Palindra Seksi 1. Proyek Tol Palindra merupakan proyek tol pertama
dan terpanjang yang menggunakan teknik vacuum preloading. Tol
ini merupakan bagian dari Tol Trans Sumatera yang
menghubungkan antara Palembang dan Indralaya dengan total
panjang 22 km. Terbagi menjadi 3 seksi; Seksi 1 (7.1 km,
Palembang-Exit Pamulutan), Seksi 2 (5 km, Exit Pamulutan-Exit
KTM)), dan Seksi 3 (9.9 km, Exit KTM-Indralaya).
Figure 5 Nilai Properties dan Parameter Deformasi Tanah

Area Seksi 1 Tol Palindra
Berdasarkan hasil pengujian laboratorium nilai angka pori

berkisar antara 1 hingga 3, sedangkan nilai koefisien kompresi tanah
berkisar antara 0.3 hingga 1.2. Tekanan pra konsolidasi berkisar
antara 50 hingga 100 kPa. Dengan elevasi finish grade timbunan
Jalan Tol pada Seksi 1 berkisar antara 2 m hingga 11 m, sehingga
tentu saja dengan kondisi tanah seperti itu maka masalah stabilitas
dan penurunan akan menjadi isu utama pada pelaksanaan proyek
Jalan tol ini. Setelah serangkaian proses pemilihan metode
perbaikan tanah yang sesuai, dengan pertimbangan masalah
stabilitas, ketersediaan material untuk timbunan, kecepatan
pelaksanaan, ketersediaan teknologi dan peralatan, harga, dan
resiko, maka diputuskan vacuum preloading sebagai metode
Figure 3 Kondisi Lapangan Proyek Tol Palembang Indralaya perbaikan tanah untuk proyek ini.
389
4. INSTRUMENTASI VACUUM PRELOADING 5. ANALISIS PENURUNAN DAN SETTLEMENT RATIO

Pada pelaksanaan pekerjaan vacuum preloading, terdapat Pada tahap perencanaan proyek ini, analisis prediksi besar
serangkaian alat instrumentasi yang digunakan, diantaranya adalah; penurunan konsolidasi telah dilakukan pada tahap perencanaan
inklinometer, piezometer (vibrating wire), vacuum gauge, settlement dengan menggunakan metode 1D Terzaghi. Kemudian pada tahap
plate, dan extensometer. pelaksanaan, tahapan pemantauan dan monitoring instrumentasi
juga dilakukan hingga semua syarat penghentian vacuum dipenuhi
dan vacuum dihentikan. Berikut merupakan perbandingan antara
hasil analisis 1D dengan hasil pemantuan lapangan dengan
settlement plate:
Table 1 Settlement, Vacuum vs 1D Terzaghi
Settlement (cm)
Ratio
Zona 1-D Terzaghi Settlement Plate (b/a)
Consolidation (a) Monitoring (b)
B 115.98 108.500 0.94
9 69.69 43.300 0.62
19 66.46 63.100 0.95
12 86.68 77.600 0.90
17 132.26 84.900 0.64
10 63.97 46.600 0.73
Figure 6 Stratifikasi Tanah Seksi 1 Tol Palindra 10 48.28 46.600 0.97
6 128.35 100.025 0.78
Monitoring dengan instrumentasi dilakukan setiap hari hingga 25 85.42 72.225 0.85
semua persyaratan penerimaan penghentian vacuum dihentikan, 25 97.54 72.225 0.74
yaitu; derajat konsolidasi sudah mencapai minimum 90%, load ratio
13 113.06 116.425 1.03
juga sudah mencapai minimum 1.2 sesuai dengan persyaratan yang
13 163.21 116.425 0.71
ada dalam Speksifikasi Khusus untuk Metode Pra-beban dan
Vacuum.
Upper Bound, S1D = Svac
Average, S1D = 0.88 Svac
Lower Bound, S1D = 0.77 Svac
Figure 8 Settlement Ratio
Dari nilai di atas, terlihat bahwa nilai settlement rasio yang

diperoleh adalah maksimum rata-rata sebesar 1, minimum rata-rata
sebesar 0.77, dan dengan trend sebesar sebesar 0.88.
6. KESIMPULAN
Untuk perencanaan vacuum, analisis penurunan akibat tekanan
vacuum dan beban preloading dapat saja diprediksi dengan metode
konvensional 1D Terzaghi, yaitu dengan mengalikannya
menggunakan suatu nilai yang disebut settlement ratio. Berdasarkan
hasil analisis dan perbandingannya dengan data settlement plate,
nilai settlement rasio untuk Zona 1 Tol Palindra adalah berkisar
antara 0.77 hingga 1, dengan rata-rata sebesar 0.88. Nilai ini masih
dalam rentang nilai settlement ratio yang diusulkan peneliti
sebelumnya yaitu sebesar 0.81 hingga 1.1.
Figure 7 Hasil Monitoring Instrumentasi Zona 2 – Seksi 1
390
Hayashi H., Nishikawa J., and Sawai. K, 2004. Improvement

7. REFERENSI Effect of Vacuum Consolidation and Prefabricated Vertical
Drain in Peat Ground. Helwany, S. 2007. Applied Soil
Bergado D. T., Chai. J. C., Miura N., and Balasubramaniam A.S. Mechanics: with ABAQUS Applications, John Wiley &
1998. PVD Improvement of Soft Bangkok Clay with Sons
Combined Vacuum and Reduced Sand Embankment Indraratna B., Redana. I. W. 2000. Numerical modeling of vertical
Preloading, Geotechnical Engineering Journal, SEAGC, 29 drains with smear and well resistance installed in soft
(1), 95-122 clay, Canadian Geotechnical. Journal, 37 (1), 132-145
Chai J. C., Carter J.P., and Hayashi S. 2006. Vacuum Consolidation Indraratna, B. 2012. Vertical drains, vacuum consolidation &
and its Combination with Embankment Loading, Can. preloading, University of Wollongong
Geotechnique, J 43, 985-996 Mesri G., and Khan A. Q. 2012. Ground improvement using
Choa V. 1989. Drains and Vacuum Preloading Pilot Test, 1347- vacuum loading together with vertical drains, Journal of
1350 Geotechnical and Geoenvironmental Engineering, 138(6),
Chu J., Yan S. W. (2005): Application of the Vacuum Consolidation 680-689
Method in Soil Improvement Projects. Griffin H., O’Kelly B. Stapelfeldt, T. 2006. Preloading and vertical drains, Electronic
C. 2014. Ground Improvement by Vacuum Consolidation, publication,
Ground Improvement, Proceeding of The Institution of Civil http://www.tkk.fi/Yksikot/Rakennus/Pohja/Preloading_and_v
Engineers Vol. 167, 274-290 ertical_drains.pdf
391
The Settlement Evaluation of Improved Soft Clay Using LECA Replacement

Technique
A. Azhani Zukri1, 2, B. Ramli Nazir2 and C. Ng Kok Shien3
1
Faculty of Civil Engineering & Environmental Resources, Universiti Malaysia Pahang, Pahang, Malaysia
2
Tropical Geoengineering Centre, Universiti Teknologi Malaysia, Skudai, Johor, Malaysia
3
Faculty of Civil Engineering, Universiti Teknologi Mara, Pulau Pinang, Malaysia
e-mail: engr.azhani@gmail.com
ABSTRACT: Soil replacement technique is the easiest and oldest method to improve the soft soils underneath the shallow foundations by
reducing the settlement by replacing weak soil (e.g. medium or soft clay and organic soils) with more competent materials such as sand,
gravel or other suitable granular materials. This paper presents the research on soft soil improvement underlain a large embankment using
Lightweight Expanded Clay Aggregate (LECA) as a replacement material which is substitute of normal aggregate. LECA has been used
successfully in geotechnical application where weight is an issue because the materials can help reduce dead loads by more than half. The
analysis of performance of LECA as a replacement material was conducted through finite elements methods by using Plaxis 3D software and
the recorded settlement magnitude then were compared to Tarzaghi’s equations for evaluating the average settlement of uniform loading. The
obtained results demonstrated that the settlement decreases with increasing of LECA replacement thickness.
Keywords: Soft clay, settlement, numerical modelling, soil improvement, replacement technique.
1. INTRODUCTION up the construction process. It does not require specialty contractors

and special machineries except excavators and rollers.
Soil replacement is one of the improvement methods which can This replacement method can be equated with the concept of
upgrade the soft soil performance. According to the review made by reinforced gravel rafts. The method involve the construction of a 1.2
M. Gaafer et al. (2015), there is an urgent need to study the meter thick compacted ‘raft’ of engineered aggregate. Once
technique of ‘remove and replace’ for improving a weak soil in term constructed the shallow gravel raft provides a stable platform which
of geotechnical requirements, (which is settlement and bearing creates a more uniform pressure distribution and reduces the
capacity) and the price to get the optimum thickness of replacement differential settlement.
layers as well as the most suitable material to minimize total cost of
foundation works (Gaafer, Bassioni, & Mostafa, 2015). However,
2.2 Thickness of replacement
the impact on environment and sustainability matters also need to be
concerned in term of suitable material selection for replacement. Generally, the determination of the thickness of soil is based on
Light weight expanded clay aggregate (LECA) are among the experience which in many cases is questionable. However, the
common lightweight materials that have been applied successfully thickness of the replaced zone should be greater than 0.5 meter,
in geotechnical application. It can reduce the weight of compacted where dimensions of a replaced zone is depend on the problem to be
geotechnical fills by up to one-half. This material is currently being diminished (Gabr, 2012). For example, to increase the bearing
used in many civil engineering works due to its low weight, high capacity and reduce the settlement of a footing on a soft soil,
strength and favourable drainage characteristics. In terms of their Lawton (2001) proposed the replaced zone length and/or width of 1
dimensions, the LECA particles can be classified as gravels. to 3 footing width and a thickness of 0.5 to 1.5 footing width. This
Previous study shows that the LECA has been used as fill material problematic soil behaviour can be improved either by fully or
for road embankment construction and filling behind retaining wall, partially replacing the inadequate soils with compacted granular fill
airport pavement subgrades, planting and storm water drainage layers.
system (Holm & Valsangkar, 1993). The full replacement is done to increase the bearing capacity and
reduce settlement, while the partial replacement done is mainly to
2. SOIL REPLACEMENT increase the stability of the side slope(Das, 2010; Han, 2015). For an
embankment over a soft soil, Broms (1979) suggested full and
2.1 General partial replacement of soft soil under the embankment as shown in
Poor soil conditions make the form of traditional construction Figure 1.
expensive, it may be economically viable to improve soil
engineering properties before construction begins. This can be done
by reducing pore pressure, by reducing the volume of voids in the
soil, or by adding stronger materials to soft soils.
2.1.1 Soil replacement
Replacement method has been commonly used to improve
geomaterials under continuous (strip) and isolated footings and also
in highway and railways construction when problematic soil are
encountered within limited areas and depths (Han, 2015). The
condition of foundation can be strengthened by replacing poor soil
with more competent materials such as sand, gravel or crushed
stone. The soil replacement under shallow foundation will reduce
consolidation settlement and increase soil bearing capacity,
eliminate expansion/shrinkage of expansive soil and the freeze–thaw Figure 1 (a) Full and (b) Partial replacement under an embankment
of frozen soil. The advantages of this simple technique over other (Modified from Broms, 1979)
improvement methods and deep foundation is it more economical,
reliable, well established and requires less time, therefore can speed 2.3 Previous researches on soil replacement
392
Many studies have been conducted on soil replacement works. Most 3.1 Constitutive model
of the studies were focused on settlement of soft soil under square or
circular footings. However, a number of researchers are also It is very important to choose the right constitutive model to
interested to evaluate the settlement behaviour of soil with column represents the soft soil behaviour during conducting the numerical
raft installation (soil replacement and stone column). analysis. Numerous models have been introduced and used by
A. K. Gabr (2012) conducted a study on the effect of researchers to analyse the behaviour of soft soil stabilized by various
replacement soil depth in controlling the settlement of square methods. In order to select the right constitutive model, several
footing through centrifuge test and numerical modelling. The three famous models have been selected for analysis in this study. The
dimensional numerical analysis was carried out using the software selection of model made based on availability of soft soil and LECA
Plaxis (2005) adopting the Mohr-Coulomb soil behaviour mode properties. While the non-existent parameters will be calculated
(Gabr, 2012). The study found that, the use of replacement soil can based on correlation or indirectly calculated using others suitable
reduce the settlement magnitude. It was observed that during the parameters, and this particular matters readily available in Plaxis
experiment, when a uniform load exceeding 100kN/m2, the soil 3D. The results obtained from the analysis will be compared to
suffered complete failure if the soil replacement was less than 0.2B analytical calculation.
(B is footing width). However, the numerical model results
indicated no signs of failure for the whole range of loads and soil 3.1.1 Comparison of soil constitutive models
replacement thicknesses, where the settlement values recorded Consolidation settlement magnitude can be predicted using
increase almost linearly and do not trend towards a constant value analytical method One-dimensional Terzahgy’s Equation.
which is does not seem logical, as there is a minimum limit to the Immediate Settlements and settlements due to primary consolidation
void ratio of any soil. This may be due to the fact that no lower limit occur during construction, while settlements due to secondary
condition is assigned for the void ratio in the Mohr-Coulomb failure consolidation and creep occur after the end of construction.
surface. However, only primary consolidation settlement will be calculated
The performance of partial replacement for soft clay under since, the immediate settlement is insignificant in clay and can be
square footing has been analysed using finite elements methods ignored, while secondary consolidation settlement is more important
(PLAXIS 3D program), where clay soil and sand material were in highly compressibility clay and organic soils such as peat. In over
simulated by the standards of linear elastic-perfectly plastic Mohr- consolidated inorganic clay, the secondary compression index is
Coulomb theory (Abbas, 2016). Sand was used as a replacement very small, of less practical significance (Das, 2015).
materials. The settlement values recorded were compared to Janbu The models selected for analysis are; Mohr Columb Model
equations which is adopted to calculate the average settlement of (MC), Linear Elastic Model (LE), Soft Soil Model (SSM), Soil
flexible foundations on saturated clay soils. The obtained results Hardening Model (SHM) and Soft Soil Creep Model (SSC). From
demonstrated that the stress-settlement behaviour increases with the sensitivity analysis conducted, Soil Hardening Model and Soft
increasing the thickness of sand layer. Similar findings were found Soil Creep Model (SSC) gave a closed result to analytical
by Abdel Salam through the study on the effect of using different calculation. However, since creep behaviour is only significant in
types and thickness of replacement layer on increasing bearing organic soft soil and peat, therefor, Soil Hardening Model will be
capacity and reducing consolidation settlement of soft clayey soil adopted to analysis the behaviour of soft soil in this study.
experimentally, where the study concluded that the vertical For untreated soil, the settlement magnitude calculated is
settlement decreased with increasing of the replacement layer 0.3153m (for 50kN/m2 uniform load) Plaxis 3D provided a result
thickness (S. Abdel Salam, 2007). slightly higher, as predicted by SS model, while MC model
The artificial neural networks (ANNs), and the multi-linear predicted lower value up to 40%. Compare to other models, SSC
regression model (MLR) have been utilised to predict the bearing and SHM estimated closed value, however SHM model provided
capacity of circular shallow footings supported by layers of granular almost equal to analytical value (0.3209m) where the percentage
replacement over natural clay soil (Ornek, Laman, Demir, & Yildiz, was less than 2%. However, since the sensitivity analysis is based
2012). The data used in running the network models have been on analytical calculations, the physical model should be carried out
obtained from an extensive series of field tests, including large-scale to see the accuracy of the settlement values estimated by each
footing diameters. The results indicate that the use of granular fill model.
layers over natural clay soil has a considerable effect on the bearing
capacity characteristics and that the ANN model serves as a simple 3.2 Methodology of numerical analysis for LECA
and reliable tool for predicting the bearing capacity of circular replacement
footings in stabilized natural clay soil. In addition, the field test
results indicate that the use of partially replaced granular-fill layers The numerical analysis of LECA replacement has been performed
over natural clay soil has considerable effects on the bearing based on following conclusions and limitations;
capacity and the settlement characteristics. For a given value of  Mesh refined used in the analysis: Fine Mesh.
replacement thickness to footing diameter ratio, H/D, the ultimate  Since the replacement of soft soil by LECA will be
bearing capacity increases in a nonlinear manner with the footing constructed on very large area with uniform distributed load,
diameter. On the other hand, for a given value of D, the magnitude the boundary distance is not critical, the replacement was
of qu increased with the increase in the compacted granular-fill layer modelled using three dimensional.
thickness, H (Ornek, Laman, Yildiz, & Demir, 2012).  Type of analysis: Drained analysis was selected in this study
to reduce the time consumed and to simulate the soft soil
3. METHODOLOGY behaviour in long-term, and also to allow for a greater
number of sensitivity and parametric analyse to be done.
The objective of the study is to evaluate the effects of a LECA  Type of constitutive model used in the study; MC (drained)
aggregates thickness, H as soil replacement substitute the normal model for LECA and SHM to represent the soft clay
aggregate in reducing compressibility, S of soft clay soil through behaviour.
numerical modelling. Three dimensional numerical analysis will be
 Construction effect was neglected in the analysis.
performed using commercial software Plaxis 3D (2017). To permit
 The total load applied on the structural system is assumed to
timely analysis in this research and to simulate long term soil
be uniformly distributed over the entire surface area of the
behaviour, drained analysis is adopted to allow for a greater number
LECA replacement surface and plate will be assigned as a
of sensitivity and parametric analysis to be done. Sensitivity analysis
medium for load transfer to the above ground. The uniform
needs to be conducted before numerical modelling is carried out.
load imposed used in this study are 50, 100 and 150 kN/m2.
393
 The analysis of settlement was focused to the centre of load, replacement technique. The results show that the increasing of depth
therefor settlement behaviour at the edge of surface loading of replacement will contribute to decreasing of settlement magnitude
was not include in the study. for all unit weight of LECA as well as normal aggregates. LECA 3,
 Unit weight of LECA: from the review of the previous study LECA 5, LECA 7 and LECA 9 represent the LECA with unit
and numerous LECA manufacturer through the world, the weight of 3kN/m3, 5kN/m3, 7kN/m3 and 9kN/m3, respectively, while
unit weight of LECA was found to be various between 3 to NA is normal aggregate. This finding is in good agreement with
10 kN/m3, which is lower than water density. Therefore, the previous study conducted utilizing other granular materials such as
settlement of soft clay will be evaluated based on the sand and gravel as a replacement material.
different value of LECA density (3kN/m3, 5kN/m3, 7kN/m3
and 9kN/m3).
 The analysis of normal aggregate used as a replacement
material was also performed as comparison.
 Depth of soft soil is 10 meter.
 Settlement analysis was analysed based on different depth of
replacement; which are 1.5m, 2.5m and 3.5m.
 Two stage of calculation involved; initial and loading phase
as shows in Table 1;
Table 1 Numerical analysis stages

Phase Calculation type Loading input
Initial phase K0 procedure Stage construction
Loading phase Plastic Analysis Stage construction
Figure 2 Settlement versus Replacement Depth for Various LECA
Density
3.3 Material properties
LECA is known as common lightweight materials with various
Manufactured Kaolin Clay was chosen to represent soft clay soil
unit weight between 3kN/m3 to 10kN/m3, which is lower than water
because it was easily obtained and could be reconstitute
density and soft clay itself. According to the previous study this
homogeneously through consolidation method. Beside, many other
materials can help reduce dead loads and lateral forces by more than
researchers also used kaolin in their research and therefore results
half in installations with soft soils. From the analysis, LECA with
could be compared. The properties of materials used in this study
lowest unit weight which is 3kN/m3 contributed to highest rate of
were tested based on British Standard and/or the American Society
settlement compare to other LECA. This proved that the lightweight
of Testing Material (ASTM), based on the suitability and
materials like LECA can be utilised as replacement material in order
availability of the equipment in the laboratory for the respective
to improve the soft soil with limited depth. It also can be seen that
tests. Table 2 represents the properties of Kaolin Clay and LECA
the LECA perform well compare to normal aggregate in term of
used in numerical analysis.
settlement reduction.
Table 3 shows the percentage of settlement improvement when
Table 2 Materials properties
LECA used as replacement material compared to normal aggregate
Parameter Kaolin LECA
(NA), while Table 4 represents the percentage of settlement when
Clay
LECA used as replacement material compared to untreated
Constitutive Model SHM MC
settlement. When the unit weight of the replacement material is low,
Type of analysis Drained Drained the percentage of settlement improvement becomes higher up to
Unit weight, ɤ (kN/m3) 16 3,5,7,9 41%, 62% and 80% for 1.5m, 2.5m and 3.5m depth of replacement,
Young's Modulus, E (kN/m2) 2420 2520 respectively. The percentage also seen to be increased with
Cohesion, c’ (kN/m2) 7 0 increasing of replacement depth for all materials.
Friction angle, φ’ (°) 25º 35ᵒ
Dilatation angle, Ψ’ (°) 0 5 Table 3 Settlement improvement percentage compared to
Poisson's ratio, ν 0.30 0.30 normal aggregate
Permeability, k (m/s) 2.58 x 10-10 2.53 x 10-2 LECA Settlement Improvement Percentage
LL (%) 54 - unit
1.5 meter 2.5 meter 3.5 meter
PL (%) 29 - weight
PI (%) 25 - LECA 3 30.82% 50.29% 70.24%
Undrained shear strength 7.5 - LECA 5 26.96% 43.43% 60.05%
(kN/m2) LECA 7 23.21% 37.03% 50.28%
Cc 0.256 -
2.60 0.77 LECA 9 19.45% 30.79% 41.03%
Specific gravity
Cr 0.058 -
e0 2.39 - Table 4 Settlement improvement percentage compared to
settlement of untreated soil
LECA Settlement Improvement Percentage
4. RESULT AND ANALYSIS unit
1.5 meter 2.5 meter 3.5 meter
Sixty models in total have been developed to simulate the soft soil weight
displacement with LECA used as a replacement material. LECA 3 40.82% 62.26% 80.25%
LECA 5 37.52% 57.06% 73.48%
4.1 SETTLEMENT LECA 7 34.31% 52.20% 67.00%
Figure 2 shows the plot of Settlement versus LECA replacement LECA 9 31.10% 47.46% 60.86%
depth for loading 50kN/m2. From the analysis, the settlement of clay NA 14.46% 24.09% 33.62%
layer was found to be improved when LECA material used as
394
Settlement improvement factor (SIF) for LECA and NA also

calculated and tabulated in Table 5 below. The factor of settlement
improvement is defined as the ratio of the settlement of an untreated
soil to the settlement of a treated soil. The factor is used to evaluate
the reduction in settlement after improvement. It is stated that the
settlement improvement factor was increased with increasing the
replacement depth. In addition, the settlement improvement factor
for normal aggregate replacement found to be lower compare to
LECA replacement.
Table 5 Settlement improvement factor for LECA and normal

aggregates
Replacement Settlement Improvement Factor =
Depth SIF =Su/Sc Figure 5 Pressure-settlement behaviour of improved soft clay by
1.5 m 2.5 m 3.5 m replacing 3.5m
LECA 3 1.69 2.65 5.06
LECA 5 1.60 2.33 3.77 The plot of settlement improvement factor (SIF) versus pressure
LECA 7 1.52 2.09 3.03 for each replacement depth also established as shows in Figure 6.
LECA 9 1.45 1.90 2.55 From the plot, it can be concluded that, the SIF of LECA reached
NA 1.17 1.32 1.51 constant value at the point pressure of 120kN/m2, while constant
value of SIF were recorded for NA replacement for all loading
The study also find out the effect of compacted shallow LECA imposed.
layers on the settlement under uniform distributed load. Maximum
settlement for each condition was measured. Figure 3, Figure 4 and
Figure 5 illustrate pressure-settlement behaviour of improved soft
clay by replacing 1.5m, 2.5m and 3.5m of the soft clay with
compacted LECA layers, respectively. The simulations results show
the vertical settlement at centre of the load imposed equals to
0.5748m or 574.8mm in resistance pressure of 150 kN/m2. The
maximum vertical settlement for improved soil with 1.5m thickness
replacement was between 0.44m to 0.45m for various unit weight of
LECA. While, settlement magnitude recorded for improved soil
with 2.5m and 3.5m were laid between 0.36m to 0.39m and 0.30m
to 0.32m, respectively. In addition, the replacement using normal
aggregate materials contributed to higher settlement value due to the Figure 6 Settlement improvement factor (SIF) versus loading
heavier aggregate weight compared to LECA. imposed
Figure 7, Figure 8 and Figure 9 present the curve of estimated

settlement based on various unit weight of LECA including NA as a
comparison under 50, 100 and 150kN/m2 uniform loading,
respectively. The plot shows that, the settlement increases linearly
by increasing the unit weight of the substitute substance. This
condition proves that the weight of the replacement layer gives
advantage to the settlement behaviour.
Figure 3 Pressure-settlement behaviour of improved soft clay by

replacing 1.5m
Figure 7 Settlement versus material unit weight for 50kN/m2 load
Figure 4 Pressure-settlement behaviour of improved soft clay by

replacing 2.5m
395
Figure 11 Stress ratio-replacement depth
5. CONCLUSIONS AND RECOMENDATIONS

Numerical analysis was performed to investigate the settlement
behaviour of treated soft clay soil in 3D models under drained
condition. A few conclusions can be drawn from this study;
 The study concluded that the increasing of depth of LECA
replacement will contribute to decreasing of settlement
magnitude for all unit weight of LECA as well as normal
aggregates. However, it can be seen that the LECA perform
well compare to normal aggregate in term of settlement
reduction. This proves that the lightweight materials like
LECA can be utilized as a replacement material in order to
4.2 SOIL STRESS improve the soft soil with limited depth.
 The obtained results concluded that increasing the thickness
A soil subjected to shearing stress offers shearing resistance which of replaced soft clay with compacted LECA layer evidently
comprises cohesion (c), dilatation angle, () and friction angle (). decreases the settlement and also decreases the resistance
The maximum resistance of a soil to shearing stresses is Shear stress. It can be concluded that use the replacement
Strength. The maximum effective shear strength of the treated technique under the loaded area is effective in increasing the
ground is also recorded from the numerical analysis. Figure 10 settlement factor.
presents the soil stress-settlement behaviour of untreated and treated  The resistance stress found to be decreased with increasing
soft clay by replacement works under 50kN/m2 loading. The results of LECA replacement layer. This condition shows that,
show the value of soil stress is decreased when the replacing layer LECA replacement is effective to improve settlement of soft
increases, while the settlement magnitude decreases. soil, however, it is proposed to utilise this method along
with stone column to solve both settlement and stress
behaviour of soft soil.
6. ACKNOWLEDGEMENT
The authors wish to thanks to Universiti Teknologi Mara,Pulau
Pinang and Universiti Teknologi Malaysia for providing facilities
during the research work and Ministry of Higher Education for
granting the financial for this research work.
7. REFERENCES
Figure 10 Stress-settlement behaviour of improved soft clay Abbas, B. J. (2016). The Settlement Evaluation of Improved Soft
Clay using Sand Columns and Partial Replacement
In addition, stress ratio was found to be increases with Technique. International Journal of Engineering Research &
increasing of replacement depth as presented in Figure 11. Stress Technology (IJERT), 5(7), 348–355. Retrieved from internal-
ratio is defined as the ratio of the maximum shear stress of an pdf://79.23.57.127/The Settlement Evaluation of Improved
untreated soil to the shear stress of a treated soil. This finding were Soft Cla.pdf
contrast with normal aggregate replacement, where the stress Das, B. M. (2010). Principle of Foundation Engineering. USA:
resistance will increased up to 12% with increasing of replacement Cengage Learning.
layer under 50kN/m2 uniform load. However, NA replacement Das, B. M. (2015). Principles of Foundation Engineering, SI. (7th
indicates a slight increase (less than 5%) of shear stress when 100 Edition, Ed.). USA: Cengage Learning.
and 150kN/m2 load are applied. Gaafer, M., Bassioni, H., & Mostafa, T. (2015). Soil Improvement
Techniques. International Journal of Scientific &
Engineering Research, 6(12), 217–222.
Gabr, A. K. (2012). The Uncertainties of Using Replacement Soil in
Controlling Settlemen. Journal of American Science, 8(12),
662–666.
Han, J. (2015). Principles and Practice of Ground Improvement.
John Wiley & Sons, Inc. Hoboken, New Jersey.
Holm, T. A., & Valsangkar, A. J. (1993). Lightweight Aggregate
Soil mechanics: Properties and Applications. Transportation
396
Research Record. Retrieved from internal-

pdf://212.153.191.145/Lightweight Aggregate Soil
Mechanics- Properti.pdf
Ornek, M., Laman, M., Demir, A., & Yildiz, A. (2012). Prediction
of bearing capacity of circular footings on soft clay stabilized
with granular soil. Soils and Foundations, 52(1), 69–80.
https://doi.org/http://dx.doi.org/10.1016/j.sandf.2012.01.002
Ornek, M., Laman, M., Yildiz, A., & Demir, A. (2012). Numerical
Analysis of Circular Footings on Natural Clay Stabilized
With a Granular Fill. Acta Geotechnica Slovenica, (June),
61–75.
S. Abdel Salam. (2007). The Effect of Replacement Soil on Reducing
Settlement of Footing on Deep Soft Clay Using Numerical
Approach. Cairo university, Giza, Egypt.
397
Performance of Helix Piled Raft Foundation in Tropical Fibrious Peat Soil Under
Traffic Loads
A. Ardy1, S. Lawalenna1, B.M. Ahmad1, B. Eko1, B.F. Andi1
1
Department of Civil Engineering, Faculty of Engineering, Hasanuddin University, Makassar, Indonesia
E-mail: ardy.arsyad@unhas.ac.id
ABSTRACT: This paper presents the investigation of settlement based performance of helix piled raft foundation under traffic load through
analytical and numerical methods. The analytical method is based on the concept of PDR analysis of piled raft where the stiffness of helix
piled raft is computed from the stiffness of helix pile and raft with a certain helix piled – raft interaction factor. In this study, the stiffness of
helix pile to peat soil is estimated by using a modified Randolph and Wroth equation, incorporating the mechanical behaviour of helix pile.
On the other hand, the stiffness of raft is determined by using Richart et al equation. In the numerical model, 3-dimension FEM model of
helix piled raft is undertaken in which helix pile is modeled as fixed end anchor, and raft as a plate above soft peat soil. The effect of helix
pile number on the bearing capacity and settlement of helix piled raft foundation system is investigated. The results reveal that the stiffness
of helix pile is too small compared to the stiffness of raft with the ratio from 1/6 to 1/4. The helix pile stifness is much influenced by the
number of helice and the radius of helice in a single helix pile. Since the stiffness of helix pile is small, such critical number of helix piles is
needed to have effect on the foundation system. Below this number, helix pile is insignificant to have effect in the helix piled raft foundation
system, and the elasticity of the raft plays important role. In general, however, it is obvious that the increasing number of helix piles can
decrease the elastic settlement of helix piled raft foundation when it is subjected with traffic load. In the stiffness ratio of helix pile to raft at
1/6, the increasing number of helix pile can reduce the settlement up to 80%, whereas that in the stiffness ratio helix pile to raft at ¼ can
reduce the settlement by 65%. These finding s would beneficial for development of alternative helix piled raft foundation for road
infrastructure in tropical fibrious peat soil.
Keywords: tropical fibrious peat soil, helix piled raft, stiffness, elastic settlement.
1. INTRODUCTION Qc  S f (DLc )cu  AH cu N c  dH eff cu (1)

Peat is a type of soil, composed of high contents of fibrous organic
materials. This soil is changed and fossilized in wetlands under where Qc is ultimate pile compressive capacity, Sf is spacing
appropriate climatic (Edil and Dhowian, 1981; Munro, 2005). Peat ratio factor; Lc is distance between top and bottom helical plates;
soil has such problematic mechanical behaviour since it has low AH is area of the helix,  is adhesion factor, Cu is undrained shear
shear strength, high compressibility and high water content. These strength of soil, Nc is compressive bearing capacity factor for
characteristics have contributed to subgrade problem in cohesive soils, Heff is effective length of pile above top helix (Heff
development road infrastructures with abundant peat soil deposits in = H – D), D = diameter of helix plate.
most lowland areas in Indonesia such as East Sumatera, Central and Stiffness of helix pile to peat soil can be estimated with pile
South Kalimantan, and West Sulawesi. head load-settlement by using a modified model of Randolph and
Typical peat soil deposit is very thick, up to 30 meters. This Wroth (1978). However, for helix pile, the model was modified to
condition has led to such difficulty in implementing several soil
improvement methods such as soil replacement and piled geotextile accomodate the behaviour of helix pile. Due to axial loading, soil
reinforcement soil (Arsyad et al., 2013). Moreover, the application below the helical plate, at soil along the cylindrical shear area, and
of PVD method is also difficult since the permeability of peat is along shaft pile will deform (Figure 1). For shallow helix pile, shaft
about 1000 × permeability of soft clay (Mesri and Ajlouni, 2007). resistance is too small compared to others resistance (Mohajeneri et
Although the application of piled raft in peat soil is still challenging al, 2016). As a result, the deformation along the shaft pile does not
(Huat et al., 2014), very low undrained shear strength and creeps of significantly contribute to pile settlement. The stiffness of helix pile
peat soil neccessitates very deep piles for obtaining sufficient
can be determined based on deformation of soil below helical plate
bearing capacity (Kazemian et al., 2011).
Alternatif method of combining raft foundation with helix piles and the deformation of soil along cylindrical shear failure as
in peat soil is needed to be investigated. This is due to previous expressed in Eq.2.
studies just focus on bearing capacity of helix pile in fibrous peat
soil (Adi et al 2016; Parlan et al. 2016). Helix piled raft may present 1
Pt  4 2Lc tanh(Lc)   4 Lc tanh(Lc )  (1)
better performance compared to just only helix piles. Thereofore,     1  
this study investigates the performance of helix piled raft foundation Gr0*wt  (1  v) r0*Lc    (1  v)r0 Lc 
*
when it is loaded with traffic loaded, in tropical peat soil.

Lc 2 (2)
2. METHOD Lc 
r0* 
2.1 Stiffness of Helix Pile
Screw Pile or helix pile is a pile foundation which consists of helices Where,
fixed to the shaft at specific spacing (Arup Geotechnics, 2005). Pt = pile load
Helix pile is used to compressive and lateral loadings with G = shear modulus of peat soil
overturning moments (Schmidt and Nasr, 2004). Analysis used to ro* = radius of cylindrical helix failure zone
estimate compressive and tensile bearing strength of helix pile in wt = settlement of pile
cohesive soils can be derived from a function of bearing strength of  =1
the end of the pile, helix plate bearing and the frictional resistance  = Poisson ratio of peat soil,
offered by the shaft-soil interface (Mooney et al., 1985; Narashima  = Soil – helix pile stiffness ratio, Ep/G
et al. 1993). The formulation of bearing capacity for compression is  =relation between the radius of influence of the helix pile
shown in Eq. 1. and the radius of the helix plate, ln (rm/r0*)
398
3. RESULT
Analysis of bearing capacity and settlement of helix piled raft in
peat soil under axial loading was conducted by incorporating
modified Randolph and Wroth (1978) method into Randolph (1994),
Poulos and Davis (1980). FEM analysis with PLAXIS 3D was also
undertaken to comprehend the result.
3.1 The Stiffness of Helix Pile to Peat Soil

Estimation of the stiffness of helix pile to peat soil was undertaken
for homogeneous peat soil. In this case,  = 7.94 kN/m 3, cu = 5 kPa,
Epeat = 150 kPa, v = 0.15 and G = 65.12 kPa. The helix pile has 3
helices with spacing distance of 0.50 m, helix radius of 0.35 m, and
length of 2 m. This has resulted in the stiffness of helix pile to peat
Figure 1 Bearing capacity of helix pile under a compressive load, soil at 334.612 kN/m. If this stiffness was compared to the
deformation patterns of helix pile in the upper and lower soil layers. experimental study based helix pile stiffness to peat soil (300 kN/m)
as investigated by Parlan et al. (2016), the result is quite closed. It
2.2 Stiffness of Helix Piled Raft and Load Sharing should be noted that the dimension of helix pile is similar between
analytical and experimental, LLL50. In the case of LL30 model, the
Raft stiffness to peat soil was estimated by using Richart et al stiffness obtained through analytical method was found at 312,43
(1970), as described by Eq. 3. kN/m, which has relatively closed to experimental based helix pile
stiffness at 269,23 kN/m.
Gs The effect of number and radius of helices on the stiffness of
Kr  z (4cd ) (3) helix pile were also examined. As shown Figure 2, the increase of
(1  vs) number of helices in helix pile would increase its stiffness. In
addition, the increase of helice radius also would increase helix pile
where: Gs is the shear modulus of the soil; vs is the Poisson’s stiffness (Figure 3). The stiffness of helix pile to peat soil depends
ratio of the soil; and t is the thickness of the raft, and c and d are on young modulus of peat soil. The larger young modulus of peat
coefficient raft dimension. soil, the higher helix pile stiffness will be.
Randolph (1994) introduced that piles are located strategically
with the raft in order to reduce differential settlement. Load sharing
between the raft and the piles can be estimated using simple method
Randolph (1994), and Poulos and Davis (1980), shown in Eq. 4.
K 
1  0.6  r 
K 
K pr   hp  K (4)
 Kr 
hp
1  0.64  
K 
 hp 
Where kpr is stiffness of piled raft; Khp is stiffness of the helix

pile group; kr is stiffness of the raft alone, and hp is raft – pile Figure 2 The effect of number of helices on the helix pile stiffness
interaction factor. to peat soil.
The proportion of the total applied load carried by the helix pile
raft is in Eq. 5.
1
 hp 
1 a
0.2  Kr  (5)
a  
 K r   K hp 
1  0.8  
K 
 hp 
Tri-linear curve was generated by computing mobilization of the

helix pile capacity according Eq. 6.
Pup (6)
P1  Figure 3 The effect of helice radius on the helix pile stiffness to
 hp peat soil.
Php   hp P  Pup (7) Raft stiffness was estimated in this study. For the raft with area
of 100 m2, in peat soil with shear modulus of 65.217 kN/m2 and v of
0.15, is 1,687 kN/m. The raft stiffness increases as the dimension of
Pr  P  Php (8) the raft and the shear modulus of peat soil aslo increase. Fot the raft
where Php is load on helix piles, Pr is load on the raft, Pup is the with area of 400 m2 and the shear modulus at 173.913 kN/m2, the
ultimate bearing capacity of helix pile group. raft stiffness would be at 8,184 kN/m, and that with area of 900 m2,
the stiffness is 12,276 kN/m.
399
3.2 Settlement of Helix Piled Raft in Peat Soil Under Traffic

Load: Analytical Model
We analysed the effect of the number of helix piles on elastic
settlement of helix piled raft foundation system. A segment of raft
with dimension of 10 m × 10 m with a 0.2 m thick was modeled and
subjected with truck loads of 65 tons. The raft lies over a very soft
peat soil with soil properties (E = 150 kPa, v = 0.15 and Cu = 5 kPa).
In the model, the helix pile stiffness was assumed at 300 kN/m. The
result is shown in Figure 4. It is found that increasing number of
helix pile would reduce the elastic settlement of the foundation
system under traffic load. However, since the helix pile has a quite
lower stiffness compared to the raft (1/6 to ¼), the effect of helix
pile is only seen with minimum number of helix pile of about 12
piles. It means that below that quantity, there is no effect of helix Figure 4 The effect of number of helix piles on the elastic
settlement of helix piled raft foundation under traffic load with helix
pile and raft foundation is still working as a single foundation
pile stiffness from 300 kN/m to 900 kN/m.
system without any support from helix pilex.
In Figure 5, it can be seen that load sharing carried by helix piles
is just 10% of the total load at the minimum number of 12 helix
piles. This will increase as the number of helix piles increases. At
the 25 helix piles under the raft foundation, the load sharing carried
by helix pile achieves 65% (Figure 6). It should be noted that the
ratio of a single helix pile stiffness to the raft stiffness is only 0.18 at
a single pile, up to 0.89 at 25 helix piles.
If the raft dimension and the stiffness of helix pile increase, the
elastic settlement would decrease. The more helix piles under the
raft, the lower elastic settlement would be (Figure 5). It is obvious
that there is critical number of helix piles where beyond this number,
their effects on the elastic settlement seems to be unchanged. The
critical number is much influenced by helix pile stiffness to peat soil. Figure 6 Load sharing ratio carried by helix piles at varies helix pile
The larger helix pile stiffness to peat soil, the smaller critical stiffness to peat soil.
number of helix piles required. The critical number of helix piles
is about 10 piles when the helix pile stiffness at 900 kN/m, whereas
that is about 21 piles when the helix pile stiffness at 400 kN/m to
peat soil.
Figure 7 presents the effect of helix pile quantity on the elastic
settlement of helix piled raft foundation system under traffic load,
with larger helix pile stiffness and raft dimension. Similar results
were found that the more helix piles, the lower elastic settlement of
helix piled foundation would be. The elastic settlement becomes
about 10 cm when 25 helix piles installed under raft foundation.
Load sharing ratio has becomes significant at large stiffness of helix
pile. It can be seen in Figure 6, helix pile stiffness of 3000 kN/m can
support more than 50% of the total load, supported by only 2 helix
piles installed. Load sharing increases to 90% when 11 helix piles
installed under raft. However, helix pile stiffness under 1000 kN/m,
the load sharing ratio seems to be much lower than that for helix pile Figure 7 The effect of number of helix piles on the elastic
stiffness above 1000 kN/m. settlement of helix piled raft foundation under traffic load, with
helix pile stiffness varies from 1000 kN/m to 1500 kN/m.
3.3 Elastic Settlement of Helix Piled Raft in Peat Soil: FEM
Model The effect of helix pile number on the elastic settlement of helix
pile raft foundation was examined. It can be seen that, at the soft
3-dimension FEM modeling of helix piled raft foundation in peat
peat soil, the increase of helix pile number in the helix piled raft
soil was undertaken. There is two models of peat soil. First, soft peat
foundation system will decrease the elastic settlement of the
soil was modeled in HS Model with a unsat of 10 kN/m3, sat of 12
foundation system (Figure 8). The larger helix pile stiffness
kPa, E50 of 50 kPa, Eoed of 50 kPa, and Eur of 200 kPa. The
generates the lower settlement of the helix piled raft foundation
undrained cohession is 1 kPa, with shear angle of 17,5. Second is system. The EA of helix pile was found to be critical at 1500 kN.
stiff peat soil which has unsat of 10 kN/m3, sat of 12 kPa, E50 of Beyond that EA, the effect was found to be constant. The critical
400 kPa, Eoed of 400 kPa, and Eur of 3600 kPa. In the second number of helix pile is about 9 to 15 helix piles, depending on the
model, the undrained cohession is 5 kPa, with shear angle of 27. helix pile stiffness.
The helix pile was modeled as fixed end anchor with EA varies from In the stiff peat soil, the result seems to be a bit different (Figure
300 kN to 3000 kN in the first soft peat soil, and 1000 kN to 10000 9). The helix pile number from 1 to 4 will yield similar elastic
kN in the second stiff peat soil. Meanwhile, raft foundation was settlement, while the helix number from 4 to 9 generates significant
modeled as plate with d of 0.20 m,  of 25 kN/m3, and E of 2.1 × reduction of elastic settlement. The EA of helix pile at 5000 kN was
106 kN/m2, and v of 0.15. The traffic load was modeled as surface found to be upper bound, where beyond the EA, the effect of helix
load of 5,37 kN/m2. pile quantity on the elastic settlement becomes less significant.
400
5. REFERENCES
Arsyad, A., Hamid, W., Yusmin, A. (2014). Case study of
reinforced embankment over a deep softsoil. Proceedings of
Softsoils, 21(23rd).
Arup Geotechnics, (2005). Design of Screw Piles: Assessment of
Pile Design Methodology. Ove Arup & Partners Ltd, London.
Edil T.B., Dhowian A.W. (1981). At-rest lateral pressure of peat
soils. Proc. Conf. Sediment. Consolidation Model, ASCE,
San Fransisco, pp. 411-424.
Huat, B.B.K. Prasad, A., Asadi, A., Kazemian,S. (2014).
Geotechnics of organic soils and peat. CRC Press, p.250.
Islam S, Hashim R (2008b). Engineering properties of peat soils in
Figure 8 The effect of number of helix piles on the elastic peninsular Malaysia. J. Appl. Sci., 8 (22).
settlement of helix piled raft foundation obtained from FEM model, Kazemian, S., Huat, B., Prasad, A. and Barghchi, M., (2011). “A
in very soft peat soil. State of Art Review of Peat: Geotechnical Engineering
Perspective”, International Journal of Physical Sciences, Vol.
6(8), 2011, pp. 1974-1981.
Munro, N. (2005). Dealing With Bearing Capacity Problems on
Low Volume Roads Constructed on Peat. ROADEX II
Northern Periphery.
Mesri G., Ajlouni M. (2007). Engineering properties of fibrous
peats. J. Geotech. Geoenv. Eng., 133(7): 851-866.
Mooney, J.S., Adamczak, S., Clemence, S.P., (1985). Uplift
Capacity of Helix Anchors in Clay and Silt. American
Society of Civil Engineers48–72.
Mohajerani, A., Bosnjak, D., Bromwich, D. (2016), Analysis and
design methods of screw piles: A review, Soils and
Foundations, Volume 56, Issue 1, Pages 115-128.
Narasimha Rao, S., Prasad, Y.V.S.N., Veeresh, C., (1993).
Behaviour of embedded screw anchors in soft clays.
Figure 9 The effect of number of helix piles on the elastic
Geotechnique 43, 605–614.
settlement of helix piled raft foundation obtained from FEM model,
Nasr, M.H., 2009. Performance-based design for helical piles. In:
in stiff peat soil.
Contemporary Topics in Deep Foundations. American
Society of Civil Engineers, USA, pp. 496–503.
4. CONCLUSION
Poulos, H.G., (2001) Piled raft foundations: design and applications,
- The increase of number of helix piles under raft foundation in Géotechnique 51:2, 95-113
peat soil can reduce the elastic settlement of the foundation from Poulos,H.G., and Davis, E.H. (1980). Pile Foundation, Analysis and
65% to 80%. Design. University of Sidney.
- Critical number of helix pile is found at about 10 piles to 21 Richart, F.E., Jr., J. R. Hall, Jr. R. D. Woods.Richart, F. E. (Frank
piles, depending of the stiffness of helix pile. Beyond this Edwin), (1970). Vibrations of soils and foundations. Prentice
number, the reduction of elastic settlement is less significant. Hall, p. 414.
From FEM Model, the critical number of helix pile is from 9 to Randolph, M.F., (1994). Design methods for pile groups and piled
15 piles, depending helix pile stiffness. rafts Proc. 13th International Conf. on Soil Mechanics and
- Load sharing carried by helix pile varies from 10% (1 pile) to Foundation Engineering, New Delhi (5):61-82.
65% (25 piles). This is influenced by the ratio stiffness of helix Randolph, M. F., & Wroth, C. P. (1978). Analysis of deformation of
pile to raft. The larger stiffness ratio of helix pile to raft, the vertically loaded piles. Journal of Geotechnical and
larger load sharing carried by helix pile will be. Geoenvironmental Engineering, 104(ASCE 14262).
- Further study is needed to investigate the mechanical behaviour Schmidt, R., Nasr, M., (2004). Screw piles: uses and considerations.
of helix pile raft through experimental model and FEM with Struct. Mag., 29–31.
SSC model.
401
Effect of Compaction on Liquefaction of River

and Sea Sand in Hai Phong City, Vietnam
C.L. Nguyen 1, Q.P. Nguyen2, H.H. Nguyen3 and D.M Nguyen4
1
Geotechnical Engineering, University of Transport and Communications, Hanoi, Vietnam
2
Civil Engineering Faculty University of Transport and Communications, Hanoi, Vietnam
E-mail: nguyenchaulan@utc.edu.vn, nguyenquangphuc@utc.edu.vn, haihaqd@ utc.edu.vn , ndmanh@utc.edu.vn @utc.edu.vn
ABSTRACT: Haiphong city is situated on the north-eastern coast of Vietnam. This city has a large coastal and sea area, which is advantage
for marine economic development. However, in order to reclamation the coastal area, river sand material is much demand which lead to face
some environmental problems due to exploitation of river sand. Thus, the intensive laboratory experiments were conducted including
physical, chemical and mechanical tests of river sand and sea sand to evaluate the potential use of sea sand in road embankment in Haiphong
city. The samples were prepared by compacting to value of 90% and 95% of maximum density defined by the Standard Proctor test.
Consolidated undrained cyclic triaxial tests were conducted for sea sand samples and river sand samples. The cyclic triaxial test results
showed that all samples are not liquefied under normal vehicle conditions which corresponding to cyclic deviator stress about of 8 kPa and
frequency of 1Hz. In addition, sea sand samples were not liquefied when increasing the compaction effort to 95% of maximum density.
Therefore, it is possible to use sea sand instead of river sand in some cases for reclamation land in Haiphong area.
Keywords: dynamic properties, road embankment, CU triaxial test, liquefaction
1. INTRODUCTION No. Group Sample Compaction ᵞd max ᵞs Note

of Ratio (R) g/cm3 g/cm3
Earth quakes and traffic loading are the most sources of cyclic
Sample
loading which may trigger the liquefaction of granular material.
10 B10 0.95 1.584 2.57
Liquefaction characteristic of granular soil was studied carefully
11 S06 B11 0.95 1.584 2.57
under the earthquake load condition. There are several methods to
determine the liquefaction which including laboratory testing or 12 B12 0.95 1.584 2.57
based on the simple data from the field such as Standard Penetration 13 S03 B13 0.95 1.655 2.675
Sea sand
Test (N1 or (N1)60) , Cone Penetration Test (CPT), and relied on 14 S05 B14 0.90 1.597 2.675
experience from the earthquake occurred during the past (Ishihara, 15 S01 B15 0.95 1.682 2.675
1977; Bolton, M. and Ignacio, 1983; P. and J., 1995; Jakka, Datta 16 S03 B16 0.95 1.635 2.670
and Ramana, 2010; Y. and J., 2014). 17 S04 B17 0.95 1.676 2.674
The primary aim of this study is to determine the liquefaction 18 S05 B18 0.95 1.676 2.658
capacity of the sea sand and river sand samples at density 19 S06 B19 0.95 1.645 2.669
corresponding to 90% and 95 % of maximum dry density. 20 S07 B20 0.95 1.649 2.670
A series of undrained cyclic triaxial tests were conducted for sea RS-
21 0.95 1.651 2.590
River sand
sand and river sand samples which were taken in Hai Phong city S01
area. These tests were considered the effect of compaction, RS-
frequency, amplitude and stress ratio to study the dynamic 22 0.90 1.713 2.570
S02
characteristic of sand fill in loading condition due to vehicle loading. RS-
23 0.90 1.681 2.590
S03
2. MATERIALS AND METHODS
2.1 Material 2.2 Method of undrained cyclic triaxial test
The sea sand and river sand material are taken from Hai Phong city. 2.2.1. Cyclic Triaxial equipment
The characteristic of materials is shown in Table 1. The sea sand
and river sand samples are prepared by compacting to reach the The cyclic Triaxial test was conducted at Laboratory of
compaction density of 90% and 95% of maximum dry density. Geotechnical Engineering Department, Hanoi University of Mining
Compaction ratio is defined of ratio between dry density for and Geology. The name of equipment for cyclic triaxial test is
prepared sample and maximum dry density which were conducted Tritech 100 from Controls – Group Italia. The advanced of the
by Standard Proctor test. The maximum dry density was from 1.584 machine is following:
to 1.644 g/cm3 and from 1.651 to 1.713 g/cm3 for sea sand and river  This equipment can determine in both the static condition
sand, respectively. (effective stress and stress line) and cyclic condition.
Table 1 Initial characteristic of sand sample  It remedies automatically the back pressure/ cell pressure
No. Group Sample Compaction ᵞd max ᵞs Note during cyclic triaxial test
of Ratio (R) g/cm3 g/cm3  Maximum control frequency is 10 Hz (based on test condition)
Sample
1 B01 0.95 1.644 2.58  Two load option/ dynamic deformation
2 S04 B02 0.90 1.644 2.58 - Maximum is ±25 mm with ±5 kN double-effect actuator
3 B03 0.95 1.644 2.58
Sea sand
4 B04 0.90 1.631 2.58 - Maximum is ±15 mm with ±14 kN double-effect actuator
5 S01 B05 0.90 1.631 2.58 2.3.1. Testing preparation.
6 B06 0.90 1.631 2.58
7 B07 0.90 1.584 2.57 Samples with 70 mm in diameter and 140 mm in height, were
8 S02 B08 0.90 1.584 2.57 compacted with optimal moisture content to ensure that the sample
9 B09 0.90 1.584 2.57 were reached the density ratio corresponding with 90% and 95% of
402
maximum dry density. All samples were conducted in undrained a = .Px = 0.5 x16 = 8 kPa
condition accordance to ASTM D 5311-92-1996.
where:
2.3.2. Testing procedure.
-  : loss dynamic stress factor from road surface to the sand layer
The samples were tested by the cyclic triaxial test and listed in Table at 1.5m depth with  = 0.5
2. The sample goes through into 3 steps: saturation process,
consolidation process and after finish consolidation steps, the vale is - Px: calculated vehicle loading, Px is taken as 16 kN/m3
closed to ensure that no water dissipated during the loading process. Thus, the effect of stress amplitude and frequency were
In the final step, the sample was subjected dynamic loading by considered with the following conditions:
acting the cyclic load with different amplitude of shear strain and
changing frequency. During the test, pore water pressure, cell  Effect of stress amplitude on liquefaction:
pressure, axial strain and axial stress will be read and noted
continuously. The testing procedure is illustrated as below:  Samples B1-B6 and Frequency f= 1Hz
 Change of Stress amplitude range: 8± 32(kPa)
- Saturation method step: Using back pressure (back) of 90 kPa
and cell pressure cell of 100 kPa  Effect of frequency on liquefaction:
To determine the stability of material when the vehicles moving
- Consolidated step: The sample is consolidated with cell pressure
on the road with difference speed with change of frequency but
(cell ) of 170 kPa; back pressure (back) of 130 kPa and the
stress amplitude remains constant.
effective stress (3’) of 40 kPa.
 Frequency f (Hz) = 1-8 Hz
- Shearing step: Dynamic loading will be applied
 Change of Stress amplitude a = ± 8(kPa)
Table 2 The cyclic triaxial test for samples
The shear strength of sand when liquefied is evaluated by
Sample Freq Cell Back Eff Stress Cyclic
(f) pressure pressure stress
conducting cyclic triaxial test in undrained test condition. In this
amplitude stress
3 (kPa) b ’c (kPa) Ratio test, the sand sample is saturated and consolidated under constant
(kPa) (kPa) CSR cell pressure and amplitude repeated stresses until the sample is
B01 1 170 130 40 0.4 deformed, or the pore water pressure reaches similar value with
32 initial cell pressure. In this state, the sample can be considered at
B02 1 170 130 40 24 0.3 unstable dynamic state, and then the liquefaction occurred.
B03 1 170 130 40 16 0.2 Normally, the liquefaction phenomena can be determined by 2 ways
B04 1 170 130 40 8 0.1 (Figure 1):
B05 1 170 130 40 16 0.2
(1) The increase of pore water pressure reaches 95 % of initial
B06 1 170 130 40 24 0.3
stress,
B07 1 170 130 40 8 0.1
B08 2 170 130 40 8 0.1 (2) or (2) The development of large deformation, are described
B09 3 170 130 40 8 0.1 with amplitude corresponding with 5% of axial strain.
B10 1 170 130 40 8 0.1
For establishing the liquefaction of specific sample (up to 95%
B11 2 170 130 40 8 0.1 pore water pressure or 5% amplitude axial strain), the number of
B12 3 170 130 40 8 0.1 cycles must be indicated in one specify case with the uniform
B13 1 170 130 40 16 0.13 dynamic loading amplitude. It is noted that to reach the liquefaction
B14 1 170 130 40 16 0.1 stage corresponding to the 95 % of pore water pressure ratio and 5%
B15 1 170 130 40 16 0.2 of axial shear strain could not occur at the same time, then two
B16 3 170 130 40 16 0.13 separate curves can be drawn.
B17 8 170 130 40 16 0.1
B18 1 170 130 40 24 0.16
B19 3 170 130 40 24 0.2
B20 8 170 130 40 24 0.2
RS- 1 170 130 40 8 0.1
S01
RS- 1 170 130 40 8 0.1
S02
RS- 1 170 130 40 16 0.13
S03
2.3.3. Type of testing

The cyclic triaxial test was conducted with two types of samples
including sea sand samples (B01-B20) and 3 river sand samples
(RS- S01 to RS- S03). To study the failure mechanism of sea sand
sample, the several dynamic loading tests were conducted with
difference amplitude (8-12kPa), corresponding to the maximum
value that can be reached for fill embankment location at 1.5 depth
which is below pavement surface. The dynamic loading caused by
vehicle is calculated according Vietnamese standard as shown
equation below:
403
100
kPa
100 Comment [ln1]: Đường mầu
90 90 xanh bên dưới bị lỗi
80 80 Ngoài ra đường ký hiệu em
70 70 cho thưa ra (khoảng cách giữa
60 Ru (%) 60 dấu tròn hoặc dấu tam giác
cho thưa ra)
Excess Pore Pressure (kPa)
Ru (%)
50 Compression Stress (kPa) 50
Extension Stress (kPa)
40 40
30 30
20 20
10 10
Figure 1 Liquefaction evaluation method 0 0
0 40 80 120 160 200
Cycle
3. RESULTS AND DISCUSSION Figure 3: Test results of sample B01, SR = 0.4; f = 1HZ, stress
3.1 Physical properties of sand amplitude a =32 kPa
Figure 2 shows particle size distribution of sea sand and river sand.
It is observed that the particle sizes of river sand samples are larger 100
kPa
100 Comment [ln2]: ĐƯờng màu
than those of sea sand samples. 90 90 xanh bị lỗi
Ru (%)
80 Excess Pore Pressure (kPa) 80
100
Compression Stress (kPa)
90 70 70
Extensive Stress (kPa)
80 60 60
Ru (%)
Percent Passing (%)
70 50 50
60 River sand sample RS-S01 40 40
River sand sample RS-S02
50 River sand sample RS-S03 30 30
Sea sand sample S01
40 Sea sand sample S02 20 20
Sea sand sample S03
Sea sand sample S05
Sea sand sample S07
10 0 20 40 60 80 100
Cycle
0 Figure 4: Test results of sample B02, SR = 0.3; f = 1HZ, stress
0.1 1 10
Seivesize (mm) amplitude a =24 kPa
Figure 2 Particle size distribution of sea sand and river sand samples
(S01 to S07 are group name of sea sand samples, see in Table 1) 100
kPa
100
3.2 Effect of stress amplitude on liquefaction capacity 90 Ru (%) 90
80 Compressive Stress (kPa) 80
To study the failure mechanism of sea sand sample, the number of Extensive stress (kPa)
70
70
dynamic loading tests were conducted with difference amplitude (8-
60 60
12kPa), corresponding to the maximum value that can be reached
50
Ru (%)
for fill embankment location at 1.5 depth which is below pavement 50
surface. 40 40
In case of amplitude value increasing from 8 to 32 kPa with the 30 30
frequency (f) of 1Hz, the results are presented in Fig. 3 to 8. From 20 20
Table 3, conclusions are given following: 10 10
0 0
 Samples with R95 are not liquefied even the sample B01 0 200 400 600 800 1000
Cycle
having the large vibration amplitude of 32 kPa.
Figure 5: Test results of sample B03, SR = 0.2; f = 1HZ, stress
 Samples with R90 are not liquefied even the sample B04
amplitude a = 16kPa
having the vibration amplitude of 8 kPa.
However, it is observed that the sea sand samples were
liquefied for R90 samples at amplitude of 16 and 24 kPa (see kPa Comment [ln3]: Vẽ lại theo
100 100
Figure 7 and 8). phần mềm
90 Excess pore pressure (kPa) 90
Table 3: Test result of various amplitude 80 Ru (%)
80
Sample Density f (Hz) a Ru Comment 60 60
Ru (%)
index (kPa) 50 50
B01 0.95 1 32 77.5 Not 40 40
liquefied 30 30
B02 0.90 1 24 95.0 Liquefied 20 20
B03 0.95 1 16 21.0 Not 10 10
liquefied 0 0
B04 0.90 1 8 13.5 Not 0 30 60
Cycle
90 120 150
liquefied
B05 0.90 1 16 100.0 Liquefied
B06 0.90 1 24 100.0 Liquefied amplitude a = 8 kPa
404
100
kPa Table 5: Test result for river sand Comment [ln4]: Đường màu
100
90 Ru (%) 90 f a Stress xanh bị lỗi
Sample Density Ru Comment
80 Compressive Stress (kPa)
80 (Hz) (kPa) ratio
70 70 RS-01 0.95 1 8 0.1 32 Not liquefied
60 60
8.2
RS-02 0.9 1 8 0.1 Not liquefied
Ru (%)
50 50
5
40 40
30
RS-03 0.90 1 16 0.2 60 Not liquefied
30
20 20
kPa
10 10 100 100
0 0 90 90
Ru (%)
0 50 100 150 200
Cycle 80 Excess Pore Pressure (kPa) 80
Compressive Stress (kPa)
Figure 7: Test results of sample B05, SR = 0.2; f = 1HZ, stress 70 Extensive Stress (kPa) 70
amplitude a = 16 kPa 60 60
Ru (%)
50 50
40 40
100
kPa
100 30 30 Comment [ln5]: Đường xanh
90 90 20 20 lá cây bị lỗi
80 80 10 10
70 70 0 0
0 200 400 600 800 1000
60 60 Cycle
Ru (%)
50 50 Figure 2: Test results of sample B07, f = 1 HZ, R90

40 40
30 30 kPa
100 100
Ru (%)
20 20
Excess Pore Pressure (kPa) 90 90
Extensive Stress (kPa) 80 Ru (%) 80
0 0 70 Compressive Stress (kPa) 70
0 5 10 15 20 25 30 Extensive Stress (kPa)
Cycle 60 60
Ru (%)
50 50
amplitude a = 24 kPa 40 40
30 30
3.2 Effect of frequency on liquefaction capacity 20 20
10 10
The vibration amplitude is remained constant in this case, to study 0 0
vehicles capacity moving on the road with difference velocity, the 0 200 400 Cycle 600 800 1000
frequency f will be changed to check the liquefaction capacity
Figure 3: Test results of sample B08, f = 2 HZ, R90
corresponding to the samples having R90 and R95. The test results
in this case are shown from Figure 9 to Figure 24.
kPa
From the test results shown in Table 3 and 4, it is clear that in 100 100
the case of the amplitude remained constant to 8 kPa, the frequency 90 Ru (%)
90
varies from 1 to 3 Hz, the pore pressure ratio is less than 95%, all 80 Compressive Stress (kPa)
80
the sea sand samples are not liquefied. 70 70
60 60
Table 4: Test result in case of different frequency of vehicles
Ru (%)
50 50
Sample Density f (Hz) a Ru Comment 40 40
index (kPa) 30 30
B07 0.90 1 8 4.5 Not liquefied 20 20
10 10
B08 0.90 2 8 4.0 Not liquefied 0 0
0 200 400 600 800 1000
B09 0.90 3 8 74.3 Not liquefied Cycle
B10 0.95 1 8 13.0 Not liquefied Figure 4: Test results of sample B09, f = 3 HZ, R90
B11 0.95 2 8 13.3 Not liquefied kPa

100 100
B12 0.95 3 8 17.0 Not liquefied 90 Ru (%) 90
B13 0.95 1 16 100.0 Liquefied 80 Compressive stress (kPa) 80
70 70
B14 0.90 1 16 100.0 Liquefied 60 60
B15 0.95 1 16 100.0 Liquefied 50 50
Ru (%)
40 40
B16 0.95 3 16 100.0 Liquefied
30 30
B17 0.95 8 16 41.5 Not liquefied 20 20
10 10
B18 0.95 1 24 100 Liquefied
0 0
B19 0.95 3 24 68 Not liquefied 0 200 400 Cycle 600 800 1000
B20 0.95 8 24 100 Liquefied

405
kPa kPa
100 100 100 100
90 Ru (%) 90 90 90
Excess Pore Pressure (kPa) Ru (%)
80 Compressive Stress (kPa) 80 80 Excess Pore Pressure (kPa) 80
Extensive Stress (kPa) Compressive Stress (kPa)
70 70 70 Extensive Stress (kPa) 70
60 60 60 60
Ru (%)
Ru (%)
50 50 50 50
40 40 40 40
30 30 30 30
20 20 20 20
10 10 10 10
0 0 0 0
0 50 100 150 200 250 300 350 400 0 200 400 Cycle 600 800 1000
Cycle
Figure 6: Test results of sample B11, f = 2 HZ, R90 Figure 10: Test results of sample B15, f = 1 HZ, R95, Amplitude 16
kPa
kPa 100
100
90 90 100
kPa
100 Comment [ln7]: ĐƯờng màu
80
Ru (%)
Excess Pore Pressure (kPa) 80 90 90 trắng và xanh nước biển bị lỗi
Compressive Stress (kPa) Ru (%)
70 Extensive Stress (kPa) 70 80 Excess Pore Pressure (kPa) 80
Compressive Stress (kPa)
60 60 70 Extensive Stress (kPa) 70
Ru (%)
50 50 60 60
Ru (%)
40 40 50 50
30 30 40 40
20 20 30 30
10 20 20
10
10 10
0 0
0 200 400 600 800 1000 0 0
Cycle 0 50 100 150 200 250 300 350
Cycle
Figure 11: Test results of sample B16, f = 3 HZ, R95, Amplitude 16
kPa
100
kPa
100 Comment [ln6]: Vẽ lại theo
90
phần mềm
90 kPa
100 100
80 80
90 Ru (%) 90
70 70 Excess Pore Pressure (kPa)
80 Compressive stress (kPa) 80
60 60
Ru (%)
70 70
50 50 60 60
Ru (%)
40 40 50 50
30 Excess pore pressure (kPa) 30 40 40
20 Ru (%) 20 30
30
10 Compressive Stress (kPa) 10 20 20
0 0 10 10
0 200 400 Cycle 600 800 1000 0 0
0 150 300 450 600 750 900
Figure 8: Test results of sample B13, f = 1 HZ, R95 Cycle
Figure 12: Test results of sample B17, f = 8 HZ, R95, Amplitude 16
kPa
kPa
100 100
90 90 100
kPa
100 Comment [ln8]: Lỗi đường
90 90 màu trắng, xanh
80 80
70 70 80 80
60 70 70
60
Ru (%)
60 60
Ru (%)
50 50
50 50
40 40
40 40
30 30
Excess pore pressure (kPa) 30 30
20 Ru (%) 20
20 Ru (%) 20
Extensive Stress (kPa) 10 Excess Pore Pressure (kPa)
Compressive Stress (kPa) 10 Extensive Stress (kPa)
0 0 0 0
0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60
Cycle Cycle
Figure 9: Test results of sample B14, f = 1 HZ, R95 Figure 13: Test results of sample B18, f = 1 HZ, R95, Amplitude 24
kPa
406
100
kPa
100 100
kPa Comment [ln9]: Lỗi hình màu
100
90 Ru (%) 90 xanh
Excess Pore Pressure (kPa) 90 90
Comment [ln11]: Dung phần
Compressive Stress (kPa) 80
80 Extensive Stress (kPa) 80 80
70 70
70 70 mềm
60 60
Ru (%)
60 60
Ru (%)
50 50
40 50 50
40
30 30 40 40
20 20 30 30
Excess pore pressure (kPa)
10 10 20 Ru (%) 20
0 0 10 Compressive Stress (kPa) 10
0 100 200 300 400 500 600 700 800
Cycle 0 0
Figure 14: Test results of sample B19, f = 3 HZ, R95, Amplitude 24 0 30 60 90 120 150
Cycle
kPa Figure 18: Test results of sample S3, f = 1 HZ, R90, Amplitude 16
kPa
100
kPa
100 Comment [ln10]: Lỗi hình
Ru (%) 90
90 Excess Pore Pressure(kPa)
80
Compressive stress(kPa)
80 3.3 Comparison of test results between sea sand and river
Extensive stress(kPa)
70 70 sand.
60 60 Sea sand and river sand samples are compared based on the cylic
Ru (%)
50 50 triaxial test results which corresponding to frequency of 1HZ and

40
density ratio of R90, and R95.
40
From the test results shown in Table 4 and Table 5, some
30 30
discussions are given as below.
20 20
10 10  All sea sand and river sand samples were not occur liquefied
under normal condition of vehicle moving on the road (With
0 0
0 150 300 450 600 750 stress amplitude (a) of 8 kPa and frequency (f ) of 1Hz)
Cycle
Figure 15: Test results of sample B20, f = 8 HZ, R95, Amplitude 24  River sand samples could be subjected under condition that the
kPa vibration amplitude is twice higher than those in normal
condition of vehicle moving on the road.
100
kPa
100
 Sea sand samples are liquefied when the vibration amplitude is
90
twice higher than those in normal condition of vehicle moving
90
on the road.
80 80
70 70 Liquefaction state are occurred corresponding to some
60 60 conditions:
Ru (%)
50 50  Sample with R90 and vibration amplitude of 16 kPa

40 40
30 30  Sample with R90 and R95 and the stress value is of 16, 24 kPa
Excess pore pressure (kPa) for B13, B14, B15 and B18 samples.
20 Ru (%) 20
Extensive Stress (kPa) Thus, in the case of embankment having compaction ratio larger
10 Compressive Stress (kPa)
10
0 0 than 95% corresponding to normal amplitude or the value of
0 50 100 150 200 250 300 amplitude is twice higher than normal conditions, the embankment
Cycle
will not occur liquefaction state. Therefore, sea sand material can be
Figure 16: Test results of sample S1, f = 1 HZ, R90, Amplitude 8
used to replace for river sand to fill embankment.
kPa
kPa 4. CONCLUSION
100 100
90 Ru (%) 90 Based on the test results, conclusions are drawn following:
70 70  All of sea and river sand samples are not liquefied under normal
60 60 vehicle moving condition (f-1Hz; stress amplitude a = 8 kPa).
Ru (%)
50 50
 Sea sand samples are liquefied under the vibration amplitude is
40 40
twice higher than normal condition of vehicle moving
30 30
20 20  It is recommended that in case of embankment having
10 10 compaction ratio R90 and R95, the embankment will not be
0 0
liquefied under normal amplitude value (a = 8 kPa).
0 200 400 Cycle 600 800 1000
 Thus, the sea sand can be used to replace river sand for
Figure 17: Test results of sample S2, f = 1 HZ, R90, Amplitude 8 embankment fill.
kPa
407
5. REFERENCES Jakka, R. S., Datta, M. and Ramana, G. V (2010) ‘Liquefaction
behaviour of loose and compacted pond ash’, Soil Dynamics
Bolton, S. H., M., I. I. and Ignacio, A. (1983) ‘Evaluation of and Earthquake Engineering, 30(7), pp. 580–590.
Liquefaction Potential Using Field Performance Data’, P., V. Y. and J., T. (1995) ‘Liquefaction and Postliquefaction
Journal of Geotechnical Engineering. American Society of Behavior of Sand’, Journal of Geotechnical Engineering.
Civil Engineers, 109(3), pp. 458–482. American Society of Civil Engineers, 121(2), pp. 163–173.
Ishihara, K. (1977) ‘Simple Method Of Analysis For Liquefaction Y., S. H. and J., Y. (2014) ‘Failure Modes of Sand in Undrained
Of Sand Deposits During Earthquakes’, Soils And Cyclic Loading: Impact of Sample Preparation’, Journal of
Foundations, 17(3), Pp. 1–17. Geotechnical and Geoenvironmental Engineering. American
Society of Civil Engineers, 140(1), pp. 152–169.
408
Proposal Of Permeability Evaluation Method Of Suspension Grout

K. Uemura1, T. Sasaki2, N. Suemasa3, K. Itoh3, K. Nagao4 and S. Shimada5
1
Graduate School of Engineering Dept. of Civil Engineering, Tokyo City University, Tokyo, Japan
2
Research and Development Division, Kyokado Engineering Co., Ltd, Tokyo, Japan
3
Department of Urban and Civil Engineering, Tokyo City University, Tokyo, Japan
4
Engineering Research Institute, Sato Kogyo Co., Ltd., Kanagawa, Japan
5
President, Kyokado Co., Ltd., Tokyo, Japan
E-mail: g1671702@tcu.ac.jp
ABSTRACT: Large earthquakes have occurred and caused liquefaction damage in urban areas in Japan every few years. Recently, the
applicability of suspension grout has been expanded because of the development of micronized technology of particles. Permeability of
suspension grout has been conventionally evaluated using groutability ratios obtained from sizes of micro-particles contained suspension
grout and particle size of soil. The problem is that permeability is affected not only by grain size of micro-particles of grout but also pore
structure. In this study, in order to evaluate permeability of micro-particles more accurately, the pore index was proposed, which can be
calculated from the Kozeny-Carman equation. And from the results of one-dimensional permeation experiments of micro-particles, it was
confirmed that permeability of suspension grout can be evaluated accurately by the pore index.
Keywords: suspension grout, coefficient of permeability, pore index.
1. INTRODUCTION Gr < 11 not possible of grouting

Gr > 15 possible of grouting
Grouting materials for chemical grout injection methods are
categorized into two types; solution and suspension. Generally, the where D15 is grain size of soil at 15% and G85 is particle size of
solution type of chemical grout has been used for the liquefaction grouting material at 85% in its grain size accumulation curve,
countermeasures under existing buildings. However, this type of respectively. Here, Gr was originally defined as filter criteria to
grout is unsuitable for residential areas because of its high prevent outflows of fine particles through a filter attached to pipe
construction cost. On the other hand, the suspension type of grout drainage (Terzaghi et al. 1948). In the previous studies of Gr
has been mainly applied for the improvement of ground under (Johnson 1958 and Mitchell 1970), the applicability of Gr was
important buildings because of its high strength. However, the discussed.
permeability of suspension grout is lower than that of solution grout However, the permeability of micro-particles is affected by not
(Yoneda et al 1994). Since the size of micro-particles of suspension only grain size of particles but also characteristics of suspension
grout is close to that of pore, there is possibility of clogging. grout. For examples, the effect of particles viscosity on permeability
Therefore, suspension grout has been used for only coarse soil. was confirmed through the injection tests where suspension grouts
Recently, however, micro-particles under sub-μm such as spherical with different viscosities were injected to sand deposit and cracks of
silica grout (Naito et al. 2010) was developed by the latest rock (Yoneda et al. 1993). The permeability was expressed as
micronization technology. Herewith, the permeability of micro- viscous fluid to analyse the behaviour of permeation of suspension
particles has been improved and the applicability of suspension grout (Yoneyama et al. 2015). While the effects of grouting pressure,
grout has been expanded. relative density and concentration of grout were indicated
For reasonable design in chemical grout injection method, it is (Akbulut et al. 2002).
necessary to confirm the permeability of chemical grout and its In addition, permeability is subjected to be affected by
gel-time beforehand. In particular, since the permeation of characteristics of pore of soil as well. It was indicated that the
suspension grout can be impeded because of clogging of micro- permeability of micro-particles changed depending on the relative
particles in the pore of soil, it is essential to confirm the density and the fine content of the soil (Zebovitz et al. 1989). A
permeability of micro-particles to soil. However, most researches in distribution of pore diameter of sand skeleton was measured by “ air
the permeability of suspension grout have mainly targeted the intrusion method” (Kamiya 1999) and based on these results, the
application to cracks of rock (Yoneyama et al. 2015). relationship between capillary spreading pore of soil and grain size
In this study, first, previous researches in the permeability of distribution was formulated using “ diameter estimating grain size
suspension grout are reviewed. Next, from the results of one- distribution” (Fukuda et al. 1997, Kamiya 1999). In the researches,
dimensional permeation experiments of micro-particles, a new the permeability of micro-particles was evaluated from a
permeability evaluation method is proposed. comparison of the size of micro-particles and the minimum size of
pore of soil which was estimated from an uniformity coefficient, a
2. PREVIOUS STUDIES ON SUSPENSION GROUT void ratio and grain size of soil at 50% in its grain size accumulation
PERMEABILITY curve. Moreover, in order to prevent an outflow of fine particles
Permeability of suspension grout has been conventionally evaluated from filters, a stability index of filters was derived from “the
using “groutability ratios” (hereinafter referred to as Gr) obtained constriction size of filter” (Kenny et al. 1985).
from size of micro-particles contained suspension grout and As known from the previous studies, the permeability of
particles size of soil (Mitchell 1981). Gr is proposed in conformity to micro-particles was evaluated by the comparison of a pore diameter
the theory that pore size is closely related to that of soil. Gr is and a diameter of micro-particles. However, a particular
expressed in Eq. (1). experimental apparatus is required to estimate pore diameter so that
it is difficult to apply these methods for practical constructions. In
D15 this study, the pore index was proposed, which is an estimation
Gr  (1) value of pore diameter obtained from a coefficient of permeability
D85 of soil. Since the coefficient of permeability can be generally
estimated from a preliminary investigation in chemical grout
injection method, this method is easier for permeability evaluation
and suitable for practical constructions.
409
In this paper, in order to establish a new evaluation method, the its applicability was confirmed through one-dimensional permeation
pore index was formulated from the Kozeny-Carman equation and experiments of micro-particles for mixed sand
Table 1 Properties of Samples
Figure 1 Grain size distribution of sand samples
3. DERIVATION OF PORE INDEX lr nd 2  w g h (4)

v' 
In order to estimate the permeability of micro-particles more l 32  w x
accurately, it is necessary to evaluate pore diameter of soil. In this
chapter, a derivation method of the pore index is proposed using the Assuming that Darcy flow can be applied for permeation in pore of
Kozeny-Carman equation which is based on the Hagen-Poiseuille the soil skeleton, Eq. (4) is converted to Eq. (5)
equation (Loudon et al. 1999).
The Kozeny-Carman equation (Emmanouil et al. 2012) was 1 l r nd 2  w g
chosen among the several previous studies regarding coefficients of v' (5)
permeability of soil in this paper because this equation is formulated 32 l w
from the pore conditions of soil such as grain size of soil, porosity n,
and correction of curvilinear pipe in pore of the soil skeleton. where k is a coefficient of permeability.
Assuming that pore of the soil skeleton is circular capillary tube, In Eq. (5), circular channel is assumed and according to Carman
actual velocity v in a circular tube is expressed as Eq. (2). (Carman et al. 1937), ratio of lr to l (lr/l) is approximately 2.5.
Therefore, the pore index d is expressed as Eq. (6).
r 2  w g h (2)
v'  w
32  w x d  4 5 k (6)
n w g
where r is diameter of circular tube (cm), ρw is density of water
(g/cm3), μw is viscosity of water (g/(cm・s)), g is acceleration of As seen from Eq. (6), the pore index d is not depending on grain
gravity (cm/s2) and ∂h/∂x is hydraulic gradient. In Eq. (2), diameter size of soil particles but the coefficient of permeability and the
of circular tube r is estimated as uniform pore diameter of soil porosity. Furthermore, containing the parameter of viscosity of fluid,
skeleton. In this paper, estimated r is defined as the pore index d; an Eq. (6) can be applied for the cases even when grouting material is
index to evaluate permeability of micro-particles. regarded as viscous fluid.
Next, actual velocity v in Eq. (2) is converted to Darcy
velocity v, to be applied for the permeation in pore of the soil 4. PERMEABILTY EVALUATION METHODS OF
skeleton. Since v is velocity of flow in pore of the soil skeleton, it MICRO-PARTICLES
is necessary to consider bending-related extension of pipe in pore of In order to confirm the applicability of the pore index for
the soil skeleton when converting. The relationship of actual permeability evaluation of micro-particles, one-dimensional
velocity v and Darcy velocity v is expressed as Eq. (3) permeation experiments were carried out. First, soil samples for
these experiments were introduced. Next, the pore indexes of soil
1 lr samples were estimated from coefficients of permeability samples.
v' v (3) Finally, from the results of one-dimensional permeation experiments
n l of micro-particles, applicability of the pore index was confirmed.
where l is a length of soil specimen (cm), lr is length of bending- 4.1 Summary of samples for permeation experiments
related extension of pipe in pore of the soil skeleton (cm) and n is
porosity of the soil. Then, lr expresses the actual length of the flow The samples used for the experiments were mixed sand with
0.075mm to 2.000mm in diameter. The conditions and the grain size
channel. Eq. (4) is obtained by substituting Eq. (3) for Eq. (2).
410
distribution of the samples are shown in Table. 1 and Figure 1. In 4.3 ONE-DIMENSIONAL PERMEATION EXPERIMENTS
order to consider the effect of uniformity coefficient; Uc and the OF MICRO-PARTICLES
grain size of soil at 15%; D15 on permeability of micro-particles,
mixed sand was prepared by mixing classified sands. The D15 of 4.3.1 Experimental conditions
samples No.1 to No.4 were made to be same. In order to confirm the One-dimensional permeability experiments of micro-particles were
effect of uniformity coefficient, Uc on permeability of micro- carried out using the sand samples in Figure 1. The flow chart of
particles, the Uc of samples No.5 to No.13 were set at 3.0 to 5.5. permeability evaluation of micro-particles is shown in Figure 4.
Then, Uc of samples No.14 to 17 were made to be same to compare Even though at least 50cm in radius is required for the ground
the change of the coefficients of permeability depending on the improvement in chemical grout injection methods, in order to save
particle sizes. time and effort, micro-particles were permeated to the specimens
with 5.0cm in diameter and 10.0cm and 15.0cm tall (hereinafter
4.2 Permeability tests of mixed sand referred to as short specimens) in first screening. From these result,
In order to calculate the pore index, permeability tests were carried the samples for second screening were decided. In second screening,
out on each sand sample shown in Figure 2. The specimens for specimens with 50.0cm or 90.0cm tall (hereinafter referred to as
permeability tests were prepared to be 5.0cm in diameter and long specimens) were prepared. When the volume of permeated
10.0cm tall and their void ratio e was set at 0.82 by air-pluviation micro-particles is greater than that of pore in a long specimen, the
method. Only in the sample No.16, the void ratio was set at e 0.70. micro-particles are determined to be permeable.
Then, CO2 gas and de-aired water were infiltrated into them to The samples of specimens are shown in Figure 1. Both the grain
heighten the saturated rates and permeation tests were conducted. size distribution of micro-particles measured by the laser diffraction
Here, the coefficients of permeability of specimens were the average particle size analyser (SALD-3100 produced by Shimadzu
of three values measured within 0.5 of hydraulic gradient. Corporation) and the grain size of micro particles at 85% G85 are
The results of permeability tests were shown in Figure 2. Since shown in Figure 5. The weight concentration C of micro-particle
the void ratios e of specimens were almost equal, the dispersion in was set by mixing with water as follows; WC-1(G50=18.5μm), WC-
the permeability seen in Figure 2 was likely to be caused by the 2(G50=14.9μm) and WC-3 (G50=8.9μm). Then, they were permeated
differences in uniformity coefficient; Uc and grain size of sand into the specimens with stirring to avoid deposition of micro-
particles. Therelationship between the calculated pore index d and particles. The concentrations of mixture C were set at 10.0% in
the grain size of soil at 15% D15 is shown in Figure 3. From the every case, which was calculated from Eq. (7).
figure, the pore index d cannot be expressed by D15 uniquely.
Figure 2 Void ratio e and coefficient permeability k of specimen
Figure 4 Decision of permeation micro-particles
Figure 3 The pore indexes d and D15
Figure 5 Grain size (µm)
411
Table 2. Experimental Conditions of Specimens

Height of Type of
Sample
Case No. Specimens micro- Gr
No.
H (cm) particles
Case1 1 14.97 - WC-1 7.85
Case2 2 14.90 - WC-1 7.75
Case3 3 15.02 50.11 WC-1 7.69
Case4 4 14.96 50.02 WC-1 7.65
Case5 5 9.97 50.02 WC-1 6.47
Case6 6 10.00 49.81 WC-1 5.11
Case7 7 9.99 - WC-1 3.98
Case8 8 9.98 - WC-1 5.68
Case9 9 10.03 50.00 WC-1 5.92
Case10 10 10.02 49.98 WC-1 8.76 Figure 6 Summary of experimental apparatus
Case11 11 10.01 49.99 WC-1 8.17
Case12 12 9.99 49.89 WC-1 6.75
Case13 13 9.95 50.11 WC-1 6.16
Case14 14 15.00 50.00 WC-1 13.33
Case15 15 15.00 - WC-1 7.29
Case16 16 - 90.00 WC-2 9.68
Case17 17 - 90.00 WC-2 4.99
Case18 17 - 90.00 WC-3 6.53
Mm (7)
C 100(%)
Mw
where Mm is weight of micro-particles and Mw is that of water,

respectively. In addition, the concentrations of mixture C and
particle size of micro-particles before/after permeation were
measured to confirm that no flocculation/separation had been
generated in the specimens. than 11 is regarded as unpermeable and Figure 7 Detail of experimental apparatus
shown in red. As seen from this table, the micro-particles in most of
the cases were determined to be unpermeable by conventional From the results, the injection pressure decreased due to
methods. clogging accompanied by increasing of injection volume and the
The apparatus of the experiment and its details are shown in decreasing trends were different depending on the physical
Figure 6 and Figure 7, respectively. In order to permeate properties of samples. In order to confirm the effect on the
microparticles uniformly, fine gravel with 2.000mm to 3.360mm in permeability of micro-particles by both characteristics of grain size
particle size was paved as a filter layer on top and bottom of the and coefficients of permeability of samples, the results of the
specimens. Mixture was stored in a column in Figure6 and injected specimens with 10.0cm tall shown in Figure 8(a) were sorted as
into the specimens by air-pressure. At this time, to avoid volume Figure 9(a)~(c) where each figure was sorted with reference to the
expansion of specimen and fracture grouting, 5kPa air-pressure was properties of Case5. Comparison was made on coefficient of
loaded on the upper piston besides the injection air-pressure. In permeability k, uniformity coefficient Uc and grain size of soil at
Case1 to Case15, the initial hydraulic gradient was set at 15% D15 in Figure 9(a), (b) and (c), respectively. Furthermore,
approximately 20. In addition, when the permeation velocity Table.3 shows the details of physical properties of indicated Cases
decreased to almost 0 due to clogging, the injection pressures were in Figure 9. Each result was discussed in the next section.
increased to 100kPa and 200kPa on short and long specimens, (a) Difference in coefficient of permeability k
respectively. Here, their limitations were set not to occur leak of In Figure 9(a), the effect of difference in the coefficient of
mixture or fracture grouting. Neither sand running out nor fracture permeability k was considered. From Table.3, even though
of specimen was visually observed during injection. In Case16 to coefficient of permeability k in Case12 was only 1.4 times as great
Case18, the permeability of micro-particles with different particle as that of Case5, the permeability of micro-particles was
sizes was confirmed under 10kPa of constant injection pressure. substantially different. Comparing between the results of Case5 and
Furthermore, the drainage was measured by a platform-scale. Case12, although clogging occurred at very early stage in Case5, the
injection velocity hardly decreased by clogging and the mixture was
4.3.2 Experimental results able to be permeated more than three times and the permeated
amount was as much as the pore volume in Case12.
Experimental results of short and long specimens are shown in (b) Difference in uniformity coefficient Uc
Figure 8(a)~(d) which indicate the permeation volume normalized In Figure 9(b), the effect of difference in uniformity coefficient Uc
by the pore volume of a specimen. In Figure 8, when vertical axis is was considered. Even though coefficients of permeability are similar
greater than 1, the pore water of the specimen was considered to be values, higher permeability of micro-particles was observed when
replaced by the mixture of micro-particle and water more than once. the uniformity coefficient was smaller. In the case of high
In addition, the relative amount of injection volume of each uniformity coefficient, it was confirmed that the sand skeleton was
specimen became smaller according as the height of specimen constructed by various sizes of sand particles mutually and
became greater. accordingly clogging occurred locally in small pores of the sand
skeleton, which led low permeability of micro-particles even at the
initial stage of injection. From this result, it was revealed the
permeability of micro-particles is affected by uniformity coefficient
Uc.
412
(c) Difference in grain size of soil at 15% D15 As observed above, the permeability of micro-particles cannot
In Figure 9(c), the effect of difference in grain size of soil at 15% be evaluated accurately only by Gr. Figure 11 shows the relationship
D15 which is used for Gr was considered. All the cases in Figure 9(c) between uniformity coefficient Uc and the pore index d normalized
showed same trend, that is, the injection velocities decreased to by the grain size of micro-particles at 85% G85. It is clear whether
almost 0 by clogging. From this result, it was confirmed that the size the micro-particles are permeable or not in Figure 11. The
of micro-particles was close to those of pore diameter even in Case5 experimental discriminant was obtained from uniformity coefficient
where the particle size was bigger than other cases. As seen from Uc, a pore index d and the grain size of micro-particles at 85% G85
Figure 9(a), on the other hand, even though the D15 of Case13 was using the following equation; Eq. (8).
smaller than that of Case5, satisfactory permeability was obtained.
From these results, it was confirmed that it is difficult to evaluate d
 0.887  U c  8.125 (8)
permeability of micro-particles only by Gr.
G85
4.4 Permeability evaluation of micro-particles
If calculated value from Eq. (8); d/G85 is plotted upper than the
From the results of experiments, the permeability of micro-particle approximate line, permeation of micro-particles could be determined
was evaluated. Figure 10 shows the relationship between Gr and to be possible. For obtaining an equation with high-precision, it is
coefficient of permeability k. A red frame indicates a range of Gr necessary to accumulate more experimental data since Eq. (8) was
less than 11, which means the permeation of micro-particles is estimated from the measured values during the experiments.
estimated to be impossible. Figure 10 indicates that permeation of
micro-particles succeeded in some cases even in a red frame.
Figure 8 Experimental result of one-dimensional regarding height of specimen
Figure 9 Experimental result regarding index (H=10.0cm) Figure 10 Relationship of Gr and coefficient permeability k
413
Fukuda, M. and Uno, T. (1997) “Analysis of “method of

classification of soils” based on proposed “diameter
estimating grain-size distribution””, Journal of JSCE,
No.582/III-41, pp.125-136. (In Japanese)
Johnson, J. (1958) “Cement and clay grouting of foundations:
Grouting with clay-cement grouts”, Journal of Soil
Mechanics and Foundations Division, Proceeding of the
American Society of Civil Engineers, Vol.84, No.SM1, pp.1-
12.
Kamiya, K. (1999) “Evaluation of void diameter distribution of
sandy soils and its utilization”, Gifu University Doctoral
Dissertation, pp.57-74. (In Japanese)
Kenny, T. C., Chahal, R., Chiu, E., Ofoegbu, G. I., Omange, G. N.
Figure 11 Relationship of d/G85 and uniformity coefficient Uc
and Ume, C. A. (1985) “Controlling constriction sizes of
granular filters”, Canadian Geotechnical Journal, Vol.22,
Table 3 Properties of Each Specimens at Figure 8 pp.32-43.
Case Specimen k D15 Mitchell, J. K. (1970) “In-place treatment of foundation soils”,
Uc Journal of Soil Mechanics and Foundations Division,
No. No. (cm/s) (mm)
5 5 7.78×10-2 0.228 5.07 Proceeding of American Society of Civil Engineers, ASCE,
6 6 7.91×10-2 0.180 5.06 Vol.91, No.SM1, pp.73-110.
8 8 7.41×10-2 0.200 5.14 Mitchell, J. K. (1981) “Soil improvement: State-of-the-art”, Proc. of
9 9 7.35×10-2 0.208 4.06 the 10th int. conf. on soil mechanics and foundation eng.,
Vol.4, pp.509-565.
12 12 1.11×10-1 0.238 5.35
Naito, M. Kishi, H. Fukuoka, N. Yamada, T. and Ishida, H. (2010)
13 13 1.02×10-1 0.217 5.51
“Development of superfine spherical silica grout as
alternative grouting material for the geotechnical disposal of
5. CONCLUSIONS long-lived radioactive waste”, Proc. of 18th int. conf. on
nuclear eng., pp.1-7.
In this study, first, the pore index d was proposed using coefficient
Loudon, C. and Mcculloh, K. (1999) “Application of the Hagen-
of permeability k and porosity of soil n. Next, one-dimensional
Poiseuille equation to fluid feeding through short tubes”,
permeation experiments of micro-particles were carried out. Finally,
Annals of the Entomological Society of America, Vol.92,
the permeability of micro-particle was evaluated by the pore index d.
No.1, pp.153-158.
The conclusion obtained is shown as follows.
Terzaghi, K. and Peck, R. B. (1948) Soil mechanics engineering
1. The pore index was formulated using coefficient of
practice, John Wiley & Sons, Inc., pp.50-51.
permeability k and porosity of soil n from the Kozeny-Carman
Yoneda, S. and Nakagawa, K. (1993) “Basic properties and
equation without grain size of soil.
penetration of cement suspension grouts with different grain
2. Permeability of micro-particles is affected by not only particle
sizes”, Journal of JSCE, No.462/VI-18, pp.101-110. (In
size but also coefficient of permeability k or uniformity
Japanese)
coefficient Uc.
Yoneda, S. Okabayashi, S. Tamura, M. and Mori, A. (1994)
3. It is difficult to evaluate the permeability of micro-particles by
“Engineering properties and micro structures of fine sands
Gr which has been used in conventional evaluation methods.
injected with suspension grouts of ultra-fine granular cement
4. Permeability evaluation of micro-particles can be expressed by
and supplementary waterglass”, Journal of JSCE, No.493/III-
the pore index d accurately.
27, pp.109-118.
Yoneyama, K. Tosaka, H. Okuno, T. and Nishi, T. (2015) “A study
6. REFERENCES
on flow model of cement grout in rock fracture”, Japanese
Akbulut, S. and Saglamer, A. (2002) “Estimating the groutability of Geotechnical Journal, Vol.10, No.4, pp.545-557. (In
granular soils: a new approach”, Tunnelling and Underground Japanese)
Space Technology, Vol.17/No.4, pp.371-380. Zebovitz, S., Krizek, R. J. and Atmatzidis, D. K. (1989) “Injection
Carman, P. C. (1937) “Fluid flow through granular beds, Institution of fine sands with very fine cement grout”, J. Geotech. Eng.,
of Chemical Engineers”, Vol.15, pp.32-48. ASCE, Vol.115, No.12, pp.1717-1733.
Emmanouil, S. Chistos, G. and Georgios, A. (2012) “Kozeny-
Carman and hydraulic conductivity of compacted clayey
soils”, Geomaterials, Vol.2, No.2, pp37-41.
414
Compacted Polymer-Enhanced Bentonite-Sand Mixture – Behaviour and Potential

Applications
S.S. Agus1 and Y.F. Arifin2
1
Engineering Consultancy Firm, Singapore
2
Civil Engineering Department, Lambung Mangkurat University, Banjarmasin, Indonesia
E-mail: agus72indo@gmail.com
ABSTRACT: An investigation into behaviour and potential applications of a polymer-enhanced bentonite-sand mixture (PEBSM) in its
compacted state had been carried out in an environmental controlled laboratory. The experiments undertaken included microscopic
visualization of the mixture, drying-wetting processes, consolidation and shear strength tests. The results obtained indicate that depending on
the overburden pressure acting on the compacted mixture and its water content, settlement in combination with swelling and shrinkage may
result in cracks or on contrary may close the initial cracks induced by the mechanical compaction in the field. The conclusions have been
drawn based on an analytical assessment performed on the field and laboratory data gathered. A discussion is also brought up on the
difference between two approaches used – single-valued effective stress and two-independent stress state variable approaches. The two
approaches predict different values of suction at which crack starts to occur and different time-settlement behaviour.
Keywords: Bentonite, swell/shrink, settlement, shear strength, unsaturated soils, cracks.
1. INTRODUCTION The bentonite presence is beneficial to the hydro-mechanical

behaviour as reported earlier by Arifin et al. (2006). As described in
Polymer-enhanced bentonite-sand mixture (PEBSM) has been various publications, the magnitude of swelling pressure of
adopted as landfill liner and/or landfill cover in several countries compacted samples containing bentonite and its hydraulic
particularly in Europe. A study on a PEBSM was undertaken conductivity depend on the dry density of the compacted material.
pertaining to this application and a field measurement was taken in a Depending on the magnitude of vertical pressure acting in the
location in Germany. This innovative material was developed for the field, the compacted mixture, when in contact with water, may
landfill usage and mainly consists of a granular material (sand) exhibit swelling or settlement and/or cracks under the influence of
mixed with a small amount of sodium type bentonite and a minute weather i.e. during drying-wetting cycles. In contrast, another
amount of polymer with certain proportions. The following figures possible scenario may occur, whereby initial cracks earlier induced
show the environmental scanning electron microscopy photos taken by mechanical compaction or some other processes close during the
and reported by Agus et al. (2007). The photos clearly depict cycles.
bentonite-polymer nets that have been formed when water is added During drying cycle, it is highly possible that suction exists in
to the mixture. The minute amount of polymer added to the mixture the compacted mixture and its magnitude changes with time and
covers the surface of bentonite clusters and sand grains. may reach zero value when saturation is attained. The addition of
polymer in the mixture improves resistance of the compacted
material to crack. Visual observation made by Schanz et al. (2004)
indicates that when exposed to the same suction value the
compacted PEBSM exhibits no apparent cracks in contrast to the
compacted mixture without polymer addition.
(a) Without polymer addition (b) With polymer addition
Figure 2 Crack developed under suction application on

compacted specimens (from Schanz et al., 2004)
Theoretically, the development of cracks in the compacted

mixture can be predicted utilising the unsaturated soil principles.
Two different approaches can be used for this purpose; namely,
using the single-valued effective stress concept and the two-
independent stress state variables approach. This is the subject
brought out in this paper. Data from a field measurement have been
Figure 1 Polymer in interaction with bentonite particles (from Agus utilised in the study. Critical suction at which cracks start to occur is
et al., 2007) also formulated.
415
2. FIELD TEST DATA 3. STRESS-STRAIN RELATIONSHIP

The field test data that have been used in the study had been As described in the preceding section, the field experimental data
gathered from a landfill site located in Germany. A test field of had been analysed based on two approaches; namely, the single-
about 2,800 m2 in area had been established at the site, where valued effective stress concept and the two-independent stress state
suction measurement by means of tensiometer had been undertaken variables approach. The formulation of both approaches is described
in the PEBSM layer for about 4.5 years’ duration. below.
The degree of saturation of the PEBSM was calculated from the
measured suction (in this case matric suction) using its soil-water 3.1 Single-Valued Effective Stress Concept
characteristic curve, which was measured and reported in Schanz et This concept is based on the framework proposed by Bishop (1959),
al. (2004). which is formulated as:
The SWCC of the compacted PEBSM portrays its good water
retention characteristic as indicated by a small drop in the degree of   (  u a )  s (1)
saturation over a “wide” range of suction (up to 100 kPa).
Where ( – ua) is the net stress,  is a coefficient which depends
mainly on the degree of saturation, Sr, soil structure and cycle of
wetting, drying, or stress change, and s is suction, which is defined
as matric suction or (ua-uw).
The  coefficient is a function of degree of saturation of soil and
the following simple form can be used:
  f (S r )  Sr (2)
Vertical and horizontal effective stresses can subsequently be

defined as below, with Ko represents the coefficient of earth
pressure at rest:
 v '  ( v  u a )  S r s (3)
 h '  ( h  u a )  S r s  K o ( v  u a )  S r s (4)
Figure 3 Measured matric suction and computed degree of
saturation of the compacted PEBSM layer As suction in the field changes, the effective stress of the liner
will also change, which follows the following equation:
 v '   h '  S r s  sS r (5)
Assuming the Ko condition applies, which is reasonable for this

case, the relationship below holds:
  v  u a   S r s   2  h  u a   S r s   (6)

1
 v or  vol 
E
where E is the modulus of elasticity with respect to change in net

stress and υ is the Poisson ratio of the soil. The following figure
shows changes in the vertical and horizontal effective stresses due to
changes in the soil suction, which have been derived using Figure 3,
Figure 4 and Equation (5).
60
Figure 4 Soil-water characteristic curve for the compacted PEBSM vertical effective stress
speciemen (from Schanz et al., 2004) 50 horizontal effective stress
s'v, s'h (kPa)
40
On site, the PEBSM was laid approximately 1 meter deep from
the final ground surface leading to a surcharge of approximately 20 30
kPa or slightly lesser. The initial conditions of the PEBMS in the
20
field test are as following:
10
 Initial dry density (d) : 1.75 Mg/m3
 Initial void ratio (e) : 0.52 0
28.07.01
05.11.01
13.02.02
24.05.02
01.09.02
10.12.02
20.03.03
28.06.03
06.10.03
14.01.04
23.04.04
01.08.04
09.11.04
17.02.05
28.05.05
05.09.05
14.12.05
 Initial water content (w) : 5%

 Initial degree of saturation (Sr) : 25.6%
With a 1-m thick recultivation and drainage layer, the total Date
vertical overburden stress acting in the middle of the compacted Figure 5 Changes in vertical and horizontal effective stresses with
PEBSM of 7 cm can be computed to be equal to 20.6 kPa. time
416
The Ko condition only holds when soil blocks do not develop as 4.3 Modulus of Elasticity with Respect to Change in Net
a result of cracks that develop due to desiccation. In the case of Stress (E)
crack, the horizontal net stress follows:
The E value has been derived using the elastic theory from the
 h  u a     v  u a   1  2  S r s
saturated oedometer test data for the unloading-reloading, which is
(7) deemed to be within the elastic range:
1    1   
e 1 1 1   1  2 
mv   (15)
The changes in vertical and horizontal strains in the event of
crack can be expressed as:
1  e0   ' E (1   )
where mv is the coefficient of volume compressibility of soil. The
  v  u a   S r s  2 S r s
1
 v  (8) average mv value obtained from the saturated oedometer test results
E of the PEBSM specimen compacted at dry side as shown in Figure 6
is 1.798x10-5 /kPa. Thus, the corresponding E value is 36,134 kPa.
1   S r s    v  u a   S r s  
1
 h  (9)
E 0.65
3.1 Two-Independent Stress-State Variable Approach 0.60
The formulation of this approach is outlined in Fredlund and 0.55
Void ratio, e
Rahardjo (1993), where changes in two state variables used (i.e. net
stress and suction) and the resulting effects are considered 0.50
separately.
In the case of the Ko condition (Δεv = Δεvol), the volumetric 0.45 Swelling at 7 kPa
strain is given as: Loading
0.40 Unloading
1   1  2    u   1     s (10) Reloading
E 1    H 1   
v a 0.35
1 10 100 1000
In the above equation, H represents the modulus of elasticity of Vertical stress, s v (kPa)
the soil with respect to suction change (the 2nd stress state variable), Figure 6 Oedometer test result on a saturated compacted
while as described before, E signifies the modulus of elasticity for PEBSM specimen
the effect of net stress (the 1st stress state variable) change.
During crack, the horizontal net stress is defined as: 4.4 Modulus of Elasticity with Respect to Change in Suction
(H)

 h  u a    v  u a   E
s (11) The H value was computed from the e-log suction plot that had been
1    H (1   ) obtained while determining the SWCC curve (Schanz et al., 2004).
The SWCC had been determined at three-dimensional free
The changes in vertical strain (Δεv) and horizontal strains (Δεh) swell/shrink conditions. Therefore, the following relationships hold:
after crack are formulated as:
 v   vol  Ct  v  u a   C a s (16)
 v  u a   2 ( h  u a )  s
1 1 (12)
E H e 1 e 1
Ct  ; Ca  (17)
1  e0    ua  1  e0  s
( h  u a )   ( v  u a )  ( h  u a )
1 (13)
E Analogous to the unloading-reloading path, from which the E
value has been derived, the corresponding drying-wetting cycles of
4. PARAMETERS USED IN COMPUTATION the PEBSM should also be used to determine Ca value. However,
The following parameters have been used in the computation and due to the unavailability of the wetting data, only the drying path
the values have been derived as outlined in this section. has been considered in the derivation of Ca value. The following
1) Coefficient of earth pressure at rest (Ko) figure shows the drying test on the material.
2) Poisson ratio (υ) 0.58
3) Modulus of elasticity with respect to change in net stress (E) Experimental (drying)
4) Modulus of elasticity with respect to change in suction (H)
0.56
4.1 Coefficient of Earth Pressure at Rest (Ko)
Void ratio, e
0.54
The Ko value was determined from the experiment carried out in
Geodelft, Netherlands. The value is equal to 0.52.
0.52
4.2 Poisson Ratio (υ)
The Poisson ratio (υ) is calculated from the elastic theory as follows 0.50
(with Ko equal to 0.52): 10 100 1000 10000 100000 1000000
Suction, st (kPa)
KO
  0.34 (14) Figure 7 Change in void ratio with increasing suction from a
1  KO drying test on compacted PEBMS
417
The void ratio-suction curve exhibits a bimodal characteristic The subsequent changes in vertical and horizontal strains were
with negligible changes for suction above 500 kPa. The linear computed based on an assumption that the changes in vertical and
portion of the curve up to 500 kPa suction is within the suction horizontal net stresses are equal to zero. This is adopted for both
range, at which the material was subject to in the field and thus can approaches when computing the evolution of deformation and
be used to determine the Ca value. From the curve, the Ca value is settlement due to changes in suction of the PEBSM layer on site.
1.017x10-4 /kPa, and thus a H value of 29,500 kPa as for this case Ca
= 3/H. 5.3 Comparison of Computed Strain and Settlement
It is shown in the two tables that the use of two different approaches
5. DEFORMATION AND SETTLEMENT
affects the initial strains and deformation computation for the case
Computation of settlement and deformation as a result of the change with cracks with the two-independent stress-state variable approach
in suction, which was measured by the tensiometer, has been carried estimating half the values compared with the single-valued effective
out using the parameters derived as described in the preceding stress concept.
section. The following figure illustrates that, for the case of Ko condition,
the use of H value from the drying curve result in much higher
5.1 Computation using Single-Valued Effective Stress settlement estimated using the two-independent stress-state variable
Concept approach. This is inconsistent since at the Ko condition, cracks do
not occur and both approaches should give approximately the same
As described in Section 2, a change in the net vertical stress ((v- result.
ua)) of 20.6 kPa was applied on the compacted PEBSM layer after
placement. The corresponding change in the net horizontal stress -0.15
((h-ua)) is therefore equal to 10.7 kPa. The initial suction of the
PEBSM was determined from the SWCC given the placement water -0.10
Settlement, D H (cm)
content and dry density of the compacted layer and the value was
4.4 MPa. Immediately after placement, the tensiometer measured -0.05
20.1 kPa and therefore the material was subjected to a significant
suction change once placed in-situ. The corresponding change in 0.00
degree of saturation of 59% can be computed from the SWCC.
The initial strain and deformation due to placement can be 0.05
computed and the values are summarised in the following table. Two-independent stress state variable
0.10
Single-valued effective stress
Table 1 Initial Strain and Deformation Computed using the Single-
0.15
valued Effective Stress Concept
28.07.01
05.11.01
13.02.02
24.05.02
01.09.02
10.12.02
20.03.03
28.06.03
06.10.03
14.01.04
23.04.04
01.08.04
09.11.04
17.02.05
28.05.05
05.09.05
14.12.05
Without cracks With cracks
Vert. strain, v (%) -3.3 -3.2
Hor. strain, h (%) 0 -3.3 Date
Figure 8 Comparison of settlement with time from two different
Vert. deform. ,dv (cm) -2.3 -2.3 approaches
Hor. deform. ,dh (cm) 0 -3.3
The variation of time-settlement plot for the two-independent
Note: -ve indicates expansion
stress state variable approach is shown in the following figure using
the realistic H value equal to 8 and 10 times that determined from
5.2 Computation using Two-Independent Stress-State the drying test as described earlier. It is demonstrated that an
Variable Approach average value of 9H provides the best fit to the settlement plot from
As mentioned earlier, the H value for the compacted PEBSM layer the single-valued effective stress concept.
was derived from the drying path. Realistically, the value for the
wetting path is much higher than that from the drying cycle. The -0.020
following table summarises the initial strain and deformation due to -0.015
Settlement, DH (cm)
placement computed using the two-independent stress state variable -0.010

approach.
-0.005
0.000 Single-valued effective stress
Table 2 Initial Strain and Deformation Computed Using the Two-
independent Stress-state Variable Approach 0.005 Two-independent stress state
0.010 variables (8xH)
Without cracks With cracks Two-independent stress state
0.015 variables (10xH)
Vert. strain, v (%) -3.3 -1.6
0.020
Hor. strain, h (%) 0 -1.7
28.07.01
05.11.01
13.02.02
24.05.02
01.09.02
10.12.02
20.03.03
28.06.03
06.10.03
14.01.04
23.04.04
01.08.04
09.11.04
17.02.05
28.05.05
05.09.05
14.12.05
Vert. deform. ,dv (cm) -2.3 -1.3

Hor. deform. ,dh (cm) 0 -1.7
Date
Note: -ve indicates expansion Figure 9 Determination of actual H value
In this computation, the realistic H value used in the The evolution of vertical strain assuming Ko condition is shown
computation was first approximated by matching the settlement plot in Figure 10 for both methods. The single-valued effective stress
from both approaches. It is later demonstrated that taking the concept consistently gives lesser vertical strain compared with that
realistic H value of 8 to 10 that of the drying is reasonable and an given by the two-independent stress-state variable approach,
average value of 9H (i.e. 265,500 kPa) was subsequently used. although the magnitude does not differ significantly.
418
Figure 11 and Figure 12 demonstrate that the two-independent sufficiently deep to facilitate full development of preferential flow
stress-state variable approach produce lower values of vertical and paths.
horizontal strains for the case when crack occurs. The suction value, at which cracks first occur is defined as the
critical suction (scr) and the value can be computed by equating the
-3.36 net horizontal stress with the maximum tensile strength (tmax).
-3.34 Schanz et al. (2004) reported that tmax value of the PEBSM is 20 kPa
(%)
on average.
-3.32 The following equations represent the critical suction for the
v
Vertical strain, e
-3.30 single-valued effective stress concept and the two-independent

stress-state variable approach, respectively.
-3.28
 v  ua   tmax  1    1
-3.26
 
scr   (18)
-3.24
independent stress state variables  1     1  2  S r
-3.22
single effective stress
 
 v  ua   tmax  H 1   
-3.20
scr   (19)
 
28.07.01
05.11.01
13.02.02
24.05.02
01.09.02
10.12.02
20.03.03
28.06.03
06.10.03
14.01.04
23.04.04
01.08.04
09.11.04
17.02.05
28.05.05
05.09.05
14.12.05
 1    E
Equation (18) must be solved iteratively since Sr is the function

Date of suction, while scr from Equation (19) can be computed
Figure 10 Vertical strain with time computed using the two straightforward since both E and H are assumed to be constant with
approaches (Ko condition) net vertical stress and suction, respectively. Nevertheless, it is
imperative to note this assumption may not be always valid since in
reality, E, H and tmax are also stress and suction dependent.
Vertical, horizontal strain,e v, e h(%)
-3.34
The critical suction can also be plotted as a function of the total
-3.32 vertical stress using both the above equations. The following figure
shows such plot for the PEBSM studied.
-3.30
-3.28 100
Single effective stress
Total vertical stress, s v (kPa)
-3.26 approach
80 Independent stress-state
-3.24
variables approach
-3.22 horizontal strain
vertical strain 60 No crack Crack
-3.20
No crack Crack
28.07.01
05.11.01
13.02.02
24.05.02
01.09.02
10.12.02
20.03.03
28.06.03
06.10.03
14.01.04
23.04.04
01.08.04
09.11.04
17.02.05
28.05.05
05.09.05
14.12.05
40
Date 20
Figure 11 Vertical and horizontal strains computed using the single-
valued effective stress concept (after crack occurs) 85 kPa 147 kPa
0
-1.68 0 50 100 150 200 250 300
Vertical, horizontal strain, ev, e h(%)
Critical suction, scr (kPa)

-1.66 Figure 13 Critical suction versus total vertical stress computed
using the two approaches
-1.64 The above figure reveals that at the applied total vertical stress
of 20.6 kPa (the air pressure is atmospheric), the scr value is 85 kPa
computed using the single-valued effective stress concept and 147
-1.62
horizontal strain kPa using the two-independent stress-state variable approach. At
vertical strain any given applied total vertical stress, the two-independent stress-
-1.60 state variable approach gives a higher value of scr compared with the
other approach. Further detailed investigation will warrant clarity at
28.07.01
05.11.01
13.02.02
24.05.02
01.09.02
10.12.02
20.03.03
28.06.03
06.10.03
14.01.04
23.04.04
01.08.04
09.11.04
17.02.05
28.05.05
05.09.05
14.12.05
which cracks really commence during drying.

Considering that the maximum measured suction is only
approximately 50 kPa (see Figure 3), cracks will not occur in the
Date compacted PEBSM layer. It may not be the case when polymer is
Figure 12 Vertical and horizontal strains computed using the two- not added to the compacted bentonite-sand mixture due to presence
independent stress-state variable approach (after crack occurs) of macro- and micro-pores within the mixture.
The mechanism of crack in the compacted bentonite-sand
mixture has been described in detail by Agus (2005). Consider a
5. CRITICAL SUCTION AT FIRST CRACK representation of a compacted bentonite-sand mixture in Figure 14.
OCCURRENCE A rapid desiccation results in movement of water in the inter-
Resistance to cracks is provided by the tensile strength of the soil aggregate pores (macro-pores), which is not accompanied by
and thus crack occurs when the negative net horizontal stress sufficient macro-pore volumetric contraction. The presence of
exceeds this value. Nevertheless, when desiccation takes place, the polymer in the compacted mixture will prevent this rapid
extent of cracks in the PEBSM layer in the field may not be desiccation process to occur, at least at low suction (i.e. below 100
419
kPa), which is practically even higher than the maximum value that 2) Strain and deformation computed using the two-independent
can occur in the field when the compacted PEBSM layer is used in stress-state variable approach are approximately half those
landfill. derived using the single-valued effective stress concept. The
former computation is realistic when adopting a realistic
modulus of elasticity with respect to change in suction
(realistic H) of 8 to 10 times H derived from the drying test.
3) Critical suction at which cracks start to occur (scr) has been
computed and the value is consistently lower for the two-
independent stress-state variable approach at any given net or
total vertical stress (with atmospheric air pressure).
8. REFERENCES
Agus, S.S. (2005) An Experimental Study on Hydro-Mechanical
Characteristics of Compacted Bentonite-Sand Mixtures.
Doktor-Ingenieur Dissertation, Bauhaus-University Weimar,
Germany.
Agus, S.S., Arifin, Y.A., and Schanz, T. (2007) “Settlement and
swells/shrink behaviour of a compacted polymer-enhanced
Figure 14 Possible representation of a compacted bentonite-sand bentonite-sand mixture as induced by field mechanical
mixture (from Agus, 2005) loading and wetting-drying cycles”, Proceeding of Brazilian
Symposium of Unsaturated Soils, Salvador, Brazil, 2007.
Arifin Y.F., Agus S.S., and Schanz T. (2016) “Swelling pressure
6. POSSIBLE APPLICATIONS and hydraulic conductivity of compacted polymer-enhanced
Due to its characteristics as described, the compacted PEBSM is bentonite-sand mixture”, Proceeding of Geotechnische
suitable for landfill application. The material is better placed under a Aspekte in Umweltschutz 2006, Stahlman and Rosenberg
drainage layer to ensure that suction remains low and deep cracks Edition. Brauschweig, 87-98.
are not present during the landfill design life. The effectiveness of Fredlund D.G., and Rahardjo H. (1993) Soil Mechanics for
placing the compacted PEBSM below the waste is questionable Unsaturated Soils. John Willey & Sons. Canada.
since interaction between the bentonite-polymer with leachate from Schanz, T., Agus, S.S., and Tscheschlok, G. (2004) “Hydraulisch-
the landfill may degrade the beneficial properties of the mixture. mechanische Eigenschaften einer polymerverbesserten Sand-
Bentonit- Mischung beim Einsatz im Deponiebau”,
Geotechnik, 27, 4: pp344-355.
7. CONCLUSION
An investigation into the behaviour of a compacted PEBSM has
been conducted by means of a field case study. The following ACKNOWLEDGEMENT
conclusions can be drawn from the outcome of the study. The authors would like to express sincere gratitude to the late Prof.
1) A maximum suction value of 50 kPa was measured during an Dr.-Ing. habil. Tom Schanz for his guidance when this research
approximately 4-year period in the case study presented in this work was carried out in Bauhaus-University Weimar, Germany. He
paper. The corresponding drop in degree of saturation as was also the authors’ PhD supervisor in the same university. He
computed from the soil-water characteristic curve (SWCC) is passed away in 2017 in Bochum, Germany, where he and his family
hardly below 80%, which indicate a good water retention lived.
capability of the material.
420
Geosynthetic Reinforced Road Structure as Fast Rehabilitation

for a Typhoon Disaster
H. Hermina1, Y. Jeff1, and S. Henry1
1
ACE Geosynthetics, Taichung, Taiwan, R.O.C
E-mail: hermina.ho@geoace.com
ABSTRACT: Disasters are increasingly difficult to predict due to severe climate change. In addition to disaster prevention, it is an urgent
task to prepare for post-disaster restoration and prevent secondary damage to the environment to reduce the impact of disasters. The
construction site in this paper is located on the primary trunk road which connects eastern and western Taiwan. Due to major typhoon
disasters, the slope below the road severely collapsed. It needs to be repaired in a short period to reduce the impact on transportation and
economy. Considering the issue of time-urgency and cost-saving, geosynthetics reinforced structure in accordance with concrete retaining
wall and pile foundation was constructed as road embankment. In addition to achieve a rapid recovery after disaster, the construction method
also lowers carbon emission environmental impact. In this paper, geosynthetic reinforced structure has been proved as a sustainable solution
for disaster recovery.
Keywords: disaster recovery, geosynthetics reinforced structure, carbon emissions, green engineering
1. INTRODUCTION and made of sand, silt and gravel. The SPT-N value of the
foundation soil is equal to 20 – 50 ((1) Chen, et al. 2017).
Since the discovery of the greenhouse effect as one of the causes of Taiwan is located in the orogenic belt from obduction of micro-
climate change, green engineering for energy conservation and oceanic plate into Eurasian continental plate. The orogeny
carbon reduction have gradually been introduced into the design and movement resulting in a fractured basement rock with several
construction of engineering projects (Heerten, 2012). In recent geologic structure such as fault and fold (Sibuet and Hsu, 2004). In
years, Reinforced Concrete (RC) retaining structures have been this case, only a fractured basement rock is observed, without any
partially replaced by Geosynthetics Reinforced Structures (GRS), occurrence of fault and fold nearby.
which excludes the casting and curing time of concrete, to shorten The site is located adjacent to Fenggang River. It is one of the
the construction time and disaster recovery process. Compared to major river contributed to a high sediment erosion in Taiwan,
GRS, RC structure also demands higher material quantity, especially during the typhoon season ((2) Chen, et al. 2017). The
transportation cost, and environmental impact (WRAP, 2010). main river has 41 km length with average elevation of 698.3 m and
This paper presents a case study of slope rehabilitation project average slope of 28.4o.
using a composite structure made of pile foundation, RC wall, and
GRS. The case is located in the southern part of Taiwan, specifically
3. DESIGN
in Pingtung County, Provincial Highway No. 9 470K+500, adjacent
to Fenggang River. In 2013, Typhoon Kong-Rey was attacking In the design part, several factor need to be considered upon
Taiwan with heavy rainfall, in which the concentrated rainfall rise (FHWA-NHI-00-043). First, the main cause of collapsed slope is
upon 250 to 300 mm within a single day (JTWC, 2013). Rapid scouring phenomenon in the toe of the slope. Therefore, the new
change of precipitation rate caused a fluctuation in the adjacent river constructed slope should be made to prevent any damage from the
water level. The impact of high energy and high velocity in the river scouring. The occurrence of typhoon and earthquake are also a main
stream was an accelerated scouring of soil. Thus, a slope failure was concern in this structure. Typhoon will results in river stream
occurred beneath Provincial Highway No. 9 (Figure 1). fluctuation, whereas earthquake will impact the overall stability of
the structure.
The collapsed slope is 150 m in length and 22 m in height. At
the bottom, all-casing pile foundations were built to deal with the
scouring phenomenon from Fenggang River. On the top of pile
foundations, a semigravity RC wall was designed to deal with the
fluctuation of river stream due to typhoon impact. For the upper
structure, two-tiered GRS consists of geogrids, soil bags, and water
culvert was installed (Figure 2).
Figure 1 Slope Failure below Highway Embankment
2. SITE DESCRIPTION
Pingtung County of Taiwan was geologically located above
Chaochow Formation, Quaternary Terrace Deposit, and Quaternary
Alluvium Deposit. Chaochow Formation consists of slate, a
metamorphic rock, with thinly bedded sandstone. Terrace Deposit
and Alluvium Deposit are younger sediment from Quaternary period Figure 2 Design Drawing of Composite Structure
421
3.1 All-casing pile foundation pressure inside the geogrid reinforced slope. The facing system of
the GRS was piled up soil bags as the vegetation medium. Above
The foundation was set with double-row cased piles with a diameter the GRS, a soil slope with a gradient of 1:0.3 and 2 meter height was
of 1.2 meters (Figure 3). The outside piles were set at intervals of 2 backfilled and compacted as the road base.
meters in a row, and the interval of rear piles was set as 5 meters.
The depth of each pile is about 10 m with an initial drill penetration
of at least 2 meters into the rock sublayer. A total of 155 piles were
installed to transfer the vertical load into metamorphic basement
rock and improve the bearing capacity of the whole structure.
Furthermore, a reduced influence of riverbed scouring was expected,
and the sliding drive force raised by the earth pressure of slope was
resisted.
Figure 5 The Installation of GRS
3.4 Drainage System

In order to discharge the water from heavy rain on the slope, the
design of the drainage box culvert under the road surface was also
carried out. The overflow from the road surface and upper slope can
be collected by the drainage system and then transported into the
adjacent river. In this structure, drainage system will prevent the
cohesion-loss from the soil, which is caused by surface water
Figure 3 The All-Casing Pile Installation infiltration (Figure 6).
3.2 Semigravity RC retaining wall

The semigravity RC retaining wall was constructed on the top of
pile foundation and connected to the pile cap (Figure 4). Taking into
account the highest water level, the wall high was raised up to 12
meters. The main function of the RC retaining wall is to prevent the
main structure from being damaged by the impact of debris flow
during typhoon season and resist the lateral earth pressure from the
backfill.
Figure 6 Drainage System for Surface Erosion Measurement
3.5 Hydroseeding
Hydroseeding, a plantation work using a mixture of seed and mulch,
was performed after the completion of GRS installation (Figure 7).
The main function of hydroseeding is to control the surface erosion
using vegetation growth, while it also provides an aesthetical
function to enhance the appearance of the whole GRS structure.
Overall, hydroseeding has a beneficial impact on the environment
Figure 4 Semigravity RC Retaining Wall Construction aspect by lowering the carbon footprint from the construction work.
3.3 Geosynthetics Reinforced Structure

The upper structure above highest water level was constructed as a
geogrid reinforced slope (Figure 5) to a certain height and then
backfilled with soil to the level of road base. According to the safety
stability analysis, the geogrid reinforced slope was considered to be
divided into two stages. The lower stage used geogrid with an
ultimate tensile strength of 240 kN/m and the upper stage utilized
160 kN/m separately. The embedded length of the geogrid for both
stages is 10 meters. The backfilled material utilized the collapsed
soil in the construction site and the dredged sand from the river.
Besides the mechanically stabilized components, the design of
drainage in the structure is greatly important as well, the drainage
system composed of drainage board, non-woven geotextile and
gravel, are placed into the structure to guide and drain the water Figure 7 Hydroseeding
422
4. SAFETY STABILIZATION ANALYSIS 6. CONCLUSION

The overall structure was analyzed by commercial software Since the structure was completed in 2014, it has been through
STEDwin for global safety and stabilization under normal, several natural disasters, such as earthquake and typhoon. In recent
rainstorm and earthquake conditions (Figure 8). The soil parameters years, the structure remains intact with no deformation is observed
for analysis were set as follows in Table 1 according to the on the site (Figure 9). Therefore, it can be seen that the GRS is a
geological report for the construction site. The analysis result green engineering method that combines rapid construction with
demonstrates that the composite structure made of pile foundation, ecological landscapes and carbon reduction. Certainly, although
RC wall, and GRS is stable under normal, typhoon, and earthquake there are so many advantages of GRS, a geological limitation might
conditions. The achieved factor of safety for normal condition be a concern. This paper presents a combination between GRS and
(FSnormal) is 1.63, typhoon condition (FSTyphoon) is 1.27, and traditional RC structure to deal with natural hazard, such as typhoon
earthquake condition (FSearthquake) is 1.38. and earthquake. This kind of structure is completely suitable for a
country like Taiwan, which has frequent natural disasters. Due to the
Table 1. Soil Parameters used for Stabilization Analysis many advantages of the composite construction method, this method
Parameters In-situ soil Backfill soil has gradually been welcomed and favoured in Taiwan for disaster
γ (kN/m3) 18.5 18.5 recovery process. This paper suggests a sustainable solution for
disaster recovery engineering in countries with similar geological
γsat (kN/m3) 20.5 20.5 conditions to Taiwan.
Cohesion, c (kPa)
10/5 5/3
General/ Storm
Friction angle, ϕ (°) 32 30
Figure 9 GRS Structure Condition in 2018

Figure 8. The Global Stabilization Analysis
(Presenting Earthquake Model as an Example) 7. REFERENCES
(1)
5. CARBON EMISSION Chen, M.H., Hsu, W.I., and Chung, W.K., 2017, The
Development and Application of Carbon Emission Estimation
The carbon emissions of the project using the composite for Highway Construction, Journal of Taiwan Highway
construction method are as shown in Table 2 ((1)Chen, et al. 2017). Engineering, Vol. 43(4), pp. 2 – 29 (in Chinese)
Regardless of planting, the carbon footprint of the traditional (2)
Chen, C-W., Oguchi, T., Hayakwa, Y.S., Saito, H., Chen, H., Lin,
construction method with entirely reinforced concrete components is G-W., Wei, L-W., and Chao, Y-C., 2017, Sediment Yield
reduced by 30%. during Typhoon Events in Relation to Landslides, Rainfall,
and Catchment areas in Taiwan, Geomorphology (2017)
Table 2. Carbon Emission of Each Construction FHWA-NHI-00-043, 2001, Mechanically Stabilized Earth Walls
Carbon Emission (TCO2e) and Reinforced Soil Slopes Design & Construction
Structure Traditional Guidelines, U.S. Department of Transportation, Federal
This Case Highway Administration, pp. 81-201.
Method
Heerten, G., 2012, Reduction of climate-damaging gases in
Cantilever retaining wall 0 715
geotechnical engineering by use of geosynthetics, Geotextile
Geosynthetics reinforced
80.73 0 and Geomembrane, Vol. 30, February 2012, pp. 43-49
structure
Join Typhoon warning Center (JTWC), 2013, Annual Tropical
Semigravity RC retaining
1336.65 1336.65 Cyclone Report, pp. 30
wall and pile foundation
Sibuet, J-C., and Hsu, S-K., 2004, How was Taiwan Created,
Total 1417.38 2051.65 Tectonophysics 379, pp. 159 – 181
Waste and Resources Action Programme (WRAP), 2010,
Sustainable geosystems in civil engineering applications,
Geosystems Report
423
424
The use of the Observational Method in Deep Excavations for the Realization of
a Residential Compound in a Complex Hydrogeological Context
M. Carassini1, F. Bucci2, A. Antiga3
1
Technical Director, Bonatti S.p.A. BU Civile & Infrastrutture, Parma, Italy
2
Engineering Manager, Bonatti S.p.A. BU Civile & Infrastrutture, Parma, Italy
3
Technical Director, Soil S.r.l., Milan – Italy
E-mail: andrea.antiga@gmail.com
ABSTRACT: The urban redevelopment works of the "Railway Station Area ex-Boschi”, was developed on an area of nearly 17.0000 m2, it
entailed the execution of deep excavations in a difficult geotechnical and hydrogeological context. In order to overcome these critical issues a
design based on an observational approach was developed. This design considered several possible hydrogeological scenarios and staged
excavations over small areas.
The Observational Method has been successfully extended from the more traditional case of tunneling works to the less usual case of deep
excavations in large construction areas.
Keywords: Observational Method, deep excavations, diaphragms
1. INTRODUCTION Real estate surfaces and the remaining € 42.3 mln delivered
according to periodic progress payment certificates, as per initial
The urban redevelopment project of the "Railway Station Area ex- contract data. The Project contract duration was 57 months. From a
Boschi” area in Parma (Italy), involved the creation of a large road technical and organizational point of view, the Project was
link, under-passing the existing eight-track railway line Station and characterized from the outset of the basic design phase, by a strong
the construction of five new buildings. interaction with the existing urban context and with the local
The works, developed on an area of nearly 17.0000 square meters, infrastructures. In fact, within the Train Station or in the immediate
entailed the execution of considerable excavations and supporting vicinity, there are:
structures.
 public and private buildings both in the north and south areas,
The hydrogeological context was complex because of soils with
 the existing railway station, whose outer masonry envelope and
poor mechanical characteristics and of two different aquifers. The
steel decorations had to be preserved according to the constraints
presence of an important railway line in operation and several
imposed by the local Superintendence of historic buildings,
existing buildings next to the excavations, completed the project
framework.  bus stops and terminals,
In order to overcome these critical issues and assure greater  road and pedestrian traffic ways,
safety and continuity of the works, the Contractor developed a  the Milan-Bologna railroad tracks, with the need to maintain at
detailed design proposal based on an observational approach, which least four lines always running at the same time, being this the
considered several possible hydrogeological scenarios and staged main Italian railway line.
excavations over small areas. As a result, the contract basic design solution was conceived to
In this case, the Observational Method has been successfully minimize as much as possible the impacts and risks associated with
extended from the case of tunnels works, more traditional, to the the execution of works.
less usual case of deep excavations in large construction areas, For example, the need to carry out deep excavations required the
showing its own peculiarities and giving important indications for execution of perimeter r.c. diaphragms along the entire project area.
future applications. This technique permitted the use of machinery and workmanship
from the top ground level before proceeding to the excavation and
2. RIQUALIFICATION PROJECT underground works.
In 2007, the Municipality of Parma (Italy), through its subsidiary 3. GEOTECHNICAL WORKS IN THE NORTH AREA
company "STU Società di Trasformazione Urbana", launched a
tender for the public works of the urban renovation of the Parma The renovation project envisaged the creation of a pedestrian and
Railway Station Area. traffic road under-passing of the rail tracks from the North to South
The contract basic design was developed by an association of areas of the Railway Station, from which the new access to the
design firms with MBM Arquitectes from Barcelona as a leader and station would be created, and two levels of underground parking in
it essentially entailed the renovation and upgrading of the existing the north area.
Railway Station of fascist epoch and of both the North and South These works covered a very large area of approximately 17,000
Areas of the Station, as identified in figures 1 and 2. This was square meters and resulted in significant excavations of varying
achieved through the following interventions: depths ranging up to approx. 15m below the existing ground level.
 redesigning the respective squares with their green areas, the The excavations were carried out in a complex and variable
traffic roads and pedestrian walkways, hydrogeological context because of the different excavated soil
 linking the two areas with the creation of an underground layers and their variable characteristics, due to the extension of the
roadway underneath the existing station and the entire railway area.
tracks park, Specifically, the geotechnical context of this area is
characterized by an intermediate silty-clay layer with poor
 providing in the northern area, formerly dedicated to railway
mechanical characteristics, while the hydrological context is
storage and maintenance, two floors of public and private
characterized by the presence of a first phreatic water table in direct
underground car parks, a newly constructed bus station and
contact with the Parma river – called A0 – and a second artesian
buildings, including residences, offices and a hotel.
water table under pressure – called A1 – both characterized by
The works were awarded to the JV Bonatti SpA - Di Vincenzo
seasonal level variations (fluctuation above 5 m annually).
under a Public-Private partnership procedure for a work amount of:
€ 98.3 mln, of which € 56 mln as compensation for the value of the
425
 excavation depth 15.30 m; Diaphragms depth 30 m x 100 cm

thickness;
 n. 5 levels of provisional type supporting brackets (harmonic
steel tie rods);
 jet grouting treatments at the foot of the excavation for a volume
9 m wide x 7,5 m deep;
 1.4 m deep r.c. foundation slab acting as permanent contrast to
the r.c. lateral diaphragms.
Figure 1 Pre-existing status
Figure 3 Excavations in the north area and railway tracks
4. GEOTECHNICAL AND HYDROGEOLOGICAL

CONTEXT IN THE NORTH AREA
In order to better appreciate the geotechnical and hydrogeological
context, it has been very useful to analyse, using historical maps, the
events that determined the current stratigraphic structure of the area;
they show that the stream of the Parma river extended, at the end of
the last century, even hundreds of meters beyond today's shores.
Of great interest is the morphological configuration of the Parma
river at the end of 1700, as represented by the Atlas Sardi (1767) in
figure 4, characterized by the presence of an important meander in
the area under examination.
The stratigraphic and hydrogeological structure of the area was
investigated to the depth of 50 meters, by carrying out many
continuous core drilling, pore pressure (piezometer) observations
and laboratory tests and also by correlations with other site
investigation surveys carried out in neighbouring areas.
As shown in figure 5, the stratigraphic succession is:
 Layer A - composed of polygenic gravel with sand and silt and it
Figure 2 Project simulation is characterized by high permeability. It is testimony of an
ancient bend of the Parma river.
The contractual basic design and the detailed design to be  Layer B - a complex of sediments with fine granulometry and
developed by the Contractor had to consider all these site issues, cohesive behaviour.
together with a context of inhabited buildings and existing  Layer C - composed of polygenic gravel, with sand and silt. It is
infrastructure adjacent to the excavations, involving the need to limit present on the whole area at almost uniform depth.
interferences, as well as ongoing and long-term settlements. The It should be noted that, inside layer B, there are 6 sub-layers
following sequence of works was utilized: made of silty clays and clayey silts, locally sandy silts, characterized
1. perimeter barrier consisting of r.c. diaphragms executed from the by significant differences in grain size distribution and in the over-
ground level, followed by consolidation ratio. The plasticity index values are within CL and
2. excavation steps of about 2 m to 4 m depth each, followed by: CH classes. The grain-size analysis showed silt content ranging
3. execution phases of supporting multistrand tieback anchors between 20% and 50% and clay content varying between 40% and
4. execution stages of jet grouting soil consolidation treatments of 70%. This variability was also found between samples taken at the
the bottom ground, and finally same depths, indicating the presence of lenses with different grain
5. excavations down to design level and casting of final supports size distribution even in the same sub-layer. For structural
works (diaphragm foot restraint foundation r.c. slab). calculations, the 6 sub-layers can be grouped in one layer (level B)
The section with the deepest excavation height occurred in the with homogeneous mechanical characteristics (figure 5). On the
North area. The contractual basic design typical cross section other hand, the variability in the grain-size distribution and in the
solution, named 1A, consisted of: plasticity index values of the layer B has an important influence on
 n.2 underground levels used for under passing roads and strand anchor bond pull-out strength, on the homogeneity and on the
walkways and car parkings (excavation from 56.50 m. down to mechanical characteristics of the jet grouting ground improvement.
41.20 m above sea level - a.s.l.); The hydrogeological analysis allowed to identify the presence of
two aquifers (A0 and A1) and to define the oscillations of water
426
levels to consider in the design. The first aquifer (A0) is present Figure 5 shows a geotechnical section (layers A, B, C -
within level A; it is a phreatic aquifer whose levels are directly boreholes (BH and CPT).
correlated with the Parma river; during summer time the The most significant geotechnical parameters are shown below:
groundwater is absent, as it reaches the ground level in the river
flood stages. The second aquifer (A1) is present within level C. This
is an important artesian aquifer of considerable hydro-potable Ɣ Փ c' Cu
interest confined by the silt-clayey layer B. (kN/m3) (°) (kPa) (kPa)
Polygenic gravel (layer A) 20.5 39 0 -
Silty clay/clayey silt (layer B) 19 21 15 80
Polygenic gravel (layer C) 20.5 39 0 -
5. MAIN DESIGN, ORGANIZATIONAL AND

CONTRACTUAL CRITICAL ISSUES
From the Design point of view, the variability of the water levels
affects the overall behavior of the structures (reinforced concrete
diaphragms) representing a variation of the external load (water
pressure); it affects the value of the resulting stresses in the
diaphragms, the actions on the strand anchors and the resulting loads
acting on the volume of solid ground in jet grouting at the foot of the
diaphragm.
The variability in the chemical-physical characteristics of level
B (point 2) does not affect much the actions on the structures, but it
has a more direct and significant influence on the effectiveness of
the consolidation interventions (mechanical properties of the jet
grouted solid ground volume) and the tensile strength of the strand
anchor bulbs).
From the Contractors point of view, the main issue was the
Figure 4A Current situation
management of the lump sum contract risk in terms of costs and
execution times. In this case, the risk was characterized by: the
verification and taking over of the Client's basic project design and
the subsequent development of the design detail, the high degree of
complexity of the works and the boundary conditions imposed by
the operating context. From an organizational point of view, the
urbanized area was a limiting factor for: the spaces dedicated to the
construction of site installations and equipment, the internal and
external road accessibility to the yard, the work shifts schedule, and
it enforced the limits of noise and vibration emissions with impact
on site productions rates etc. ...
In the light of the Client's basic project, substantially based on
simplified average calculations and assumptions, and the actual
variability of projected project hypotheses, the Company considered
it appropriate to study, within the detailed design by the consultant
designers, a proposal for a design improvement variation order, with
the aim of reducing the risk in terms of time and cost of the project
and increasing the safety factors of geotechnical works.
Figure 4B Eighteenth century (Atlante Sardi) This was achieved by:
 developing a construction site and logistics plan respectful of the
phases of the works and road access;
 deciding a rational sequence of the work phases with activities to
be completed in one single shift (casting & pouring, drillings,
injections, etc. ...);
 realization of propaedeutic field testing for geotechnical works
and implementation of the results and sequences of the works in
the sizing of the works themselves;
 introducing a variability concept in the design interventions and
execution phases and using an “active tests and monitoring
system” during construction to adapt the design to actual site
conditions.
6. BASIS OF THE CONTRACTOR’S DESIGN

IMPROVEMENT PROPOSAL
According to the Italian law in force at the time (2009-2010), the
Contractor could, during the course of the work, propose to the
Client modifications/ improvements to the works as long as:
1. they led to a reduction in the original amount of work,
Figure 5 Stratigraphic section north area
427
2. they were intended to improve the functional aspects, as well as probable" resistance characteristics of the jet-grouting and strand
the individual technological elements or components of the anchor bulbs to be used in the project.
project, Afterwards, during the works, specific testing and control fields
3. they did not entail a reduction in the performance of the Project were established and carried out to verify the actual anchor bulbs
itself and kept the work execution time and the worker's safety tensile strength and jet grouting mechanical resistance in areas
conditions unaltered. where, even after preliminary investigations, the greatest variability
In addition, for the Italian and European standards in force remained.
(Eurocode n.7 1997, NTC2008), the adoption of the so-called The strand anchors tensile test field has determined the
Observational Method in the design of provisional and final minimum design resistances of the ground layers interested by the
geotechnical works was innovative, if compared with the previous anchor bulbs (superficial deposits and gravel, clayey limestone-clay)
technical regulations, under the following main conditions: according to the drilling diameters, see figure 6. The jet-grouting
 that the particular complexity of the geotechnical situation and test field has determined the technique (Bi-fluid), cement type and
the importance of the work was established; quantity, execution parameters (feed rate and rotation speed,
 that the acceptability limits were established of the values of injection pressure, etc.), see figure 7.
those parameters values which represent the ground-structures The introduction of local variability and execution phases in the
system’s overall behavior; detailed design related to the geotechnical and foundational works,
 that other alternative solutions congruent with the project were together with the execution of such tests during the works, allowed
verified as possible and envisaged; to satisfy loading hypothesis of more conservative and wider
 that an appropriate monitoring system was set in place during variance and to absorb the uncertainties of geotechnical aspects of
works and control plans aimed to confirm the assumptions made direct relevance on ground support interventions (strands anchors,
and the validity of the design solution adopted or the adoption of jet grouting base r.c. foundation slab excavation and casting).
one of the proposed alternatives if the acceptability limits were Furthermore, as a result of the various studies, it was found that the
reached. basic design assumptions on material resistance could be improved
As a result of the comparison between all the various available and opening the finale excavation level (foundation slab) for limited
data and design proposals developed in a preliminary form, the phases and dimensions would have created improved conditions for
choice was made in favor of an Observational Method of design the perimeter diaphragm stability, as further explained below.
(Peck 1969, Ciria 1999).
For the choice of the design parameters, the approach
recommended by Peck was used with the adoption of "most
probable" values, the evaluation of "most unfavorable scenarios”
and the definition of mitigation measures for such situations.
Schematically, the solution adopted, according to the national
regulation requirements, was developed according to the following
macro steps:
1. design and calculation of works by adopting "most probable"
water levels;
2. analysis of possible scenarios with rising water levels up to "most
unfavorable" values and definition of mitigation measures for
such situations;
3. active monitoring during works of the two water table levels, to
identify the real scenario.
We want to point out that the unavoidable geotechnical
uncertainties identified in the previous chapter (with two possible
scenarios: "more probable" and "most unfavorable") made it
absolutely unsatisfactory to use a classic design approach based on a
single "predefined design".
Indeed, in this case, the generalized use of "most unfavorable" Figure 6 Strand anchor tests
values would have resulted in excessive costs. On the other hand,
less prudent values, such as the generalized use of "most probable"
values would have led to a risk level considered unacceptable.
With regard to the maximum piezometric load, detailed analyses
and simulations were carried out on the basis of the historical data
available from the hydro-potable wells in the Municipality of Parma,
which showed a strong variability in the aquifer levels with annual
oscillations up to 5m.
At the end of the analysis, for the detailed design it was decided
to consider:
 as “most unfavourable” occurrence: the value of 48 m a.s.l. for
the artesian aquifers (A1) and 51.50 m a.s.l. for the phreatic
aquifer (A0),
 as “most probable” occurrence: the value of 45.50 m a.s.l. for
the artesian aquifer (A1) and 49.00 m a.s.l. for the phreatic
aquifers (A0).
The contractual basic design considered a single value equal to
the value of 45.50m a.s.l. for the artesian aquifer (A1) and 49.00m
a.s.l. for the phreatic aquifer (A0).
Regarding the aforementioned influence of ground composition
and plasticity values variability, field tests have been performed
before the detailed design phase in order to evaluate the "most Figure 7 Jet grouting tests
428
7. CONSTRUCTION PHASES OF THE CONTRACTOR

The excavation methods involved progressive excavations stages of
about 3m depth, starting from the existing ground level, down to the
drilling and setting of the various strand anchor arrays down to the
last level (+43.50 m a.s.l.). Specially studied construction site
internal circulation, required continuous operability of the main
access ramp from the only access situated to the North-west side of
the site, adjacent to the B-C side of the diaphragms. For this
purpose, six main stages of internal road configurations and
execution phases were studied and implemented into the detailed
design, of which phase no. 2 is reproduced as an example in fig. 8.
The excavation down to the bottom to the level of r.c.
foundation slab from the last anchor strands array level (+43.5 m
a.s.l.) was carried on according to finite square-shaped elementary
trenches of small dimensions (approximately 10x10m ÷ 20x20m).
The choice of limited dimension trenches allowed to combine
multiple needs, among which:
 to avoid congestion of the work equipment circulation inside the
yard and perform daily r.c. slab casting and pouring phases
completed within single shifts;
 to optimize the resources used in the various trenches phases
(excavation, lean concrete, waterproofing, reinforcement laying,
casting and concrete pouring) and limiting the reinforcement
overlapping between adjacent bars;
 to handle sudden increases in the level of the two aquifers.
In addition, the excavation and casting sequence of the r.c.
foundation slab proceeded from the center of the north area up to the
border of the perimeter, leaving to the last excavation stage a ground
berm against the r.c. diaphragm in order to create the necessary
contrast.
After the casting of the foundation slab in the central portion of
the North area, the perimetral ground berm has been excavated
according to a continuous "open and close" mode, for non-
contiguous sections or cones, according to a "comb" scheme, to
better manage the criticality due the perimeter supporting structures.
Figure 9 shows this concept.
Figure 8 Step n.2 of internal circulation and execution phases

Figure 9 Execution sequence for the r.c. foundation slab
429
This also allowed the beneficial 3D effects to be developed, as 8.1 Variability in the sizing of the supporting works
described in the following paragraph.
The design improvement proposal establishes two “project limit
configurations”: a “minimum configuration” and a “maximum
8. DESIGN DETAILS OF THE CONTRACTORS
configuration”, as a result of the approach described in the geology
section referring to the levels of aquifers (they identify the “most
Following the "open and close" mode of execution, we could rely on probable” occurrence and the “most unfavorable” occurrence of
a set of beneficial static effects otherwise unavailable with a general water levels for aquifers A0 and A1).
excavation: Typical sections of the excavation support works have been defined,
1. three dimensional effect of contrast in the foot of the diaphragm where a variability has been hypothesized for ground anchor
wall due to the presence of the unexcavated perimetral berm characteristics (free strand length, bond length and drilling
and/or the foundation slabs; diameter), for jet grouting (geometries and spacing of columns,
2. plate behavior of the jet grouting band (the whole jet grouting execution quotas) and for the execution procedures of the
volume is affected by a limited section of diaphragm than would excavation steps below height 43.50 m a.s.l. as per figure 11.
be with a general excavation); By varying these parameters, it was possible to cover the
3. overload effect behind the diaphragm due to the weight of the cast combined effect of:
foundation slab that allows the increase of the passive earth 1. possible variations in the aquifers (A0, A1) level (i.e. load) within
pressure by increasing the effective stresses; the considered range;
4. undrained behaviour for the cohesive layer B during the 2. possible variations in the mechanical properties of jet grouting
construction phases below the excavation quota of 43.5 m. a.s.l. and anchor bond pull-out strength (due to the possible variations
Three-dimensional numerical analysis, using 3D Flac software, in layer B characteristics).
was performed to evaluate the beneficial effects on the stress-strain The method set out that the support works are dimensioned and
behaviour of the reinforced concrete diaphragm wall produced by verified in detail within the defined range (minimum to maximum),
the excavation and concrete pouring stages of small size segments depending on actual site conditions encountered during works.
(see figure 10). Two-dimensional analyses were not suitable to The definition of support works within the range defined in the
capture the confinement effects provided by the unexcavated berm detailed design occurred at each single excavation stage and in the
and the diffusion of the stresses due to the jet-grouting band; indeed, presence of significant increases in the aquifers level, for which “in
in a traditional two-dimensional calculation, the excavation of the progress detailed design” documentations were issued.
perimetral berms, even of a single segment, is equivalent to the
complete removal of the berm through the whole longitudinal
development of the diaphragm wall, situation this which does not
correspond to reality.
Figure 11 Typical transverse section with perimetral berm
8.2 Monitoring procedures during the works

In applying this approach, it was therefore crucial to implement a
specific system of controls and monitoring during the works.
This monitoring system played an active role in the realization of
Figure 10A 3D model the work because, through a constant interpretation and analysis of
the collected data, it allowed the modulation of the interventions
within the “minimum” and “maximum” design configuration.
The parameters monitored during works with an active role were:
a) aquifers A0 and A1 levels, b) homogeneity and mechanical
characteristics of the jet grouting ground improvement and c) anchor
bond pull-out strength.
A topographic monitoring system to control the settlements of
buildings and lateral diaphragms wall displacements during
excavation was also installed, but this monitoring system plays a
very much passive role, its aim was only to check original
predictions and for general safety precautions, but it didn’t have any
role in the design modulation.
A piezometric network was installed to control the groundwater
level. Specifically: n.6 open standpipe piezometers to measure the
level of aquifer A0, n.7 open standpipe piezometers to measure the
level of aquifer A1, n.4 Casagrande Standpipe Piezometer to
Figure 10B horizontal displacements caused by berm excavation
measure pore-water pressure within the cohesive layer B.
430
8.3 Field Testing of jet-grouting soil improvement and tieback In these occasions activities proceeded either on less critical
stranded anchors work fronts, or excavations proceeded according to the above
strategies 2.ii or 2.iii.
After the execution of the preliminary tests, by proceeding with the
excavations, the execution of field tests during construction were
also carried on according to an observational approach.
This was done along the perimeter, where there were expected
particular situations, in order to assess in detail the influence of
geotechnical and stratigraphic characteristics.
Detailed notes for the interpretation of the results of individual test
fields and calibration of project support works have been issued.
The execution of the field tests and controls during construction,
followed by the issuance of the specific “in progress detailed
design” reports, in conjunction with the "key" moments and
execution phases, allowed to eliminate the geotechnical and
stratigraphic uncertainties.
8.4 Parametric analysis based on groundwater levels

In the subsequent phases of work, the only variability is therefore
considered to be the one linked to the groundwater levels.
Figure 12 Aquifer levels A0 and A1 during excavation works
This data keeps margins of uncertainty as it can be described as a
range of values.
In the following, some pictures of excavation and pouring of r.c.
Such uncertainty requires to consider as always possible, albeit
foundation slab are presented.
unlikely at comparable time scale of the individual work phases, a
sudden and unexpected rise in the groundwater levels up to the
maximum design limit.
If it was planned to proceed according to lower groundwater
conditions than the maximum, this implied lower intensity support
works than those required for the maximum configuration, with a
reduction in acceptable values of the groundwater excursion.
In such hypothesis, in order to cope with possible increases in
the water levels up to the maximum values of the project range, it
was necessary to define a system of extraordinary measures of
intervention and mitigation.
Many numerical analyses were carried out simulating possible
rise in groundwater level for each possible excavation phase and for
each possible configuration of support works, in order to define a set
of mitigation actions to be taken when unexpected and sudden rise
in the ground level occurred, with works in a configuration lower
than the maximum floor level.
By proceeding this way, it was always possible to restore the
starting safety level and allow for further work.
8.5 Implementation of mitigation actions and emergency

measures Figure 13A Excavation and execution of r.c. foundation slab
In principle, to cope with unexpected increases in the groundwater

levels and in presence of support works performed in a lower
configuration than the one required by the actual groundwater level
(for example, it was proceeding on a configuration lower than the
maximum and groundwater levels suddenly increased to the
maximum values) it was necessary to either:
1. proceed immediately to the backfill of the excavated area adjacent
to the diaphragm
2. or continue construction with more conservative phases and
support works than those in agreement between groundwater level
and existing design configuration; this was done by:
2.i. reducing the height of excavation needed to install the single
array of anchor strands;
2.ii. raising the level from which (normally 43.50 m a.s.l.) the
berm /finite trench excavation could start, for the execution
of the bottom r.c. slab;
2.iii. reducing dimensions of the last excavation trench for the
execution of the bottom slab.
As it can be seen from the figure 12 below, the “most
unfavourable” levels for the aquifer A1 have been nearly reached a
couple of times during and near the end of the excavation works. Figure 13B Excavation and execution of r.c. foundation slab
431
This situation has shown its own peculiarities, which may affect
the definition of mean values, most probable values and the limits of
acceptability of project hypotheses as well as the alternative
remedial solutions.
The Observational Method has also proved to be a mostly
suitable design criterion when dealing with situations in which the
construction methods, operating sequences and the statics of the
works are particularly related and the quantities are to be
rationalized.
The potential optimization of the quantities that the Method
offers, must necessarily be supported by detailed studies and
verifications, which ultimately lead to greater reliability of the
project, both in terms of cost and execution times.
In this sense, the Observational Method may also be framed into
Projects according to "cost-plus fee" or “fixed-rate” contractual
schemes, and also according to "lump sum" logic as the present case,
for the main benefit of the Client, especially where the limits of
variability of the design parameters can be clearly established.
Figure 13c Last excavation phase in east side diaphgram wall
9. CONCLUSIONS
The design improvement proposal based on the Observational
Method as developed by the Contractor together with its design
consultant Soil srl and approved by the Client, allowed to finish the
excavation works on time and with no significant interruptions.
Under the more specific point of view of the Design process, it
allowed for the main following benefits:
1. to include in the design possible "most unfavorable" groundwater
conditions and absorb the uncertainties about the geotechnical
aspects, extending the calculation hypothesis for provisional
works, which have direct effects on the quality of the support
structures, specially the maximum water table levels, in order to
ensure a greater continuity of the site activities (excavations and
casting of the foundation slab);
2. to rationalize the sizing and the dimensioning of the provisional
works (strands number, diameter and barrel length and jet Figure 14 Works near completion in the north area
grouting extension) as a partial compensation of the execution
burden related to the extension of the calculation hypothesis; 10. REFERENCES
3. to offer greater guarantees and reliability in general terms of risk
Ciria - Report 185 (1999) - Observational Method
management related to the execution of the project (timing of the
Eurocode 7 (BS EN 1997-1:2004) “Geotechnical design - Part 1:
work and execution costs) both for the Client and for the
General rules”
Company.
Decreto legislativo 12 aprile 2006, n. 163, Codice dei contratti
Therefore, in this case, the Observational Method is a valid
pubblici
design support for the execution of complex geotechnical works and
NTC2008 - Norme tecniche per le costruzioni - D.M. 14 gennaio
projects.
2008
Its application has been successfully extended from the case of
Peck, R. B. (1969). Advantages and limitations of the observational
underground works, more traditional, to the less usual case of large
method in applied soil mechanics. Geotechnique, 19: 171-
“open-sky” excavation works.
187.
432
Reliability Assessment on Deep Braced Excavations Adjacent to High Slopes in

Mountain Cities
RH. Zhang1 WG.zhang12 ZJ.Hou1 and W.Wang1
School of Civil Engineering, Chongqing University, Chongqing, China
1
2
Key Laboratory of New Technology for Construction of Cities in Mountain Area, Chongqing University, Chongqing, China.
E-mail: cheungwg@126.com
ABSTRACT: Due to rapid urbanization, the land available for construction becomes more and more scarce. Within a built-up environment,
the construction safety of a deep excavation becomes more crucial with the ever-increasing building density. For deep excavations in
mountain cities, the areas of the foundation pit to be excavated are generally the passive soil pressure zones for the upper existing slope.
Construction disturbance, weakening of the passive area, as well as the formation of even higher slope through superposition of foundation
pit to the upper slope, will result in more deformation and even failure of the slope. This study numerically investigates the influences of
excavation geometries, the system stiffness and the distance between the excavation and slope and develops simplified ultimate and
serviceability limit state models with regard to the overall factor of safety and the maximum lateral wall deflection of the supporting system,
respectively. Considering the uncertainties of the design parameters, a probabilistic framework combining the estimation models with First-
Order Reliability Method (FORM) is proposed to determine the probability that a threshold factor of safety or the maximum wall deflection
is exceeded. The study presents preliminary guidelines for reliability assessment of ultimate and serviceability limit state designs for deep
braced excavations adjacent to high slopes in mountain cities.
Keywords: braced excavation, upper slope, wall deflection, factor of safety, reliability assessment.
1. INTRODUCTION
2. FINITE ELEMENT ANALYSIS
With promotion of the Belt and Road Initiatives in China, the fast
development of urbanization progress has brought the great demand 2.1 Numerical modeling
of commodity housing and public transport facilities. Thus it is
inevitable that more deep excavations to be constructed for The PLAXIS2D software was utilized for the numerical simulations.
residence, commercial buildings as well as the sky scrapers designed The Mohr-Coulomb constitutive model was selected for the soil. A
aside the existing slopes in densely populated mountainous cities typical cross-section of the excavation and slope system, the
such as Chongqing and Guiyang. It is well known that for deep geometries as well as the properties of the soil and the supporting
excavations in mountain cities, the areas of the foundation pit to be elements are shown in Figure 1.
excavated are generally also the passive soil pressure zones for the
upper existing slopes. Consequently, the construction disturbances,
weakening of the passive area, as well as the formation of the even
higher slope through superposition of foundation pit to the upper
slope, will result in more deformation and even collapse of the
slope. Nevertheless, there are few investigations of the interaction
between braced excavation and the adjacent slope and the influence
of such interaction on the overall stability. Li et al. (2011)
investigated the stability of supporting system and the safety of deep
braced excavation adjacent to slope, through analyzing the influence
of excavation of Shangshuijing station Shenzhen Metro Line 5 on
side slope using FLAC3D. Wang et al. (2011) examined the
deformation characteristics and behaviors of retaining structures for Figure 1 Cross-sectional soil and wall profile
a complex geotechnical system comprising of a high building slope
and a nearby deep excavation, based on field instrumentations. The analyses considered a plane strain excavation supported by
Varzaghani and Ghanbari (2014) presented a new analytical model a retaining wall system near an unreinforced slope. The soil was
to determine the seismic displacements of the shallow foundations modeled by 15-noded triangular elements. The structural elements
adjacent to slopes. However, there is still a lack of systematic were assumed to be linear elastic with the wall represented by 5-
investigation of the key influential factors and the effects on the noded beam elements and 3-noded bar elements were used for the 6
ultimate limit state and serviceability limit state of the excavation levels of struts located at depths of 1 m, 4 m, 7 m, 10 m, 13 m and
and slope system. 16 m below the original ground surface. The nodes along the side
In this study, the global factor of safety FS obtained via the boundaries of the mesh were constrained from displacing
shear strength reduction (SSR) technique (also called c/φ reduction horizontally while the nodes along the bottom boundary were
method) is used as the criterion for the ultimate limit state and the constrained from moving horizontally and vertically. The left and
calculated maximum lateral wall deflection is adopted as the right vertical boundary extend far from the excavation to minimize
serviceability limit state criterion. It then numerically investigates the effects of the boundary restraints. The ranges of the design
the influences of the excavation geometries, the supporting system parameters varied are shown in Table 1.
stiffness, the distance between the new excavation and the existing The strut stiffness per meter EA is assumed as a constant at
slope on excavation responses including the global FS and the wall 3.0×105kN/m since the influence of strut stiffness on wall deflection
deflection using PLAXIS software. Estimation models with regard is not very significant when the strut is stiff (Poh and Wong 1997).
to both the ultimate and serviceability limit states are developed. A total of 162 hypothetical cases were analysed.
Probabilistic framework combining the proposed estimation models The construction sequence comprised the following steps:
with the First-Order Reliability Method (FORM) is adopted to  the wall is installed (“wished into place”) without any
determine the probability that a threshold factor of safety or the disturbance in the surrounding soil;
maximum wall deflection is exceeded. This proposed approach
enables a cost-effective analysis to be conducted for a rational
design of excavation system adjacent to an existing high slope.
433
 the soil is excavated uniformly 1 m below each target strut

level prior to adding the strut support with struts at 3 m
vertical spacing until the final depth He is reached.
 Each phase of strut installation is followed by a subsequent
phase of global safety factor calculations by SSR method.
Details are listed in Table 2.
Table 1 Parameters considered and the ranges
Parameters Ranges
*System stiffness S 3.794, 4.605, 5.187
Excavation width B (m) 20, 30, 40
a) He=0 m FS=2.341
Excavation depth He (m) 14, 17, 20
Wall thickness d (m) 0.6,0.9,1.2
distance between braced excavation and 5,10,15,20,30,40
side slope B1 (m)
Penetration ratio D/He 0.50, 0.76, 1.14
* Influence of wall stiffness was studied by varying wall thickness d while
keeping the Young’s modulus of the wall constant (E=1.20106 kN/m2).
The corresponding natural logarithm of the system stiffness ln(EI/γwh4avg),
denoted by S for the wall thickness of 0.6, 0.9 and 1.2 m with average
vertical strut spacing havg=3 m
Table 2 Construction procedures

b) He=14 m FS=2.214
Phases Construction details
Initial Generate the initial effective stress, pore pressure
Phase and state parameters.
Phase 1 Calculated the global safety factor by SSR method
Phase 2 Install the diaphragm wall
Phase 3 Reset displacement to zero, excavate to 2 m below
the ground surface inside the excavation, install
strut at 1 m below the ground surface
Phase 4 Excavate to 5 m below the ground surface
Phase 5 Install strut at 4 m below the ground surface
Phase 7 Excavate to 8 m below the ground surface c) He=17 m FS=2.095
Phase 19 Excavate to 20 m below the ground surface d) He=20m FS=1.705
Phase 20 Calculated the global safety factor by SSR method Figure 2 Contour of slip surface and FS for different excavation
depths He for B=30 m, B1=5 m, S=4.605
2.2 Numerical results
Figure 3 presents some typical plots of the FS decrease for
The numerical results include the factor of safety FS and the different B1 for He=20 m, S=4.605. Generally, the FS decrease
maximum lateral wall deflection δhm. FS is solved through SSR becomes less significant as the distance between the excavation and
technique, in which the shear strengths are systematically reduced the existing slope B1 increases and converges to 0, indicating that the
until failure occurs. This procedure was proposed by Zienkiewicz et further away the excavation is from the existing slope, the much
al. (1975), and improved by Brinkgreve and Bakker (1991). It has safer the whole system is. In addition, for the model dimensions
been verified by Lian et al. (2001) that the SSR FE method can be considered in this study, the braced excavation has no impact on
widely applied in the engineering practice since this method takes stability of the adjacent slope when the separation B1 is greater than
advantages over the conventional limit equilibrium method. Cheng 40 m. As for the lateral deflections of the retaining wall on the slope
et al. (2007) and Dawson et al. (1999) proved that the SSR side, Figure 4 shows the maximum lateral wall deflection δhm for
technique perform well in many slope cases. different distances B1 for case of He=20 m, S=4.605. It is clear that
Figure 2 plots the variation of slip surface contours as δhm has a tendency to grow with excavation width B while it
excavation proceed, for case of B=30 m, B1=5 m, S=4.605. The FS decreases with the increase of the separation B1.
values for excavation depths He of 0, 14, 17, 21 m are also
calculated, respectively. It can be observed that as excavation
proceeds, FS values decrease. The smallest FS is about 1.705 with a
decrease of 0.636 from the original 2.341. In addition, it is also clear
that a larger slip surface occurred when the excavation depth He
becomes greater.
434
Figure 3 Decrease of factor of safety FS on different B1 for He=20

m, S=4.605
Figure 5 Comparison between FS_FEM and FS_EST
3.2 Serviceability limit state model

Similarly, a Logarithmic Regression (LR) model for predicting the
maximum lateral wall deflection δhm is developed and shown in Eq.
(2), with fairly high coefficient of determination R2=0.946, as below
δhm=0.1133B0. B1-0.223(He)2.1247(D/He)0.0568S-0.4448 (2)
1086
Figure 6 plots the estimated maximum lateral wall deflections

hm_EST values against the calculated results hm_FEM. Also shown are
the 100% agreement line and the 20% error lines, indicating that Eq.
(2) is fairly accurate in predicting the maximum wall deflections
induced by deep braced excavations adjacent to high slopes.
Figure 4 Max. lateral wall deflection δhm for different B1 of case

He=20 m, S=4.605
3 ESTIMATION MODELS FOR THE LIMIT STATE

FUNCTIONS
For the performance in deep braced excavations, especially for the
excavations adjacent to high slopes, both the ultimate limit state
(ULS) and the serviceability limit state (SLS) should be satisfied. In
the following sub sections, the limit state functions for ULS and
SLS are developed respectively, based on the numerical results in
the previous section.
3.1 Ultimate limit state model
Based on the calculated FS results, a Polynomial Regression (PR)
model has been developed for estimating the factor of safety FS as a
function of four input parameters: B, B1, He and S in Eq. (1), with a
coefficient of determination R2 of 0.881, as below
FS=7.35×10-2B-1.57×10-1B1+3.51×10-2He+1.02S-2.5×10-6B2-
Figure 6 Comparison between δhm_FEM and δhm_EST
2.29×10-4B12-8.07×10-4He2-2.22×10-2S2+2.4×10-4BB1-4.08×10-
3BH +1.65×10-4BS+8.01×10-3H B -1.56×10-3SB -2.8×10-2H S-
e e 1 1 e
1.84×10-2B(D/He)+4.83×10-2B1(D/He)+9.4×10-2He(D/He) -3.55×10-1 4. PROBABILISTIC ASSESSMENT OF THE LIMIT-
S(D/He) (1) STATES
In many civil engineering applications, the assessment of safety is
Figure 5 plots the estimated factor of safety FS_FEM values
made by firstly establishing a relationship between the load S of the
against the calculated FS_EST results. Also shown are the 100%
system and the resistance R. The boundary separating the safe and
agreement line and the 10% error lines, indicating that Eq. (1) is
‘failure’ domains is the limit state surface (boundary) defined by
fairly accurate in predicting the global factor of safety for deep
G(x)=R-S=0, where x is vector of the random variables.
braced excavations adjacent to high slopes since the majority of data
Mathematically, R > S or G(x) > 0 would denote a ‘safe’ domain.
points are within the error lines.
An unsatisfactory or ‘failure’ domain occurs when R < S or G(x) < 0.
Calculation of Pf involves the determination of the joint probability
distribution of R and S and the integration of the Probability Density
435
Function (PDF) over the failure domain. Considering that the PDFs 2007). For nonnormals, the nonnormal distributions are replaced by
of the random variables are not known in most geotechnical an equivalent normal ellipsoid, centred at the equivalent normal
applications and the integration is computationally demanding when mean. Cells D3:E5 are parameters which are set corresponding to
multi-variables are involved, an approximate method, known as the the normal distribution in this study. The correlation matrix R in
First-Order Reliability Method (FORM) (Hasofer and Lind, 1974), cells G3:I5 are used to define the correlations between B, He and S.
is commonly used to assess the probability failure Pf. (Low 1996) The ni vector in cells J3:J5 contains equations for (xi ‒ uiN) / iN.
has shown that Microsoft EXCEL spreadsheet can be used to The design point (x* values) was obtained by using the
perform the minimization and determine reliability index. spreadsheet’s built-in optimization routine SOLVER to minimize
The reliability index  and the probability of failure Pf for both the cell, by changing the x* values, under the constraint that the
the ultimate and the serviceability limit states can be performed performance function G(x*) = 0. Prior to invoking the SOLVER
using FORM based on the built PR and LR models. The ULS model search algorithm, the x* values were set equal to the mean values
Eq. (1) is incorporated into an EXCEL spreadsheet environment (30, 17, 4.5) of the original random variables. Iterative numerical
based on the approach by (Low and Tang 2007), from which the derivatives and directional search for the design point x* were
reliability index can be determined. Figure 7 shows a sample automatically carried out in the spreadsheet environment.
spreadsheet for computing the factor of safety FS where the Probabilistic assessment of SLS in Figure 8 is almost the same
statistics of the design parameters are the same as those used in the as Figure 7 except the G(x) formulations. For the detailed
previous section. The spreadsheet cells B3:B5 allows the selection procedures in performing the FORM spreadsheet framework to
of various distribution types for the input variables, including derive  and the corresponding Pf, the paper published by Zhang
normal, lognormal, triangular etc. as explained in (Low and Tang and Goh (2012 can be referred to.
Figure 7 Calculation on β and Pf for ultimate limit state using FORM spreadsheet
Figure 8 Calculation on β and Pf for serviceability limit state using FORM spreadsheet
4.1 Probabilistic assessment of the ultimate limit state Figure 9 plots the influence of the various design parameters on the
 and Pf of ULS. It is clear that both the coefficient of variation of
For either the braced excavation or the slope, there are design
the system stiffness COVS and the critical factor of safety FS_cr
guidance with regard to the choice of the critical factor of safety.
significantly influence the  and Pf. In addition, the influence of
However, for the excavation and slope system, there are no
guidelines for the determination of such critical safety factor values. COVS on  and Pf is also as significant as that for FS_cr. The plots
Thus the influence of the critical factor of safety FS_cr on  and Pf in Figure 10 indicate that the influence of either B1 or COVS on 
of ULS is examined in this study. and Pf is also obvious when different excavation widths B of 20, 30,
40 m are considered.
436
(a) (b)
Figure 9 Influence of COVS and FS_cr on (a)  and (b) Pf for B1=5m, B=30m, He=17m, S=4.5
B=20m B=30m B=40m
B=20m B=30m B=40m

Figure 10 Influence of COVS and B1 on  and Pf for B=20, 30, 40 m, He=17m, S=4.5, FS_cr=2.0
B1=20 m B1=30 m B1=40 m

Figure 11 Influence of COVS and δhm_cr on  for B=30 m, He=17m, S=4.5, B1=20, 20, 40 m
437
Figure 10 also compares the influence of both the COVS and B1 Regression models for the ultimate and serviceability limit states
on  and Pf for B=20, 30, 40 m, respectively, for He=17m, S=4.5, are developed respectively. Through the use of the automated
and chosen critical factor of safety FS_cr=2.0. It is obvious that  spreadsheet search algorithm to determine the design point, to meet
becomes greater with increase of the excavation width B1 while the different target performance levels, the critical FS or the
decreases with the increase of excavation width B. Meanwhile, Pf threshold max. lateral wall deflection can be obtained. The
decreases as the excavation becomes further away from the slope. A influences of the key parameters, as well as the design uncertainties
greater excavation width B generally results in a larger Pf. Generally on the reliability index and the probability failure are examined. The
Pf converges to 0 when the separation is sufficient. However, procedures outlined in this paper can be used to obtain a rational
different B causes different convergence speeds. design of braced excavation adjacent to high slope and a cost-
effective analysis.
4.2 Probabilistic assessment of the serviceability limit state
There are also discussions as for the choice of the threshold lateral 6. REFERENCES
wall deflections for serviceability considerations. Figure 11 plots the
influence of COVS and the critical max. wall deflection δhm_cr on  Brinkgreve, R.B.J., Bakker, H.L.(1991). "Non-linear finite element
and Pf for B=30 m, He=17m, S=4.5 and B1=20, 20, 40 m, analysis of safety factors. " In: Proc. 7th Int. Conf. on
respectively, indicating that both COVS and δhm_cr significantly Computer Methods and Advances in Geomechanics, vol. 23,
influence the  and Pf. However, the influence of COVS on  and Pf pp. 1117–1122.
is not as significant as that for δhm_cr, especially when COVS is Cheng, Y. M., et al. (2007). "Two-dimensional slope stability
greater than 0.20.  has a tendency to grow with the critical analysis by limit equilibrium and strength reduction
maximum lateral wall deflections δhm_cr since the probability that a methods." Computers and Geotechnics 34(3): pp 137-150.
greater threshold is exceeded is much lower.  decreases with the Dawson, E. M., et al. (1999). "Slope stability analysis by strength
increase of COVS. In addition, it can be observed that the influence reduction." Geotechnique 49(6): pp 835-840.
Hasofer, A.M., Lind, N.(1974). "An exact and invariant first-order
of B1 on  is also significant since  increases substantially with the
reliability format." J. Eng. Mech. ASCE 100 (1), pp 111–121.
separation B1.
Li, Y. H., et al. (2011). "Stability and Safety Analysis of Braced
Figure 12 shows the influence of COVS on  for He=17m, S=4.5,
Excavation for Subway Station during Construction under the
δhm_cr=23 mm, B=20,30, 40 m and B1=10,15m respectively. It is
Condition of Side Slope." Applied Mechanics and Materials
clear that  decreases as the variation of the system stiffness 99-100: pp 1166-1170.
becomes greater. It is logical that  increases when the excavation is Lian, Z.-y., et al. (2001). "Stability analysis of excavation by
becoming further away from the slope. strength reduction FEM." Chinese Journal of Geotechnical
Engineering 23(4):p 5.
Low,B.K.(1996)."Practical probabilistic approach using spreadsheet.
" In: Shackelford, C.D., Nelson, P.P., Roth, M.J.S. (Eds.),
Uncertainty in the Geologic Environment, GSP 58. ASCE,
Reston, pp. 1284–1302.
Low, B.K. and Tang, W.H. (2004), "Reliability analysis using
object-oriented constrained optimization."Struct. Saf., 26(1),
pp 69-89.
Low, B.K. and Tang, W.H. (2007), "Efficient spreadsheet algorithm
for first-order reliability method." J.Eng. Mech. ASCE,
133(12), pp 1378-1387.
Poh TY, Wong IH, Chandrasekaran B, (1997) "Performance of two
propped diaphragm walls in stiff residual soils. "Journal of
Performance of Constructed Facilities, 11(4): pp 190–199.
Varzaghani, M. I., & Ghanbari, A. (2014). "A new analytical model
to determine dynamic displacement of foundations adjacent
to slope. " Geomechanics and Engineering. pp 561-575.
Figure 12 Influence of COVS on  for He=17m, S=4.5, δhm_cr=23 mm, Wang, Q., et al. (2011). "analysis of slip-risk and dynamic
B=20,30, 40 m and B1=10,15m monitoring of a high building slope fringed a deep foundation
pit." journal of safety and environment .11(2):p 6.(in Chinese)
5. SUMMARY AND CONCLUSIONS Zhang, W. and A. T. C. Goh (2012). "Reliability assessment on
ultimate and serviceability limit states and determination of
This paper presents numerical investigations about influence of critical factor of safety for underground rock caverns."
braced excavation on the existing slope, from perspectives of the Tunnelling and Underground Space Technology 32: pp 221-
global factor of safety and the maximum lateral wall deflections. It 230.
also proposed probabilistic framework for quantitative assessment Zienkiewicz, O.C., et al.(1975). "Associated and non-associated
of both the ultimate and the serviceability limit states in view of visco-plasticity in soil mechanics. "Geotechnique 25 (4), pp
some design and construction uncertainties. 671–689.
438
Numerical Investigation of Wall Deflections Induced by Braced Excavations in Sands

Zhongjie Hou2, Wengang Zhang*12, Runhong Zhang2, Wei Wang2
1
Key Laboratory of New Technology for Construction of Cities in Mountain Area, Chongqing University, Ministry of
Education, Chongqing 400045, China
2
School of Civil Engineering, Chongqing University, Chongqing 400045, China
E-mail: cheungwg@126.com
ABSTRACT: Most previous studies focused on the performance of braced excavations in clays, where limited publications involved braced
excavations in sands. In this study, to better understand the performance of braced excavations in sand, a series of two-dimensional (2D) and
three-dimensional (3D) finite element analyses using the Hardening Soil (HS) model of PLAXIS software were performed to investigate the
influences of soil properties, wall stiffness and excavation geometries, on the maximum wall deflection induced by braced-excavation in
sand.. Results show that the maximum wall deflections calculated by 3D analysis are greater than those from 2D. Based on the numerical
results, regression models were developed for estimating the maximum wall deflections induced by braced excavation in dense sand and
medium dense sand, respectively.
Keywords: Braced excavation, wall deflection, sand, finite element analysis, Hardening Soil (HS) model, regression models
1. INTRODUCTION excavations in sand based on extensive plane strain and 3D

numerical analyses with HS models. The influences of soil
Deep excavations are increasingly carried out in urban areas with properties, wall stiffness and excavation geometries on the
development of underground space. However, the excavation maximum wall deflection induced by braced excavation in sand
process inevitably alters the stress states underground and may were investigated. Based on the numerical results, Logarithmic
introduces significant wall deformations and ground movements, Regression (LR) models and Polynomial Regression (PR) models
which would cause potential damage to the adjacent properties. To have been developed for estimating the maximum wall deflections
reduce the excavation-induced deformations, appropriate retaining induced by braced excavation in dense sand and medium dense sand,
walls and supporting systems should be designed through adequate respectively.
construction methods. Based on the Nicoll highway collapse in
Singapore and the Xianghu foundation pit accident in Hangzhou,
2. NUMERICAL MODELLING
China, an important design issue is to ensure the reliability of the
structural systems, a common design criterion is to limit the The FE software PLAXIS 2D (2017) and PLAXIS 3D (2017) were
maximum wall deflection to a fraction of the excavation depth He, used to perform the excavation simulations. Figure 1 shows a typical
typically in the range of 0.5~1.5%. However, most previous studies cross-section and plan view for the cases considered. The embedded
(Wong and Broms 1989, Kung et al. 2007, Wang et al. 2008, Xuan retaining wall together with a five-level strut system for He =17 m is
2009, Hwang et al. 2012, Moh and Song 2013, Hsieh and Ou 2016, also plotted in Figure 1. Relevant design parameters shown in the
Zhang et al. 2015, Finno et al.2016, Goh et al. 2017) have figure include: excavation lengths L of 36, 60 and 84 m, excavation
investigated the behavior of wall affected by deep excavations in widths B fixed at 30 m, excavation depths He of 17m, wall
soft to stiff clays. There are only limited studies involving braced penetration depths into underlying stiff clay D of 5 m, thickness of
excavation in sands. sand layer T1 and thickness of stiff clay layer T2 of 25 and 30 m,
A number of researchers have adopted finite element method respectively, horizontal strut spacing SH of 4 and 5 m, and vertical
(FEM) to study the behavior of retaining systems in deep strut spacing SV of 3 m.
excavations. Nakai et al. (1999) conducted 2-D model test and the
results were compared with FE analysis. It was found that the
computed results describe well with the model test results. They also
indicated that the stiffness of the wall, the wall friction, and the strut
significantly influence the performance of the wall. Based on the
back analysis using the MIT S1 soil model, Nikolinakou et al. (2011)
analyzed excavations in Berlin sand and examined the structural
systems and soil properties of the excavation. Khoiri and Ou (2013),
Han et al. (2017) and Hsiung (2016) compared the FEM predicted
results with measured data for some excavation case histories and
validated the reliability of numerical analysis with HS model for
predicting the wall deformations and ground settlements in sandy
soils. Hsiung and Dao (2014) has made a comparison of the
performance of three constitutive soil models, i.e. Morh-Coulomb
model (MC model), Hardening soil model (HS model) and
Hardening soil model with small-strain considerations (HSS model), Figure 1 Cross-section and plan view of the numerical model
in predicting excavation-induced wall deflection in sands. Results
indicated that the HSS model and HS model perform better 2.1 Numerical models
predictions than those of the MC model. Sabzi and Fakher (2015) For 2D analysis, only half of the excavation model was developed
studied the performance of buildings adjacent to excavation due to symmetrical conditions for both the excavation sequence and
supported by inclined struts in sand soils, it was found that soil geometry. A fine mesh size was adopted for 2D analysis to improve
strength parameters (c and ) significantly affect the horizontal wall the accuracy of calculations. For 3D analysis, only a quarter mesh
displacement and the soil stiffness affects the ground surface and a medium mesh size were used to reach a balance between
settlement considerably. accuracy and the processing time. Fig. 2 shows a typical 3D mesh
As discussed previously, the performance of braced excavations plot, comprising of 93713 nodes and 63343 15-noded wedge
in sand is still a hot topic due to limited research. The main purpose elements.
of this study is to study the behaviour of wall affected by deep
439
Table 1 Wall properties for 2D and 3D analyses

Parameters Wall types
flexible medium
Plane strain (2D) FE parameters
 0.06 0.1 0.2 1.0
Wall stiffness EI (kNm2/m) 3.224104 5.407104 1.081105 5.407105
Compressive stiffness EA (kN/m) 2205954 3.767106 7.532106 3.767107
Poisson’s ratio, ν 0.15 0.15 0.15 0.15
Three-dimensional (3D) FE parameters
Young’s Modulus (kPa) E1 5.252106 8.754106 1.751107 8.754107
E2 2.62610 5 4.37710 5 8.754105 4.377106
G12 2.62710 5 4.37810 5 8.757105 4.379106
Shear Modulus (kPa) G13 8.754105 1.459106 2.918 106 1.459107
G23 2.626105 4.377105 8.754105 4.377106
Poisson’s ratio ν 0 0 0 0
Table 2 Properties of shoring system

Parameters Struts Walers
Young’s Modulus E (kN/m2) 2.1108 2.1108
Unit weight γ (kN/m3) 78.5 78.5
Cross section area A (m2) 0.007367 0.008682
Moment of inertia (m4) I2 5.07310-5 3.66810-4
I3 5.07310-5 1.04510-4
The boundary conditions for the cases considered were: roller

fixities at side boundaries to allow the vertical displacements;
pinned at the bottom boundary to restrain any movements; the top
boundary was free to move in all directions. For both 2D and 3D
models, the lateral boundaries in the side directions were defined as
Figure 2 Typical quarter 3D mesh of excavation 90 m away from the centre of the excavation to minimize the
boundary effect of the mesh. The original ground water level inside
In this study, the embedded retaining wall is simulated using 5- the excavation was assumed at a depth of 5 m below the ground
noded elastic plate elements for 2D analysis, while for 3D analysis, surface, which was progressively lowered with the excavation of the
the wall is simulated using 8-noded quadrilateral plate elements with soil during each phase. Identical construction procedures of the
six degrees of freedom per node. Considering that the relatively simulation were applied as described in Table 3.
flexible wall was generally designed and constructed for braced
excavations in sands in engineering practice, two types of wall with Table 3 Typical construction sequence for 2D analysis
four different stiffness values were considered for each soil type, as Phases Construction Details
listed in Table 1. A stiffness coefficient  was utilized to represent Phase 1 Install the excavation wall
walls with varying rigidities (Zapata-Medina and Bryson 2012). The Phase 2 Excavate to 2 m below ground surface
baseline bending stiffness EI used to develop the different models
Phase 3 Install strut system at 1 m below ground surface
was 540,675 kNm2/m, which refers to a wall of medium stiffness
Phase 4 Excavation to 5m below ground surface
based on databases of Long (2001) and Moormann (2012).
Therefore, =1.0 represents the cases with medium wall stiffness in Phase 5 Install strut system at 4 m below ground surface
this study. For flexible walls in sands, the baseline bending stiffness Phase 6 Dewatering and excavation to 8 m below ground surface
EI of 540,675 kNm2/m was multiplied by smaller  values of 0.06, Phase 7 Install strut system at 7 m below ground surface
0.1, and 0.2. Meanwhile, based on method adopted by Finno et al. Phase 8 Dewatering and excavation to 11 m below ground
(2016), the wall thickness of 0.42 m was set to be constant so that Phase 9 surfacestrut system at 10 m below ground surface
Install
the moment of inertia I and area A of the plate were kept constant, Phase 10 Dewatering and excavation to 14 m below ground
and only elastic modulus E was varied. Besides, the soil-structure Phase 11 surfacestrut system at 13 m below ground surface
Install
interactions are simulated through interfaces on both sides of the Phase 12 Dewatering and excavation to 17 m below ground
wall, which allow for the specification of a reduced friction between surface
wall and the soil. 2.2 Constitutive model and model parameters
The shoring system of the braced excavation comprised of struts
and walers. The struts were simulated via fixed end anchors in 2D The hardening-soil (HS) model was used to simulate the constitutive
analysis. For 3D analysis, beam elements were used to model the behavior of the two types of sands. This model involves frictional
struts and walers, which have six degree of freedom per node. For hardening characteristics to model plastic shear strain when
the braced excavations considered in this study, the struts were subjected to primary deviatoric loading, and cap hardening to model
designed horizontally at a regular spacing of 4 m in two directions to plastic volumetric strain in primary compression. Failure is still
form a strong frame. The walers transfer the forces from the defined by the M-C failure criteria. It should be noted that the
retaining wall to the struts and also stiffen and align the wall. The hardening soil parameters of sands considered in this parametric
properties of shoring system are tabulated in Table 2. study were based on the data extracted from Brinkgreve et al.
(2010). The underlying stiff clay with average undrained shear
strength cu = 125 kPa was based on the Gault clay found in
Cambridge (Ng 1992). The HS parameters for different soils are
440
Table 4 Hardening Soil parameters used for Finite-Element modelling
Hardening soil parameter Medium dense sand Dense sand Stiff clay
Parameters Unit (drained) (drained) (undrained)
γunsat kN/m3 drained
17 18.2 20
γsat kN/m3 19.8 20.3 20
kx=ky=kz m/day 110-8 110-8 110-8
E50ref kN/m2 30000 48000 14847
Eoedref kN/m2 30000 48000 14847
Eurref kN/m2 90000 144000 44540
c kN/m2 0 0 0.05
 ° 34.3 38 33
Ψ ° 4.3 8 0
νur [-] 0.3 0.35 0.2
pref kN/m2 100 100 100
m [-] 0.544 0.45 1
K0nc [-] 0.436 0.384 0.455
Rf [-] 0.938 0.9 0.96
Rinter [-] 0.8 0.8 1
A series of FE simulations using Hardening Soil (HS) model different excavation stages from 2D analyses in medium dense sand
were carried out to investigate the influences of soil properties, for wall stiffness α = 0.06 and 1.0. For the 3D rectangular braced
excavation geometries, wall stiffness  on the wall deflections excavation, only the profiles of the horizontal wall deflection at the
induced by excavation. For brevity, the numerical results of the centre of excavation are presented because the maximum lateral
study are presented with the main findings, as described in the next displacement occurs at this particular centreline location for
sections. symmetry. For brevity, only wall deflections caused by braced
excavations in medium dense sand with L/B = 2.8 for α = 0.06 and
3. FINITE ELEMENT ANALYSES 1.0 are plotted in Figure 4.
Figure 3 shows the typical wall deflection profiles corresponding to
(a) α = 0.06 (b) α = 1.0

Figure 3 Wall deflection profiles for different excavation stages in medium dense sand from 2D analyses (α = 0.06,1.0)
(a) α = 0.06 (b) α = 1.0

Figure 4 Wall deflection profiles for different excavation stages in medium dense sand from 3D analyses (α = 0.06,1.0)
441
Table 5 Typical maximum wall deflection for He = 17 m and =0.06
Excavation Maximum wall deflections (mm)

Soil stage  = 0.06  = 1
type (m) 2D 3D 2D 3D
L/B = 1.2 L/B = 2.0 L/B =2.8 L/B = 1.2 L/B = 2.0 L/B =2.8
2 1.38 1.22 1.61 1.90 1.38 1.14 1.79 1.93
Medium 5 4.36 5.05 4.89 5.31 4.36 3.30 L 5.02 5.46
dense 8 18.67 19.17 18.36 18.51 10.1 9.23 / 11.53 10.24
sand 11 37.91 45.89 43.15 41.71 20.5 20.75 B 20.25 20.58
14 65.49 93.82 83.2 77.25 35.1 37.93 33.76 37.54
17 106.2 192.4 149.7 136.7 55.5 64.73 = 54.34 63.44
2 0.78 1.04 1.53 1.81 0.8 1.26 1.55 1.84
5 3.61 3.08 4.35 4.93 3.81 3.39 2 4.42 5.15
Dense 8 7.60 9.17 8.83 9.13 7.81 6.40 . 8.28 9.52
sand 11 18.88 26.17 24.22 21.17 12.5 11.24 0 12.93 14.59
14 35.05 55.61 48.55 42.85 18.9 21.40 21.53 21.08
17 60.56 112.8 88.58 78.48 33.8 38.41 L 40.09 39.10
/
B
The distribution patterns of wall deflection profiles induced by For brevity, only some of the main results from 2D and 3D
excavations obtained from 2D simulations and 3D simulations are analyses are presented in Table 5. =
It can be found that the maximum
almost similar in shape, following a general trend that increasing wall deflection has a tendency to2grow with the soil strength and
wall stiffness leads to smaller maximum wall deflection in medium decrease as the wall stiffness increases.
. Results in Table 5 also
dense sand. The wall behaved in cantilever-mode first, and then indicated that the wall displacements
8 increase continuously as
changed to prop-mode after the struts were installed. The strut excavation proceeds. Generally, small horizontal displacement
installation generally restrains the displacement of wall above the appears in the first two excavation stages, and becomes considerable
level of installed strut, especially for flexible and medium walls, so in the subsequent stages. Maximum difference of wall deflection
that the wall deflection profiles at various excavation stages almost occurs in the last excavation stage, for medium dense sand, the
coincide with each other above the installed struts. This agrees with maximum difference of wall deflections range from 93.82 mm to
the previous research by Hsiung et al. (2016). Generally, for the 192.4 mm in the final excavation stage for L/B=1.2 with wall
flexible walls, the wall deflection profile has a bulging shape with stiffness  of 0.06.
the maximum wall deflection between the excavation level and the
toe of the wall.
(a) α = 0.06 (b) α = 0.1
(c) α = 0.2 (d) α = 1.0

Figure 5 Maximum wall deflection for various L/B ratios in medium dense sand (= 0.06, 0.1, 0.2 and 1.0)
442
The values of the coefficients of Eq. (1) are shown in Table 6.

Tables 7 and 8 tabulate the values of the coefficients of Eq. (2) for
medium dense sand and dense sand, respectively.
Table 6 Coefficients for δh*
Soil type a0 a1 a2 a3
Medium dense sand 0.1896 0.36193 -0.18585 1.757
Dense sand 0.1067 0.2847 -0.09736 1.8802
Figure 7 shows the plot of the maximum wall deflection

estimations using Eq. (1) versus the 2D FEM values, Eq. (1) is
reasonably accurate with a high coefficient of determination R2 of
Figure 6 Maximum wall deflection for various excavation stage in 0.9250 for medium dense sand and 0.9567 for dense sand,
two types of sands (d for dense sand while m for medium dense respectively. Similarly, Figure 8 shows the plot of the maximum
sand), L/B=2.8. wall deflection estimations using Eq. (2) versus the 3D FEM values,
where there is a high coefficient of determination (R2) of 0.9553 for
Figure 5 plots the maximum wall deflections of different medium dense sand and 0.9539 for dense sand, indicating the
excavation stages for various L/B ratios in medium dense sand (= applicability and accuracy of the proposed Polynomial Regression
0.06, 0.1, 0.2 and 1.0). In general, the horizontal wall displacements (PR) models in predicting maximum wall deflections. It should be
calculated by 3D simulations are greater than results from 2D noted that there is a great scatter in points representing the LR
simulations. It is also observed that for wall stiffness α≤0.2, the model estimation against the FEM calculated results.
wall deflection calculated by 3D simulations for various L/B ratios
approximates to the results calculated by 2D simulations at
excavation depth no more than 11 m, however, it has a tendency to
grow with L/B ratios when excavation depth is more than 11 m
below the ground level. Figure 6 compares the maximum wall
deflections at various excavation stages with different wall stiffness
in dense sand and medium dense sand for L/B=2.8. The
observations show that the maximum wall deflection increases when
soil strength decreases and decreases as the wall stiffness increases,
similar to the trend described previously. It is evidently that the
maximum wall deflection varies noticeably when excavation depth
is greater than 5 m yet less difference when excavation depth is less
than 5 m.
4. ESTIMATION MODELS
Based on the numerical results, Logarithmic Regression(LR) models
and Polynomial Regression (PR) models have been developed for
estimating the maximum wall deflection δhm induced by braced
excavations in dense sand and medium dense sand, respectively.
Four input parameters (Sv, L/B, α and He) are considered. For 2D
numerical results, the optimal regression equation for wall
Figure 7 Predicted maximum wall deflection δh* versus δhm by 2D
deflection takes the following form:
FEM
 h* = a0S va1(L/ B ) a2 α a3 He a3 (1)
For 3D results, the optimal regression equation takes the form of:
 h* = b0 + b1S v + b2S v2+ b3(L/ B )+ b4(L/ B ) 2+ b5 α+ b6 α 2+ b7He +

b8He2+ b9S v (L/ B )+ b10S v α+ b11S v He+ b12(L/ B ) α+ b12(L/ B ) He+
b 6 α He (2)
Table 7 Coefficients for δh* in medium dense sand

b0 b1 b2 b3 b4 b5 b6 b7
0 4.484 0.8894 -109.593 -1.7618 -0.4462 1.4837 113.654
b8 b9 b10 b11 b12 b13 b14
0.6079 -0.6089 -0.8631 0.293 6.288 -1.0801 -5.2397
Table 8 Coefficients for δhm in dense sand
b0 b1 b2 b3 b4 b5 b6 b7
0 4.1095 -1.6948 -59.823 -1.7105 -0.51504 1.0046 65.028
b8 b9 b10 b11 b12 b13 b14
0.38243 0.090104 -1.3302 0.2179 4.5105 -0.65509 -2.9924
443
Hsiung, B. C. B., & Dao, S. D. (2014). "Evaluation of constitutive

soil models for predicting movements caused by a deep
excavation in sands". Electronic Journal of Geotechnical
Engineering, 19, pp95-111.
Hsiung, B. C. B. (2016). "Impacts from three-dimensional effect on
the wall deflection induced by a deep excavation in
kaohsiung, taiwan". Japanese Geotechnical Society Special
Publication, 2(45), pp1602-1607.
Hwang, R. N., Lee, T. Y., Chou, C. R., and Su, T. C. (2012).
"Evaluation of performance of diaphragm walls by wall
deflection paths". Journal of Geoengineering, 7(1), pp1-12.
Khoiri, M., & Ou, C. Y. (2013). "Evaluation of deformation
parameter for deep excavation in sand through case histories".
Computers and Geotechnics, 47(47), pp57-67.
Kung T. C., Juang C. H., Hsiao C. L. and Youssef M. A. H., (2007).
"Simplified model for wall deflection and ground-surface
settlement caused by braced excavation in clays". Journal of
Geotechnical and Geoenvironmental Engineering, ASCE,
133(6), pp731-747.
Long, M. (2001). "Database for retaining wall and ground
Figure 8 Predicted maximum wall deflection δh* versus δhm by 3D movements due to deep excavations". Journal of
FEM Geotechnical and Geoenvironmental Engineering, 127(3),
pp203-224.
5. SUMMARY AND CONCLUSIONS Moormann, C. (2012). "Analysis of wall and ground movements
due to deep excavations in soft soil based on a new
In this study, a series of 2D and 3D simulations using the HS model worldwide database". Soils and Foundations -Tokyo-, 44(1),
for braced excavations in sands were carried out. In general, the pp87-98.
horizontal wall displacements calculated by 3D simulations are Moh, Z. C., and Song, T. F. (2013). "Performance of diaphragm
greater than results from 2D simulations, the maximum wall walls in deep foundation excavations". In: First International
deflection has a tendency to grow with soil strength and decreases as conferences on case histories in geotechnical engineering,
the wall stiffness increases. Meanwhile, it is found that small Missouri University of Science and Technology; pp1335–43.
horizontal displacement appears in the first two excavation stages, Nakai, T., Kawano H., Murata K., Banno M., and Hashimoto T.
and becomes considerable in the subsequent stages. On the other (1999). "Model Tests and Numerical Simulation of Braced
hand, based on the results, simple regression models are developed Excavation in Sandy Ground: Influences of Construction
for estimating the maximum wall deflections induced by braced History, Wall Friction, Wall Stiffness, Strut Position and
excavation in dense sands and medium dense sand, respectively. As Strut Stiffness". Soils and Foundations, 39 (3): pp1–12.
shown in Eq.(1) and Eq.(2). These models relate the maximum wall Ng, C.W.W. 1992. "An Evaluation of Soil-Structure Interaction
deflection to various parameters including the vertical strut spacing Associated with a Multi-propped Excavation". Ph.D thesis,
Sv, L/B ratios, the wall stiffness  and excavation depth He, the high University of Bristol, U.K.
coefficient of determination (R2) indicates the applicability and Nikolinakou, M. A., Whittle, A. J., Savidis, S., & Schran, U. (2011).
accuracy of the proposed Polynomial Regression (PR) models in "Prediction and interpretation of the performance of a deep
predicting maximum wall deflections. excavation in berlin sand". Journal of Geotechnical and
Geoenvironmental Engineering, 137(11), pp1047-1061.
6. REFERENCES Sabzi, Z., and Fakher, A. (2015). "The performance of buildings
adjacent to excavation supported by inclined struts".
Brinkgreve, R., Engin, E., and Engin, H. K. (2010). "Validation of International Journal of Civil Engineering, 13(1), pp1-13.
empirical formulas to derive model parameters for sands". In: Wang, I. W., Teng, F. C., Seed, R. B., and Ou, C. Y. (2008). "Using
Numerical Methods in Geotechnical Engineering, pp137– buttress walls to reduce excavation-induced movements".
142.
Geotechnical Engineering, 161(4), pp209-222.
Finno, R. J., Blackburn, J. T., and Roboski, J. F. (2016). "Three- Wong K. S, and Broms B. B. (1989). "Lateral wall defections of
dimensional effects for supported excavations in clay". braced excavations in clay = déplacements latéraux des parois
Journal of Geotechnical and Geoenvironmental Engineering, d'excavations avec butons dans l'argile". Journal of
133(1), pp30-36.
Geotechnical Engineering.
Goh, A. T. C., Zhang, F., Zhang, W., and Chew, O. Y. S. (2017). Xuan F. (2009). "Behavior of diaphragm walls in clays and
"Assessment of strut forces for braced excavation in clays reliability analysis". M.Eng.Thesis, Nanyang Technological
from numerical analysis and field measurements". Computers University, Singapore.
and Geotechnics, 86, pp141-149. Zapata-Medina, D. G., & Bryson, L. S. (2012). "Method for
Han, J. Y., Zhao, W., Chen, Y., Jia, P. J., and Guan, Y. P. (2017). estimating system stiffness for excavation support walls".
"Design analysis and observed performance of a tieback Journal o`f Geotechnical & Geoenvironmental Engineering,
anchored pile wall in sand". Mathematical Problems in 138(9), pp1104-1115.
Engineering, 9, pp1-23. Zhang, W., Goh, A. T. C., and Xuan, F. (2015). "A simple
Hsieh, P. G., Ou, C. Y. (2016). "Simplified approach to estimate the prediction model for wall deflection caused by braced
maximum wall deflection for deep excavations with cross excavation in clays". Computers and Geotechnics, 63, pp67-
walls in clay under the undrained condition". Acta Geotech; 72.
11: pp177–89.
444
Singapore Downtown Line 3 - Tunnelling Challenges in Soft Soil and under

Conserved Structures
Michael McGowan1, Sofren Leo Suhaendi2 and Gordon Lee3
1
Director, Arup Singapore Pte. Ltd., Singapore
2
Senior Tunnel Engineer, Arup Singapore Pte. Ltd., Singapore
3
Associate, Arup Singapore Pte. Ltd., Singapore
E-mail: Micahel.McGowan@arup.com1, Leo.Suhaendi@arup.com2
ABSTRACT: The opening of Downtown Line 3 (DTL3) in October 2017 marked the longest underground and driverless MRT line in
Singapore. The complicated technical challenges in DTL3, especially Package A, were associated with the nature of its geological strata and
construction of underground infrastructure below existing buildings and structures which were conserved by the Authority to some extent of
the alignment. Due to its vertical alignment, tunnelling works in DTL3-A was carried out within the Kallang Formation which was recent
deposits consisting of soft/loose soils. At some extent of the alignment, tunnelling works was significantly close to conserved shophouses
which were very sensitive to induced settlements. Settlement predictions, risk assessments and mitigation measures were therefore carried
out to ensure safety during TBM tunnelling. The monitoring regime and protection measures were successfully implemented with actual
ground movements to be well within the design predicted movements. This paper presents some of the technical challenges in DTL3 Package
A in relation to tunnelling works in soft soil condition and under conserved shophouses at some extent of the alignment.
Keywords: Downtown Line Stage 3, control tunnelling, face pressure, damage assessment, building protection measures.
1. INTRODUCTION This paper presents tunnelling works challenges encountered in

DTL3A, particularly ones related to tunnelling in soft soil and under
The Downtown Line (DTL) is the fifth Mass Rail Transit (MRT) conserved structures.
line in Singapore and was opened in three stages (i.e. DTL Stage 1,
2 and 3). The opening of DTL3 in October 2017 marks DTL as the
2. DOWNTOWN LINE 3 PACKAGE A
longest underground rail line at 42 km with 34 stations. It connects
the north-western and central-eastern regions to the downtown of 2.1 Tunnel Alignment
Singapore enhancing linkages between the historic Chinatown and
Little India districts of Singapore and enhances access to the historic Due to various site constraint along DTL3A alignment, some extent
civic areas of Fort Canning and Jalan Besar district. It is the longest of the bored tunnels configuration have to apply stacked
rapid transit line to use completely automated driverless trains. arrangement before transitioning to parallel one at the station
interfaces. This arrangement is quite prominent in Contract C933
where the tunnelling works needs to be carried out in a very close
proximity to quite a number of sensitive structures along Jalan Besar
area. The depth of tunnel to its axis is in the range of 16.5 m to 38
m.
Figure 1 DTL3 alignment route Figure 2 DTL3A – C933 alignment layout

Arup was appointed by the Land Transport Authority (LTA) as 2.2 Geological Condition
the lead design consultant to provide Architectural/ Engineering
(A/E) Consultancy Services for the 5 km long Downtown Line The geological condition encountered in DTL3A tunnelling works is
Stage 3 Package A (DTL3A) comprising of five underground summarised in Table 2.
stations running from Chinatown to Kallang Bahru. Details of civil
contracts in DTL3A is shown in Table 1. Table 2 DTL3A Geological Conditions along Tunnel Alignment
Contract Geology at Tunnel Face
Table 1 DTL3A Civil Contracts C937 Kallang Formation, Jurong Formation and FCBB
C935 Kallang Formation, Old Alluvium and FCBB
Contract Scope Contractor C933 Kallang Formation and Old Alluvium
Fort Canning Station and At some of the alignment extent, tunnelling works will encounter
C937 GS E & C full face of Kallang Formation beneath the made ground along
Associated Tunnels
C936 Bencoolen Station Sato Kogyo DTL3A, notably in Contract C933. Underlying the Kallang
Jalan Besar Station and Leighton Offshore – John Formation is the Old Alluvium formation. The Kallang Formation
C935 soils itself comprise the marine clay, fluvial sand, fluvial clay and
Associated Tunnels Holland JV
Bendemeer Stattion and estuarine clay. Marine clays notably present at Jalan Besat and
C933 Penta Ocean Construction Kallang Bahru area, sandwiching the fluvial sand and/or fluvial clay
Associated Tunnels
C932A Kallang Bahru Station China State Construction layers. The existence of both upper and lower marine clays are
found at Jalan Besar area.
445
soil condition. The target TBM face pressures for each tunnel bound
and drive are shown in Figure 4.
Figure 3 Geological profile along DTL3A – C933 bored tunnel

2.3 Bored Tunnel Segmental Lining
In general, the bored tunnels are lined with 1.4 m-wide, 275 mm-
thick precast segment, incorporating Grade 8.8 M24 bolts at both
segment to segment joint and ring to ring joint.
Table 3 DTL3A Segment Reinforcement and Erector System
Item C937 & C935 C933

Steel Fibre Reinforced
Concrete (SFRC) in
general;
Precast Reinforced
Hybrid Reinforced
Segment Concrete (RC)
Concrete @ tunnel
opening and close tunnel
proximity
Segment
Erector Mechanical Erector Vacuum Erector
System
3. TUNNELLING IN SOFT SOIL

3.1 Control Tunnelling Figure 4 Target TBM face pressure in DTL3A Contract C933
A total of four Earth Pressure Balance (EPB) TBMs were deployed
to perform the tunnelling works in DTL3A Contract C933 covering 3.2 TBM Specifications
approximately 4 km tunnel drives. The face pressure of EPB TBM is
maintained by proper combination of propulsion thrust and removal One specific requirement on EPB TBM in DTL3A is that they have
of muck at the correct rate matching the TBM advance rate. As to be equipped with an automatic face control (AFC) system where
highlighted by Shirlaw et al (2013), TBM face pressure has to be bentonite slurry will be automatically pumped into the cutter head
applied at all times to prevent face collapse during tunnelling in pressure chamber to compensate for deficiencies in the face support
Kallang Formation. Furthermore, a face pressure provision in the pressure, i.e. when pressure at the tunnel face drops below the
range of 0.9 to 1.2 total overburden pressure will also reduce the agreed minimum target face pressure and maintain minimum
potential settlements during the tunnelling particularly in this soft pressure in the plenum chamber independently of the shield jacks,
screw rotation or other soil conditioning system. This system will
446
ensure that no loss of face pressure will be encountered resulting in damage classification table first put forward by Burland et al. (1977)
excessive ground movement during TBM excavation. as shown in Table 4.
Figure 5 AFC schematic diagram inside EPB TBM
4. TUNNELLING UNDER CONSERVED STRUCTURES

4.1 Old Shophouse Along Jalan Besar
In Contract C933 and C935, DTL3A tunnel alignment runs below
Jalan Besar, one of the area safeguarded as conservation districts
under Secondary Settlement category by the Urban Redevelopment
Authority (URA). The area is located outside the central city district
developed after the end of World War I and have established its own
distinct identities over time. The shophouses sandwiching the tunnel
alignment are two- to four-storeys high, built in contiguous blocks
with common party walls in the late 19th & early 20th centuries and
traditionally designed to provide for business premises on the
Figure 6 Damage assessment methodology to buildings
ground floor and residential accommodation on the upper storeys.
The shophouses architecture styles comprise of Late, Transitional The third stage evaluation of the structure will be carried out if
and Art Deco styles. the building damage assessment is not satisfactory after the second
Other characteristic of these shophouses is the distinct five-foot stage assessment, considering more factors such as: soil structure
way feature, indented into the ground floor of shophouses from the interaction, structural continuity, foundation systems, construction
road with overhanging upper floors, serving as a sheltered space for sequence, orientation to alignment, etc.
social activities and circulation.
Case study on the tunnelling works under conserved shophouses Table 4 Building Damage Classification after Burland et. al.
has been highlighted by Poea, et. al. (2014).
Building damage classification (for masonry walls, cladding and
4.2. Damage Assessments finishes)
Limiting
The empirical prediction of ground movements due to bored Predicted Approximate
tensile
tunneling works assumed transverse ground settlement profile above degree of Description of damage width of
strain
a single tunnel is of normal probability distribution curve, or damage cracks (mm)
(%)
Gaussian form which generally produce a greenfield settlement
0
trough. The ground displacement is assumed to occur at constant Hairline cracks < 0.1 0 to 0.05
Negligible
volume, specified as a ‘volume ground loss’ at the tunnel. For
multiple tunnels effect, the movements induced by each tunnel can Fine cracks that are
easily treated during
be superimposed. Better prediction of ground movements can be
normal decoration.
carried out using advanced methods of numerical analyses, based on
Damage generally
the finite element method, such as Plaxis and Oasys FREW 1 0.05 to
computer programs. restricted to internal wall 0.1 to 1
Very slight 0.075
The LTA Civil Design Criteria for Road and Rail Transit finishes. Close inspection
may reveal some cracks
Systems prescribes a three-staged approach to the assessment of
in external brickwork or
damage to buildings for ground movements due to deep excavation
masonry.
and tunneling as shown in Figure 6.
The first stage applies a very simple and conservative approach Cracks easily filled.
for the preliminary assessment. Ground surface settlement contours Redecoration probably
along the alignment can be determined to filter all buildings required. Recurrent
experiencing a maximum slope of 1:500 and a settlement of less cracks can be masked by
than 10 mm. This approach is quite conservative as it neglects any suitable linings. Cracks
2 0.075 to
interaction between the stiffness of the buildings and the ground. may be visible externally 1 to 5
Slight 0.15
In the second stage, the building is represented by a simple and some repointing may
beam whose foundations are assumed to follow the displacement of be required to ensure
the ground in accordance with the greenfield site assumption. The weather-tightness. Doors
maximum resultant tensile strains are calculated for both the and windows may stick
hogging and sagging settlements. This result is then plotted on the slightly.
relevant interaction diagram of deflection ratio versus horizontal
strains. The resultant damage category is compared to the building
447
Building damage classification (for masonry walls, cladding and

finishes)
Limiting
Predicted Approximate
tensile
degree of Description of damage width of
strain
damage cracks (mm)
(%)
The cracks may require
some opening up and can
be patched by a mason.
Repointing of external
brickwork and possibly a
5 to 15 or
3 small amount of 0.15 to
several cracks
Moderate brickwork may need to 0.3
> 3mm
be replaced. Doors and
windows sticking.
Service pipes may Figure 7 Seismic logging set-up
fracture. Weather
tightness often impaired.
Extensive repair work
involving breaking-out
and replacing sections of
walls, especially over
doors and windows. 15 to 25 but
4 Windows and door also depends
> 0.3
Severe frames distorted, floor on number of
sloping noticeably. Walls cracks
leaning or bulging
noticeably, some loss of
bearing in beams.
Service pipes disrupted.
This requires a major
repair job involving
partial or complete
> 25mm but
5 rebuilding. Beams lose
also depends
Very bearing; walls lean badly
on number of
Severe and require shoring.
cracks
Windows broken with Figure 8 Seismic logging result
distortion. Danger of
instability
The greenfield surface and sub-surface ground movements due

to tunnelling works were predicted after which each building along
the alignment was assessed for possible building damage based on
the described methodology above.
Table 5 Building Damage Assessment Summary in DTL3A

DA
C937 C935 C933
Stage
2nd 14 buildings 81 buildings 25 buildings
Stage pass pass pass
3rd 11 buildings 13 buildings Figure 9 Trial Pit
-
Stage pass pass
4.4. Protection Measures

4.3. Site Investigation
Building protection measures were installed for conserved
At some occasion, foundation detection works (i.e. seismic logging shophouses along Jalan Besar area that did not pass the damage
and trial pit) needs to be carried out as part of investigation works in assessment criteria. The reason of these buildings not passing the
order to verify the type of foundation and to ascertain whether it damage assessment criteria was often due to the differential
impedes the TBM tunnel construction. settlement between five-foot way façade and the rest of the building.
A typical seismic logging set-up and test result are shown in The adopted strategy was then to allow the differential settlement
Figure 7 and Figure 8, respectively. Based on the logging test result, and install adjustable props to stabilize the five-foot way façade and
the toe level of bored pile was estimated at13.5m below ground provide tie beams at first floor level to anchor the base of the five-
level, approximately 9 m above the tunnel. foot way façade to the main building. Extensive instrumentation and
Figure 9 shows trial pit involving a careful mining process to monitoring regime was implemented to facilitate better observation
expose the foundation. Existence of bakau pile foundations was on the affected structures. A typical arrangement for building
found confirming foundations were of shallow type and will not protection propping is shown in Figure 10.
impede the tunnel construction.
448
4.5. Instrumentation & Monitoring

Instrumentation comprising of strain gauges was installed to the
building protection proppings for close monitoring of the loadings in
the propping systems. Two sensors were installed at the
instrumented props to measure the strain values. The load carried by
the props was estimated by the changes in strain values using the
following formula:
strain (ave)= [(sensor1) + (sensor2)] / 2 (1)
raw= strain initial (ave) – strain(ave) (2)
= raw * calibration factor (3)
Load (F)=  * E * Area (4)
where E= Young’s Modulus of structural steel (i.e. 205 GPa)
Figure 10 Typical arrangement for building protection propping
Based on visual inspection report, a number of shophouses were
found to be in poor condition. The defects recorded included very
large structural cracks and large areas of concrete spalling. Fot these
particular shophouses, enhanced strengthening and protection works
were installed prior to commencement of the bored tunnelling
works.
Figure 13 Typical Strain Gauges Arrangement
Figure 14 Plot of strain gauge at one of the prop

As shown in Figure 14, the loading in the one of the shophouses at
Jalan Besar was quite insignificant during the period when TBM
tunnelling works was carried out. This result trend was found to be
Figure 11 Typical enhanced external building protection measures consistent for the shophouses along Jalan Besar and Lavender
Street.
In addition to the building protection measures, temporary five-
foot way propping system was installed to restrain potential 5. CONCLUSION
differential movement of the five-foot way columns as shown in
Figure 12, comprising of a steel casing wrapped around the five-foot Tunnelling in soft ground and under conserved structures pose its
way column joined to a steel frame that was then tied into the own challenges and complications. Proper design, planning
existing shophouse frontage. (including site investigation, mitigation measures & specifications)
and execution prove to be the decisive factors that allows tunnelling
works to be successfully completed without causing significant
impact to any of the existing structures in the vicinity.
6. REFERENCES
Burland, J.B., et. al. (1977) “Behaviour of foundations and
structures”. Proceedings of the 9th International Conference of
Soil Mechanics and Foundation Engineering, pp495-546.
New, B.M., and O’Reilly, M.P. (1991) “Tunnelling induced ground
movements: Predicting their Magnitude and Effects”, 4th
International Conference on Ground Movements and
Structures, pp671-697.
Shirlaw, J.N., et. al. (2003) “Local settlements and sinkholes due to
EPB tunnelling”, Proceedings of the Institution of Civil
Engineers, Geotechnical Engineering 156, pp193-211.
Land Transport Authority (LTA) (2010) Civil Design Criteria for
Road and Rail Transit Systems, February 2010, Rev A1.
Arup. (2011) C933 Geotechnical Interpretative Baseline Report
(Tender).
Figure 12 Five-foot way protection system
449
HKIE Geotechnical Division. (2014) Ground Control for EPB TBM Ong, C.W., et. al. (2015) “A case study of twin bored tunnelling
Tunnelling, GEO Report No. 298. under mixed-face soil – Bendemeer MRT station project
Poea, J., et.al. (2014) “Protection of old shophouses due to MRT (Downtown Line 3), Singapore”, Proceedings of the 15th
excavation & tunnelling works on Downtown Line Stage 3 Asian Regional Conference on Soil Mechanics and
C933”, Proceedings of Underground Singapore 2014, Session Geotechnical Engineering, pp176-181.
6.5, pp33. Urban Redevelopment Authority (URA). (2017) Conservation
Guidelines, December 2017 edition.
450
Seasonal Variation of Water Content and Pore-water Pressure Distribution in

Vegetated Soil Slope
N. Gofar1, H. Rahardjo1, A. Satyanaga1
1
School of Civil and Environmental Engineering, Nanyang Technological University, 50 Nanyang Avenue, Singapore. 639798.
E-mail: nurlygofar@ntu.edu.sg
ABSTRACT: This paper presents soil water content and pore-water pressure distribution in a vegetated residual soil slope in response to a
one-year seasonal variation in Singapore. The measurements were taken by tensiometers and soil moisture sensors TM1-SM1, TM2-SM2,
TM3-SM3 and TM4-SM4 installed at vertical distances of 2.0, 2.0, 1.37 and 0.23 m from slope surface. Seasonal variation was represented
by rainfall and actual evaporation calculated based on data collected by a weather station installed at the site. The field monitoring shows
that the soil water content and pore-water pressure measurements by TM1-SM1, TM2-SM2, TM3-SM3 were representative of soil-water
characteristic curve (SWCC) of the residual soil. On the other hand, the measurements by TM4-SM4 shows a lower pore-water pressure
during dry period and a lower volumetric water content during wet period as compared to SWCC of residual soil. The study showed that the
response recorded by TM4-SM4 was representative of top soil used as media for vegetation planting. The study also showed that the larger
variation of pore-water pressure recorded by TM4 in December 2016 and January 2017 was due to the long dry period with high temperature
which is not normal in Singapore during these months.
Keywords: Field monitoring, soil water content, pore-water pressure, soil-water characteristic curve, vegetated soil slope.
1. INTRODUCTION recommended the use of updated Penman-Monteith Method or

FAO-56 Method (Zotarelli et al, 2009) to calculate the rate at which
Soils in tropical regions commonly consist of residual soils with readily available soil water is vaporized from a specified vegetated
negative pore-water pressure in the zone above ground water table surface ETo. The effect of different species of vegetation can be
(Rahardjo et al. 2005). The pore-water condition is influenced by considered by modifying the equation with a crop coefficient, Kc
flux boundary conditions such as rainfall, evaporation, and (Lazzara and Rana, 2010), thus (ETc = ETo × Kc).
transpiration processes. Earlier studies considered only rainfall as The transient volumetric water content and pore-water pressure
the flux boundary condition to avoid more complexity in their in soil are measured by soil moisture sensor and tensiometer
analysis of pore-water pressure and slope stability (e.g. Chipp et al., respectively. The state of moisture in the unsaturated zone of soil
1982; Rahardjo et al., 1998; Ng et al., 2003; 2008 and Li et al., between the ground surface and the water table (Blight, 1997) is
2005; Lee et al., 2008). However, not all rainfall become infiltration. controlled by the water balance between the water input into and
From the definitions, the rainfall may be separated into four water output from the soil. The change in water stored in soil lead to
components, i.e. runoff, infiltration, interception (rainfall that is the change in both volumetric water content and pore-water
caught on the vegetation surfaces), and evapotranspiration (ET) pressure. Long term real-time monitoring is required to capture the
(Joel et al., 2002). Thus, some researchers took further steps to effect of different weather conditions throughout the year.
consider the portion of precipitation that infiltrates into the soil (Ng The water flow into the soil is governed by the coefficient of
et al. 2003; Rahardjo et al. 2005; Gofar et al. 2008). Other permeability with respect to water (kw) (Fredlund and Rahardjo,
researchers also considered the effect of evaporation on the 1993). Unlike saturated soils, the permeability of an unsaturated soil
prediction of pore-water pressure variation in soil slope (e.g. Gasmo is a non-linear function of the volumetric water content of the soil.
et al., 2000; Gitirana et al, 2005 2006; Yunusa et al, 2014). When the soil approaches saturation, the permeability becomes
Numerical analysis performed by Yunusa et al. (2014) on one-year constant and equal to the saturated coefficient of permeability, ks.
data showed a better agreement with field response of unsaturated Thus, the response of soil to the flux boundary conditions is
soil when the input data was the combination of rainfall and controlled by the soil-water characteristic curve (SWCC) which is
evaporation rates. Rahardjo et al (2017) included the effect of non-linear with respect to the negative pore-water pressure
evapotranspiration in their study of pore-water pressure variation in (suction). In other words, the ability of the unsaturated soil to retain
vegetated soil slope. These studies suggested rainfall and water varies with soil suction. The SWCC follows different paths
evaporation as well as transpiration play important roles in affecting during drying and wetting (hysteresis) in nature. The soil on the
the pore-water pressure distribution within residual soil slope. drying path has a higher water content than the soil on the wetting
The response of a soil to flux boundary conditions can be path at a given matric suction. Thus, the field data of volumetric
evaluated by field monitoring of the flux boundaries as well as the water content and pore-water pressure could be plotted within the
transient volumetric water content and pore-water pressure in the hysteretic of the drying and wetting curves of the SWCC.
soil. The intensity of rainfall that falls on the slope is normally This paper presents the response of a residual soil slope to flux-
monitored using a tipping-bucket rain gauge. In contrast to rainfall, boundary conditions based on the field monitoring data collected
evapotranspiration rate from vegetated surface cannot be easily from an instrumented slope in Singapore. The rainfall data collected
measured. Potential evaporation rate (PE) can be measured in the by rain-gauge and evapotranspiration (ETc) were used as flux
field using Lysimeter or predicted based on climate data (Penman, boundary conditions while the responses were identified by
1948) equation. The PE is an evaporation from an open water body volumetric water content and pore-water pressure recorded by soil
whereby the relative humidity is equal to unity. Sattler and Fredlund moisture sensor and tensiometers. The responses were compared to
(1991) showed that the actual evaporation (AE) from a soil surface the soil-water characteristic curve (SWCC) of the corresponding soil
is related to the relative humidity in the soil at ground surface, for surrounding the soil moisture sensors and tensiometers tip.
dry climate like in Saskatchewan, it is about 70% of the PE.
In addition to the relative humidity of soil surface, transpiration
should be considered for the effect of flux boundary conditions on
2. THEOTERICAL BACKGROUND
vegetated surface. Several empirical methods are available to obtain
evapotranspiration (ET) rate. Weather station can be installed to The evaporation from ground surface covered with certain
capture the meteorological variables required to calculate the species of vegetation (ETc) can be calculated by considering
evapotranspiration such as air temperature, solar radiation, relative standard meteorological variables and a crop coefficient (Kc)
humidity and wind speed (Rahardjo et al., 2014). Allen (1998) (Lazzara and Rana 2010), thus ETc is calculated as follows:
451
ET C  K C  ET O (1) protection rod and the pyranometer were installed on top of the
where ETo is the grass reference evapotranspiration in mm/day mast. The rain-gauge was installed at a height of 1.6 m from the
which can be calculated as follows: ground surface together with the data logger. Soil temperature (ST)
was measured at depths of 0.10 m, 0.15 m, 0.20 m, and 0.25 m.
900
0 , 408  ( R n  G )   u 2 (e S  ea ) The tensiometers (TM) and soil moisture sensors (SM) were
(T  273 ) (2)
ET O  installed within residual soil layer at a depth of 2 m from crest and at
   (1  0 ,34 u 2 ) distances 0.4 m; 2.4 m; 3.8 m and 4.6 m from the slope face. The
jet-fill tensiometers were calibrated properly prior to installation and
where T = mean air temperature (oC); u2 = wind speed (m/s) at 2 m then inserted into a tube to the specified depth. The reading from
above the ground; Rn = net radiation flux (MJ/m2/day; G = sensible transducer was verified by a bourdon gauge attached to it. The
heat flux into the soil (MJ/m2/day); es = saturation vapour pressure response of tensiometers to pore-water pressure change must be
(kPa); ea = actual vapor pressure (kPa); ∆ = slope of saturation checked regularly to ensure the quality and physical performance of
vapour pressure curve; and γ = psychrometric constant. The detailed high air-entry ceramic tips. In this study, regular maintenance of the
calculation of this method can be found in Zotarelli et al. (2009). tensiometers was conducted twice a week by refilling the jet-fill
Typical Kc values are given in Lazzara and Rana (2010), for grass, reservoir with de-aired water and flushing the tensiometers to
Kc= 1. remove the accumulated or trapped air in the tubes caused by
The water flow through an isotropic unsaturated soil is cavitation of water and air diffusion through the ceramic tip. Soil
formulated using Darcy’s law as follows (Fredlund and Rahardjo, moisture sensor used in this study was of TDR type which was
1993): capable of measuring soil moisture up to saturation (0–100%) with
accuracy of 1% and response time of 0.5 second. All soil moisture
  hw    hw  hw sensors were tested in water and air environment and their values
 k w   m 2w  w g (3)
 kw  corresponded to the values specified by manufacturer which was
x  x  y  y  t about 70 to 90% in pure water and zero in air.
with,
uw (4)
hw  z 
wg
where hw = hydraulic head; z = elevation head; uw = pore-water

pressure (kPa); x and y are the Cartesian coordinates in the x- and y-
directions, respectively; kw = permeability function (m/s); w=
density of water (Mg/m3); g = gravitational acceleration (m/s2);
m2w= coefficient of water volume change with respect to a change in
matric suction (ua-uw); ua= pore-air pressure (kPa), and t = time
(second).
Equation 4 shows that the unbalanced flow of water through a
soil element is equal to the change in water volume in the soil Figure 1. Plan view residual soil slopes with instrumentation
element. Unlike saturated soils, the permeability of an unsaturated locations
soil is not constant (Fredlund and Rahardjo, 1993). The coefficient
of permeability with respect to water for a soil is a non-linear The readings of all measuring instruments were calibrated and
function of the volumetric water content of the soil. When the soil checked before they were connected to a data acquisition system
approaches saturation, the permeability becomes constant and equal (DAS) to obtain instrumentation readings in real time. The data
to the saturated coefficient of permeability with respect to water, ks. logger was powered by solar panel and battery. The cables were
In addition, the volumetric water content of the soil is dependent on protected by corrugated tubing to prevent damages from rainfall and
the negative pore-water pressure in a non-linear fashion (soil-water insects. The data logger recorded readings at a 10-minute interval
characteristic curve or SWCC). In other words, the ability of the regardless of rainfall events. The data logger sent all data through
unsaturated soil to retain water varies with soil suction. SWCC general packet radio service (GPRS) to a web page for remote
follows different paths during drying and wetting (hysteresis) in monitoring. Figure 2 shows the weather station and data acquisition
nature where the soil on the drying path has a higher water content system used at the study site.
than the soil on the wetting path at a given matric suction.
2. METHODOLOGY
2.1 Field Instrumentation
The investigated slope in this study is located at the central part of
Singapore within residual soil from Bukit Timah Granite. Figure 1
shows diagram of residual soil slope and the field instrumentation
carried out in this study. The instrumentations can be divided into
two parts; the first part was weather instrumentation for climatic
data measurements while the second part was instrumentation for
quantifying soil response to the flux boundary conditions. The
monitoring period was one year (1st July 2016 to 30th June 2017).
The weather instrumentation includes tipping bucket rain gauge
for rainfall, pyranometer for solar radiation, wind monitor for wind
speed and direction, as well as temperature and relative humidity
probes. The weather station for climatic data measurement was
located near the toe. A 10 m high galvanized mast was erected on a
concrete footing. The wind monitor, the solar panel, the lightning Figure 2. Data acquisition system used at the study site
452
2.2 Soil Properties than the mean monthly rainfall for the months in Singapore.
November 2016 was the wettest month during the monitoring period
The slope was formed by residual soil with a thin layer of top soil as with 21 rainfall days accumulating to 311 mm rain, while August
media for growing vegetative cover. The soil properties required for 2016 was the driest month with only 8 days of rainfall accumulating
this study were investigated by performing laboratory tests on to 98.6 mm of monthly rainfall. Past data (1981 – 2010) showed that
samples taken from the field. The grain size distribution shows that the wettest month in Singapore was November while the driest
about 50% of the material is clay. The unified soil classification month were June and February. Figure 5 shows that the maximum
system USCS (ASTM D2487-00) categorized the soil to be highly daily rainfall occurred on 23rd January 2017 i.e. 103.8 mm. The
plastic clay (CH). The bulk density of the residual soil was 1.8 month of January 2017 represented extreme condition because it
Mg/m3. started with a two-week dry period followed by very wet period
The saturated permeability of the soil was determined using the towards the end of the month. The cumulative rainfall in January
flexible-wall saturated permeability test (ASTM D5084-10). The 2017 was 298.4 mm. Even though the number of days with rainfall
saturated permeability of the soil is 6×10-7m/s. In order to perform in January was 17, much of the rainfall (217.8 mm) was actually
analyses related to water flow in unsaturated soil, permeability occurred on 18th to 24th January.
functions of the soil need to be obtained. Permeability function can
be determined from the soil-water characteristic curve (SWCC)
using a statistical method proposed by Childs and Collis-George
(1950). The procedure for the prediction can be found in Fredlund
and Rahardjo (1993). The SWCC of the soil from the site was
obtained using Tempe cell and pressure plate according to ASTM
D6838-02. The wetting and drying SWCCs of the residual soil are
shown in Figure 3. The permeability functions obtained using the
prediction method based on wetting and drying SWCCs are shown
in Figure 4.
Figure 5. Daily rainfall from July 2016 to June 2017.
Figure 3. Drying and wetting SWCCs of residual soil
Figure 4. Drying and wetting permeability curves of residual soil
3. RESULTS AND DISCUSSION Figure 6. Mean monthly and number of rainfall days
3.1 Rainfall 3.2 Climatic data

st th
Figure 5 shows daily rainfall recorded from 1 July 2016 to 30 The climatic data observed from the weather station from July
June 2017. The cumulative yearly rainfall was 2819 mm which was 2016 to June 2017 including air temperature (Ta), relative humidity
higher than the average annual rainfall in Singapore based on long (RH), solar radiation (SR) as well as wind speed (WS) are shown in
term record from NEA Singapore (1981–2010) i.e. 2166 mm. The Figures 7. The climatic data obtained from the study site were also
number of rainfall days during the monitoring period was 178 days, compared with the typical climatic data reported by NEA Singapore
which was higher than the average annual number of rainfall days in based on data from 1981 to 2010.
Singapore (167 days). Rainfall monitoring at the study site indicated The minimum and maximum air temperatures during the
that the monthly rainfalls were quite different from the typical trend monitoring period were 22.9 and 36.9oC, respectively. The range of
in Singapore as shown in Figure 6. air temperature was wider and higher than the mean range of air
For example, the monthly rainfall in December 2016 (214.3 temperature in Singapore i.e. 23–33oC. As for the rainfall, the
mm) was too low as compared to the mean monthly rainfall in variation of air temperature during the monitoring period was
Singapore for December (318 mm). On the other hand, the monthly different from the typical values recorded in Singapore. Both
rainfall in May 2017 (319 mm) was very high as compared to the maximum and minimum temperatures (37 and 23.2oC) were
average rainfall in Singapore for May (171 mm). The monthly measured in in January 2017 followed by December 2016 with
rainfalls in September, October and November 2016 were higher maximum of 36.9oC and minimum of 23.6oC). Therefore, based on
453
both rainfall and air temperature, the month of January 2017 3.2 Evapotranspiration
represent an extreme condition. Past data indicated that the
maximum temperature usually occurred in March to May while Figure 8 shows the ETc calculated using FAO-Penman or FAO-56
minimum temperature occurred in December and January. In this Method for ETo and coefficient of 1 for Kc, plotted together with
study, the trend of higher temperature in March to May was rainfall data. The minimum, mean and maximum daily potential
overcome by heavy downpours (Figure 6). evaporations in the study area were 0.66, 1.92, and 3.56 mm,
Soil temperature was measured at depths of 0.10, 0.15, 0.20 and respectively. The highest daily ETc occurred in June 2017 while the
0.25 m from ground surface. Field measurements performed in this lowest was in November 2016. The highest monthly ETc in June
study indicated that the soil temperature varied with depth and the was 65.29 mm) while the lowest monthly ETc in November was
time of the day. The least variation with the time of the day was 47.77 mm. The ETc obtained in this study were slightly less than
obtained in the measurement of soil temperature at depth of 0.20 m those reported by Rahardjo et al. (2017) based on their studies in
from the ground surface. Figure 7a shows the variation of soil different parts of Singapore. The minimum and maximum ETc
temperature measured at depth 0.20 m as compared to the maximum calculated in their studies were 0.55 and 4.3 mm/day, respectively.
and minimum daily temperatures. The soil temperature was In their study, the highest daily ETc occurred in February while the
relatively low in November, December and January, and relatively lowest was in June 2017.
high in May, June and July.
Relative humidity plays an important role in the prediction of
ET. The minimum and maximum relative humidity recorded in the
study site were 67.4% to 99% with a mean value of 80%. The
recorded RH was higher than the typical range recorded in
Singapore i.e. 60 to 95%. The maximum relative humidity occurred
in November due to the high number of days with rainfall. The
lowest relative humidity was recorded in August due to the least
number of rainfall days. Figure 7b shows the maximum and
minimum daily RHs during the monitoring period. Observation of
Figure 7b indicates that low relative humidity was maintained
Figure 8. Evapotranspiration on slope surface at the study site from
during the first two weeks in both December 2016 and January 2017
July 2016 to June 2017
due to lack of rainfall.
The minimum and maximum solar radiations during the
3.3 Pore-water pressure measurements
monitoring period were 1.12 and 14.76 MJ/m2/day, respectively
with a mean value of 7.93 MJ/m2/day. These values are the typical The response of soil slope to rainfall and evapotranspiration from
values for Singapore. The maximum solar radiation was recorded in vegetated surface in terms of pore-water pressure is presented in
the dry month of August 2016. The variation of solar radiation at the Figure 9. The figure shows that the measurements at TM1 – TM3
study site during the monitoring period is shown in Figure 7c. were quite consistent, while the measurements at TM4 varied
Observation of Figure 7c also indicated that high solar ration was significantly especially from November 2016 to January 2017.
maintained during the first two weeks in both December 2016 and Based on the evaluation of rainfall and climatic data, this period
January 2017. represented extreme changes from dry to wet conditions.
The maximum wind speed (4.92 m/s) was recorded in February As shown in Figure 1, TM1 and TM2 have vertical distances of
2017. The recorded wind speed was lower than the range of wind 2 m from crest while TM3 was 1.37 m below the slope face. TM4
speed in Singapore i.e. 0–13.375 m/s with a mean value of 2.65 m/s. was very close to the slope face i.e. 0.4 m horizontal distance or
This may be because the location of the study site was surrounded only 0.23 m vertical distance. Thus, TM4 was more affected by flux
by high rise buildings. The variation of wind speed throughout the boundary conditions as compared to the other tensiometers.
year is also presented in Figure 7c.
*Note: Refer to Figure 1 for instrumentation locations

Figure 9. Pore-water pressure recorded at the study site from July
2016 to June 2017
3.4 Soil’s response to flux boundary conditions
Field data of pore-water pressure and volumetric water content

throughout the monitoring period was plotted together with SWCC
of residual soil in Figure 10. It can be seen that measurements by
TM1-SM1, TM2-SM2 and TM3-SM3 were plotted well within the
hysteretic of the SWCC. The range of suction was quite narrow i.e.
4 – 27 kPa. The volumetric water content ranged from 27% to the
saturated volumetric water content of the soil (51%). This shows
that the measurements at TM1 to TM3 are representative of the
Figure 7. Variation of climate data from July 2016 to June 2017 residual soil forming the slope. On the other hand, the measurements
454
by TM4 were below the SWCC wetting curve of the residual soil. It water pressure to be recorded by TM4 which was located closest to
was deduced that the soil response at TM4-SM4 was more the ground surface (Figure 9).
representative of the top soil instead of the residual soil. Readings of pore-water pressure and soil water content recorded
Soil samples were collected from the slope surface to a depth of by TM1-SM1, TM2-SM2, TM3-SM3 are representative of the
20 cm. SWCC and saturated permeability tests were carried out on SWCC of the residual soil. Thus, the readings are representative of
the samples using the same procedure as for the residual soil. It was the soil where the instrumentations were installed. However, the plot
found that the saturated volumetric water content of the top soil was for TM4-SM4 located at a vertical distance of 23 cm from the
lower than that of the residual soil. The top soil had a lower density ground surface was more representative of the top soil which has a
as compared to that of the residual soil, thus the higher void ratio lower density and a higher coefficient of permeability as compared
shifted the SWCC to the left, resulting in a lower air-entry value. to the residual soil. This showed that the readings of instrumentation
The saturated coefficient of permeability of the top soil was 1 × 10-5 closer to the ground surface was affected by the flux boundary
m/s which was higher than that of the residual soil (6 × 10-7 m/s). conditions and the consistency of surface soil as well as vegetative
cover.
4. REFERENCES
ASTM D5084-10 Standard Test Method for Measurement of
Hydraulic Conductivity of Saturated Porous Material using a
Flexible Wall Permeameter, ASTM International, West
Conshohoken. PA.
ASTM D6838-02 Standard Test Method for Determination of
SWCC for Desorption using Hanging Column, Pressure
Extractor, Chilled Mirror Hygrometer or Centrifuge, ASTM
International, West Conshohoken. PA.
ASTM D 2487-00 Standard Practice for Classification of Soils for
Engineering Purposes (Unified Soil Classification System) ,
ASTM International, West Conshohoken. PA.
Figure 10. Plot of pore-water pressure and volumetric water content
Allen R.G., Pereira L.S., Raes D., Smith M. (1998) Crop
measured at the study site from July 2016 to June 2017 in SWCC of
evapotranspiration: guidelines for computing crop water
residual soil.
requirements Food and Agriculture Organization of the
United Nation, Rome.
The field data from TM4-SM4 were plotted on the SWCC
Blight G.E. (1997) The “Active” Zone in Unsaturated Soil
curves of the top soil as shown in Figure 11. The figure shows that
Mechanics. 1st GRC Lecture, Nanyang Technological
the field data could be plotted well inside the hysteresis of the
University, Singapore.
SWCC of the top soil. The range of suction was from 8 to 90 kPa,
Childs E.C., Collis-George G.N. (1950) “The permeability of porous
while the range of volumetric water content was from 13 to 30%.
materials”. Proc. Royal Society of London 210A. pp. 392-
The maximum volumetric water content was less than the saturated
405.
volumetric water content of the surface soil. It may be attributed to
Chipp, P.N., Henkel, D.J., Clare, D.G. & Pope, R.G. (1982) Field
some water intercepted by vegetation (grass) on the ground surface.
Measurement of Suction in Colluvium Covered Slopes in
Hong Kong. Proc. 7th Southeast Asian Geotechnical Conf.,
Hong Kong, pp.49-62.
Fredlund D.G., & Xing A. (1994) “Equations for the Soil-Water
Characteristic Curve”. Canadian Geotech. Journal 31 pp. 521-
532.
Fredlund, D. G. & Rahardjo, H. (1993). Soil Mechanics for Un-
saturated Soils. New York: John Wiley & Sons, Inc.
Joel, A., Messing, I., Seguel, O., & Casanova, M. (2002) “Meas-
urement of surface water runoff from plots of two different
sizes”. Hydrological Processes. 16(7) pp. 1467-1478.
Lazzara P. & Rana G. (2010) “The use of crop coefficient approach
to estimate actual evapotranspiration: a critical review for
major crops under Mediterranean climate”. Italian Journal of
Figure 11. Plot of pore-water pressure and volumetric water content Agrometeorology 2 pp. 25-39.
measured by TM4 from July 2016 to June 2017 in SWCC of top Lee, M.L. Gofar N. & Rahardjo H. (2009) “A Simple Model for
soil. Preliminary Evaluation of Rainfall-Induced Slope Instability.
Engineering Geology”. 108(3-4) pp. 272-282.
4. CONCLUSIONS Li, A.G., Yue, Tham, L.G. and Lee C.F., (2005) “Field-monitored
variations of soil moisture and matric suction in a saprolite
Field measurements of flux boundary conditions and soil response slope”. Canadian Geotechnical Journal. 42, pp. 13–26.
were carried out in a residual soil slope at the central part of Gasmo, J.M., Rahardjo, H. and Leong, E.C. (2000) “Infiltration
Singapore. The measurements presented in Figures 5 and 7 specified Effects on Stability of a Residual Soil Slope”. Computer and
that the seasonal variation of rainfall was quite different from the Geotechnics, 26, April, pp.145–165.
typical trend in Singapore (Figure 6). The dissimilar trend also Gitirana Jr., G., Fredlund, D.G., Fredlund, M. (2006) “Numerical
observed for the other flux boundary conditions such as air modeling of soil-atmosphere interaction for unsaturated
temperature and relative humidity. Highly variable conditions surfaces”. Proc. 4th Intl. Conf. Unsaturated soils, US., pp.658-
occurred during the months of December 2016 and January 2017. 669.
This could be identified by unusually high air temperature and long Gitirana Jr., G., Fredlund, D.G., Fredlund, M. (2005) “Infiltration-
period of dry condition which are different from the normal runoff boundary conditions in seepage analysis”. Proc. 58th
conditions in Singapore. The long dry period in combination with Canadian Geotech. Conf. Canada, pp.516-523.
low relative humidity and high solar radiation allowed high pore-
455
Gofar N. Lee, M.L. & Kassim, A. (2008) “Response of Suction Rahardjo, H., Satyanaga, A. and Leong, E.C. (2013). “Effects of
Distribution to Rainfall Infiltration in Soil Slope”. Electronic Flux Boundary Conditions on Pore-water Pressure
J. of Geotechnical Engineering. 13 (E), pp.1-13 Distribution in Slope”. Engineering Geology, 165, October,
National Environment Agency, 2010.Meteorological Services Data. pp. 133–142.
National Environment Agency, Singapore. Rahardjo, H., Lee, T.T. Leong, E.C. and Rezaur, R.B. (2005).
Ng, C.W.W., Springman, S.M. and Alonso E.E. (2008). “Response of a Residual Soil Slope to Rainfall”. Canadian
“Monitoring the Performance of Unsaturated Soil Slopes”. Geotechnical Journal, 42(2), pp. 340–351.
Geotechnical and Geological Engineering. 26(6), pp.799-816. Rahardjo, H., Leong, E.C., Gasmo, J.M. and Tang, S.K. (1998)
Ng. C.W.W., Zhan, L.T., Bao, C.G., Fredlund, D.G. and Gong, “Assessment of rainfall effects on stability of residual soil
B.W., (2003) “Performance of an Unsaturated Expansive Soil slopes”. Proc. 2nd Intl. Conf. Unsaturated Soils, China, pp.
Slope Subjected to Artificial Rainfall Infiltration”. 280-285.
Géotechnique. 53(2), pp. 143-157. Sattler, P.J. & Fredlund, D.G. (1991). “Modelling vertical ground
Penman, H.L (1948) “Natural evapotranspiration from open water, movements using surface climatic flux”. Proc. Geotechnical
bare soil and grass”. Proc. Royal Society London Ser. A.193, Engineering Congress, US., pp. 1292-1306.
pp.120-145. Yunusa, G.H. Kassim, A. & Gofar, N. (2014) “Effect of Surface
Rahardjo, H., Amalia, N., Leong, E.C. Harnas, F.R. Lee, T.T. and Flux Boundary Conditions on Transient Suction Distribution
Fong, Y.K. (2017) “Flux Boundary Measurements for the in Homogeneous Slope”. Indian J. Science & Technology,
study of Tree Stability”. Landscape Ecol Eng. 13:81–92 7(12), 2064–2075
Rahardjo, H., Satyanaga, A. Harnas, F.R. and Leong, E.C. (2014) Zotarelli L, Dukes MD, Romero CC, Migliaccio KW, Morgan KT
“Comprehensive Instrumentation for Real Time Monitoring (2009) Step by step calculation of Penman-Monteith
of Flux Boundary Conditions in Slope” Proc. 3rd Italian evapotranspiration (FAO-56 Method), University of Florida.
Workshop on Landslides, Italy.
456
Strength Characteristics of Cement-treated Peat in Sumatera Island, Indonesia

Hirochika Hayashi1, Takahiro Yamanashi1, Hijiri Hashimoto1,
Eddie Sunaryo2, Fahmi Aldiamar2, Maulana Iqbal2 and Dea Pertiwi2
1
Civil Engineering Research Institute for Cold Region (CERI), Sapporo, Japan
2
Institute of Road Engineering (IRE), Bandung, Indonesia
E-mail: hayashi@ceri.go.jp
ABSTRACT: When constructing road embankment over peat, ground improvement is needed to avoid problems such as bearing capacity
failure and excess post-construction settlement. The cement stabilization technique is one of the new proven ground improvement which can
greatly improve ground in a short period of time. In this study a series of laboratory testing using unconfined compression test on peat mixed
with some type of cements with different chemical compositions, including a special cement which contains large amount of SO 3 than other
type of cement, was conducted. The tested peat was collected in a site at Dumai in Sumatera Island, Indonesia. As a result, the cement
stabilization has considerable potential to improve strength for the peat from very soft to stiff and hard consistency.
Keywords: ground improvement, cement stabilization, peat, unconfined compression strength, modulus of elasticity.
1. INTRODUCTION 2.2 Cement used

Fibrous and highly organic peat, which is very soft and problematic Three types of cement (Portland cement, blast–furnace slag cement
soil, is widely distributed in Southeast Asian Countries. Especially and a special cement) were used as the binder. Table 2 shows
in Indonesia, extremely large peat deposit is found in Sumatera, chemical composition of these cements. The Portland cement and
Java, Kalimantan and Papua (IRE, 2001). Peat hinders the slag cement were produced according to Japanese Industrial
implementation of construction project, due to its unique Standard. The special cement (product name: ET201) was
engineering properties. commercially developed for stabilizing highly organic soil by a
When constructing road embankment over peat, ground Japanese cement company. The feature of ET 201 is that it includes
improvement is needed to avoid problems such as bearing capacity more sulfur trioxide (SO3) and has greater specific surface area as
failure and excess post-construction settlement. The cement compared to the ordinary cements.
stabilization technique is one of the new proven ground In this study, the content of additive cement (cement ratio, CR),
improvement which can greatly improve ground in a short period of defined in Equation (1), were 20%, 30%, 40% and 50%. Where, WC
time (CERI, 2017). However, there is a possibility that organic is the dry weight of the cement and WP is the wet weight of peat.
material included in peat hinders the solidification of cement. The
unexpected effect may leads to insufficient strength of cement- Cement Ratio (%) = (WC / WP）×100 (1)
treated peat against the required strength. The problem should be
solved by exercising various type of cement to increase the quality
of soil stabilization works and give more contribution in practical
works.
In this study a series of unconfined compression tests on peat
mixed with several type of cements composed of different chemical
compositions, including a special cement for ground improvement
produced in Japan was conducted as a fundamental research. This
composition can become a hint of future development of local
cement in Indonesia.
2. METHODOLOGY
A laboratory mixing test was performed on Dumai peat with
ordinary and special cement commercially available in Japan for
cement stabilization in order to clarify their effect on the peat and
confirm the suitability of implementation this technology in
Indonesia, especially with similar condition with Dumai peat.
2.1 Soil Property of Peat tested Figure 1 Location of sampling site
Peat sample for the laboratory test was collected from Dumai,
Sumatera Island, Indonesia. (Figure 1). Table 1 shows engineering Table 1 Engineering Soil Properties of Peat Sample
soil properties of the peat sample which resulted high water content Specific gravity Gs 1.51
of over 700%, ignition loss of over 95% and low pH value of 3.4. Natural water content Wn (%) 769
The peat included a lot of organic materials such as Bitumen and Iginition loss Li (%) 95.2
Humic acid that known to hinder the solidification of cement (e.g., PH 3.4
Okada et al, 1983; Noto, 1991; Hayashi and Nishimoto, 2005). The Degree of decomposition (von Post) H3-H4
physical and chemical properties of the peat are difficult soil
Content of organic material (%) Bitumen 8.3
condition for cement stabilization.
Humid acid 56.2
Unconfined compression strength UCS (kPa) No. 1 7.3
(undisturbed sample before stabilization) No.2 12.8
457
Table 2 Chemical Composition of Used Cement

Specific Chemical Composition (%)
SAC
Type of Cement surface area SiO2 Al2O3 CaO SO3 Others
ratio
(cm2/g)
Ordinary Portland 3310 21.1 5.7 63.9 2.1 7.2 0.12
Blast Furnance Slag 3810 26.0 8.7 55.4 1.8 8.1 0.19
Special Cement (ET201) 6250 22.6 8.2 49.8 12.5 6.9 0.42
SAC ratio = (Al2O3+SO3)/CaO
2.3 Test Procedure

According to the “Practice for Making and Curing Stabilized Soil
Specimens without Compaction” (JGS 0821-2009: JGS, 2016), a
standard defined by the Japanese Geotechnical Society, specimens
5 cm in diameter and 10 cm in height were made as follows. First,
peat and cement slurry with W/C (W: weight of water, C: dry
weight of cement) of 0.6 for each CR were poured into an electric
mixer (Fig. 2) and mixed well for 10 minutes. Next, the mixture was
put in a mold in three layers, without compaction. Then, the
specimens were tamped to avoid voids.
The unconfined compression test (JGS 0511-2009: JGS, 2015)
was conducted after 7 days and 28 days of laboratory curing at a
temperature of 20°C.

3.1 Strength of Cement-treated Peat
Figure 3 shows a typical result of the unconfined compression test
on undisturbed peat before stabilization (original peat), stabilized
peat using Portland cement and ET 201 of CR = 50% after 28 days
curing. For the original peat, no clear peak in the stress-strain curve
was found and the unconfined compression strength (compression
stress at failure: UCS) was very low. The result is a typical
mechanical behavior of peat (Noto, 1991; Huat et al., 2014).
Meanwhile, the stress-strain curve of the stabilized peat was
Figure 2 Electric mixer used in this study
characterized by significant higher UCS and smaller axial strain at
failure as compared with the original peat. The results show that the
cement stabilization has considerable potential to improve strength
for the peat from very soft consistency (UCS<24 kN/m2) to stiff
consistency (UCS ranges from 96-192 kN/m2) for Portland cement
and hard consistency (UCS>383 kN/m2) for ET 201 according to
Terzaghi and Peck (1967).
Figure 4 (a) and (b) show the relationship between the CR and
the UCS of stabilized peat after 7 and 28 days curing respectively.
The UCS of stabilized peat increased with the increase of the CR. It
should be noted that the UCS of stabilized peat varied depending on
type of cement. In cases in which Portland cement and slag cement
were used, the UCS was approximately 100 kN/m2, even at CR =
50%, and it was low improvement effect. It is thought that the
unexpected effect is caused by hindering cement solidification of the
organic materials included in peat.
When ET 201, a special cement containing large amounts of SO3
and has greater specific surface area was used, the UCS after 28
days curing besides CR = 20% was higher than that using the
ordinary cements. Hayashi and Nishimoto (2005) conducted a series
of laboratory tests on stabilized Japanese peat with different types of
cement. They pointed out chemical reactions for increasing the UCS
of stabilized peat with a special cement containing large amounts of
SO3 as follows. The SO3 in a special cement is contained as gypsum
(CaSO4). A typical reaction, in which a hydration product is
produced through the binding of gypsum and a large amount of
Figure 3 Typical result of unconfined compression test on original
water, is shown below.
peat and stabilized peat after 28 days curing
3CaO-Al2O3＋3CaSO4＋32H2O 3CaO-Al2O3-3CaSO4 -32H2O
458
Figure 4 Relationship between Cement Ratio and UCS of stabilized peat
3CaO-Al2O3-3CaSO4-32H2O is a hydrogen product called

ettringite (Fig. 5). It is characterized by large, needle-shaped crystals
unlike ordinary hydrogen products. It binds together with a large
amount of water during its formation process, which lowers the
water content of the soil. It is thought that the increase in the
strength of stabilized soil progresses with the entanglement of these
needle-shaped crystals and peat. This reaction is also unlikely to be
hindered by organic substances in the soil. This is a reason that the
ET201 was effective for the peat.
In case of all types of cement, the UCS after 28 days curing
increased from that after 7day curing as shown in Fig. 4. To clarify
this trend, the relationship between the UCS after 7 days and 28
days is shown in Fig. 6. For most of the data except for some cases
in the ET201, the UCS after 28 days curing ranged from 1.5 to 3.0
times the UCS after 7 days curing. In these phenomena, no
significant difference was found due to the difference in cement type.
Figure 5 Ettringite, a needle-shaped hydrogen product (SEM photo
3.2 Modulus of Elasticity of Cement-treated Peat of stabilized peat after Hayashi and Nishimoto, 2005)
When analyzing deformation of ground improved by using a deep

cement stabilization method by numerical modeling, it is important
to accurately determine stiffness of cement-treated soil. Therefore,
modulus of elasticity of the cement-treated peat is described in this
Section. The modulus of elasticity (E50) as defined in Equation (2) is
mean stiffness in range from small strain to strain at failure, and is
calculated using the stress-strain curve as shown in Fig.2. Where,
unit of the E50 is kN/m2, unit of the USC is kN/m2 and 50 is the
axial strain at compression stress of (USC/2) (%). In analyzing static
deformation of improved ground due to embankment loading, the
E50 is often used for evaluating the stiffness of the cement stabilized
soil.
E50 = ((USC/2)/50) x 100 (2)
Figure 7 (a) and (b) show the double-logarithm relationship

between the E50 and the UCS of stabilized peat after 7 and 28 days
curing respectively. Without depending on the type of cement, the
E50 increased linearly with the increase of the UCS. Kitazume and Figure 6 Relaintionship between UCS of stabilized peat after 7 days
Terashi (2013) presented that a similar tendency is observed for curing and after 28 days curing
various types of cement-stabilized clay. This relationship in this
study can be approximated by Eq. (3) and (4). In case of 7 days curing: E50 = 18.0 UCS1.45 (3)
In case of 28 days curing: E50 = 13.7 UCS1.49 (4)
459
4. CONCLUSIONS
In this study, a series of laboratory mixing testing on peat collected
at a site of Dumai in Sumatera Island, Indonesia using different
types of cement was conducted. The main results can be
summarized as follows:
1. The tested peat has high water content of over 700%, ignition
loss of over 95% and low pH value of 3.4. The physical and
chemical properties are difficult soil condition for cement
stabilization.
2. In cases in which Portland cement or blast furnace cement were
used, small strength enhancement from unconfined compression
strength (UCS) resulted from curing time of 7 and 28 days.
Stabilized peat can only achieve stiff consistency.
3. When special cement (product name: ET201) containing large
amounts of sulfur trioxide or aluminium oxide was used,
enhancement of high UCS result was obtained. Stabilized peat
can achieve hard consistency.
4. For most of the data except for some cases in the ET201, the
UCS after 28 days curing ranged from 1.5 to 3.0 times the UCS
after 7 days curing.
5. Without depending on the type of cement, the modulus of
elasticity (E50) increased linearly with the increase of the UCS.
Based on the result, an experimental correlation between the E50
and the UCS is presented.
This study was collaboratively carried out based on
“Agreement on Research Exchange and Cooperation between Civil
Engineering Research Institute for Cold Region (CERI), Japan and
Institute of Road Engineering (IRE), Indonesia”, in order to
technically support national projects of highway construction over
peat in Indonesia.
5. REFERENCES
Civil Engineering Research Institute for Cold Region (CERI). (2017)
Manual for Countermeasure for Peat Soft Ground, pp.1-6. (in
Japanese)
Hayashi, H. and Nishimoto S. (2005) Strength Characteristic of
stabilized peat using different types of binders, Proceedings
of The International conference on Deep Mixing Best
Practice and Recent Advances (CD-R), Stockholm.
Huat, B. B. K., Prasad, A., Asadi, A. and Kazemian, S. (2014)
Geotechnics of Organic Soils and Peat, CRC press, pp.81-95.
Institution of Road Engineering (IRE). (2001) Guideline of
Indonesian Road Construction over Peat and Organic Soils
Figure 7 Relationship between UCS and modulus elasticity Japanese Geotechnical Society (JGS). (2016) Laboratory Testing
Standards of Geomaterials, Vol.2.
Japanese Geotechnical Society (JGS). (2015) Laboratory Testing
Standards of Geomaterials, Vol.1.
Kitazume, M. and Terashi, M. (2013) The Deep Mixing Method,
CRC press, pp.83-84.
Noto, S. (1991) Peat Engineering Handbook, CERI, pp.102-110.
Okada, Y., Kutara, K. and Miki, H. (1983) Effect of humic acid on
soil stabilization, Proceedings of the 53rd Annual Conference
of Japan Society of Civil Engineers, pp.467-468. (in
Japanese)
Terzaghi. K., and R.B. Peck. (1967). Soil Mechanics in engineering
practice, Wiley New York p.729.
460
Basement Excavation in Soft Marine Clay in Bukit Tinggi, Klang

S. Sharmeelee1 and EG Balakrishan2
1
GCU Consultants Sdn Bhd, Selangor, Malaysia
2
GCU Consultants Sdn Bhd, Selangor, Malaysia
E-mail: sharmeelee@gcu.com.my
ABSTRACT: Thick deposits of very soft marine clay can be found in Klang which lies along the West Coast of Malaysia. Basement excavation
up to 3.5m below the existing ground level has been carried out in Bukit Tinggi, Klang using three ground retention methods namely propped
sheet pile wall, cantilever sheet pile wall and cantilever contiguous bored pile wall. This paper elaborates on the implemented ground retention
methods and explores the ground’s response to the excavation in soft marine clay through the recorded instrumentation data.
The estimated wall deflections from the 2D FE analysis were compared against the measured field data to verify the adopted soil model and
the FE analysis. The results show that the adopted soil parameters are realistic and the sequence of work adopted in the FE analysis and
subsequently in the basement excavation work has minimized the wall deflections.
Keywords: Soft clay, propped excavation, cantilever excavation,
1. INTRODUCTION
The coastal areas of Peninsular Malaysia are typically composed of
3. SOIL CONDITION
thick deposits of very soft marine clay generally deposited during the
Holocene period. The soft marine clay can extend to depths exceeding The subsoil at this site based on the boreholes that were carried out
50m below the existing ground level close to the coast. indicates that the site is underlain by very soft clay with depths
Bukit Tinggi in Klang, Selangor is a fast developing varying from 24m to 30m below the existing ground level. Thin sand
township which is predominantly composed of low-rise residential lenses were found at selected locations across the site. The hard clay
and commercial development. Basements are rare in this area due to layer is typically found at depths ranging from 32m to 40m below the
the difficulties and high cost usually associated with deep excavation existing ground level.
in very soft marine clay. Table 1 shows the summary of the soil description and
Mixed commercial and residential development has been properties where γ is the bulk density of the soil (kN/m3), Su is the
proposed in this area in two (2) adjoining plots i.e. Plots A and B with undrained shear strength of the soil (kPa), N is the SPT N of the soil
one (1) level of basement. The plots are flanked by existing (blows/ft) and Eu is the Undrained Young’s Modulus of the soil (kPa).
commercial and residential development on either sides. The
basement excavation works were carefully planned to minimise the Table 1 Summary of the soil description and properties
settlement of the surrounding and to avoid distress to the adjacent
structures. Depth Soil γ Su Eu
A soil model was developed using the available soil (m)
investigation (SI) data and a Finite Element (FE) analysis was carried 0–8 Very Soft Clay 16 10 1200
out by simulating the full excavation and construction process. 8 - 20 Very Soft Clay 16 15-25 2200-3500
Instrumentation was installed to monitor the ground movement. The
field behaviour of the retention system is compared with the FE 4. PROJECT DESCRIPTION
analysis and discussed.
The proposed development comprises of two (2) plots which share a
2. SITE GEOLOGY boundary as shown in Figure 2.
The proposed site is underlain by Quaternary Deposit as shown in
Figure 1.
This formation consists of extensive deposits of
unconsolidated/semi-consolidated marine deposit consisting of
mainly CLAY deposited under marine environment. This is likely to
be underlain by the Pre-Quaternary sedimentary formation.
Figure 2 Site Layout
Plot A is located 5m away from the closest residence and

10m away from a five (5) storey shopping mall with one (1) level of
basement car park while Plot B is located 5m away from a primary
school and 10m away from a five (5) storey shopping mall with one
(1) level of basement car park.
A 1.2m diameter functioning water pipe which is the main
Figure 1 Site Geology water supply for this area is located along the site boundary at Access
461
Road 1 located approximately 1m below the existing ground level.

Access Road 1 is a busy public road catering to the residential area
and the school. Access Road 2 is a private road which predominantly
caters to the shopping mall. The mall can also be accessed from
several other entry points apart from Access Road 2.
5. ANALYSIS & DESIGN OF RETENTION SYSTEM
For both Plots A and B, one (1) level of basement was

proposed with maximum excavation depth of 3.5m from the existing
ground level. The permanent basement wall comprised of a reinforced
concrete (RC) wall integrated with the building structure.
The piling for both plots comprised of jack-in spun piles
which were installed to set from the piling platform level.
To cast the basement slab and permanent RC wall,
temporary ground retention was required. Taking into consideration
the functioning water pipe along Access Road 1, the temporary Figure 4 Stage 2 Construction Sequence
ground retention systems for Plots A and B were proposed as follows.
Stage 3 : Install remaining spun piles from base of excavation and
• Plot A : Propped temporary sheet pile wall cast pile cap and base slab up to the slope toe. Install waler and strut
• Plot B : Cantilever permanent contiguous on sheet pile wall and prop against basement slab.
bored pile (CBP) wall along Access
Road 1 &
Cantilever temporary sheet pile wall
along Access Road 2
The analysis of the propped temporary sheet pile wall, cantilever

sheet pile wall and the cantilever CBP wall was carried using Finite
Element Analysis by considering an undrained analysis for the soft
clay.
5.1 Plot A
Propped temporary sheet pile wall was proposed for Plot A. The
sequence of works for Plot A is shown in Figures 3 - 8 and is
elaborated below.
Stage 1 : Install temporary sheet piles along project boundary. Allow Figure 5 Stage 3 Construction Sequence
1m working space from temporary sheet pile to permanent RC wall.
Install first two (2) rows of foundation piles to set using jack-in Stage 4 : Excavate temporary slope and cast pile cap.
method from existing platform level
Figure 6 Stage 4 Construction Sequence
Stage 2 : Excavate to form stable temporary slope with slope gradient

1V:4H. At the base of excavation, remove and replace 1.5m thick of
soft marine clay with suitable compacted fill.
At the toe of the slope, remove and replace 2m thick of soft marine
clay with suitable compacted fill.
462
Stage 5 : Cast remaining slab and RC wall. Allow for opening in the
wall for the strut. Repeat Stages 4 and 5 between alternate struts
without removing the soil berm beside these struts until the full slab
and RC wall has been cast.
Stage 6 : Fill the gap between the RC Wall and the sheet pile wall up
to maximum 2m using suitable fill.
Figure 9 Permanent CBP wall construction sequence
The sequence of works for the cantilever temporary sheet pile wall is
shown in Figure 10 and is elaborated below.
Stage 7 : Remove the temporary strut, waler and sheet pile wall.
Backfill to the existing ground level.
5.2 Plot B
Along the site boundary adjoining Access Road 1, cantilever CBP
wall was installed taking the functioning water pipe into
consideration. Along the site boundary adjoining Access Road 2,
cantilever sheet pile wall was installed.
The sequence of works for the cantilever CBP wall is shown in Figure
9 and is elaborated below.
Stage 1 : Install 600mm diameter bored piles at 675mm centres and

15m length to form the CBP wall from the existing ground level. The
top of the CBP wall is 1m below the existing ground level.
Stage 2 : Construct the capping beam. Excavate to form stable
temporary slope with height of 1m and slope gradient 1V:4H behind Figure 10 Temporary sheet pile wall construction sequence
the CBP wall. The exposed height of the CBP wall is 2.5m
Stage 3 : Excavate to base of excavation in front of the CBP wall. Stage 1 : Install 12m length temporary sheet piles along project
Remove and replace 1.5m thick of soft marine clay with suitable boundary at Access Road 1. Allow 1m working space from temporary
compacted fill. sheet pile to permanent RC wall. The top of the sheet pile is 1.5m
below the existing ground level.
Stage 4 : Install foundation spun piles from base of excavation and
cast pile cap, base slab and skin wall. Stage 2 : Excavate to form stable temporary slope with height of 1.5m
and slope gradient 1V:4H behind the sheet pile wall. The exposed
Stage 5 : Reinstate the ground behind the CBP wall to the finished height of the sheet pile wall is 1.5m
platform level.
Stage 3 : Excavate to base of excavation in front of the sheet pile wall.
Remove and replace 1.5m thick of very soft marine clay at base of
excavation with suitable compacted fill.
463
Stage 4 : Install foundation spun piles from base of excavation and pipe was monitored during the excavation works and no distress was
cast pile cap, base slab and RC wall. observed on the pipe at the end of the construction works. The
Stage 5 : Fill the gap between the RC Wall and the sheet pile wall cantilever CBP wall and sheet pile wall also allowed for speedier
with suitable granular fill. construction.
Stage 6 : Remove the temporary sheet pile. Reinstate the ground 8. CONCLUSION
behind the RC wall to the finished platform level.
Basement excavation up to 3.5m below the existing ground level has
6. INSTRUMENTATION been carried out in Bukit Tinggi, Klang using three ground retention
methods namely propped sheet pile wall, cantilever sheet pile wall
Instrumentation in the form of ground settlement markers and and cantilever contiguous bored pile wall.
inclinometers were installed behind the temporary sheet pile walls A FE analysis was carried out to confirm the details of the
and CBP wall for both Plots A and B. retention system. The estimated wall deflections from the 2D FE
The monitoring was carried out on a daily basis during the analysis were found to be comparable to the measured field
excavation works to monitor the wall deflection and ground instrumentation data thereby validating the adopted soil model, the
settlement. The monitoring was terminated when the ground floor FE analysis and the construction sequence.
slab was fully constructed. The ground retention systems elaborated in this paper
provide simple and economical methods of retaining very soft clay
7. FIELD PERFORMANCE OF RETENTION SYSTEM where excavation depths do not exceed 3.5m below the existing
ground level.
7.1 Plot A
9. REFERENCES
The performance of the temporary sheet pile wall was evaluated by
studying the recorded instrumentation data and a comparison has O’Rourke, T. D., (1993) “Base stability and ground movement
been made between the recorded wall deflection and the data prediction for excavations in soft clay”. Retaining structures,
extracted from the FEM analysis, as shown in Table 2 below. Thomas Telford, London, pp. 131-139.
Ukritchon B., Whittle, A. J., and Sloan, S. W., (2003) “Undrained
Table 2 Summary of Estimated & Recorded Wall Deflection for stability of braced excavation in clay”. ASCE J. Geotech.
Plot 1 Geoenviron. Eng. Vol. 129, No. 8, pp. 738-755.
Bjerrum, L. & Eide, O. (1956) “Stability of strutted excavations in
Location Estimated Deflection Actual clay”. Geotechnique, 6, pp. 32-47.
from FEM Deflection Chang, M. F. (2000) “Basal stability analysis of braced cuts in clay”.
(mm) (mm) J. Geotechnical and Geoenvironmental Engineering. ASCE,
Along Access Road 1 15 7 126(3), pp. 273-275.
CIRIA C580 (2003). Embedded retaining walls – guidance for
Along Access Road 2 15 15 Economic design. Graba A.R., Simpson, B., Powrie, W. &
Beadman,
As can be observed from Table 2, the estimated wall deflections are Goh, A. T. C. (1994) “Estimating basal-heave stability for braced
quite comparable to the actual wall deflections recorded on site. The excavations in soft clay”. J. Geotech. Engrg. Div., ASCE,
actual wall deflections were lower along Access Road 1 as the 120(GT8) pp. 1430-1436.
excavation for the temporary slope restraining the temporary sheet Karlsrud, K. (1986) “Performance monitoring of deep supported
pile wall was carried out in very short intervals of 6m as compared to excavations in soft clay”. Proc. 4th Int. Geot. Seminar, Field
the excavation along Access Road 2 which was carried out at 12m instrumentation and in-situ measurement’, NTU, Singapore,
intervals. pp. 187-202.
Lim, P. C. & Tan, T. S. (2003) “A Floating-Type braced excavation
7.2 Plot B in soft marine clay”. Proceedings of Underground Singapore
2003, Singapore, pp. 326-337.
Mana, A. I. & Clough, W. G. (1981) “Prediction of movements for
The performance of the temporary sheet pile wall and the permanent
braced cuts in clay”. J. Geotech. Engrg. Div., ASCE, 107(GT6)
CBP wall was evaluated by studying the recorded instrumentation
pp. 759-778.
data and a comparison has been made between the recorded wall Terzaghi, K. (1943). Theoretical soil mechanics. John Wiley, New
deflection and the data extracted from the FEM analysis, as shown in York.
Table 3 below.
Table 3 Summary of Estimated & Recorded Wall Deflection for

Plot 2
Location Estimated Deflection Actual
From FEM Deflection
(mm) (mm)
Along Access Road 1 – 20 19
Permanent CBP Wall
Along Access Road 2 – 30 31
Temporary Sheet Pile
Wall
As can be observed from Table 3, the estimated wall deflections are

quite comparable to the actual wall deflections recorded on site. A
higher wall deflection was allowed along Access Road 2 as there were
no critical structures located here. Along Access Road 1, the water
464
Effect of Rapid Impact Compaction Energy on Unconfined

Compressive Strength Properties of Fill Clay Soil
Arifin Beddu1, Lawalenna Samang2, Tri Harianto3 and Achmad Bakri Muhiddin3
1
Doctoral Student, Civil Engineering Department, Hasanuddin University, Indonesia
2
Professor, Civil Engineering Department, Hasanuddin University, Indonesia
3
Associate Professor, Civil Engineering Department, Hasanuddin University, Indonesia
E-mail: arifin_beddu@yahoo.com
ABSTRACT: In order to advantages the Rapid Impact Compaction (RIC) test method, a number of Unconfined Compression Tests (UCT)
were performed on the RIC compaction samples. The UCT sample size is 6 cm in diameter and 12 cm in height, which was drilled out from
the mould compaction that has been previously compacted with RIC Method. The compaction process is carried out in a standard modified
mould (15 cm in diameter and 25 cm in height) by varying number of 5 - 35 blows for each compacted sample with frequency 30 - 40 blows
per minute. All compaction processes are controlled by electro-mechanical new development system of repeated RIC method. The test
results represented that at the same energy level with larger number of blow variation of RIC results produce higher unconfined compressive
strength value in comparing to that of the energy level for larger masses with smaller number of blows. It is possible to make relative
compaction relation which is the result of the achievement of UCS value using Rapid Impact Compaction method to UCS value of proctor
method that fulfill a linear line equation. Based on these results, it could be created a simple relation of energy compaction in determining
unconfined compressive strength value with appropriate process and lower RIC energy consumption.
Keywords: rapid impact compaction, unconfined compression test, clay soil.
1. INTRODUCTION where maximum dry density is reached, namely maximum dry

density and optimum water content (Rajasekhar, et al., 2016).
The selection of the compaction method is one of the important Increasing the compaction effort will increase maximum
stage in the embankment construction. Several methods of dynamic density but will reduce the optimum water content. In general, the
compaction for a thick soil layer is known today among them are the compaction curve shows that the ratio of air voids remains the same
Rapid Impact Compaction. This method is a dynamic soil at maximum density, so at high moisture content, there is only a
compaction with Low Energy Dynamic Compaction concept. slight increase in the density obtained by increasing the compaction
(Becker, 2011), (M M Mohammed, et al., 2013). The main effort. The clay soil has a much higher optimum water content,
equipment of Rapid Impact Compaction consists of masses of consequently reducing the maximum dry density. The effect of
hammer which is dropped from a certain height. increasing the compaction effort is also much greater in the case of
In the laboratory the effect of compaction effort is generally clay soil (Rajasekhar et al., 2016).
tested using standard proctor and modified proctor with reference to
compaction curve. Has been commonly known that if the energy of
2.2 Rapid Impact Compaction,RIC
the compaction effort of per unit volume of soil is changed then the
tendency of a curve of compaction will also change with increasing Rapid Impact Compaction Equipment, RIC consists of three main
compaction effort and main parameters is the dry density and water components namely the foot hammer; rig where the hammer is
content (Das, 2010). attached, and hammer its self (Kristiansen & Davies, 2004)
In relation to the Rapid Impact Compaction method and the (M Mohammed et al., 2010). On the method of RIC, the hammer is
commonly used compaction standard that is standard proctor and dropped on a foot hammer which is placed above the surface of the
modified proctor then to obtain optimal implementation in the use of soil with the frequency of the blows 30 - 60 /min as a characteristic
compaction energy, it is necessary to conduct a research to extent of rapid impact compaction (Falkner, et al., 2010).
the effect of Rapid Impact Compaction energy, especially on the The parameters of the compaction energy of RIC methods is
compaction of fine gradation soil such as clay and silt. determined by the weight of the hammer, the weight and area of the
geometry of the foot hammer, the frequency and the cumulative
2. LITERATURE REVIEW amount of the blows in one cycle of compaction at the fixed point.
One important thing, in accordance with the results of previous
2.1 Characteristics of Compaction research (Koohsari, et al., 2016) that in the impact method the effect
Proctor (1933) (Das, 2010), has observed that there is a definite of hammer weight is more significant than the falling height.
relationship between moisture water content and the dry. In general, Similar to the magnitude of the compaction energy according
for a variety of soil types, there is a value of optimum moisture to Falkner on the method of dynamic compaction, then the
content to achieve maximum dry density. Dry density at optimum cumulative amount of blows energy per area of contact area of the
moisture content is defined as maximum dry density. Any increase foot hammer (kNm/m2), determined by the impact area of the foot
in water content that exceeds the optimum water content tends to hammer (anvil) (m2), weight of hammer, weight of hammer foot,
reduce dry density. In addition to the water content and the soil type, velocity of blows, and number of blows (Falkner et al., 2010).
other important factors that affect the compaction is the energy per
unit volume (Das, 2010). The compaction energy per unit volume 3. MATERIAL and METHODS
used for the Proctor test is determined by the amount of blows of the
3.1 Characteristics of Compaction
layer; number of layers per mould and per unit volume of mould.
The general relationship between water content, density and The local soil samples are used, the physical properties of soil
compaction energy can be predicted from a compacted curve. In an samples are summarized in Table 1 and Figure 1.
effort compaction, increasing water content will be relatively easy to
change the structure of the soil, and the achievement of dry density
will be increased with a given compaction effort. There are two
main parameter quantities obtained from the compaction curve
465
Table 1 Physical Properties of The Materials

Properties Values
Loose Wet density,(kN/m3) 10.30
Water content (%) 38.36
Specific gravity 2.73
Liquid Limit, (%) 55.89
Plastic limit, (%) 47.48
Plasticity index (%) 8.41
Sieve Analysis Hydrometer Analysis

#4 #10 #20 #40 #60 #100 #200 Clay
100
Figure 3 Compaction apparatus test: a. Cylindrical mould test;
90
b. Anvil
80
70
3.3 Test carried out
60
Percent Passing (%)
50 The soil in the compaction mould was obtained from the remoulded
40 sample, the required amount of soil was mixed with water and left
30 under a plastic for air drying to obtain a water content in accordance
20
with the standard proctor compaction of 15-35% and 0.046 kN of
10
soil was filled in the compaction mould without compacting by
0
controlling the loose density of 10.4 kN/m3. The compaction process
10,0000 1,0000 0,1000 0,0100
Diameter (mm)
0,0010 0,0001 with rapid impact compaction was performed using various energy
levels by applying the various impact blows.
Figure 1 Grain size distribution
In the compaction process, the mass of compactor was
periodically dropped over the anvil with ranging of frequency
3.2 Apparatus of Testing 30 - 40 blows per minute and adjustable falling height were set in
In this experimental test, the equipment of repeated load impact 15 cm, 20 cm and 25 cm for each sample tested, the impact process
used air-pneumatic piston. The main components of the equipment being carried out by varying blow ranging from 5 to 35 amount of
consisted of air compressor, piston double-acting air pneumatic- the blows for each mould compaction at specified height.
pressure, valve solenoid, panel controller and mass of compactor as The UCT sample size is 6 cm in diameter and 12 cm in height,
shown in Fig 2. Repeated load impact equipment has 150 kg vertical which was drilled out from the mould compaction that has been
lift capacity of mass compactor and falling height can be set 10 cm previously compacted. In this research, all compaction process were
to 30 cm, the time period of the impact compaction is being 2 - 3 controlled by a new developed electro-mechanic system of rapid
seconds per blow. For testing stability, the apparatus was attached impact compaction.
and positioned on a steel frame.
Figure 4 Program of testing: a. Sample preparation; b. Set up of

anvil; c. Compaction proses; d. Measurement of UCS

4.1 Energy of Proctor and Rapid Impact Compaction
The compaction energy of proctor methods are based on compaction
efforts that carried out in a 101.6 mm diameter compaction mold
Figure 2 Schematic of model equipment of electro-mechanic system with a height of 116.43 mm having a volume of 944 cm3. In the
rapid impact compaction model standard proctor test the amount of compaction energy is 594 kJ/m3
obtained from the compaction of 3 layers per mould using a pounder
To produce energy required, the weight of the mass compactor weighing 2.5 kg that dropped from a height of 30.05 cm with the
was used by 45 kg and 70 kg of steel blocks which are attached at amount of 25 blows for each layer, for modified proctor using a
the ends of the piston rod. These compactor mass can be dropped at weight of 4.5 kg for a falling height of 45.7 cm, with 5 layers per
a height in accordance with the setting of the height on the control mould using 25 blows per layer, the compaction energy size is 2,695
panel. For testing mould, the property of mould used were inner kJ/ m3.
diameter of 15 cm, 25 cm in height, volume of the mould 0.0044 m3. While for Rapid Impact Compaction method in this research
For the compacting process, the mould were equipped with use mould diameter 15 cm and height 25 cm with volume 4,417 cm3.
a pedestal pad as foot compactor or anvil which has adjusted size to As for the weight scheme of hammer weight used is 70 kg, by
the diameter of the mould. The anvil was by steel plate which has 2 varying the falling height respectively 15 cm, 20 cm and 25 cm for
cm of thickness, the rod connection was 15 cm in height and the number of 5, 15, 25 and 35 blows. All compaction processes are
diameter of 7.5 cm. carried out using only 1 layer on each mould in the test, so from this
466
RIC compaction process the energy of compaction effort varies 12,50

between 111 kJ /m3 for the smallest and 1,365 kJ/m3 for the highest RIC-01 = 0,20 ESP
energy or when compared to the standard energy of the proctor 12,00 RIC-02 = 0,59 ESP
varies between 0.20 - 2.3 times. RIC-03 = 0,98 ESP
11,50 RIC-04 = 1,38 ESP
Dry Density, g d (kN/m 3 )

Table 2 Energy Compaction Scheme Based on Compaction Type
Type of Hammer Falling Blows Layer Volume Compaction Energy 11,00
Compaction Weight Height Per Per Mould Energy Ratio to
(kg) (m) Layer Mould (m3) (kJ/m3) Proctor
10,50
Standard
2.5 0.31 25 3 0.0009 594.29 1.00
Proctor
Modified 10,00
4.54 0.46 25 5 0.0009 2695.13 4.54
Proctor
RIC-01 70 0.15 5 1 0.0044 117.05 0.20
RIC-02 70 0.15 15 1 0.0044 351.15 0.59 9,50
RIC-03 70 0.15 25 1 0.0044 585.26 0.98
RIC-04 70 0.15 35 1 0.0044 819.36 1.38
RIC-05 70 0.20 5 1 0.0044 156.07 0.26 9,00
RIC-06 70 0.20 15 1 0.0044 468.20 0.79 10,00 15,00 20,00 25,00 30,00 35,00 40,00
RIC-07 70 0.20 25 1 0.0044 780.34 1.31 Water Content, w %
RIC-08 70 0.20 35 1 0.0044 1092.48 1.84
RIC-09 70 0.25 5 1 0.0044 195.09 0.33
Figure 5 The density relations to optimum moisture content
RIC-10 70 0.25 15 1 0.0044 585.26 0.98 and energy levels
RIC-11 70 0.25 25 1 0.0044 975.43 1.64
RIC-12 70 0.25 35 1 0.0044 1365.60 2.30
4.3 Unconfined Compressive Strength vs Optimum
4.2 Dry Density and Optimum Water Content at Different Compaction Energy
Energy Levels
Tables should be presented as indicated in Table 1. Their layout Unconfined compressive strength, UCS for optimum energy level
should be consistent throughout. Horizontal lines should be placed were taken for moisture content of 25%. To obtain the optimum
above and below label headings, below subheadings and at the end energy level then the observation of Rapid Impact Compaction
of the table. Vertical lines should be avoided. compaction is done at different energy levels as shown in Figure 6.
First, to explain the relationship of dry density variables,
optimum water content compaction and energy level of the 410
compaction effort, it has been observed with the relationship of
Unconfined Compressive Strength,UCS (kPa)
360
these three variables, in this case the relationship is expressed as a
comparison to the energy of standard proctor method which is 310
594.29 kJ/m3 to the changes in the moisture content of each 260

compaction mould according to the amount of energy of Rapid
210
Impact Compaction given as stated in the Table 3.
160
Table 3. Dry Density at Different Energy and Water Content Levels 110
Type of Hammer Falling Blows Cummulative Energy Water Dry 60 intercept energi line
Compaction Weight Height Energy Ratio to Content Density
(kg) (m) (kJ/m3) Proctor (%) (kN/m3)
10
15 10.52
0,0 0,5 1,0 1,5 2,0 2,5 3,0 3,5 4,0 4,5 5,0
20 10.55
RIC-01 70 0.15 5 117.051 0.20 25 10.66 Energy Ratio to Proctor Standar Energy
30 10.32
35 9.31 Figure 6 Linearization of UCS values from the relation of the
15 11.05
20 11.05
proctor compaction energy line to the RIC energy line
RIC-02 70 0.15 15 315.153 0.59 25 11.33
30 10.86
35 9.80 The energy line compaction Rapid Impact Compaction which
15 11.30 in this case according to Eq. (1), the other line is the energy line
20 11.53
RIC-03 70 9.15 25 585.256 0.98 25 11.74 connecting the standard energy level of the proctor and the modified
30 10.90 proctor which has previously been examined the relationship of its
35 10.00
15 11.60 unconfined compressive strength value which has been linearly and
20 12.09 satisfies the Eq. (2).
RIC-04 70 0.15 35 819.358 1.38 25 12.16
30 11.27
35 10.20 UCSRIC = 61.175(ER) + 82.477 (kPa) (1)
UCSProctor = 13.25(ER) + 189.99 (kPa) (2)
Based on Table 3 it can be known and stated that the tendency
of dry density relation is influenced by energy level of compaction From the Eq. (1) and Eq. (2) in as the Figure 6, there are three
effort and moisture content, at least this can be seen in Figure 4. The important parts are visible: first is for lower energy levels of RIC
characteristics of the relation are illustrated by the prediction line of energy than the proctor energy in this case showing that the value of
the optimum moisture content corresponding to the energy level of UCS obtained is also lower than the UCS of the proctor energy, then
the compaction effort given. The results show that the addition of the next section is on the energy level zone greater than the standard
energy tends to reduce the value of the optimum water content. energy of the proctor but still smaller than the proctor compaction
energy line in this section shows the level of attainment of UCS
values in general equal to the attainment of UCS at the proctor
compaction energy line to a point of intersection, while the last part
is the comparison of energy rapid impact compaction greater than
the proctor energy line, where in general the results of achievement
of its UCS value is also higher than the energy of compaction
proctor.
467
Based on these three conditions, it is possible to obtain an 6. REFERENCES

optimum compaction energy level equivalent to the optimal UCS
values corresponding to the intersection line energy proctor method Becker, P. J. (2011). Assessment of rapid impact compaction for
to the Rapid Impact Compaction energy line and the proctor transportation infrastructure applications, Graduate Theses
compaction energy line, i.e. a point where if applied greater energy and Dissertations. Paper 10261. Iowa State University.
ratio it will get greater compaction result from UCS of proctor Das, B. M. (2010). Principles of Geotechnical Engineering (7 th).
method, and otherwise so the turning point of comparison is as Cengage Learning 200 First Stamford Place, Suite 400
reference of optimum energy where in this research got optimum Stamford, CT 06902 USA.
point of energy ratio is equal to 2.25 . Falkner, F., Adam, C., Paulmichl, I., Adam, D., & Fürpass, J.
(2010). Rapid Impact Compaction for Middle-Deep
Improvement of the Ground – Numerical and Experimental
5. CONCLUSION Investigations. From Research to Design in European
Some conclusions that can be drawn from the results of this study Practice, (June), 2–11.
Koohsari, H., Alielahi, H., Najafi, A., & Adampira, M. (2016).
are:
Evaluation of factors affecting the estimated improvement
1. The tendency of the relationship of dry density is influenced by
depth of dynamic compaction using fuzzy method and PSO.
the energy levels of compaction effort and water content. The
Journal Soils and Foundations 2016, The Japanese
characteristics of relationship can be described with the line of Geotechnical Society. http://doi.org/10.1016/
prediction of the optimum moisture content corresponding with j.sandf.2016.08.012.
the energy level of a given compaction effort. Kristiansen, H., & Davies, M. (2004). Ground Improvement Using
Rapid Impact Compaction. 13th World Conference on
2. It is possible to make Relative Compaction relation which is the Earthquake Engineering, 13th(August 1-6), 496.
result of the achievement of UCS value using Rapid Impact Mohammed, M. M., Hashim, R., & Salman, A. F. (2010). Effective
Compaction method to UCS value of proctor method that improvement depth for ground treated with rapid impact
satisfies a linear line equation. compaction. Scientific Research and Essays, 5(18), 2686–
2693.
3. By connecting the energy comparison line of Rapid Impact Mohammed, M. M., Roslan, H., & Firas, S. (2013). Assessment of
Compaction to the energy line of the proctor method, it can be rapid impact compaction in ground improvement from in-situ
determined the amount of optimum energy that is equal to the testing. Journal of Central South University, 20, 786–790.
achievement of the UCS compaction value by the RIC method http://doi.org/10.1007/s11771-013-1549-0.
and the Proctor method. Rajasekhar, C., Naga Sai baba, A., & Kameswara Rao, M. (2016).
To Develop a Correlation Between CBR and Dynamic Cone
Penetration Value. International Journal For Technological
Research In Engineering, 4(1), 11–16.
468
Investigations of Soft Soil Stabilization by Mass Stabilization Methods for

Construction of Dike and Rural Roads
V. A. Phung1, V. V. Nguyen 2, V. T. Le1, T. Q. Do1
1
Center for Underground Construction, Hydraulic Construction Institute, Vietnam Academy for Water Resources
2
Vietnam Academy for Water Resources
E-mail: anphung.vtc@gmail.com , vietnguyenvu@vawr.org.vn2, levantuan112@gmail.com3, thequynh2711@gmail.com4
1
ABSTRACT: Nowadays, costs for soft soil improvement account for a large proportion of the total construction expenses. According to
statistics of construction cost in Viet Nam, these costs were about 20% - 35% of the total expenses. As a result, advanced technologies have
been applied to reduce the costs; one of them was mass stabilization technology. This paper introduces a study on soft soil stabilization using
mass stabilization method experimented via a full-scale physical model, and comparison between settlement results computed by analytical
formulas and the physical model.
Keywords: Mass stabilization, Deep mixing, Soft soil improvement, Binder, Dike
1. INTRODUCTION in recent years (Vietnam standard TCVN, 2012). In this standard,

the calculation method for embankment on soft soil used mass
There are several popular methods to create reinforcing mass in soft stabilization technology is described as follows (Nguyen et al.,
soil treatment, three of them are (Phung et al., 2015): (1) the deep 2014):
mixing method creating continuous mass; (2) digging the soil to mix
with binder, then spray down and compact; (3) Mass stabilization 2.1 Selection of strength and thickness of reinforced mass
using Allu’s equipment. In these methods, the third method has
2.1.1 Preliminary selection of strength of reinforced mass is not less
many advantages in accordance with conditions of Vietnam,
than qu according to the following formula:
especially in Cuu Long area. An Allu's equipment system consists of
three parts (Phung et al., 2015): power mixer (PM) – connected to
the excavator; pressure feeder (PF) to push the binder through the qu  Fs  w H (1.1)
tube into the ground; and the control system (DAC) - measurement, in which:
control, data supply during construction. qu = strength of reinforced mass (kN/m2);
w = wet weight of embankment (kN/m3);
H = height of embankment (m);
Fs = factor of safety, Fs=1,1
embankment
mass stabilization
T
Figure 1 Allu's equipment system
ALLU Systems uses compressed air to transfer dry binder from Figure 2 Diagram for settlement
the container into the soil throughout a tube that connected to power
2.1.2 Selection the appropriate thickness of reinforced mass (HT) in
mixer. Through the DAC device, it can control both the PF function
order to make the ground behaviour to be in the linear phase and
and adjust the volume of the binder pump to the ground. This helps
weak soil under the reinforced mass ensured load bearing capacity.
the soft soil improvement processing obtain the good quality and
Then the following conditions are described:
economical. To create the reinforcing mass, the blades cut soil
slowly from the top and mix with the binder. The Allu's technology
is especially effective in muddy areas because it does not need any P  Rtc (1.2)
additional solutions. Thus, this technology has the advantage of fast in which:
construction, economic efficiency, material, and energy savings. In P = load acting on the ground under reinforced mass (kN);
addition, this method is easy to compatible with surrounding Rtc = standard strength of soil below reinforced mass (kN),
structures and environment (with no settlement differences). was determined by the following formula:
Besides, because of using soil in place, it makes transportation costs
reduced as well as no vibration, no noisy and non-polluted Rtc  m A b  B H  Dc  (1.3)
environment during construction. The disadvantage is that the initial in which:
investment cost of equipment is quite high. m = coefficient of working condition, m=1;
BT = width of reinforced mass (m);
2. CACULATION METHODS FOR EMBANKMENT AND HT = depth of reinforced mass (m);
MASS STABILIZATION c = cohesion of soft soil (kN/m2);
For creating a legal basis to apply mass stabilization technology in  = the unit weight of the ground (=wđ if above
Vietnam, The Ministry of Agriculture and Rural Development groundwater level, and =đnđ if below groundwater
(MARD) has developed the based standard TCVN 04: 2014/VTC level);
based on the experience of applying deep-mixing technology in A, B, D = coefficients depend on the friction angle of the soil,
Vietnam (Phung et al., 2016) combined with several research results selected in the table 1.1.
469
The load acting on the reinforcing mass was defined as follows: in which :
lE = the length of circle sliding in embankment (m);
P   w H  qxe B (1.4) lI = the length of circle sliding in reinforced mass (m);
in which: lS = the length of circle sliding in soft soil (m);
RR = radius of circle sliding (m);
w = the weight of the embankment (kN/m3);
WE = the weight of the embankment (kN);
HT = depth of reinforced mass (m);
XE = the horizontal arm of the embankment;
qxe = live-load distribution of the vehicle on the embankment
(kN/m2); E = shear strength of the embankment (kN);
B = width at top of embankment (m). I = shear strength of weak ground (kN);
S = shear strength of soft soil (kN).
Table 1.1 Value of A, B, D coefficients
At present, the facto of safety can be quickly estimated by using
Friction Coefficients numerical simulations in Geotech such as Geo-slope, Plaxis, etc.
angle  (o) A B D 2.3 Calculation by the second limiting state
0 0 1,00 3,14
Total of settlement S were determined by settlement of reinforced
2 0,03 1,12 3,32 mass and the settlement of soft ground below the reinforced mass:
4 0,06 1,25 3,51
6 0,10 1,39 3,71 S  h1  h2 (1.6)
8 0,14 1,55 3,93 in which:
10 0,18 1,73 4,17 S = total settlement of reinforced mass (m);
12 0,23 1,94 4,42 h1 = settlement of reinforcement mass (m);
14 0,29 2,17 4,69 h2 = settlement of soft ground under stabilization mass (m).
16 0,36 2,43 5,00
18 0,43 2,72 5,31 load q (kN/m2)
2.2 Calculation by the first limiting state
T
2.2.1 Stability examination of circular sliding surface Mass stabilization
2.2.1.1 Embankment material: There are 3 cases for the
embankment, including: (1) Embankment was bulk material with a
friction angle > 30o. For this case, according to British Standard BS
s
8003, it was not essential to examine the overall stability of the
slide; (2) Embankment was cohesive soil, weak in-place that
embank in multiplied stages or one stage using geotextile. In this
case, as embankment met the stability requirement or geotextile met Figure 4 Calculation diagram for reinforcement mass
the loading capacity, no overall stability is required. However, it is
difficult to determine when it was enough and when it is not enough. h1 was determined as as follows:
Therefore, it was necessary to monitor the overall stability of the
slope in this case. qH T
h1  (1.7)
ET
2.2.1.2 Reinforced mass: This mass was considered as a
homogeneous, isotropic material. This mass directly transfers the
load from the embankment to the ground and played the most  wH  Q
q (1.8)
important role in stabilizing of the embankment. BT
in which:
2.2.1.3 Depending on the case, the circular sliding can be cut only q = distributed load on the reinforced mass (kN/m2);
on the embankment or cut through the reinforced mass. Stable Q = concentrated load on the reinforced mass (kN);
condition k  [k]. HT = depth of reinforced mass (m);
ET = elastic modulus of reinforced mass (kN/m2).
h2 was determined as follows:
CC  '  q'
h2  H S 1g 0 (1.9)
1  e0  '0
qBT (1.10)
q' 
H
BT  S
2
Figure 3 Calculation schema of overall stability in which:
h2 = calculated settlement of ground (m);
2.2.1.4 The overall sliding stability coefficient FS was defined by: q’ = loading on soft soil below reinforced mass (kN/m2);
Hs = the thickness of soft soil below reinforced mass (m);
RR (1E  E  1I  I  1S  S Cc = compression index of soft soil under pile tip;
Fs  (1.5) e0 = porosity ratio of soft soil under pile tip;
X EWE 0’ = effective stress (kN/m2).
470
3. EXPERIMENT TO DETERMINE SETTLEMENT BY Table 1.3 Shear resistance of the soil model
PHYSICAL MODEL
Shear resistance Su
3.1 Experimental purposes TT Depth (m) Remark
(kg/cm2)
The experiment purposes: (1) evaluate effectiveness of mass
stabilization method; (2) verify the results calculated by analytical 1 0,2 0,11 The results are
resolution. comparable to soil
2 0,4 0,09 tests at the sampling
site and in the
3.2 Site of Tests
3 0,6 0,10 laboratory
The experiments were built at Thanh Oai industry zone, near the Ba
La street, Ha Dong town, 10km from Ha Noi capital. This location
was selected because of its wide layout for carrying out the tests. Before carrying out the compression tests, the shear test was
carried out for monitoring and evaluation the shear resistance Su to
3.3 Test design compare between two experiments. If the Su was not suitable, the
moisture of soil could be adjusted. The location of shear test as
Digging soil to the depth of 40cm to build test models. The test shown in figure 5 and the test results were shown in the Table 1.3.
models were constructed by brick and cement M100. The thickness
of model wall was 35 cm and the height was 2.0 m. The container
dimensions were 1.0 x 8.0 x 2.0 m. Front of container was made by
glass thickness 15 mm and laid on steel frame system I20 by 5 mm
rubber bearings in thickness and 12 mm in width. Loading system
was placed inside the container. There were two I10 shaped steel
bars connected two sides of steel frame.
3.4 Treating soil in model

The model soil was the same as the soil in reinforced mass. Before
treatment, taking 12 samples for tests in the laboratory. The test
results were shown in Table 1.2. After treatment, the soil was placed
in 10cm thick into the model. At the same time, a layer of sawdust Figure 6 Soil in the model after processing
used for marking was also placed on the glass. Sawdust layers were
taken photos to mark the displacement of the soil during loading 3.5 Fabrication of reinforcing mass
(PIV technique).
The soil for making reinforcement mass and using in the laboratory
Table 1.2 Physical criteria of the experimental soil were taken in the ground, then transported to laboratory of
Hydraulic construction Institute (HyCI). Here, the soil was dried and
TT Parameters Unit Value pounded before return to the test model. Before processing on the
1 Natural Moisture (W) % 54,9 model to simulate the reinforcement mass, dried soil + cement +
2 Natural density ) g/cm3 1,6 roadcerm mixture were created based on experience [6]. Then water
3 Dry density c) g/cm3 1,04 was added to create a mass that water content was the same as
4 Density ( g/cm3 2,62 natural soil. The whole mixture was poured into a 0.5 m3 mixer box.
5 Degree of saturation (G) (%) 94,20
This mixture was put into the model to make the reinforcement
mass.
5 Porosity (n) eo 60,37
- Cement: Cement used in the test was PCB 40 with the content of
6 Porosity ratio 1,52 200 kg/m3.
7 Soil viscosity (Is) 0,82 - Additive roadcerm: 1% of cement volume.
8 Angle of interior friction ( Độ 500’
9 Cohesion (c) (kG/cm2) 0,09 3.6 The experiment equipment
10 Compressibility coefficient (a) (cm2/kG)
a0÷0,25 0,23
a0,25÷0,5 0,19
a0,5÷1,0 0,15
a1,0÷2,0 0,10
After each test, the model soil was re-processed as shown in

Figure 6. The soil was then consolidated for a period of 3 months
before carrying the tests.
Figure 7.1 Hydraulic jack Figure 7.2 Pressure sensor
33 803
brick wall M100# 33
steel I 10 cement-motar M75 thicked 1,5cm steel I 10 Laboratory equipment includes:
- Loading jack: To increase the loading load on the reinforced
33
33
33
150
mass at different levels based on the experimental design. The
33 33 jack could be loaded 20 tons. A gauge with the accuracy of 0.2
103
103
bullet-pfoof glass 15mm cutting position A ton was attached on the jack to read data that produced by China
270 as shown in Figure 7.1.
- Stress sensor (sensor): To measure the stress at the bottom of
33
33
33
33
45
45
steel I 10 steel frame I 20 steel I 10 reinforced mass in each load level. The sensor has a diameter of
45
33 227 260 45 227 33 28 mm, a thickness of 6.5 mm, could be measured stress in range
Figure 5 Shear test location
471
of 0 ÷ 0.2 MPa, made by China, as shown in Figure 7.2. The 3.8 Experiment design
layout of sensors S1, S2, S3, S4 and S5 were shown in Figure 8.
- The Data sensor (Data Taker DT80) was used to record the data After installing experiment model as shown in Figure 10, carrying
from a pressure transmitter then transfer to the computer, made out the experiment with load level in the Table 3 for two cases:
in Australia, which can be connected and recorded - case 1: compression soil without mass stabilization;
simultaneously with five sensors, as shown in Figure 7.3. - case 2: compression soil with mass stabilization.
- Specific Gravity Test: To measure the vertical displacement of
the compression table and the ground surface at different load
levels. The range of specific gravity test was up to 50 mm and
have a precision of 0.01 mm, was made in China and was
arranged in both symmetric sides through the centre, far 10 cm
from the edge of the table.
- Steel compression table was used to transfer the concentrated
load from the hydraulic jack to the foundation. This compression
table was made of steel, 1cm in thickness, 1m x 1m in
dimension, was reinforced at both side by I10 shaped I10 steel
bars to increase the hardness, as shown in Figure 7.4.
- Laptop: storing and analysing data from the reader.
- Digital Cameras: A digital camera with a projection light and a
cover to capture the displacement of points in the background.
Figure 10 Model installation
Table 3 Loading design

TT Load level Loading Time to maintain the
(ton) (ton/m2) load (minute)
1 0 0 0
2 0,4 0,40 30
Figure 7.3 Data reader Fig 7.4 Compressed steel 3 0,4 0,80 30
4 0,4 1,20 30
3.7 Standards used for experiments 5 0,4 1,60 30
Vietnamese standard TCVN 9354-2012. Soils - In situ test methods 6 0,4 2,00 30
for determination of deformation module by plate loading. 7 0,4 2,40 30
Japanese standard JSF: T25-80T. Experimental method for field 8 0,4 2,80 30
bench press for ground. 9 0,4 3,20 30
10 0,4 3,60 30
3.8 Experimental diagram 0,4 30
11 4,00
Layout of experimental equipment was arranged as below:
1m 3.9 Experimental results and discussion
steel plate thickness 1cm - Settlement results: Measuring the settlement at the end of each
P
loading stage and the final settlement for the case of nature soil
and mass stabilization. The results are shown in the table 4:
S2
0,3m
S1 S5,S3 S4
Table 4 Settlement measurement of two experimental cases
0,25 0,25
Single-point Average settlement of Average settlement
TT
0,5 0,5 load (T) foundation MS (mm) of ground (mm)
soft soil ground 1 0 0 0

Figure 8 Layout of stress measuring device
2 0,40 0,30 1,88
3 0,80 1,3 3,95
4 1,20 3,87 8,93
5 1,60 7,47 20
6 2,00 13,23 38,28
7 2,40 21,75 59,70
8 2,80 40,51 83,71
9 3,20 63,13 120,11
10 3,60 87,71 163,25
11 4,00 120,12 -
Figure 9 Devices measuring of installation settlement 12 4,20 133,53 -
472
The settlement results were graphed as follows: h1 was calculated according to Equation 1.7:
qH T
h1   0.36 (mm)
ET
in which:
q = distributed load on mass stabilization, q= 4,2 (T/m2);
HT = thickness of mass stabilization, HT=0,3 (m);
ET = elastic modulus of mass stabilization, was determined by
taking samples and conducting tests in the laboratory, ET=
3500(T/m2).
In case of ground was fully consolidated, h2 was calculated by
Equation 1.9 :
CC  ' q'
h2  H S 1g 0  144,08 (mm)
Figure 11 Stress-displacement relations of two experimental cases: 1  e0 0'
in which:
It is can be seen that: for natural ground with the intensity limit
q’ = loading on soft soil below mass stabilization, q’=2,27
of 3,6 (T/m2), the settlement was 163,25 mm, and for ground with
(T/m2);
30 cm thickness of mass stabilization (MS) and intensity limit of 4,2 Hs = thickness of soft soil below mass stabilization, Hs =1,7(m);
(T/m2), the settlement was 133,53 mm. Thus, with the mass Cc = compression index of soft soil under pile tip, Cc=0,29;
stabilization (MS), the bearing capacity of ground was increased by
e0 = porosity ratio of soft soil under pile tip, e0= 1,52;
15% and the settlement was decreased by 46.64% in compared with
0’ = effective stress, 0’ =0,51(T/m2).
those of natural ground.
The results showed that the error between the analytical formula
and the experiment was 7.56%. However, analytical formula was
applied for the case of full consolidation. In fact, the experiment
maintained the compression level at 30 minutes for each load level.
Therefore, the soft soil was under full consolidation. Hence, the
error between test results and analytical equation could be less than
7.56%.
4. CONCLUSIONS
Mass stabilization technique is an effective method for soft soil
improvement. Especially for works such as dikes (embankments),
roads in soft soil conditions and difficulties in finding construction
materials. Because of several advantages such as construction
without additional equipment, using in-place soft soil, and fast
construction speed. Mass stabilization technology is a potential
Figure 12 Displacement of ground without reinforcement that
method in Vietnam in general and the Mekong Delta in particular.
marked with sawdust
The calculation method was relatively simple, can use analytical
formulas as presented to design. In addition, it is possible to use
geotechnical software to calculate.
The results showed that the error between the analytical formula
and the experiment was less than 7.56%. This value was acceptable.
5. ACKNOWLEDGEMENT
The work presented in this article was supported by the Viet Nam
Ministry of Agriculture and Rural Development and following two
studies:
1. Research on solutions of soft soil treatment technology by
mixing in-place soil with inorganic binder for construction of
dikes in the Mekong River Delta.
2. Research on dike crack phenomenon and solutions
improvement to protect dike combined rural road.
Figure 13 Displacement vector field of the ground at the destructed

time was treated with PIV. 6. REFERENCES
Phung V. A. et al., (2010) "Research project on deep mixing method
- Comparison of settlement results for the case of mass
and ability applying for Mekong river delta". Report on the
stabilization with analytical formula:
studies of the Ministry of Agriculture and Rural
Overall settlement S of mass stabilization was determined by Development.
Equation 1.6: Basic standards TCCS 04:2014/VTC (2014) “Hydraulic
construction - Soft soil treatment by mass stabilization
technology - Design, construction and acceptance
S  h1  h2  114,45 mm requirements”.
473
Japan standards. Experimental method for field bench press for Allu and Rambol. Mass stabilisation manual. Allu Finland Company.
ground JSF: T25-80T. EuroSoilStab. Design Guide Soft Soil Stabilisation. CT97-
Vietnam standard TCVN 9354-2012. Construction soil - Method for 0351 Project No: BE 96-3177.
determining deformation modulus at field by flat plate. Le V. T., Yonglai Z., ShuXin D., Phung V. A., Ha T. L. (2015): “An
Phung V. A. et al., (2015) "Research on solutions for soft soil Numerical Analysis of the Influence of Replacement Area
treatment by soil mixing equipment with inorganic binder for Ratio in Foundation Reinforced with Soil Cement Columns”.
hydraulic construction". Report on the studies of the Ministry The Electronic Journal of Geotechnical Engineering ISSN
of Agriculture and Rural Development. 1809-3032. Vol.20.8. pp 3821-3828. 5.2015 (EI)
Phung V. A. (2014) “Research on solutions of soft soil treatment Phung Vinh An, Vu Ba Thao, Nguyen Quoc Dung (2016) “An
technology by mixing in-place soil with inorganic binder for Analytical Approach for Determining the Bearing Capacity of
construction of dykes in the Mekong River Delta”. Report on Soil Cement Column Using Jet Grouting Technology”.
Water Resources Science and Technology for the Mekong 19th Southeast Asian Geotechnical Conference & 2nd
Delta and the South East Provinces. AGSSEA Conference (19SEAGC & 2AGSSEA) Kuala
Nguyen Q. D., Phung V. A., Vu N. B. (2014) “Research on Ground Lumpur 31 May – 3 June 2016. pp 937-941.
Improvement in Ca Mau with cement and cement additives
combined”. Journal of Water Resources Science and
Technology 2009-2014, ISBN 978-604-67-0392-1.
474
Study for Structural Performance of Steel Sheet Piles Used for Retaining Wall
P.S. Eka1, M. Nobuyuki2, T. Tomoya3
1
PT Nippon Steel and Sumitomo Metal Indonesia, Indonesia
2
Nippon Steel & Sumitomo Metal Southeast Asia Pte.Ltd., Singapore
3 Nippon Steel & Sumitomo Metal Vietnam Co., Ltd., Ho Chi Minh City, Vietnam
E-mail: eka.2tb.susanto@id.nssmc.com1; matsui.c4d.nobuyuki@sg.nssmc.com2; tominaga.5xg.tomoya@vn.nssmc.com3
ABSTRACT: This paper describes structural performance of steel sheet pile walls based on laboratory and full-scaled field test. The steel
sheet piles are widely used as a permanent and temporary retaining wall. In order to realize reliable structure of the walls, the actual
performances should be evaluated properly in design especially for in case height of walls are deeper and close proximately to existing
structures. From this point of view, a reduction of sectional stiffness of U-type due to insufficient transmission of shear force in interlocks are
regulated in some of design standards. Depending on soil condition and restraint against slippage of interlocks, U-type should be considered
reduction factor for section modulus and moment of inertia in case of composite section. On the other hand, Hat-type sheet piles fully
perform the stiffness because interlocks are located on outermost surface where the shear force is negligible. The test results show the actual
stiffness performed according to sheet pile shapes.
Keywords: Steel sheet pile, retaining wall, stiffness
1. INTRODUCTION 2. SUMMARY OF SSP

The necessity of earth retaining works in Southeast Asian countries 2.1 Types of SSP
such as Indonesia, Singapore, and Malaysia, has been increasing
with the incremental of infrastructure projects such as subway train, In this paper, it is discussed about the sheet pile which name is Hat
port, bridge, and basements of building. The construction methods type steel sheet pile (hereafter, Hat type) as shown in the Figure 2
for these underground structures is required a high safety and and conventional U type steel sheet pile (hereafter, U type) as in the
reliability because accidents have a high possibility of causing fatal Figure 1. As conventional steel sheet pile which is developed in
results. At the same time, in parallel, the improvement of 1931, U type has 400 mm width. Hat is developed in 2005, and it
productivity of construction is also considered for enhancing has 900 mm width. These sheet pile can be used as both temporary
effectivity of economic growth. and permanent earth retaining wall. The interlocking joints of the U-
Conventionally reinforced concrete structure has been main type sheet pile is left-right symmetric so that it required connecting
material for deep excavation wall. It is reinforced by steel bar or H each SSP alternately reversed. On the other hand, the interlocking
beam so that high rigidity can be achieved and cast in place joints of Hat-type, with its 900mm width, has asymmetric shape so
construction also can be applied. Concrete wall is also the one of that it allows one-directional connection as shown in the Figure 3.
affordable and versatile choice so that it is used widely in the
construction.
Beside of concrete wall, Steel Sheet Pile (hereafter, SSP) is also
another choice for retaining wall. SSP is connected by integrated
joint clutches, it will construct a line of wall. SSP has thinner and
lighter section in comparison with concrete wall. The features gives
cost-effectiveness and rapid construction including easy handing at
the site. SSP also is reusable member after the settlement of
underground structure as a temporary structure. In addition, thanks
to tight joint interlocking, SSP is able to provide water tightness.
Earth retaining works can be classified into 2 usage such as for
temporary and permanent usage. As temporary structure, it is
applied in the underground structure to provide the soil stability
from excavation stage until completion of construction. As
permanent works, earth retaining works can be applied in quay wall Figure 1 U-type sheet pile
of seaport, basement wall.
When the SSP supports earth and water pressures, bending
moment and shear force is acting on sheet pile section. However, it
has been discussed that these forces causes slip of interlock
depending on its shape, which leads reduction of sectional
performance designated, such as section modulus and moment of
inertia. This can be referred as “lack of interlock integrity”, and the
degree of reduction is considered as the reduction factor for
sectional properties used in design. Therefore, when designing SSP
retaining wall, it is necessary to consider the reduction factor
properly to ensure the structural stability and safety.
In this paper, theory of reduction factor is proved. Then
conducting both structural experiment and numerical analysis,
degree of reduction factor were confirmed.
Figure 2 Hat-type sheet pile
475
H-section
Hat-type steel
Figure 3 Comparison between Hat and U type sheet pile
Fillet 900 mm
weld
Table 1 Sectional Properties of HAT Type Sheet Pile
Figure 4 Outline of the Hat+H pile
Type Dimension Property per meter of wall
Effective Effective Thick- Sectional Moment Section Unit
width height ness area of Inertia Modulus mass 2.2 Installation methods
2 4 3
mm mm mm cm cm cm kg/m
NS-SP-10H 900 230 10.8 122.2 10,500 902 96
Hat type can be installed by ordinary piling methods such as
vibratory driving method and the press-in method. The vibratory
NS-SP-25H 900 300 13.2 160.4 24,400 1,610 126
diving method is the major method to reduce ground friction by
NS-SP-45H 900 368 15.0 207.8 45,000 2,450 163 vibration as shown in Figure 5 This method can shorten the driving
NS-SP-50H 900 370 17.0 236.3 51,100 2,760 186 time and drive SSP into hard ground. This method is useful not only
driving but also pulling out the pile. The press-in method shown is
Hat type is applicable for both temporary and permanent Figure 6 is using hydraulic mechanism by grasping the middle place
retaining walls, and able to contribute to enhance structural safety of sheet pile while taking reaction by holding driven piles. Driving
and productivity in view of aspects below. machine is compact. It is applicable for the narrow and low
<Efficient pile-driving work> clearance construction site. This method provide low noise and low
Since Hat-type is more rigid than U type in terms of 1 pile, pile vibration of construction. For hard soil layer, additional method
deformation while the pile is driven into the ground is effectively such as water jet cutter or augering is applicable.
restrained, and hence even a longer pile can be driven in efficiently
<Higher structural reliability>
Since the pile joints are at the outermost part of the wall
structure, the neutral axis of each of the piles coincides with that of
the wall structure formed. Therefore, Hat type is able to display its
structural performance under various construction conditions.
<Applicable to deep excavation and high retaining wall >
Hat type is able to provide higher section modulus and moment
of inertia by combining with H shape by simple fillet welding as
shown in Figure 4. The fabrication is easily done at the site. Thanks
to high stiffness, it is possible to simplify additional supports such as
strut and tie rod, which leads to enhance construction productivity.
In addition, rapid installation is possible by using percussion
hammer or vibratory hammer, compared to existing piles i.e
concrete pile or steel pipe.
HAT+H is composite section, so moment of inertia is calculated
as follows.
I Is As ys2 IH AH yH2
I I /w (1)
I = moment of inertia of the Hat-type and H-shape per pile Figure 5 Vibro hammer (In Indonesia)
Is = moment of inertia of the Hat-type per pile
As = section area of the H-shape section
IH = moment of inertia of the H-shape section
AH = section area of the H-shape section
yS = distance from the neutral axis of the Hat-type and H-shape to
the neutral axis of the Hat-type per pile
yH = distance from the neutral axis of the Hat-type and H-shape to
the neutral axis of the Hat-shape
I’ = moment of inertia of the Hat-type and H-shape per pile wall
width
w = effective width of the Hat+H pile (900 mm)
In order to specify the intermittent fillet welding, it is necessary

to calculate and define the required welding ratio and leg length as
shown below. The shear and flexural resistances of the sheet pile is
considered in order to ensure the integrity of the Hat-pile and H-
shape. In general, the welding ratio should be from 40% to 60% and
the leg length should be from 6 mm to 8 mm. Figure 6 Silent piler (In Singapore)
476
3 STRUCTURAL PERFORMANCE OF SSP The Flexural tests for Hat+H were conducted in order to verify
3.1 Influence due to joint slippage due to shear force the sectional performance under bending force as well. The sheet
pile and H-shape were combined by fillet welding with a leg length
Steel sheet pile must be integrated with each other with its joint
of 6 mm. The flexural moment was generated by having vertical
interlocking for showing stiffness performance. When SSP resist the
soil and water pressure, bending moment and shear force occur. forces acting on the flange portion of the H-shape as shown in
These force cause joint slippage, so that SSP cannot perform actual Figures 9. The vertical load acted on the upper flange of the H-
nominal section properties i.e. section modulus and moment of shape. Figure 10 shows the relationship between the bending
inertia. This situation, so-called “lack of joint integrity” depends on moment and the curvature obtained by the loading test. The
the shape and joint location of SSP wall. The percentage of curvature,  was calculated by the measured displacements at the
reduction is designated as the reduction factor. For that reason,
loading point and center of the specimen. It shows that the bending
consideration of reduction factor in the SSP design is very important
to keep the structure safety and stability. In this paper, theory of moment of the Hat+H pile is almost the same as theoretical curve up
reduction factor between Hat and U type is proved through to the yield bending moment of the composite section. This large
structural experiment and numerical analysis and its reduction factor section did not cause local buckling before yielding. In addition, the
is confirmed. strain distribution in the section along the vertical direction kept the
As conventional SSP, when U type is subjected to bend due to plane, and it was not discontinuous like a built-up section as shown
soil pressure or other horizontal load, a large bending shear force in Figure 11. In addition, the strain curve is also equivalent to that of
occurs in the its interlocking joints because its interlocks are located
the calculated value of a composite wall. These results shows Hat+H
at the center of the wall, defined as neutral axis position. In these
cases, the shear force does not transmit sufficiently between the can be treated as a composite section by simple fillet welding for the
adjacent piles because the interlocking joints slip from each other, Hat-type and H-shape.
and its possessed sectional properties, the moment of inertia and
section modulus of the wall, are reduced by the reduction factor.
Hat as relatively new steel sheet pile has many advantage.
Beside of its 900mm width perform less number of piling than U
type for rapid construction, Hat also has been proved able to
perform 100% joint integrity as discussed in Eurocode. It means that
the actual nominal value of stiffness i.e section modulus and
moment inertia can be inputted without considering the reduction
factor due to the shape factor and site condition. On the other hand,
Hat-type sheet piles does not slip because interlock located on the
outermost side of the wall. So, Hat-type wall is no need to consider
reduction factor. Figure 7 illustrates interlocking joint behaviours
both U-type sheet piles and Hat-type sheet piles.
Figure 8 Bending resistance of HAT
Figure 9 Outline of the flexural test
Table 2 Specimens and sectional properties

Sectional Properties(Per 1m length)
H
Sheet Moment of Section
shape Weight
Pile Inertia Modulus
(H×B×tw×tf) (kg/m) 4
Figure 7 Theoretical behavior due to interlock integrity (cm /m) (cm3/m)
10H 400×200×9×12 169 88,074 2,320
3.1 Sectional Performance of Hat and Hat+H H : Depth of section, B : Width of section, tw : Web thickness tf : Flange
thickness.
In order to evaluate the bending resistance characteristic of the Hat-
type sheet pile, a single 10H pile and two 10H piles jointed side by
side were subjected to a bending test. The load-displacement curves
obtained from the test are shown in Figure 8. The vertical axis
represents the load converted in terms of the sheet pile width per
meter. Up to the yield load calculated from material test results, the
load-displacement relationships are almost the same as theoretically
estimated. In addition, despite the fact that the 10H pile is thinner
and has a large sectional area, it demonstrated sufficient plastic
deformation performance devoid of local buckling until the total
plasticization load was reached. Like the single 10H pile, the jointed
10H piles showed a similar load-displacement relationship in the
bending test. Thus, it is confirmed from bending test that just as with
the Hat-type sheet pile, the joint efficiency can be omitted from
consideration. Figure 10 Bending moment to curvature
477
Then, the shear force T in the interlock is:
eAM
T 
2I 1  y 0 eA 
(4)
Next, the stress w occurred at the web is:
M  2Ty 0
w   y1   2T (5)
I1 A
From (3) (4), the stress w is described:
Figure 11 Distribution of strain at allowable load

w  
M
 y1  e  (6)
I 1  y 0 eA
3.2 Estimation of the reduction factor
In the Figure 12, there is illustration about the sectional strain Thus, the moment of inertia I is:
distribution related the position of the neutral axis of the wall.
I  I 1  y 0 eA (7)
Neutralaxisofasinglepile
Neutralaxisofa wall
y1
Also, the section modulus zw at the edge of the sheet pile web is:
e
y y0
I 1  y 0 eA
zw  (8)
M y1  e
(a)full shear (b)zeroshear (c)partial shear
transfer condition transfer condition transfer condition
Furthermore, the stress t at the interlock is obtained from (1)
Figure 12 Illustration of relationship between the interlock slippage
(y= y0) and (3):
and section strain distribution
y1 = the distance between the edge of the web side and the neutral t  
M
 y0  e  (9)
I 1  y 0 eA
axis of a single sheet pile,
y0 = the distance between the neutral axis of a single sheet pile and
Thus, the section modulus zt at the interlock is:
the edge of the interlock side,
e = the distance between the neutral axis of a single sheet pile and
I 1  y 0 eA
the neutral axis of the wall. zt  (10)
y0  e
The actual sectional strain distribution depends on the
transmission of the shear force T in the interlock. In the full share In the full shear transfer condition, the neutral axis of the wall
transfer condition, the neutral axis of the wall (e=y0)(Figure 10.a). located the center of the wall (e = y0). Then, the stress w0, the
On the other hand, in the zero shear transfer condition, the neutral section modulus zw0 at the web and the moment of inertia is:
axis of the wall corresponded to that of a single sheet pile
(e=0)(Figure 10.b). When the partial shear force transfers in the
interlock, the neutral axis of the wall located within the region  w0  
M
 y1  y 0  (11)
I 1  Ay 0
2
bounded by the full and zero shear transfer conditions (0<e<y0)
(Figure 10.c).
The bending moment M is caused by the soil pressure acts on I 1  Ay 0
2
the wall, the stress σy at the distance y from the neutral axis of the z w0  (12)
single sheet pile is y1  y 0
y 
M  2Ty 0 
y
2T (2) I 0  I 1  Ay 0
2
(13)
I1 A
Thus, the factor accounting for the possible reduction of the
where I1 is the moment of inertia of a single sheet pile, A is the moment of inertia due to lack of shear force transmission in the
sectional area of a single sheet pile. interlocks D is defined as:
Since the stress y is zero at the center of the wall (y = e):
I I  y 0 eA
D   1
M  2Ty 0 
y
2T I 0 I 1  Ay 0 2
(14)
I1 A (3)
478
Regarding the factor accounting for the possible reduction of the 4. STRUTURAL TEST
section modulus due to lack of shear force transmission in the
4.1 Test purpose
interlocks B, there are two equations, i.e., w at the web and t at
the interlock: The purpose of this test is to verify the lack of interlock integrity of
U-type sheet pile by comparing its deformation in actual situation.
Zw y  y1 I 1  y 0 eA y 0  y1
w   0   D (14)
Z w0 y1  e I 1  Ay 0 2 y1  e 4.2 Test method
Zt y  y1 I 1  y 0 eA y 0  y1 Both of NS-SP-25H (hereafter, 25H) and IV is installed into the

t   0   D (15) ground to construct cofferdam as shown in the Figure 12 then it is
Z w0 y1  e I 1  Ay 0 2 y0  e excavated inside as shown in the Figure 13. After that the horizontal
displacement of each sheet pile wall is measured and compared
The section modulus is adapted on the side occurred maximum between its possessed stiffness and actual deformation.
stress, which is depends on the e:
y0  y1
e ; the maximum stress occurs at the edge of the web,
2
y  y1
e 0 ; the maximum stress occurs at the interlock.
2
In the figure 13, it is shown about the relationship between the
position of the actual neutral axis of the wall e and reduction factors
 in the case of NS-SP-IV (hereafter, IV). When the neutral axis of
the wall located the center of the wall (e = y0), the full shear transfer
condition ( = 0) is occurred. During the partial shear transfer
condition, the neutral axis of the wall shifts from the center of the
wall to the neutral axis as single sheet pile. The reduction factor for Figure 14 Construction cofferdam
moment of inertia D decreases linearly with decreasing of e. When
the neutral axis reaches the half of the effective sheet pile height (e
= (y0- y1)/2), the maximum stress side changes from the web side to
the interlock side. After the neutral axis exceeds the half of the
effective sheet pile height (e < (y0- y1)/2), the reduction factor for the
section modulus B decreases rapidly with changing from w to  t.
In this situation, the maximum stress at the interlock increases
rapidly.
3.3 Reduction factors stipulated in design codes Figure 15 Excavation inside the cofferdam
In the table 3, it is shown that the reduction factors stipulated in 4.3 SSP properties
design codes in Japan and Eurocode. In Japan, D as the reduction
factor for the moment of inertia is set as 0.45, with no welding or In the Table 3 it is shown about SSP type, length, and sectional
crimped at each sheet pile. In Eurocode, it is set as 0.30 to 0.40, with properties of this test. Both IV and 25H has same 12 m length with
cantilever, uncrimped, and favorable or unfavorable soil conditions. embedded 11.1 m into the ground. Inside excavation depth is 5.8m.
Possessed moment of inertia of IV is approx. 1.58 times higher than
its 25H.
Table 4 SSP properties
Sheet pile Embedded Excavatio Moment
Type length length n depth of inertia
(m) (m) (m) (cm4/m)
U-type NS-SP-IV 12.0 11.1 5.8 38,600
Hat-Type NS-SP-25H 12.0 11.1 5.8 24,400
4.4 Soil condition

In the Figure 14 it is shown about the relationship between the
embedded pile depth and the soil level and SPT-N value. Silt and
clay is exiting this test site.
Figure 13 Relationship between e and 
Table 3 Stipulated reduction factor in some design code

D B Remarks
Japan Road Association
Japan Design guideline of 0.45 0.6 Not weld
temporary structure
Singles or uncrimped
Eurocode-3 0.30- 0.40- doubles, no structural
Europe
Design of steel structures 0.40 0.60 support, unfavorable
soil condition Figure 16 Soil condition
479
4.5 Results 5. Numerical analysis

In the Figure 15 it is shown about the relationship between the 5.1 Finite Element Method (FEM)
excavated depth and the horizontal displacement of the SSP. The
displacement is obtained as the average value of three measuring FEM analysis by PLAXIS is carried out to estimate the lack of
methods. If a measured value is out of standard variation of three integrity of type IV. Firstly, the soil condition is adjusted as shown
measured values, the rest two values are used to calculate the in the Table 4 in order to make corresponding the displacement of
average. 25H by PLAXIS with that test value, because the stiffness of 25H
wall is clear under full shear transfer condition. Then, the stiffness
of IV wall is varied to adjust the PLAXIS result with the
displacement measured in the test. When the displacement of IV
obtained by PLAXIS matches the test value, this stiffness means the
actual stiffness of IV wall and the reduction factor could be
obtained.
Then the structure and its condition is modelled as in the Figure
17. The sheet pile is embedded into ground 11.1 m, and the
excavation width is 4m. The water level is set on G.L. -1.1 m
(H.W.L).
Figure 17 Relationship between the excavated depth and lateral

displacement
Before the excavated depth reached -2 m, the displacement of all

walls is small. After the excavated depth exceeded -2m depth, the
displacement increased relatively large. Finally, the excavated depth
reached -5.8m and it is shown that the displacement of the IV wall
shows maximum and the displacement of 25H wall is 45% smaller
Figure 18 Overview of PLAXIS excavation simulation model
than displacement of IV. It means that the actual wall stiffness
composed by 25H is higher than IV even though the nominal
The excavation protocol is illustrated in the Figure 18. After the
moment inertia of 25H is 37% lower than IV as shown in Table 3.
sheet piles are placed (Step2), excavated depth is deepened
Therefore, it is said that the IV wall is occurred the lack of interlock
gradually (Step3), finally reach to 5.8m (Step4).
integrity.
Table 7 Coefficients for δh* in medium dense sand

Shear strength Mod. of elasticity
Weathering Permeability Poisson’s
Thickness G.L Unit weight Undrained Drained Undrained Drained Ko
grade Ratio
cu c’ Φ’ Eu E’ k
m m kN/m3 kN/m2 kN/m2 deg MN/m2 MN/m2 m/s
0
1.5 Fill-1 17.0 0 0 37.0 14.0 11.7 1.0x 10-8 0.3 0.40
-1.5
-8
1.3 -2.8 Fill-2 17.0 0 0 41.4 56.0 46.7 1.0x 10 0.3 0.34
1.6 -4.4 Fill-3 17.0 0 0 34.7 28.0 23.3 1.0x 10-8 0.3 0.43
3 -7.4 Clay-1 15.2 45 0 0 60.0 50.0 1.0x 10-9 0.3 1.00
2.2 -9.6 Clay-2 16.1 38 0 0 60.0 50.0 1.0x 10-9 0.3 0.45
0.6 -10.2 Silt-1 19.6 0 0 36.5 56.0 46.7 1.0x 10-9 0.3 0.40
6.9 -17.1 Silt-2 19.6 0 0 32.5 16.8 14.0 1.0x 10-9 0.3 0.46
5.9 -23.0 Silt-3 19.6 0 0 39.5 173.6 145 1.0x 10-9 0.3 0.36
5.2 Results
As the results, in the Figure 19 it is shown about the deformation
view in the -5.8 m excavation depth from G.L. The deformation of
IV is larger than 25H. In the Figure 20 it is compared about
displacement from PLAXIS with the actual test data. In -5.8m from
G.L, the displacement simulated by PLAXIS is close to the test data
rested 16 hours. The moment of inertia of IV is adjusted 0.4 times
from its nominal value which means that the reduction factor for
moment inertia is 0.4.
Step1 Step2
480
6. CONCLUSION
Through the excavation test, the stiffness between Hat-type sheet
pile and U-type sheet pile is compared in actual situation. The
horizontal displacement of the sheet piles wall in increasing in
parallel with the incremental of excavation depth. This excavation
test verified that 25H (HAT) show higher stiffness than U type IV in
the same actual situation.
It is proved that U type IV sheet pile experience lack of joint
integrity because displacement of the U type IV wall is larger than
25H (HAT). This lack of joint integrity condition is stated in the
reduction factor for the moment inertia which is estimated 0.4 by
FEM analysis (PLAXIS); this value is approximately stay in range
of several standards value. For that reason, it is necessary to
consider the reduction factor for U type sheet pile to ensure the
Step3 Step4 structural stability and safety in steel sheet pile design stage. In other
Figure 19 Excavation protocol hand, it is not necessary to consider reduction factor for stiffness in
case of HAT type sheet pile design.
7. REFERENCES
Harata, N. (2008). Development of Hat-Type Sheet Pile 900,edited
by Tatsuta M, Kurosawa T, Nishiumi K, Taenaka
S,Wakatsuki T, Miura Y, and Eda K. Nippon Steel Technical
Report, Japan.
Otsushi K. et al. (2016) “Line-up Expansion of Hat-shaped Steel
Sheet Pile (NS-SP-45H, 50H)”, Nippon Steel & Sumitomo
Figure 20 Comparison the PLAXIS result and Test result Metal Technical Report, No.113, pp57-63.
Matsui et al. (2015) “Novel Compound Steel Sheet Pile for Earth
Retaining Works” The IES Journal Part A : Civil & Structural
Engineering
Nippon Steel & Sumitomo Metal Corporation. (2016). Product
brochure, STEEL SHEET PILES.
Japan Road Association. (1999) Design guideline of temporary
structure.
Eurocode3. (2007) Design of steel structures, Part5, Pilling
Susanto,Eka et al. (2017). “Study on Seismic Reinforcement
Method for Coastal Dyke Using Hat Steel Sheet Pile”.
HATTI (21th Annual National Conference on Geotechnical
Engineering Jakarta, 7-8 November 2017). Indonesia
Vietnam National Standard. (2013): TCVN 9860:2013 Steel sheet
Figure 21 Deformation view of the PLAXIS result pile structures in transportation engineering – Design
(Left: 25H, Right: IV) requirement.
481
Protection of the Existing Railway Tunnels from an Adjacent Deep Excavation

Jung-Feng Chang1, I-Chou Hu1, Hsiao-Chou Chao1, Chun-Seng Hsu1 and Huei-Ting Chang2
1
Department of Geotechnical Engineering & Geomatics, MAA Group Consulting Engineers, New Taipei City, Taiwan
2
North District Project Office, Department of Rapid Transit Systems, TCG, Taipei City, Taiwan
E-mail: jungfeng.chang@maaconsultants.com
ABSTRACT: With the booming of urban development, construction of new infrastructures faces challenges on rapidly disappearing space.
More and more underground constructions are inevitably conducted adjacent to existing buildings, and the associated safety concerns
become an important issue. This paper presents a particular case where a deep excavation is conducted in the proximity of operating public
transportation facilities. The underground construction is a joint-development for major substation of the Taipei MRT Circular Line, intercity
bus terminal, and commercial purposes. The close-by operating facilities include tunnels for Taipei MRT Blue Line, Taiwan Railway (TRA),
and Taiwan High Speed Rail (THSR) and none of them can be suspended during the excavation. The paper first briefs the environment
constraints. It then discusses the design considerations of protection measures against excavation-induced risks to the operation and safety of
the public transportation, including demolition of existing underground structure, trench protection of diaphragm wall (i.e., D-wall), strutting
and groundwater control, and monitoring. Conclusions are provided at the end of the paper
Keywords: Deep excavation, shield tunnels, underground structure, pumping, protection measures.
1. INTRODUCTION could thus be restricted within a relatively small region that would
make the distribution and change of groundwater easy to control.
The project is located at the north side of Banqiao Train Station in This paper first introduces the background the project, including the
New Taipei City. The site was an intercity bus terminal and is adjacent existing railway and MRT tunnels, and geological and
planned to merge with the adjacent lands to construct a joint- groundwater condition. It then illustrates the design consideration of
development building for the original bus terminal, main substation the protection measures for excavation. Preliminary monitoring
of the Taipei MRT Circular Line and commercial purposes. The results are presented followed by the concluding remarks.
scope of the project includes the construction of the building from
the basement to the fifth floor, above which will be constructed by
2. BACKGROUND OF THE PROJECT
the future developer. Since there are quite a few important
infrastructures around the site, least construction effect on these 2.1 Site Layout
facilities will be the first priority. Furthermore, the bus terminal
shall be relocated to not only remain its functions but also allow the To meet the requirement of the structure layout for the joint
project to proceed. development and the substation, the depth of the excavation is
The deep excavation of the building is a critical and key issue to designed to be approximately 33.3 m and 29.8 m, respectively.
the success of the project. The surrounding facilities include the Figure 1 shows that the site is a rectangular shape with about 157 m
tunnels of the MRT Blue Line, joint-tunnels of Taiwan Railway in the east-west direction, 27 m in the north-south, and the total area
(TRA) and Taiwan High Speed Rail (THSR), and three underpasses of about 3,750 m2. The basement of the building is a single wall
connecting the bus terminal and Banqiao Train Station. The system constructed using the bottom-up method. The D-wall is
excavation therefore is required not only to keep the facilities in adopted for both retaining and permanent structure purposes with
operation but also to remain the bolts on the top slab of the joint- 1.5 m thick and 52 m deep into the Gravels. Besides cross-wall and
tunnels that were pre-installed for the piers of the elevated Circular buttress, the inner strutting is adopted to reduce wall deformation
Line in a tolerant position. Further, since part of the piers of the during excavation. A ten- and nine-level strutting system is planned
elevated station that situates right next to the joint-development for the joint-development building and substation, respectively.
building is overlapped with the existing underpasses it is of 2.2 Overview of Adjacent Railway and MRT Tunnels
particular concern to the construction interaction between the
excavation and the foundation. In response to such complicated As depicted by Figure 1, on the west side of the site is the operating
construction constraints, the retaining structure adopted for the Blue Line of Taipei MRT; the south side is Banqiao Train Station
project include diaphragm wall (i.e., D-wall), cross-wall, and which is connected to the site through three underpasses; the east
buttress. In addition, deep guide wall and ground improvement were side is the operating tunnels for TRA and THSR; the north side is a
conducted at some regions to reduce ground disturbance by D-wall temporary bus terminal substituted for the one situated on the site.
excavation. Among the surrounding facilities, the tunnels for MRT, TRA, and
Since the joint-development building is constructed on the THSR are of most importance. There is literally no substitution of
existing bus terminal, conflicts of the new and existing structure transportation for the tunnels and protecting them from excavation-
would be inevitable. The D-wall of the building is especially induced effects is the only feasible way.
designed for both retaining and permanent use and thus its quality The up- and down-track of the Blue Line are constructed by two
shall be strictly controlled during the removal process of conflicted single shield tunnels that operate daily for as long as 18 hours and
structure and backfill. serve the trains as short as in every 3 minutes during the rush hours.
In addition, the building is seated on the Songshan Formation, a The outer diameter of the tunnels is 6.1 m with clearance of about 7
sand-clay interbedded alluvium that is underlain by the highly m and overburden of about 18 m within the region of the site. The
permeable Jingmei Gravels with abundant groundwater. In view that tunnels are aligned roughly in the south-north direction passing by
the project only covers the building up to the fifth floor, the control the site with the closest distance of about 6.4m on the southwest
of hydraulic pressure on the bottom of the excavation becomes a corner.
particular concern for both excavation stage and the complete stage On the other hand, the joint-tunnels for TRA and THSR are box
of the project before the developer finishes the upper stories of the structure with one and two story below ground surface, respectively.
building. Based on the deep excavation experience accumulated on Both tunnels adopted D-wall as retaining structure with 0.8-m-thick,
the Taipei Basin (e.g., Young et al., 1996; Huang et al., 1997; Chen 30-m-deep for TRA and 1-m-thick, 40-m-deep for THSR. The
et al., 1999; Moh, 2004), it was suggested except for base grouting tunnels pass by the site with the closest distance of about 6 m on the
to introduce cross-wall to panel the excavation site and dewatering southeast corner. There are 9 places on the top slab of the tunnels
482
THSRConference on Geotechnical Engineering. Jakarta - INDONESIA, 6 -7 November 2018. ISBN No. 978-602-17221-6-9
Tunnel Banqiao Train Station
TRA Foundation of MRT Elevated Station
Tunnel
Underpass
Existing Underground Structure
24m
27m
Joint-Development Building
& Major Substation
157m
Banqiao Temporary
Intercity Bus Terminal
Pre-Embedded Bolts for the Piers of MRT Blue Line

the Elevated MRT Circular Line Shield Tunnels
Figure 1 Schematic of project site and adjacent facilities
and concourse level for the Banqiao Train Station that were distribution with depths. It is especially noted that there is no sign of
embedded bolts for the future piers of the elevated Circular Line. hydraulic pressure in the Gravels that is below the hydro-static state
(e.g., Chao et al., 2015; Hu et al., 1996). The hydraulic pressure has
3. GROUND AND GROUNDWATER CONDITION been recovered in the Gravels after deep well pumping was banned
for domestic and industrial water use in late 1960s.
According to the results of geological investigations, the depth of
the Gravels on the site is about 50 m below ground surface. Except
4. PROTECTION MEASURES FOR CONSTRUCTION
for the fills, the Gravels are overlain by 5 alternating sand and clay
layers. It is identified as the sublayer I to V in the Songshan In the deep excavation where D-wall is adopted as the retaining
Formation but the sublayer VI is not clearly shown on the site (Hu et structure, the induced ground deformation generally originates from
al., 1996). The soil profile for the site is generally divided into 7 trench excavation of D-wall, base excavation and strutting, and
layers that from top to the bottom are: dewatering. The project is particularly challenging in view of the
(1) Fills – distributed from ground surface to 4.0 m below with existing underground structure and foundation for bus terminal and
an average thickness of about 4.3 m. underpasses on the site that shall be demolished during excavation.
(2) Silty Sand – distributed from 4.0 m to 11.0 m below ground In response to such environmental constraints, any possible
surface with about 7.3 m in average. construction risks were evaluated and the corresponding protection
(3) Silty Clay – distributed from 11.0 m to 20.0 m with a measures were planned to meet the regulated requirements. They are
significantly varying thickness and about 6.7 m in average. illustrated in the following sections and summarized in Table 2.
(4) Silty Sand – distributed from 20.0 m to 28.0 m in depths
with an average thickness of about 6.7 m. Table 2 Critical Issues and Strategies for Excavation
(5) Silty Clay – distributed from 28.0 to 36.0 m with about 6.8
m thick in average. (a) Removal of existing underground structure
(6) Silty Sand – distributed from 36.0 to 50.0 m with about 11.3 Existing Piled Wall  Removal of existing piled wall
m thick in average. and backfill with grouting
(7) Gravels – distributed from 50.0 m deep and below at least materials.
till the bottom of the borehole at the greatest depth of 59.1  Partly destruction of existing
m with the largest grain size of about 10 cm. Major
underground structure to
Substation build deep guide wall for D-
A simplified soil profile and its engineering properties adopted wall excavation
for excavation analyses are presented in Table 1.  Site excavation and demolish
Existing Underground the remaining underground
Structure
Table 1 Simplified Soil Profile and its Engineering Properties structure
Soil
SPT- (b) Retaining system
Layer Depth Classification t su '
N D-Wall
USCS
1 4.0 SF 8 19.7 - 28
2 11.0 SM 9 20.0 - 30  D-wall
3 20.0 CL 9 19.4 62 32  Cross-wall
4 28.0 SM 15 19.7 - 31.5 Cross-Wall  Buttress
Buttress
5 36.0 CL 13 19.5 115 33  Steel strutting
6 50.0 SM 32 20.0 - 32.5
7 > 50.0 GM 50 22.0 - 38
Note: Unit for depth, unit weight, undrained shear strength, and effective (c) Trench protection and construction monitoring during D-wall
friction angle is m, kN/m3, kN/m2, and degree, respectively. excavation
In addition, based on the recordings obtained from the
observation wells and piezometers during the investigation period
between May and August of year 2009, the groundwater level is
located at about 5 m below ground surface with a hydro-static
483
MRT Blue Line piled wall within the site and cavity was backfilled with grouting
Shield Tunnels
 Depth of micro-piles: 30m materials. For those in conflict with the new D-wall, the separation
 Low-pressure grouting: panels and retaining structure were constructed, respectively, within
depth 4 m ~ 20 m and outside the existing basement as the guide wall for the D-wall.
 Automatic monitoring: total They were demolished after the D-wall was complete.
Major Substation
station, reflector prism, beam
sensor, and convergence 4.2 Trench Protection During D-Wall Excavation
point for the shield tunnels of
Low-Pressure The project is in close proximity of the tunnels for MRT, TRA, and
Grouting for MRT Blue Line
Trench Protection Micro-Piles THSR with minimum clearance of about 6 m. Since D-wall was
used for the excavation of the TRA/THSR box tunnels, it is
estimated that the effect of ground deformation induced by trench
(d) Groundwater control
excavation on the site would be relatively slight on the box tunnels.
Pumping For the MRT’s shield tunnels that exhibit smaller volume and have
D-Wall
Strut
no retaining structure as a protection, the effect would be significant.
 Base grouting (existing Figure 2 shows that a series of low-pressure grouting with depths
Excavation Base basement floors must be varying from 4.0 m to 20.0 m was therefore conducted accompanied
penetrated in partial area) by 30-m-deep, steel-tube micro-piles to strengthen the soils outside
 Pumping within D-wall the trench. The ground deformation could be reduced as well as the
Buttress Cross-Wall
 Section dewatering in effect on the surrounding soils.
Base Grouting excavation site
4.3 Strutting and Groundwater Control During Excavation
Pumping Well Within D-Wall
Strutting and groundwater control are the two major issues for the
stability of excavation. As depicted by Figure 2, except for steel
4.1 Removal of Existing Underground Structure struts used in the retaining system the inner buttress and cross-wall
If the removal of existing underground structure is not fully are additionally introduced to the site to reduce the excavation-
complete, the remaining part would result in potential risks in induced wall deflection. The inner buttress is a pure concrete mass
excavation. It would create a passage for groundwater leakage with depths of 50 m on the side close to the MRT shield tunnels and
Banqiao Train Station
MRT Blue Line

Shield Tunnels
Low-Pressure
MRT Circular Line Grouting for Trench
Elevated Station Protection of D-Wall
11.5m 14.5m
Joint-Development Major
Building Substation
21m 33m 21m 37m Micro-Piles

(L=30m)
: D-Wall(2)(52m)
: Cross-Wall (50m; 33.3m/16.7m) : Low-Pressure Grouting for Trench Protection of D-Wall
: Cross-Wall (50m; 29.8m/20.2m) (Depths: 4m~20m)
: Buttress (40m) : Micro-Piles (L=30m)

: Buttress (50m) : 20 Pumping Wells into Gravels (L=90m)
(a) Plan
Low-Pressure
Grouting for
Trench Protection
SF Micro-Piles
SM (L=30m)
CL/ML
SM
Shield Tunnels for
CL Blue Line
Cross-Wall
4m D-Wall (L=52m)
SM
3m
5m 3m
2m
GM
Base Grouting
(b) Section
Figure 2 Schematic layout of protection measures for excavation
resulting in additional ground deformation around the site. 40 m for the rest of the site. The cross-wall is 50 m deep with pure
Therefore, the casing method was adopted to remove the existing and reinforced concrete for depths above and below excavation
484
levels, respectively. The cross-wall is used not only as part of mass in the Gravels. To effectively control the dewatering efficiency,
retaining system but also as panels for dewatering in the project. the site was panelled by six pumping area through cross-walls.
Since the groundwater level is about 5 m deep and the site is
excavated up to 33.3 m below ground surface, it was estimated that 4.4 Construction Monitoring
piping and uplift would be the issues against the stability of
excavation. Furthermore, the pumping output would increase as the To keep the site and adjacent structure in a safe condition over the
excavation depth closes to the underlain aquifer (i.e., the Gravels). It excavation period, a construction monitoring plan was implemented
is thus particularly crucial to maintain the stability on the excavation and focused on the significant factors. During the excavation, the
base. The base grouting was adopted to block the recharge of monitoring targets, instruments, and the associated frequency and
groundwater from the Gravels to the bottom of the excavation. The management levels involved are summarized in Table 3.
permeability of the improved soils is required to be below 1×10-5 Figure 3 shows the monitoring results from the inclinometer
cm/sec. Pumping wells were installed on the site to lower the installed in the D-wall close by the shield tunnels and in between the
groundwater table below the excavation level. Besides, additional buttresses. Due to the support of ground improvement, micro-piles,
pumping wells were installed within the D-wall and penetrated and barrettes, the resulting wall deflection was significantly reduced
down to 90 m deep with inlets opened at depths greater than 60 m to with a maximum of about 16.6 mm in comparison of the
control the hydraulic pressure on the bottom of the improved soil corresponding calculated free-field value of about 35 mm. The
Table 3 Summary of monitoring system

Management Level Recording
Target Instrument Label Frequency
Alert Action Type
Surface & set tlement point SM Once a week 32 mm 40 mm M anual
ground inclinometer & twice 34 mm 42 mm M anual
movement SIS during
excavation
Structure stand point SB Once a week 80% Action Note 3 M anual
movement & tiltmeter TI & twice level M anual
deformation crack meter CG during visible 1 mm (concrete) M anual
crack gauge excavation 3 mm M anual
CM
(brick/masonry)
Groundwater observation Twice a week Initial value Initial value M anual
OW
level & pore well & once a day ±1.0 m ±2.0 m
pressure stand-pipe during Notes 1 & 2 Alert level M anual
PS pumping
piezometer (within +1.0m (within
electronic excavation) excavation) Automatic
piezometer Initial value Initial value
ELP ±1.0 m ±1.0 m
(outside (outside
excavation) excavation)
Strut load strain gauge Twice a week 90% design load 125% design M anual
VG
& once a day load
D-wall stress rebar stress during 250 MPa 300 MPa M anual
transducer excavation (60 Grade) (60 Grade)
RS
170 MPa 200 MPa
(40 Grade) (40 Grade)
inclinometer Once a week 34 mm 42 mm M anual
& once a day
SID
during
excavation
Excavation heave indicator once a day 100 mm 150 mm M anual
base HI during
excavation
Tunnel total station & 10-30 min. 0.96/1,000 1.2/1,000 Automatic
RP
deformation reflector prism (9.6mm/10m) (12mm/10m)
convergence 10-30 min. 16 mm 20 mm Automatic
CB
point
horizontal 10-30 min. 0.96/1,000 1.2/1,000 Automatic
electronic beam EBS (9.6mm/10m) (12mm/10m)
sensor
Note: 1. Factor of Safety (FS) against piping and uplift shall not be smaller than 1.5 & 1.2, respectively; 2. FS against buoyancy
shall be 1.03 & 1.07 for temporary and permanent condition, respectively. 3. See table below.
Table 3a Settlement control criteria for buildings & structures

Current tilt Current tilt
Settlement 
Building Type Tilt  falling b/w 1/500 greater than
(mm)
& 1/300 (65%) 1/300 (40%)
Temporary 13/10,000
40 1/500 1/1,250
(timer or iron) (~1/769)
13/20,000
Brick 25 1/1,000 1/2,500
(~1/1,538)
13/10,000
RC (footing) 40 1/500 1/1,250
(~1/769)
13/10,000
RC (mat foundation) 45 1/500 1/1,250
(~1/769)
485
Deflection (mm)
0 10 20 30 40 50
0
EXC. 1 (2015.10.06)
5 EXC. 2 (2015.10.20)
EXC. 3 (2015.11.03)
10 EXC. 4 (2015.11.13)
EXC. 5 (2015.11.20)
15
EXC. 6 (2015.12.22)
EXC. 7 (2015.12.25)
20
EXC. 8 (2016.01.05)
EXC. 9 (2016.01.22)
25
dhmax =
Depth (m)
FINAL EXC. (2016.01.29)
35mm 30 d max=16.6mm
35
40
45
50
55
(a) Calculation (b) Measurement
Figure 3 Comparison of measured and calculated wall deflection closest to MRT tunnels
effect on the shield tunnels was therefore greatly reduced as well.
According to the readings of instruments mounted on the tunnels,
the deformation and differential deflection were remained within the 7. REFERENCES
management levels over the excavation period. Chao, H.C., Yang, G.R., Su, T.C., and Chen C.H., (2015) "Joing
Dewatering for MRT Deep Excavations – A Case Study".
5. CONCLUSION Sino-Geotechnics, 146, pp65-74.
In the early stage of design for the project, the deep excavation Chen, Y.W., Duann, S.W., Hwang, R.N, and Kong, S.K., (1999)
experience on the Taipei Basin was collected and studied. The "Influence of Excavation on Adjacent Piles", Proceedings of
protection measures, such as cross-wall, inner barrette, and base the 8th Conference on Current Researches in Geotechnical
grouting, were proposed based on the past experience, site Engineering, Pingtung, Taiwan, pp1425-1432.
characteristics, and environmental constraints. Up to the first half of Hu, I.C., Chin, C.T., and Liu, C.J., (1996) "Review of the
year 2018, the construction of basement is almost complete and the Geotechnical Characteristics of the Soil Deposits in Taipei".
effects of D-wall and base excavation were fairly controlled with no Sino-Geotechnics, 54, pp5-14.
interference to the tunnel operation. The relevant monitoring results Moh, Z.C., (2004) "Protection System to the Neighbouring
are still under summary and study. It is hoped to be shared in the Buildings in Taipei Metro", Proceedings of the Symposium
near future. on City and Underground Space and Underground
Construction, Beijing, China.
Wong, L.W., and Chin, C.T., (1997) "Groundwater Control for
6. ACKNOWLEDGEMENTS
Deep Excavation". Sino-Geotechnics, 63, pp15-22.
Permission for publishing this paper from Moh and Associates, Inc. Yang, G.R., Wong, L.W., and Wang, S.N., (1995) "Groundwater
and Department of Taipei Rapid Transit Systems of Taipei City Problems for Deep Excavations in Gravel Deposits in Taipei
Government is gratefully appreciated. Special thanks to Dr. R.N. Basin", Proceedings of the International Symposium on
Hwang and Mr. C.R. Chou who helped to prepare the manuscript. Underground Construction in Gravel Formations, Taipei,
Taiwan, pp4-21~4-28.
486
BORED PILE RETAINING WALL SOLUTIONS FOR EARTHQUAKE SLIP 6 AT

OHAU POINT, KAIKOURA, NEW ZEALAND
Senthuran Arulanantham1, Anthony Fairclough2, Sam Glue3and Jodi Edwards4
1
Senior Geotechnical Engineer, Stantec, Christchurch, New Zealand
2
South Island Geotechnical Manager, Tonkin & Taylor Ltd, Christchurch , New Zealand
3&4
Senior Geotechnical Engineer, Tonkin & Taylor Ltd, Christchurch, New Zealand
ABSTRACT: The 14 November 2016 Mw 7.8 Kaik�ura Earthquake triggered a number of landslips along the Kaik�ura coastline. Ohau
Point is located approximately 26 km north of Kaik� ura and was the site of massive landslide which destroyed the road and rail transport
corridors. This paper describes the bored pile retaining wall solutions that were developed as part of the North Canterbury Transport
Infrastructure Recovery (NCTIR) earthquake recovery program of works for the New Zealand Transport Agency (NZTA) and KiwiRail.
The subject site is named as NCTIR Site 6 and the subject bored pile retaining walls are named RTW 3.
Keywords: Bored pile retaining wall, Kaikoura earthquake and bored pile construction
1. INTRODUCTION The North Canterbury Transport Infrastructure Recovery

(NCTIR) is an alliance partnership between Fulton Hogan, Downer,
The Mw 7.8 Kaik�ura Earthquake occurred on 14 November 2016 at Higgins HEB Construction and the New Zealand Government. The
12.02 am. The earthquake initiated along a complete network of New Zealand Transport Agency and KiwiRail are board members
several existing faults with the rupture trending towards north for a and funders for the programme and represent the New Zealand
distance of approximately 170km. The faults which were ruptured Government on the NCTIR project. The objective of the NCTIR
during the 2016 Kaikoura earthquake included the Humps, Conway- program of works is to restore the State highway and rail transport
Chartwelll, Upper Kowhai, Hundalee, Fidget, Jordan, Kekerengu, networks which were damaged by the Kaik�ura earthquake and
Papatea, Hope and Needles faults. create a more resilient infrastructure along Inland route 70, State
Highway (SH1) and the Main North Line (MNL) railway.
Figure 1 shows the location of the faults which are currently
known to have ruptured during the 2016 Kaikoura earthquake. The major landslide which was located at the south side of Ohau
Point was named as Slip 6 by NCTIR. The Slip 6 landslide material
had a total volume of approximately 110,000 m3. Figure 2 shows the
Slip 6 landslide a few days after the Kaik�ura earthquake.
Figure 2-The Site 6 landslide a few days after the 14 November

2016 Earthquake
At Site 6 the NZTA carriageway is located on the seaward side

and the KiwiRail railway is located on the landward side of the
repair corridor. Figure 3 shows the location of NCTIR Site 6.
Figure 1-2016 Kaikoura Earthquake Fault rupture map.
The Kaikoura earthquake caused over NZD $ 1.2 billion of

damage to the public transport infrastructure along Inland route 70,
State Highway (SH 1) and main north line Railway (MNL). This
damage was primarily due to major landslides and slips above
and/or below the transport corridors. N
The MNL and SH 1 were closed between Clarence and NCTIR
Kaik�ura SH 1. This is the main transport link between the Picton
and Christchurch and the only rail link. The alternative transport
Site 6
link is via state highways 63, 6, 65 and 7 (alternate route). This
route is an alpine route and subjected to closure over winter months.
The route is also not designed to accommodate large volumes of
heavy traffic. A number of small communities were isolated and the
closures had significant impact on the regional economy.
Figure 3-Satellite image showing the location of NCTIR Site 6 -
Image from Google Earth Pro ©
487
KiwiRail specifications and BM3. Table 1 below summarises the

As part of the proposed repair works in the slip 6 area the road is target wall deflection, movement and settlement criteria which was
to be widened and a new geogrid reinforced soil and mass block adopted during the design of Site 6 RTW 3.
seawall constructed. New road and rail alignment is proposed for a
safe transport corridor. The new road and rail realignment at Site 6
also requires a new bored pile retaining system to be constructed Table 1: Summary of Geotechnical Target Design Criteria
between the new road and rail alignments in order to retain the
Parameter Limit
ground above the road and support the loads from rail infrastructure,
rail operations and a new rock fall protection bund. The proposed
wall enabled a reduction in height of an average of 3m over more Track Settlement (static) 20mm
than 100m. The retained height of this bored pile wall which is
named RTW 3, typically varies between 2.0m and 5.0m high. RTW
3 has been subdivided into eight sections and named RTW 3 A1 to Tracks to be checked & re-
RTW 3 D3. This subdivision was primarily made due to reasons of
construction progress and sequence. Figure 6 and 7, which are Track Settlement (Seismic) levelled after any significant
towards the end of this paper, present typical cross-sections through
the Slip 6 repair works seismic event
Figure 4 below provides an oblique drone photograph of the Maximum Wall Deflection
Site 6- RTW 3 location, which was taken during late 2017. The new 100mm
Site 6 seawall (under construction), old rail and road alignments are (ULS Seismic)
clearly visible in Photograph 4.
Minimum Slope Stability

Approximate Location of 1.50
Retaining Wall RTW3 (Wall is Tunnel 19
(FOS)-Static
located between road and rail
alignment
Minimum Slope Stability
1.10
(FOS)- Seismic
192m
123m
33m
3.1 Design Assumptions / Constraints
Shag Rock
The following special issues and constraints were duly considered
Figure 4-New Site 6 seawall & RTW 3 Bored Pile Retaining Wall and accommodated in the design for Site 6 RTW 3:
(under construction) 1. Construction programme, Kaik�ura and small
communities to the north had been isolated since the
It is worth noting that several new structures and mitigation Kaikoura earthquake severely impacting the wider region.
have been implemented at Site 6 including: a rock fall bund, debris The New Zealand Government and the affected
flow structures, rock fall mitigation catchment fences, sluicing of community wanted ‘full operation’ to be reinstated as
the landslide debris and blasting of the rocks. These works were quickly as possible.
designed by the NCTIR ‘Slope Team’ and were substantially
2. Limited space was available between the proposed new
completed before construction of the new structures and RTW3
commenced, to mitigate the risk of landslide and rockfall during rail and road alignment. This was one of the primary
construction. drivers for the selection of a bored pile solution.
3. The seaward side of Shag Rock is a nesting area for shags
and the proposed retaining structure and road alignment
2. SITE CROSS-SECTIONS needed to be built around the existing rock outcrop and
The Site 6 RTW 3 is divided into eight sub-sections, based on the have minimum to nil impact on this culturally significant
site geology, construction program and /or various retaining heights topographic feature.
and different loading cases on top of the walls. 4. Settlement and deformation of the rail infrastructure
behind the retaining structure (Rail track and rock fall
Figures 6 and 7 present typical present cross sections through bund) had to be maintained below acceptable limits as
the transport corridors, seawalls and RTW3 at Site 6 published by KiwiRail.
3.2 Design Life

3. DESIGN CRITERIA AND PHILOSPHY
In accordance with NZTA and KiwiRail requirements a 100 year
NZTA Bridge Manual V3 (BM3) Table 2.2 has been adopted as the design life was adopted for the design of Site 6 RTW 3.
design basis for the RTW 3 retaining structures. In accordance with
the recommendations outlined in BM3 Table 2.2, the proposed Site
6 RTW 3 works have been designed as an importance level 3
(1/2500 year earthquake event) structure as they are to support a
primary lifeline route (i.e. the MNL ).
, the RTW 3 retaining walls were designed to meet the

deflection and settlement criteria requirements of the relevant
488
4. LOADING CRITERIA
4.1 Static and Live Loads
In general accordance with the wider project design philosophy
4.1.1 Rail Loading and Rockfall Protection Structure (based on records and codes), a vertical seismic acceleration (PGAV)
(Tunnel Extension) value of 0.31g (approximately 40% of the horizontal PGAH value)
A rail live load of 90kPa was adopted for the design of this project was used in the pseudo-static models in conjunction with the
in accordance with the recommendations which are published in the horizontal acceleration.
KiwiRail Specification. A load of 100kPa has also been applied,
where appropriate, to account for the weight of the rockfall The design value for the vertical acceleration is in general
protection structure (tunnel extension) which is to be constructed at accordance with the recommendation published in Eurocode 8.
the southern approach to Tunnel 19. This allowance was selected in
collaboration with the designers of the tunnel extension structure. 5. GEOLOGY AND GROUND MODEL
4.1.2 Live Vehicle Loading 5.1 Ground Conditions
A live load of 30kPa has been adopted, where appropriate, during The road and rail corridor is located between the South Pacific
design of the retaining walls, to account for heavy construction Ocean and the foothills of the Kaik�ura mountain range. The
traffic and equipment. following sub-sections provide a general description of the
geological, hydro – geological and geotechnical conditions which
4.1.3 Rock Fall Protection Bund and Attenuation Energy are inferred to underlie the subject site.
Load 5.2 Geological Setting
The following load allowances have been made for the rock fall At the location of Site 6 the major geological unit is Pahau Terrane
protection bund and rock fall energy attenuation loads. These loads (Ktp). This unit is described in the relevant published geological
were provided by the designers of the Rock fall bund. map as well bedded and poorly bedded sandstone. These units are of
 Rock fall bund dead load = 55kPa. early cretaceous age and are approximately 145 million years old.
The site-specific geotechnical investigation results indicate a thin
 Rock fall energy attenuation energy load = 80kPa to
layer of beach gravels and/or colluvium overlies the sandstone
340kPa.
bedrock beneath most locations along the proposed Site 6 RTW 3
4.2 Seismic Loads footprint.
Immediately following the Mw Kaik�ura Earthquake, the damage at 5.3 Ground and Surface Water Conditions
Site 6 and to the existing road and rail infrastructure was observed to Fifteen machine drilled borehole investigations have been
be predominantly due to landslip debris impact that originated in undertaken within the general area around Site 6. The bedrock level
sandstone/mudstone materials above these assets. The peak beneath the Site 6 retaining walls varied between 11 m and 14 m
horizontal ground acceleration at the site during the Mw 7.8 below the existing ground surface level (bgl) and around 1m bgl in
earthquake of 14 November 2016 is inferred from data published by the Shag Rock area (refer to Figure 4 for approximate location). The
the University of Canterbury as being in the order of 0.50g. As material overlying the bedrock is generally described as colluvium
such, the site is considered to have been recently tested to between a comprising medium dense silty sandy gravel and occasionally Silt
1 in 500 and 1 in 1000 year return period seismic event. with some gravel and boulders. The ground water level was
In terms of AS/NZS1170.5 the site is assessed to be a Class B – generally observed to be sitting on the colluvium rock interface and
Rock site as it meets the following criteria: at a depth of between 11 and 14m bgl.
The subgrade material comprises rock with a compressive 5.3 Geotechnical Parameters
strength of between 1 and 50 MPa, and, an average shear wave
The geotechnical design parameters which were adopted during the
velocity greater than 360 m/s. Further, this site is underlain by
analysis of RTW 3 are summarised in Table 2 below.
materials having a compressive strength greater than 0.8 MPa and
shear wave velocity greater than 300 m/s as per Class B Rock site. Table 2: RTW 3 Geotechnical Design Parameters
Unit Friction Young’s

The design peak horizontal seismic loading value Unit Soil
Weight
Cohesion Tension
Angle Modulus
has therefore been estimated using the following equation: Name Model (kPa) (kPa)
(kg/m3) () (MPa)
Modified
Bedrock Hoek- 2548 GSI = 30, mi = 15,σ ci = 50MPa 300
(BM3 section 6.2.2) (1) Brown
Where: Mohr-
Colluvium 1800 7 0 35 50
Coulomb
= 1.00 for a Class B site, Concrete

Mohr-
2550 535 1795 35 23500
Coulomb
= 0.55 for the Kaik�ura area, and,

The colluvium is described in the borehole logs as a mix of silt,
Ru (1/2500) = 1.8 sand, gravel and boulder material and has uncorrected SPT N values
of between 4 and 50+. As such, this material was considered to be,
Using the above equation and assumptions, the following ULS on average, a medium dense soil with some cohesion.
seismic design values for the peak horizontal -acceleration were
calculated: These parameters were also verified by the back-analysis of
select slopes.
489
Table 3: Summary of the Site 6 RTW 3 configuration

Retaining Wall ID Maximum Approx Proposed load or Structure above Proposed retaining system
Retaining Wall the final wall
Height Length
RTW 3A1 3.0m 43.0m New rail alignment 0.8m diameter pile at 1.8m C/C spacing ,
12.0m long
RTW 3A2 2.0m 41.0m New rail alignment 0.8m diameter pile at 2.0m C/C spacing ,
9.0m long
RTW 3B (rock) 2.0m 20.0m New rail alignment 0.8m diameter pile at 2.0m C/C spacing ,
7.0m long
RTW 3B(colluvium) 68.0m 0.8m diameter pile at 2.0m C/C spacing ,
9.0m long
RTW 3C 4.0m 123.0m New rail alignment/ proposed 0.9m diameter pile at 1.5m C/C spacing ,
extension of tunnel/ rock fall bund 13.0m long
RTW 3D1 5.0m 20.0m New rail alignment 0.8m diameter pile at 0.85m C/C spacing,
12.0m long, with one row of Anchor at
1.7m spacing (32mm Anchor, 200kN pre
stress Load)
RTW 3D2 5.0m 26.0m Existing sloping ground 0.8m diameter pile at 0.85m C/C spacing,
8.0m to 12.0m long with Base Slab
RTW 3D3 5.0m 7.0m Existing sloping ground 0.8m diameter pile at 0.85m C/C spacing,
8.0m to 12.0m long with Base Slab
Existing rail location

Proposed
final rail Proposed Retaining Wall RTW 3B
location (The Primary subject of this paper)
New reinforced earth

and mass block facing
seawall
Figure 6: Typical Cross section through Site 6 RTW 3B
Existing and
Proposed
Final Rail Proposed Retaining Wall RTW 3C
maintenance
Location (The primary subject of this paper)
Access track
New reinforced earth

and mass block facing
seawall
Figure 7: Typical cross section through Site 6 RTW 3C
490
option as it was expected to meet all of the

6. DESGN OPTIONS performance criteria and have the highest level
of constructability.
6.1 Assessment of Options
The preliminary options for the Site 6 RTW 3 were assessed and the 6.2 Selection of Preferred Option
preferred solution selected based on due consideration of the
following key criteria: The conventional bored pile retaining wall option (Option 7 above)
 The likely performance of the retaining wall. was identified as the preferred option primarily because it was
 The constructability of the retaining walls within the judged to have the lowest level of construction risk, could be
timeframes stipulated by the New Zealand Government to constructed within the available corridor width, and could achieve
most if not all of the design requirements in particular rapid
open the road and rail transportation corridors. reconnection of this lifeline route post- earthquake. All of the other
 Consideration of the temporary live operation of the rail options were judged to have some technical fatal flaw.
on the old alignment and,
 Cost. A reinforced concrete bored pile wall with capping beam was
identified from the options assessment process as the best available
solution to retain the ground adjacent to the proposed road. The
The following retaining wall options were considered during the
ground between the piles was designed to activate an arching
options assessment phase of the Site 6 RTW 3 design. mechanism. Megaflow drainage and concrete shotcrete facing was
also provided between the piles to provide long-term durability,
Option 1 - No retaining walls and provide stable batters. control ground water pressures, and, ensure an appropriate level of
This option was found to be infeasible due to the in-service performance was obtained. For aesthetic reasons an
unavailability of space between the new road architectural facing is proposed to dress the exposed section of the
and rail alignment. wall, however, this architectural facing was not finalised at the time
of writing this paper. Due allowance for an architectural dressing
Option 2 - Cantilever pole retaining wall designs was made during the detailed design of the wall super structure.
comprising steel or timber poles. These wall
designs were found to be unsuitable for reasons 7. GEOTECHNICAL DESIGN METHODOLOGY
of durability, strength and/or deflection.
The geotechnical stability and structural demands on the proposed
Option 3 - Gabion basket retaining wall. As with Option 2, retaining walls were analysed using the specialized software
preliminary analysis indicated that this solution package FLAC version 8.00.448 and Mohr-Coulomb soil models.
was unlikely to achieve a satisfactory level of Figure 8 provides an example typical output from the FLAC Model,
performance with respect to deflection, in showing total wall displacements.
particular as insufficient space was available to
provide geogrid ‘tie backs’ and enable the The structural demands on wall RTW 3 which were obtained
fastest possible reinstatement of this critical from the FLAC analysis were cross checked using the geotechnical
lifeline route (i.e. similar issues to options 4 software package Wallap V6.06. For the purposes of the Wallap
below). assessment, the cantilever walls were assessed to act as a stiff
system and the anchored wall were assessed to act as a rigid system
Option 4 - Reinforced earth wall (soil nail or MSE). This
option was assessed to have significant Table 3 above summarises the key aspects of the final design
constructability issues as it required a significant which was developed for each of the eight sub sections of RTW 3.
excavation to be made behind the wall and the
placement of geogrids and engineered fill which
would have required the relocation or removal
of the rail line and ballast. This in turn had the
RTW3D1
potential to cause significant disruption to the RTW3C
rail service and delay reconnecting of this
critical lifeline route. RTW3B
Option 5 - Secant bored piles (SBP). This option was
assessed to be infeasible due to construction Shag Rock
constraints such as available cutters for
secondary piles and duration. RTW3D2& D3
Option 6 - Contiguous flight auger piles (CFA). The
primary advantage of CFA piles was the method RTW3A1 RTW3A2
does not require temporary casing to support the
pile holes. However, this option was assessed to Debris Flow Structure N6
be infeasible due to constructability issues in the
colluvium and rock materials. Figure 5: Aerial Photograph of complete portion of RTW 3 which is
located adjacent to Shag Rock and Debris flow Structure N6
Option 7 – Conventional bored pile wall with shotcrete
infill panels. This was selected as the preferred
491
 Construct bored pile – Provide a temporary casing to all

pile excavations which extends 1.0 m minimum above the
adjacent working platform surface.
 Construct capping beam.
 Complete excavation in front of the wall in 1.0 m deep
stages; Construct the wall lagging and drainage as
appropriate after each excavation stage.
 Construct handrails, drainage controls.
 Install architectural dressing to front of the wall.
Figure 8- FLAC Model

In the FLAC analysis, all the sections and their respective
construction sequences were modelled as per the construction
programme that was provided by the Delivery Team (Construction
crew). Usually such program was driven by the need to have a
temporary road and rail passage open as quickly as possible, with
further work undertaken at a later date to achieve the permanent
cross-sections and topographic profiles.
The structural demands such as bending moment, shear forces,

deformations of the wall and settlement of the road and rail
corridors were estimated by and inferred from the FLAC models. Figure 9- Bailey bridge installed above the bored pile retaining wall
The results from these models was also used to optimize the spacing to enable traffic to pass safely, during the RTW 3 and debris flow
of the bored piles, pile size and propping systems such as anchors
bridge construction
and concrete slabs The slope stability in both static and seismic
design cases also were analysed using the FLAC models and check
make for ensure the design criteria. 9.2 Bored Pile Retaining Wall with Ground Anchor
 Complete all slope stabilisation works as appropriate,
Site observations after the Kaik�ura earthquake sequence
above the RTW 3 construction site.
indicated the slopes above the site had experienced significant slope
failures and the existing seawall and slopes below the road surface  Construct temporarily bailey bridge crossings adjacent to
had generally performed very well. The new Site 6 RTW 3 structure RTW 3 (see figure 9)
has been designed to increase the geotechnical performance of the  Construct bored pile - Provide a temporary casing to all
slopes below the road surface in future earthquakes, and is assessed pile excavations which extends 1.0 m minimum above the
as unlikely to exhibit significant damage after a future ULS seismic adjacent working platform surface (see figure 10).
event.
 Pre-fabricate ground anchor unit and grout internal
An observational approach to the design of RTW 3 and the annulus over full bonded length.
adjacent seawall indicates that the proposed design will improve the  Drill anchor hole and install anchor assembly. (Grout was
geotechnical performance of Site 6 without trying to restrain the pumped in to the hole via grout tubes from the bottom and
whole slope, which would have been prohibitively expensive. i.e. pumped upwards, to prevent trapped air bubbles).
the road and rail platforms have been designed to act as a single  Grout internal annulus below ground level.
sliding block, and, is expected to exhibit high levels of seismic
performance.  Grout external annulus over full anchor length.
 Construct capping beam.
A risk of debris flow and / or rockfall damage due to failure of  Grout internal annulus with-in the capping beam.
slopes high above the road and rail platforms has not been addressed  Test anchors when anchor and capping beam grout and
in this paper. Mitigation of the risk which is associated with upslope concrete had achieved the specified minimum 28 day
rockfall and debris flows has been addressed by the NCTIR Slope
strength.
Hazards Team via the design of catch structures, division channels
and bridge structures.  Tension and lock off the anchors at the specified prestress
load.
 Complete excavation in front of the wall in 1.0 m deep
8. CONSTRUCTION SEQUENCES
stages; Construct the wall lagging and drainage as
The following construction sequences were developed and appropriate after each excavation stage.
recommended to maximise safety levels at all stages of the wall
 Construct handrails, drainage controls.
construction.
 Check anchor prestress loads, install anchor end caps and
8.1 Bored Pile Retaining Wall grease packs.
 Complete all slope stabilisation works as appropriate,  Install architectural dressing to front of the wall.
above the RTW 3 construction site.
492
11. CONCLUSION
NCTIR Site 6 was severely damaged by landslides originating
above the road surface during the November 2016 Kaik�ura
Earthquake.
The overarching project goal of reopening SH 1 and the MNL

coastal corridors as quickly and safely as possible was a key
consideration during the options assessments phase of this project
and ultimately contributed to the selection of the preferred solution
for RTW 3.
The bored pile option, although expensive compared to other

options was developed in order to open the road with minimum
disruptions to the existing road and rail networks. It was also
considered the best in terms of constructability and providing a
Figure 10-Bored pile construction (at RTW 3D1, an anchored robust durable solution in a coastal environment.
section of RTW 3
The proposed bored pile solutions create a stable transport
9. SAFETY IN DESIGN CONSIDERATIONS infrastructure for any future earthquake events.
The following features were considered and incorporated into the

design of RTW 3 to improve the safety of the structure during ACKNOWLEDGEMENTS
construction and for the end users. The authors would like to thank to North Canterbury Transport
 Bored piles were adopted, amongst other reasons to Infrastructure Recovery (NCTIR Alliance), the New Zealand
minimise the live rail disruptions and risk to rail Transport Agency (NZTA) and Kiwi Rail for providing their
operations once the temporary rail reconnection have been support and permission to write this paper.
made.
 A concrete nib was provided on the top of the capping REFERENCES
beam to prevent ballast from falling into SH1, and, to
AASHTO(2016)-LFRD Bridge Design Specifications, Seventh
reduce the risk of tool fall from KiwiRail maintenance Edition 2014 with 2016 Interim Revisions.
activities.
 The capping beam was designed to provide a safe working Civil Engineering Standard, Retaining Walls, Issue 1, 30-06-2016,
KiwiRail
platform for the KiwiRail maintenance workers
 A sturdy safety handrail / barrier was provided on the Eurocode 8-Part 1EN 1998-1:2004+A1:2013 and Part 5 EN 1998-
outside edge of the capping beam to provide fall 5:2004.
protection for future maintenance personnel. FLAC, Version 8.0.448, Copyright © 2016 Itasca Consulting
 A temporary steel casing was specified by the designers Group, Inc.
which extended 1.0 m minimum above the ground surface Fault Rupture Map-
to improve the safety of the construction workers and https://www.geonet.org.nz/news/6JadI02Pi8YUAOiiKUOgO2
avoid any pile hole collapse.
Google Earthpro
Kiwirail Civil standards for Retaining walls (C-ST-RW-4104) and
Ground Engineering and Earthworks(C-ST-GE-4105).
NCTIR-https://www.nzta.govt.nz/projects/kaikoura-earthquake-
response/
NZTA, ‘Bridge Manual (SP/M/022),’ third edition, Amendment 2,
May 2016.
Rattenbury, M.S.; Townsend, D.; Johnston, M.R. (compilers) 2006:
Geology of the Kaikoura area: scale 1:250,000 geological map.
Lower Hutt: GNS Science. Institute of Geological & Nuclear
Sciences 1:250,000 geological map 13. 70 p. + 1 folded map
Wallap, Version 6.06, Copyright © 2017 D.L.Borin Distributed by
Geosolve.
Figure 11-State Highway 1 opened after road debris flow bridge

construction, the debris flow channel construction underway.
493
Effects of Cross Wall to the System Stiffness of Deep Excavations in Clay

Zih Yun Wang1, Bin Chen Benson Hsiung2, Hsii-Sheng Hsieh3 and Louis Ge4
1
2
Department of Civil Engineering, National Kaohsiung University of Applied Science, Kaohsiung, Taiwan
3
Trinity Foundation Engineering Consultants, Co., Ltd., Taipei, Taiwan
4
E-mail: louisge@ntu.edu.tw
ABSTRACT: In urban area with soft clay deposit, it is required that the diaphragm wall deflection of a deep excavation be limited to
minimize damages to adjacent buildings. The auxiliary measures such as buttress wall, cross wall and ground improvement have been
applied to deep excavation. The so-called three-dimensional effect accounts for the presence of buttress wall and cross wall as well as the
size of the excavation area and depth. When considering the three-dimensional effect, the diaphragm wall deflection can be much reduced
compared with its plane strain condition. This study uses three-dimensional finite element analysis, by means of PLAXIS 3D, to predict the
wall deflection. The numerical results are compared with the outcome from a one-dimensional analysis and field measurements. In this study,
the effect of auxiliary measures is incorporated into the system stiffness of the excavation, where the presence of auxiliary measures is given
in form of increasing the undrained shear strength of the soil mass within the excavation zone. In turn, it leads to an increase of the factor of
safety against basal heave and a reduction of the wall deflection. One case history in soft clay are presented in this paper to evaluate the
effectiveness of the three-dimensional finite element analysis considering the effect of auxiliary measures.
Keywords: deep excavation, three-dimensional effect, wall deflection
1. INTRODUCTION elastoplastic foundation but is not capable to simulate the 3-

dimensional behavior of deep excavation that use auxiliary
Diaphragm wall is a retaining system to retain the pit, protect measurements. However, a simplified approach was developed in
adjacent buildings and prevents excessive ground movements during order to transform the effects of buttress wall and cross wall into
excavation. There are several methods of reducing wall deflections changing soil strength parameters based on the system stiffness,
in deep excavation, such as buttress wall and cross wall. In this which allows the designers to incorporate the complex effects of
study a three-dimensional finite element program, PLAXIS 3D, was buttress wall and cross wall in 1-dimensional excavation analysis.
used to simulate the wall deformation during construction for an Details of simplified approach are described by Hsieh and Lu
excavation with additional cross wall. The numerical results were (1999). The same approach was adopted in order to conduct
compared with the results from the one-dimensional beam-spring analyses using TORSA in this study.
model analysis and field measurements. The effects of cross wall are
incorporated into the system stiffness by increasing the undrained
shear strength of the soil mass within the excavation zone for beam-
spring model stated above. The purpose of this paper is to evaluate
the effectiveness of the real three-dimensional analysis and the one-
dimensional analysis with consideration of system stiffness for an
excavation with details stated above.
2. BACKGROUND
2.2 System stiffness
Clough et al. (1989) first presented the relationship between wall
displacement, system stiffness and factor of safety against basal Figure 1 Relationship between maximum wall movement and
heave (Fb). As shown in Figure 1, wall displacement is controlled by system stiffness (after Clough et al. 1989)
system stiffness and Fb. If the system stiffness is constant, as the
value of Fb increases, the maximum wall displacement would 2.4 PLAXIS 3D program
significantly reduce. The factor of safety against basal heave, Fb, is
calculated based on the equations defined by Terzaghi (1967). The PLAXIS 3D is a finite element program intended for three-
Y-axis in Figure 1 is the ratio of the maximum wall displacement dimensional analysis of deformation and stability in geotechnical
(δhm) to the excavation depth (He), while the X-axis is the system engineering. In three-dimensional model, it can consider three-
stiffness (S) defined by Clough et al. (1989) as follows. dimensional effect, like buttress wall and cross wall. A 3D FEM
model of excavation site was used and the Mohr-Coulomb soil
EI (1) model was applied in analysis. The length, width and depth of the
S model is 120 m, 120 m and 59.9 m, respectively, as presented in
rw xh 4 avg
Figure 2.
Where EI is the flexural stiffness of the retaining wall, rw is the 3. CASE HISTORY
unit weight of water, and havg is the average vertical spacing of
bracing system. It could be seen from Figure 1 that induced 3.1 Site description
maximum wall movement could be interpreted by excavation depth, Case A is a 17-story office building with 4 basement levels. The
S and Fb. Further, having the same S and excavation depth, the project site is about 40 m in length 38 m in width and 59.9 in depth,
greater Fb leads to smaller wall movement. and the excavation depth of the basement is 17.1 m. Diaphragm wall
is 0.8 m in thickness and 34 m in depth together with 5 levels of H-
2.3 TORSA program steel bracing as the retaining system. The typical horizontal spacing
of H-steel bracing is 6 m, and each level of bracing was preloaded to
TORSA is a one-dimensional numerical program which considers
the behavior of retaining wall as a beam resting upon an 50% of its allowable axial capacity. The depth of the cross wall
494
extends from GL.-16 m to GL.-34 m. There are six inclinometer maximum wall displacement were G.L.-19.04 m and G.L.-1.63 m.
casings installed in the perimeter diaphragm wall as shown in Figure In other words, the wall behavior at SID-1 and SID-4 were similar,
3. The ground water level was at 2 m below surface level. and SID-3 was coincided with SID-5. As shown in Figure 3, both of
SID-1 and SID-4 were installed in the wall which attached to cross
wall. This might be the reason that the induced displacement could
become very small but unlikely to move outward at final excavation
stage. In contrast, the location of SID-3 and SID-5 were not close to
the cross wall so the wall displacements were larger than SID-1 and
SID-4.
Figure 2 The model used for 3D FEM analyses
3.2 Wall displacement of field observation

The wall displacement measured from 4 selected inclinometers,
SID-1, SID-3, SID-4 and SID-5 are used for further evaluation and
locations of these inclinometers are shown in Figure 3. The depth of
toe of these inclinometers are installed at the same depth of wall toe.
Unfortunately, only data at final excavation stage are available for
comparisons and details are drawn in Figure 4. It is realized that
SID-5 had the largest wall displacement, was 7.762 mm, and the
depth of maximum wall displacement was at G.L.-10 m to G.L.-11
m. The maximum wall displacement of SID-3 was about 4.180 mm,
Figure 3 Layout of diaphragm wall and cross wall for Case
and the depth of maximum wall displacement was at G.L.-10.65 m.
Additionally, the wall moves outward instead of inward at SID-1
and SID-4. The maximum wall displacement of SID-1 and SID-4
were -3.433 mm and -5.056 mm, respectively, and the depth of
Table 1 Material properties of the soil layers

Depth from Depth to SPT-N E
Layer Type
m m
1 CL 0 -3.5 18.7 5 54 0 0 27000
2 SM -3.5 -9.3 20 8 0 31 0 16000
3 CL -9.3 -24.2 18.6 3 32.4 27 0 16200
4 CL -24.2 -30.4 18.6 6 74.5 30 0 37250
5 SM -30.4 -32.8 18.7 9 0 31 0 18000
6 CL -32.8 -45.4 18.3 10 98.1 31 0 49050
7 CL -45.4 -48.4 18.2 15 107.9 0 0 53950
8 CL -48.4 -56.1 19.1 19 135.4 0 0 67700
9 GM -56.1 -59.9 21.6 >100 0 37 0 912000
Table 2 Bottom-up excavation sequence of Case A

Phase Activity Remarks
1 Excavation to GL.-1.7 m First exc. Stage
2 Install strut at GL. -1.0 m H300*300*10*15 mm Preload: 490kN/ea
3 Excavation to GL.-4.8 m Second exc. Stage
4 Install strut at GL. -4.1 m H400*400*13*21 mm Preload: 980kN/ea
5 Excavation to GL.-8.0 m Third exc. Stage
6 Install strut at GL. -7.3 m 2H400*408*21*21 mm Preload: 980kN/ea
7 Excavation to GL.-11.2 m Fourth exc. Stage
9 Excavation to GL.-14.4 m Fifth exc. Stage
11 Excavation to GL.-17.1 m Final exc. Stage
12 Cast B4FL and FS at GL.-14.4 m and GL.-17.1 m
13 Remove the fifth floor support
14 Cast B3FL at GL.-11.2 m (t=40 cm)
15 Remove the fourth floor support
17 Remove the third floor support
19 Remove the first and second floor support
20 Cast 1FL at GL.0 m (t=25 cm)
495
Table 3 Material properties of the diaphragm walls, cross walls and basement floors
Diaphragm Cross Cross
Parameters Unit F1 B1F B4F FS
Wall Wall (a) Wall (b)
m 0.8 0.8 0.8 0.25 0.4 0.2 0.6
5.65 5.65 5.65 0 0 0 0
2.75E+07 1.37E+07 2.40E+07 2.75E+07 2.75E+07 2.75E+07 2.75E+07
2.46E+07 1.74E+07 2.30E+07 2.46E+07 2.46E+07 2.46E+07 2.45E+07
0.2 0.2 0.2 0.2 0.2 0.2 0.2
Table 4 Material properties of the anchors stress analysis, as the clay layers are under undrained type C of total
stress analysis due to limitation of available data from site
Parameters Unit H300x300 H400x400 2H400x408
investigation. As shown in Table 1, the effective Young’s modulus
-7.3, -10.5, of sand layers was determined by the following equation after
Depth m -1 -4.1
-13.7 Hsiung (2009).
E kgf/cm2 2.04E+06 2.04E+06 2.04E+06
cm2 E  2000 N (kPa) (2)
A 118.4 218.7 501.4
EA kN 2.37E+06 4.38E+06 1.00E+07 where N is the number of standard penetration teat (SPT). The
kN undrained Young’s modulus of clay layers was obtained by the
Preload 490 980 980 following empirical equation as reported by Bowles (1996), Lim et
al. (2010), Likitlersuang et al. (2013), Khoiri and Ou (2013).
E u  500 su (kPa ) (3)
where su is undrained shear strength of clay. The ground profile

and cross section of the excavation are presented in Figure 5. As
indicated in Figure 5, the excavation has to be delivered mostly in
clay, very occasionally in thin layer of sand. Further, as indicated in
Table 1, SPT-N of clay above final excavation level is less than 10
which is comparatively soft. By the same reason, additional cross
wall is thus considered in order to restrain the wall movements
induced. Parameter of structures, such as walls, floors and anchors
are shown in Table 3 and Table 4. The excavation sequence of Case
A is shown in Table 2, including the sizes and preloads of horizontal
struts.
Figure 4 Wall displacements of field observation

Figure 5 Soil stratigraphy and construction sequence for Case A
4. RESULTS OF CASE HISTORY
4.1 PLAXIS 3D numerical model and material properties 4.2 Results of PLAXIS 3D numerical analysis
Details of dimension of analytical model used are described The results of wall displacement of real 3D numerical analysis at
previously and the excavation is approximately 40 m by 38 m. Soil final excavation stage at various locations are shown in Figure 6. It
properties of each layer are listed in Table 1 and elastic-perfect is realized that the wall displacement of above excavation level is
plastic Mohr-Coulomb model is chosen for ground behavior larger than the wall displacement of below excavation surface. This
simulation. The sand layers are under drained condition of effective might be connected with depth of cross wall as the cross wall was
496
constructed from one meter above final excavation level down to the
same depth of the wall toe which expects to limit wall displacement
at same depth. In general, the maximum wall displacement is less
than 20 mm at final excavation stage which is insignificant.
Additional corner effect formed by construction of cross wall also
reduces the overall wall displacement, no matter the wall is attached
by the cross wall directly or not.
Figure 7 Compare the wall displacement of PLAXIS 3D and

field observation
4.4 Compare with the results of TORSA numerical analysis

As stated above, due to limit of time and budget, one-dimensional
analyses are widely adopted instead of real 3D FE analyses. That is
the reason that a simplified approach has to be developed in case 3-
dimensional characteristic of the excavation, such as cross-wall has
Figure 6 Wall displacements of PLAXIS 3D analysis at final
to be simulated by one- dimensional analytical software.
excavation stage
Comparisons of analytical results of PLAXIS 3D and TORSA
together with field observations at final excavation are presented in
4.3 Compare with field observation
Figure 7. It can be seen that PLAXIS 3D performs much better
Wall displacement on each side of the site were analyzed by rather than TORSA does. Because TORSA, one-dimensional
PLAXIS 3D and compared with field measurements, as shown in numerical program, has a limitation in modeling three-dimensional
Figure 7. It is found that the results of PLAXIS 3D are slightly behavior of cross wall, though a simplified approach has been
higher than field observation but the trend of entire wall recommended.
displacement are close. Especially, the SID-1 is in a good agreement It is also seen from Figure 8 that prediction of TORSA shows the
with field data. As SID-1 is located in the wall attached to the cross wall should move outward instead of inward during the excavation
wall directly, the displacement is small and likely the inclinometer which is not very reasonable. It is thus suggested that there is still a
toe is fully fixed and it might be the reason that both observational space to discuss how to run TORSA properly if additional cross wall
data and analytical results can be consistent. Also shown in Figure has to be adopted in the excavation.
7, predictions of SID-3 and SID-5 indicate that wall/inclinometer toe The maximum wall displacements of both analytical results and
shall have certain movements but toe is assumed to be fully fixed for observations are shown in Table 5. The TORSA simulated results
field measurement. The field measurement may thus be are overestimated, as PLAXIS 3D results are much closer to the
underestimated due to toe movement and this might be the reason to field data, while the observed maximum wall displacements are less
lead the differences between analytical results and field than 10 mm. The ratio of the maximum wall movement to the
observations. excavation depth can vary from 0.05% (field measurement) to
0.15% (TORSA analyses).
497
Figure 9 Wall displacement of SID-1
Figure 8 Comparisons of wall displacements from field

measurements and analyses
Table 5 Comparisons of analytical results with field observation
Max. wall Max. wall

Program displacement displacement/excavation depth
mm %
Field data 7.76119 0.045387
TORSA 25.1685 0.147184
PLAXIS
17.77 0.103918
3D
5. DISCUSSIONS
As described before, only data at final excavation stage are
available, it is thus difficult to have a full discussion for an
excavation in clay with additional cross wall. Through conclusions
in Section 4.4, it is agreed that simulation using TORSA still has a
room to be improved. Therefore, stage by stage deformations from
real 3D analyses are selected for further discussions herein. The wall
deformations of PLAXIS 3D analysis for each excavation stage are
shown in Figure 9, 10 and 11, where the dash line is the depth of
excavation surface. As further excavated, the wall deformation
would become larger. The results of wall deformation of SID-3 and
SID-5 are similar, with maximum wall displacements in the vicinity
of 15 mm. However, it is obvious that the results of SID-1 are
smaller than SID-3 and SID-5, and the maximum wall displacement
is 4 mm. It might be attributed to an apparent corner effect at the
location of SID-1. Figure 10 Wall displacement of SID-3
498
(4) It is likely that a real 3D FE analyses can provide a reasonable

prediction of wall displacements for an excavation with cross
wall. The difference between observational data and analytical
results might be caused by inclinometer toe movement.
(5) Even though a simplified approach is provided, it still has a
room to be improved if one-dimensional analytical software
TORSA has to be adopted to simulate wall displacement for an
excavation with cross wall.
7. REFERENCE
Terzaghi, K. (1967). Theoretical Soil Mechanics, John Wiley &
Sons, New York.
Clough, G.W., Smith, E.M., and Sweeney, B.P. (1989). "Movement
control of excavation support systems by iterative design".
Proceedings, ASCE Foundation Engineering: Current
Principles and Practices, 2, 869-884.
Ou, C.Y., Chiou, D.C., and Wu, T.S. (1996). "Three-dimensional
finite element analysis of deep excavations". Journal of
Geotechnical Engineering, ASCE, 122(5), 337-345.
Finno, R.J., Blackburn, J.T., and Roboski, J.F. (2007). "Three-
dimensional effects for supported excavations in clay".
Journal of Geotechnical and Geoenvironmental Engineering,
ASCE, 133(1), 30-36.
Hsieh, H.S. and Lu, F.C. (1999). "A note on the analysis and design
of diaphragm wall with buttresses". Sino-Geotechnics, 76, 39-
50 (in Chinese).
Hsieh, H.S. and Huang Y.H., Hsu, W.T. and Ge, L. (2017). "On the
system stiffness of deep excavation in soft clay". Journal of
GeoEngineering, Vol. 12, No. 1, pp. 21-34.
Hsiung, B.C. (2009). "A case study on the behavior of a deep
excavation in sand". Computers and Geotechnics, 36, pp.
665-675
Figure 11 Wall displacement of SID-5 Bowles, J.E. (1996). Foundation analysis and design, 5th Edition,
McGraw-Hill Book Company, New York, USA.
6. CONCLUSIONS Khoiri, M. and Ou, C.Y. (2013). "Evaluation of deformation
Through this study, following conclusions are made. parameter for deep excavation in sand through case histories".
(1) System stiffness (S) in excavation is an index which includes Computers and Geotechnics, Vol. 47, pp.57-67.
flexural stiffness of the retaining wall, the unit weight of water, Lim, A., Ou, C.Y., and Hsieh, P.G. (2010). "Evaluation of clay
and the average vertical spacing of bracing system. constitutive models for analysis of deep excavation under
(2) By considering S, safety factor against basal heave (Fb) and undrained conditions". Journal of GeoEngineering, TGS, Vol.
excavation depth, the maximum wall displacement could be 5, No. 1, 9-20.
predicted. In addition, having the same S and excavation depth, Likitlersuang, S., Surarak, C., Wanatowski, D., Oh, E., and
larger Fb should lead to smaller displacement. Balasubramaniam, A. (2013). "Finite element analysis of a
(3) It is aware that wall displacement could be much smaller at the deep excavation: a case study from the Bangkok MRT". Soil
place once cross wall directly attached to it. However, the depth and Foundations, Vol. 53, No. 5, pp. 756-773.
and height of cross wall also affect the magnitude and shape of
wall displacement. The wall displacement might still be
generated if there is no cross wall at said depth.
499
Ground Surface Settlement Induced by Diaphragm and Buttress Walls Installation:

Numerical Study
Aswin Lim1 and Pio-Go. Hsieh2
1
Department of Civil Engineering, Universitas Katolik Parahyangan, Bandung, Indonesia
2
Department of Assets and Property Management, Hwa Hsia University of Technology, New Taipei City, Taiwan
E-mail: spg@cc.hwh.edu.tw
ABSTRACT: In construction practices, diaphragm walls are a cast-in-situ reinforced concrete retaining wall that are constructed using a
slurry supported trench method. The installation process includes slurry supported trench excavation, placing the reinforcement cage,
concrete casting and curing. This installation process would modify the in-situ stress state in the soil close to the trench and generate ground
surface settlements, which might be significant compared to those induced by the main excavation. Also, the construction of buttress walls, a
concrete wall that perpendicular to diaphragm walls, might generate additional ground surface settlement, and this issue has not been
investigated. For clarify this issue, a series of three-dimensional finite element analysis was performed to quantify the amount of ground
surface settlement induced by the diaphragm and buttress walls installation process using the Wall Installation Modeling (WIM) method.
Results show that the installation of buttress walls inside or outside the excavation zone did not yield significant additional ground surface
settlement outside the excavation zone because the diaphragm wall was completed first before the construction of the buttress wall. But, the
construction of outer buttress walls could widen the settlement zone.
Keywords: Excavation, Settlement, Installation, Buttress wall, Diaphragm wall.
1. INTRODUCTION restrained while they were restrained in all directions at the bottom
of the geometry.
In construction practices, diaphragm/buttress/cross walls are a cast-
in-situ reinforced concrete retaining wall constructed using a slurry
supported trench method. The installation process includes slurry
supported trench excavation, placing the reinforcement cage,
concrete casting and curing. This installation process would modify CL (1)
the in-situ stress state in the soil close to the trench and generate
ground surface settlements, which might be significant compared to Diaphragm wall
those induced by the main excavation. panel
The installation effects of diaphragm wall have been G CL (2)

investigated using the three-dimensional numerical analyses and
found that the soil stress redistribution might be generated due to the Buttress wall
installation of diaphragm walls and its quality of construction GM panel
65 m
(Gourvenec and Powrie 1999, Ng and Yan 1999, Comodromos et al.
2013). Schäfer and Triantafyllidis (2006) compared the results from
the Wall-Installation-Modelled (WIM) method and the Wished-In-
Place (WIP) method using three-dimensional finite element analysis
of TNEC excavation project in Taipei basin (Ou et al. 1998). They
Figure 1 Finite element model for the WIM model analysis
concluded the WIP method would underestimate 15-20% of the
ground surface settlements and the wall deflections compared to the
Table 1 Soil Input Parameters for Analyses
WIM method. Also, the construction of buttress walls might
generate additional ground surface settlement, and this issue has not
been investigated. Depth t ' E50ref ref
Eoed Eurref
Soil layer m
For clarify this issue, three-dimensional finite element analyses (m) (kN/m3) (deg) (kPa) (kPa) (kPa)
(Brinkgreve et al., 2013) were performed to quantify the amount of
0-2 18.25 30 7033 4923 1 21100
ground surface settlement induced by the diaphragm and buttress
walls installation process using the WIM method. CL(1) 2-4 18.25 30 6826 4779 1 20479
4 - 5.6 18.25 30 6631 4642 1 19894
2. WALL INSTALLATION MODELLING (WIM) 1
CL(2) 5.6 - 45 18.5 30 9488 6642 28470
METHOD
GM 45 - 65 19.6 37 85000 121000 256000 0.5
The WIM method analysis followed the procedures which were
Note: Rf = 0.9;  ur = 0.2
done by Schäfer and Triantafyllidis (2006). Fig 1 presents the three-
dimensional finite element model for the WIM model. The depth of The plane section of the considered diaphragm wall consists of
the trench (Ht) was 33 m, and the excavation length was 56 m. The fifteen diaphragm wall panels and seven buttress wall panels with a
model represents a plane section which comprises fifteen diaphragm selected length of 4 m and 6 m, respectively, as shown in Fig 2. The
wall panels and seven buttress wall panels. The Hardening Soil (HS) thickness of the diaphragm and buttress walls was assumed 0.6 m. A
model (Schanz et al, 1999) was adopted to simulate the soil number nearby each panel indicates the construction stage, for
behavior, including the clay (CL) and the silty gravel (GM) under example, the panel H was first constructed (stage 1), followed by the
the undrained and drained conditions, respectively. The model panel D and the panel L (stage 2), then the panel A and the panel O
parameters of soils were typical values for the Taipei silty clay and (stage 3), and so forth. After all of the diaphragm wall panels were
the Taipei silty gravel (Lim and Ou 2017; Hsieh et al 2016). Table completed, then the panel buttress-H was constructed (stage 8),
1 lists the input parameters for the WIM model analysis. 10-node followed with the panel buttress-D and the panel buttress-L (stage
tetrahedral elements were employed to simulate the soil and trench 9), and closed by the panel buttress-B, the panel buttress-F, the
volume. Soil movements normal to the four vertical sides were panel buttress-J and the panel buttress-N (stage 10).
500
X=0 m
L dw =4m L bw =6m
Note: a number nearby each panel indicates the construction stage.
Figure 2 Construction stages of the diaphragm wall and buttress wall panels
For each stage, three additional steps should be conducted to model not necessary to be considered because the soil would be excavated
the WIM method such as: soon after the retaining wall system was constructed. Moreover, the
outer buttress wall trench excavation widened the ground settlement
(1). The excavation under slurry support was modeled by zone but the additional ground surface settlement induced by the
deactivating the respective finite elements inside the trench and outer buttress walls trench excavation was insignificant. Thus, it
applying the distributed loads on the surface of the trench could be concluded that the installation of buttress walls has no
walls. The magnitude of the loads corresponds to the significant effect on the additional ground surface settlement
hydrostatic slurry pressure with a bulk unit weight of b=10.3 induced by the buttress walls trench excavation because the
kN/m3. diaphragm wall was completed first before the construction of the
(2). On the subsequent process of concrete pouring, the distributed buttress wall.
loads were increased from the slurry to the fresh concrete
Uz (mm)
pressure (c). The pouring process was modeled following the
10
bilinear approximation by Lings et al. (1994), which adopts a 8
hydrostatic pressure distribution up to a critical depth (hcrit) of
6
20–30% of the panel depth. Below hcrit, the pressure gradient
4
corresponds to bulk unit weight b of the bentonite slurry:
2
  c  z , z  hcrit 0
c   (1)
 b  z   c   b   hcrit , z  hcrit
-2
-4
-6
where c is the bulk unit weight of the concrete, b the bulk unit
(a) (b) -8
weight of the slurry, z the depth below surface ground level.
-10
(3). The finite elements representing the fresh concrete were -12
activated inside the trench, and the distributed loads are -14
removed. The increased stiffness of the concrete due to aging is -16
considered by a suitable evolution of Young’s modulus, E, and -18
the Poisson ratio,  in the course of 28 days. -20
-22
-24
3. RESULTS AND DISCUSSIONS -26
-28
Fig 3 shows the contour of ground surface settlements induced by
(c) (d) -30
the diaphragm and buttress walls installation. As shown in Fig 3, the
installation of buttress walls inside the excavation zone did not yield Figure 3 The contour of ground surface settlements induced by: (a)
significant ground surface settlement outside the excavation zone. diaphragm wall installation, (b) diaphragm wall with single buttress
However, it could generate the ground surface settlement inside the wall installation, (c) diaphragm wall with three buttress walls
excavation zone. Furthermore, buttress walls also modeled outside installation, (d) diaphragm wall with seven buttress walls
the excavation zone. Fig 4 shows the contour of ground surface installation
settlements induced by the diaphragm and outer buttress walls
installation. As shown in Fig 4, based on the maximum settlement Furthermore, the ground surface settlements at each cross
point of view, the installation of outer buttress walls yielded section of the diaphragm wall with inner and outer buttress walls
insignificant additional ground surface settlements outside the were plotted in Fig 5 and Fig 6, respectively. The ground surface
excavation zone. But, the construction of outer buttress walls settlement (vw) and the distance behind the diaphragm wall (d) are
widened the settlement zone. normalized with the depth of the trench (Ht). The main influence
Table 2 summarizes the maximum ground surface settlement range of settlement was 0.3 to 0.5Ht from the diaphragm wall trench
induced by the diaphragm and buttress walls installation inside and panel, and small settlement occurred beyond 1.0Ht from the panel.
outside the excavation zone. The maximum ground surface This observed settlement characteristic was also reported by Ou and
settlement occurred at the center of the diaphragm wall section Yang (2000) in which they monitored the settlement induced by the
(x=0). The maximum ground surface settlement induced by construction of the diaphragm walls for the excavations in the Taipei
diaphragm wall installation was 16 mm. It was apparent that the Rapid Transit System. In addition, at some cross-sections, small
installation of seven inner and outer buttress walls only increased amount of ground surface heave was detected from the computation
1.1 mm and 0.6 mm of ground surface settlements, respectively, and results, especially at the location between 0.5 to 1.0 d/Ht. According
they were minimal. The inner buttress wall trench excavation only to author experiences, it seems that the ground heave was unlikely to
induced ground surface settlement in the excavated zone but it was be occurred in the field. The possible reason might due to the
501
limitation of HS model. Moreover, the WIM method yielded the vw

was 0.05% Ht while Ou and Yang (2000) reported the vw was in the
range of 0.05% to 0.13% Ht, depends on the progress of completed
diaphragm panels. Although the computed vm might underestimate
the field condition, at least the installation effect of buttress walls
could be well captured. (a) (b)
Table 2 Summary of the Maximum Ground Surface Settlement

Induced by the Diaphragm and Buttress Walls Installation
Settlement behind the D-wall (mm)

Description x= x= x= x= x= x= x=
0m 4 m 8 m 12 m 16 m 20 m 24 m (c) (d)
D-wall only 16.0 11.3 14.8 11.8 12.4 11.8 15.5 d/Ht
0 1 2 3 4
D-wall + 1 Inner
16.5 11.4 15.0 11.9 12.5 11.9 15.5 -0.02
B-wall
0
/Ht (%)
D-wall + 3 Inner
16.6 11.3 15.3 12.2 12.9 12.3 15.8 0.02
B-walls
vw
0.04
D-wall + 7 Inner
17.1 11.8 15.9 12.8 13.4 12.7 16.3 (e) 0.06 (f)
B-walls
D-wall + 1 Outer
16.2 13.5 15.2 11.4 13.7 11.3 15.3
B-wall
D-wall + 3 Outer Legend:
16.3 13.3 15.5 11.8 14.3 11.9 15.6
B-walls
D-wall + 7 Outer
16.6 13.6 16.0 12.10 14.7 12.1 16.5
B-walls
Note: x indicates the distance away from the center section of the diaphragm (g)
wall
Figure 5 Profile of ground surface settlements at each cross section
of the diaphragm wall and inner buttress walls: (a) x = 0 m, (b) x = 4
Uz (mm) m, (c) x = 8 m, (d) x = 12 m, (e) x = 16 m, (f) x = 20 m, (g) x = 24 m
10
8
6
4
2
0
-2
-4
-6
-8 (a) (b)
(a) (b) -10
-12
-14
-16
-18
-20
-22
-24 (c) (d)
-26
-28 d/Ht
0 1 2 3 4
-30
-0.02
(c) (d)
0
/Ht (%)
Figure 4 The contour of ground surface settlements induced by: (a) 0.02
diaphragm wall installation, (b) diaphragm wall with single outer
vw
0.04
buttress wall installation, (c) diaphragm wall with three outer
0.06 (e) (f)
buttress walls installation, (d) diaphragm wall with seven outer
buttress walls installation d/Ht
0 1 2 3 4
Legend:
-0.02
4. CONCLUSION
0
As discussed above, the Wall-Installation-Modelling method would 0.02

substantially increase the complexity and the running time of the 0.04
analysis. Hence, for simplification, the widely used Wish-In-Place 0.06 (g)
method (Hsieh et al. 2016, Dong et al. 2016, Goh et al. 2017) was
adequate for the simulation of diaphragm wall and buttress wall with Figure 6 Profile of ground surface settlements at each cross section
the consideration of the weight of the concrete from the diaphragm of the diaphragm wall and outer buttress walls: (a) x = 0 m, (b) x = 4
wall and buttress wall over the existing soil. m, (c) x = 8 m, (d) x = 12 m, (e) x = 16 m, (f) x = 20 m, (g) x = 24 m
502
5. ACKNOWLEDGMENTS analysis." Journal of Tunneling and Underground Space

The authors acknowledge the support provided by the Ministry Technology, 63, pp118-132.
of Science and Technology in Taiwan via grant number: MOST Lings, M. L., Ng, C. W. W., Nash, D. F. T. (1994) "The lateral
106-2221-E-146-002. pressure of wet concrete in diaphragm wall panels cast under
bentonite". Proceedings of the Institution of Civil Engineers
4. REFERENCES Geotechnical Engineering, 107(3),pp163–172.
Ng, C. W. W., Yan, R. W. M. (1999) "Three-dimensional modelling
Brinkgreve, R. B. J., Engin, E., and Swolfs, W. M. (2013) PLAXIS of a diaphragm wall construction sequence". Geotechnique,
3D Manual; Delft, Netherlands, PLAXIS. 49(6),pp825-834.
Comodromos, E. M., Papadopoulou, M. C., and Konstantinidis, G. Ou, C. Y., Liao, J. T., Lin, H. D. (1998) "Performance of Diaphragm
K., (2013) "Effects from diaphragm wall installation to Wall Constructed Using Top-Down Method." Journal of
surrounding soil and adjacent buildings". Computer and Geotechnical Engineering and Geoenvironmental
Geotechnics, 53, pp106–121. Engineering 124,pp798-808.
Dong, Y. P., Burd, H. J., and Houlsby, G. T. (2016) "Finite-element Ou, C. Y., Yang, L. L. (2000) "Ground movement induced by the
analysis of a deep excavation case history". Geotechnique, construction of diaphgram wall". Geotechnical Research
66(1), pp1–15. Report No GT200005, Department of Construction
Goh, A. T. C., Zhang, F., Zhang, W., Chew, O. Y. S. (2017) Engineering, National Taiwan University of Science and
"Assessment of strut forces for braced excavation in clays Technology, Taipei, Taiwan, R.O.C
from numerical analysis and field measurements". Computer Schäfer, R., Triantafyllidis, T. (2006) "The influence of the
and Geotechnics, 86,pp141-149. construction process on the deformation behaviour of
Gourvenec, S. M., Powrie, W. (1999) "Three-dimensional finite- diaphragm walls in soft clayey ground". International Journal
element analysis of diaphragm wall installation". for Numerical and Analytical Methods in Geomechanics.
Geotechnique, 49(6),pp801-823. 30,pp563-576.
Hsieh, P. G., Ou, C. Y., and Hsieh, W. H. (2016) "Efficiency of Schanz, T., Vermeer, P. A., and Bonnier, P. G. (1999) "Formulation
excavations with buttress walls in reducing the deflection of and verification of the Hardening-Soil model". Beyond 2000
the diaphragm wall". Acta Geotechnica, DOI in Computational Geotechnics. Brinkgreve ed, Rotterdam
10.1007/s11440-015-0416-6. Balkema, pp281-290.
Lim, A., and Ou C. Y. (2017) "Stress paths in deep excavations
under undrained conditions and its influence on deformation
503
Evaluation of Surface Settlement and Lateral Displacement During Tunnel

Construction Using 3D Numerical Modelling
A. Fahmi1, Desyanti1, I. Masyhur2, H. Bigman2, S. Endra2, M. Riska3 and M. Weni3
1
Institute of Road Engineering (IRE), Bandung, Indonesia
2
Bandung Institute of Technology, Bandung, Indonesia
3
Mass Rapid Transit Jakarta Project, Jakarta, Indonesia
E-mail: fahmi.aldiamar@pusjatan.pu.go.id
ABSTRACT: Shield tunnelling construction in urban area could influence surrounding structures. Several empirical equations and
laboratory testing using centrifuge model already published to determine effect of tunnel excavation, especially due to the shield loss and tail
loss effects. Gap between shield machine and tunnel segment lining is the main subject that determine the amount of surface settlement and
lateral displacement. In this study, case study for evaluating ground deformation during tunnel construction is located in Mass Rapid Transit
Jakarta project around diaphragm-wall Bendungan Hilir station. Evaluation consists of comparison between prediction analysis using 3-
dimensional finite element numerical model and actual deformation from surface settlement and inclinometer are conducted to evaluate the
modelling effectiveness.
Keywords: shield tunneling, ground loss, surface settlement, lateral displacement and mass rapid transit Jakarta
1. INTRODUCTION then conducted to evaluate best fit result between predicted and
actual measured deformation.
Mass Rapid Transit Jakarta is the first shield tunnelling using Earth
Pressure Balance (EPB) method constructed in Indonesia. EPB
method are commonly used for fine grained soils (<0,06 mm) such
as clay-silt and silt-sand soft to stiff consistency and active support
pressure control conducted to ensure low influence to settlement and
lateral deformation (B. Maidl, et al., 2012). Anticipation of possible
problems especially to surrounding structure due to excessive
settlement, cave in and building damage should be considered
during design and construction phase. According to Mair and Taylor
(1997) and Nagen Loganathan, et al (2011) deformation due to
tunnel construction mainly caused by face loss, over cutting, shield
loss and tail loss. Even though gap usually determined by
effectiveness of shield machine, soil condition could also influence
displacement that happened. In this study, case study for evaluating
ground deformation during tunnel construction is located in Mass
Rapid Transit Jakarta project around diaphragm-wall Bendungan
Hilir station. Surface settlement marker and inclinometer were
placed above tunnel alignment and near the tunnel cross section to
measure surface settlement and lateral displacement. Evaluation
consists of comparison between prediction analysis using 3-
dimensional finite element numerical model and actual deformation
from surface settlement and inclinometer are conducted. Since
tunnel in this location constructed into hard silty sand material, two
constitutive models using Mohr-Coulomb and hardening-soil model
Figure 1 Location of monitoring site
were evaluated to determine which of the following constitutive
model best suit the actual deformation occur on the field during
construction. 2.1 Soil Parameter
2. METHODOLOGY There were 3 borings conducted in the tunnel section at Bendungan
Hilir station, with distance range from 50 m to 92 m from
Numerical modelling 3-dimension finite element method using
diaphragm-wall (D-wall). First boring was conducted during design
Midas Gts Nx software is conducted to predict and simulated
stage on 2010, while the other boring conducted in 2016 and 2017.
excavation phase in stage construction. Deformation acting around
According to the evaluation from all of the field testing, tunnel
tunnel excavation then compared with surface settlement and lateral
construction was predicted going through hard silty sand and below
deformation measured using digital waterpass combined with
ground water level. According to ASTM D2487-17, soil at project
robotic total station and inclinometer that placed 5 m from outer
location mostly have liquid limit more than 50% and plasticity index
tunnel cross section. Ground condition that modelled in this paper
plots below “A”-line which can be defined as elastic silt (MH) as
based on primary borehole, in-situ test and laboratory test also
seen in Figure 2, while prediction of Young’s modulus according to
secondary data collected by MRT project. All of the data then
pressuremeter test can be seen in Figure 3. Soil parameter used in
review to gain soil parameter that suitable for Mohr-Coulomb and
the modelling can be defined in Table 1 and Table 2.
hardening-soil modelling. Actual and numerical matching method
504
100 70,000
MRTJ (2010) MRTJ (2010)
90
BH-1 (2016) 60,000 Bowles (1997)
80 BH-2 (2017) BH-2 (2017)
P r e s u r e m e t e r M o d u lu s, ( k P a )
70 50,000
Plasticity Index (PI)
60
40,000
50
30,000
40
y = 756.91x + 2301.8
30 20,000 R² = 0.649
MH or OH
20
10,000
10
CL or ML
ML or OL 0
0
0 10 20 30 40 50 60 70 80 90 100 110 0 10 20 30 40 50 60
N-SPT
Liquid Limit (LL)
Figure 3 Elastic modulus from Pressuremeter test
Figure 2 Plasticity chart data
Table 1 Mohr- Coulomb Soil Parameters

γ c' Ø' E
Name Description
kN/m3 kPa deg kN/m2
Layer-1 Stiff silty clay 15.68 5 25 7,600
Layer-2 Stiff clayey silt 15.54 10 27 9,114
Layer-3 Very stiff silty sand 15.50 1 30 17,440
Layer-4 Hard silty sand 16.00 1 30 47,716
Layer-5 Very hard clayey silt 16.00 15 27 47,716
Table 2 Hardening Soil Parameters

Secant Tangent Unload/reload Shear modulus
Name Description stiffness, E50 stiffness, Eoed stiffness, Eur K0 at small strain
(kPa) (kPa) (kPa) (kPa)
Layer-1 Stiff silty clay 4,180 4,180 12,540 0.58 31,660
Layer-2 Stiff clayey silt 5,013 5,013 15,038 0.55 24,713
Layer-3 Very stiff silty sand 9,592 9,592 28,776 0.50 18,957
Layer-4 Hard silty sand 26,244 26,244 78,732 0.50 14,217
Layer-5 Very hard clayey silt 26,244 26,244 78,732 0.55 23,822
2.2 Numerical modelling Surface elevation in tunnel modelling are according to topography at
site to evaluate influence of overburden height (H), diameter of
Numerical modelling using finite element method is usually used for tunnel (D) is 6.55 m and advance of excavation (d) is 1.50 according
evaluating soil structure interaction between tunnel boring machine to lining segment placement as illustrated in Figure 5.
and the surrounding soil, complex model also to estimate
deformation around tunnel on design stage. German Tunneling
Committee (ITA-AITES, 2016), recommended 3 dimensional (3D)
numerical modelling with stage construction to evaluate tunnel
excavation phase. 3D model can give a more thorough information
not only deformation around tunnel excavation, also face stability
each step of excavation progress. In this paper, thickness of ground
loss is according to site project information as seen in Figure 4 and
modelled as soil interface that reduced in term of strength.
Figure 4 Shield loss and tail loss at MRT Project

(Shimizu-Obayashi-Wijaya-Jaya,SOWJ, 2015) Figure 5 Numerical modelling scheme
505
3. RESULTS AND DISCUSSION displacement from all of 10 sections show that the trend is displaced
and the value decline along with the distance away from tunnel
3.1 Surface settlement measurement cross section. This trend is relatively similar with previous
According to average value of surface settlement monitoring for 10 researcher (Sven Moller, 2006), and (Loganathan, et al, 2011).
cross section, daily measured starting from November 21, 2016 until Surface settlement that occurred are relatively small. This is
April, 9 2017 as seen in Figure 6, it is shown that at every section possibly due to the high overburden above tunnel construction and
trend of surface settlement curve is quite different. 0 number in x- less disturbance of ground loss from shield and tail loss. Empirical
axis is denote centreline of tunnel section position which shown approach as mentioned in (JSCE, 2006) describe that minimum
largest surface settlement (until -1.39 cm) occurred during overburden that could cause large surface settlement is less than 1.5
monitoring time frame. In some position of surface settlement times tunnel diameter.
marker section, heaving occurred, nevertheless average
Figure 6 Result of surface settlement measurement
3.2 Inclinometer measurement A

(Direction to Meridien B
(Direction to HSBC
Building)
Building)
Inclinometer position are set into four direction i.e. Meridien Displacement (mm)
Building to HSBC Building for cross sectional evaluation and D- -10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10
0
wall to Setiabudi Station for longitudinal evaluation of tunnel
1
construction influence as seen in Figure 7. Shield tunnel position is
2
passing through inclinometer position at December 8, 2016 and
3
inclinometer indicate lateral displacement (2.6 mm) perpendicular to
4
the tunnel direction or to direction A. Lateral displacement
5
increasing until 3.15 mm on December 10, 2016 and reduced until
6
0.6 mm on April 06, 2017 as seen in Figure 8. This behaviour
7
indicate ground loss during passing through of tunnel construction
8
and application of grouting radial to the tunnel surrounding that
9
filling void around tunnel excavation and reduce lateral Ground water level (-10.10m)
10
displacement.
11
12
Cumulative displacement (mm) 13
A (Direction to Meridien Building)
-10 14
D ep th (m )
-9 15
3.45 m
-8 16
-7 17
-6 18
Tunnel direction
-5 19
-4
20
-3 6.65 m 21
Displacment
direction -2
22
10 9 8 7 6 5 4 3 2 1 0 -1
-1 -2 -3 -4 -5 -6 -7 -8 -9 -10
D (Direction C (Direction 23
0
to D-Wall) to Setiabudi
1 24
Station)
2 Tunnel position 25
3 (Not Scaled) 26
4 27
5 28
6 29
November 28, 2016 (08:00 AM) 7
December 08, 2016 (01:00 AM) 30
December 09, 2016 (01:00 AM) 8 November 28, 2016 (08:00 AM)
December 10, 2016 (01:00 AM) December 08, 2016 (01:00 AM)
December 11, 2016 (04:00 PM) 9
April 06, 2017 (01:00 PM)
December 09, 2016 (01:00 AM)
10 December 10, 2016 (01:00 AM)
December 11, 2016 (04:00 PM)
B (Direction to HSBC Building) April 06, 2017 (01:00 PM)
Figure 7. Result of top view cumulative displacement Figure 8. Result of lateral displacement during tunnel construction
506
3.2 Numerical simulation result -30 -25 -20 -15 -10 -5 0 5 10 15 20 25 30 35

3.00
Numerical modelling using Mohr-Coulomb and hardening soil CROSS SECTION 1-10 (21 NOV 2016-9 APR 2017)
model give a different result as seen in the total displacement Figure ACTUAL DEFORMATION (AVERAGE VALUE) 2.00
9 and Figure 10. Mohr-Coulomb model give a smaller displacement MOHR-COULOMB MODEL
D e fo rm a tio n (c m )
HARDENING SOIL MOEL
at ground surface while hardening soil model give the opposite 1.00
result.
0.00
-1.00
-2.00
-3.00
Distance (m)
Figure 11. Comparison between numerical and actual surface
settlement
A
(Direction to Meridien B
Building) (Direction to HSBC
Building)
Displacement (mm)
-10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10
0
1
2
3
4
Figure 9. Numerical simulation result using Hardening soil Model 5
6
7
8
9
Ground water level (-10.10m) 10
11
12
13
14
D ep th (m )
15
3.45 m
16
17
18
19
20
6.65 m 21
22
23
24
Tunnel position 25
Figure 10. Numerical simulation result using Mohr-Coulomb model (Not Scaled) 26
27
28
Comparison between numerical and actual deformation
29
according to measured instrumentation result are describe in Figure
30
11 and Figure 12. Mohr-Coulomb model fit in great correlation with
November 28, 2016 (08:00 AM) December 08, 2016 (01:00 AM)
average surface deformation result, while hardening soil close match December 09, 2016 (01:00 AM) December 10, 2016 (01:00 AM)
with the maximum surface ground deformation. Opposite result December 11, 2016 (04:00 PM) April 06, 2017 (01:00 PM)
shown in the lateral displacement, i.e. Mohr-Coulomb model fit in HARDENING SOIL MODEL MOHR-COULOMB MODEL
great correlation with maximum lateral displacement, while Figure 12. Comparison between numerical and actual lateral
hardening soil model give a smaller result. displacement
507
4. CONCLUSIONS 5. REFERENCES
In this study, numerical analysis compared with actual measurement ASTM D2487-17. Standard Practise for Classification of Soils for
data from instrumentation of surface ground settlement and lateral Engineering Purposes (Unified Soil Classification System).
displacement. The main results can be summarized as follows: B. Maidl, M. Herrenknecht, U. Maidl, G. Wehrmeyer. (2012)
(1) Mohr-Coulomb model give a good result to evaluate lateral Mechanised Shield Tunnelling, 2nd Edition, Wilhelm Ernst
displacement on tunnel construction but give a less and Sohn. Berlin, Germany.
conservative result on the surface ground settlement. ITA-AITES. (2016) Recommendations for face support calculations
(2) Hardening soil model give a smaller lateral displacement on for shield tunneling in soft ground, German Tunneling
tunnel construction but give a more conservative result on the Committee.
surface ground settlement JSCE. (2007): Standard specifications fortTunneling: Shield tunnels.
Japan Society of Civil Engineers.
This study was collaboratively carried out based on Mair, R. J., Taylor, R. N. (1997): Bored tunneling in urban
“Agreement on Research Exchange and Cooperation between environment, Plenary Session 4, Proc. 14th Int, Conf. on
Institute of Road Engineering and MRT Jakarta in order to develop SMFE, Hamburg, Vol. 4, pp. 2353-2385.
technical guideline for shield tunnelling using earth pressure balance Nagen Longanathan. (2011): An Innovative Method for Assessing
method. Tunnel-Induced Risks to Adjacent Structures. Parsons
Brickerhoff Inc. New York.
Sven Moller. (2006) Tunnel induced settlements and structural
forces in linings. Phd thesis. Universitat Stuttgart. ISBN-10:
3-921837-54-5.
508
Aplikasi Metode Elemen Hingga dan Kriteria Runtuh Papaliangas Batuan Berkekar:
Problem-Problem Analisa Terowongan, Kestabilan Lereng dan Wellbore Stability
dalam Mekanika Batuan
G.L. Wilham1
1
PT Signifikan Solusi Integrasi and Lecturer at Postgraduate Program of Civil Engineering Dept.,
National Institute of Science and Technology (ISTN), Jakarta, Indonesia.
1
E-mail: wilham@signtegra.com and wilham.george@gmail.com
ABSTRACT: On the early days of Finite Element Method research, Prof. Zienkiewicz et.al. (1968) in his monumental article, implemented
a no-tension material for jointed rock mass (material that cannot withstand tensile stress) into existing Mohr-Coulomb Joint Rock criterion
for shear stress analysis. With background from the geologists / geophysicists research between 1960s to1990s, Papaliangas et.al. (1996)
proposed a new breakthrough in joint rock failulre criterion, that is the Brittle-Ductile Transition. The old Mohr-Coulomb equation had been
used in various rock mechanic problems (Zienkiewicz et.al. 1971, Locat et.al. 2000) and in petroleum engineering / engineering geologists
(Hetcher 1995, Aoki et.al. 1993, McLean 1998, Ramsay & 2000), and it does not include the Brittle Ductile Transition element. In this paper,
the author tries to implement a new algorithm of Papaliangas into the older Finite Element Method code of Zienkiewicz-Pande (1977), so
this assistance can help to solve the rock mechanics problems (in Slope Stability, Tunneling, wellbore stability, etc.). Some examples have
been presented. Failure Indices and Failure Algorithm have been proposed and explained.
Keywords: joint rock, Tunneling, Failure Indices, Britle-Ductile Transition (Papaliangas), Slope Stability, Wellbore Stability
ABSTRAK: Di masa permulaan riset Metode Elemen Hingga Prof. Zienkiewicz dkk. (1968) dalam artikel monumentalnya mengusulkan
kriteria kekar/batuan sebagai material yang tidak dapat menahan tegangan tarik (no-tension material) kedalam kriteria runtuh kekar Mohr-
Coulomb untuk analisa tegangan gesernya. Dengan memakai latar belakang riset dari para ahli geologi / geofisik antara masa 1960-an s/d
1990-an, Papaliangas dkk. (1996) mengusulkan suatu terobosan baru dalam rumus dasar keruntuhan kekar dengan mengimplementasikan
pengaruh Transisi Getas-Daktail (Brittle-Ductile Transition) pada kekar/batuan. Kriteria runtuh lama (Mohr-Coulomb) yang dipergunakan
para pakar mekanika batuan (Zienkiewicz et.al. 1968, Locat dkk. 2000) dan ahli geologi teknik / teknik perminyakan (Hetcher 1995, Aoki
dkk. 1993, McLean 1998, Ramsay & 2000) tidak memasukan unsur transisi Getas-Daktail. Penulis mencoba mengimplementasikan
algoritma Papaliangas yang baru ini kedalam code (listing program Fortran) Metode Elemen Hingga lama dari Zienkiewicz-Pande (1977),
sehingga hasilnya diharapkan dapat bermanfaat dalam pemecahan problem-problem mekanika batuan, perencanaan terowongan (tunneling),
kestabilan lereng dan stabilitas lubang bor (wellbore stability). Beberapa contoh telah ditampilkan dan dibandingkan disini. Pengenalan akan
Indeks keruntuhan dan algoritma keruntuhan juga diterangkan disini.
Keywords: Kekar/batuan, Penerowongan, Indeks Keruntuhan, kiteria runtuh Britle-Ductile (Papaliangas), Kestabilan Lereng, Kestabilan
Lubang Bor
1. PENDAHULUAN ahli Geologi Teknik / Teknik Permiyakan (Hetcher 1995, Aoki dkk.
1993, McLean 1987, Ramsay 2000) belum bisa menerapkan
Banyak pekerjaan teknik sipil dan pertambangan melibatkan fenomena Transisi Getas-Daktail (brittle-ductile transition).
pembuatan perkuatan lereng, terowongan dibatuan, seperti Umumnya tinjauan dipisahkan dalam masing-masing masalah.
terowongan untuk “spillway” bendungan, ruang penyimpanan mesin Seperti McLean (1987), hanya meninjau keruntuhan batuan utuh
pembangkit listrik pada bendungan, terowongan pada tambang- (intact rock) karena umumnya hal itu terjadi dalam problem
tambang. Demikian pula untuk dunia perminyakan, analisa wellbore rekayasa perminyakan (simulasi lubang bor / “wellbore”). Problem
stability juga sangat penting. Umumnya batuan yang ditemukan “wellbore” menyangkut kedalaman yang besar, maka umumnya
adalah batuan utuh (intact rock) dan batuan berkekar (joint rock). dijumpai keruntuhan pada batuan utuh (intact rock); walaupun
Perencanaan pembangunan proyek terowongan batuan melibatkan keruntuhan kekar (joint rock) juga terjadi.
berbagai bidang ilmu, dan satu diantaranya adalah mekanika batuan Sedangkan Locat dkk. (2000) sudah menyuguhkan grafik
dan pemodelan konstitutip (kriteria runtuh) dari material batuan utuh keruntuhan batuan utuh dan kekar batuan dalam satu gambar, namun
/ berkekar tsb. karena maslah yang dihadapi adalah kestabilan lereng dengan
Era teknologi informasi telah membawa kita/ para insinyur, kedalaman rendah maka yang dijumpai umumnya keruntuhan pada
kepada era penyelesaian dengan dibantu komputer dan metode kekar batuan (joint rock).
numerik dengan berbagai algoritma perhitungan yang kompleks. Berkaitan dengan itu penulis, dalam software Newmo3962_2000,
Prof. Zienkiewicz et al. (1968) mengusulkan kriteria batuan mencoba mengimplementasikan algoritma baru (Louhenapessy
sebagai material yang tidak dapat menahan tarik (no-tension 2000), yang diharapkan dapat bermanfaat dalam memecahan
material) dan memakai kriteria runtuh Mohr-Coulomb untuk problem-problem mekanika batuan. Beberapa contoh numerik akan
tegangan gesernya. Papaliangas (1996) melakukan riset-percobaan ditampilkan dan dibandingkan dalam sub-bab sub-bab berikut ini.
kekar batuan dan mengusulkan suatu terobosan baru dalam rumus Adapun sistematika pembahasan makalah ini dapat digambarkan
dasar keruntuhan batuan. Tentu saja Papaliangas didukung riset-riset seperti pada bagan dibawah ini (Gambar 1).
para pakar Geologi dan Geofisik sebelumnya (Greenwood &
Williamson 1966, Byerlee 1978, Scholz 1990). Model konstitutip
baru Papaliangas segera disambut dengan pemodelan numeric
2. KRITERIA RUNTUH LAMA (MOHR COULOMB)
konstitutip dari batuan utuh dan kekar batuan oleh penulis
(Papaliangas 1999, Louhenapessy 2000). 2.1 Kriteria runtuh Mohr-Coulomb
Paper ini memperkenalkan algoritma yang diusulkan penulis
sebelumnya (Louhenapessy 2000) dan juga memberikan beberapa Pemodelan kriteria runtuh Mohr-Coulomb unuk kekar batuan
hasil dari contoh simulasi numerik. Dari studi literatur, kriteria adalah:
runtuh Mohr-Coulomb yang dipergunakan para pakar mekanika
batuan sebelumnya (Zienkiewicz et.al. 1971, Locat dkk. 2000) dan F = |j | + n tan  j = 0 (1)
509
dimana, dimana,
j = sudut geser kekar batuan (joint rock friction angle),
j = tegangan geser pada bidang kekar, dan  
log 10  nT  (5)
n = tegangan normal pada permukaan kekar.  n 
tan  tan o
 
log 10  nT 
  no 
dan, no = tegangan normal (normal stress) pada permukaan

kekar batuan; m = sudut geser dalam pada kekar batuan yang bukan
merupakan sudut dilational (independent of normal stress). Lihat
Tabel 1;  = simbol sudut dilasional pada saat puncak dari tegangan
geser; o = sudut dilasional puncak (maksimum) pada suatu
tegangan normal (no) yang tidak menyebabkan terjadinya
deformasi (asperity); nT = tegangan normal efektif yang
mengakhiri semua sudut dilasional (Gambar 2).
Table 1 m, Sudut Geser Dalam pada Kekar yang Bukan

Merupakan Sudut Dilasional (Papaliangas dkk. 1996)
Jenis Batuan Range m
Sandstone Dari 26.2° sampai 41.6°
Gambar 1 Sistematika Pembahasan
Limestone Dari 34.6° sampai 48.6°
2.2 Kriteria runtuh Patton Siltstone Dari 22.5° sampai 36.6°
Pada konperensi mekanika batuan pertama di Lisbon, tahun 1966, Granite Dari 27.7° sampai 37.3°
Patton mengusulkan tambahan sudut dilasional (dilational) kedalam
sudut geser kekar. Hal ini didapatnya dari studi dan evaluasi yang
terperinci dari problem-problem kemiringan lereng (Patton 1966).
Patton menunjukan hubungan langsung antara tegangan geser,  dan
sudut dilasional, i:
j = n tan ( r + i) (2a)
dimana,
r = sudut geser yang tersisa (residual) pada kekar batuan,
j = tegangan geser pada bidang kekar, dan
n = tegangan normal pada permukaan kekar.
Gambar 2 Grafik garis runtuh dan sudut geser dalam, m
Sehingga kriteria runtuhnya menjadi : (Papaliangas dkk. 1996)
F = |j | + n tan ( r + i) = 0 (2b) 4. KRITERIA RUNTUH BATUAN UTUH

Kriteria runtuh yang dipergunakan untuk batuan utuh adalah Kriteria
2.3 Kriteria runtuh Barton-Choubey (1977) runtuh Mohr-Coulomb sbb,
Pada tahun 1977, Barton dan Choubey memperbaiki criteria Patton
diatas dengan memperkenalkan ketergantungan tegangan normal, F = |i| - Co + n tan o = 0 (6)
n pada sudut dilasional, i dalam bentuk persamaan empiris sbb:
dimana,
i = JRC log10 (JCS/n) (3) ni = tegangan normal pada bidang keruntuhan daidalam batuan
utuh,
Dengan JRC = Koefisien kekasaran permukaan kekar (Joint o = sudut geser dalam dari batuan utuh,
Roughness Coefficient), dan JCS = Koefisien kekuatan tekan Co= kohesi batuan utuh tsb.
permukaan kekar (Joint Wall Compressive strength).
Agar dapat dicari besarnya dalam gaya-gaya dengan berorientasi
3. KRITERIA RUNTUH BRITLE-DUCTILE pada sumbu lokal, maka diperlukan bentuk persamaan dalam bentuk
(PAPALIANGAS) tegangan-tegangan invariant, maka persamaan (6) menjadi:
Model kriteria runtuh Getas-Daktail untuk kekar batuan yang _
diusulkan oleh Papaliangas adalah: F = (cos+(sin sino)/ 3 )-msin o-Cocoso= 0 (7)
_
F = |j | + n tan (m+) = 0 (4) dimana  ,  dan m adalah komponen-komponen invariant.
510
5. METODE ELEMEN HINGGA DAN KESULITAN-

KESULITAN DALAM PEMODELAN
Kesulitan-kesuliltan dalam pemodelan Metode Elemen Hingga,
khususnya dalam memodelkan struktur batuan dan struktur didalam
batuan, antara lain (Gambar 3):
1. Kesulitan dalam mengimplementasi kekasaran kekar (dan sesar)
didalam pemodelan,
2. Bagaimana memodelkan RQD
3. Bagaimana memodelkan jarak-antar kekar dan jumlah set kekar
(joint set)
4. Dip Angle (arah kekar)
5. Tegangan-tegangan/regangan-regangan awal (in-situ
stress/strain) disekitar terowongan
6. Pengaruh air (pore pressure)
7. Pengaruh waktu (Rheology of rock) Gambar 4 Algoritma tahap-tahap pengambilan keputusan untuk
menghasilkan Indeks Keruntuhan Tarik (Louhenapessy 2000)
Gambar 3 Latar Belakang: Kesulitan-kesulitan dalam pemodelan

“properties of rock” (Louhenapessy 2000 & Louhenapessy 2017)
6. ALGORITMA INDEKS KERUNTUHAN

Tegangan-tegangan yang terjadi pada sebuah titik disebuah
kedalaman suatu struktur di massa batuan (jointed rock mass) yang
diakibatkan galian pembuatan terowongan (tunneling), gaya-gaya
dari dalam bumi: tegangan lateral in-situ (in-situ stress), kehadiran
air, dsb. akan dianalisa oleh perhitungan numerik (Metode Elemen
Hingga). Proses selanjutnya masuk kedalam algoritma pemilihan
keruntuhan, yang dibuat untuk memutuskan apakah kemungkinan-
kemungkinan keruntuhan berikut ini dapat terjadi:
(a) keruntuhan batuan utuh. Parameter-parameter kuat geser batuan
utuh diuji dan diuji pula fungsi runtuhnya: yaitu Persamaan (7),
dan (a)
(b) keruntuhan kekar-kekar batuan: Parameter-parameter kuat geser
kekar batuan diuji dan diuji pula fungsi-fungsi runtuhnya: yaitu
Persamaan (1), (4) dan (5). Demikian pula,
(c) kehadiran tegangan tarik dicek.
Suatu algoritma yang menentukan tempat-tempat dimana
keruntuhan terjadi, misalnya “apakah pada batuan utuh atau kekar
batuan atau tarik?”. Hal ini diusulkan dalam paper ini dan secara
ringkas ditampilkan dalam Gambar 4 dan 5. Secara detail dapat
dilihat pada publikasi lainnya (Louhenapessy 1997, 1998, 2000,
2017). Prosedur ini diulang lagi sampai semua titik buhul (node, dari
pemodelan metode elemen hingga) dianalisa.
7. CONTOH NUMERIK
7.1 Terowongan (Tunnels)
Dalam contoh ini sebuah terowongan lingkaran digali pada sebuah
kedalaman didalam massa batuan berkear (jointed rock mass) yang (b)
memiliki satu set kekar (one sets of joints). Hal ini ditampilkan Gambar 5 a) Algoritma tahap-tahap pengambilan keputusan untuk
untuk contoh perhitungan Indeks Keruntuhan (Failure Indices). menghasilkan Indeks Keruntuhan Kekar Batuan; b) Algoritma
Gambar 6 dan Tabel 2 menunjukan geometri dan material tahap-tahap pengambilan keputusan untuk menghasilkan Indeks
properties dari contoh numerik ini. Keruntuhan Global (Louhenapessy 2000).
511
Daerah Indeks keruntuhan (Failure Indices, seperti yang disebut

dalam thesisnya W.G. Louhenapessy (2000)) seperti pada Gambar 8
ini dapat dipergunakan dalam perencanaan disain penyangga
terowongan. Gambar 8.a s/d Gambar 8.c. menunjukan daerah Indeks
keruntuhan disekitar terowongan untuk bermacam arah kekar,  = 0o,
45 o dan 90 o dengan memakai kriteria runtuh Papaliangas dengan m
= 30 o. Simbol-simbol indeks ini ialah:
 “0” untuk daerah yang tidak runtuh
 “1” untuk daerah runtuh akibat kekuatan batuan utuh terlewati,
 “2” untuk daerah runtuh akibat kekuatan geser kekar batuan
dilewati, dan
 “5” adalah daerah runtuh akibat kekuatan tarik kekar batuan
dilewati.
Gambar 8.a menunjukan Indeks Keruntuhan disekitar dinding
terowongan dan terlihat bahwa pada atap (untuk arah kekar batuan,
 = 0o dan tegangan insitu Ko = 0.333) tidak terjadi keruntuhan Pada
hal yang lain, dalam Gambar 8.c, keruntuhan tarik terlihat pada atap
dari dinding terowongan (untuk arah kekar batuan,  = 90o dan
tegangan insitu Ko = 0.333). Perlu dicatat pula, bahwa umumnya -
pada dinding permukaan terowongan-, keruntuhan terjadi akibat
geser pada kekar, tetapi ada juga beberapa kondisi dimana kekar
runtuh akibat tarik atau keruntuhan batuan utuh, khususnya
diterowongan yang berada pada kedalaman yang besar. Gambar 8.d.
adalah hasil untuk contoh aplikasi Tambang Batubara/Coal Mining
Gambar 6 Geometri Terowongan dan Set Kekar (joint rock)
(Louhenapessy 2017).
Table 2 Parameter Material Batuan (Rockmass)
(a)
Gambar 7 menunjukan bentuk typical dari Jaring Elemen

Hingga (Finite Element Mesh) yang digunakan dalam analisis ini,
yang terdiri dari 736 titik buhul (nodes) dan 224 Elemen
Isoparametrik ber-buhul-delapan (eight noded isoparametric
elements).
(b)
Gambar 7 Jaring Elemen Hingga (Finite Element Mesh) (c)
512
7.3 Stabilitas Borehole (lubang bor) untuk aplikasi

Geomekanik
Dalam contoh wellbore stability, Indeks Keruntuhan memegang
peranan penting juga. Wellbore stability adalah analisa kestabilan
lubang bor pada pekerjaan pengeboran minyak di batuan. Dengan
hal ini penulis memakai contoh yang sama dengan contoh numerik
dari Aoki dkk. (1993) dengan mempergunakan software yang
berbeda: yaitu Newmo3962_2000 (Louhenapessy & Pande 2000).
Tabel 3 menunjukan data-data borehole tsb.
Tabel 3 Data-data untuk analisa Lubang Bor (Wellbore Stability)
In-situ Tegangan
Tegangan In-situ Pore Lumpur
Kasus
v H h Pressure / Mud
(d) (MPa) (MPa) (MPa) Pressure
(MPa) (MPa)
Gambar 8. (a-c) Identifikasi jenis keruntuhan untuk massa batuan
dengan satu set kekar dan berbagai orientasi (arah) kekar. Horizontal 72 72 54 30 33
(d) Contoh Untuk Tambang Batubara (Louhenapessy 2017)
Pemodelan numerik dengan metode elemen hingga dapat dilihat
pada Finite Element Mesh pada Gambar 10 dan juga terlihat gaya-
7.2 Lereng Batuan Berkekar gaya luar yang bekerja pada sistim borehole tsb. Hasil-hasil analisa
Dalam contoh ini, sebuah lereng digali pada massa batuan berkekar dapat dilihat pada Gambar 11. Disini terlihat dimana keruntuhan yag
(jointed rockmass) yang memiliki satu set kekar (one sets of joints), terjadi dan jenisnya. Hal tsb. dibandingkan dengan output dari Aoki
dengan arah, = 15o. Tinggi lereng dari kaki sampai puncaknya 70 dkk. (1993). Terlihat disini hasil dari Algoritma runtuh yang
m. Sedangkan rasion tegangan lateral in-situ yang ada adalah Ko diusulkan overlapping dengan analisa wellbore dari Aoki et.al.
=0.333. Data properti batuan juga diasumsikan sama dengan Tabel 2.
Daerah Indeks keruntuhan (Failure Indices) seperti pada Gambar 9
ini dapat dipergunakan dalam perencanaan / solusi disain Kestabilan
Lereng. Symbol “0” untuk daerah yang tidak runtuh; Symbol “1”
untuk daerah runtuh akibat kekuatan batuan utuh terlewati, Symbol
“2” untuk daerah runtuh akibat kekuatan geser kekar batuan dilewati,
dan, Symbol “5” adalah daerah runtuh akibat kekuatan tarik kekar
batuan dilewati.
Gambar 10 Mesh Elemen Hingga Untuk Analisa Wellbore

Stability
Gambar 9 Identifikasi jenis keruntuhan untuk massa batuan dengan

satu set kekar dan arahnya, θ = 15o. a) Kontur Indeks Keruntuhan,
b) Indeks Keruntuhan dalam angka.
Gambar 9 menunjukan Indeks Keruntuhan disekitar lereng.

Tampak pada kaki lereng, keruntuhan yang diakibatkan ‘keruntuhan
tarik’ (simbol Indeks Keruntuhan = 5) mendominasinya walaupun
ada lokasi- disekitarnya yang terjadi keruntuhan akibat geser Gambar 11 Hasil-hasil perhitungan analisis “wellbore stability”. a)
(simbol Indeks Keruntuhan = 2). Pada permukaan miring lereng dan Hasil dari Newmo3962_2000 (Louhenapessy & Pande 2000),
juga beberapa meter didalamnya terjadi keruntuhan akibat b)Hasil dari Aoki et.al (1993).
“keruntuhan batuan utuh terlewati” (simbol Indeks Keruntuhan = 1)..
513
8. KESIMPULAN Louhenapessy, Wilham. G. (2000) “Analysis of Tunnel Supports

using the Finite Element Method”; Ph.D. Thesis. University
Penjelasan teori-teori kriteria runtuh Papalaiangas (getas- of Wales Swansea, Swansea, UK.
daktail/brittle-ductile), algoritma indeks keruntuhan, indeks Louhenapessy, W, & Pande, G. N. (1997) “A Rational Finite
keruntuhan pada batuan dengan berbagai arah orientasi kekar, jenis- Element Analysis (FEA) Based Procedure for The Analysis
aplikasi keruntuhan pada analisa kestabilan lereng, terowongandan of Pressure on Tunnel Supports”; Internal report no:
kestablian wellbore (wellbore stability) dalam rekayasa CR/964/97. Department of Civil Engineering: Univ. of
perminyakan/Geomekanik telah disajikan. Wales, Swansea.
Penyerdehanaan penampilan jenis dan daerah / zone keruntuhan Louhenapessy, Wilham G. (1998) “A Rational Finite Element
dalam bentuk angka-angka indeks (“0”, “1”, “2” dan “5”) Analysis Based Procedure for The Analysis of Pressure on
memungkinkan analisis dan disain masalah-masalah geoteknik Tunnel Supports”; Pages 236-243 from “Proc. Canadian
dengan lebih pasti. Soc. Mech. Engineering FORUM 1998”, vol. 2. Ryerson
Pemodelan kriteria runtuh dengan transisi getas-dektail lebih Polytechnic University, Toronto.
presisi dibandingkan Mohr-Coulomb atau Modifikasi dari Patton Louhenapessy, Wilham G. and Pande, G. N. (2000)
dan Barton. Penggunaan metode elemen hingga memungkinkan “Newmo3962_2000: User's Instruction Manual”; Rep
penyajian yang menarik bagi zone dan type runtuh material dan juga No.CR/1022/00. Civil Eng. Dept., Univ.of Wales Swansea,
fleksibilitas dalam pemodelan kriteria runtuh. Swansea, UK.
McLean (1987) “Numerical Analysis of Well bore Instability”; PhD
9. DAFTAR PUSTAKA / REFERENCES Thesis University of London, 1987.
Aoki, T.; Tan, C.P. and W.E. Bamford (1993) “Effects of Natau, O, Buhler, M, Keller, S, & Mutschler, T. (1995) “Large
Deformation and Strength Anisotropy on Borehole Failures in scale triaxial test in combination with a FEM analysis for the
Saturated Shales”, Int. J. Rock Mech. Min. Sci. & Geomech. determination of the properties of a transversal isotropic rock
Abst., Vol 30 (1993), No.7, pp. 1031-1034. Pergamon mass”; Pages 635--643 from; Fuji, T (ed), 8th International
Press.Ltd. Great Britain. Congress on Rock Mechanics, vol. 2. ISRM, Tokyo.
Barton, N. dan Choubey, V. (1977) “The shear strength of rock Oberti, G, Bavestrello, F, Rossi, P, & Flamigni, F. (1986) “Rock
joints in theory and practice”; Rock Mechanics, Vol. 10, pp. Mechanics Investigation, Design and Construction of the
1-54.. Ridracoli Dam”; Rock Mechanics & Rock Engineering, 19,
Bieniawski, Z.T. (1990) Engineering rock mass classification. 1st 113--142.
ed. Chichester: Wiley. Palmstrom, A. (1995) “RMi - a system for characterizing rock mass
Byerlee, J. (1978) "Friction of Rocks”; Pure and Appl. Geophys. strength for use in rock engineering”; J. of Rock Mech. &
Vol. 116, pp. 615-626. Tunnelling Tech.), 1(2), 69--108 [1]
Deere, D.U. (1968). “Geological consideration” Pages 1--20 from Pande, G.N., & Williams, J.R. (1990) “Numerical Methods in Rock
Rock Mechanics in Engineering Practice. Stagg, K.G., & Mechanics”; Chichester: John Willey. 327p.
Zienkiewicz, O.C. (eds), London: John Wiley & Sons. Papaliangas, T.T., Lumsden, A. and S. Hencher (1996) “Prediction
Greenwood, J.A. & Williamson, J.B.P (1966) "Contact of nominally of in situ shear strength of rock joints”, In EUROC’96, Barla
flat surfaces. Proc. Royal Society, A. 295, pp. 300-319. (ed.), p.143-149, A.A.Balkema, Rotterdam.
Hatcher, R.D. (1995) "Structural Geology: Principles Concepts and Papaliangas, T.T. (1999) - Personal Communication
Problems (2nd Edition)"; Prentince Hall, New York. Patton, F.D. (1966) “Multiple modes of shar failure in rock” In Proc.
Hoek, E, Kaiser.P.K., & Bawden, W.F. (1995) “Support of 1st Congress ISRM, Vol. 1. Int. Society of Rock Mechanics,
Underground Excavations in Hard Rock”; 1 edn. Rotterdam: Lisbon.
A.A.Balkem Ramsay, J. & Lisle, R.J. (2000) “The Techniques of Modern
Locat, Leroueil & Picarelli, (2000) "Some Considerations on the Structural Geology, Vol 3 “, Academic Press. London
Role of Geological History on Slope Stability and the Scholz, C.H. (1990) “The Mechanics of Earthquake and Faulting,
Estimation of the Minimum Apparent Cohesion of a Rock Cambridge”, University Press, Cambridge.
Mass”; In Proceedings of the 8th International Symposium on Zienkiewicz, O.C.; Valliapan, S. and King, I.P. (1968) “Stress
Landslides. Held in Wales. Thomas Telford Ltd. Great Analysis of Rock as a 'no-tension Material”; Geotechnique
Britain. Vol. 18, pp. 56-66.
Louhenapessy, Wilham. G. (2017) “Tunnel Support Design With Zienkiewicz, O.C., & Pande, G.N. (1977) “Time dependent multi-
Finite Element Analysis:”; PPT Presentation at The laminate model of rocks - a numerical study of deformation
Conference and Exhebition of SUMATRA MINER 2017; and failure of rock masses”; Int. J. Numerical and Analytical
22-23 March 2015, Novotel Hotel, Palembang, South Meth. in Geomech., 1 (1), 219--247.
Sumatra, Indonesia .
514
The Application of Active Lime & Cement as Soil Stabilization Materials

Renggo Ginanjar1, Wilham G.Louhenapessy2, Asrul Ahdar3
1
Geotechnical Engineer - PT. Rekayasa Industri,
2
Senior Geotechnical Engineer, AECOM Indonesia and
Lecturing at The Postgraduate Program of Civil Engineering of ISTN
3
Chief Field Engineer, PT. Rekayasa Industri
Email: renggo_ginanjar@rekayasa.co.id, 2wilham.loehenapessy@aecom.com, 2wilham.george@gmail.com,
1
3
asrul_ahdar@rekayasa.co.id
ABSTRACT: This paper is elaborating the application of active lime and cement as soil stabilization material, especially in remoulded
residual soil. The main focus of this paper is measuring the mixing effects of original soil + active lime + cement in soil strength and
workability perspective. The original soil, which is used as filing materials, consists of sandy clay (high water content) with gravel. The
active lime is supposed to have reaction with the clay material and also to reduce the water content, while cement is supposed to have
reaction with sand/granular material. Mixing formula for the materials are: original soil (Sandy silt/clay with sand & gravel) + 10% Active
lime + 3% Cement. The criterion for active lime is it has more than 90% of CaO which will react with the clay minerals such as Sodium or
Potasium.
Keywords: Active lime, cement, clay, sand, stabilization.
1. PREFACE
Weather condition in mountainous area is quite unpredictable. The
morning sky is clear, but heavy rain could come suddenly in the
noon, and can last until evening. This event can happened even
between April to September in the same year, when normally
considered as dry season. Obviously this case was a disruption for
the construction works, which has tight schedule. The original soil,
that'll be used as filling materials, is always wet due to the heavy rain
and couldn’t be compacted by hand or Vibro-Compactor. Several Figure 1.2 Original Soil Compaction Curve
methods have been proposed to overcome this issue, such as; (Source: BPJN 3 Padang Soil Lab)
Maximum Optimum Moisture Content (OMC): 31%
1. Importing granular material or red clay Maximum Dry Unit Weight (γd): 1.34 t/m3
2. Waiting the original soil to drained naturally
3. Speeding up the drying process by mixing the original soil with
As mentioned in abstract section above, the active lime contains
something that makes it quickly dry and well compacted.
CaO more than 90% and was made of limestone which is burnt in the
The last afore mentioned method has been chosen, based on +1000°C stove. After burning process in stove, the burned limestone
consideration of the weather, schedule and cost. The Original become active lime and has different characteristic from its original
material is the residual soil where is located at the mountainous area form as follows:
(+1410 MASL) and has characteristic as follows:  Chemical compostition:
Soil Description : Sandy Silt 1. Minimum CaO and MgO (%): 90
USCS Chart : MH 2. Maximum CO2 (%): 5
Specific Gravity (GS) : 2.59 3. Maximum water content (%): 5
Wet Unit Weight (γm) : 1.34 t/m3  Physical properties:
Dry Unit Weight (γd) : 0.58 t/m3 1. 100% passing 1” sieve.
Water Content (ωn) : 131% 2. Unit weight (kN/m3): 13
Liquid Limit (LL) : 157.3% The active lime-soil reaction can be explained as follow; active
Plastic Limit (PL) : 46.75 lime addition in to the soil gives the calcium (Ca2+) and magnesium
Plasticity Index (PI) : 110.6 (Mg2+) to the soil. These ions tend to change the existing ions in the
Passing no.200 : 66% soil such as sodium (Na+) or potasium (K+). This changing will
reduces the plasticity index significantly and also adsorbs the water
which indicates the chemical reaction between lime & soil. Beside
reducing the plasticity index and water content, the grain size
distribution also changed. The new grain size, which is comprises of
slightly cemented clay aggregate, mostly floculated and become to
sand size particle.
2. METHODOLOGY
Several samples of original soils, active lime and cement were
prepared prior to mixing and testing in laboratory. These materials
have its percentage composition (By weight) in the mixing process
based on its particle size distribution. The percentage of active lime
is 10% and for cement is 3%. Before mixing process, the original
Figure 1.1 Grain Size Distribution soil, active lime & cement have been separated and measured by
(source: PT. SOILENS) weight based on defined percentage. The active lime was broken by
hammer until meet the grain size criteria (max 5 cm). After breakage
of active lime; original soil, water & cement with certain percentage
were poured together in bucket and mixed manually by hand (with
To check the performance of original soil if being compacted, the gloves). The mixing process has been taken carefully since the active
original soil also was tested using standard proctor test as follow: lime is very reactive and feels so hot in the hand. The mixing process
515
stopped after the materials mixed well and could be indicated by 3.3 Unsoaked & Soaked CBR Test
colour and texture changing. The mixed materials were put into the
closed bucket and left for 1 day for curing time. After curing process, Since internal plant road laid on the mixed material fill, the CBR
the mixed materials are ready to be tested. There are several samples value of this materials also was tested as part of the road design. The
of mixed material with different water content added. mixed materials were checked by soaked CBR test using SNI 1744-
The laboratory tests for mixed materials are: 2012 with curing time 24 hours for unsoaked CBR and 96 hours for
1. Modified Proctor soaked CBR. The result as follow:
2. UU Triaxial
3. Unsoaked & Soaked CBR Laboratory
4. Consolidation
5. Grain Size Distribution
3. RESULT AND ANALYSIS

3.1 Modified Proctor Test
The first test for mixed materials was Modified Proctor Test, because
the maximum dry density and optimum moisture content was
obtained by this test.
Figure 3.3 Unsoaked CBR Value

(Source: BPJN 3 Padang Soil Lab)
95% γd max : 1.32 t/m3
Designed CBR : 66%
Figure 3.1 Modified Proctor Test of Mixed Material

(Source: BPJN 3 Padang Soil Lab Test)
Maximum γdry: 1.39 t/m3
Optimum Moisture Content (OMC): 26%.
After mixing process between original soil, lime & cement, we
can see the maximum dry unit weight increased from 1.34 t/m3
become 1.39 t/m3, and the optimum moisture content also changed
from 31% become to 26%.
3.2 UU Triaxial Test
After γdry and OMC was obtained, a constitutive mixed soil sample
could be formed and tested in UU Triaxial apparatus.
Figure 3.4 Soaked CBR Value

95% γd max : 1.32 t/m3
Designed CBR : 118%
From the tests above (Figure 3.3 and 3.4), it can be seen that the
soaked CBR value is higher than unsoaked CBR, this is mostly
related with the curing time process, which the longer curing time
will give the higher result of CBR value. Based on the road design
requirement (Rigid pavement), the minimum CBR at the subgrade
must be at least 6%. Thus we can conclude that the designed CBR of
this mixed materials has strong enough to fulfill the requirement
whether it’s unsoaked or soaked CBR.
Figure 3.2 UU Triaxial Test (Source: BPJN 3 Padang Soil Lab) 3.4 Consolidation
Undrained Cohesion (Cu): 141.7 kN/m2
Internal Friction Angle (φ): 13.08° A constitutive sample of mixed material also was made for
consolidation test (ASTM D 2435), and the result as follow (Figure
The cohesion value which was obtained by UU Test is 141.7 3.5):
kN/m2. Regarding this value, the mixed materials could be Coefficient of Consolidation (Cv) : 0.744 m2/year
considered as very stiff material. Coefficient of Compressibility (Mv) : 0.004 m2/MN
Permeability (k) : 8.927x10-9 m/s
516
Figure 3.5 Consolidation Curve

From the consolidation test, which mv was obtained, the mixed Figure 3.6 Grain Size Distribution
material is classified as very low compressibility materials since the (Source: BPJN 3 Padang Soil Lab)
actual mv is less than 0.05 m2/MN. The prediction of settlement (δ)
will depend on the linear approach with mv as the main parameter, 4. CONCLUSION
  m v H  ' (1) There are significant differences between original and improved
characteristics of the soil. The remarkable characteristic differences
where: are water content, grain size distribution and also shear strength.
δ = Settlement These advantages obviously help the construction works which was
mv = Coefficient of Compressibility supported by this method, regarding the weather and construction
H = Layer thickness schedule. The mixing process at field also has been done with tight
Δσ’ = Effective stress increases QC supervision, especially focusing on water content, uniformity of
materials, maximum lift thickness and also actual field CBR and
3.5 Grain Size Analysis density after compaction. Beside the percentage of lime and cement,
the curing time is also important to increases soil shear strength. 1
The grain size distribution also was tested to identify the particle size day has been taken as curing time since water content has reached the
distribution after mixing (Figure 3.6). optimum, even more days is better (see Figure 3.4 soaked CBR), the
Table 2.1 Grain Size Distribution increases of the strength also have been contributed by cement
contain in the mixed material. The mixed materials at site have been
Sieve Size (mm) Retained (%) Passing (%) treated carefully to maintain its water content during the curing time,
30.48 0.00 100.00 such as providing the tarpaulin sheet on the banks and also providing
the sufficient drainage around the banks. These mixed material in
25.40 0.00 100.00
this paper are purposed as filling material underneath the power plant
19.05 5.16 94.84 facilities such as cooling tower, water treatment plant, warehouse,
12.70 13.95 86.05 road, etc. which required the good bearing subgrade. Since there is
9.53 28.83 71.17 limitation of this paper scope, author suggests the further
4.76 48.99 51.01 investigation/experiments of this mixed material should be done to
get better understanding of active lime & cement mixing behaviour
2.00 54.23 45.77 on each case.
0.42 64.01 35.99
0.01 74.29 25.71 5. REFERENCES
Pan 100.00 0.00
(Source: BPJN 3 Padang Soil Lab) Hardiyatmo, Hary Christady, Dr, Ir., 2013, ”Stabilisasi Tanah untuk
Gravel : 48.99% Perkerasan Jalan” Gadjah Mada University Press.
Sand : 25.3% Wesley, D. Laurence., 2012, “Mekanika Tanah untuk Tanah
Silt & clay : 25.71% Endapan & Residu” Penerbit Andi Yogyakarta.
Balai Pelaksana Jalan 3 Soil Laboratory, 2017 “Laporan Hasil
Already known that the majority of original soil is consist of Pengujian Tanah Timbunan”, Padang.
silt/clay, but after mixing with active lime and cement, it’s altered PT. Soilens Soil Laboratory, 2017, “Soil Investigation Report”,
became granular material (mostly) and it’s aligned with explanation Bandung.
in section I (The Preface) above. Bowles, J. E., 1996, Foundation Analysis and Design, McGraw
HillBook Company, New York, pp. 66–68.
517
Figure 4.2 Applications of Mixed Materials on Site

(Source: Author’s Photograph)
Figure 4.1 Applications of Mixed Materials on Site

(Source: Author’s Photograph)
518
Perbandingan Analisa Perkuatan Lareng dengan Jangkar: Program dan Manual

K. Wawan1 and G.L. Wilham2
1
Lecture at The Department of Civil Engineering , National Institute of Science and Technology (ISTN), Jakarta, Indonesia
2
Finite Element Lecture at The Department of Civil Engineering,
Postgraduate Program, National Institute of Science and Technology (ISTN), Jakarta, Indonesia
dan Senior Geotechnical Engineer, AECOM Indonesia
E-mail: wilham.george@gmail.com1, wilham.loehenapessy@aecom.com1, wawankuswaya@yahoo.com2
ABSTRACT: Dalam kegiatan pelaksanaan pekejaan dilapangan ada hal-hal yang dapat memberikan kelancaran dalam pelaksanaan yaitu
pengetahuan praktis yang di lakukan perhitungan secara manual di bantu dengan program excel untuk dapat mendampingi perhitungan
dengan Software (Program Computer) Slide3D. Pada kasus ini dicoba melakukan kajian kestabilan lereng dilakukan perkuatan dengan
jangkar,dalam kajiannya tahap awal dilakukan analisa dengan program oleh perencana, kemuadian dilakukan analisa dengan cara Fellenius
dengan didang longsor yang dihasilkan dari perhitungan program kindisi statis dan gempa . Dari hasil analisa di peroleh hampir dekat antara
hasil program software (Rocscience’s SLIDE) dan manual (Microsoft Exce)l,untuk kondisi statis: SFsoftware = 1,728 (Fillenius) dan 1,339
(Bishop dan GLE), sedangkan SF untuk hitungan manual, SFmanual = 1,450. Untuk kondisi dinamis, SFsoftware-DYN = 0.861 (Fillenius)
dan 0,998 dan 0.999 (Bishop dan GLE), sedangkan SF untuk hitungan manual, SFmanual-DYN = 1.036. Sementara, besar gaya jangkar
untuuk mencapai harga safety factor, SFsoftware-DYN = 1,095 (Fillenius) dan 1.125 (Bishop) dengan kondisi dinamis gaya jangkar
(software dan manual) adalah sebesar 15 ton. Telah dibandingkan pula hasil analisa 2D dan 3D.
Keywords: Jangkar, Limit Equilibrium, Safety Factor (Faktor Keamanan), Kestabilan Lereng
1. PENDAHULUAN bidang longsor dari analisa ini seperti gambar dibawah ini dengan
bidang longsor terjadi pada lereng.
Analisa pekuatan lereng ini di dasarkan atas hasil analisa yang telah
dilakukan oleh kami dengan penyelidikan Tanah oleh Tarumenegara Table 1 Resume Data Lapisan Tanah
Bumiyasa dan Jurusan Teknik Sipil Universitas Cendrawasih.
Analisa ke satabilan lereng atas pendekatan data hasil uji Tanah Berat isi Sudut ges Kohesi
oleh Tarumenegara Bumiyasa dan hasil uji tanah Jurusan Sipil Lapis Tebal (m) Level
3
T/m dalam t/m3
universitas local, dengan model lereng seperti gambar dibawah ini,
kimiringan lereng berkisar atas perbandingan H = 2 dan V = 5. 1
Sesuai
1.62 14° 1.750
Sesuai
(Gambar 1). gambar gambar
2 2.00 1.68 38° 1.610 +(21-23)
+(15.2-
3 5.80 1.68 36° 2.078
21)
+(14.2-
4 1.00 1.72 39° 2.152
15.2)
+(4.6-
5 10.60 1.68 33° 4.865
14.2)
-7.8 s-
6 12.40 1.68 32° 6.034
+4.6
Gambar 1. Potongan Lereng Sebagai Dasar Analisa Kestabilan
2. DATA TANAH DALAM ANALISA KESTABILAN

Data lapisan tanah yang menjadi data anlisa kestabilan lereng di
dekati hasil penyelidikan pada kaki lereng dengan sebaran nilai
parameter tanah adalah seperti pada Tabel 1 dan Gambar 2
3. HASIL ANALISA PROGRAM DAN GAMBAR

BIDANG LONGSOR
Hasil analisa kestabilan lereng dengan bantuan Program dengan dua
(2) kali analisa kestabilan, nilai kestabilannya di peroleh nilainya
Dari hasil analisa di peroleh hampir dekat antara hasil program
software (Rocscience’s SLIDE) dan manual (Microsoft Exce)l,untuk
kondisi statis: SFsoftware = 1,728 (Fillenius) dan 1,339 (Bishop dan
Gambar 2 Hasil Uji Bor (Borelog) dan Nilai SPT
GLE), sedangkan SF untuk hitungan manual, SFmanual = 1,450. Letak
519
yaitu Gambar 2 dan kami peroleh seperti gambar pendekatanya

dibawah ini (Gambar 5).
Gambar 3. Bidang Kelongsoran Yang mendapatkan SF Terendah

Gambar 5. Pendekatan Bidang Lonsor Sebagai Dasar Analisa
Hasil Analisa (Lihat Appendix A)
4.2 Pembagian Vias* dan Data-data Analisa
Melihat nilai kestabilan hasil hitungan ke dua ini sebesar
SFsoftware_DYN = 0,875 dengan gempa, suatu nilai yang dirasa Dari gambar bidang longsor di atas selanjutnya melakukan
terlalu kecil membuat kekhawatiran sehinga perlu di lakukan pembagian Vias dan mendapatkan data-data sebagai dasar analisa.
perbaikan lereng sedemikian sehingga nilai kestabilan di rasa cukup Bidang longsor akan di bagi menjadi 6 vias, vias No 1 paling atas
(bisa naik). dan vias No 6 paling bawah, kemudian setiap vias di hitung
beratnya sesui dengan berat isi masing-masing lapisan (γ t/m3) dan
4. ANALISA PERKUATAN LERENG di ukur nilai α (sudut kemiringan bidang longsor setiap vias) serta
panjang bidang longsor (L), hasil dari perhitungan di peroleh serti
Dalam menganalisa perkuatan lereng ini adalah untuk meningkatkan
pada table dibawah ini.
tahanan atau perlawanan terhadap longsor sehinga nilainya dirasa
cukup, dasarkan analisa mengacu pada rumus ke stabilan lereng
Tabel 2 Berat dan Sudut kelongsoran setiap Vias.(Slice)
seperti dibawah ini.
C * L  W * Cos * tan  No Wi (ton) αo L (m)

SF  (1)
W * Sin 1 23.612 54.81 10.00
2 14.670 53.13 2.60
Melihat rumus diatas perlawanan longsor adalah sebesar, C*L + 3 68.120 49.24 8.00
W*Cosα*tanφ, dari elemen rumus ini kami memilih meningkatkan 4 15.860 45.00 1.50
nilai W yaitu nilai yang diperoleh dengan melakukan penjangkaran
yang membuat sudut sedemikian sehinga arah gaya menjadi 5 151.320 38.60 16.00
perlawanan longsor dengan rumus Fjrk*Sin Ø + Fjrk*Cos Ø*Tanφ 6 7.661 30.76 4.400
(sketsa gaya-gaya pada Gambar 4).
4.3 Analisa Kestabilan
Atas nilai berat dan sudut kelongsoran pada table diatas dan nilai
parameter kekutan geser tanah sesuai dengan data lapisan tanah
(kohesi C t/m2 dan sudut geser φ o), maka dapat di hitung nilai
ketsabilan seperti dengan sistim table dibawah ini.
Tabel 3 Komponen Kestabilan Lereng

W*sinα
No. C*L (ton) W*cosα*tanφ (ton)
(ton)
1 17.500 19.291 3.386
2 4.186 11.736 6.874
3 16.627 51.567 32.339
4 3.227 11.213 9.083
5 77.847 94.423 76.797
6 26.549 3.861 4.137
Jumlah 145.936 192.091 132.616
Gambar 4. Gaya-gaya sistim penjangkaran.
4.1 Pendekatan Bidang Longsor Sebagai Dasar Analisa
Untuk melakukan analisa ini maka perlu membuat gambar
pendekaran dengan bidang longsor seperti pada hasil analisa diatas
*
Vias = Slice
520
Dari table diatas maka dapat di hitung nilai kestabilan lereng maka besarnya gaya penahannya seperti dari penjumlahan dibawah
dari bidang longsor pendekatan di atas tanpa beban gempa yang ini.
nilainya adalah sebesar.
Fpenahan = ∑(Fjkr SinØ + Fjkr*CosØ*Tanφ) (2)
SF = 145 ,963  132 .616  1,450
192 .091 Dengan memasukkan nilai gaya Fjkr secara coba-coba seperti
pada table dibawah ini, maka di peroleh nilai Fjkr adalah sebesar 15
(lebih besar dari hasil program computer SF = 1,301) ton.
Melihat nilai FS = 1,45 diatas lereng cukup aman, namun jika Tabel 5 Perhitungan Gaya-gaya Penahan
bidang longsor ini di kenakan beban gempa dengan kegempaan 0,2* dari Sistim Jangkar Dalam Ton.
W (20% dari berat, W), maka akan di peroleh SF menjadi sangat No. Ø sin Ø cos Ø Fjkr*sinØ Fjkr*cosØ*tanφ
kecil dengan gempa ini akan menambah gaya pelongsor seperti pada 1 73 0.956 0.292 14.34 3.5458
Table 4 dibawah ini. 2 68 0.927 0.375 13.905 4.556
Tabel 4 Gaya Pelongsor Akibat Gempa. 3 63 0.891 0.454 13.365 4.42
0.2*W 0.28W*Cosα 0.2*W*sinα 0.2*W*cosα*tanφ 4 59 0.857 0.515 12.855 5.014
4.72 2.720 3.860 3.860 54.465 17.537
2.93 1.760 2.350 2.350 Total : 72.002
13.62 8.900 10.310 10.320
3.17 2.240 2.240 2.240 Nilai tambahan gaya penahan dari sistim penjangkatran adalah
57,624 ton yang membuat sedemukian sehinga nilai SF menjadi
30.26 23.670 18.880 18.890
1,25 dari yang sebelumnya 1,036.
1.53 1.320 0.770 0.770
40.610 38.420 38.430 5. KESIMPULAN
Besar keamanan longsor akaibar gempa ini menjadi seperti Dari hasil analisa perhitungan terlihat bahwa perbedaan nilai factor
rumus di bawah ini keamanan (SF) antara hasil software dan analisa manual hampir
145 ,963  132 .616 mendekati (Tabel 6), yaitu:
SF =  1,036 1. Kondisi statis SFsoftware = 1,339 (Bishop 2D), hitungan manual
192 .091  38, 42  38, 43
1,450
(sedikit lebih besar dari SF=0.998) . 2. Kondisi dinamis SFsoftware = 0. 861 (Bishop 2D) hitungan
manual 1,036.
Nilai SF = 1,036 suatu suatu nilai yang sangat kecil perlu di
3. Besar gaya jangkar untuuk mencapai SFmanual = 1,25 dalam
lakukan perbaikan dengan minimal SF menjadi 1,25 dan sehinga
kondisi dinamis besar gaya jangkar manual 15 ton , gaya ini
penambahan gaya untuk menahan kelongsoran adalah sebesar :
Tambahan Gaya = 1,25*(192.091+38.433+38.418) sama dengan dari Software, 15 ton, namun SFsoftware = 1.095
– ( 132.616+145,963) (Fillenius) dan 1.125 (Bishop).
4.
= 57,624 ton. Tabel 6 Summary Hasil Perhitungan
SFAnali SF Analisa Program
SF Syarat
Jangkar
4.4 Perkuatan Dengan Sistim Jangkar. sa Komputer
Gaya
Manual
Prinsip Pekuatan dengan sistim jangkar adalah seperti sket gambar Fillenius Bishop GLE
Fillenius
di bawah ini. 2D 2D,3D 2D, 3D
(Excel)
Tanpa Jangkar
Static 1.45 1.728 1.339 1.339 1.5 -
1.488 1.446
Gempa 1.036 0.860 0.875 0.873 1.0 -
0.998 0.999
Dengan Jangkar
Static 1.825 1.71 1.729 1.731 1.5 15
1.811 1.800
Gempa 1.25 1.095 1.155 1.146 1.0 15
1.125 1.126
6. REFERENSI
RocNews (Spring 2017) “Slide3 - Introducing 3D Limit Equilibrium
Slope Stability Analysis , Rocscience, Toronto, Canada”
technology.pitt.edu (2013) “Microsoft Excel 2013 Fundamentals
Workshop”
Gambar 6 Sket gaya-gaya pada sistim Jangkar.
Gaya penahan yang di timbulkan oleh sistim penjangkaran

sesuai dengan sket gambar di atas dimana kemiringan
penjanjangkaran adalah adalah tegak lurus bidang lereng galian,
521
APPENDIX A: Hasil Perhitungan SLIDE3D (Rocscience)
522
On the Weak Limestone Rock Slope Stability Analysis
G.L. Wilham1
1
PT Signifikan Solusi Integrasi and Lecturer at Civil Engineering Dept.,
Postgraduate Program, National Institute of Science and Technology (ISTN), Jakarta, Indonesia
1
E-mail: wilham@signtegra.com and wilham.george@gmail.com
ABSTRACT: The natural rock slope of Weak Limestone has been studied, modelled analysis and reviewed. This paper describes the Rock
Slope Stability Analysis for the construction of a Generating Station. Four scenarios were analysed i.e.: original proposal of 75o; continuous
slope of 60o; and double step slope of 60o and 60o with a bench; double Steep slope of 65o and 75o with a bench. Slope reinforcements (End
Anchored and High Tensile Mesh) were recommended. The analysis is with Limit equilibrium method, LEM and Finite element method,
FEM.
Keywords: Limestone, Rock Slope Stability, LEM, FEM.
1. INTRODUCTION
This paper describes the Slope Stability Analysis for the
construction of a Generating Station. This paper aims to analyse the
rock slope stability: A steep slope early proposed 75o, and then
propose the following;
a. Steep slope of 60o
b. Double Steep slope of 60o and 60o with a bench
c. Double Steep slope of 65o and 75o with a bench
d.Recommendation of slope reinforcement (End Anchored and
High Tensile Mesh) if needed.
The main analysis conducted is aimed at finding the Factor of
Safety (FS), through the Limit Equilibrium method using
Rocscience’s slope stability Analysis Software Slide 3D and with
Finite Element Method (Griffiths 2015), and a Microsoft Excel
Software.
2. CASE STUDY
2.1 Location
The drilling hole point at the nearest future Slope is at BHX
depicted in Figure 1. Figure 2 showed the Borelog BH-X
Figure 2 An example only for a Borehole BH-X, with lithology of

LIMESTONE that shown the Low Values of RQD (ranging from 0
to 25%)
Figure 3 Site (Previously) Planned Cross Section Profile

Figure 1 Assumed Geotechnical Investigation Area
2.2 Seismic Condition

The Earthquake Strong Ground Motion properties was calculated
using the expected peak ground acceleration for Probability
Exceeding 2% in 50 Years, according to the Indonesian Earthquake
Hazard Map (Ministry of Public Work 2017) for the maximum
credible earthquake at our study site in Eastern Indonesia. The
resulting PGA from the Earthquake Hazard Map was 0.5 g. A
moment magnitude of Mw = 8.0 was used based on W.G. Housner’s
chart (1971). Figure 4 and Figure 5 depict the Indonesian Figure 4: The Peak Ground Acceleration Map in Bedrock (Sb) for
Earthquake Hazard Map and Housner’s Recommendation Table for Probability Exceeding 2% in 50 Years (For our site, The PGA =
Strong Ground Motion. 0.51 g, and the Acceleration Multiplier is 0.5).
523
Figure 8. Forces that work in the Slices*

Figure 5: GW Housner's Table and Calculation for Mw and
Earthquake Duration - Maximum Peak Ground Acceleration at All these methods consider the moment and force equilibrium in
Bedrock each slice. If the moment and force equilibrium is satisfied in
each slice, the overall moment and force equilibrium will be
2.3 Laboratory Test Results satisfied automatically. The basic concept in these methods is the
same; the difference lies in the assumption of the interslice forces. If
Based on the Rock Laboratory and Point Load tests, we found out both moment and force equilibrium are satisfied, the assumption on
that the Limestone in the area is weak Limestone, with UCS = 1800 interslice forces should have only small effect on the factor of safety
kPa, Young’s Modulus, E = 250000 kPa and Poisson’s Ratio, v = obtained.
0.23. In the Morgenstern and Price method, an assumption is made
Through additional analysis of Rockmass Properties we found regarding the relationship between interslice shear and normal
out that the Rock mass is FAIR (Q or NGI Classification), with GSI forces. After obtaining the computer output based on this
= 41. (See Appendix C). assumption, all the computed quantities, including the inter-slices
forces, must be examined to determine whether they seem
2.4 Ground (Soil) Resistivity Test Results reasonable. If not, a new assumption must be made.
Based on the Soil Resistivity test, we found out that the Limestone All these methods can be applied to both circular and
in the area is a weak Limestone, Figure 6 showed the summary of noncircular failure surfaces.
it.
3.2 The failure criterion for intact rock (Limestone)
In rock mechanics practice the use of Hoek-Brown failure criteria is
common. The Hoek-Brown properties (GSI, mi and s) can be
converted into Mohr-Coulomb intact rock properties ( and c).
a. Mohr Coulomb Failure Criterion
The failure criterion for intact rock used is the Mohr-Coulomb
criterion as follows, i = ni tan o + co ; where ni is the normal
stress on the failure plane and o and Co are material constants for
intact rock (Appendix C).
Figure 6. Summary of Soil (Ground) Resistivity Results (Showed

the type of Limestone that is in the range of weak rock, which is
anomaly from the common Limestone found elsewhere.)
3. METHODS OF ANALYSIS
3.1 Limit Equilibrium Method (LEM)
LEM is a method that uses the principle of force equilibrium. This
method of analysis first assumes the field of sliding that can occur,
the field of assumption that is assumed to be circular and non- Figure 9. Algorithm for Failure (Fi) Calculation of the Intact Rock
circular (Figure 7) (Limestone).
LEM is the most popular approach in slope stability analysis.
This method is well known to be a statically indeterminate problem, b. Generalized Hoek-Brown Failure Criterion (constitutive Model)
and assumptions on the interslice shear forces are required to render For a constitutive Model of the Generalized Hoek-Brown (Hoek
the problem statically determinate. Based on the assumptions of the 1995; Eberhardt 2012), the equation is as follows:
internal forces and force and/or moment equilibrium, there are
more than 10 methods developed for slope stability analysis. a
The famous methods include those by Janbu (1957, 1973),   
Spencer (1967) and Morgenstern and Price (1965).  1   3   ci  mb 3  s   0 (1)
  ci 
Where 1 dan 3 are major and minor principal stress, ci is

unconfined compressive strength (UCS) or the maximum axial
compressive stress that a right-cylindrical sample of material can
Figure 7. Failure Line: Circular & Non-Circular+ withstand under unconfined conditions — the confining stress is
+The calculation is done by dividing layers of rock / soil that are in zero, and
the field of landslides into slices.
524
 GSI  100  and Franklin (1984) showed that the Safety Factor for earthquake
mb  mi exp  (2) loading should be no less than 1.0. See (Right Column Above).
 28  14 D 
5. RECOMMENDATION AND CONSLUSION
 GSI  100 
s  exp  (3) It is recommended that using The Slope Angle of 60os without
 9  3D  Shotcrete (but instead will use High Tensile Mesh) and using end-
anchor of 5 m x 5 m, with 15 m long and 100 kN tensile capacity.
Table 1 and Table 2 below details the results of the slope
1 1   GSI    20 
a  exp   exp  (4) stability analysis (72 Run) along with the loadings (static and
2 6   15   3  seismic) and some of the reinforcements that were provided.
Table 3 showed that from the Matrix of Percentage for Factor of
mi is a material constant for the intact rock, GSI (the Geological Safety compared with Required Factor of Safety, the best option for
Strength Index) relates the failure criterion to geological design is using the double slope of 60o dip angle. While a Table in
observations in the field, and D is a "disturbance factor" which Appendix D, is the FEM Summary Results.
depends upon theo of disturbance to which the rock mass has been The above recommendation is still a bit higher than the one
subjected by blast damage and/or stress relaxation. It varies from 0 recommended by Rodrigues (Appendix B), where in his Table 6.5,
for undisturbed in situ rock masses to 1 for very disturbed rock Rodrigues et.al. recommend that (for broken Limestone) a double
masses. slope with first slope angle is 53o and a higher level slope is 45o
angle. The chosen of broken Limestone is due to the Rock
3.3 The Finite Element Method (FEM) Mechanics Lab. Results and the broken/low values of RQD (Figure
2).
The FEM is a numerical method for solving problems of It is recommended to check the tension crack and make sure if
engineering and mathematical physics. Typical problem areas of there are joint rock appears in the weak limestone the dip angle of
interest in engineering and mathematical physics that are solvable the joint rock should be lower than 20o (Appendix E).
by use of the finite element method include structural analysis,
geotechnical engineering (inc’l Slope Stability, Tunneling etc.), heat Table 1 Summary of Without Shotcrete (Instead, now with: High-
transfer, fluid flow, mass transport, and electromagnetic potential. Tensile steel wire and Erosion Control)
The majority of slope stability analyses performed in practice
still use traditional limit equilibrium approaches involving methods
of slices that have remained essentially unchanged for decades.
Then, in 1967, Whitman & Bailey (1967) set criteria for the then
emerging methods to become readily accessible to all engineers
(Louhenapessy 1995). The FEM represents a powerful alternative
approach for slope stability analysis which is accurate,versatile and
requires fewer a priori assumptions, especially, regarding the failure
mechanism. The author will use Griffiths’ FEM Software, name:
SLOPE64 (Griffiths 2004, Griffiths & Lane 1999) to check the
Limestone Rock Slope Stability.
4. ASSUMPTIONS
The calculation of the factor of safety of the given slope was
conducted, which is done depending on the rock material properties
(based on Mohr-Coulomb and/or Hoek-Brown Constitutive Model),
the slope geometry, static (live) load (10 kPa on top of the slope),
Seismic Load (0.255 g) and if necessary the slope reinforcement.
The two loadings that were used were Non-Earthquake (NE)
loading, and Earthquake (E) loading. Earthquake loading was
provided from Figure 4, The Peak Ground Acceleration Map in
Bedrock (Sb) for Probability Exceeding 2% in 50 Years (Peta
Percepatan Puncak di Batuan Dasar (Sb) Untuk Probabilitas
Terlampaui 2% Dalam 50 Tahun) by The Public Work Department
Republic of Indonesia.
Recommended values of safety factors for rock slopes use, SF >
1.5 reference RSNI “Persyaratan Perancangan Geoteknik BSNI
2017. Recommendation of earthquake values of safety factors using,
SF > 1.0 Hynes-Griffin and Franklin (1984). (Figure 10).
Figure 10: Safety Requirement: Above Left Column, shows that

earthquake loading Factor of Safety requirement is 1.5 (Permanent
Condition), whilst other common practice, such as Hynes-Griffin
525
Table 2 Summary (with Shotcrete: Not Recommended, due to sShotcrete is Prone to Groundwater and Can Crack)
Notes: = Limit Equlibrium Method, FEM = Finite Element Method.
Table 3 Matrix of Percentage for FS compared with Required FS

PS: The tables bellow each Slope Profile is taken from the Table 1
526
6. REFERENCES APPENDIX A – Portion of LEM

Eberhardt, E. (2012) "The Hoek-Brown Failure Criterion", Rock
Mech Rock Eng Vol. 45:981–988.
Griffiths, D.V.& Lane,P.A. (1999) "Slope stability analysis by
finite elements", Geotechnique Vol. 49, No. 3, pp. 387-
403.
Griffiths, D.V. (2015) SLOPE STABILITY ANALYSIS BY
FINITE ELEMENTS A guide to the use of Program
Slope64, Geomechanics Research Center Colorado School
of Mines, September 2015
Haynes-Griffin, M.E. and Franklin, A.G. (1984) "Rationalizing
The Seismic Coefficient Method", Dept. of The Army, US
Army Corps of Engineers
Hoek, E., Kaiser, P.K. and Bawden, W.F. (1995) “Support of
Underground Excavations in Hard Rock”, AA Balkema,
Rotterdam.
Louhenapessy, W. G. The Computer Algorithm Failure Index
for joint rock mass in Tunnelling Design, Slope Stability
and Well bore Stability Jurnal Teknik Sipil: No.2,Tahun Figure A.1 Portion of LEM Results (by Slide Software)
ke IX, Jul. 2003, hal.: 277-293, Tarumanagara University,
Jakarta, Indonesia. APPENDIX B – Slopes Recommended for Cuts
Louhenapessy, W. G. & Pande, G.N. (2000) -
Newmo3962_2000: User's Instruction Manual, Internal
Rep.No.CR/1022/00. Dept. of Civil Eng., Univ. of Wales
Swansea, 2000.
Louhenapessy, W. G. (1995) “Undrained Analysis of Rock
Structures”, MSc Thesis, University of Wales, Swansea.
RocNews (Spring 2017) “Slide3 - Introducing 3D Limit
Equilibrium Slope Stability Analysis , Rocscience,
Toronto, Canada”
Rodriquez, A.R., Del Castillo, H. and Sowers, G. (1988) "Soil
Mechanics in Highway Engineering", 2nd Edition, Trans
Tech Publication.
Rockscience - https://www.rocscience.com/software/slide3
RSNI (2017) “Persyaratan Perancangan Geoteknik BSNI 2017
Rodrigues et.al. recommend that (for broken Limestone) a

double slope with first slope angle is 53o and a higher level slope
is 45o angle.
527
APPENDIX C – Rock Classification with Mohr Coulomb APPENDIX D –Finite Element Method: SLOPE64
and Hoek-Brown Intact Rock Properties
528
APPENDIX E –Tension Crack Analysis
529
530
Slope Stability Analyses using Probabilistic Approach

(A Case study in Grobogan, Central Java)
D.R. Febryansyah1, M.P. Yusep1, S.S. Niken1, and D. Noegroho1
1
Department of Civil Engineering, Universitas Sebelas Maret, Surakarta, Indonesia
E-mail: febryansyahdr@gmail.com, ymuslih@yahool.com, nikensilmisurjandari@gmail.com
ABSTRACT: Naturally, soil parameters are uncertain random phenomena in a certain range. This uncertainty is also caused by the process of
soil testing. In a slope stability analyses, the probabilistic approach may become an alternative method to take into account this uncertainty.
The safety factor is optimized with the probability of failure (Pf) that is matching with the soil parameter distribution function using Chi-
Square, Kolmogorov-Smirnov and Anderson-Darling methods. In this study, a number of slopes in Grobogan, Central Java, were analyzed
using Monte Carlo simulation and simplified Bishop with the available soil parameters from laboratory tests. The result indicates that the
greater the safety factor, the smaller the probability of failure. In this study, category of high risk impact (maximum Pf = 25%) produces a
safe slope angle recommendation for maximum slope reaches 5 m with a slope angle up to 55°. The category of low risk impact (maximum
Pf = 50%) produces a safe slope angle recommendation for maximum slope reaches 6 m with a slope angle of less than 60°.
Keywords: Monte Carlo, parameter uncertainty, probability of slope failure.
1. INTRODUCTION factor less than 1 (Pf = P (SF<1)). The probability of failure

standard of slope failures is referred to SRK 2010 (Arif, 2016).
The stability of the slope is influenced by soil parameter, which According to this standard, slope can be categorized as a high
consist of the unit weight, cohesion, and friction angleThe value of risk impact with allowed Pf up to 25% and a low risk impact with
these parameters are uncertainty with the random phenomena allowed Pf up to 50%. Output besides Pf is the reliability index ()
causing random variables. In this situation, the value of paramaters
which has the formula = (SF – 1) / SF , with FS is mean of SF
cannot be determined with certainty and tend to be in a certain range
and SF is standard deviation of SF. The standard value  used for
of values (Ang, 1976). The slope stability analysis is generally
slope stability is 2 (US Army, 1999).
conducted deterministically.
However, this method cannot answer what is the probability of
failure occurance due to the uncertainty of the parameters. 3. METHODOLOGY
Optimization is needed in the deterministic analysis, so that safety The sequence of the study as follows: data validation, fitting model
factor can increase the level of confidence in the calculation. With distribution, geometry modeling, and slope stability analysis with
probabilistic approach, the uncertainty of soil parameters are taken probabilistic approaches. According to the laboratory report, each
into consideration as random variables. soil properties have 100 data. The soils were categorized as clay
This paper presents a case study of slope stability analyses using with very soft to medium consistency. Those data are presented in
probabilistic approach. Slopes and soil properties of the slopes were Table 1.
taken from a Soil Investigation Report in Project of Electric
Transmission in Mojoagung village to Warukaranganyar along 30
Table 1. Summary of Data
km, Grobogan, Central Java (Fig. 1).
Parameter (kN/m3) c (kN/m2) (°)
n 100 100 100
Soil Investigation of Project Electric minimum 14,27 4 3
Transmission, Grobogan
maximum 21,14 70 24
Based on the Table 1, each soil parameter has minimum and

maximum value. The value of each parameter must be validation
with typical value for very soft to medium consistency. Typical
view of slope failure in Grobogan is presented in Figure 2.
Legend:Borehole
Figure 1. Location of study
2. SCOPE AND OBJECTIVES

The soil parameters in this study are unit weight (, cohesion (c),
and friction angle (. Based on the data, between a depth of 1 m to
8 m, no groundwater was found. To find the best fit of soil Figure 2. Typical view of slope failure in Grobogan
parameter distribution function, in this study used Chi-Square,
Kolmogorov-Smirnov and Anderson-Darling methods. Slope 3.1 Data Validation
stability analysis was conducted using simplified Bishop method
with the Entry and Exit procedure. Monte Carlo simulation for Firstly, soil properties validated with typical value of each soil
probabilistic calculation used 2000 trial (Malkawi et al in Khan parameter for clay soil with very soft to medium consistency. The
(2013). The probability of failure is defined as probability of safety result is shown in Table 2.
531
Table 2. Soil Properties
Parameter (kN/m3) c (kN/m2) (°)

data (before validation) 100 100 100
data (after validation) 90 99 98
minimum 14,27 4 3
Frequency
maximum 18,79 26 19
3.2 Fitting distribution

The smallest distribution value between the Chi Square (C-S),
Kolmogorov-Smirnov (K-S), and Anderson-Darling (A-D) methods
is the best fit distribution parameter. Fitting distribution used
@RISK trial version program based on data are as follows:
2
a) Unit weight ( Cohesion (kN/m )
Figure 4. Normal distribution of cohesion

The most suitable distribution for unit weight is lognormal that
shown in Table 3. Based on figure 4, mean value = 14,11 kN/m2, standard deviation =
4,86, skewness = 0 (center).
Table 3. Distribution Value for Unit Weight ()
Method Normal Lognormal Triangular Uniform c) Friction angle ()
C-S 9,9778 9,9778 12,4222 54,9556
The most suitable distribution for the friction angle is lognormal that
K-S 0,0743 0,0934 0,1342 0,3222
shown in Table 5.
A-D 0,5803 0,3984 1,5615 12,6980
Table 5. Distribution Value for Friction Angle ()
In this study, lognormal distribution is utilized. Because the
Method Normal Lognormal Triangular Uniform
value of soil properties is always positive (Consolata, in Shen 2012).
Lognormal distribution for unit weight is shown in Figure 3. C-S 109,2041 51,5102 48,5102 81,8163
K-S 0,2385 0,1051 0,2509 0,4184
A-D 7,5010 1,5259 - 24,4814
Normal distribution for cohesion is shown in Figure 4.

Frequency
Frequency
3
Unit weight (kN/m )
Figure 3. Lognormal distribution of unit weight
Based on figure 3, mean value = 15,92 kN/m3, standard

Friction angle (°)
deviation = 0,91, skewness = 0,65 (+/tend to right).
Figure 5. Lognormal distribution of friction angle
b) Cohesion (c)
Based on figure 5, mean value = 8,02°, standard deviation =
The most suitable distribution for cohesion is normal. Lognormal
5,17, skewness = 3,47 (+/tend to right).
distributions for cohesion parameters cannot be analyzed because if
The recapitulation of statistical parameter values based on the
plotted into data, the minimum relative value becomes too negative
results of best fitting of distribution for each parameter that will be
and does not meet the lognormal distribution requirements (x > 0).
used for slope stability with probabilistic approach is shown in
Distribution value of cohesion is shown in Table 4.
Table 6.
Table 4. Distribution Value for Cohesion (c) Table 6 Parameters for Probability Analyses
Method Normal Lognormal Triangular Uniform Parameter (kN/m3) c (kN/m2) (°)
C-S 8,2222 N/A 13,7778 44,4444 mean 15,924 14,11 8,29
K-S 0,0850 N/A 0,0710 0,3222 distribution lognormal normal lognormal
A-D 0,4814 N/A 0,5139 5,2086 standard deviation 0,91 4,86 5,17
minimum 14,60 5,00 4,00
Normal distribution for cohesion is shown in Figure 4. maximum 17,43 22,00 18,00
532
3.3 Geometry model Based on figure 7, the slope have probability failure (Pf) about
28,8%. Its means that 28,8% x 2000 = 576 have SF < 1, and then the
In this study used 36 variation geometry models.
2000-576 = 1424 have SF > 1. Reliability index () produced (1,176
 Slope height (H) : 2.5 m, 4.5 m, 5 m and 6 m, - 1) / 0,28 = 0,62.
 Slope angle () : 45o to 85o. (with range 5 o) The probability of failure in all model variations is then plotted
with low risk (Pf = 50%) and high risk (Pf = 25%) in the SRK 2010
3.4 Slope Stability Analysis with Probabilistic Approach standard, that shown in figure 8.
The example result of slip surface obtained from slope stability
analyses is shown in Figure 6. In the example is the result for slope
with height is 6 m and angle is 45°.
Low risk
Height (m)
High risk
Length (m)
Figure 6. Slip surface deterministic analysis (H = 6 m,  = 45°)
Figure 8. Probability failure for variations slope model that
Based on figure 6, the safety factor based on deterministic compared with SRK 2010
analysis with mean value of each parameter is 1,185. Then, the
probability of failure from probabilistic analysis using 2000 trial of Figure 8 shows the variation of slope model based on the high
Monte Carlo simulation is shown in probability density function risk impact category with allowed probability of failure up to 25%.
(PDF) and cumulative density function (CDF) in figure 7. Slopes with height 2.5 m are safe to an angle of slope of 85°. Slope
with height 4 m are safe up to 70° slope angle. Slopes with height 5
m are safe up to of 55° slope angle. Slopes with height 6 m are not
safe until the slope angle is 45°. From this result, it is indicated that
the slope angle must be determined to be less than 45°.
For low risk impact category of slope with allowed probability of
failure up to 50%, slopes with height 2.5 m and 4 m, are safe up to
85° slope angle. Slopes with height 5 m are safe up to 70° slope
angle. Slopes with height 6 m are safe with slope angles of less than
60°.
The relationship between the mean safety factor and the
probability of failure is shown in Figure 9 below:
Figure 9. Safety factor and failure probability relationship
Figure 9 shows that the greater the safety factor, the smaller the
probability of failure. According to Shen (2012), the relationship
between safety factor and probability of failure (Pf) is influenced by
the overlap that occurs between probability density function of
resisting and forces in slope failure. The relationship between safety
factor and Pf is inversely proportional. Figure 9 indicate that the
higher the slope, the variation of the probability failure is greater
Figure 7. Probability of failure (H = 6 m,  = 45°)
than the variation in safety factor. The frequency of safety factor of
less than 1.0 for 6 m slopes is greater than 5 m, 4 m and 2.5 m slope.
533
The relationship between the probability of failure and the Arif, Irwandy (2016). Geoteknik Tambang. Jakarta: PT. Gramedia
reliability index on variations in research slopes is shown in figure Pustaka Utama.
10 below. Azizi. (2011). Aplikasi Probabilistik untuk Analisis Kestabilan
Lereng Tunggal (Studi Kasus di PT. Tambang Batubara Bukit
Asam Tbk. Tanjung Enim, Sumatera Selatan). TPT XX
PERHAPI. Mataram: ResearchGate.
Azizi, M., & Handayani, H. (2011). Karakterisasi Parameter
Masukan untuk Analisis Kestabilan Lereng Tunggal (Studi
Kasus di PT. Tambang Batubara Bukit Asam Tbk. Tanjung
Enim, Sumatera Selatan). Prosiding Seminar Nasional AVoER
ke-3. Palembang: ISBN : 979-587-395-4.
Begemann. (1965). The maximum pulling force on a single tension
pile calculated on basis of results of the adhesion jacket cone.
proceedings of the 6th international conference on soil
mechanics and foundation engineering. Montreal.
Bowles, J.E. (1997). Analisa dan Desain Pondasi, Edisi keempat
Jilid 1, Erlangga, Jakarta.
Christian, J., Ladd, C., & Baecher, G. (1994). Reliability applied to
slope stability analysis. Jounal of Geotechnical Engineering,
2180-2207.
Dirjen Pertambangan Umum (1999). Keputusan Menteri
Figure 10. The relationship between reliability index and probability Pertambangan dan Energi Nomor : 555.K/26/M.PE/1995.
failure Tentang Keselamatan dan Kesehatan Kerja Pertambangan
Umum.
Figure 10 is a scatter diagram between the reliability index as the x GeoStudio. (2004). Stability Modelling with GeoStudio. Canada:
axis and the probability of failure as the y axis. Reliability and GeoSlope International, Ltd.
probability indexes are obtained from the results of slope stability Gibson, W. (2011). Probabilistic Methods For Slope Analysis and
analysis with a probabilistic approach on all variations of the slope. Design. Australian Geomechanics Vol 46 No.3.
Then all the points are made trend line so that the trend line can be Khan, F.S, & Malik, A.A. (2013). Probability and Sensitivity
determined from the point of the plot of the reliability index and Analysis of the Slope Stability of Naulong Dam. Pak. J. Engg
probability of failure of all variations. It is indicated that the & Appl. Sci. Vol 13, 54-64.
reliability index and probability of failure have a relationship; that is Laboratory, S. M. (1983). Buku Data Laporan Penyelidikan Tanah
the smaller the probability of failure, the greater the reliability index. Tiang Transmisi 500 KVA Ungaran-Krian Bagian Jawa
This result is consistent with the theory of the relationship index of Tengah. Semarang: Fakultas Teknik, Universitas Diponegoro.
reliability index and probability of failure in Christian (1994). Trend Palisade. (2016). Risk User Guide. NY USA: Palisade Corporation.
line in the Figure 10 produces an empirical formula of the Santoso, E., Arif, I., & Wattimena, K. (2013). Aplikasi Pendekatan
relationship between reliability index (x) and probability of slope (y) Probabilistik dalam Analisis kestabilan Lereng pada Daerah
as y = 3.2418 x2 - 29.967 x + 47.2226 with a value of R2 = 0.9872. Keidakstabilan Dinding Utara di PT. Newmont Nusa
Tenggara. Prosiding TPT XXII PERHAPI 2013, (pp. 283-
4. CONCLUSION 291).
Shen, H. (2012). Non-Deterministic Analysis of Slope Stability
Based on this case study, it can concluded that:
Based on Numerical Simulation. Faculty of Geosciences,
1. The greater the safety factor, the probability of failure is getting
Geoengineering and Mining of The Technische Universitat
smaller. The smaller the probability of failure, the greater the
Bergakademie Freiberg.
reliability index.
Standar Nasional Indonesia. (1995). SNI 03-3997-1995 : Klasifikasi
2. High risk slope category (SRK, 2010) with a permissible
Tingkat Kecuraman Lereng. Indonesia: Badan Standardisasi
probability slope limit of 25% has a maximum slope
Nasional (BSN).
recommendation up to a height of 5 m with a slope angle of 55°.
Tang, W., & Ang, A.-S. (1984). Probability Concepts in
3. Low risk slope category (SRK, 2010) with allowable slope
Engineering Planning and Design. New York: Wiley:
probability limits of 50% has a maximum slope
Decision, Risk, and Reliability Vol 2.
recommendation up to 6 m high with a safe slope angle of less
U.S. Army Corps of Engineer. (1999). ETL 1110-2-556, Risk-based
than 60°.
analysis in geotechnical engineering for support of planning
studies. http://www.usace.army.mil/publications/eng-tech-
5. REFERENCES ltrs/etl-cw.html, Appendix A, page A11 & A12.
Ang, A., Yucemen, M. S., & Tang, W. (1976). Probability Based Whilliam T dan Whitman, Robert V. (1962). Soil Mechanics.
Short Term Design of Soil Slopes. Canadian Geotechnical Whitman, R. (1984). Evaluating Calculating Risk in Geotechnical
Journal, 201-215. Engineering. J. Geotech Engrg, 110: 143-188.
534
Perancangan Proteksi Galian Dalam Dengan Soldier Pile. Studi Kasus: Underpass
Pada Pusat Perbelanjaan Di Kawasan Jakarta Selatan
R. Siska1 dan A.L. Ryan2
1
Fakultas Teknik Program Studi Teknik Sipil, Universitas Katolik Parahyangan, Bandung, Indonesia
2
Fakultas Teknik Program Studi Teknik Sipil, Universitas Katolik Parahyangan, Bandung, Indonesia
E-mail: 1siska@unpar.ac.id, 2alexanderlyman95@gmail.com
ABSTRAK: Pekerjaan galian tanah merupakan pekerjaan dalam ilmu geoteknik yang memerlukan perhatian dan penanganan yang cukup
matang dalam perencanaan maupun pelaksanaanya. Pada makalah ini, peneliti ingin mengetahui kondisi kritis dari suatu sistem galian dalam
sebelum terjadi keruntuhan. Galian dalam pada penelitian ini diperkuat dengan struktur penahan tanah berupa soldier pile. Analisis dilakukan
menggunakan metode elemen hingga yaitu program komputer PLAXIS 2D. Gaya dalam dan deformasi merupakan komponen yang
dianalisis dalam makalah ini. Dari hasil analisis, pada kedalaman pemancangan 7,5 m untuk kondisi short term stability dan long term
stability memberikan momen lentur maksimum masing-masing 397.37 kNm dan 662.34 kNm.
Keywords: soldier pile, metode elemen hingga, metode konvensional, underpass, galian dalam.
1. PENDAHULUAN
Ketersediaan lahan parkir yang cukup pada gedung bertingkat tinggi
(high rise building) menjadi hal yang harus diperhatikan dengan
baik oleh pihak developer. Semakin kompleks struktur bangunan
yang akan dikonstruksi, maka semakin besar juga lahan parkir yang
diperlukan untuk menampung kendaraan yang ada. Secara umum,
lahan parkir dapat berada diatas permukaan tanah dan dibawah
permukaan tanah (basement). Pada makalah ini, pekerjaan galian
yang akan dibahas yaitu galian konstruksi underpass pada akses
menuju lahan parkir di salah satu pusat perbelanjaan di Jakarta
Selatan.
Konstruksi underpass tersebut dimaksudkan untuk
mempermudah akses menuju fasilitas lahan parkir pada basement
serta memperlancar arus lalu lintas di sekitar pusat perbelanjaan di
Gambar 1 Konstruksi Underpass dan Titik Bor NBH-22 Pada
Jakarta Selatan. Sistem perkuatan dan proteksi galian tanah sedalam
Lokasi Pembangunan Proyek
kurang lebih 15 meter menggunakan struktur penahan tanah yang
dalam perancangannya turut memperhatikan jenis tanah, sifat tanah,
lokasi sekitar proyek, kedalaman penggalian tanah yang cukup,
3. METODE ELEMEN HINGGA
metode pelaksanaan penggalian, metode analisis serta pemodelan
yang sesuai dan menyerupai kondisi pada lapangan. 3.1 Model Material Tanah
Struktur penahan tanah yang digunakan pada konstruksi
underpass adalah dengan soldier pile. Pemilihan jenis soldier pile Dalam analisis dengan metode elemen hingga, analisis kondisi
dengan sistem pengeboran menjadi pilihan yang baik sebagai sistem parameter tanah secara umum terbagi atas dua, yaitu kondisi total
proteksi galian pada tanah dasar yang relative padat. Disamping itu, stress analysis dan kondisi effective stress analysis yang masing-
pelaksanaan konstruksi soldier pile akan lebih praktis, cepat, dan masing kondisi berdasarkan tipe material dapat dibagi atas Drained
pengontrolan terhadap mutu beton pada tiang akan lebih mudah dan UnDrained.
(GEC, 2013). Pada total stress analysis (TSA), efek dari tekanan air pori atau
Adapaun tujuan dari penulisan makalah ini yaitu: muka air tanah dapat diabaikan dalam analisis. Hal tersebut
1. Memperoleh besaran gaya – gaya dalam pada soldier pile yang disebabkan oleh efek dari air telah tercakup kedalam parameter total
muncul akibat beban kerja arah lateral; stress dimana parameter air dan tanah ditinjau sebagai suatu
2. Memperoleh dan mengontrol defleksi pada soldier pile akibat material tunggal. Sedangkan pada effective stress analysis (ESA),
beban kerja arah lateral; analisis dilakukan dengan memperhitungkan efek dari tekanan air
3. Memperoleh penurunan muka tanah akibat penggalian. pori atau muka air tanah. Hal tersebut disebabkan oleh efek dari air
belum tercakup kedalam parameter effective stress dimana
2. LOKASI PROYEK parameter air dan tanah ditinjau sebagai dua buah material yang
berbeda.
Data yang dipergunakan dalam penulisan makalah ini berasal dari Secara umum, terdapat beberapa kondisi parameter tanah yang
pembangunan konstruksi underpass pada pusat sentra bisnis di dapat digunakan dalam analisis yaitu: UnDrained A, UnDrained B,
kawasan Jakarta Selatan. Pembangunan konstruksi underpass ini UnDrained C dan Drained.
dilakukan setelah bangunan utama yang terdiri atas pusat
perbelanjaan, hotel dan office tower telah selesai dikonstruksi. Tabel 1 Penentuan Parameter Tanah Berdasarkan Tipe Material
Pembangunan konsturksi underpass ini dimaksudkan untuk (Tjie Liong, 2011)
mengurangi kemacetan lalu lintas pada Jl. Kyai Moh. Syafii
Hadzami (Jl. Terusan Gandaria – Jakarta Selatan) yang disebabkan
oleh antrian kendaraan yang hendak masuk menuju fasilitas lahan
parkir basement. Pada penelitian ini, penulis menggunakan data
hasil uji bor dengan kode titik yaitu NBH-22 yang memiliki
kedalaman bor hingga 29.5 m dengan lokasi muka air tanah pada
keadalaman 8 m.
535
3.2 Penentuan Parameter Tanah Kondisi Total Sterss Analysis 3.4 Model Elemen Hingga dan Tahapan Penggalian
dan Effective Stress Analysis
Proses analisis sistem proteksi galian dalam dilakukan dengan
Berdasarkan tabel 1 desain pada kondisi long term stability, memodelkan sistem galian pada program komputer PLAXIS 2D
parameter tanah pada umumnya akan ditentukan dengan sedemikian rupa mendekati proses konstruksi secara nyata
menggunakan kondisi effective stress analysis. dilapangan yakni dengan membagi pekerjaan konstruksi menjadi
Pada kondisi desain short term stability, parameter tanah pada beberapa tahapan yang dirangkum dalam tabel berikut, yaitu:
umumnya akan ditentukan dengan menggunakan kondisi total stress
analysis. Namun pada penelitian ini, parameter tanah lempung Tabel 4 Pemodelan Tahapan Konstruksi
ditentukan dengan menggunakan kondisi effective stress undrained Tahapan
analysis (ESUA) dengan pertimbangan bahwa terdapat pengaruh Deskirpsi Tahapan Konstruksi
Konstruksi
muka air tanah dan tekanan air pori.
Tahap 1 konstruksi soldier pile dan aplikasi beban kerja
Tahap 2 penggalian pertama hingga kedalaman 1 m
Tahap 3 penggalian kedua hingga kedalaman 3 m
Tahap 4 penggalian ketiga hingga kedalaman 5 m
Tahap 5 penggalian keempat hingga kedalaman 7 m
Tahap 6 penggalian kelima hingga kedalaman 7.5 m
Tahap 7 disipasi tekanan air pori ekses (long term stability)
Adapun model akhir dari tahapan penggalian dapat dilihat pada
gambar berikut:
Gambar 2 Kondisi Tipe Material Dalam Analisis Galian Dalam

Besar parameter tanah yang dipergunakan dalam input pada
program komputer PLAXIS 2D adalah sebagai berikut.
Tabel 2 Rekapitulasi Input Parameter Tanah Berdasarkan Hasil Uji
NBH-22 (Rahardjo, 2016)
Gambar 3 Meshing Pada Model Galian Dalam

4 HASIL DAN DISKUSI
Hasil analisis dari program PLAXIS 2D akan ditinjau pada kondisi
maksimum yaitu pada tahapan galian ke-5 (5th Excavation) untuk
analisis short term stability dan tahap disipasi tekanan air pori ekses
untuk analisis long term stability.
4.1 Deformasi Lateral Soldier Pile
Dari hasil analisis yang dilakukan, dapat diperoleh besaran
deformasi arah horizontal pada soldier pile. Rekapitulasi hasil
3.3 Penentuan Parameter Soldier Pile analisis deformasi maksimum lateral soldier pile dapat dilihat pada
tabel berikut:
Soldier pile yang digunakan pada analisis dengan program komputer
PLAXIS dibedakan menjadi 4 jenis, yaitu berdasarkan variasi Tabel 5 Rekapitulasi Hasil Analisis Defromasi Lateral Maksimum
terhadap kedalaman pemancangan yang masing – masing parameter Soldier Pile
ditentukan sebagai berikut:
Deformasi Lateral
Tabel 3 Rekapitulasi Input Parameter Desain Soldier Pile Panjang Soldier
Maksimum [cm]
Pile [m]
Short Term Long Term
13,125 6.34 12.81
15 6.04 11.15
16,875 5.97 10.63
18,75 5.71 9.73
Adapun profil deformasi lateral sepanjang soldier pile pada

kondisi long term stability dan short term stability yang bervariasi
untuk berbagai panjang soldier pile dapat dilihat pada gambar
berikut ini
536
Dengan melakukan ploting pada grafik hubungan diatas

diperoleh bahwa hasil analisis tidak berada pada range dari syarat
deformasi maksimum menurut Ou et al., 1993.
Hal tersebut mungkin terjadi karena adanya keterbatasan data
hasil pengujian yang diperoleh dan mengakibatkan penentuan
parameter tanah yang rekatif lebih konservatif sehingga
menghasilkan deformasi lateral dinding yang relatif besar.
4.2 Penurunan Tanah
Dari hasil analisis yang dilakukan, dapat diperoleh besaran
penurunan permukaan tanah di sekitar galian dalam. Rekapitulasi
hasil analisis penurunan permukaan tanah dapat dilihat pada tabel
berikut:
Tabel 6 Rekapitulasi Hasil Analisis Penurunan Permukaan Tanah
Penurunan Permukaan
Panjang Soldier
Tanah [cm]
Pile [m]
Short Term Long Term
13.125 1.48 1.21
15 1.42 1.04
16.875 1.40 1.00
18.75 1.35 0.87
Adapun profil penurunan permukaan tanah disekitar galian
dalam pada kondisi long term stability dan short term stability yang
bervariasi untuk berbagai panjang soldier pile dapat dilihat pada
gambar berikut ini
Gambar 4 Profil Deformasi Lateral Soldier Pile dengan Panjang

Bervariasi Saat: (a) L = 13.125 m; (b) L = 15 m; (c) L = 16.875 m;
dan (d) L = 18.75 m
Dari hasil analisis yang telah diperoleh, akan dilakukan evaluasi
menggunakan hubungan kedalaman galian (He) dengan deformasi
maksimum lateral pada dinding (δhm) berdasarkan kondisi short term
analysis.
Gambar 6 Profil Penurunan Muka Tanah di Sekitar Galian Dalam

dengan Panjang Bervariasi Saat: (a) L = 13.125 m; (b) L = 15 m; (c)
L = 16.875 m; dan (d) L = 18.75 m
Dari hasil analisis yang telah diperoleh, akan dilakukan evaluasi
menggunakan hubungan perbandingan kedalaman galian (He)
dengan deformasi maksimum lateral pada dinding (δhm) dan
perbandingan kedalaman galian dengan penurunan maksimum
permukaan tanah (δvm) berdasarkan kondisi short term analysis.
Gambar 5 Hubungan Kedalaman Galian dan Deformasi Maksimum

Dinding (Ou et al., 1993)
Menurut Ou et al., (1993), besarnya deformasi lateral
maksimum pada dinding penahan dapat diestimasi berdasarkan
persamaan
δhm = (0.2% - 0.5%)He (1)
dimana He merupakan kedalaman penggalian yang dilakukan. Nilai
batas atas pada persamaan (1) pada umumnya diperuntukkan pada
jenis tanah lempung dengan konsistensi yang lunak, sedangkan
batas bawah pada persamaan (1) diperuntukkan pada jenis tanah Gambar 7 Hubungan Penurunan Permukaan Tanah dengan
pasir. Deformasi Lateral Dinding (Ou et al., 1993)
537
Pada umumnya, hubungan antara δhm dengan δvm didasarkan 5. KESIMPULAN

pada persamaan
δvm = (0.5 - 0.7) δhm (2) Berdasarkan penelitian yang telah dilakukan, dapat disimpulkan
bahwa:
dimana batas bawah diperuntukkan tanah pasiran dan batas atas 1. Analisis galian dengan metode elemen hingga menggunakan
untuk tanah lempung. Namun terkadang dijumpai juga bahwa pada program komputer PLAXIS 2D dengan menggunakan model
kasus galian pada tanah sangat lunak, besarnya penurunan muka Mohr-Coulumb;
tanah maksimum (δvm) akan sama dengan besarnya deformasi lateral 2. Analisis dilakukan pada 2 kondisi yaitu short term stability dan
dinding maksimum (δhm). long term stability untuk kedalaman pemancangan yang
Sama halnya dengan penurunan muka tanah, dilakukan juga bervariasi yaitu 13.125 m; 15 m; 16.875 m dan 18.75 m;
ploting pada grafik hubungan diatas dan diperoleh bahwa hasil 3. Hasil analisis menggunakan kondisi short term stability
analisis tidak berada pada range dari syarat menurut Ou et al., memberikan hasil yang pada umumnya lebih kecil dari hasil
1993. long term stability baik untuk analisis deformasi lateral soldier
Karena pada grafik hubungan ini melibatkan deformasi lateral pile dan gaya – gaya dalam;
maksimum pada dinding yang hasil perhitungannya tidak tepat, 4. Akibat penggunaan model Mohr-Coulumb, terdapat hasil yang
maka hasil perhitungan penurunan tanah memungkinkan tidak konsisten pada analisis penurunan maksimum muka tanah
menghasilkan nilai yang tidak tepat juga akibat dari penentuan untuk kondisi short term stability dan long term stability;
parameter tanah yang terlalu konservatif. 5. Model Mohr-Coulumb tidak dapat secara akurat mempredikasi
hasil perhitungan tekanan air pori ekses dengan baik sehingga
4.3 Gaya Dalam Soldier Pile untuk kasus galian dalam, disarankan untuk menggunakan
Dari hasil analisis yang dilakukan akan diperoleh output besaran model material lain yang lebih advanced;
gaya dalam yang terdiri atas momen lentur, gaya lintang dan gaya 6. Perlu diperhatikan ketersediaan data tanah yang lengkap serta
normal. Besaran gaya – gaya dalam maksimum untuk variasi penentuan korelasi untuk memperoleh parameter tanah yang
kedalaman yang ada dapat dilihat pada tabel berikut ini baik dan tepat dalam melakukan analisis.
Tabel 7 Rekapitulasi Gaya – Gaya Dalam Soldier Pile Untuk

Berbagai Variasi Kedalamanan
Hasil evaluasi gaya dalam khususnya bidang momen akan dilakukan

dengan cara menentukan terlebih dahulu besarnya kapasitas dari
soldier pile yang digunakan. Soldier pile yang digunakan memiliki
dimensi sebesar 80 cm dengan konfigurasi tulangan longitudinal
yaitu 14 D25. Dengan menggunakan program komputer spColumn,
diperoleh bahwa kapasitas penampang soldier pile (φMn) adalah
sebesar 1229 kNm. Besarnya momen maksimum soldier pile untuk
setiap variasi kedalaman pada kondisi analisis short term stability
dan long term stability apabila dibandingkan dengan kapasitas
penampang dapat dilihat pada gambar berikut ini
6. DAFTAR PUSTAKA
Geothechinal Engineering Center (GEC), (2013). Manual Pondasi
Tiang. 4th ed. Deep Foundation Research Institute,
Parahyangan Catholic University, Bandung, Indonesia.
Ou, Chang Yu, (2006). Deep Excvation: Theory and Practice.
Taylor & Francis Group, London, U.K.
Rahardjo, Paulus P. (2016). In Situ Testing and Soil Properties
Correlation 2nd Edition. Universitas Katolik Parahyangan,
Bandung, Indonesia.
Tjie-Liong, Gouw. (2011). Deep Excavation Failures, Can They Be
Gambar 8 Bidang Momen Soldier Pile Berbagai Variasi Prevented ?, International Symposium On Sustainable
Kedalaman Pada Kondisi: (a) Short Term Analysis dan (b) Long Geosynthetics and Green Technology for Climate Change 2011
Term Analysis (SGCC 2011), 7-8 December, 2011, Bangkok.
538
Study of Slope Failure and Reinforcement Analysis to Restore and Increase the Slope
Stability in Flores, East Nusa Tenggara Using Plaxis 2D and 3D
Rendra Priatno1, Ali Iskandar2, and Jo Lian Huat3.
1
Geotechnic Engineer, PT. Tarumanegara Bumiyasa, Jakarta, Indonesia.
2
Director of Engineering, PT. Tarumanegara Bumiyasa, Jakarta, Indonesia.
3
President Director, PT. Tarumanegara Bumiyasa, Jakarta, Indonesia.
E-mail: engineering@tarumanegara.co.id
ABSTRACT: The steeper modification of a natural slope due to construction demand required additional reinforcement to minimize the
slope failure potential. The excavation and removing of sloping soil create new internal instability which means slope failure potential
increase and must anticipated with additional reinforcement. The reinforcement as internally stabilized systems intended to decrease the
stress release during and after excavation activity. Steeper modification of natural slope does need proper geotechnical design and
construction method as well post construction monitoring. In Flores, East Nusa Tenggara, the slope fails after end of excavation with a height
of about 20m and a slope angle of 60o without reinforcement. Internally stress release of slopes increases gradually and causing slope
failures and cracks in the around of collapsing area. To restore the modification slope is required the reinforcement of the slope such as
ground anchor, soil nailing and crib wall as additional support the slope stability. Comprehensive analysis of slope reinforcement is done
using finite element method in Plaxis 2D and 3D program. The design parameters were obtained from the results of the soil investigation in
the study area and field observation. According to slope analysis model for slope failure condition with crack around the collapsing area,
safety factor is less than 1. The process of analysis is done by back fill slope which serves as counter weight then done excavating gradually
and follow by installing the combination of ground anchor, soil nailing, and crib wall. The length of ground anchor and soil nailing is
designed to exceed the slope failure pattern that occurs and ends in the rock layer so that the final SF obtained is more than 1.5 or is in a safe
condition. The stage of construction should be implemented based on the on engineering design of slope stabilization to avoid failure.
Keywords: Slope stability, finite element method, ground anchor, soil nailing.
1. INTRODUCTION 2. PREMILINARY STUDIES

The study location is located Flores - East Nusa Tenggara with 2.1 Regional Geology
° °
existing steep slope about 30 to 45 (Van Zuidam, 1983). To
achieve the elevation level of the plan, the existing slopes are Based on the geographical position, the study location is in the
°
excavated at about 20 m with an angle of 60 . Based on preliminary Physiographic Zone of the Lesser Sunda Islands, where the Small
study and field observation, slope conditions are unstable. This is Islands physiographic zone includes Bali, West Nusa Tenggara and
shown to be found some signs of deformation of the slope, including East Nusa Tenggara (Van Bemmelen, 1949). These zones are small
initial crack above the site, repaired road and tilted electric pole due islands formed by tectonic activity of the Indo-Australian plate that
to sliding, and road rectified due to slope deformation. So that in the moves northward, urging the Eurasian plate. The deformation
construction Stage of stripping required reinforcement of slopes in resulted in the basement of the continent which was originally below
the form of crib wall, soil nailing, and drainage at the edge of the the average surface of the land to be raised and formed a group of
highway to stabilize the slope and reduce erosion on the slope. Lesser Sunda Islands. In addition, increased volcanic activity
In the construction Stage, the initial work undertaken is contributed to the formation in some parts of the archipelago
stripping existing slopes to the elevation level of the planned because of its position on the Sunda Volcanic Arc.
buildings with temporary road below the main road. High The study location is composed by Tertiary Volcanic Rock
°
excavation slopes about 20 m with an angle reaching 60 . During the Formation. The stratigraphic aspect of the project area is structured
by one formation, the Dacitic Tuff Unit, based on the Geological
slope stripping process, the planned slope reinforcement is not
carried out. These conditions make the slope becomes prone to slope Sheet Map of Komodo, Nusa Tenggara (Ratman et al., 1978). The
failures. Finally, the potential slope failure becomes true. The Dacitic Tuff Unit (Tmdt) generally consists of volcanic sediment
deformation of the slopes is caused by the slope loss the axial and material, composed by the dacitic tuff mostly bedded and partly
massive, contain intercalations of green tuff, calcareous tuff,
lateral extension (stress release) due to the excavation, so it is easy
to be triggered for slope failures. limestone, tuffaceous sandstone, breccia, and lava. The composition
From the slope failure pattern it can be classified into Multiple of the lava is partly dacite and partly andesite. This unit is estimated
Rotational Debris Slides. This is because the slope failure geometry to be middle Miocene with the marine deposition environment. The
Dacitic Tuff (Tmdt) unit is deposited not aligned with the volcanic
consists of one crown and scrap but there is also a fracture that
forms a crown behind the avalanche. Therefore, the situation is very rock units (Tmv) composed of lava and the dacitic breccia and the
dangerous around the location if the construction activity is laminated Limestone (Tml) units composed by layered limestone
continued without any mitigation to the collapsing area. Response to with intercalations tuffaceous limestone, quartz sandstone, tuffs, and
conglomerates.
the slope failure slope should be done by restore the slope before the
construction continued. During the process of restoring the slope, it
is necessary to observe periodically on the fractures formed by slope 2.2 Site Observations
failures. The Slope reinforcement that can be done, such as Based on field observations before constructions phase started, the
backfilling and gabion installation at the bottom of the slope to the slope conditions are unstable. This is shown by several signs of
collapsing area serving as counterweight, installing an anchor at the slope deformation, including initial crack above the site, repaired
top of slope, and installing soil nailing on the slope to the elevation road and tilted electric pole due to sliding, and road rectified due to
of the plan. slope deformation. So that in the stage of stripping construction
need slope reinforcement such as crib wall, soil nailing, and
drainage on the edge of the highway to stabilize the slope and
reduce erosion on the slope.
539
Slope Failure
Cracks
Figure 4: Land Situation and vertical cross section.

Figure 1: Site condition from Google Earth observation (Source:
Street View of Google Earth) Based on the result of monitoring the slope deformation carried
out at 6 monitoring point, it can be seen that the slope deformation
in the cracks has decreased. The slope deformation observation for 9
days after the slope failure shows a downward trend until it stops.
From the result of slope deformation, it can be concluded to be
classified as Extremely Slow, with an average deformation speed of
9.65x10-9 mm/sec  0.3 mm/year.
Figure 2: Initial crack above the site project
3. POST-SLOPE FAILURE OBSERVATION

After the excavated slope collapsed, there are appear cracks in some
location around the crown. This is an indication that the slope is still
in unstable condition and there is a possibility of further slope
failure potential. So that it is necessary to conduct intensive
observation of slope deformation during the process of slope
reinforcement until the construction phase is complete. Figure 5: Classification of sliding deformation rate (Cruden and
Varnes, 1996)
4. DATA AND METHOD

Approximate soil parameter and cross section for slope stability
analysis assumed from the geotechnical investigation, consists of 11
deep boring and laboratory test. The slope stability analysis is based
on finite element method, which is performed on the Plaxis 2D and
Plaxis 3D program.
To calculate the reinforcement for the slope stability design, the
soil parameter required as input into the program for analysis. The
soil parameters are interpreted based on the result of the
geotechnical investigation. The analysis carried out by considering
one condition, is a long term or drained condition in Plaxis 2D and
Plaxis 3D, and uses the main parameters such as (c ', ', E ', ') for
Figure 3: Post-Slope failure Condition. cohesive soil and uses drained material for non-cohesive soils with
parameters (c ', ', E ', ') for the modulus value in the excavation
are used unload-reload modulus whose value is assumed to be three
to four times the effective modulus.
540
Table 1. Soil Parameter
Soil ɤunsat E’ref v cref ɸ

Layer [kN/m³] [kN/m²] [-] [kN/m²] [°]
1 Layer-1 18.00 52700 0.30 20 35

2 Layer-2 20.00 90000 0.30 44 38
With the local collapsing of the slope in the field, then the back
analysis of the soil parameters is reduced on the deformation plane
by adding an interface line on the layer based on the slope failure
and crown patterns that occur around it. The existing safety factor
obtained will be lower than the equilibrium condition (SF = 1).
The reinforcement alternatives used in slope back analysis are
fill material, Ground Anchor, Soil Nailing, Crib Wall, and Gabion.
Table 2. Reinforcement Parameter
Mohr- ɤunsat E’ref v cref ɸ 

Coulomb [kN/m³] [kN/m²] [-] [kN/m²] [°] [-]
Fill
1 13.00 37000 0.30 20 35 -
Material
L Fmax, Fmax,
Length EA
Structure Spasi Type Tens Comp
[m] [m] [kN/m] [kN] [kN]

Ground Figure 7: Back analysis of post-slope failure phase, Top: Plaxis 2D,
23 2 Elastic 1.399E+05 - -
Anchor Bottom: 3D
Soil Nailing 20 2 Elastoplastic 8.042E+04 314 1 2. The second phase is the early treatment phase where to backfill
the fill material that serves as counterweight up to +28.7m and
d Type EA EI ν installation of gabion on the toe of counterweight. At this stage
Structure the main road is shifted and the road temporarily stopped.
[m] [kN/m] [kNm2/m]
Crib Wall 0.201 Elastic 1.990E+06 6729.89 0.15
Layer 1: Gravelly
Sand, N = 36-50
Layer 2: Medium
Weathered – Fresh
Dacite, N = >50
Figure 6: Slope profile and cross section.
The Slope stability improvement analysis shall meet the

requirement for the safety factor condition with a factor value of 1.5.
5. RESULT AND DISCUSSION

5.1 Back Analysis Result
The back analysis in Plaxis programs is done in various phases. The
phases are the construction stage that will be done in the field if the
analysis results have been qualified. The slope geometry is based on
post-slope failure cross section.
1. In the early phase is the initial load or gravity load stage in the
post-slope failure condition. This stage still includes the traffic Figure 8: Early treatment phase, add fill material for counterweight
load of main road and the temporary road to the location. and gabion, Top: Plaxis 2D, Bottom: 3D
541
3. The Slope reinforcement phase 1. The installation of ground 5. Slope reinforcement phase 3, the installation of soil nailing and
anchor and crib wall for slopes close to the main road. crib wall down to +9.8m elevation and external load (traffic on
the main road has been running normally).
Figure 9: Slope Reinforcement Phase 1, Top: Plaxis 2D, Bottom:

3D. Figure 11: Slope Reinforcement Phase 3, Top: Plaxis 2D, Bottom:
4. Slope reinforcement phase 2, the installation of soil nailing and 3D.
crib wall down to +18.5m elevation. In the installation of soil 6. The final phase is applying earthquake load ah= 0.288g.
nailing and crib wall, counterweight excavated gradually then
follow by installing the soil nailing and crib wall.
Figure 10: Slope Reinforcement Phase 2, Top: Plaxis 2D, Bottom: Figure 12: Earthquake load phase, Top: Plaxis 2D, Bottom: 3D.
3D.
542
5.2 Early Warning System

factor result of Plaxis 2D and Plaxis 3D is SF of Plaxis 3D larger
For Further construction and post-construction, it need to install than SF of Plaxis 2D.
early warning system (EWS) for slope failure hazard mitigation. The Based on the result of back analysis, we can conclude that the
Early warning system instrument consists of crackmeter, additional reinforcement can restore the collapsing area and increase
extensometer, tiltmeter, potensiometer, and accelerometer which can the slope stability. The synchronization between geotechnical slope
be monitored remotely anytime. The installation of Early Warning stability design and construction activity is needed to minimize the
system (EWS) on the slopes as early warning and can minimize the slope failure potential.
risk in case of significant slope deformation or failure. In addition, the installation of Early Warning system (EWS) on
the slopes as an early warning and can minimize the risk in case of
6. CONCLUSION significant slope deformation or failure.
The slope failure that has occurred in site, caused by the slope loss
the axial and lateral extension (stress release) due to the excavation 7. REFERENCES
without installing additional reinforcement to the slope. According Brinkgreve, RBJ, PLAXIS 2D ver 8: User’s manual, Delft
the analysis of restoring collapsed slope need backfill serves as University of technology & PLAXIS b.v. Belanda.
counter weight, excavating gradually, and follow by installing Carter, M. & S.P. Bentley, "Correlations of Soil Properties".
combination of ground anchor, soil nailing, and crib wall. The result CGGS, 2012. ”Manual Kestabilan Lereng”, Universitas Katolik
of back analysis with Plaxis 2D and Plaxis 3D, the safety factor of Parahyangan, Bandung.
back analysis of various phases is as below. GEC, Insitu Testing and Soil Properties Correlations, Insitu 2001,
Bali.
Table 3. Safety Factor Result Holtz, Robert D., William D.Kovacs & Thomas C.Cheahan, "An
Safety Factor Safety Factor Introduction to Geotechnical Engineering", 2nd Edition.
Phase
(2D) (3D) Hunt, Roy E., 2005, Geotechnical Engineering Investigation
Post-Slope failure 0.99 1.07 Handbook, Second Edition”, Taylor and Francis Group.
Early Treatment 1.70 1.62 Ratman, Nana & Yasin, Aswan, 1978, Geologic Map of Komodo
Slope Reinforcement 1 1.67 1.94 Quadrangle, Nusa Tenggara, Geological Survey of
Slope Reinforcement 2 1.96 2.37 Indonesia, Indonesia.
Final: Slope Reinforcement 3 1.62 3.22 Sanglerat, G., "The Penetrometer and Soil Exploration", Amsterdam
Earthquake load 1.08 1.47 1972.
Sanglerat, G., Gilbert Olivari & Bernard Cambou,"Practical
The result of safety factor from Plaxis 2D and Plaxis 3D, the Problems in Soil Mechanics and Foundation
safety factors meet the requirements SF=1.5 in final condition, and Engineering",Amsterdam, 1984.
for dynamic condition (earthquake load), safety factor of Plaxis 2D Van Bemmelen . R. W, 1970. “Geology of Indonesia. Vol 1.
close to the safety factor requirement, SF=1.15, but safety factor of General Geology Adjaeent Archipelago,” Government office.
Plaxis 3D meets the requirement. The difference between the safety The Haque.
543
Study on The Effects of Vibration due to Pile Driving by Empirical Formulas

and Continuous Monitoring
Stefanus Diaz Alvi1, Ricky Setiawan2, Andy Sugianto3, Paulus P. Rahardjo4
1
Engineer, PT Geotechnical Engineering Consultant, Bandung, Indonesia
23
3
4
Professor of Geotechnical Engineering, Universitas Katolik Parahyangan, Bandung, Indonesia
E-mail: diaz.stefanus@gmail.com, rickysetiawan.geo@gmail.com, andysugianto.sorong@gmail.com, rahardjo.paulus@gmail.com
ABSTRACT: Pile driving using hydraulic hammer will induce large vibration so that the effects of pile driving needs to be analyzed to
avoid building damage and malfunction of the sensitive device. This paper discusses a case study in South Sumatera, Indonesia regarding the
effect of pile driving on a sensitive gas plant instrument and building surrounding the project area. The effect of vibration due to pile driving
is calculated using empirical formulas (Attewel & Farmer, 1973 and J.M. Ko et al, 1990). However, there is correction factor of the empirical
formulas that are dependent to soil types. Continuous monitoring has been carried out in the project area by using geophone sensors to
measure vibrations and PDA (Pile Driving Analyzer) to measure actual energy of driving. Peak particle velocity (PPV) from empirical
formulas are approaching PPV value from monitoring test at the same energy of pile driving. The results will be used to back analyse to
modify the correction factor of the empirical formulas.
Keywords: Peak Particle Velocity (PPV), Pile Driving, Vibration Monitoring.
1. INTRODUCTION
When pile driving operations are carried out near other structures,
several technical investigations must be carried out to evaluate the A
vibration impact of the pile driving. This investigation must take
into account the ground vibration due to pile driving, the BH-08
Sensitive Device
transmission through the soil media to the nearest structure and how
the vibration will affect the structure or sensitive devices. BH-10
In general, the type of vibration propagation through the soil is
more predictable than its effect on the structure. This is because it
BH-01 B’ A’
depends on the quality of the structure, foundation system and B BH-07
configuration. Ground vibrations that can affect the structure include: BH-02 BH-04
 Soil settlement and foundation due to compaction. BH-03
 Vibration in the structure due to impact vibration.
 Vibration in sensitive devices. BH-05 BH-06
Determination of whether the structure will experience a
decrease or horizontal vibration due to pile driving can be estimated. Figure 1. Geotechnical Investigation Location
Identification of local soil conditions can provide an indication of
the potential decline of buildings. The movement of buildings or Soil stratification on this project is represented by Section A-A’
sensitive devices near pile driving location is a function of soil type, and Section B-B’ as follows.
vibration frequency, and amplitude.
Therefore, when the pile is placed near the existing structure or
BH-08 BH-10
sensitive devices, it is the responsibility of the engineer to evaluate
N-SPT N-SPT
its potential for damage to the structure and the devices. For this 0 10 20 30 40 50 60 0 10 20 30 40 50 60
reason the engineer must understand how ground vibrations are 0
15
0
3
caused by the pile-up process and how soil parameters such as 4
2
5 5 5
damping can give effect. 11
9
4 Silty Sand 11
On this paper, case study is taken from a project located in South 10 13 10 9
22
Sumatera, Indonesia. This project uses a type of pile foundation with 9
12
Depth [m]
Depth [m]
15 13 15
a 400 mm diameter spun pile. The hydraulic hammer for driving that 10
46
27
will be carried out has vibration issues that have the potential to 20
11
20 17
20
interfere with the performance of a sensitive instrument that is near 60
60 60
that location and existing structure. The closest distance between the 25 Coal 60
60 25 60
60
60 60
pile and the instrument is about 186 m. Vibration monitoring is Carbonaceous 60 51
30 60 30
carried out to estimate the amount of vibration arising from the pile Clay 60 54
Clay
60
driving of a 400 mm spun pile using the hydraulic hammer. Besides 35 35
that, PDA monitoring was conducted to determine the actual energy
40 40
of the hammer that entered the pile foundation during pile driving.
Figure 2. Section A-A’
2. PROJECT DATA
Soil stratification on section A-A’, which are represented by
2.1 Soil Condition BH-08 and BH-10, shows that around 22 m of the top layer is loose
to medium silty sand. Layer 2 is coal with 6 m thickness. Layer 3 is
Before analyze the vibration effect due to pile driving, it is a dark brown brittle carbonaceous clay. Layer 4 is a layer of clay
important to understand the soil conditions at the project site. with very hard consistency. In this section, it is shown that coal
Geotechnical investigation is consists of drill testing and SPT layer tends to be uniform even though there are different elevation.
(Standard Penetration Test). The location of the test is shown on
Figure 1.
544
BH-02 BH-03 BH-04 BH-06 BH-07

N-SPT N-SPT N-SPT N-SPT N-SPT
0 10 20 30 40 50 60 0 10 20 30 40 50 60 0 10 20 30 40 50 60 0 10 20 30 40 50 60 0 10 20 30 40 50 60
0 0 0 0 0
10 5
3 8 Sandy Silt 3
11 Silty Sand 8
7
1 22
5 8 5 5 5 7 5
7 8 11
Clay 20 5 9
7
17
26 8
10 10 13 10 12 10 10 7
31 7
Sandy Silt33 9 5Silty Sand
10
Carbonaceous Clay Silty Sand 7
9 14
Depth [m]
Depth [m]
Depth [m]
Depth [m]
Depth [m]
45 9
15 33 15 15 14 15 15 15
60 12
52 Coal 60 12 9
46 11
20 20 20 14 20 20 22
60 25 20
55 14
49 Clay 41 Coal 52 19
46
25 35 25 30 25 50 25 17 25
55 27
60 50 26
Carbonaceous Clay 59 45 60
60 Clay 51 60 Coal 60
30 30 30 30 60 30 7
60 36
60 60 30
60 60 60 60
35 35 51 35 41 35 35
Clay 60
54 51 60 Clay
Clay 52
60 52
40 40 40 40 40
Figure 3. Section B-B’
Section B-B’ shows different soil stratification from Section A- Figure 4. Correlation between Distance from Piling Operation and
A’ where the coal depth is vary. In the upper layer there is sandy silt Peak Particle Velocity (Eurocode 3)
and silty sand. The layer below is a coal layer that is brittle with
thicknesses ranging from 6-10 m at various depths. The layer below The other empirical formulas derived by J.M. Ko et al. (1990)
is carbonaceous clay. The layer is found between the layers of hard are as follows:
clay. Hard soil layers are found at different depths in each area.
exp  α  (r  1)
1
v = 70  (2)
2.2 Piling Method r
where :
The foundation system on this project is 400 mm diameter spun pile. v = particle velocity (mm/sec)
The pile driving will be done with a hydraulic hammer with hammer r = distance from the vibration source to the location
specifications as follows. observed (m)
 Hammer type : NH-70 α = damping coefficient (depends on the soil), about 5-20%
 Hammer weight : 7 tons
 Energy efficiency : 60-70%
 Blow speed : 25 blows / minute 4. STUDY ON THE EFFECTS OF VIBRATION BY
 Hydraulic pressure : 210 kg/cm3 CONTINUOUS MONITORING
Piling with a hydraulic hammer will induce great vibration so
that the vibration impact of the piling done needs to be analyzed so 4.1 Vibration Monitoring
that the pile driving does not cause damage to the building, Vibration monitoring is carried out by a geophone sensor as an
malfunctioning of the existing gas plant instrument, and failure of accelerometer. The geophone is used to recording particle velocity
concrete settings after pouring concrete on the need for casting the (v) and frequency (f). The data is displayed against time so that
shallow foundation in the surrounding area. variations in particle velocity or frequency with time can be
obtained along piling. The reading of particle velocity and frequency
is done in 3 axis directions (x, y, and z) so that analysis can be
3. STUDY ON THE EFFECTS OF VIBRATION BY carried out on the three directions. From the measurement of the
EMPIRICAL FORMULAS three directions, the resultant magnitude of the three directions will
Peak particle velocity (PPV) values can be calculated by empirical be calculated.
formula from Attewel & Farmer (1973) or J.M Ko (1990). Attewell Vibration monitoring is carried out using 2 sensors. The first
and Farmer (1973) conducted an analysis of the results of tool is used to measure vibration in the sensitive device (existing
measurements of vibration due to the pile driving on several GTC). The second tool measures in 4 different directions with
variations of soil and pile type. They suggested that the peak particle varying distances to obtain a vibration relation to the distance from
velocity (v) due to pile pile driving could be predicted with the the point of piling. From the test at each point, the particle
following formula. displacement, particle velocity, acceleration, and frequency will be
obtained. The monitoring results will be concentrated on the final
Wo set condition as a critical condition.
v= C (1)
r Pile driving is carried out in 2 stages. In the first stage the piling
where : is located at 200 m from the existing GTC where the vibration
v = peak particle velocity (mm/s) measurements were carried out at distances varying from the point
C = constant value of pile driving (30 m, 40 m, 60 m, 90 m, 115 m, 120 m, and 200 m).
Wo = hammer energy (Joule/blow) The measurements at various distances is used to evaluate safe
r = distance of the point of purging to the location being distance. Pile driving in the second stage was carried out at New
reviewed (m) GTC Unit 1 which was as far as 186.19 m from the existing GTC.
The equation states that the partivle velocity at any point is The pile that was placed on the New GTC Unit 1 was the location
proportional to the square root of hammer energy and inversely with the closest distance between the location of the existing GTC.
proportional to the distance of a point at the stake location. This In other words, the measurement of vibration due to pile driving at
vibration reduction has a linear tendency to distance in the the New GTC Unit 1 will obtain vibration data in critical conditions.
logarithmic scale. The proposed equation is then adopted by The data to be used in this paper only refers to the data taken in the
Eurocode 3. The constant C is used to express the variation in soil final set condition.
conditions and type of pile type. The value of C ranges from 0.25- The location of stage 1 piling point (loc pilling) and stage 2
1.50. Eurocode 3 recommends a value of 0.7 for piling vibratory. At piling point (New GTC Unit 1) to the existing GTC location can be
impact piling, the value of C is around 1.0 for solid or firm soil and seen in Figure 5 and Figure 6. The value of peak particle
0.5 for loose or soft soil. Here is the relationship of the distance of displacement, peak particle velocity, peak particle acceleration, and
one point from the location of the location of the peak particle the maximum frequency measured from monitoring is shown in
velocity according to Eurocode 3. Table 1 and Table 2.
545
falling height of 1.28 m. The height of the hammer falls when

EXISTING GTC testing is set at 0.32 m.
The energy produced by the hammer blow can be calculated
theoretically by multiplying the weight of the hammer by the height
of falling hammer. However, the energy received by the pile
foundation will be smaller than theoretical energy. The reduction in
hammer energy (efficiency) is calculated by dividing the amount of
PILING LOC energy measured on the pile foundation divided by the theoretical
maximum energy in percent. The PDA can monitor the amount of
energy received by the pile foundation by doing integral to the
NEW GTC UNIT 1
product measuring force (f) multiplied by velocity (v) against the
time up to 2L / c which is formulated as follows.
E  f.v.dt  (3)
Figure 5. Piling and Existing GTC Location The pile driving was carried out with a hydraulic hammer where
7 tons ram weight and 0.32 m ram stroke, so that theoretical energy
was obtained at 2.24 t.m. PDA test equipment shows that with the
same hammer weight and falling height the actual energy is in the
range of 1.2 - 2.01 (see Figure 7) and is taken on an average of
1.501. Energy efficiency was taken based on the actual average
energy divided by theoretical energy, so that the energy efficiency
E D A
obtained in the hydraulic hammer was obtained by 67%.
(105 meter) (90 meter) (30 meter)
PILING LOC
Energy vs Blow Number

F C B 2.5
(120 meter) (60 meter) (40 meter)
MR
(42 meter)
2.0
Energy [t.m]
1.5
Figure 6. Vibration Monitoring Location
1.0
Table 1. Monitoring Data (First Stage) 0.5

Peak Peak
Peak Particle Maximum
Distance Particle Particle 0.0
Displacement Frequency 0 10 20 30 40 50 60 70 80 90 100 110 120
[m] Velocity Acceleration
[mils] [Hz] Blow Number
[mm/s] [m/s2]
30 2.37 5.35 0.978 37.5 Actual Energy Theoretical Energy
40 4.41 3.44 0.588 37.5 Figure 7. Energy Measured from PDA Test
60 2.64 1.14 0.120 23.5
90 1.26 0.67 0.095 15.5
115 2.91 0.49 0.048 20.5 5. THE EFFECTS OF VIBRATION DUE TO PILE DRIVING
120 2.24 0.36 0.052 27 ON SENSITIVE DEVICE AND EXISTING BUILDING
200 1.63 0.29 0.104 66
5.1 The Effect of Vibration to Sensitive Device on Existing GTC
Table 2. Monitoring Data (Second Stage) Around the project site, there is a gas plant that is very sensitive to
Peak Peak vibration. If the vibration received by the gas plant exceeds the
Peak Particle Maximum
Distance Particle Particle allowable limit, there will be a shutdown of the gas plant. The
Displacement Frequency
[m] Velocity Acceleration closest distance between the pile and the gas plant is 186 m. The
[mils] [Hz]
[mm/s] [m/s2] maximum vibration acceleration allowed is 0.4 mm/s2 and the
186.19 2.11 0.22 0.05 90.5 maximum allowable frequency is 1.2 Hz. The maximum particle
speed is determined by the following equation.
v =  a.dt
4.2 Energy Monitoring (4)
1
v = a.
Pile Driving Analyzer (PDA) is a useful tool to measure and f
determine the effect of hammer blows on pile foundations. The a
hammer blow given to the pile head is intended to penetrate into the v=
f
ground (pile foundation) or only to test the pile bearing capacity.
So that the required peak particle speed is 0.33 mm/s.
PDA is consist of the transducer sensor (to measure strain) and
Based on the monitoring data, it was found that the pile driving
accelerometer sensor (to measure acceleration of pile foundation).
was carried out at the closest distance, the peak particle velocity and
There are two or four sides of the pile foundation head
acceleration that occurred in the GTC area was smaller than required.
consisting of transducer sensor and accelerometer sensor. PDA
The frequency that occurs exceeds the required frequency. However,
computers are tools for recording measured data from transducer
the particle velocity and particle acceleration occurs small, so that
sensors and accelerometer sensors. Based on these data, the hammer
the existing GTC is in a safe condition.
energy transferred to the pile can be interpreted.
The pile hammer used to carry out the spun pile is a hydraulic
5.2 The Effect of Vibration to Existing Building
hammer coded NH-70 where the weight of the hammer is 7 tons
with a high gradual fall (multiples of 0.16 m) and a maximum Criteria for damage to the building structure as in Figure 8 shows
that the peak particle velocity that occurs due to pile driving is
546
included in the safe category. Nevertheless, these criteria are only The amount of energy is determined based on theoretical energy,
used to determine the structural conditions of the building so that it which is 2.24 t.m. With a large energy of 2.24 t.m or equal to 22400
cannot be used for evaluation of other needs such as the effect of Joule, a modification of the Attewel & Farmer equation is obtained
pile driving on the gas plant instrument. for this project which can be seen in the following equation.
10
SERIOUS
240
CRACKING
9
220
v = 15 Wo (5)
MAJOR r 1.75
8 DAMAGE 200
where :
Particle Velocity – mm/sec

CRACKING
v = peak particle velocity (mm/s)
Particle Velocity – in/sec
7
DAMAGE 180
Wo = hammer energy (Joule/blow)

160
6 r = distance of the point of purging to the location being
FINE CRACKS MINOR 140 reviewed (m)
& LOOSE DAMAGE
5 PLASTER
120
FALLS
The relationship between distance and PPV value from the
4 100
CAUTION modified formula forms a linear line in the logarithmic scale as
3
CAUTION CAUTION 80 shown in the following figure.
60
2
NO
40
NOTICEABLE
Hydraulic Hammer
Peak Particle Velocity (PPV) [mm/s]

1
DAMAGE SAFE
SAFE 100
LIMIT 20
Energy = 2.24 t.m
0
LANGEFORS EDWARDS BUMINES
(SWEDEN) (CANADA) (USA)
10
Figure 8. Damage Criteria for Building due to Vibration

(J.F.Wiss,1981)
1
Concrete material has a limit of receiving vibration with the

amount of PPV (Peak Particle Velocity) of 50 mm/s. This limit is
based on the structural limit of the structure to vibration in Figure 8. 0.1
The closest distance between piling point and the existing building 10 100 1000
is 30 m. Based on the monitoring result, the PPV is less than Distance from pile driving location [m]
requirement (50 mm/s). Hence, the existing building is safe. Monitoring Data Back Analysis
Figure 10. Relationship Between Distance from Piling Operation
6. BACK ANALYSIS and Peak Particle Velocity from Back Analysis
The following is the peak particle velocity (PPV) obtained from the
Attewel and Farmer Method (1973), J.M. Method. Ko (1990), and Based on Figure 10, it can be obtained that to meet the
the results of monitoring in the field. maximum particle speed requirements of 0.33 mm / s, the required
minimum GTC distance from the minimum purging point is 150 m.
From these results it can be obtained that the particle velocity at a
distance of 186.19 m with hammer energy 2.24 t.m meets the safe
Peak particcle Velocity [mm/s]
100.00 Energy = 2.24 t.m criteria. Radius of vibration influence due to pile driving on this case
= 22400 Joule
is 300m.
10.00 7. CONCLUSION
 The case study is taken from South Sumatera, Indonesia
1.00
regarding the effect of pile driving on a sensitive gas plant
instrument and building surrounding the project area. Soil
Hydraulic condition in the project area is consists of silty sands in the
Hammer
upper layer, coal and carbonaceous clay in the middle, and the
0.10
1 10 100 1000
underlying layer is hard clay.
Distance [m]
 The effect of vibration due to pile driving is calculated using
Attewel & Farmer (1973) J.M.Ko (1990) Data Monitoring
empirical formulas (Attewel & Farmer, 1973 and J.M. Ko et al,
Figure 9. Relationship Between Distance from Piling Operation and 1990). However, there is correction factor of the empirical
Peak Particle Velocity from Empirical Formulas and Continuous formulas that are dependent to soil types. To overcome the
Monitoring uncertainty of the correction factor, continuous monitoring by
geophone sensors to measure vibrations and PDA (Pile Driving
Based on Figure 9, it can be seen that the relationship between Analyzer) to measure actual energy of driving.
peak distance and particle velocity (PPV) from the results of  The pile driving is done using a hydraulic hammer which uses
monitoring in the field forms a linear line in the logarithmic scale. NH-70 with 7 tons ram weight and 0.32 m ram stroke. The
The relationship of distance and PPV obtained by empirical theoretical energy obtained is 2.24 t.m while the average actual
formulas from the Attewel and Farmer (1973) forms a linear line energy from the PDA test is 1,501 t.m, so that the energy
whereas from the J.M. Ko (1990) does not form a linear line on a efficiency of the hydraulic hammer is 67%.
logarithmic scale. With the similarity of the linear function form, it  In existing GTC, the maximum vibration acceleration allowed is
is necessary to do a back analyze the empirical formula from 0.4 mm/s2 and the maximum allowable frequency is 1.2 Hz. The
Attewel and Farmer (1973) based on the results of monitoring as a maximum allowable particle velocity is 0.33 mm/s. Based on the
reference for determining the PPV value for various distances on monitoring data, it was found that the pile driving was carried
certain energy quantities. out at the closest distance, peak acceleration, and particle
547
velocity that occurred in the GTC area was smaller than Eurocode 3, (1992). “Design of steel structures, chapter 5, piling.”
required. The frequency that occurs exceeds the required DD ENV 1993-1-1:1992
frequency. However, due to small particle switching, the FHWA, 2016, “Design and Construction of Driven Pile Foundations
existing GTC is in a safe condition. – Volume I” FHWA-NHI-16-009
 Radius of vibration influence due to pile driving on this case is Ko, J.M., Luk, S.T., and Cheng, C.Y., (1990), “Vibration and Noise-
300m. Measurement Prediction and Control”, Proceedings of
Australian Vibration and Noise Conference 1990.
PT GEC, 2017, “Analisis Vibrasi Akibat Pemancangan, Sumatera
8. REFERENCES Selatan”, Final Report
PT GEC, 2017, “Laporan Hasil Monitoring Vibrasi dan PDA Test”,
Attewell, P. B. & Farmer, I. W., (1973) “Attenuation of Ground Final Report
Vibrations from Pile Driving”, Ground Engineering, vol. 3, Wiss, J.F., 1967.“Damage Effects of Pile Driving Vibration”,
No. 7, July. Highway Research Record, Number 155, P.14-20
Deep Foundation Research Institute (DFRI). (2017) “Manual Wiss, J.F., 1981, “Construction Vibrations: State-of-the-Art”,
Pondasi Tiang Edisi 5”. Universitas Katolik Parahyangan. Journal of the Geotechnical Engineering Division, ASCE,
Centre of Excellence for Geotechnical Engineering Vol. 107, No. 6T2, pp. 167-181, February,.
(GEOCENTER)
548
Pengaruh Tahanan Ujung Tiang dan Faktor Aman Terhadap Lendutan Pelat
Terpaku Tiang Tunggal Pada Lempung Lunak
P. Anas1, O. Ferly2, M. Roza3
1
Pengajar Program Studi Teknik Sipil,Universitas Islam Riau, Riau, Indonesia
2
Mahasiswa Program Sarjana Teknik Sipil, Universitas Islam Riau, Riau, Indonesia
E-mail: anaspuri@eng.uir.ac.id1, ferlyoktavia@student.uir.ac.id2, rozamildawati@eng.uir.ac.id3
ABSTRAK: Perkerasan Sistem Pelat Terpaku merupakan salah satu alternatif solusi untuk mengatasi permasalahan konstruksi jalan yang
melalui tanah lunak. Kenaikan kapasitas dukung tanah dasar akibat pengaruh dukungan tiang-tiang pada pelat akan mengurangi kebutuhan
tebal perkerasan beton dan memperkaku sistem perkerasan. Kapasitas dukung tiang tersebut kontribusi dari tahanan ujung dan tahanan gesek.
Lazimnya tahanan ujung diabaikan untuk tiang berada pada tanah lunak. Penelitian ini bertujuan untuk mempelajari pengaruh tahanan ujung
tiang dan faktor aman (SF) terhadap lendutan pelat terpaku tiang tunggal. Dimensi struktur pelat terpaku dan data material mengacu pada
Puri (2017). Pelat terpaku tiang tunggal mempunyai ukuran pelat 120 cm x 120 cm, tebal 15 cm, diperkuat dengan tiang bulat beton masif
tunggal diameter 20 cm, dan panjang tiang 150 cm. Sistem ini dibebani dengan pembebanan tekan sentris. Perhitungan lendutan dan gaya
dalam pelat menggunakan metode BoEF (Beam on Elastic Foundation), berdasarkan tambahan modulus reaksi tanah dasar ekivalen yang
dibandingkan dengan lendutan hasil pengamatan. Hasil analisis menunjukkan bahwa besaran faktor aman mempengaruhi nilai lendutan
hitungan. Peningkatan SF cenderung meningkatkan lendutan hitungan. Penggunaan SF = 1,0 sudah memadai. Perhitungan lendutan dengan
memperhitungkan tahanan ujung tiang (Qb> 0) lebih baik dibanding tanpa memperhitungkan tahananan ujung tiang (Qb = 0). Momen pelat
memberikan perbedaan tidak signifikan untuk hitungan tanpa atau dengan memperhitungkan tahanan ujung tiang.
Keywords: perkerasan kaku, pelat terpaku, diameter tiang, reaksi subgrade, lendutan pelat.
1. PENDAHULUAN tanah dan tiang, karena penurunan relatif tanah dan tiang kecil maka
tahanan gesek tiang belum sepenuhnya optimum, sehingga
Perkerasan jalan di atas tanah lunak sering dibangun menggunakan digunakan pendekatan faktor perpindahan tiang (α) dalam
perkerasan kaku. Lapis perkerasan tersebut langsung menumpu pada menentukan modulus reaksi subgrade (∆k). Menurut Puri, dkk
tanah dasar lunak atau di atas lapis fondasi jalan. Perkerasan jalan (2012), sulit untuk menentukan faktor perpindahan tiang (α), serta
terletak pada tanah dasar yang tidak stabil seperti lempung lunak mengusulkan sebuah pendekatan dengan mempertimbangkan
atau tanah ekspansif, banyak menimbulkan masalah pada kinerja penurunan toleransi perkerasan kaku. Dalam menentukan kapasitas
jangka panjang perkerasan (Hardiyatmo, 2014). Perkerasan jalan dukung tiang diizinkan biasanya diambil 1/2,5 kapasitas ultimit.
yang terletak pada tanah dasar lunak atau ekspansif, sering Nilai 2,5 adalah faktor aman, dengan mengambil α = 1/SF = 1/2,5
mengalami deformasi yang berlebihan oleh beban kendaraan berat, dengan asumsi perpindahan kepala tiang sama dengan penurunan
sedangkan menurut Hardiyatmo (2012), metode perancangan toleransi perkerasan pelat kaku (δp = δa; maka β = 1 / δa).
perkerasan jalan yang telah ada, umumnya diasumsikan bahwa Analisis variasi faktor aman pada modulus reaksi subgrade
tanah-dasar dalam kondisi stabil, sehingga tebal komponen struktur ekivalen untuk model skala penuh Pelat Terpaku satu baris tiang
perkerasan hanya didasarkan pada daya dukung tanah-dasar yang dengan pembebanan tekan sentris telah dilakukan Puri (2017).
dinyatakan oleh nilai CBR atau modulus reaksi subgrade vertikal. Sistem tersebut berada pada media tanah lempung lunak. Belum
Perkerasan jalan tidak hanya menerima beban siklik oleh kendaraan, dilakukan analisis penerapan variasi faktor aman pada Pelat Terpaku
namun juga menerima beban akibat temperatur yang membuat tiang tunggal dengan memperhitungkan tahanan ujung tiang.
perkerasan jalan mengalami momen lentur bolak-balik (Puri, 2015). Penelitian ini akan menerapkan nilai modulus reaksi subgrade
Sistem Pelat Terpaku (Nailed Slab System) yang ditemukan oleh ekivalen (k’) dengan pendekatan variasi faktor aman pada nilai
Prof. Dr. Ir. Hary Christady Hardiyatmo, M. Eng., DEA. (2008) tambahan modulus reaksi subgrade (∆k) dan pengaruh tahanan
adalah suatu perkerasan beton bertulang (tebal antara 12 - 20 cm) ujung tiang.
yang didukung oleh tiang-tiang mini (panjang 150 - 200 cm dan
diameter 15 – 20 cm). Tiang-tiang dan pelat beton dihubungkan
2. LANDASAN TEORI
secara monolit dengan bantuan tulangan-tulangan. Interaksi antara
pelat beton-tiang-tanah disekitarnya menciptakan suatu perkerasan 2.1 Modulus Reaksi Subgrade Ekivalen (k’)
yang lebih kaku, yang lebih tahan terhadap deformasi tanah-dasar.
Sistem ini merupakan salah satu alternatif solusi untuk mengatasi Modulus reaksi subgrade ekivalen merupakan salah satu cara
permasalahan konstruksi jalan yang melalui tanah lunak. Sistem analisis yang digunakan pada perkerasan sistem pelat terpaku.
Pelat Terpaku cocok digunakan untuk perkerasan yang tanah Modulus reaksi subgrade ekivalen adalah modulus reaksi akibat
dasarnya dipengaruhi oleh penurunan tidak seragam, karena adanya tiang-tiang beserta pelat. Nilai modulus reaksi subgrade
interaksi tanah-tiang-pelat membuat pelat lebih kaku, sehingga ekivalen (k’) diperoleh dengan menjumlahkan nilai modulus reaksi
mengurangi terjadinya beda penurunan permukaan perkerasan subgrade tanah (k) dan nilai tambahan dari modulus reaksi dari tiang
(menciptakan permukaan perkerasan beton yang selalu rata) tunggal (Dk). Hardiyatmo (2011) mengusulkan metode penentuan
(Hardiyatmo, (2016). Pelat berfungsi ganda yaitu sebagai struktur tambahan modulus Dk. Puri, dkk, (2012b) menyederhanakan metode
perkerasan sekaligus sebagai pile cap, sedangkan tiang-tiang selain yang diusulkan oleh Hardiyatmo tersebut dengan
berguna untuk menaikkan daya dukung tanah-dasar, juga menjaga mempertimbangkan penurunan izin dari perkerasan. Istilah modulus
agar pelat beton tetap kontak dengan baik pada lapis pondasi bawah. reaksi subgrade ekivalen (k’) digunakan untuk membedakan
Perancangan Pelat Terpaku dapat menggunakan Metode modulus reaksi subgrade klasik (yaitu dari uji beban pelat standar)
Hardiyatmo Modifikasi (Puri, dkk., 2012). Metode ini merupakan dan modulus subgrade pelat yang sudah dipengaruhi oleh dimensi
salah satu metode dalam menentukan nilai modulus reaksi subgrade pelat yang lebih lebar dan perlawanan kelompok tiang yang
ekivalen, dan merupakan modifikasi dari metode Hardiyatmo mendukung pelat.
(2011). Oleh karena tiang berada pada tanah lunak dan dimensi Pada sistem pelat terpaku, pendekatan penentuan modulus reaksi
tiang kecil maka tahanan ujung dibaikan (Hardiyatmo, 2011; Puri, subgrade ekivalen (k’) didefinisikan sebagai modulus reaksi
dkk. 2012, Puri, 2015). Menurut Hardiyatmo (2011), penurunan subgrade yang memperhatikan konstribusi tiang dalam menahan
pelat beban lebih besar dari pada penurunan atau perpindahan relatif gesekan perpindahan vertikal pelat ditentukan sebagai berikut
549
(Hardiyatmo, 2011; Dewi, 2009; Puri, et.al., 2011b, 2012a dalam modulus reaksi subgrade karena memperhitungkan tahanan ujung
Puri, dkk, 2013): tiang (Qb) dinyatakan oleh persamaan berikut:
k '  k  k (1) 0 , 4 (Q S  Q b )
k  (5)
dengan :  a A PS
k’ = modulus reaksi subgrade ekivalen pelat terpaku (kN/m3)
k = modulus reaksi subgrade dari tanah (kN/m3) atau,
Δk = tambahan modulus reaksi subgrade karena adanya tiang
(kN/m3) 0 ,4 ( f S A S  f b Ab )
k  (6)
 a A PS
2.2 Reaksi Subgrade
Koefisien reaksi subgrade adalah hubungan konseptual antara Sehingga modulus reaksi subgrade ekivalen dengan
tekanan tanah dengan defleksi (Bowles, 1982 dalam Muhu, 2007). memperhitungkan tahanan ujung tiang dapat dihitung dengan:
Salah satu besaran yang dapat digunakan dalam analisis lendutan
pelat adalah koefisien reaksi tanah dasar. Rumus dasar perhitungan 0 , 4 ( f S A S  f b Ab )
k' k  (7)
nilai koefisien subgrade (kv) untuk pelat kaku adalah:  a A PS
Untuk pelat terpaku yang berada di atas tanah lunak, maka
q
kv  (2) tahanan ujung tiang diabaikan. Dalam hal ini tambahan modulus
 reaksi subgrade karena adanya tiang di bawah pelat ditentukan
dengan: (Puri, dkk, 2012a; Puri, 2015) sebagai berikut:
q = tekanan tanah (kN/m2)
δ = lendutan yang terjadi (m) 0 ,4 a d c u A S
k  (8)
 a A PS
Menurut Puri, dkk (2013), koefisien reaksi tanah dasar arah
vertikal (kv) dapat digunakan dalam hitungan lendutan. Koefisien ini Sehingga modulus reaksi subgrade ekivalen dapat dihitung dengan :
ditentukan sebagai tekanan fondasi (q) yang dibagi dengan
penurunan yang bersesuaian (δ) dari tanah di bawahnya. Dengan 0,4a d cu AS
kata lain, reaksi tanah dasar tidak lain adalah distribusi reaksi tanah k' k  (9)
(q) di bawah struktur fondasi rakit guna melawan beban fondasi. a APS
Reaksi subgrade terdistribusi tidak linier akibat beban merata Faktor adhesi ad dapat ditentukan berdasarkan kurva Tomlinson
fondasi. Pada lempung, distribusi reaksi tanah berbentuk cembung pada Grafik McClelland (1974) dengan menggunakan nilai cu
dengan reaksi maksimum di sekitar pinggir fondasi dan reaksi yang (Gambar 1).
lebih kecil pada tengah-tengah fondasi.
Modulus reaksi subgrade dari uji beban pelat (k) biasanya
menggunakan pelat lingkaran dan harus dikoreksi terhadap ukuran
dan bentuk pelat dari pelat terpaku (Puri, dkk, 2012). Koreksi
ukuran dan bentuk pelat mengacu pada metode Das (2011) seperti
pada persamaan berikut:
Koreksi terhadap ukuran pelat dinyatakan sebagai :
 0,3 
k v  k 0, 3   (3)
 B
Koreksi terhadap bentuk pelat dinyatakan sebagai :
 0,5B 
k ( B  B ) 1  
kv   L  (4)
1,5
dengan: Gambar 1. Faktor adhesi untuk tiang pancang pada tanah lempung
k = modulus reaksi tanah dasar dengan lebar B (kN/m3), (McClelland, 1974)
k0,3 = modulus reaksi tanah dasar pelat beban 0,3 m x 0,3 m
(kN/m3), Mobilisasi perlawanan gesek tiang masih dalam zona elastis
B = lebar pelat (m), (Puri, dkk (2012). Menurut aturan praktis dalam menentukan
kapasitas dukung tiang diizinkan biasanya diambil pada 1/2,5
= modulus reaksi tanah dasar pelat beban B x B (kN/m3),
kapasitas ultimit. Dimana 2,5 adalah faktor aman (SF). Dengan
L = panjang pelat (m). mengambil α = 1/SF = 1 /2,5 dengan asumsi perpindahan kepala
tiang sama dengan penurunan toleransi perkerasan pelat kaku (δp =
δa; maka β = 1 / δa). Oleh karena itu, tambahan modulus reaksi
2.3 Modulus Reaksi Subgrade Ekivalen Metode Hardiyatmo
subgrade karena adanya tiang dibawah pelat dimodifikasi dengan
Modifikasi
(Puri, dkk. 2012) :
Metode Hardiyatmo (2011) menggunakan pendekatan reduksi
tahanan tiang dalam penentuan Dk. Perpindahan relatif antara tanah f S AS (10)
dan tiang, dan reduksi tahanan tiang telah diperhitungkan. Untuk k 
2,5 a APS
keperluan desain praktis, penentuan reduksi tahanan tiang sulit
diperhitungkan. Puri, dkk, (2012a) mengusulkan suatu pendekatan Dan persamaan (3.10) tersebut ditulis menjadi (Puri, 2015) :
dalam penentuan tambahan modulus reaksi subgrade dengan
f S AS (11)
menggunakan penurunan izin/toleransi pelat perkerasan. Tambahan k 
SF a APS
550
Sehingga modulus reaksi subgrade ekivalen dapat dihitung dengan :
f S AS (12)
k'  k 
SF a APS
Dengan fs : tahanan gesek satuan tiang (kN/m2), fs = ad.cu; ad :
faktor adhesi, c : kohesi (kN/m2), δa : toleransi penurunan pelat
perkerasan kaku sebesar 5 (mm), As : luas selimut tiang (m2), Aps:
luasan zona pelat yang didukung oleh satu tiang (m2), Aps= s2, s :
jarak antar tiang (m), dan 2,5 : nilai faktor aman (SF).
Di lapangan sistem ini diaplikasikan untuk perkerasan kaku
bertulang menerus (CRCP). Untuk menjaga agar pelat beton tidak
mengalami regangan tarik berlebihan maka lendutan toleransi (δa)
maksimum 5 mm atau 1/256 bentangnya (Puri, 2015). Balok lantai
biasanya dibatasi hingga lendutan maksimum 1/360 bentangnya
agar menghindari keretakan pada beton menurut Singer (1985).
3. Metode Penelitian
3.1 Metode Pengumpulan Data
Penelitian dilakukan menggunakan data sekunder hasil uji skala
penuh Sistem Pelat Terpaku tiang tunggal pada lempung lunak
(Puri, 2017). Gambar 2 adalah skema uji dengan sifat-sifat tanah dan
beton ditunjukkan pada Tabel 1. Kolam uji berukuran 7 m x 3,54 m
x 2,5 m yang diisi lempung lunak setebal 2,15 m. Pelat terpaku tiang
tunggal terdiri atas satu buah tiang masif beton bertulang
berdiameter 0,20 m dan panjang 1,5 m. Tiang ini terhubung secara
monolit pada pelat beton bertulang di atasnya. Pelat berukuran 1,2 m
x 1,2 m dengan tebal 0,15 m dimana di bawah pelat juga diberi
lantai kerja setebal 0,05 m. Penjelasan konstruksi dan metode
pelaksanaan uji Pelat Terpaku dapat merujuk pada Puri (2017). Gambar 2. Perkerasan Sistem Pelat Terpaku Tiang Tunggal (Puri,
2017).
3.2 Tahapan Pelaksanaan Penelitian
3.3 Cara Analisis
Penelitian dimulai dengan pengumpulan data sekunder hasil uji
skala penuh Sistem Pelat Terpaku tiang tunggal pada lempung lunak Analisis dilakukan semi-manual yang didasari pada perkerasan pelat
(Puri, 2017), kemudian perhitungan modulus reaksi subgrade terpaku tiang tunggal, perhitungan nilai modulus reaksi subgrade
ekivalen (k’) menggunakan Metode Hardiyatmo Modifikasi dengan ekivalen (k’) dengan tambahan modulus reaksi subgrade (∆k)
variasi faktor aman. Selanjutnya, perhitungan berdasarkan lendutan Metode Hardiyatmo Modifikasi sebagaimana Persamaan (1) dan (8),
pengamatan (Puri, 2015) sebagai lendutan izin (δa = δs). dan perhitungan lendutan dan gaya dalam dilakukan dengan bantuan
Perhitungan lendutan dan gaya dalam menggunakan metode program “BoEF.xls” Program Versi 1.4.
BoEF (Beam on Elastic Foundation) dengan bantuan program
“BoEF.xls” Program Version 1.4 dengan meng-input nilai modulus Table 1 Sifat-sifat Lempung dan Beton (Puri, 2017)
reaksi subgrade ekivalen (k’). Hasil analisis berupa lendutan pelat
dan gaya dalam berupa momen. Dari pembahasan dapat No. Parameter Unit Rerata
A. Lempung Lunak
disimpulkan pola lendutan akibat penerapan nilai modulus reaksi
1. Berat Spesifikasi, Gs - 2,55
subgrade ekivalen (k’) Metode Hardiyatmo Modifikasi dengan
2. Batas-batas Konsistensi :
pendekatan variasi faktor aman, serta membandingkan pola lendutan - Batas Cair, LL % 88,46
perhitungan saat Qb = 0 dan Qb > 0 berdasarkan lendutan - Batas Plastis, PL % 28,48
pengamatan (Puri, 2017) sebagai lendutan izin (δa = δs) dan lendutan - Batas Susut, SL % 9,34
izin maksimum (δa = 5 mm). Lebih jelasnya tahapan-tahapan - Indeks Plastisitas, PI % 59,98
penelitian dirangkum ke dalam sebuah bagan alir sebagaimana - Indeks Cair, LI % 0,36
Gambar 3. 3. Kadar Air, w % 54,87
4. Kadar Lempung % 92,93
5. Kandungan Pasir % 6,89
6. Berat Volume Basah,  kN/m3 16,32
7. Berat Volume Kering, d kN/m3 10,90
8. Kuat Geser Undrained, Su
- Tak Terganggu kN/m3 20,14
- Remolded kN/m3 11,74
9. CBR % 0,83
10. Klasifikasi Tanah :
- AASHTO - A-7-6
- USCS - CH
B. Beton
11. Pelat beton bertulang :
- Kuat tekan MPa 29,21
- Kuat tarik kPa 3.13
12. Kuat tekan beton tiang MPa 17,4
13. Kuat tekan lantai kerja MPa 14,5
551
Tabel 2 Nilai ∆k dan k’ Tanpa Memperhitungkan Tahanan Ujung

Tiang (Qb = 0)
Beban Modulus SF
(kN) Reaksi
Subgrade 1,0 2,0 2,5 3,0
(kN/m³)
5 ∆k 54441,80 27220,90 21776,72 18147,27

k' 58941,80 31720,90 26276,72 22647,27
10 ∆k 27620,37 13810,19 11048,15 9206,79

k' 32120,37 18310,19 15548,15 13706,79
20 ∆k 13402,76 6701,38 5361,11 4467,59

Gambar 3. Bagan Alir Penelitian k' 17902,76 11201,38 9861,11 8967,59
40 ∆k 4470,62 2235,31 1788,25 1490,21

4. HASIL DAN PEMBAHASAN
k' 8970,62 6735,31 6288,25 5990,21
4.1. Pembebanan
60 ∆k 2248,28 1124,14 899,31 749,43
Beban yang digunakan sebesar 5 kN, 10 kN, 20 kN,40 kN dan 60
kN mengacu pada Puri (2017). Hubungan beban dan lendutan pelat k' 6748,28 5624,14 5399,31 5249,43
diberikan pada Gambar 4. Perhitungan berdasarkan lendutan
pengamatan sebagai lendutan izin (δa=δs) ditinjau pada pembebanan k' 9870,67 7185,33 6648,27 6290,22
tengah pelat.
Tabel 3 Nilai ∆k dan k’ dengan Memperhitungkan Tahanan Ujung
Tiang (Qb > 0)
Beban Modulus SF
(kN) Reaksi
Subgrade 1,0 2,0 2,5 3,0
(kN/m³)
5 ∆k 70774,35 35387,17 28309,74 23591,45

k' 75274,35 39887,17 32809,74 28091,45
10 ∆k 35906,48 17953,24 14362,59 11968,83

k' 40406,48 22453,24 18862,59 16468,83
Gambar 4. Hubungan P- Pelat Terpaku tiang tunggal (Puri, 2017)
20 ∆k 17423,59 8711,80 6969,44 5807,86
4.2. Nilai k’ dan ∆k k' 21923,59 13211,80 11469,44 10307,86
Penyederhanaan hitungan untuk BoEF dilakukan sebagai berikut: (a) 40 ∆k 5811,81 2905,90 1169,11 974,25
lantai kerja diabaikan, guna desain pada zona lebih aman, (b) pelat
penebalan/pelat konektor juga diabaikan. Perhitungan Δk dengan 2 k' 10311,81 7405,90 5669,11 5474,25
cara perhitungan yaitu menggunakan lendutan pengamatan sebagai 60 ∆k 2922,76 1461,38 1169,11 974,25
lendutan izin (δa=δs). Berdasarkan data yang diketahui, dilakukan
perhitungan nilai ∆k untuk Qb = 0 (Persamaan (11)) dan untuk Qb > k' 7422,76 5961,38 5669,11 5474,25
0 (Persamaan (6)), kemudian perhitungan nilai k’ dengan Persamaan
k' 11173,05 7836,53 7169,22 6724,35
(1). Hasil dari perhitungan disajikan pada Tabel 2 dan Tabel 3.
552
4.3. Hubungan Beban dan Lendutan 4. DAFTAR PUSTAKA

Hasil perhitungan berdasarkan lendutan pengamatan sebagai Afriliyani, N., Puri, A., Ardiansyah, R. (2017). “Penerapan Modulus
lendutan izin (δa=δs) ditinjau pada pembebanan di tengah pelat Reaksi Subgrade Ekivalen Metode Puri, dkk (2012) dalam
diberikan pada Gambar 5. Gambar ini memberikan hasil hitungan Perhitungan Lendutan Pelat pada Perkerasan Sistem Cakar
dengan variasi SF dan Qb. Terlihat bahwa lendutan tanpa Ayam Modifikasi”. Prosiding Konferensi Nasional Teknik
memperhitungkan tahanan ujung tiang (Qb= 0) lebih besar Sipil dan Perencanaan (KN-TSP) 2017, Program Studi Teknik
dibanding lendutan dengan memperhitungkan tahananan ujung tiang Sipil, Universitas Islam Riau, Pekanbaru, 9 Februari 2017, pp.
(Qb > 0), dan semua lendutan hitungan lebih besar daripada 29-35.
lendutan pengamatan sebagaimana Puri (2017). Terhadap SF, Agustin, D.R., Puri, A., Ardiansyah, R. Perhitungan Lendutan
tampak bahwa hasil hitungan cenderung over-estimated dengan Perkerasan Jalan Sistem Cakar Ayam Modifikasi dengan
bertambahnya nilai SF baik untuk Qb= 0 maupun Qb > 0. Hasil Variasi Faktor Aman pada Tambahan Modulus Reaksi
lendutan hitungan untuk SF = 1,0 sudah sangat mendekati Subgrade. J. Saintis, Vol 17 No. 1, 2017, pp. 15-23.
pengamatan. http://journal.uir. ac.id/index.php/saintis/article/view/1761.
Das, B.M. (2011). Principle of Foundation Engineering, 7ed.,
Stamford: Cencage Learning
Dewi, D.A. (2009). Kajian Pengaruh Tiang Tunggal Terhadap Nilai
Koefisien Reaksi Subgrade Ekivalen Pada Uji Beban Skala
Penuh, Tesis, Program Studi Teknik Sipil, Program
Pascasarjana UGM, Yogyakarta, Indonesia.
Hardiyatmo, H.C. (2008). “Sistem‘PelatTerpaku’ (Nailed Slab)
Untuk Perkuatan Pelat Beton Pada Perkerasan Kaku (Rigid
Pavement)”. Prosiding Seminar Nasional Teknologi Tepat
a) SF = 1,0 b) SF = 2,0 Guna dalam Penanganan Sarana-prasarana, MPSP JTSL FT
UGM., pp. M-1—M-7.
Hardiyatmo, H.C. (2009). Metoda Hitungan Pelat dengan
Menggunakan Modulus Reaksi Tanah Dasar Ekivalen untuk
Struktur Pelat Fleksibel, sub mitted in Dinamika Teknik.
Hardiyatmo, H.C. (2011). “Method to Analyze the Deflection of the
Nailed-Slab System”. International Journal of Civil and
Environmental Engineering IJCEE-IJENS. Volume 11 Number
4, pp.22-28.
Muhu, H.L.Y., 2007, “Kajian Lendutan Pada Sistem Cakar Ayam
c) SF = 2,5 d) SF = 3,0 Akibat Variasi Lebar Pelat (Model Sistem Cakar Ayam dari
Pelat Baja)”, Tugas Akhir, Program Studi Teknik Sipil dan
Gambar 5. Hubungan P- untuk variasi SF dan pengaruh Qb untuk
Lingkungan, UGM,Yogyakarta, Indonesia.
δa=δs.
Puri, A., 2015, “Perilaku Sistem Pelat Terpaku Pada Tanah Lunak”,
Disertasi, Program Studi Teknik Sipil, Program Pascasarjana
4.4. Momen Pelat
UGM Yogyakarta, Indonesia
Pengaruh SF dan Qb terhadap momen pelat diberikan pada Gambar Puri, A. (2017). “Developing the Curve of Displacement Factor for
6. Terlihat bahwa tidak terdapat perbedaan signifikan antara momen Determination The Additional Modulus of Subgrade Reaction
tanpa memperhitungkan tahanan ujung tiang (Qb= 0) dengan on Nailed-slab Pavement System”. International Journal of
momen saat memperhitungkan tahananan ujung tiang (Qb > 0). Technology (IJTech), Vol. 8 No. 1 (2017)124-133. ISSN2086-
9614. http://ijtech.eng.ui.ac.id/article/view/232.
Puri, A., 2017, “ Effect of Safety Factors on The Calculated
Deflection of 1-Pile Row Full Scale Nailed-Slab Pavement
System Resting on Soft Clay Due to Concentric Loadings,
ICoSET & ICoSEEH UIR 2017, Pekanbaru.
Puri, A., Hardiyatmo, H.C., Suhendro, B., dan Rifa’i, A., 2013,
“Penerapan Metode Analisis Lendutan Pelat Terpaku Pada
Model Skala Penuh dan Komprasi dengan Uji Pembebanan”,
Konferensi Nasional Teknik Sipil 7 (KoNTeKS7), Universitas
Gambar 6. Pengaruh SF dan tahanan ujung tiang terhadap momen Negeri Sebelas Maret, Surakarta, 24-26 October 2013, pp.
pelat untuk δa=δs G201-G211.
Puri, A., Hardiyatmo, H.C., Suhendro, B., dan Rifa’i, A., 2012a,
5. KESIMPULAN “Determining Additional Modulus Of Subgrade Reaction
Based On Tolerable Settlement For The Nailed-Slab System
Berdasarkan hasil dan pembahasan dapat disimpulkan sebagai Resting On Soft Clay”, International Journal of Civil and
berikut Environmental Engineering IJCEE-IJENS,Vol. 12 No. 03,
1. Hasil analisis menunjukkan bahwa besaran faktor aman pp.32-40. http://ijens.org/
mempengaruhi nilai lendutan hitungan. Peningkatan SF IJCEE%20Vol%2012%20Issue%2003.html.
cenderung meningkatkan lendutan hitungan. Penggunaan faktor Puri, A., Hardiyatmo, H.C., Suhendro, B., Rifa’i, A., 2015,
aman SF = 1,0 sudah memadai. “Perilaku Perkerasan Sitem Pelat Terpaku Pada Tanah Dasar
2. Perhitungan lendutan dengan memperhitungkan tahanan ujung Lempung Lunak”, Prosiding Konferensi Nasional Pascasarjana
tiang (Qb> 0) lebih baik dibanding tanpa memperhitungkan Teknik Sipil (KNPTS) 2015, Institut Teknologi Bandung,
tahananan ujung tiang (Qb = 0). Bandung, 12 Nopember 2015, ISSN 2477-0086, pp. 7-17.
3. Momen pelat memberikan perbedaan tidak signifikan untuk Young, W.C., Budynas, R.G. Roark’s Formula for Stress and Strain.
hitungan tanpa memperhitungkan tahanan ujung tiang (Qb= 0) 7th ed., McGraw-Hill, New York, (2002).
dibanding dengan hitungan yang memperhitungkan tahananan
ujung tiang (Qb > 0).
553
Application of Distributed Fibre Optic Sensor (DFOS) in Bi-directional Static Pile

Load Tests
Lee Siew Cheng1, Tee Bun Pin1, Chong Mun Fai1, Hisham Mohamad2 and Ang Koh An3, Paulus P. Rahardjo4
1
Smart Sensing Technology Sdn. Bhd., Malaysia.
2
Civil & Environmental Engineering Dept., Universiti Teknologi PETRONAS, Malaysia.
3
GDS Instruments Sdn. Bhd., Malaysia.
4
PT. Geotechnical Engineering Consultant, Indonesia.
E-mail: sstechsb@gmail.com
ABSTRACT: This paper describes a case study of a bi-directional load test on a working pile located at limestone formation area. The test
pile was instrumented with Distributed Fibre Optics Strain Sensors (DFOS) to measure the change in strain and to determine the pile shaft
friction and end bearing. This paper highlights the advantages and limitations of DFOS in measuring the continuous strain profile of a test
pile. Interpretation on the anomalies detected through the DFOS results is discussed. The paper aims to introduce to the industry, the superior
information obtained using the innovative fibre optic technology for geotechnical testing and monitoring.
Keywords: Bidirectional Static Load Test, Distributed Fibre Optic Strain Sensor, pile load test.
1. INTRODUCTION Maintain load tests (MLT) assigned to evaluate such a high

capacity piles significantly increase the time, cost and risk of the
Pile load tests are very important to assess the actual behaviour of pile test. Testing these highly loaded piles using conventional MLT
piles and to verify the geotechnical design parameters and methods will require the construction of massive reaction systems,
construction approaches. The Bidirectional Static Load Test i.e. dummy piles, platforms, reaction beams or concrete blocks. The
(BDSLT) was introduced as one of the Maintain Load Test (MLT) conventional MLT methods with such a massive reaction system
options and applied since 1980’s. Although the BDSLT offers time, involve very high cost and risks, sometimes even human cost. There
cost and technical advantages over the conventional MLT, this are many literatures on MLT failures due to platform collapse,
approach is not widely accepted by the industry due to the fact that tension bar snapped, etc (FPS, 2006).
it is not a top loaded test. The Bidirectional Static Load Test (BDSLT) was first
To compute the skin friction and end bearing of the pile introduced by Pedro Elisio and Jorj Osterberg in the 1980’s as an
instrumentations such as vibrating wire strain gauges, electrical- alternative approach to the conventional MLT such as Kentledge
resistance strain gauges, load cells and tell-tale extensometer are system and reaction pile system. The main difference between
used to obtain the strain and displacement at a discrete point within BDSLT and conventional MLT is the position of jacks and the
a pile. Recent advancement in the fibre optics technology has reaction systems. The conventional MLT systems apply load from
opened up more opportunities for studying the actual pile-soil the pile top. While, in the BDSLT, the jacks are located in an
interaction behaviour. The distributed fibre optic strain sensors are equilibrium point of the pile shaft, separating the pile body into
able to acquire detailed continuous strain data along a pile, and this upper section and lower section. When pressure is applied by the
unique nature of the DFOS has an advantage in helping engineers hydraulic pump, the bidirectional jack expands and push the upper
locate anomalies such as deformation and non-uniform distributed shaft upwards (to mobilize shaft friction) and lower shaft
soil structure interaction forces. These anomalies might be caused downwards (to mobilize the shaft resistance and end bearing).
by heterogeneous ground formation or non-uniform of pile stiffness Therefore, in a BDSLT, the jack capacity is only half of the test load
(Soga et al, 2015), which is difficult to be detected by using point which leads to a significant saving in cost. The typical schematic
wise sensors. Although the DFOS system has been utilized in many diagram of BDSLT set up is shown in Figure 1.
different areas for a long time, the understanding of its full
capabilities in civil engineering applications is still lacking.
The authors understand the importance of knowledge and
experience sharing in order to raise the awareness and confidence
level of the industry to embrace and to specify newer technologies.
Hence this paper describes the successful application of the DFOS
systems to measure the strain within a bored pile during the BDSLT
and its corresponding interpretation of the continuous strain profile
in the assessment of the bored pile condition.
2. GENERAL PRINCIPLES
This section describes the general principles of bidirectional static
load test (BDSLT) and distributed fibre optic sensor (DFOS).
2.1 Bidirectional Static Load Test (BDSLT)
Figure 1 Typical Schematic Diagram of Bi-Directional Static Load
Rapid urbanization process is now a global phenomenon in response Test
to a fast-growing population and development. Construction of
skyscrapers and mega infrastructure projects can be seen in many In a conventional MLT, sometimes it is not easy to determine
developing countries to serve specific functions in meeting the the ultimate shaft friction and end bearing of rock socket in a bored
urban need and to improve the quality of life. The construction of pile, as the magnitude of top down load significantly reduces before
these gigantic structures in urban areas with space constraints often it reaches the pile toe to fully mobilizing the bored pile. In order to
adopts very long bored piles of large diameter, in order to limit the optimize the design of a rock socketed bored pile, designers are keen
number of piles within a pile group. Therefore, tremendous loads to determine/verify the ultimate shaft friction and end bearing of the
will be imposed on these piles. rock layer. This can be achieved relatively easier in a BDSLT as the
554
jack can be placed in the rock layer. The load will be directly measured for many years, hence the DFOS is also very good for
exerted on the rock layer before it is transferred to the soil layer. long term monitoring. Finally, as compared to conventional sensors,
Another advantage of BDSLT is the improved safety as the the DFOS system provides a continuous strain measurement along
sacrificial jack is embedded within the pile body located deep down the pile rather than discrete measurements at several designated
in the ground. There is no need for platform preparation or stacking points.
of concrete blocks on the bored pile and this can eliminate the risk The limitations of the DFOS system is that it is not suitable for
of collapse due to soft ground. Besides, for BDSLT, there is no need dynamic measurements as the current technology requires several
to extend the pile up to ground level (especially for pile with deep minutes to obtain a complete measurement. Also, the cost of the
cut off level), hence there is a saving in time and cost. The bored analyzer is high as compared to conventional loggers. However, the
pile can be cast up to cut off level only as the jack exert forces price for DFOS system will surely be more competitive when the
within the pile body. system is more commonly adopted.
Without the requirements for massive reaction systems, BDSLT
can be performed with minimal space and is able to significantly
reduce time, cost and risk.
2.2 Distributed Fibre Optics Sensors (DFOS)

The application of fibre optic (FO) in the civil engineering industry
is still new as compared to its application in telecommunications and
the medical industry (Annamdas, 2011). For structural monitoring
purposes, there are three main FO sensing approaches, i.e. the
Fabry-Perot interferometers (discreet sensing points), Fibre Bragg
sensors (quasi-distributed sensor system), and Distributed Fibre
Optics Sensing system, DFOS (continuous sensor system).
This paper focuses specially on DFOS in measuring strain along
the pile body during the bidirectional static load test. The
Distributed Fibre Optic Sensor (DFOS) system is a new approach
which provides flexibility and capability in structural monitoring
based on two major motivations, integrity monitoring and Figure 3 BOTDA Analyser (OZ Optic Ltd)
performance-based monitoring. For instrumented test pile, DFOS is
used to measure strain, which is then computed to pile shaft friction The optical fibre cable used for the bored pile is as shown in
and pile end bearing through analyses. The DFOS functions based Figure 4. The single core single mode optical fibre is reinforced with
on the properties of spectrum of the backscattered light within the several strands of steel wires and polyethylene cable jacket. The
optical fibre, using the Brillouin Optical Time Domain Analysis diameter of fibre optic sensor is 0.125 mm with 0.25 mm cable
(BOTDA) system. The principle of BOTDA sensor system is as coating, while the overall cable diameter is 5 mm. The glass core is
show in Figure 2 and the typical BOTDA analyser is as show in firmly fixed together with the coating in order to ensure the full
Figure 3. The BODTA technology is a proven robust system in transfer of the strain from coating to inner glass core.
determining the load distribution along the pile shaft and pile toe.
Light waves travelling in the fibre optic cable reacts with the glass
material in the fibre. The reaction causes a change in density and
also a change of frequency, i.e. the Brillouin frequency shift. By
resolving the frequency shift and the propagation of time, a
continuous full strain profile can be determined.
Figure 4 Configuration of strain sensing optical cable
3. METHODOLOGY
This section describes the geological information, pile information
and the procedures of testing. Criteria for the termination of load test
is also included in this section.
Figure 2 Principle of BOTDA sensor system (Lan et al, 2012) 3.1 Geological Information
The main strengths of fibre optic sensors are the nature of glass This paper describes the case study of an instrumented pile load test
which is immune to electromagnetic interference, corrosion resistant using DFOS in a residential building project. The test pile is a
and inert to chemical reactions. The FO sensors are very versatile working pile located at limestone formation. The bidirectional static
and can be safely used in many harsh environments such as high- load test aims to verify the geotechnical design parameters for the
voltage, marine or explosive environments. Besides, the DFOS highly fractured limestone formation.
cables are thin and therefore simplifies the installation on the The soil profile consists of loose to medium dense silty clay,
reinforcement cage without disrupting the concreting procedures. with SPT-N values range from 0 to 28 as shown in Figure 5. The
The DFOS cables embedded in concrete piles are durable and can be limestone layer was found at approximately 61 m below ground
555
level. The Rock Quality Designation (RQD) of the limestone the reinforcement cage at four sides, at 90-degree interval around
samples are all 0 %, indicating highly fractured limestone, with total the circumference of the reinforcement cage.
core recovery (TCR) ranging from 47 % to 100 % as illustrated in
Figure 5. 3.3 Testing Procedures and Criteria
The BDSLT was conducted in accordance to ASTM
D8691/D8691M-18 (Standard Test Methods for Deep Foundations
Under Bi-Directional Static Axial Compressive Load) and in
accordance to load schedule and requirements provided by the
design engineer. The test pile is a working pile designed to a factor
safety of 2. Therefore, the test load of the pile is twice the working
load (WL=12,800 kN and TL=25,600 kN). The loading sequence is
The increment or decrement of loading can only be executed if
the rate of the pile settlement is not more than 0.0625 mm per 15
minutes, or 0.25 mm per hour. If this criterion is not met, the
maintaining time of the loading stage shall be extended.
Table 1 Loading cycle of BDSLT

Loading Working Duration Loading Working Duration
Cycle Load Cycle Load
% Hour % Hour
First 0 0 Second 0 0
Loading 20 1 Loading 20 1
Cycle 40 1 Cycle 40 1
Figure 5 Geological Information of pile location 60 1 60 1
80 1 80 1
3.2 Test Pile Information 100 12 100 1
80 1 120 1
The test pile is 1.35 m diameter and was designed to a length of 57.3
60 1 140 1
m with a 6.8 m rock socket length. The pile was cast up to cut off 40 1 160 1
level only, i.e. 10.2 m (empty bored) from the ground level. The 20 1 180 1
bidirectional jack was embedded at 5.8 m from pile toe as shown in 0 2 200 24
Figure 6. Concrete grade 40 was used and the main reinforcement 160 1
120 1
consists of fifteen number of T25 steel bars extending from top to 80 1
toe of the pile. The boring and coring procedures took 40 1
approximately 4 days, using bentonite as drilling fluid. 0 2
This bidirectional static load test shall be terminated if one the

following conditions occur:
1. Maximum stroke of the jack is reached (150 mm).
2. Test load is fully applied.
3. Instruction from design engineer.
In this test, excessive upward movement (>30 mm) recorded on

the pile above the jack with applied test load still less than
designated maximum test load. The design engineer instructed the
termination of test in order not to damage the working pile. The jack
and gap beside the jack were post-grouted with epoxy and non-
shrink cement grout after the load test to ensure the integrity of the
working pile.
4. DATA ANALYSIS
This section describes the strain profile, pile shaft friction and the
pile movement. The potential causes of the low shaft friction is also
discussed in this section.
4.1 Strain Profile
Figure 7 shows the continuous axial strain profile at several loading
stages, i.e. 60%, 100%, and 150% of the working load (WL). Along
the pile, the DFOS measurement indicates consistent strain
increment as the load increased. It is interesting to observe that at
the depth between 16 m to 43 m, spikes were recorded at the same
localise depth for the four sensing cables located at the four different
corners of the pile. They are visible at the same depth in all cables
Figure 6 Schematic diagram of the BDSLT set up despite each having an independent baseline profile. This spike is
likely due to the variation of pile stiffness (EcAc+EsAs) (caused for
Prior to the construction of bored pile, a bidirectional jack was example by features such as local change in concrete, Ec and steel
attached to the reinforcement cage. The DFOS cables were tied to modulus, Es; cross sectional area of concrete, Ac and steel, As or
556
isolated stiffening bands in the reinforcement cage) (Soga et al, Other possible reason is the concrete may have collapsed during
2015). the construction process when the machinery unintentionally hit the
If the spike is due to the inhomogeneity of concrete, the concrete surface of the bored hole. For such a long pile, the concrete might
modulus can be back calculated by assuming consistent pile be slightly contaminated and compromised the quality of concrete
diameter and steel area along the pile. Figure 8 shows the back (e.g. Young’s Modulus, E) at these localized areas. Hence, much
calculated concrete modulus at selected depth along the pile. At higher strain measurements were recorded.
depth without spikes, etc. 30.0 m, 35.0 m and 46.5 m, the concrete Generally, the back calculated concrete modulus, Ec is lower
modulus is generally at range of 20 to 21 kN/mm2. At depth with than usual 28 days concrete modulus as suggested in some of the
spikes, etc. 27.7 m, 36.5 m and 41.65 m, the concrete modulus was literature. It could be due to the pile was tested at only 14 days after
much lower, at about 10 kN/mm2. These could indicate that the concreting. The concrete had achieved required strength but the
concrete strength at those pile depths with spike are likely lower concrete modulus may have not fully gained yet. Local change in
than the average strength. It is likely due to muck from pile toe that steel cross sectional area, reinforcing links or isolated stiffening
was not completely flushed above the cut-off-level and caused bands in the reinforcement cage may also cause the recorded spikes.
contamination to concrete on certain depths of the pile. If the muck Those localise depth with spike is excluded from the
was completely flushed, localise spike on strain profile should had interpretation of load transfer and mobilised shaft friction analysis
been recorded at the 1 to 2 m overcast concrete above the cut-off- as the pile stiffness are uncertain.
level.
4.2 Pile Shaft Friction
The strain data in this paper is analysed to obtain the axial forces. It
is important to note that uniform axial stiffness along a bored pile is
nearly impossible due to the generally non-uniform shape of bored
pile and the minor inconsistency in the concrete stiffness. However,
the pile stiffness, EA is assumed as constant along the pile to
facilitate the interpretation of pile load transfer, mobilised shaft
friction and mobilised end bearing mechanism.
Figure 9 illustrates the load distribution along the test pile. The
jack was embedded within the limestone bedrock with 0.8 m rock
cover above the jack. However, from Figure 10 and Figure 11, it is
observed that the 0.8 m rock layer above the jack did not contribute
to an expected high shaft friction to the pile. Maximum mobilised
ultimate shaft friction at the rock layer above the pile is only 39
kN/m2, which is lower than the soil layer above it that has mobilised
shaft friction up to 62 kN/m2. The ultimate soil friction value was
estimated based on SPT-N method. Initially, the ultimate soil
friction was estimated based on a constant ultimate shaft resistance
factor, Ksu of 4 throughout the soil layer. However, the denser soil
layer only achieved Ksu value of 2.5; while, the looser soil layer
achieved Ksu value of approximately 3.5.
Figure 7 Change in strain along the pile during BDSLT
Figure 8 Back Calculated Concrete Modulus Figure 9 Load distribution along test pile
557
40
30
Upward and Downward Movement (mm)

20
10
Actual Bi-Directional Load (kN)

0
0 2000 4000 6000 8000 10000 12000 14000
Figure 10 Mobilized Unit Shaft Friction (Upper Pile Section)
against Average Strain
-10
The ultimate pile shaft friction of the rock socket below the jack
(Figure 11) was also lower than initial proposed friction value but is -20 Average Pile Top Settlement (Upward) 1st Cycle
near to proposed limestone friction value in the literature (Neoh, Average Pile Top Settlement (Upward) 2nd Cycle
Average Pile Top Settlement (Downward) 1st Cycle
1998) based on the RQD value of the rock. From 52.5 m to 54.0 m,
Average Pile Top Settlement (Downward) 2nd Cycle
the mobilised shaft friction is 176.5 kN/m2 and 260.5 kN/m2 at 54.0
m to 56.5 m. These two layers have mobilised ultimate shaft friction Figure 13 Upward and downward movement against applied load
near to the suggested value in literature which suggested 300 kN/m2
for rock with RQD <25%. However, at the last rock socket layer,
56.5 m to 57.8 m, a mobilised shaft friction of 667.3 kN/m2 (>300
kN/m2) was achieved.
Figure 11 Mobilized Unit Shaft Friction (Lower Pile Section)

against Average Strain
Figure 14 Equivalent top settlement curve
Figure 12 indicates that the end bearing of the test pile was not
fully mobilised. The mobilised end bearing was 3339.5 kN/m2 up to Figure 14 shows the equivalent pile top settlement curve,
maximum test load. converted using an equivalent conversion method. The construction
of equivalent top load versus movement curve is according to
Specification for Static Loading Test of Foundation Pile – Self-
balanced Method (JT/T 738-2009) and subjected to the following
assumptions:
1. Pile is elastic;
2. Based on the designed pile capacity, test pile is evenly divided
into upper and lower parts, with a jack placed at the equilibrium
point of the two sections;
3. Loading mechanism of the bottom section of the pile (below the
jack) is similar to the loading mechanism of a top-loaded pile.
Therefore, it has the same load-movement behaviour as a top
loading pile;
4. Shortening of the upper section of pile, Δ is equal to the sum of
elastic shortening caused by the (i) downward load and the (ii)
Figure 12 Mobilised Unit End Bearing against Toe Movement
upward load – self weight of upper pile.
4.3 Pile Movement
Δ = Δ1 + Δ2 (1)
As shown in Figure 13, the upper pile section started to experience Where,
drastic movement at 140% working load. As this was a working pile, Δ = Shortening of the upper section of pile.
the test was terminated at 150% to avoid the working pile being Δ1 = Elastic shortening caused by the downward load.
loaded to failure. At 150% working load, the upper pile movement Δ2 = Elastic shortening caused by the (upward load – self weight of
was 33.24 mm, while, the lower pile section moved 11.12 mm. upper pile).
F L
1  (2)
EC A p
558
( F equivalent  W ) L 4.4 Potential Causes of Low Shaft Friction

2  (3)
2 EC ApY Despite the rock socket length of the test pile being 5.8 m, the result
where, indicates that the test pile did not achieve the designed pile unit shaft
F↓ = Downward load of jack, friction of 600 kN/m2 at the limestone layers except at the final
segment of the pile, i.e. approximate 1.5 m rock socket from the pile
F↑equivalent = Equivalent upward load obtained from the upward
toe. This section discusses the potential causes of the low shaft
movement versus load curve. (i.e., upward load, friction.
where the upper pile section displaced at the same
magnitude as downward movement of jack when 4.4.1 Karstic Limestone Feature
downward load is F↓),
The test pile is located in Kuala Lumpur, Malaysia, on a limestone
L = Length of upper pile section, formation. Extra considerations need to be taken to design a pile in
Ec = Concrete modulus, Kuala Lumpur limestone due to the extremely karstic feature.
Ap = Pile cross sectional area, For the first 4 m length rock socket, the Rock Quality
W = Self weight of upper pile body, Designation (RQD) of the limestone samples are less than 25% and
γ = Upper shaft friction adjustment factor. total core recovery (TCR) are generally low, indicating the pile was
socketed into highly fractured limestone. This may result in lower
ultimate rock friction value.
Soil investigation result also shows that the test pile was located
Measured Side Shear Load -
Movement Curve on an inclined limestone surface as shown in Figure 6. The top of
rock levels around the pile differed by a range of about 1.5 m.
Therefore, the effective/actual pile rock socket length is likely less
Movement, s (mm)
than 5.8 m, resulting in lower total mobilised rock friction obtained

during the BDSLT.
ࡲ↑ ࢋ࢛ࢗ࢏࢜ࢇ࢒ࢋ࢔࢚ The good practice of checking the solid and flat bed rock layer
through dipping procedures can be carried out to ensure the pile was
Net Load, F (MN)
fully rock socketed as per design, and thus achieve expected shaft
ࡲ↓ friction.
࢙↓
Measured End Bearing Load -

4.4.2 Construction Method
Movement Curve The other common potential causes of low shaft friction is the
construction method which include the time taken for boring and
coring, quality of stabilizing fluid, concrete quality etc. Stabilizing
fluid shall not be stagnant in the bored hole for a long period of time
Figure 15 Upward and downward movement vs load curves (Tucker, 1984) to prevent the reduction in shaft friction.
In this case, the construction records and concrete test results
showed that the pile was constructed in accordance to the proper
F 
Combining the equations (1), (2) and (3):
  1   2 
 equivalent

 W / Y  2 F L
(4)
procedure. Therefore, the lower RQD, lower TCR and inclined
bedrock is likely the biggest contributing factor of the resulting low
2 EC A p shaft friction of this test pile.
The equivalent top load, Ftop can be calculated:
F equivalent
W  5. CONCLUSION
Ftop   F (5)
A bidirectional static load test on a working pile was carried out
Y
Therefore, the equivalent pile top settlement, Stop is: with instrumented Distributed Fiber Optic Strain Sensor cables. The
test pile was constructed in Kuala Lumpur limestone formation.
S top  s    (6)
According to Neoh (1998), for limestone with RQD<25%, the
where, suggested ultimate rock friction value is 300 kN/m2. The mobilised
stop = Equivalent pile top settlement ultimate rock friction for this pile are generally ranged from 175 to
s↓ = Downward movement of jack 260 kN/m2 for first 4 m rock socket, which very close to the value
suggested. The highest mobilised ultimate shaft friction achieved
Figure 15 shows the upward and downward curves obtained was at segment near the pile toe (last 1.8 m), approximately 700
from BDSLT and Figure 16 shows the equivalent top loaded kN/m2, which also with RQD< 25% but with generally higher
settlement curve (after conversion). (>50%) TCR value.
The novel way of instrumented load test using Distributed Fibre
Equi. Top Loaded Load (MN) Optics Strain Sensing indicated significant advantage by obtaining
ࡲ࢚࢕࢖ the continuous strain profile of the entire pile length easily as
E q u i. To p S e ttle m e n t (m m )
compared to discrete data from conventional strain gauges attached

at pre-determined points along the pile, which often require data
࢙࢚࢕࢖ interpolation between limited sensing points, laborious installation
time and data problems arising from local erroneous/anomaly
measurements. The continuous strain profile obtained using DFOS
not only allow engineers to understand the overall performance of
End Bearing and Side Shear
Components
the tested pile but also the quality of concrete/integrity of the pile.
The continuous strain profile of the entire pile length ensures
that the engineer would not miss out on any important data along the
pile shaft that may affect any computation of the skin friction and
Figure 16 Conversion curve of top load versus movement end bearing.
559
6. REFERENCES
Annamdas, V.G.M. (2011) "Review on Developments in Fiber Soga, K; Kwan, V; Pelecanos, L; Rui1, Y; Schwamb, T; Seo, H and
Optical Sensors and Applications". International Journal Wilcock, M. (2015) "The Role of Distributed Sensing in
of Materials Engineering, vol.1(1), pp1-16. Understanding the Engineering Performance of
Federation of Piling Specialists (FPS). (2006) "Handbook on Pile Geotechnical Structures". Proceedings of the XVI
Load Testing". ECSMGE Geotechnical Engineering for Infrastructure and
Lan, C.G.; Zhou, Z. and Ou, J.P. (2012) "Full-Scale Prestress Loss Development, Keynote Paper, pp13-48.
Monitoring of Damaged RC Structures Using Distributed Transportation Industry Specification of China. (2009) "JT/T 738-
Optical Fiber Sensing Technology". Sensor 2012, vol. 12, 2009 - Specification for Static Loading Test of Foundation
pp5380-5394. Pile – Self-balanced Method", pp1-12.
Neoh, C.A. (1998) "Design & Construction of Pile Foundation in Tucker, K.L. and Reese, L.C. (1984) "Research Report Number
Limestone Formation". Journal of Institution of Engineers, 351-1F - The Effect of Bentonic Slurry on Drilled Shafts",
Malaysia, Vol. 59, No. 1, pp23–29. Federal Highway Administration (FHWA), pp1-85.
560
Design of Simple Drapery Systems for Rock Cuts and Natural Slopes
Matteo Lelli1, Alberto Grimod2, Riccardo Laneri1, Deny1, Vincent Setiawan2, Dinda Mutiara Savitri1
1
PT Maccaferri Indonesia, Indonesia,
2
Maccaferri France,
E-mail: m.lelli@id.maccaferri.com , agrimod@maccaferri.fr2, r.laneri@id.maccaferri.com3, deny@id.maccaferri.com4,
1
v.setiawan@id.maccaferri.com5, d.safitri@id.maccaferri.com6
ABSTRACT: Simple drapery mesh (or simple drapery system) is commonly installed to provide rockfall protection in open pit mines or
along the highways and railways. It is generally used for rock cuts or natural rock faces. The system is a fast and cost-effective rockfall haz-
ards mitigation measure, and it consists of a steel mesh fixed at the top of the slope with anchors and a longitudinal suspension cable. This
paper describes a new design approach to define all the components of the system. A case study of a drapery system recently designed and
installed in Indonesia is presented.
Keywords: rockfall, remediation, metallic mesh, drapery, Indonesia
1. INTRODUCTION Subsequently, the drapery system may be secured at the toe of

the rock cut as well, in order to form a sort of “pocket” where the
The natural process of weathering generates geological instabilities debris and the rocks may be pile up after their fall (figure 2). Other-
which frequently expose mining areas, and infrastructures to a wide wise, in order to reduce the stress on the mesh, as well as, the main-
range of shallow instabilities, which may vary from erosion to rock tenance costs, the bottom of the mesh can be left opened. In this
falls. Despite they mostly cause small size falling, the shallow insta- second case, a ditch (figure 3) or a barrier is required to collect or
bilities cannot be underestimated because they use to happen with contain the fallen materials.
high frequency on large surfaces. Consequently, the probability of The effect of this kind of intervention is to control the falls of
accidents is high. In this situation, designs must necessarily guaran- the rocks and debris, which are driven toward the bottom with slow
tee the efficiency of the remedial solution in terms of high perform- velocity and reduced energy. In comparison to other types of rock-
ance and low maintenance costs. In spite of the technical literature fall interventions, they are simpler to install, cheaper, and their
furnishes many precious information based on analytical and em- maintenance is easier. Nevertheless, they cannot be considered as a
pirical observation (i.e. Muhunthan and al., 2005), it does not seem remediation for shallow instabilities.
possible to codify a common design procedure for all mesh typolo-
gies. In this sense the designer experience is always needed in order
to evaluate a cost-effective intervention.
2. SIMPLE DRAPERY MESH

2.1 System description
A simple drapery mesh (or simple drapery system) consists to install
a rockfall net along rock slopes (i.e. rock cuts or natural faces). The
drapery is hung as a curtain, suspended by a longitudinal rope and
anchors at the crest (figure 1). The distance between the anchors
depends on the design and the prevailing instability conditions at the
site. The anchors are commonly aligned and fitted with suitable
terminations (often eye nuts or plates or similar) to accept the crest
rope.
The top supports are generally steel bars (full or hollow core) or
flexible cable anchors (with single or double leg). In both cases, the Figure 2. Debris and rock piled up at the bottom of a simple drapery
anchors are placed in drilled-holes and then fully grouted for their system fixed at the toe of the slope by anchors and a down-slope
entire length. The supports ae considered passive, because they do longitudinal cable
not require any pre-tensioning, and they start to work only if they
are stressed by the loads acting on the mesh (they shall not be con-
sidered bolts or tie-back anchors). Once the crest anchors and the
upper longitudinal cable are installed, the mesh can be fixed to them
and left free all along the slope.
Figure 3. Example of a simple drapery mesh with a ditch (high-

lighted with the yellow line) at the base to collect the debris falling
Figure 1. Sketch of a simple drapery system application from the slope
561
In comparison to other types of rockfall interventions (i.e. se- According to the literature (Muhunthan et al., 2005; Sasiharan et
cured drapery systems, rockfall barriers, etc.), the simple drapery is al. 2006), the inclusion of vertical ropes reduces the stress concen-
more cost-effective, and its maintenance is easier, but on the other tration on the mesh, only if they are woven and not simply applied at
hand it cannot be considered the remedy for shallow instabilities, the job site. The mesh coupled with interwoven cables is fit to trans-
because it can only mitigate the effect of the falling trajectories. fer the loads directly to the top anchor system, thus the tension on
This system is usually installed on high rocky slopes, where pin the wires is reduced. The figure below describes the load effect on a
drapery systems are not too costly, or where rockfall fences or rock- simple drapery system in the case of a mesh without cables (cases
fall embankments cannot be installed due to the uneven or high A, above), and in the case of a mesh with woven edge cables (cases
gradient of the slope. B, below).
The figure defines 3 different load conditions on the system:
theoretical case (left images), considering only the weight of the
mesh (center images), and considering also the debris accumulation
at the toe (right images). It is possible to observe that in case B the
presence of the vertical cables minimizes the deformation in the
center of the panel, and the ropes transfer directly the forces to the
top anchor system. Therefore, the real condition (bottom-right image
in figure 6) looks like the theoretical one (left images on figure 6).
For this reason, SteelGrid, which is a woven composite mesh made
of steel wire and metallic ropes woven together during the produc-
tion of the hexagonal double twist wire mesh, is effective.
Figure 6 illustrates that SteelGrid mesh is ideal for use on high
rock faces and slopes with a long drop or where large volumes of
debris are expected. The longitudinal steel ropes enable the efficient
transmission of loads to the crestline ropes and anchors, with mini-
mal mesh deformation. This aspect allows the mesh to be stressed
with larger loads or be less maintained, thanks to the increased ca-
pability of the system to retain debris at the toe.
Figure 4. Simple drapery system for temple protection (India).
3. TYPE OF MESH
During the design, the first question that consultants must solve is
“Does the mesh have a quasi-static or a dynamic behaviour?”. The
market offers a wide portfolio of meshes, such as single twist (chain
link) or double twist wire meshes, steel geocomposites with cables
and wires (SteelGrid), cable meshes, cable panels and ring net pan-
els. The graph in figure 5 recaps, in a semi quantitative way, the
performances of the main meshes in the case of static and dynamic
applications. The graph shows a non-direct proportion between the
tensile resistance (quasi-static), and the dynamic resistance, which
depends on the mesh deformability.
For most applications, the dynamic resistance is useless, and the
required tensile resistance is pretty low. Other interventions require
steel geocomposite (SteelGrid) to reduce the stresses on the suspen-
sion system. If high dynamic stresses are predicted (“dynamic Figure 6. Comparison between a mesh without interwoven ca-
shield”) High Energy Absorption (HEA) cable panels should be bles (case A, above) and with interwoven cables: SteelGrid (case B
applied, because of their high performance in dynamic conditions, below), considering 3 different load conditions
such as for the attenuator systems (Arndt et al., 2009). If the dy-
namic impacts are extremely severe, rings nets are required. Another important aspect to be evaluated during the mesh de-
sign is the capacity of the mesh to avoid the unravelling. This is an
intrinsic property of the mesh fabric. The mesh must be able to in-
hibit the propagation of the tears between the wires. In order to
avoid this issue, double twist wire meshes are usually preferred to
simple twist meshes (i.e. chain link). Researches and laboratory tests
show that damage to a double twist wire mesh remains localized and
the mesh does not unravel (figure 7) (Agostini et al., 1988).
Figure 5. Comparison between the tensile (red pattern) and the

dynamic (blue pattern) resistance of 4 different meshes (Grimod et
al., 2013). The values are related to the most common mesh used per
each family of nets. Notes: (1) Dynamic tests carried out on samples Figure 7. Comparison between a double twist mesh (left) and a
2.0x2.5 m, completely restrained on 4 sides - Maccaferri internal single twist mesh (right) after a wire is cut. Double twist mesh con-
report; (2) tensile resistance determined in accordance with the Ital- tains the propagation of the wire cut. Single twist mesh (i.e. chain
ian standard UNI 11437:2012. link) unravels.
562
4. DESIGN
4.1 Factors influencing the design
The main factors affecting the proper choice of the mesh aimed at
the simple drapery system are the following:
1. Slope morphology: the probability to have dynamic impacts
against the drapery increases if the slope is uneven and/or
steep. For instance, in case of very uneven slopes, the drapery
can only push on the crest lines and convexities, whereas the
debris can freely run down on gullies and concavities. In this
situation drapery has a negligible capacity in erosion control,
and the falling stones can reach high velocities. Therefore, the Figure 8. Example of a simple drapery system with cable panels and
installation of the drapery requires a particular care in order to double twist mesh. The proper weight of the meshes (approx. 5.0
maximise the contact between ground and mesh, or the slope kg/m2) can avoid the rolling of single boulders (red arrows) along
must be preventively regularized. the slope.
2. Top anchor spacing: this factor is strictly related to the previ-
ous one, in fact the crest anchors shall be placed in order to 4.2 Design goals
guarantee a good contact mesh-ground. Moreover, the resis-
tance of the crest supporting system must be increased if the The main design goal of simple drapery systems is getting a propor-
space between the top anchors rises. In order to generate over tioned protection system that gives the possibility to the single com-
stress on the anchoring system (anchors and cable), the anchor ponents to work all together properly. Only the top anchor system
spacing should be generally lower than 3 - 4 m. could be slightly oversized in order to guarantee the safety of the
3. Prevalent instability: if the erosion represents the main prob- foundations in case of the collapse of the lower part. Another goal is
lem, the most suitable mesh should have small opening size recognizing the limit load of the drapery system. This allows to
and enough weight to push down and stabilize the loose predict the maximum height of the drapery, and when the cleaning
ground surface. As soon as the contact between mesh and of the debris pocket is required.
ground is reached, the drapery becomes quite effective as ero-
sion control. In fact, under this condition, it is possible to have 5. CALCULATION APPROACH
the growth of the vegetation (to obtain a better result the mesh The design of simple drapery depends on different variables related
can be coupled with an erosion control mat) and the retention to the geometry of the slope, the type of mesh and the hypothetical
of debris and boulders. If the slope is vertical, the drapery debris accumulation at the bottom of the system. Nowadays, the
must be quite strong to absorb the impacts and guide the de- only research carried out to give a design guideline for these appli-
bris towards the bottom. In case of large blocks, a “dynamic” cations was done by the Washington State Department of Transpor-
drapery, like cable panels or ring nets, should be foreseen. tation (Muhunthan et al. 2005).
Whereas in case of small blocks (i.e. thin layered limestone Using these studies, and the results obtained from several labo-
cliff) lighter draperies, like geocomposite SteelGrid or double ratories and field tests, Maccaferri has developed a new calculation
twist wire mesh, might be sufficient. approach (implemented in the software MacRo 2) which give the
4. Expected life span of the drapery and maintenance costs: re- possibility to design the type of mesh, the up-slope longitudinal
garding the life span, the design shall consider the exposure of suspension cable and the characteristic of the crest anchors (diame-
the mesh to aggressive environments (i.e. salted winds, water ter, type of steel and length). This tool allows designer to have a
etc.) and the possible abrasion due to the debris movements. If quick and easy, yet reliable solution to the problem. Sometimes, a
the drapery is applied for temporary protection (i.e. mining complex numerical analysis has to be done, but this is not practical
industry) light corrosion protections (i.e. simple galvanisation for all projects, especially if the intervention has a modest size and
with zinc) can be applied. Whereas, heavier corrosion protec- has to be done in a limited period of time (emergency protection).
tions (i.e. zinc-aluminum alloys, polymeric coatings) may be The equations and the procedures at the base of this design ap-
required for permanent infrastructures applications. In term of proach are quite simple, but they give reliable and fast results.
maintenance, designers shall predict the maximum size of the
debris pocket at the toe of the mesh or the dimension of the 5.1 Mesh design
ditch at the base of the system. Once the designed accumula-
tion limit is reached, the debris shall be removed (from the The simple drapery system is a passive system capable to contain
mesh pocket or from the trench). the debris at the bottom of the slope. It has to be designed by taking
5. Weight: the weight affects the capacity of the system to stabi- into account all the loads able to generate a stress on the mesh:
lize single boulders. If the mesh is too light, its capacity to 1. The weight of the chosen mesh;
push down and stabilize large blocks is reduced. Heavy 2. The weight of the debris accumulated at the toe of the mesh;
meshes (i.e. cable panels) may avoid the rolling/falling of 3. External weights, like the snow or ice accumulation on the
block along gentle slopes (figure 8). drapery.
563
be accumulated on such a steep slope (Swiss Guideline, 2007; Mu-

hunthan et al. 2005).
Ws  s ts Hs / sin  sin   cos  tan s  (4)
Where:
s = snow (or ice) unit weight;
Ts = snow (or ice) thickness, considered homogenous along the
entire area to be protected;
s = friction angle between soil and snow (or ice) (typical values
are between 30° and 40°).
Figure 9. Geometrical input data used to calculate the stresses on the
To design the drapery system at the limit equilibrium state, three
mesh due to the debris accumulation
partial coefficients have to be introduced in the calculation to in-
crease the acting forces and decrease the resisting ones:
These three loads may be described by the following formulas,
1. mts = partial coefficient which reduces the tensile strength of
based on the researched of the U.S. Department of Transportation
the mesh (≥ 1.0; from the in-situ evidences and in-situ and
FHA (Muhunthan et al. 2005).
laboratory tests, this factor should not be lower than 2.0).
Firstly, the total load due to the mesh (Wm) has to be defined:
2. vl = load partial coefficient which increases the variable
loads, like the snow thickness and the debris accumulation (≥
Wm   m Hs / sin  (sin   cos  tan  ) (1)
1.0; suggested value according to the Eurocode 7, 1997 =
1.5).
Where: 3. pl = load partial coefficient which increases the permanent
m = steel mesh unit weight; loads, like the weight of the mesh (≥ 1.0; suggested value ac-
Hs = total height of the slope (figure 9); cording to the Eurocode 7, 1997 = 1.3).
 = inclination of the slope (figure 9); The total stress acting on the netting (Sw) is therefore:
 = friction angle between mesh and slope, typical values are
shown in Table 1. Sw  Wd  Ws vl  Wm pl (5)
Table 1. Typical Values of The Friction Angle Mesh-ground (Sasi- The Serviceability tensile strength of the mesh (Rm) is calculated as:
haran et al., 2006).
Characteristic Value of the Notes Rm  Tm / mts (6)

of the slope friction angle
soil-mesh () Where, Tm is ultimate longitudinal tensile strength of the mesh (de-
Rough ≥ 60° The slope surface is very irregular fined by laboratory tests, i.e. according to UNI 11437:2012 or BS
and undulating and has many EN 10223-3:2013, see figure 4).
and/or prominent protrusion on the The design condition is satisfied if:
surface.
Undulating 36° to 59° The slope is undulating but there Rm  Sw  0 (7)
are few and/or small abrupt protru-
sion on the surface. Thus, the safety factor of the mesh, which shall be higher than 1, is
Planar 25° to 35° The slope is planar, and the surface equal to:
is fairly smooth and has few undu-
lations.
FS MESH  Rm / Sw  1 (7.a)
Then, it is possible to identify the stress transmitted from the debris

to the mesh (Wd) as follows:
5.2 Crest cable design
 1 1  The mesh is secured to the transversal up-slope suspension cable,

sin   cos  tan  d 
1
Wd  d Hd 2   (2)
 
which is fixed to the crest supports (anchors). Designer shall calcu-
2  tan Bd tan
late the maximum load acting on the drapery (defined in the previ-
ous paragraph: Wi (1) (2) (4)) and the spacing between the up-
Where: slope anchors in order to calculate the deformation and the stress
d = debris unit weight; distribution within the rope. This method utilizes the principle of the
Hd = debris accumulation height (figure 9); catenary to verify if the tensile strength of the cable is sufficient to
d = debris friction angle; support the total weight of the system (Wm + Wd + Ws).
d = debris external inclination value (figure 9): Thus, the cable check is passed if the following equation is satis-
fied:
Bd  arctanHd / Td  Hd / tan   (3)
TWLC  FCABLE  0 (8)
Where, Td is debris accumulation width (figure 9).
Where, TWLC is cable working load limit, and FCABLE is maximum
Finally, the load acting on the mesh due to snow or ice presence tensile strength acting on the cable (calculated with the catenary
shall be calculated (Ws). It is important to notice that that for slopes solution).
with an inclination on the horizontal () higher than 55-60 degrees
the load due to the snow could be neglected since the snow cannot TWLC  TCABLE /  CABLE (9)
564
Where TCABLE is ultimate tensile strength of the designed rope (it Where, Sbar(j) is the working shear resistance of the anchor j, N(j)
depends on the steel grade, the type of rope core and the diameter of is force that the cable and the mesh transfer to the anchor j (calcu-
the rope, i.e. see ASTM A1023/A 1023M, 2002 or UNI EN 12385- lated with the catenary solution), and j is position of the anchor (in-
4:2008), and CABLE is safety coefficient which reduces TCABLE (≥ termediate or lateral).
1.0).
The safety coefficient of the cable is therefore:
Sbar ( j )  (Ybar ( j ) / st ) / 3 2
1
(11)
FSCABLE  TWLC / FCABLE  1 (8.a)
Where, Ybar(j) is the yield load of the steel bar j, and st is safety
Moreover, using the catenary model it is possible to define the coefficient for the steel strength of the bar (> 1.0).
maximum length of the rope and its maximum sag between two
consecutive anchors. Ybar( j )  ESS ( j )  ADM ( j ) (12)
Where, ESS(j) is the effective area of the steel bar j, andADM(j) is

yield stress of the steel of the bar j.

ESS ( j )   / 4 e( j )  2tc ( j )  i ( j )2
2
 (13)
Where:
e(j) = external diameter of the steel bar j;
tc(j) = thickness of corrosion on the external crown of the steel
bar j;
i(j) = internal diameter of the steel bar j.
Thus, the safety coefficient of the different anchors may be calcu-
lated as follows:
Figure 10. Example of the deformation of the up-slope cable
(red line) between two anchors (A and B) calculated using the cate-
nary model. FS ANCHOR ( j )  Sbar ( j ) / N ( j )  1 (10.a)
5.3 Crest anchors design

The design of the crest anchors can be divided in 2 different steps. 5.5 Anchor length
The first step allows to design the anchor diameter and it takes into The evaluation of the anchor length shall take into account the fol-
consideration the shear stresses transmitted by the drapery system. lowing:
The second step allows to define the minimum anchor length, which • The anchor plays an important role because it has to support the
is a function of the soil characteristics and the drilling diameter.
entire system. Its length must be enough to reach the stable
rock mass.
5.4 Anchor size
• The steel bar and the grout are exposed to weathering agents
With the catenary theory it is possible to determine the maximum (rain, salinity, temperature variations, etc.).
force acting on the intermediate and lateral anchors. The forces on The minimum theoretical length is derived with the following equa-
these two anchors differ because the supporting cable is considered tion:
as a catenary (figure 11). The intermediate anchors are less stressed
because the load can be divided in 2 directions (acting to the right LTOTAL( j )  Ls ( j )  Lp (14)
and to the left of the anchor). Instead, the lateral anchors are more
stressed because they must support the entire load transmitted by the
rope. Where, Ls is minimum foundation length (in the stable rock) calcu-
late with the Bustamante-Doix formulation (equation 15), Lp is
safety length in order to increase the depth of the anchor (i.e. length
of hole with plasticity phenomena: portion of the stable rock mass
that loses its strength –plasticisation- due to the bend deformation of
the anchor bar when it is stressed. Hence, this portion is not consid-
ered as a foundation for the anchors bars).
Lsj  Pj / ( drill  lim / gt ) (15)
Figure 11. Distribution of the forces on the up-slope supporting Where:

cable and crest anchors. The cable is considered as a catenary. drill diameter of the drill-hole (usually no lower than 40 mm);
lim = adherence tension between grout and rock (bond stress);
Then, these two forces must be related to the working shear re- gt = safety coefficient of the adhesion grout – rock;
sistance of the selected anchors. P = pullout force (calculated with the catenary theory) for the
internal and lateral anchors.
Sbar ( j )  N ( j )  1 (10)
The length of the nail determined at this point is a preliminary
value. The final suitable length of the bars shall be confirmed after
in-situ pull out tests (figure 12).
565
Figure 12. Example of a pull-out test to verify the designed length of Figure 14. Sketch representing the slope cross section and the type
of rockfall protection to be implemented.
the top anchors.
To accommodate the height of the rock cut the mesh was manu-
6. CASE STUDY: MAROS ELEVATED ROAD factured in customized rolls with variable length (from 15m to 50m)
long. In this way it was possible to decrease the installation time and
The construction of a new road corridor in Maros needed some mas- minimize transversal connections between rolls of mesh along the
sive cutting of the existing rock slopes to gain room for the elevated slope. In this way, the overall cost of the intervention was consid-
road (Figure 13). During the period of 2017 – 2018, the local Public erably reduced.
Works department issued a tender to design and install rockfall
drapery mesh along the exposed cut slopes in order to mitigate the
potential of rockfall hazards to vehicles and people the base of the
slopes, some of which exceed 30 m in height.
Figure 15. Detail of the cable woven in the double twist mesh to
form the SteelGrid (the 8mm cables are interwoven in the mesh with
30 cm spacing in order to reach very high tensile resistance, up to
177 kN/m).
According to the research done in 2005 by the Washington State

Department of Transportation, in cooperation with U.S. Department
Figure 13. View of the rock slopes during cutting operations for the of Transportation (Muhunthan et al., 2005), the presence of these
new Maros elevated road. steel cables, woven within the mesh during the manufacturing proc-
ess, enabled better stress distribution in the supporting cable and
Engineers, together with the local Public Works technical de- reduced the strain in the drapery system (figure 6). The reduction of
partment based in Makassar, designed the components of the rock- the stress on the mesh could increase the total load capacity of the
fall drapery system. The client wanted a drapery system that could mesh and consequently improving the life time and significantly
be installed in one single piece that would drape the entire slope. A reducing the maintenance costs. Moreover, the hexagonal double
drapery system with steel meshes was selected for clear cost- twist mesh provided high resistance to the impacts of rocks avoiding
effective advantages compared to other traditional solutions such as the unraveling in the event of wire breakage. In order to comply
shotcrete. The contractor in charge of the execution selected the with the durability requirements of the client, wires and cables of the
SteelGrid as netting material. The material selected such as netting, geocomposite were coated with class A Zinc95%-Aluminum5%
anchors, grouting, ropes, and accessories were in compliance with alloy (according to EN 10244-2).
the recently developed national specification for Public Works rock- According to the calculation principle illustrated in the paper,
fall protection projects (Spesifikasi Khusus Interim SKh. 1.3.16: the upper longitudinal cable and the crest anchoring systems have
Jaring Kawat dan Jaring Kabel Sebagia Pengaman Lereng Batuan). been designed. To hold the mesh at the top of the slope a high ten-
Figure 14 illustrates the typical cross section of the rockfall pro- sile strength 22 mm cable was installed between the crest cable-
tection system. The system was divided in two different types of anchors (32 mm in diameter, 4.0 m long, and disposed every 3.0 m).
intervention: an active system for the upper part of the slope (Zona The selected anchors were type BJTS 50.
A) and a passive system for the lower portion of the slope (Zona B). Hereafter are summarized the steps of the design of the simple drap-
The protection system applied to Zona A and Zona differs by the ery system.
fact that in Zona A, the rockfall netting is secured to the rock slope The input data used during the calculation are listed below:
by means of rock anchors.
566
Figure 16. Input data related to the geometry.
Figure 19. Installation of rockfall netting for Maros elevated road.
The design assumptions regarding the rock-grout adhesion

(bond stress) were verified with in-situ pull-out tests.
Figure 17. The chosen mesh is a SteelGrid (type HR 30), character-

ized by a nominal tensile resistance of 177 kN/m (test according to
EN 10223-3:2013) and a unit weight of 2.2 kg/m2.
Figure 20. The rock-grout
7. CONCLUSIONS
Figure 18. The supporting system is composed by 32 mm steel bar
A simple drapery system is a cost-effective measure to mitigate
anchors, spaced 3.0 m, and a 22 mm up-slope cable (type
rockfall hazards. This system has been successfully and extensively
6x19+IWRC – 1770 MPa).
used in several scenarios (e.g. mining, highway and railway protec-
tions), and in complex conditions (slope height up to 150 m). The
The table below summarize the safety factors of all the system
aim of this kind of intervention is to control the falls of the unstable
components obtained with MacRo 2:
rocks and debris, which are driven toward the bottom with reduced
velocity and therefore reduced energy. In comparison to other types
Table 2. Output from MacRo 2 calculation.
of rockfall interventions such as secured drapery systems or rock
Element Factor of safety or value anchors, they are simpler to install, and their maintenance is easier.
Nevertheless, they cannot be considered as a remediation meas-
Mesh Capacity: FoS = 2.8 ≥ 1.0 - Satisfied
ure for shallow instabilities. Based on the researches carried out by
Supporting up-slope cable: FoS = 1.09 ≥ 1.0 - Satisfied Muhunthan et al. 2005 and in-situ and laboratory tests, Maccaferri
Intermediate crest anchors: FoS = 3.25 ≥ 1.0 - Satisfied has developed a calculation approach (implemented in the software
Lateral crest anchors: FoS = 1.39 ≥ 1.0 - Satisfied MacRo 2) able to optimize all the components of a drapery system
Min. length of the crest anchors: Lmin = 4.74 m (drill=40mm)
(mesh, supporting cable and crest anchors), and estimate the re-
quired maintenance. Moreover, the case of the simple drapery sys-
(adopted: L=5.0 m,(drill =
76mm) tem installed for the protection of the new Maros elevated road in
South Sulawesi (Indonesia) has been presented. References
567
8. REFERENCES EN 1997 Eurocode 7. Geotechnical design – Part 1: General rules.

Grimod A., Giacchetti G. (2013). New design software for rockfall
Agostini R., Mazzalai P., Papetti A. 1988. Hexagonal wire mesh for simple drapery systems. 23rd World Mining Congress, 255. 11-
rock-fall and slope stabilization. Pubblication edited by Officine 15 August 2013. Montreal QC, Canada.
Maccaferri SpA, Bologna. Italy Muhunthan, B., Shu, S., Sasiharan, N., Hattamleh, O. A., Badger,
Arndt, B., Ortiz, T. and Turner, A. 2009. Colorado’s Full-Scale T.C., Lowell, S.M. and Duffy, J.D. 2005. Analysis and design of
Field Testing of Rockfall Attenuator Systems. Transportation wire mesh cable net slope protection – Final research report.
Research Circular E-C141 Washington State Transportation Commission – U.S. Depart-
Bertolo, P., Oggeri C. and Peila, D. 2009. Full-scale testing of ment of transportation – Federal Highway Administration.
draped nets for rock fall protection. Canadian Geotechnical Sasiharan, N., Muhunthan B., Badger, T.C., Shu, S., Carradine,
Journal, 46: 306-317. D.M. 2006. Numerical analysis of the performance of wire mesh
Bonati, A. and Galimberti, V. 2004. La valutazione sperimentale di and cable net rockfall protection systems. Engineering Geology
sistemi di difesa attiva dalla caduta massi. In Peila (ed.), 88: 121-1132.
Bonifica di versanti rocciosi per la protezione del territorio: 177- Spesifikasi Khusus Interim SKh. 1.3.16. Jaring kawat (wire mesh)
189. Torino: GEAM. dan jaring cabel (cable net) sebagai pengaman lereng batuan.
Bustamante, M. and Doix, B.1985. Une Méthode pour le calcul des UNI 11211-4. 2012. Rockfall protective measures: definitive and
tirante set des micropieux injectés. Bull. Liasion Labo. Ponts et executive design.
Chaussés Paris 149 UNI 11437. 2012. Rockfall protection measures : Tests on meshes
EN 10223-3 :2013. Steel wire and wire products for fencing and for slope coverage. UNI Ente Nazionale Italiano di
netting – Part 3: hexagonal steel wire mesh products for civil Unificazione.
engineering purposes. UNI EN 12385:4. 2008. Steel wire ropes – Safety. Part 4: standard
EN 10244-2 :2009. Steel wire and wire products – non-ferrous me- ropes for general lifting applications. UNI Ente Nazionale
tallic coatings on steel wire, Zinc or Zinc alloy coatings. Italiano di Unificazione.
EN 1991 Eurocode 1. Action on structures.
568
Emergency Works and Landslide Remediation using

Geosynthetics Reinforced Soil Structures – Recent Indonesian Case Studies
Deny1, Matteo Lelli2, Dinda Mutiara Savitri3
1
PT Maccaferri Indonesia, Indonesia
2
PT Maccaferri Indonesia,
3
PT Maccaferri Indonesia,
E-mail: deny@id.maccaferri.com1, m.lelli@id.maccaferri.com2, d.safitri@id.maccaferri.com3
ABSTRACT: Landslides like any other natural disaster are unpredictable. These events may compromise the function of some infrastruc-
tures or structures around the affected area. Mitigation measures or emergency works must be performed as soon as possible to regain the
function of the affected infrastructure. Reinforced soil structure is one of the most suitable solution to be applied to restore the landslide slope
and prevent future landslide disaster. Reinforced soil structures built using geosynthetics have shown many advantages for emergency works
or landslide remediation if compared to conventional retaining structures such as mass gravity and cantilever walls. This paper aims to give
explanation about the advantages of performing remediation works using reinforced soil structure with polymer as the primary reinforce-
ment. Two completed projects are presented to provide author’s experience regarding the application of the reinforced soil structure and
proper construction process for emergency works and landslide remediation in Indonesia.
Keywords: landslide, remediation, polymer, reinforcement
1 INTRODUCTION 2 REINFORCED SOIL STRUCTURE

The term of landslide refers to several forms of mass wasting that Conventional gravity and cantilever wall systems made of masonry
include a wide range of ground movements such as rockfall, deep- or concrete act as rigid units and have served the industry well for
seated slope failures, mudflows and debris flows. Numerous factors centuries. This system works by resisting the lateral earth pressure
can cause soil slope to landslide such as earthwork which change the by virtue of their large mass (Figure 1). However, a new system of
shape of slope, erosion, rainfall, or earthquake. In Indonesia, most of retaining wall was introduced in the 1960s by H. Vidal with Rein-
the cases of landslide occur during the heavy rainfall. Consequently, forced Earth (Koerner, R.M. 2005). Reinforced soil structure works
the saturation process that causes by rainfall weakening the slope by utilizing the tensile strength from several layers of reinforcement
stability due to its excess pore water pressure. Furthermore, due to a elements in the compacted soil filling to improve the stability of the
rise of ground water level, the pore water pressure will increase and structure. Reinforced soil structures usually consist of the tensile
finally reducing the shear strength of the soil until the retaining reinforcing elements in the filling soil, facing panel unit and drain-
structure or soil slope cannot withstand its own mass. age system.
Certainly, the occurrence of landslide has impacts on the sur-
rounding environment, both socially or economically. For instance,
if a landslide occurs in the slope of a mountainous area, the whole
access to the mountain would be blocked and eventually would dis-
rupt the local economy around. By performing a quick remediation
technique and procedure the risk and losses can be minimized.
There are several types of remediation for stabilizing the landslide
slope which are:
1. Geometry modification;
2. Surface erosion control;
3. Drainage techniques;
4. Retaining structure, such as mass gravity or reinforced soil Figure 1 Typical section of a conventional gravity retaining walls
structure. (source: https://commons.wikimedia.org/wiki/File:Gravity_Walls.jpg)
Reinforced soil structure is one of the permanent solution to
construct cost and time effective soil structures in an environmen- There are several types of facing unit in the reinforced soil struc-
tally friendly and sustainable manner. This method consumes less ture, namely, concrete panel, anchored gabion, wrap around geotex-
time in terms of construction period due to its simplicity and does tile or geogrid. Each facing system has its own advantages and dis-
not require time to be able to regain its strength (curing time), unlike advantages. Accordingly, the decision of the facing type shall con-
concrete structure. Another advantage is that it permits the use of sider all the aspects from installation, availability of facing material,
any available filling-material on site which usually the landslide soil site difficulties, economical aspects, to the drainage system. In order
material itself. In addition, this system also does not require any to allow the run-off water to flow before saturating the structural
special equipment or specific labour expertise, the only heavy backfill soil, the structure must be permeable enough. For instance,
equipment needed are excavator and soil compactor. Hence, rein- by contrast to the impermeable behaviour of concrete material,
forced soil structure is the recommended method to restore the granular soil is more permeable so that it is required to be placed
slopes after landslide and any other emergency works related. behind the concrete facing or panel.
569
While, the most commonly used for reinforcement materials are information, the project site is classified as the loose and medium
steel meshes, steel strips and geosynthetics in the form of woven dense material.
geotextiles, geogrids and geocomposites. But, in cases of the land-
slide remediation or emergency works, steel strips are not suitable
due to its stiff and heavy form that require a lot of effort and time to
be installed. Therefore, geosynthetic considered as the most suitable
material for this occasion because of its form and weight that easy to
handle and install.
Geosynthetic have been known a long time ago and well ac-
cepted in this construction industry. For the retaining soil structure,
the primary function of the geosynthetics needed is reinforcement.
This reinforcement works thanks to its tensile strength that can resist
the stress and deformation in the structure. Some type of geosyn-
thetic that can be applied for the reinforcement function are geogrid, Figure 4 Soil stratigraphy based on bore hole test.
geotextile, or geocomposite.
Reinforced soil structure offers more flexibility that the structure 3.2 Solution
can accommodate differential settlement and give the structure bet-
The proposed solution is a 8 m reinforced soil structure, using
ter capability in terms of drainage. Recently, this system is becom-
woven geogrids with the commercial name MacGrid as a primary
ing more often applied in private and government projects even for
reinforcement and anchored gabions (Terramesh System) will act as
landslide remediation or emergency works. The main factors that
a front wall fascia unit to maintain stability in the front of the struc-
driving the choice of this systems are their cost-effectiveness, flexi-
ture and protect the structure from surface erosion.
bility and most importantly the construction period to construct this
Terramesh System are reinforcing elements made up of hexago-
structure that is relatively faster in terms of implementation than the
nal double twisted steel wire mesh type 8x10, forming a facing ga-
other permanent solutions such as concrete and masonry walls.
bion and a variable length reinforcement according to the project.
The steel wire used for manufacturing Terramesh System units is
heavily galvanized. A polymer coating in compliance with EN
10245-1 having a nominal thickness of 0,50 mm is then applied to
provide additional protection against environmental agents (UV
rays, rain, etc.). The reinforcement base panel, front panel and ga-
bion box lid of the Terramesh System are manufactured from one
continuous panel of mesh; the back panel is factory connected to the
base panel. The facing section of the unit is formed by connecting
the back panel and diaphragms to the main unit; the gabion facing is
divided into uniformly partitioned cells by internal diaphragms.
Figure 2 Typical section of a reinforced soil slope combining poly- Steel ring fasteners or coated lacing wire are used to connect units
meric reinforcements (geogrids) and steel wire mesh reinforcements and to close and secure them.
(anchored gabions and steel wrap around units)
3 Leuwigoong Dam Access Road - Garut

3.1 Project Background
Leuwigoong Dam was built for irrigation purpose for citizens agri-
cultural lands in Garut regency, West Java Province, Indonesia. As
seen from figure below that the stone masonry at the access road
along 45 meters collapsed on 24th November 2017. This incident
happened after heavy and continuous rain the night before.
Figure 5 Terramesh System as facing unit
The manufacturing speed rate for producing Terramesh is rela-

tively fast depending on needs. Units that delivered to site were
folded and compressed in bundles weighting differently according to
the unit as well as wire/mesh size and measuring approximately
3.1x2.1 m in plan and 0.4 m in height. Moreover, type and size de-
termine the number of units per bundle. The Terramesh can be de-
Figure 3 Site condition on 27th November 2017 livered in large numbers at once with this bundle form.
This structure allegedly collapsed due to poor quality of back-

filling compaction so that the run-off water from the rain can infil-
trate inside the embankment. This also compounded by the poor
drainage system on the retaining wall which not enough permeable
to allow water flow from the backfill to outside the structure.
The client needs a low-impact, environmentally friendly and
time-cost effective integrated solution to prevent any construction
delay of the restoration of the gravity wall due to the urgency of the
usage of the access road. The Geotechnical investigation given by
the client is only the Standard Penetration Test (SPT) since the
structure was already built a few years ago and according to that Figure 6 Bundles of Terramesh System.
570
Anchored gabions (Terramesh) are filled at the site with suitable

stones to form a flexible and permeable modular system suitable as
reinforcement and fascia of MSE walls. Polyester woven geogrids
(MacGrid) with an Ultimate Tensile Strength (UTS) of 150 kN/m
were used as primary reinforcement, enabling the soil to perform
better than it would in its unreinforced state so that the structure can
accommodate greater loads and stand at steeper angles. Geogrids
were installed with a vertical spacing of 1 m and a length of 6 m.
Reinforcement of Terramesh systems rely on the strength of friction
and interlocking between plaiting panels and increasing the shear
strength of the soil in the form of the embankment behind the
Teramesh unit and serves to cut the local collapse line. To prevent
the backfill soil to penetrate out to the facing, non-woven geotextiles Figure 9 Installation of Gravel as a Drainage Material
which acts as separator and filter were applied behind all the
Teramesh facing. The figure below the typical cross section of the The construction process is repeated until it reaches the designed
proposed solution. height. The total construction time of this reinforced soil structure
were completed in 6 weeks from January to February 2018 (see
Figure 9).
Figure 10 View of the Reinforced Soil Structure during construction

(left) and almost completed (right)
Figure 7 Typical Retaining Structure Cross Section.
4 Bukit Kemuning Road – Batam
The backfilling soil was transported by trucks and unloaded with
4.1 Project Background
the help of heavy equipment. The use of heavyweight machinery
close to the facing (within 1 m from the facing) as well as track The project site is located in Bukit Kemuning, Batam and adjacent
vehicles in contact with geogrid reinforcement shall be avoided. to the local housing. The area is subjected to high intensity of rain-
Vibratory rollers of adequate dimensions was used in function of the fall and surrounded by hills and mountains. Due to the heavy rain-
filling material (as per ASTM D3282). Light vibration compactor falls and poor drainage system of the road, the rain water run-off
has been used for the area close to the facia to minimize any damage from the mountain and triggered a massive landslide. The landslide
to the facia units. occurance affected the local residence near the area. Some houses
were damaged, and residents were evacuated. Moreover, the acci-
dent has also disrupted the traffic. The Ministry of Public Works
planned to repair the damage road and rehabilitate the existing soil
profile in order not to cause greater losses.
Figure 11 Rain Water Run-off Situation (left) and Condition after

Figure 8 Installation of Terramesh and MacGrid as Reinforced Soil Collapse (right)
Structure
In this project, soil investigations were carried out using the
In addition, the internal drainage system inside the structure has Standard Penetration Test (SPT). The SPT test was performed at a
been done by placing available permeable material, namely, rock or depth of 5 m from the existing road and the testing was executed to
gravel. The thickness of the rock material is 30 cm and wrapped a depth of 9 m. From the results of the SPT test, it was found that
with geotextile as separator. And for the horizontal drainage applied the value of N SPT was quite high and the type of soil was domi-
by using drainage pipe with proper inclination to collect and channel nated by gravelly silt. The groundwater elevation indicated in 1.4
the water out. meters depth. The SPT test results obtained are as follows.
571
Figure 14 Paralink Installation (Left) and Terramesh Installation

(Right)
Selected backfill material for the embankment body was silty

sand soil, compacted every 25 cm thick lift. Structural filling instal-
lation and compaction procedures are illustrated in Figure 15.
Figure 12 Standar Penetration Test (SPT) Result
4.2 Solution
The Public Works decided to repair the road and improve the drain-
age system, so that this problem can be avoided in the future. The Figure 15 Backfill and Compaction Procedures
original proposal from the consultant was a gravity mass retaining
structure made from gabion with woven geotextile as the primary
reinforcement and separately placed behind the gabion wall as the
separator. Due to its massive structure, the remediation act shall be
difficult to realize due to limitation of funds and time, so that the
client asked for an alternative solution in order to generate both cost
and construction schedule optimization. It is important to note that
the client must restore the function of the road as soon as possible to
reduce the risk of loss of the local community.
Therefore, a retaining soil structure with the combination of an-
chored gabion facing and geogrid reinforcement was adopted. The
structure itself can reach 14 m maximum height with the combina- Figure 16 View of the Completed Access Road Remediation
tion between anchored gabion facing and polymer reinforcement.
The proposed system was the combination of Maccaferri’s Ter- 5 Conclusions
ramesh System and High Strength Geogrids Paralink. The geogrid Remediation works after landslide needs to be performed immedi-
ultimate tensile strength (UTS) is 300 kN/m. The system was pro- ately to minimize the risk or losses that can be caused. Reinforced
vided with an internal drainage using a 30 cm thick rock layer soil structure is one of the most suitable solution to restore the
wrapped with geotextile. slopes after landslide and any other emergency works related. This
This new technical proposal can be concluded that the retaining method proved to be more effective in terms of construction time
soil structure was found to be more cost and time effective com- and cost compared to the other solutions. This paper presented two
pared to the original one. The retaining soil structure proposed re- cases that applied reinforced soil structures with polymer reinforce-
quires only one layer of anchored gabion as the facing instead using ment as the remediation method after landslide occurrence in Indo-
the massive gabion structure as the mass gravity wall. nesia.
The first case history is the remediation of 45 m access road for
Leuwigoong Dam in Garut regency, West Java. Landslide happen
after heavy rain all night long and allegedly collapsed due to poor
drainage system of the existing retaining structure. Reinforced soil
structure was applied using woven geogrid as the primary rein-
forcement and anchored gabion as a structural facing and secondary
reinforcement. The internal drainage system inside the structure has
been done by placing available permeable material, namely, gravel.
The thickness of the gravel material is 30 cm and wrapped with
geotextile as separator. Furthermore, the horizontal drainage applied
Figure 13 Typical Cross Section of Reinforced Soil Structure Bukit by using drainage pipe with proper inclination to collect and channel
Kemuning Road in Batam the water out.
The second case is the remediation of 100 meters road in Bukit
Static and seismic analysis had been performed, checking global Kemuning road in Batam. The existing slope collapsed under con-
stability, internal stability and stability as retaining wall (sliding, tinuous exposure to high intensity rainfall and as a result, adjacent
overturning and bearing capacity) using Maccaferri internal soft- residential areas were damaged. Consequently, the Reinforced soil
ware Macstars W. Maccaferri provided support to the contractor structure was chosen as the alternative solution to optimize cost and
with product installation training and supervision. construction schedule from the original solution which was mass
The geogrids are supplied in rolls with 4.5 m width and 200 m gravity wall in the form of gabions. This choice allowed the owner
long. The total weight of each roll geogrid (Paralink) approximately to deliver the project on time and within the target budget. The se-
around 900 kg. Geogrids are arranged before the Terramesh facing lection of the system was also guided by the system fascia perme-
units, from the front of facing up to the length required by the de- ability, flexibility and construction ease. Reinforced soil structure
sign. The laying direction will be in the rolling direction that having proposed was anchored gabion and high strength geogrid 300 kN/m.
greater resistance has to be placed perpendicularly to the facing. The construction itself was completed in approximately 4 months.
572
6 References
Koerner, R.M. (2005). Designing with Geosynthethics, Pearson and Metallic Reinforcements using Limit Equilibrium Meth-
Prentice Hall, New Jersey, 5th Edition. ods, Indonesian Geosynthetics Conference 2016, Indonesian
Lelli, M., Cerro, M., Scotto. M., and Rimoldi. P. (2017). Geosyn- Chapter of International Geosynthetics Society, Jakarta.
thetic-reinforced soil structures, recent Indonesian experi- Lelli, M., Gouw,T.L., and Laneri, R. (2017). The Important Role of
ences, Indonesian Chapter of International Geosynthetics So- Execution and CQA/CQC Plans in Building with Geosynthet-
ciety, Jakarta. ics, Geosintetik Indonesia 2017 seminar on Quality in Con-
Lelli, M., Gouw, T.L., Laneri, R., Cerro. M., and Meinata L.E. struction using Geosynthetics, by Indonesian Chapter of In-
(2016). Design of Earth Structures Reinforced with Polymeric ternational Geosynthetics Society, Century Hotel, Jakarta.
573
Geotechnical Properties of Cement-Stabilized Mine Tailings from Brgy. Gango,

Libona, Bukidnon, Philippines
Einstine M. Opiso, Reinerio P. Supremo, Reymar T. Rejas and Jemima R. Perodes
Department of Civil Engineering, College of Engineering, Central Mindanao University, Musuan, Bukidnon, Philippines
E-mail: einstineopiso@cmu.edu.ph
ABSTRACT: Improving the geotechnical engineering properties of mine tailings using cement stabilization process can be adopted to
reduce the volume of tailings and consequently the risk of toxicity it poses to people and the environment. Geotechnical properties of
cement-stabilized mine tailings from Brgy Gango, Libona, Bukidnon, Philippines were investigated to assess its applicability for reuse as a
construction material. Mine tailings were stabilized using ordinary Portland cement in four different formulations—0%, 6%, 8% and 10% by
weight of soil and were subjected to compaction test, California bearing ratio (CBR) test, permeability test, unconfined compressive strength
(UCS) test and direct shear test in accordance with ASTM standards. Based on the AASHTO Soil Classification System, the soil was
classified as clay of low plasticity. The maximum dry density and optimum moisture content showed a decreasing trend from the varied
mixture which resulted that adding cement did not improve the compaction characteristics of the stabilized mine tailings. Addition of OPC
up to 6% and 8% had a significant effect on the increase of CBR and UCS values, respectively. In permeability test, it was observed that the
rate of flow of water increases as the percent of OPC increases. The increase of the coefficient of permeability was influenced by the
cementitious bond which forms new solid material within the treated sample. The soil particle becomes bigger in size and the rate of flow of
water increases. Up to 6% addition of OPC is the optimum cement percentage to stabilized the mine tailings based on direct shear testing. In
overall, the cement-stabilized mine tailings from Brgy. Gango, Libona, Bukidnon, Philippines can be safely used in sub-base road
construction and as typically stabilized wastes for land disposal.
Keywords: geotechnical engineering, mine tailing, cement-stabilized soil, road construction.
1. INTRODUCTION One of the sites in Mindanao, Philippines where the problem of

ASGM tailings disposal is a great environmental concern is in
Mining industry is one of the most prominent and productive
Libona, Bukidnon. Libona is a first class municipality and known to
economic sector in the Philippines. Specifically, artisanal small-
have abundant gold deposits. Small-scale mining operations began
scale gold mining (ASGM) is the most common operation activity
to operate in Sitio Manlahuyan, in Barangay Gango in 1986. Large
which contributes to about 70-80% of the total gold production gold
amount of ASGM tailings are still present in the site that need to be
production in the country (Israel and Asirot, 2002). The extraction
managed properly in order to minimize its negative effects to the
of gold extraction is via the amalgamation process which relies on
community and to the environment (Canencia et al., 2015).
heavy use of mercury. This result to production of contaminated
mine tailings that have been proven to contain up to 250 grams per In order to address the disposal of ASGM tails in Libona, mine
ton of mercury (Appel, et al., 2011). wastes will be stabilized using Ordinary Portland Cement (OPC)
and will be assessed for its applicability for reuse as a construction
Despite its economic benefits, mining operations often face
strong opposition from affected communities because of its adverse material. Hence, the main objective of the study is to determine the
environmental and social effects (Adajar and Zarco, 2013). These geotechnical properties of cement-stabilized ASGM tailings in Sitio
wastes carry toxic material which may harm the people and the Manlauyan, Gango, Libona, Bukidnon. The cement-stabilized mine
environment if not properly managed. The disposal of the enormous wastes were subjected to compaction, California bearing ratio
amounts of ASGM tailings regularly produced from mining (CBR), falling head permeability, unconfined compressive strength
operations is the main environmental issue associated with mining (UCS) and direct shear in accordance with American Standard and
activities. The mine wastes might have sufficient amounts of
Testing Materials (ASTM) and its specific construction application
mercury to cause danger to all living organisms as it pollutes the soil
and water. was determined.
574
2. METHODOLOGY
2.1. Experimental Design

Figure 2. Collection of Soil Sample in Gango, Libona, Bukidnon.
Three (3) soil samples at each of the replacement were prepared and
mixed at 0%, 6%, 8%, and 10% of OPC using soil mixer with a
water-cement ratio of 0.4 as the design mix of the study presented in
Table 1. Digital weighing balance was used in weighing the soil 2.3. Determination of Grain Size Distribution
samples based on the designed mixture.
The Grain-size tests of the soil were performed in accordance with
the ASTM D421 (Standard Method for Sieve Analysis of Fine and
Coarse Aggregates). The soil was sieved through the nest of fine
Table 1. Design Mix of the Study. sieves, sieve numbers 3”, 1”, ¾”, ½”, 4, 10, 20, 40, 60, 100, and
200. A mechanical shaker was used having a minimum of 10
Treatment Description minutes shaking. The soil fraction retained on each sieve in a
separate container was collected and masses were taken. The
T0 0% additional OPC
percentage retained, cumulative percentage retained, and the
percentage finer was determined.
2.4. Determination of Atterberg Limits
The Atterberg Limits were determined in accordance with ASTM
D4318 - Standard Test Method for Fall Cone Test. The soil sample
is placed in a 55 mm diameter, 40 mm deep metal cup. The cone is
2.2. Sample Collection
released for 5 seconds so that it may penetrate the soil. Liquid limit
Mine tailings were collected in the common mine tailing pond at is defined as the water content of the soil which allows the cone to
Gango, Libona, Bukidnon with a latitude of 80 20’ 25” N and penetrate exactly 20 mm during that period of time. The sample was
longitude 1240 43’58” E (Figure 1). Actual collection of soil allowed to dry on the glass plate until it become plastic enough to be
samples were shown in Figure 2. The soil samples were air-dried at shaped into a ball. The sample-ball was molded and rolled between
the laboratory. The dry disturbed soil samples were pulverized using the palms of the hands until cracks on the surface of the sample
rubber mallet and undergone sieve analysis through U.S sieve no. appeared. Then, sample was divided into small pieces and rolled to
40. form thread having 3.2 mm diameter under enough pressure. The
first crack appeared on the thread surface was determined as plastic
limit. Plasticity index was calculated as the difference between the
liquid limit and the plastic limit which measures the plasticity of the
soil.
2.5. Soil Stabilization
An amount of water equivalent to 0.4 water to cement ratio was

added to the four (4) treatments of pulverized soil sample. The
sample was mixed for about 2 minutes until homogenous mixtures
were attained and were stored and zipped.
Figure 1. Location of the study (common mine tailings pond in

Gango, Libona, Bukidnon).
2.6. Curing of Soil Samples
575
The stabilized samples were cured for 28 days to let the cement was attached to the syringe to the stand pipe. The valves were first
hydrate and react with the soil. The samples were zipped to prevent locked to ensure that no water may pass through the cylinder. Water
the moisture to evaporate during the curing period and stored away was poured above the stand pipe through a funnel. The safety valve
from direct sunlight. All the samples after 28 days curing were was removed leaving a hole above the cylinder and the upper valve
subjected to testing. was turned on. Bubbles came gushing out on the hole at the safety
valve as water flowed through the cylinder. The lock was then
attached back to the safety valve. The upper valve was also turned
off when there were no more bubbles which were signs of air voids.
2.7. Compaction Test After it was turned off, a timer was set together when the upper and
lower valve handle attached above and below the cylinder was
The modified proctor test was conducted to all the samples based on
turned on and was observed until it penetrates the soil. When it
ASTM 1557 and AASHTO T180 in order to determine
penetrated the soil, a drop of water was be noticed from the faucet
experimentally the optimum moisture content at which the tailings
attached below the cylinder. The timer was stopped and recorded
became most dense and achieved its maximum dry density. The
and the upper faucet was turned off to record the initial height of
proctor compaction test consists of compacting soil samples at a
water at the stand pipe. Another timer was set one (1) minute after
given water content in a standard mould with standard compaction
the faucet was turned on again for three (3) trials to record initial
energy. It uses a 4-inch-diameter mould with the compaction of five
and final heights. The coefficient of permeability was then
separate layers of soil using 56 blows by a 4.5 kg hammer falling 18
determined by using the formula given below.
inches.
Each of the soil samples were first air dried and then separated
into 5 samples. The water content of each sample were adjusted by
(1)
adding water (3% - 5% increments or more depending on the type of
the soil). The soil samples were then placed and compacted in the
Proctor compaction mould in five different layers where each layer
received 56 blows of the standard hammer. Before placing each new
Where
layer, the surface of the previous layers were scratched in order to
ensure a uniform distribution of the compaction effects. At the end
K = coefficient of permeability in cm/s
of the test, after removing and drying of the sample, the dry density
and the water content of the samples were determined for each a = inside cross-sectional area of stand pipe in cm2
Proctor compaction test.
L = length of specimen in cm
h1 = initial head in cm
2.8. California Bearing Ratio Test
A = cross-sectional area of specimen in cm3
The California Bearing Ratio was determined in accordance with
ASTM D1883 (Standard Test Method for California Bearing Ratio t = time interval in seconds in which the head drops from h1 to
of Laboratory –Compacted Soils). The soil specimens were h2
compacted so that their compacted densities ranges from 95% to
100%. The samples were cured for about four days. Penetration h2 = final head in cm
pistons were applied to the samples. The optimum mixture of soil
and cement was determined based on the maximum CBR observed
from the several mixtures that was used.
2.10.Unconfined Compressive Strength Test
The prepared specimens followed the procedures provided by

2.9. Falling Head Permeability Test ASTM D2166 -Standard Test Method for unconfined compressive
strength of cohesive soil. The treated samples were added with
The falling head permeability test was conducted using Standard water according to the attained optimum moisture content and
Test Methods for Measurement of Hydraulic Conductivity of maximum dry density. The compacted stabilized soil was extruded
Saturated Porous Materials Using a Flexible Wall Permeameter from the tube sampler and were cut to obtain the length to diameter
(ASTM D5084). The soil samples and the water required for mixing ratio (L/d) to be approximately between 2 to 2.5, and were placed
were first weighed according to its data on maximum dry density inside the cellophane. The diameter and length of each specimen
and optimum moisture content from the compaction test results. were measured, and weighed. The specimen was placed in the
After the mixing process, it was compacted inside the cylinder with compression device and was centered at the bottom plate. The
3 layers and 25 blows per layer using a rubber mallet. The sample device was adjusted so that the upper plate made contact with the
was then confined inside the cylindrical mould and the permeameter specimen and the load and deformation dials were set to zero. The
576
cylinder of soil without lateral support was tested to failure in 3.2. Soil Classification
simple compression, in a constant rate of strain.
Table 3 shows the soil classification of the samples based on
American Association of State Highway and Transportation Official
(AASHTO) and Unified Soil Classification System (USCS). T0 and
2.11.Direct Shear Strength Test T3 sample were both classified under the group A-7-6, while T1 and
T2 sample were under group A-6. All samples belong to fine-grained
The test was performed in the laboratory according to the standard soil category for having 50% or more percent passing through sieve
test procedure ASTM D3080 – Direct Shear Test of Soils for no. 200, and were classified as CL or clay of low plasticity/lean clay
Consolidated Drained Conditions. Treated soil samples was mixed based on USCS soil classification system.
with water according to the moisture content result from compaction
test. It was compacted progressively by 1/3 of its total volume
having 25 blows. The compacted soil was carefully removed from
the sampler and placed inside a clean wrapper. Table 3. Soil Classification of the Samples
The dimension of each specimen was measured. Area and Treatment AASHTO USCS Soil Plasticity
volumes were calculated. The specimen was placed in smooth
layers. The upper grating on stone and loading block on top of soil T0 A-6(13) CL Medium Plasticity
were positioned. Desired normal load was applied before shear pin
was removed. The dial gauge which measures the change of volume T1 A-7-6(19) Cl Medium Plasticity
was then attached. Initial reading of the dial gauge and calibration
values were recorded. As the motor was operated, the reading of the T2 A-7-6(19) CL Medium Plasticity
shear force was recorded as well as the volume change readings
T3 A-6(14) CL Medium Plasticity
until failure.
3.3. Atterberg Limits

T0 sample as shown from Figure 3 has the least value of liquid limit
of 37.45%. Lean clays have correspondingly low liquid limit which
3.1. Grain Size Analysis indicate low compressibility and low in shrinks swell tendencies.
Increasing trend in LL was observed as percent of OPC content
The percentages of gravel , sand and clay-silt were summarized in increases. Liquid limit values vary from 37.45% to 41.01%,
Table 2. There was an increase in clay and silt percentage from 97. improvement of liquid limit attributed that more water is required
14% to 97.59% when 6% of OPC was added to the sample. for the cement treated soil to make it fluid. Figure 4 shows the
However, the percent passing gradually decreases as 8% and 10% of plastic limit of all the samples which vary from 23.83% to 26.1%.
cement were added to the soil sample giving the percent passing Figure 5 indicates that as cement content of the sample increases,
values of 97.48% and 95.88% respectively. This was probably due the plasticity index also increases up to 8%. All the samples were
to the chemical reaction between cement and water that produces classified soil with medium plasticity. High plasticity soil exhibits
cementitious compound which has larger particle size compared to low strength and low compressibility. This indicate that the
the tailing sample which resulted to the decrease of percentage controlled sample (T0) has the best plasticity characteristic among all
passing (Das, 2015). the treatments..
Table 2. Soil Type Percentages

45
Treatment Gravel Sand Clay and Silt
Liquid Limits (%)
T0 0 2.86 97.14 41,01 41,01

40
T1 0.18 2.23 97.59 37,45 39,11
35
T2 0.16 2.36 97.48
0% 2% 4% 6% 8% 10%
% of OPC
T3 0.38 3.74 95.88
Figure 3. Liquid Limit of the Soil Samples.
577
30
Plastic Limits (%)
26,1
25,31
25
23,83 23,83 14
Optimum Moisture Content, %

13,5 13,57
20 13
0% 2% 4% 6% 8% 10% 12,72
12,5
% of OPC
12
11,5 11,53 11,56
Figure 4. Plastic Limit of the Soil Samples. 11
0 2 4 6 8 10
% of OPC
Figure 6. Variation of OMC with varying amount OPC
20
Plasticity Index (%)
17,18 17,18
15
13,01 3.5. California Bearing Ratio
12,14
10
The addition of OPC in samples has increased the CBR values as
5 observed Figure 7. Bowles (1992) stated that the subgrades having
CBR value of 3-7% are poor to fair quality of subgrade and greater
0
0 2 4 6 8 10 12 than 50% CBR value can be used as base material . T0 is under poor
to fair subgrade with the CBR value of 4.87%. T1, T2 and T3 have
% of OPC 105.67%, 79.84% and 93.81% CBR values, respectively. All percent
additives are excellent subgrade that can be used as base.
Figure 5. Plasticity Index of the Soil Samples.
3.4. Compaction
120
The cement-stabilized soil samples showed an increasing trend of 105,67
California Bearing Ratio, %
OMC from 11.5% to 12.72%, smaller compared to controlled 100

93,81
sample as shown in Table 4. These reduction of OMC and MDD
80 79,84
valued is attributed to the addition of fines in the form of cement.
The study of Manjunatha and Sunil (2013) showed similar effect of 60
decreasing result of OMC and MDD using different stabilizing
agent. 40
20
4,87
0
Table 4. Compaction Test Result. 0% 2% 4% 6% 8% 10% 12%
% of OPC
Treatment Optimum Moisture Maximum Dry Density
Content (%) (kg/m3)
T0 13.57 1751.39
Figure 7. California bearing ratio of cement stabilized mine tailings.
T1 11.53 1587.51
T2 11.56 1524.29
3.6. Permeability Test Result
T3 12.72 1475.77
578
The coefficient of permeability (K) increased when added with OPC

influenced by the time it was started to be tested and the percent of 350 320,8
cement added. The K values of T1 and T3 were lower compared to 293,24
300
Unconfined Compressive
T0 and T2. Setting aside the soil sample reduces its moisture content
250
due to evaporation and causes the optimum moisture content to
Strength, KPa)
239,92
shrink. 200
130,04
150
The decrease or increase of K was influenced by the formation 100
of cementitious bonds which creates new solid material within the
50
treated mine tailings. These tailings are considered to have low to
very low permeability typical of silty sand. This factor increases the 0
0% 2% 4% 6% 8% 10%
average pore size to create a macro-pore structure which causes an
% of OPC
increase in K (Tay & Goh 1991).
Figure 9. Variation of unconfined compressive strength with OPC.
0,00065497
3.7. Direct Shear Stress
0,0007 0,0006491
0,0006 It was observed in Figure 10 the relationship of shear stress with
Coeffiicient of Permeability , K
0,0005 time increases until failure occurs which is at its peak. Compared to
0,00035289 all other treatments, T2 has the maximum shear stress which is
0,0004
166.11 kPa that was observed at 30 seconds of the testing and
(cm/sec)
0,0003 ultimately settles at 59.52 kPa beginning at 40 seconds. This

0,0002 0,00019442 suggests that addition of 6% OPC gave the values of shear stress
0,0001
that can be mobilized within a soil mass without failure taking place.
0
0 2 4 6 8 10 12
% of OPC 180
160 T0
Figure 8. Coefficient of Permeability of Soil Samples. 140 T1

Direct Shear STress, KPa
120
T2
100
T3
80
3.6. Unconfined Compressive Strength 60
40
Unconfined compressive strength is a control parameter for road fill 20
design. Achieving a compressive strength greater than 345kPa can 0
help to significantly reduce the potential for settlement in deep fills 0 30 60 90 120 150 180 210
(Ferguson and Levorson, 1999). Sample added with 8% of OPC has Time, s
the highest value of UCS with 320.8 kPa, not reaching the desired
value for deep fill purposes. However, the stabilized soil were Figure 10. Time vs. Average direct shear stress graph of soil
classified as very stiff (200kPa- 400KPa) clay consistencies based samples.
on the classification of Das (2010). According to Amini et al.
(2011), the effect of longer curing time (from 14 to 28 days) has
greater effect to the mixture.Thus, the sudden decrease of strength
may be also caused by the number of curing days. The UCS of soil 4. CONCLUSION
significantly increases with the curing period and it is more
Based on the experimental results, the following conclusions were
consistently increasing up to the first 14 days according to Saha et
drawn:
al. (2013).
i. The physical property of the artisanal gold mine tailings found
in Gango, Libona, Bukidnon, Philippines were classified as olive
brown and identified as silts with traces of sandy clays by the
results in the grain size distribution. It has the value of 37.45%,
25.31% and 12.14% for liquid limit, plastic limit and plasticity,
579
respectively. The soil was classified as A-6 (13) soil according to ASTM D6913 Standard test methods for particle-size distribution
ASSHTO classification and CL soil based on USCS classification. (gradiation) of soils using sieve analysis.
The soil was low plasticity that implies low compressibility and
lower tendency in swelling which was a good attribute for the ASTM D2487-11 Standard practice for classification of soils for
acceptability of earthen materials in road construction applications; engineering purposes (Unified Soil Classification).
ii. Cement causes the decrease in both MDD and OMC of the mine ASTM D2166 -Standard Test Method for unconfined compressive
waste. Controlled (T0) sample acquired the largest maximum dry strength of cohesive soil.
density and optimum moisture content value of 1751.39 kg/m3 and
13.57%, respectively. ; ASTM D3080 – Direct Shear Test of Soils for Consolidated Drained
Conditions
iii. Stabilized soil samples have the CBR values of greater than
50% classified as excellent quality for sub-grade that can be used as Adajar, M.A. and Zarco, M.A.. 2014. An empirical model for
base materials; predicting hydraulic conductivity of mine tailings, Dec.,
2014, Vol. 7, No. 2 (Sl. No. 14), pp. 1054-1061.
iv. As the percent of OPC increases, the coefficient of permeability
also increases. Stabilization using OPC gave a great significance in Appel, P., Na-Oy, L., Hatsukawa, Y., Osawa, T., Kystol, J., and
the results yet it produced higher permeability results than that of Sorensen, L. (2011) “Cleaning mercury polluted mine
the controlled treatment because OPC formed bonds between tailing tailings in the Philippines”.
particles that creates new solid material and caused faster discharge
Alhomiar, S. (2016). Thesis-Hydraulic Conductivity of Fly Ash-
of water through the sample.
Amended Mine Tailings. Retrieved:
v. Sample added with 8% of OPC has the highest value of UCS https://dspace.library.colostate.edu/bitstream/handle/1021
with 320.8 kPa, not reaching the desired value for deep fill purposes. 7/176619/Alhomair_colostate_0053N_13648.pdf?sequenc
However, the stabilized soil were classified as very stiff (200kPa- e=1.
400kPa) clay consistencies;
Bowles, J. (1992). Engineering properties of soil and their
vi. As the cement content for each treatment increases, there is an measurements, 4th Edition.
evident decrease in soil strength but values were still higher than of
Canencia, O., Palmes, N., Ibonia, S., Talampas, W., Lammawin, V.,
the controlled sample; and
Kawaling, G., Yañes, S., and Savaña, MJ. (2015) “Minin
vii. The study indicates that cement can be very effective in Industry in Northern Mindanao: Its Environmntal, Social
stabilizing the ASGM tailings up to 6% addition of cement that can and Health Impact Toward Responsible Mining. 2015”.
be safely used in sub-base and base road construction and in typical
Das, B.M. 2010. “Principles of geotechnical engineering”. 7th Ed.
stabilized wastes for land disposal.
Stamfold: Cengage Learning”.
Das, B. M. (2015). Fundamentals of Geotechnical Engineering, 4th

Edition.
5. REFERENCES Ferguson G. and Levorson S.M. (1999). “Soil and Pavement Base
Stabilization with Self-Cementing Coal Fly Ash.
American Coal Fly Ash Association International,
Alexandria, VA
ASTM D421 – Standard Method for Sieve Analysis of Fine and
Coarse Aggregates. Israel, C. and Asirot J., 2000. Mercury pollution due to small-scale
gold mining in the philippines: an economic analysis,
ASTM D3080 - Direct Shear Test of Soils for Consolidated Drained 2000-06, 1-62.
Conditions.
Sivakugan, N.,Rankine, R.M., Rankine, K.J., and Rankine, K.S.,
ASTM D4318 – Standard Test Method for Fall Cone Test. (2006). Geotechnical considerations in mine backfilling in
Australia. Journal of Cleaner Production, 1168-1175.
ASTM C136. “Standard method for sieve analysis of fine and coarse
aggregates”. ASTM International. Supriya Saha and Sujit Kumar Pal. (2013). “Influence of Fly Ash on
Unconfined Compressive Strength of Soil and Fly Ash
ASTM D1883. “Standard test method for California Bearing Ratio Layers Placed Successively”.
of laboratory – compacted soils”. ASTM International.
580
Tay, J. H., & Goh, A. T. (1991). “Engineering properties of

incinerator residue. Journal of environmental engineering”
117(2), 224-235.
Acknowledgment:
The authors would like to the Central Mindanao University for the
support in the implementation of the study as well as the
Department of Science and Technology – Philippine Council for
Industry, Energy, and Emerging Technology Research and
Development (DOST-PCIEERD) which funded this research.
581
A New Breakthrough – Application of Control Modulus Column for Settlement and

Stability Control under Soft Soil and High Embankment Load at Pemalang-Batang
Toll Road
Ryan Rahmat Setiaji1, Panji Utomo2, KM. Abuhuroyroh3

1
2
3
Independent Technical Expert for Ground Improvement, Bandung, Indonesia
E-mail: ryan.setiaji@menard-asia.com; panji.utomo@menard-asia.com; roy.yusfi@gmail.com
ABSTRACT:
Pemalang-Batang (Pematang) Toll Road as a part of Trans Jawa Toll Road is currently being constructed. The Pematang Toll Road which has
39.2 km length with 13 embankments before bridges (bridges approach / oprit) will be constructed above very soft river sedimentation soil.
The construction of embankment on the soft soil leads to excessive settlement and high susceptibility toward slope stability failure. Ground
improvement using Controlled Modulus Columns (CMC) is adopted to reinforce the soil, increase the strength thus reduce the settlement and
stabilize the embankment. The application of CMC at Pemalang-Batang Toll Road is indeed a breakthrough in history of toll road construction
as a new construction technology in Indonesia, where majority the settlement/stability issue commonly solved by structural or conventional
solution such as: slab on pile, cerucuk matras beton, PVD/Vacuum Preloading etc. Control Modulus Column is one of ground improvement
technique installed by forming a regular grid of semi-rigid inclusion to enhance global stiffness of soil mass so that the problem such as:
settlement, bearing capacity can be controlled. The installation of CMC itself performed using special design auger element that is able to force
the soil displace laterally induce volumetric expansion resulted a lateral compaction of surrounding soil. In Indonesia, Menard has been
successfully implementing CMC in several projects for different sector. The CMC’s design technology makes them uniquely efficient for the
immediate support of road for public transport, large liquid or bulk solid storage tanks warehouse, and other infrastructures building or facilities.
This paper gives a brief description of application of CMC for supporting embankment to achieve settlement criteria, special design concept,
and recommendation are briefly discussed.
Keywords: control modulus column, very soft soil, ground improvement, embankment, toll road
1. INTRODUCTION several projects in different sectors. CMC are vertical inclusions

installed in the treated ground to form a mass of semi rigid inclusion
Human has found ways of dealing with poor soil conditions using column. The composite mass (soil-CMC) has greater stiffness than
simple techniques such laying wooden log, conventional dynamic the initial untreated ground reducing global deformability, hence
compaction using human power. Significant efforts have been made reducing settlements induced by the weight of the structure to within
in developing variety and more advanced ground improvement allowable ranges.
methods. The developments took advantage of comprehensive Rather than transferring the load directly, the CMC utilizes a
research, innovative technology in equipment, advanced methods, well compacted granular fill with typical thickness varying from 0.4
sophisticated instrument to provide faster execution and reliable to 1.0m which serves as Load Transfer Platform (LTP). Figure 1
results. illustrates the load transfer mechanism of ground improvement using
It is rather common available large area with reasonable CMC as semi rigid inclusion in comparison with piling system. The
investment required for road, airport, and port facilities are located at CMC are suitable for structures with distributed loads to large surface
unpleasant location where significant weak soils such as peat, very areas (reservoir, warehouse, highway embankment) underlain with
soft clay, organic soil, loose sand is present. Often, in many industrial soft to medium soil.
practices when the time becomes a major constraint, two methods are
usually adopted; either installing piles to bypass the weak soil; or
performing engineering soil modification.
Ground improvement techniques are commonly categorized
into two categories; ground improvement without addition of
material such as consolidation and compaction method; ground
improvement with addition material such as stone column method,
soil mixing, controlled modulus columns, etc. Preliminary studies
therefore are conducted to determine the most adequate technique
with several considerations such as loading, ground condition, design
criteria to be achieved, project location, construction period.
In the beginning of 1990s, Menard patented the technology of
Controlled Modulus Columns (CMC). CMC is very common in
many countries in Europe, and it is either newly introduced or yet to
be widely used in Southeast Asia. Certain countries e.g. Malaysia
have about more than 15 years of history of using CMC. In Figure 1. Comparison of Load Transfer Mechanism between
Indonesia, Menard has been successfully implementing CMC in Pile (left) and CMC (right) (Asiri National Project, 2012)
582
Some features of the CMC technology are:

 Installation is performed using auger. In this manner, the
installation is quiet, vibration free and generates minimum
spoil.
(a) Lateral sliding failure (surface failure)
 Elasticity modulus of the CMC elements is typically 50-
3,000 times that of the soil (weakest stratum)
 A load transfer platform of generally granular fill (LTP) is
placed over the CMC reinforced ground.
 In granular soils, densification due to lateral displacement (b) Deep seated global failure
may occur between the columns by virtue of the
displacement drilling process.
2. CMC INSTALLATION METHOD (c) Lower layer extrusion with sinking
CMC are installed using a specially designed displacement auger.

Unlike the conventional auger which excavate the soil material and
produces large spoil and create disposal problem. This special auger
(d) Lower layer extrusion with ‘snapping’
displaces the soil laterally and densifies surrounding soil, increases
load bearing capacity and produces minimum spoil during Figure 3. Typical Failure of Embankment (Smith et al., 2016)
installation. The application of CMC was intended to increase soil bearing
When the auger reaches the designed depth, grout is pumped capacity thus reducing displacement into several limits and prevent
through the hollow stem of the auger and into the soil bore as the slope bearing capacity failure. High embankment generates axial
auger is retracted with calibrated pre-defined rate to avoid necking. forces (self-weight and traffic weight) and lateral forces. While the
Quality control of CMC is conducted with real time monitoring of CMC is commonly used for supporting axial load, due to its material
the following installation parameters: behavior. Therefore, additional material such as geosynthetic,
 Speed of rotation wiremesh, CMC with reinforced steel will be probably required.
 Rate of penetration and retraction of the auger Figure 4 below illustrated the forces of CMC below embankment.
 Torque The following section briefly describe technical aspect and design
 Depth of CMC concept for CMC to support high embankment:
 Installation time
 Grout pressure
 Volume of cement mixture with function of depth.
Figure 4. Typical Cross Section of CMC (Asiri National Project,

2012)
3.1 Force Equilibrium
The CMC has a different load transfer mechanism compared to pile

foundation. It does not transfer the load directly to the structures but
it utilizes some well compacted granular material as a load transfer
platform. The dimensions, spacing, and material of the CMC are
based upon the development of an optimal combination. The detailed
Figure 2. Installation of CMC by Displacement Method explanation of CMC's load transfer mechanism due to structural load
(Plomteux,. et al., 2000) are described below (Combarieu, 1988):
1. The imposed load is transferred to the CMC through LTP.
3. DESIGN CONCEPT CMC FOR HIGH EMBANKMENT The vertical displacement of the soil occurs due to
consolidation.
Placing embankment above soft layer shall be performed with slow 2. With the consolidation of the soil, vertical displacement of
increment within long time to let the soil release the excess pore surrounding soil is larger than CMC, therefore generating
water pressure and increase bearing capacity gradually. In case the negative skin friction along the CMC.
embankment was built recklessly, the low bearing capacity soil will 3. Vertical displacement of soil decreases gradually until
generate excessive settlement at the center and lateral deformation at equal to CMC (neutral plane), below neutral plane the
the slope due to slope failure as illustrated at figure below: vertical displacement of the soil is less than CMC at the
certain depth, thus CMC are able to generate positive skin
friction.
583
Figure 5. Load Transfer Mechanism (Plomteux,. et al., 2000)
The design concept of CMC should satisfy the principle of an

equilibrium state of load distribution over the full length of CMC
inclusions with the main components of acting forces:
1. The vertical load (Q) at the top of the CMC.
2. The negative skin friction (Fn) acting on the upper zone of
the CMC.
3. The positive skin friction (Fp) acting on the lower zone. Figure 8. Load Distribution Mechanism with LTP (Asiri
4. The vertical reaction at the tip (Qp). National Project, 2012)
3.3 Wiremesh
If horizontal forces caused by the active earth forces of an

embankment are not fully dissipated, it is necessary to install a
horizontal reinforcement at the embankment base, for additional
safety measures. This reinforcement (usually in the form of
geosynthetic or rebar-mesh) might also serve to limit lateral soil
displacements and therefore any displacement of the inclusions.
ASIRI provide a guideline to compute the lateral forces of the
embankment reinforcement (as illustrated in figure below). The
maximum tensile force Td;thrust;max mobilized in the geosynthetic by
the embankment lateral active forces is calculated below the
embankment crest:
Figure 6. Force Equilibrium of CMC (Abuhuroyroh,. Et al
2014) ; ; = 0.5 . . +2 . +2 . .
3.2 Load Transfer Platform (LTP) Where:

HR: embankment height above the geosynthetic,
Load Transfer Platform (LTP) is a well compacted clean granular fill Ka: coefficient of active earth pressure of the embankment:
placed over the CMC column to distribute the load from structural : unit weight of the embankment;
load to the CMC column. Another purpose is to provide stable g: permanent excess vertical load;
working platform for rig movement during CMC installation. Figure q: variable excess load;
7 describes typical cross section of CMC system. : partial factor on the soil unit weight;
: partial factor on the permanent actions;
: partial factor on the variable actions.
Figure 7. Typical Cross Section of CMC System (Abuhuroyroh,.

Et al 2014)
The thickness of LTP shall be designed to transfer the load Figure 9. Configuration examined at the edge of the
efficiently. The load transfer’s efficiency is increasing with the embankment (Asiri National Project, 2012)
thickness until optimum thickness. If the optimum thickness is
exceeded, the thickness will become additional load. The efficiency Beneath the embankment slopes, the tensile force Td;thrust
is also influenced by the spacing between CMC and shear strength mobilized in the geosynthetic by the embankment lateral active
characteristic of LTP material (elastic modulus and friction angle). pressure decreases with distance from the slope crest; this force may
Typical thickness of LTP is commonly ranged from 0.3m to 1.0m. be calculated at any point based on the embankment slope and
Figure 8 illustrates load distribution mechanism within LTP. thickness above this given point.
It is also important to note that this configuration may also
induce lateral displacements of the compressible soil capable of
creating bending moments in the inclusions that also require
verification.
584
3.4 CMC with Reinforcement 3. Check if reinforcement required:
In cases where the force is large enough compared to the resistance  If

N rd  N ed , reinforcement not required
capability of the CMC column to the lateral force, reinforcement
 If
N rd  N ed , reinforcement required
needs to be added to the CMC column. The lateral force is normally
found in is CMC near to the toe of the embankment slope. 4. If reinforcement is required, then the CMC’s reinforcement need
An analysis of the calculation of resistance capabilities of the to be designed using Med.
CMC column on lateral loads can be done with the following steps
referring to BS EN 1992-1-1.2004 12:
1. Calculate the axial design load and design moment that work with
4. PROJECT CASE STUDY: UP BOJONG
the following equation
N ed  N plaxis  spacing   G 4.1 Project Description
M ed  M plaxis  spacing   G Pemalang-Batang Toll Road with 39.2-kilometer highway connects
Pemalang with Batang Area, located in Central Java Province,
Where: Indonesia (Figure 13). The toll road is part of Trans-Java
Ned, Med : axial and moment acting on CMC Expressway (Figure 14 that connect Merak Port to Banyuwangi. The
NPLAXIS, MPLAXIS : axial and moment obtained from Plaxis Pemalang-Batang Toll Road has 13 bridge approach, embankment
Spacing : Spacing of CMC near bridge with height ±6.0m, where 7 bridge approach is treated
γG : safety factor; 1.2 - 1.4 with CMC. One of the bridge approach treated using CMC methods
are located at UP Bojong (STA 354+870) with estimated area 4,200
The axial load and bending moment are obtained from inner forces m2.
for every CMC columns as described at chapter 5.4 The Pemalang-Batang Toll Road is planned to be used during
Eid Al-Fitr, hence the construction of embankment need to be
2. Calculate resistance CMC (Nrd): accelerated using CMC method. The design criteria to be achieved
for ground improvement are listed below:
a. Eccentricity (e) with following equation = ,
 The maximum allowable residual settlement shall be 100
illustrated as below: mm after the 10 years of ground improvement
 The factor of safety against slope failure shall be equal or
more than 1.5 static condition.
 The loading component shall be calculated from fill
embankment soil from existing ground level to the final
platform plus the 15 kPa traffic loading.
The following figure shows site condition prior installation of

CMC:
Figure 10. Illustration of Eccentrical Load Distance
b. Calculate resistance area (Aref) with following equation:

Aref  R 2  2  sin 2 
 R
  arccos e Figure 11. Section of Bridge and Embankment UP Bojong
Where:
e : load eccentricity (m)
R : radius of CMC column (m)
c. Calculate resistance area (Aref) with following equation:

N rd  Aref  f cd
where:
Aref : area resistance (m2)
fcd : concrete strength, e.g. 20MPa
Figure 12. East Area of UP Bojong
585
Figure 13. Indicative Location of Pemalang-Batang
Figure 14. Layout of Trans-Java Toll Road
4.2 Soil Condition Table 1. Elevation Reference of UP Bojong
Additional soil investigation campaign consist of two boreholes and Elevation Elevation Fill thickness
In situ test
one CPT was conducted in 2017 by Menard to verify data consistency EGL (+m) FPL (+m) to FPL (m)
and provide more detail information for design purposes. Figure 15 BH-23 +7.2 +15.2 ± 8.0 m
shows the location of additional soil investigation. BH-24 +7.0 +15.0 ± 8.0 m
CPT-07 N/A N/A N/A
Table 2. Ground Water Level of UP Bojong
Elevation Elevation
Depth of
In situ test EGL (m) GWL
GWL (m)
(+m)
BH-23 +7.2 3.95 +3.25
BH-24 +7.0 5.6 +1.4
CPT-07 N/A N/A N/A
According to soil investigation results, a soft silt layer was
observed at 6.0m depth from existing level; while stiff silt layer was
Figure 15. Location of Additional Soil Investigation. found at ± 20.0m depth. The NSPT result, soil layering are shown
The following table briefly summarized the elevation reference below figure:
and ground water level for design purposes. For design, 8.0m fill
thickness was used, while ground water level was conservatively
assumed at 1.0m below existing level.
586
Table 3. Parameter Design for UP Bojong (1)
k  c’
No SOIL TYPE  ‘
(m/day) (kN/m³) (kPa)
1 LTP (Sirtu) 8.64 x 10-1 19 1 35

-4
2 SILT NSPT= 4 8.64 x 10 16 5 20
3 SILT NSPT = 9 8.64 x 10-4 17 5 20
-4
4 SILT NSPT = 7 8.64 x 10 17 5 20
-4
5 SILT NSPT = 11 8.64 x 10 17 7 21
-4
6 SILT NSPT = 15 8.64 x 10 17 7 21
7 SILT NSPT = 20 8.64 x 10-3 17 7 21
-3
8 SILT NSPT = 14 8.64 x 10 17 7 21
-2
9 Sand NSPT = 14 8.64 x 10 17 2 30
-3
10 SILT NSPT = 14 8.64 x 10 17 7 21
11 Backfill
Cu E50 Eoed Eur

No SOIL TYPE
(kPa) (kPa) (kPa) (kPa)
1 LTP (Sirtu) 35,000

2 SILT NSPT= 4 20 3,000 2,400 9,000
Figure 16. Combine Plot NSPT UP Bojong
3 SILT NSPT = 9 45 6,750 5,400 20,250
4 SILT NSPT = 7 35 5,250 4,200 15,750
5 SILT NSPT = 11 55 8,250 6,600 24,750
6 SILT NSPT = 15 75 11,250 9,000 33,750
7 SILT NSPT = 20 100 15,000 12,000 45,000
8 SILT NSPT = 14 70 10,500 8,400 31,500
9 Sand NSPT = 14 - 12,600 10,080 37,800
10 SILT NSPT = 14 70 10,500 8,400 31,500
11 Backfill
The concrete strength used for CMC were 20 Mpa. While for
material design used in PLAXIS calculate using following equation
and summarized in table below:
Eref(E50) = 2700 ∛ 28
Where:
Eref(E50) = modulus elasticity of CMC, kPa
F28 = grout srength after 28 days, MPa
Steps Analysis Unit

Figure 17. Soil Stratification of UP Bojong
Concrete quality 20 MPa
Modulus of CMC, Ey 7,329 MPa

5. GROUND IMPROVEMENT SOLUTION
Ey used for design 7,000 MPa
As described previously, CMC combined with additional
wiremesh, and CMC reinforced steel was decided to support the high
embankment. The material parameter used for analysis is
summarized below:
587
5.1 Ground Improvement Design Steps 5.2 Settlement Analysis-Without Ground Improvement
The elevation reference and loading calculation is shown in Figure Analysis Settle3D analysis was performed using information from
18. The calculation was started from settlement analysis without Figure 18 and parameter from Table 3 and Table 4. The settlement
ground improvement due to design load using Settle3D software. In of bridge approach due to embankment load and traffic load without
this condition, the settlement due to embankment and traffic load ground improvement ranged from 600 to 800 mm which exceed
exceed 100 mm. Ground improvement was required at bridge settlement criteria 100 mm. The analysis result is illustrated at below
approach area, hence first step of CMC analysis shall be performed figure:
using Finite Element Method-axisymmetric to predict CMC single
column behavior such as settlement and verification of compression
capacity of CMC.
Figure 18. Cross Section of Embankment
The axisymmetric analysis serving as guideline to prepare

further rigorous analysis using plane strain model. The finite element
software, PLAXIS was used to analyze on plane strain condition with
purposes to obtain settlement, factor of safety at slope, tensile forces
acting along wiremesh, and inner forces within CMC columns. All
the input from plane strain analysis let the designer to design
wiremesh and cmc with reinforced steel. All the calculation Figure 19. Settle3D Model
procedure is summarized in following table:
Table 6. Calculation Procedure of CMC for High Embankment
Steps Analysis Method Purposes

Settlement
Elastic Method-
Without
1. Software 1. Settlement
ground
Settle3d
improvement
Settlement FEM- 1. Settlement
2. With ground Axisymmetric – 2. Stress acting
improvement PLAXIS 2D in CMC
1. Settlement
FEM- Plane 2. Factor of
Strain – safety against
PLAXIS 2D failure. Figure 20. Settlement Curve- Without Ground Improvement
3. Slope stability
CMC – using 3. Tensile load
Embedded Pile on wiremesh 5.3 Axisymmetric Model
Row 4. Inner force in Where the settlement of bridge approach without ground
CMC improvement were exceeding 100 mm, the CMC was planned to
FEM- Plane increase soil stiffness, thus reducing the settlement below 100 mm.
Strain – Construction sequence for axisymmetric analysis are listed as below.
Wiremesh
Wiremesh PLAXIS 2D The construction period at FEM analysis was determined
4. configuration
design Wiremesh conservatively, the actual construction process require longer period
(Chapter 3.3)
equation than analysis.
checking 1. Initial phase
FEM- Plane 2. CMC installation – 7 days
Strain – Rebar 3. LTP Installation using sirtu with additional rebar – 7 days
CMC with
5. PLAXIS 2D configuration 4. Backfilling process to embankment final level – 30 days
rebar design
Rebar equation (Chapter 3.4) 5. Applying of operational load – 10 years of service period.
checking
The settlement obtained using axisymmetric model was 99 mm,
less than 100 mm (Figure 21). It should be noted that settlement
obtain using axisymmetric model are conservative due to all loading
on the top is fully distributed to the bottom of model. The
compression stress acting within CMC column was 5.2 MPa. ASIRI,
2012 limited the compression stress on CMC shall be column shall
be less than 7 MPa.
588
Figure 23. Typical Plane Strain Model
Figure 21. Settlement Result – Axisymmetric Model
Figure 24. Settlement Result – Plane Strain Model
Figure 22. Compression Stress on CMC Column
5.4 Plane Strain Model

The main purpose of plane strain model is simulating global behavior
of embankment supporting by CMC. Therefore, analysis output such
as settlement and compression stress acting within CMC column
shall be compared with axisymmetric model.
After several trial and error, the maximum settlement obtained
from plane strain analysis was 95 mm, which closed to axisymmetric
model, 99 mm. For compression stress checking, the maximum axial
forces acting on CMC column was 337.9 kN/m (Figure 25) that
generated pressure 5.367 MPa, which is close to axisymmetric model
output. The safety factor against slope failure was analyzed for short
term and long term; 1.543 and 1.703, more than minimum value of
1.5 (Figure 26 and Figure 27). Figure 25. Axial Forces CMC Column
589
5.6 Checking CMC Required Reinforced Steel
The bending moment and axial force inside the KGM is limited to
FoS – Short Term: 1.543 certain value to ensure limited strain envelope inside the CMC. The
procedure to check limit of axial and bending is using BS EN 1992-
FoS – Long 1-1:2004 §12. The axial forces and bending moment forces for each
Term: 1.703 CMC column were obtained from FEM model (Figure 25 and
Figure 29).
From the rebar checking, we conclude that rebar is required to
be placed inside the CMC for two outer most CMC columns.
Figure 26. Factor of Safety Against Slope Failure
Figure 29. Bending Moment Forces Acting on CMC Column

Figure 27. Location of Slope Failure
5.7 Configuration of CMC
5.5 Wire Mesh Design After conducted all the analysis, the optimum design configuration
Designing wire mesh was conducted in plane strain model, for CMC at UP Bojong are illustrated and summarized below figure:
where wire mesh was simulated using geogrid model. From the  Diameter: 42 cm
analysi, maximum tensile forces acting along wire mesh was 241  Length KGM 6 to 11 are 20m, KGM 1 to 4 varies from
kN/m (Figure 28). Therefore, the configuration of required wire 8.5m to 20m
mesh was determined using equation below:  2 layer of Steel wire mesh 8mm diameter with grid 15cm x
30cm
.  Compacted gravel sand (sirtu) 1m thick
= . .
4
Where:
fe = stress yield stress, kPa
n = number of steel bars per meter
dy = bar diameter in longitudinal section, m = 0.008m
spacing = 0.15m
Figure 30. CMC Design for UP Bojong
Figure 28. Tensile Forces Acting along Wire Mesh
590
6. QUALITY CONTROL
Several quality control program was conducted to ensure material

and installation quality consist of:
1. Prior installation of CMC

- Workability criteria: using flow table test, with
concrete after 3 hours to ensure pumability of the grout
during injenction.
- Strength criteria: using compressive test after 28 days.
2. During installation of CMC
- On board quality systems to monitor pumping pressure,
Figure 33. Settlement Vs Load for PLT Test at UP Bojong
drill pressure, installation depth, date of installation,
etc.
7. CONCLUSION
 Ground improvement using CMC was installed to increase

global stiffness of the soil, thus reducing the settlement to
acceptance limit.
 The application of CMC on toll road project is first time in
Indonesia.
 Quality control programs was perfomed in CMC project to
ensure the installation quality.
 The design concept and analysis procedure are briefly
described in this paper.
ACKNOWLEDGEMENT
The author would like to thank Mr. Abi Hakim and Ms. Febrini
Hartianty Adinda for helping to improve the document presentation
and collection of references, as well as their continuous support for
the development of this paper.
Reference:
ASIRI NATIONAL PROJECT, Multiple Authors. 2012.

Figure 31. Example of On Board Quality System Output Recommendations for the Design, Construction and Control of
Rigid Inclusions Ground Improvement.
3. After installation
- Vertical Plate Loading Test (PLT) is perfomed with Plomteux, C. & Liausu, P. 2007. Construction of Flat Glass Float
several purposes: Factory on Loess Deposits Using Controlled Modulus Columns
- Verification design assumption: soil condition, in Seismic Condition., Published in the Proceedings of 16th
parameter design Southeast Asian Geotechnical Conference, Kuala Lumpur,
- Verification construction procedure: drilling Malaysia.
pressure, penetration speed, etc
- Confirmation the achievement of design criteria Plomteux, C. & Lacazedieu, M. 2007. Embankment Construction
on Extremely Soft Soils Using Controlled Modulus Columns for
The typical setting up of PLT and testing result of Highway 2000 Project in Jamaica. Published in the Proceedings
Bojong are illustrated at following Figures 32 and of 16th Southeast Asian Geotechnical Conference, Kuala
Figures 33. The settlement obtained during 100% load Lumpur, Malaysia.
is 16 mm, less than 100 mm. This value indicated the
soil condition, parameter design, concrete quality, Smith, C.C, Tatari, A. 2016. Limit Analysis of Reinforced
construction procedure such as drilling pressure, Embankment of Soft Soil.
penetration speed is appropriate to be applied in this Abuhuroyroh, KM & Bechet, O. 2014. Application of Controlled
area. Modulus ColumnsTM (CMC) for Large Storage Tank in
Indonesia. Published in HATTI Conference 2014, Jakarta,
Indonesia.
Figure 32. Setting Up of PLT / Vertical Plate Loading

Test
591
Estimasi Modulus Resilien dari uji CBR pada Tanah Lempung

distabilisasi Abu Vulkanis dan Kapur
Devi Oktaviana Latif - Dr.,ST.,M.Eng
Departemen Teknik Sipil, Sekolah Vokasi, Universitas Gadjah Mada
Ahmad Rifa’i- Dr., Ir.,MT.,
Departemen Teknik Sipil dan Lingungan, Fakultas Teknik, Universitas Gadjah Mada
Latif Budi Suparma –Ir.,M.Sc., Ph.D
Departemen Teknik Sipil dan Lingungan, Fakultas Teknik, Universitas Gadjah Mada
ABSTRAK: Tanah merupakan bagian penting pada konstruksi jalan sehingga kekuatan dan stabilitas subgrade
sangatlah diperlukan untuk mendukung beban jalan raya tersebut. Pada umumnya penentuan kekuatan,
kekakuan dan stabilitas bahan subgrade pada konstruksi jalan dinyatakan dengan nilai CBR (California
Bearing Ratio) dan Modulus Resilient (Mr). Modulus resilien merupakan perbandingan antara tegangan siklis
dengan regangan. Saat ini pengujian modulus resilien sangat jarang digunakan untuk menentukan kekakuan
tanah dasar. Hal tersebut dikarenakan ketersediaan alat triaxial siklik sangat kurang, pengujian yang rumit dan
biaya yang besar, sehingga untuk meminimalisir kekurangan tersebut, maka diusulkan suatu prediksi nilai
modulus resilien (Mr) dengan nilai CBR. Prediksi nilai modulus resilen dan CBR bertujuan sebagai acuan
pada perencanaan nilai Mr pada tanah ekspansif yang distabilisasi dengan abu vulkanis dan kapur.
hasil pengujian triaksial siklis dan CBR pada penelitian ini diperoleh korelasi antara Mr dan CBR
dengan menggunakan regresi linier berganda.
Kata Kunci: Modulus resilien, stabilisasi tanah, abu vulkanis
ABSTRACT: Soil is an important part of road construction, so subgrade strength and stability are needed to
support the load of the highway. In general, the determination of the strength, stiffness and stability of subgrade
materials in road construction is stated by the value of CBR (California Bearing Ratio) and Resilient Modulus
(Mr). Resilient modulus is a comparison between cyclic stress and recovery strain. Currently testing resilient
modulus is very rarely used to determine subgrade stiffness. This is because the availability of cyclic triaxial
equipment is very lacking, complicated testing and large costs, so as to minimize these deficiency, a prediction
of resilient modulus (Mr) is proposed with the CBR value. The prediction of resilient modulus and CBR values
is intended as a reference for planning the value of Mr. on expansive soil stabilized with volcanic ash and lime.
The results of the cyclical triaxial test and CBR in this study obtained the correlation between Mr and CBR
using multiple linear regression.
Keywords: Resilien modulus, soil stabilization, volcanic ash.
1 PENDAHULUAN alat triaxial siklik sangat kurang, pengujian
yang rumit dan biaya yang besar, sehingga
Tanah merupakan bagian penting pada untuk meminimalisir kekurangan tersebut,
konstruksi jalan sehingga kekuatan dan maka diusulkan suatu prediksi nilai modulus
stabilitas subgrade sangatlah diperlukan untuk resilien (Mr) dengan nilai CBR.
mendukung beban jalan raya tersebut. Pada Prediksi nilai modulus resilien dengan CBR
umumnya penentuan kekuatan, kekakuan dan sebenarnya telah banyak dilakukan oleh
stabilitas bahan subgrade pada konstruksi jalan beberapa peneliti. Namun korelasi tersebut
dinyatakan dengan nilai CBR (California hanya dapat digunakan untuk jenis tanah
Bearing Ratio) dan Modulus Resilient (Mr). tertentu. Sedangkan untuk jenis tanah lempung
Modulus resilien merupakan perbandingan ekspansif yang distabilisasi dengan abu
antara tegangan siklis dengan regangan. Saat ini vulkanis dan kapur tidak dapat menggunakan
pengujian modulus resilien sangat jarang korelasi tersebut. Sebelum melalukan korelasi
digunakan untuk menentukan kekakuan tanah antara Mr dan CBR, dilakukan beberapa tahap
dasar. Hal tersebut dikarenakan ketersediaan penelitian yaitu tahapan penentuan kadar abu
592
vulkanis dari uji karakteristik tanah yang dengan,

menghasilkan campuran yang digunakan
adalah 20-30 % abu vulkanis dan 3-5% kapur z = Tambahan beban pada kedalaman
(Latif dkk,2016), pengamatan pengaruh tertentu (kPa)
penambahan abu vulkanis dan kapur terhadap Q = beban kendaraan (kN)
peningkatan nilai kekakuan tanah r = jarak dari pusat beban (m)
menggunakan triaxial sikilis (Latif dkk.,2017) z = kedalaman lapisan tanah
dan melakukan korelasi Mr dan CBR
2.2 Modulus Resilien
2 TINJAUAN PUSTAKA Desain dan evaluasi kinerja perkerasan
2.1 Tegangan pada Subgrade sangat dipengaruhi oleh sifat mekanis dan
komposisi lapisan tanah dasar. Sifat mekanis
Tegangan yang diterima oleh Tanah dasar
tanah tersebut adalah kekakuan, kekuatan, dan
atau Subgrade akibat beban roda kendaraan
deformasi permanen. Untuk memperoleh sifat
yang ada diatas permukaan perkerasan jalan.
mekanis tanah dasar (subgrade) membutuhkan
Gaya yang bekerja pada permukaan perkerasan
sejumlah besar data pendukung seperti
berupa gaya dinamik beban kendaraan yang
lewat secara berulang-ulang. Tegangan yang karakteristik pembebanan lalu lintas, sifat
material, kondisi lingkungan dan prosedur
terjadi pada subgrade akibat beban kendaraan
konstruksi. Saat ini, korelasi empiris
dan tegangan kkakng pada kedalaman tertentu
dikembangkan antara lapangan dan propertis
menggambarkan besarnya tegangan deviator
bahan laboratorium yang digunakan untuk
dan tegangan kekang yang dialami tanah pada
pengujian triaksial siklis. Distribusi tegangan memperoleh karakteristik kinerja subgrade.
korelasi ini tidak memenuhi persyaratan desain
akibat beban kendaraan yang terbesar terjadi
pada bagian atas dan berkurang dengan dan analisis karena mengabaikan semua
mekanisme kegagalan yang mungkin di
bertambahnya kedalaman seperti pada Gambar
lapangan. Metode tersebut menggunakan
1 Lapis perkerasan diatas tanah dasar harus
mampu mereduksi tegangan yang diterima oleh California Bearing Ratio (CBR) dan Nilai
tanah dasar, sehingga dapat mencegah Dukungan Tanah untuk memperoleh nilai
terjadinya penurunan atau deformasi yang kekakuan tanah tapi tidak mewakili kondisi
berlebihan. (Hardiyatmo, 2015) perkerasan yang mengalami beban lalu lintas
berulang. Association of State Highway and
Transportation Officials (AASHTO) 1986 dan
1993 merekomendasikan panduan desain
penggunaan modulus resilient (Mr) untuk
karakteristik base course dan subgrade untuk
merancang perkerasan lentur. Modulus resilien
(Mr) yang merupakan ukuran kekuatan tanah
dibawah tingkatan tegangan beban lalu lintas.
Nilai modulus resilien diperoleh dengan
melakukan pengujian triaksial siklik
dilaboratorium yang mensimulasikan beban
Gambar 1 Penyebaran tegangan akibat beban lalu lintas dengan tingkatan tegangan deviator
roda kendaraan. sehingga diperoleh regangan aksial dipulihkan
Analisa tegangan yang terjadi dalam massa dan regangan aksial permanen.
tanah akibat pengaruh beban kendaraaan Konsep modulus resilien sebuah material
dipermukaan dapat dilakukan dengan awalnya diperkenalkan oleh (Seed, et al., 1962)
menggunakan teori Boussinesq (1885). sebagai rasio tegangan deviatoric dinamis σd
Besarnya tegangan vertikal akibat beban roda yang diterapkan, untuk komponen tegangan
kendaraan tidak tergantung pada modulus resilien atau elastik (kembali) r. Modulus
elastis (E) dan poisson ratio () dan dinyatakan Resilien digunakan untuk menggambarkan
dalam persamaan : nilai tegangan tanah. Nilai tersebut tergantung
�
�� ⁄� dari modulus elastisitas tanah yang berbeda di
∆��= ��( �+(�⁄ )�) ……………. (1) bawah beban lalu lintas. Beberapa penelitian
�
telah banyak dilakukan untuk menyelidiki
pengaruh Modulus Resilien (Mr) suatu material
593
pada perancangan perkerasan. Hasil dari Mr 1. South African Council on Scientific and
memberi pengaruh yang sangat besar terhadap Industrial Research (CSIR)
perancangan ketebalan lapis fondasi dan M r ( psi)  3000 CBR 0.65 .......................(4)
lapisan aspal.
Modulus Resilien digambarkan sebagai fase 2. U.S. Army Waterway Experiment station
yang diperoleh dari kemiringan tegangan dan M r ( psi)  5409 CBR 0.71 ………….......(5)
regangan yang terjadi selama pembeban
berlangsung yaitu beban kendaraan melewati 3. Transport and road research laboratory
perkerasan. Uji modulus resilien mengukur M r ( psi)  2555 CBR 0.64 …..................(6)
kekakuan pada benda uji berbentuk silinder 4. Heukolom and Fuster (1960)
dengan menerapkan beban berulang. Untuk
tegangan deviatorik yang diberikan modulus
M r ( psi)  1565 CBR ............................(7)
resilien didefinisikan sebagai kemiringan kurva 5. Hopkins (1991)
tegangan deviatorik dan regangan aksial M r ( psi)  2596 CBR 0.874 ......................(8)
(Gambar 2)
Persamaan tersebut diatas digunakan untuk
jenis tanah berbutir halus sehingga persamaan
diatas tidak dapat digunakan untuk jenis tanah
yang distabilisasi dengan abu vulkanis dan
kapur. dengan demikian perlu dilakukan
pengembangan model persamaan yang dapat
digunakan untuk tanah lempung ekspansif yang
distabilisasi dengan abu vulkanis dan kapur.
3 METODE PENGUJIAN
Uji modulus resilien dilakukan sesuai
standar AASHTO T307 menggunakan alat
triaxial siklis pada Gambar 3. Prosedur
Gambar 2 Definisi modulus resilien pengujian untuk tanah dasar terdiri dari dua
��
tahap. Tahap pertama adalah tahap kondisi
��= .................................................(2) 1000 siklus. Tahap kedua terdiri dari 15 beban
��
dengan, kompresi siklik urutan dengan berbagai tekanan
�� : tegangan maksimum siklis (kN/m2),
kekang. Tekanan aksial maksimum
� : regangan elastis (%) menekankan selama fase ini berkisar dari
�
sekitar 14 kPa sampai 70 kPa, dan tekanan
kekang berkisar dari sekitar 14 kPa sampai 42
2.3 Model korelasi Modulus resilien dan CBR kPa.
Nilai modulus resilien diperoleh dengan
pengukuran dilaboratorium menggunakan alat
triaxial siklis. Namun, ketersediaan alat triaxial
siklis terbatas sehingga nilai modulus resilien
ditentukan dengan korelasi empirik baik
dengan nilai CBR maupun nilai-R. AASHTO
mengusulkan korelasi Mr dan CBR untuk untuk
tanah berbutir halus :
M r ( psi)  1500 CBR .................................(3) Gambar 3. Alat triaxial siklis
Korelasi ini efektif untuk nilai CBR
unsoaked < 20 % dan CBR Soaked <10% yang 4 HASIL DAN PEMBAHASAN
dapat digunakan untuk perancangan Korelasi nilai CBR dan modulus resilien
perkerasan. Selain itu, ada juga berbagai bertujuan untuk memprediksi nilai modulus
persamaan lain yang digunakan untuk resilien yang dapat digunakan sebagai acuan
memperkirakan nilai Modulus resilien perencanaan dilapangan untuk jenis tanah
berdasarkan hasil uji CBR yaitu : ekspansif distabilisasi menggunakan abu
vulkanis dan kapur. Untuk memperoleh nilai
594
tersebut maka dilakukan analisis regresi yang Analisa statistik diperlukan untuk
terdiri dari satu variabel bebas yaitu variabel mengembangkan pengembangan model
nilai CBR laboratorium. Setiap data pengujian prediksi yang dapat diandalkan sebagai
di plotkan kedalam grafik dan dianalisis dengan alternatif korelasi antara nilai modulus resilien,
cara regresi untuk mendapatkan nilai secara dan CBR dengan faktor pengaruh tegangan
umum pada model konstitutif yang disajikan utama mayor, tegangan minor atau tegangan
pada Gambar 4 dan Gambar 5. Berdasarkan kekang.
hasil dari prosedur regresi nonlinier dengan Analisa statistik diperlukan untuk
model fungsi berpangkat, analisis dilakukan mengembangkan pengembangan model
untuk menentukan model terbaik untuk prediksi yang dapat diandalkan sebagai
memprediksi nilai Mr dari regresi koefisien alternatif korelasi antara nilai modulus resilien,
berdasarkan Nilai CBR yang disajikan pada tegangan utama mayor, tegangan minor atau
Tabel 2. tegangan kekang dan CBR. Pada hasil analisa
hubungan antara modulus resilien dengan CBR,
d dan  diperoleh persamaan sebagai berikut
……………………………………………..(9)
Untuk mengetahui variabel b,d,f maka
dilakukan analisa regresi berganda dengan
bantuan excel. Variabel dependen dan
independen disajikan pada Tabel 2.
Tabel 1 Persamaan hubungan Mr dan CBR

No Keterangan Tingkat Persamaan R2
tegangan
Gambar 4. Hubungan antara modulus resilien 1 Merapi VA Tinggi Mr = 631.52 0.67
dan CBR pada tanah campuran abu vulkanis + 3% Kapur CBR0.365
Merapi dan kapur 3%. Sedang Mr = 436.38 0.66
CBR0.4346
Rendah Mr = 150.24 0.81
CBR0.78
2 Merapi VA Tinggi Mr = 143.79 0.96
+ 5% Kapur CBR 0.8716 0.96
Sedang Mr = 120.37 0.61
CBR0.8758
Rendah Mr = 412.01
CBR0.2577
Gambar 5. Hubungan antara modulus resilien

dan CBR pada tanah campuran abu vulkanis
Merapi dan kapur 3%
Dari hasil yang ditampilkan pada Tabel 1

menunjukkan korelasi modulus resilien dan
CBR berbeda-beda. Hal tersebut menunjukkan
Gambar 6 Pengaruh tegangan deviator dan
terdapat faktor pengaruh tegangan-tegangan
tegangan kekang terhadap modulus resislien
yang diterapkan pada benda uji terhadap nilai
pada campuran 20% abu vulkanis Merapi dan
modulus resilien. Pengaruh tegangan yang
5% kapur
diterapkan pada benda uji ditampilkan pada
Gambar 6 tersebut menunjukkan semakin besar
tegangan deviator (d), tegangan kekang (3)
dan tegangan utama yang diterapkan maka nilai
modulus resilien semakin besar
595
Tabel 2.Variabel Independen dan Dependen pada 5 KESIMPULAN

Analisis Regresi Linier Berganda
Korelasi antara modulus resilien dan nilai
Tipe Notasi Deskripsi Range CBR pada campuran tanah abu vulkanis dan
Variabel kapur dengan tujuan sebagai acuan pada
Independen Mr Modulus 590.94- perencanaan nilai Mr dilakukan dengan analisis
Resilien 3155.2 MPa regresi dimana tegangan utama mayor dan
Dependen CBR California 5-16 %
tegangan kekang merupakan variabel yang
Bearing Ratio
 Tegangan utama 28 -12 kPa mempengaruhi nilai modulus resilien
mayor menghasilkan formulasi :
 Tegangan 14 – 42 kPa
utama minor/
tegangan kekang
Hasil analisa regresi linier berganda DAFTAR PUSTAKA - REFERENCES

menghasilkan koefisien b,d dan f sehingga
persamaan 9 menjadi : ASTM. 2007. Annual Book of ASTM Standards,
section 4, Volume 04 09, Philadelphia, USA.
Bell, F.G. (1993). Engineering Treatment o Soil, 1 st
edition , E & FN Spon, London
(10) Ingles, O. & Metcalf, J. (1972). Soil stabilization:
principles and practice. Sidney, Melbourne,
Untuk mengetahui keandalan Persamaan maka Brisbane: Butterworths
dilakukan analisa dengan membandingkan nilai Lambe, T.W., and Whitman, R.V. (1969), Soil
modulus resilien hasil pengujian dengan nilai Mechanics, SI Version, John Wiley& Sons, New
York.
modulus resilien prediksi yang disajikan pada Latif, D.O., Rifa’i, A. & Suryolelono, K.B. 2016.
Gambar 7 Chemical Characteristic of volcanic ash in Indonesia
For Soil stabilization: Morphology and Chemical
Content. International Journal of Geomate, vol 11
issue 26, pp 2606. ISSN: 2186-2982(Print), 2186-
2990(Online), Japan, DOI :
10.21660/2016.26.151120http://geomatejournal.com
/sites/default/files/articles/2606-2610-151120-Devi-
Oct-2016-c1.pdf
Latif, D.O., Rifa’i, A. & Suryolelono, K.B. 2016. Effect
of Kelud Volcanic Ash Utilization on The Physical
Properties as Stabilizer Material for Soil
Stabilization. Electronic Journal of Geotechnical
Gambar 7. Perbandingan Mr prediksi dengan Engineering, vol 20.4 issue 26, pp 1679-1687.
Mr eksperimen abu vulkanis Merapi http://www.ejge.com/2016/Ppr2016.0146mb.pdf
Latif, D.O., Rifa’i, A. & Suryolelono, K.B. 2017. Impact
Pengujian dilaboratorium menghasilkan data of vocanic ash and lime adding on expansive soil for
modulus resilien yang digunakan untuk subgrade layer. International Review of Civil
menvalidasi korelasi antara parameter Engineering vol 8 no 5 pp 255-260.
Latif, D.O., Rifa’i, A. & Suparma, L.B. 2017. Influence
dependen dan independen dengan analisis of Volcanic ash and Lime on Resilien Modulus For
statistik menggunakan regresi linier berganda Subgrade. Warmadewa University International
Analisis statistik yang dilakukan pada hasil uji Conference of Architectural and Civil Engineering pp
abu vulkanis Merapi, Gambar 7 menghasilkan 165-169.
korelasi yang baik antara parameter model dan Seed, H., Chan, C. & Lee, C., 1962. Resilience
characteristics of subgrade soil and their relation to
sifat tanah. Perbandingan nilai modulus resilien fatique failure in Asphalt Pavements. Michigan,
yang diprediksi dan eksprimen menunjukkan International conference on the structural design of
bahwa korelasi yang diajukan oleh penelitian Asphalt Pavements, University of Michigan.
ini adalah akurasi yang dapat diterima.
596
Seasonal Variation of Water Content and Pore-water Pressure Distribution in

Vegetated Soil Slope
N. Gofar1, H. Rahardjo1, A. Satyanaga1
1
School of Civil and Environmental Engineering, Nanyang Technological University, 50 Nanyang Avenue, Singapore. 639798.
E-mail: nurlygofar@ntu.edu.sg
ABSTRACT: This paper presents soil water content and pore-water pressure distribution in a vegetated residual soil slope in response to a
one-year seasonal variation in Singapore. The measurements were taken by tensiometers and soil moisture sensors TM1-SM1, TM2-SM2,
TM3-SM3 and TM4-SM4 installed at vertical distances of 2.0, 2.0, 1.37 and 0.23 m from slope surface. Seasonal variation was represented by
rainfall and actual evaporation calculated based on data collected by a weather station installed at the site. The field monitoring shows that the
soil water content and pore-water pressure measurements by TM1-SM1, TM2-SM2, TM3-SM3 were representative of soil-water characteristic
curve (SWCC) of the residual soil. On the other hand, the measurements by TM4-SM4 shows a lower pore-water pressure during dry period
and a lower volumetric water content during wet period as compared to SWCC of residual soil. The study showed that the response recorded
by TM4-SM4 was representative of top soil used as media for vegetation planting. The study also showed that bigger variation of pore-water
pressure recorded by TM4 in December 2016 and January 2017 was due to long dry period with high temperature which is not normal in
Singapore during these months.
Keywords: Field monitoring, soil water content, pore-water pressure, soil-water characteristic curve, vegetated soil slope.
1. INTRODUCTION recommended the use of updated Penman-Monteith Method or FAO-

56 Method (Zotarelli et al, 2009) to calculate the rate at which readily
Soils in tropical regions commonly consist of residual soils with available soil water is vaporized from a specified vegetated surface
negative pore-water pressure in the zone above ground water table ETo. The effect of different species of vegetation can be considered
(Rahardjo et al. 2005). The pore-water condition is influenced by flux by modifying the equation with a crop coefficient, Kc (Lazzara and
boundary conditions such as rainfall, evaporation, and transpiration Rana, 2010), thus (ETc = ETo × Kc).
processes. Earlier studies consider only rainfall as the flux boundary The transient volumetric water content and pore-water pressure in
condition to avoid more complexity in their analysis of pore-water soil are measured by soil moisture sensor and tensiometer
pressure and slope stability (e.g. Chipp et al., 1982; Rahardjo et al., respectively. The state of moisture in the unsaturated zone of soil
1998; Ng et al., 2003; 2008 and Li et al., 2005; Lee et al., 2008). between the ground surface and the water table (Blight, 1997) is
However, not all rainfall becomes infiltration. From the definitions, controlled by the water balance between the water input into water
the rainfall may be separated into four components, i.e. runoff, output from the soil. The change in water stored in soil lead to the
infiltration, interception (rainfall that is caught on the vegetation change in both volumetric water content and pore-water pressure.
surfaces), and evapotranspiration (ET) (Joel et al., 2002). Thus, some Long term real-time monitoring is required to capture the effect of
researchers took further steps to consider the portion of precipitation different weather conditions throughout the year.
that infiltrates into the soil (Ng et al. 2003; Rahardjo et al. 2005; Gofar The water flow into the soil is governed by the coefficient of
et al. 2008). Other researchers also consider the effect of evaporation permeability with respect to water (kw) (Fredlund and Rahardjo,
on the prediction of pore-water pressure variation in soil slope (e.g. 1993). Unlike saturated soils, the permeability of an unsaturated soil
Gasmo et al., 2000; Gitirana et al, 2005 2006; Yunusa et al, 2014). is a non-linear function of the volumetric water content of the soil.
Numerical analysis performed by Yunusa et al. (2014) on one-year When the soil approaches saturation, the permeability becomes
data showed a better agreement with field response of unsaturated soil constant and equal to the saturated coefficient of permeability, ks.
when the input data was the combination of rainfall and evaporation Thus, the response of soil to the flux boundary conditions is
rates. Rahardjo et al (2017) included the effect of evapotranspiration controlled by the soil water characteristic curve (SWCC) which is
in their study of pore-water pressure variation in vegetated soil slope. non-linear with respect to the negative pore-water pressure (suction).
These studies suggested rainfall and evaporation as well as In other words, the ability of the unsaturated soil to retain water varies
transpiration play important roles in affecting the pore-water pressure with soil suction. The SWCC follows different paths during drying
distribution within residual soil slope. and wetting (hysteresis) in nature. The soil on the drying path has a
The response of the soil to flux boundary conditions can be higher water content than the soil on the wetting path at a given matric
evaluated by field monitoring of the flux boundaries as well as the suction. Thus, the field data of volumetric water content and pore-
transient volumetric water content and pore-water pressure in the soil. water pressure could be plotted within the hysteretic of the drying and
The intensity of rainfall that falls on the slope is normally monitored wetting curves of the SWCC.
using a tipping-bucket rain gauge. In contrast to rainfall, This paper presents response of residual soil slope to flux-
evapotranspiration rate from vegetated surface cannot be easily boundary conditions based on field monitoring data collected from an
measured. Potential evaporation rate (PE) can be measured in the instrumented slope in Singapore. The rainfall data collected by rain-
field using Lysimeter or predicted based on climate data (Penman, gauge and evapotranspiration (ETc) were used as flux boundary
1948) equation. The PE is an evaporation from an open water body conditions while the responses were identified by volumetric water
whereby the relative humidity is equal to unity. The actual content and pore-water pressure recorded by soil moisture sensor and
evaporation (AE) from a soil surface is related to the relative humidity tensiometers. The responses were compared to the soil-water
in the soil at ground surface is about 70% of the PE (Sattler and characteristic curve (SWCC) of the corresponding soil surrounding
Fredlund, 1991). the soil moisture sensors and tensiometers tip.
In addition to the relative humidity of soil surface, transpiration
should be considered for the effect of flux boundary conditions on
2. THEOTERICAL BACKGROUND
vegetated surface. Several empirical methods are available to obtain
evapotranspiration (ET) rate. Weather station can be installed to The evaporation from ground surface covered with certain species of
capture the meteorological variables required to calculate the vegetation (ETc) can be calculated by considering standard
evapotranspiration such as air temperature, solar radiation, relative meteorological variables and a crop coefficient (Kc) (Lazzara and
humidity and wind speed (Rahardjo et al., 2014). Allen (1998) Rana 2010), thus ETc is calculated as follows:
597
��= � � × �� (1) response of tensiometers to pore-water pressure change must be

where ETo is the reference evapotranspiration in mm/day which can checked regularly to ensure the quality and physical performance of
be calculated as follows: high air-entry ceramic tips. In this study, regular maintenance of the
900
0.408  (��−�)+ (�+273 ) �2 (��−��) tensiometers was conducted twice a week by refilling the jet-fill
��
�= (2) reservoir with de-aired water and flushing the tensiometers to remove
+ (1+0.34 �2 )
the accumulated or trapped air in the tubes caused by cavitation of
where T = mean air temperature (oC); u2 = wind speed (m/s) at 2 m water and air diffusion through the ceramic tip. Soil moisture sensor
above the ground; Rn = net radiation flux (MJ/m2/day; G = sensible used in this study was of TDR type which was capable of measuring
heat flux into the soil (MJ/m2/day); es = saturation vapour pressure soil moisture up to saturation (0–100%) with accuracy of 1% and
(kPa); ea = actual vapor pressure (kPa); ∆ = slope of saturation vapour response time of 0.5 second. All soil moisture sensors were tested in
pressure curve; and γ = psychrometric constant. The detailed water and air environment and their values corresponded to the values
calculation of this method can be found in Zotarelli et al. (2009). specified by manufacturer which was about 70 to 90% in pure water
Typical Kc values are given in Lazzara and Rana (2010), for grass, and zero in air.
Kc= 1.
The water flow through an isotropic unsaturated soil is formulated
using Darcy’s law as follows (Fredlund and Rahardjo, 1993):
� �ℎ � �ℎ �ℎ �
(� �) + (� �) = �2� � g (3)
��
with
��
ℎ� = �+ (4)
� g
where hw = hydraulic head; z = elevation head; uw = pore-water
pressure (kPa); x and y are the Cartesian coordinates in the x- and y-
directions, respectively; kw = permeability function (m/s); w= density
of water (Mg/m3); g = gravitational acceleration (m/s2); m2w=
coefficient of water volume change with respect to a change in matric
suction (ua-uw); ua = pore-air pressure (kPa), and t = time (second).
Equation 4 shows that the unbalanced flow of water through a soil Figure 1 Plan view residual soil slopes with instrumentation
element is equal to the change in water volume in the soil element. locations
Unlike saturated soils, the permeability of an unsaturated soil is not
constant (Fredlund and Rahardjo, 1993). The coefficient of The readings of all measuring instruments were calibrated and
permeability with respect to water for a soil is a non-linear function checked before they were connected to a data acquisition system
of the volumetric water content of the soil. When the soil approaches (DAS) to obtain instrumentation readings in real time. The data
saturation, the permeability becomes constant and equal to the logger was powered by solar panel and battery. The cables were
saturated coefficient of permeability with respect to water, ks. In protected by corrugated tubing to prevent damages from rainfall and
addition, the volumetric water content of the soil is dependent on the insects. The data logger recorded readings at a 10-minute interval
negative pore-water pressure in a non-linear fashion (soil-water regardless of rainfall events. The data logger sends all data through
characteristic curve or SWCC). In other words, the ability of the general packet radio service (GPRS) to a web page for remote
unsaturated soil to retain water varies with soil suction. SWCC monitoring. Figure 2 shows the weather station and data acquisition
follows different paths during drying and wetting (hysteresis) in system used at the study site.
nature where the soil on the drying path has a higher water content
than the soil on the wetting path at a given matric suction.
2. METHODOLOGY
2.1 Field Instrumentation
The investigated slope in this study is located at the central part of
Singapore within residual soil from Bukit Timah Granite. Figure 1
shows diagram of residual soil slope and the field instrumentation
carried out in this study. The instrumentations can be divided into two
parts; the first part was weather instrumentation for climatic data
measurements while the second part was instrumentation for
quantifying soil response to the flux boundary conditions. The
monitoring period was one year (1st July 2016 to 30th June 2017).
The weather instrumentation includes tipping bucket rain gauge Figure 2 Data acquisition system used at the study site
for rainfall, pyranometer for solar radiation, wind monitor for wind
speed and direction, as well as temperature and relative humidity 2.2 Soil Properties
probes. The weather station for climatic data measurement was
The slope was formed by residual soil with a thin layer of top soil as
located near the toe. A 10 m high galvanized mast was erected on a
media for growing vegetative cover. The soil properties required for
concrete footing. The wind monitor, the solar panel, the lightning
this study were investigated by performing laboratory tests on
protection rod and the pyranometer were installed on top of the mast.
samples taken from the field. The grain size distribution shows that
The other instruments were installed at a height of 1 m from the
about 50% of the material is clay. The unified soil classification
ground surface together with the data logger. Soil temperature (ST)
system USCS (ASTM D2487-00) categorized the soil to be highly
was measured at depths of 0.10m, 0.15m, 0.20m, and 0.25m.
plastic clay (CH). The bulk density of the residual soil was 1.8 Mg/m3.
The tensiometers (TM) and soil moisture sensors (SM) were
The saturated permeability of the soil was determined using the
installed within residual soil layer at depth of 2 m from crest and at
flexible-wall saturated permeability test (ASTM D5084-10). The
distances 0.4 m; 2.4 m; 3.8 m and 4.6 m from slope face. The jet-fill
saturated permeability of the soil is 6×10-7m/s. In order to perform
tensiometers were calibrated properly prior to installation and then
analyses related to water flow in unsaturated soil, permeability
inserted into a tube to the specified depth. The reading from
functions of the soil need to be obtained. Permeability function can
transducer was verified by a bourdon gauge attached to it. The
598
be determined from the soil-water characteristic curve (SWCC) using

a statistical method proposed by Childs and Collis-George (1950).
The procedure for the prediction can be found in Fredlund and
Rahardjo (1993). The SWCC of the soil in the site is obtained using
Tempe cell and pressure plate according to ASTM D6838-02. The
wetting and drying SWCCs of the residual soil are shown in Figure
3. The permeability functions obtained using the prediction method
based on wetting and drying SWCCs are shown in Figure 4.
Figure 5 Daily rainfall from July 2016 to June 2017.
Figure 3 Drying and wetting SWCC of residual soil
Figure 4 Drying and wetting permeability curve of residual soil
3. RESULTS AND DISCUSSION Figure 6 Mean monthly and number of rainfall days
3.1 Rainfall
3.2 Climatic data
Figure 5 shows daily rainfall recorded from 1st July 2016 to 30th June
2017. The cumulative yearly rainfall was 2819 mm which is higher The climatic data observed from the weather station from July
than the average annual rainfall in Singapore based on long term 2016 to June 2017 including air temperature (Ta), relative humidity
record from NEA Singapore (1981–2010) i.e. 2166 mm. The number (RH), solar radiation (SR) as well as wind speed (WS) are shown in
of rainfall days during the monitoring period was 178 days, which Figures 7. The climatic data obtained from the study site were also
was higher than the average annual number of rainfall days in compared with the typical climatic data reported by NEA Singapore
Singapore (167 days). Rainfall monitoring at the study site indicates based on data from 1981 to 2010.
that the monthly rainfalls are quite different from the typical trend in The minimum and maximum air temperature during the
Singapore as shown in Figure 6. monitoring period was 22.9 and 36.9oC respectively. The range of air
For example, the monthly rainfall in December 2016 (214.3 mm) temperature was wider and higher than the mean range of air
was too low as compared to the mean monthly rainfall in Singapore temperature in Singapore i.e. 23–33oC. As for the rainfall, the
for December 318 mm). On the other hand, the monthly rainfall in variation of air temperature during the monitoring period was
May 2017 (319 mm) was very high as compared to the average different from the typical values recorded in Singapore. Both
rainfall in Singapore for May (171 mm). The monthly rainfall in maximum and minimum temperature measured in this study occurred
September, October and November 2016 were higher than the mean in January 2017. This is in accordance with the rainfall condition in
monthly rainfall for the months in Singapore. November 2016 was which January 2017 represent an extreme condition. Past data
the wettest month during the monitoring period with 21 rainfall days indicated that the maximum temperature usually occur in March to
accumulating to 311 mm rain and while August 2016 was the driest May. In this study, the trend of higher temperature in March to May
month with cumulative monthly rainfall of 98.6 mm. Figure 5 shows was overcome with heavy downpour (Figure 6).
that the maximum daily rainfall occurred on 23rd January 2017 i.e. Soil temperature was measured at depths of 0.10, 0.15, 0.20 and
103.8 mm. The month of January 2017 represented extreme condition 0.25 m from ground surface. Field measurement performed in this
because it started with a two-week dry period followed by very wet study indicated that the soil temperature varies with depth and the
period towards the end of the month. time of the day. The least variation with the time of the day was
obtained in the measurement of soil temperature at depth of 0.20 m
from ground surface. Figure 7a shows the variation of soil
temperature measured at depth 0.20 m as compared to maximum and
minimum daily temperature. The soil temperature was lower in
November, December and January, and higher in May, June and July.
Relative humidity plays an important role in the prediction of ET.
The minimum and maximum relative humidity recorded in the study
site were 67.4% to 99% with mean value of 80%. The recorded RH
599
was higher than the typical range recorded in Singapore i.e. 60 to 3.3 Pore-water pressure measurements
95%. Maximum relative humidity occurred in November due to high
number of days with rainfall. The lowest relative humidity was The response of soil slope to rainfall and evapotranspiration from
recorded in August due to the least number of rainfall days. Figure 7b vegetated surface in terms of pore-water pressure is presented in
shows the maximum and minimum daily RH during the monitoring Figure 9. The figure shows that the measurements at TM1 – TM3
period. were quite consistent, while measurements at TM4 varied
The minimum and maximum solar radiation during the significantly especially from November 2016 to January 2017. Based
monitoring period was 1.12 and 14.76 MJ/m2/day respectively with on the evaluation of rainfall and climatic data, this period represents
mean value of 7.93 MJ/m2/day. These values are the typical of extreme changes from dry to wet condition.
Singapore. The maximum solar radiation was recorded in the dry As shown in Figure 1, TM1 and TM2 have vertical distances of 2
month of August 2016. The variation of solar radiation at the study m from crest while TM3 was 1.37 m below slope face. TM4 was very
site during the monitoring period is shown in Figure 7c. close to slope face i.e. 0.4 m horizontal distance or only 0.23 m
The maximum wind speed (4.92 m/s) was recorded in February vertical distance. Thus, TM4 is more affected by flux boundary
2017. The recorded wind speed was lower than the range of wind conditions as compared to the other tensiometers.
speed in Singapore i.e. 0–13.375 m/s with mean value of 2.65 m/s.
This may be because the location of the study site was surrounded by
high rise buildings. The variation of wind speed throughout the year
was also presented in Figure 7c.
*Note: Refer to Figure 1 for instrumentation locations
Figure 9 Pore-water pressure recorded at the study site from July

2016 to June 2017
3.4 Soil’s response to flux boundary conditions
Field data of pore-water pressure and volumetric water content

throughout the monitoring period was plotted together with SWCC of
residual soil in Figure 10. It can be seen that measurements by TM1-
SM1, TM2-SM2 and TM3-SM3 were plotted well within the
hysteretic of the SWCC. The range of suction was quite narrow i.e. 4
– 27 kPa. The volumetric water content ranged from 27% to the
saturated volumetric water content of the soil (51%). This shows that
the measurements at TM1 to TM3 are representative of the residual
Figure 7 Variation of climate data from July 2016 to June 2017 soil forming the slope. On the other hand, the measurement by TM4
below the SWCC wetting curve of the residual soil. It was deduced
3.2 Evapotranspiration that the soil response at TM4-SM4 is more representative of top soil
instead of the residual soil.
Figure 8 shows the ETc calculated using FAO-Penman or FAO-56
Method for ETo and coefficient of 1 for Kc, plotted together with
rainfall data. The minimum, mean and maximum daily potential
evaporations in the study area were 0.66, 1.92, and 3.26 mm
respectively. These values are slightly less than reported by Rahardjo
et al. (2017) based on their study in different part of Singapore. The
minimum and maximum ETc calculated in their study was 0.55 and
4.3 mm/day. The highest daily ETc occurred in February and June
2017 while the lowest was in June. However, based on the monthly
ETc, the variation of ETc follows weather variation on site. The
highest monthly ETc occurred in June (65.29mm) while the lowest
was in November (47.77mm).
Figure 10 Plot of pore-water pressure and volumetric water content

measured at the study site from July 2016 to June 2017 in SWCC of
residual soil.
Soil samples were collected from the slope surface to a depth of 20

cm. SWCC and saturated permeability tests were carried out on the
samples using the same procedure as for the residual soil. It was found
that the saturated volumetric water content of the top soil was lower
than that of the residual soil. The top soil has lower density as
compared to the residual soil, thus the higher void ratio shifted the
Figure 8 Evapotranspiration on slope surface at the study site from
SWCC to the left, resulting in a lower air-entry value. The saturated
July 2016 to June 2017
600
coefficient of permeability of the top soil was 1 × 10-5 m/s which is Allen R.G., Pereira L.S., Raes D., Smith M. (1998) Crop
higher than that of the residual soil (6 × 10-7 m/s). evapotranspiration: guidelines for computing crop water
The field data from TM4-SM4 were plotted on the SWCC curves requirements Food and Agriculture Organization of the United
of the top soil as shown in Figure 11. The figure shows that the field Nation, Rome.
data could be plotted well inside the hysteresis of the SWCC of the Blight G.E. (1997) The “Active” Zone in Unsaturated Soil
top soil. The range of suction was i.e. 8 – 90 kPa, while the range of Mechanics. 1st GRC Lecture, Nanyang Technological
volumetric water content was from 13 to 30%. The maximum University, Singapore.
volumetric water content was less than the saturated volumetric water Childs E.C., Collis-George G.N. (1950) “The permeability of porous
content of the surface soil. It may be attributed to some water materials”. Proc. Royal Society of London 210A. pp. 392-405.
intercepted by vegetation (grass) at the surface. The same reason that Chipp, P.N., Henkel, D.J., Clare, D.G. & Pope, R.G. (1982) Field
the evapotranspiration (ET) is less than the actual evaporation (AE) Measurement of Suction in Colluvium Covered Slopes in
from base soil surface. Hong Kong. Proc. 7th Southeast Asian Geotechnical Conf.,
Hong Kong, pp.49-62.
Fredlund D.G., & Xing A. (1994) “Equations for the Soil-Water
Characteristic Curve”. Canadian Geotech. Journal 31 pp. 521-
532.
Fredlund, D. G. & Rahardjo, H. (1993). Soil Mechanics for Un-
saturated Soils. New York: John Wiley & Sons, Inc.
Joel, A., Messing, I., Seguel, O., & Casanova, M. (2002) “Meas-
urement of surface water runoff from plots of two different
sizes”. Hydrological Processes. 16(7) pp. 1467-1478.
Lazzara P. & Rana G. (2010) “The use of crop coefficient approach
to estimate actual evapotranspiration: a critical review for
major crops under Mediterranean climate”. Italian Journal of
Agrometeorology 2 pp. 25-39.
Figure 11 Plot of pore-water pressure and volumetric water content Lee, M.L. Gofar N. & Rahardjo H. (2009) “A Simple Model for
measured by TM4 from July 2016 to June 2017 in SWCC of top Preliminary Evaluation of Rainfall-Induced Slope Instability.
soil. Engineering Geology”. 108(3-4) pp. 272-282.
Li, A.G., Yue, Tham, L.G. and Lee C.F., (2005) “Field-monitored
variations of soil moisture and matric suction in a saprolite
4. CONCLUSIONS slope”. Canadian Geotechnical Journal. 42, pp. 13–26.
Field measurements of flux boundary conditions and soil response Gasmo, J.M., Rahardjo, H. and Leong, E.C. (2000) “Infiltration
were carried out in a residual soil slope at the central part of Effects on Stability of a Residual Soil Slope”. Computer and
Singapore. The measurements presented in Figures 5 and 7 specified Geotechnics, 26, April, pp.145–165.
that the seasonal variation of rainfall was quite different from the Gitirana Jr., G., Fredlund, D.G., Fredlund, M. (2006) “Numerical
typical trend in Singapore (Figure 6). The dissimilar trend also modeling of soil-atmosphere interaction for unsaturated
observed for the other flux boundary conditions such as air surfaces”. Proc. 4th Intl. Conf. Unsaturated soils, US., pp.658-
temperature and relative humidity. Highly variable conditions 669.
occurred during the months of December 2016 and January 2017. Gitirana Jr., G., Fredlund, D.G., Fredlund, M. (2005) “Infiltration-
This could be identified by unusually high air temperature and long runoff boundary conditions in seepage analysis”. Proc. 58th
period of dry condition which are different from the normal Canadian Geotech. Conf. Canada, pp.516-523.
conditions in Singapore. The long dry period allowed high pore-water Gofar N. Lee, M.L. & Kassim, A. (2008) “Response of Suction
pressure to be recorded by TM4 which is located closest to the ground Distribution to Rainfall Infiltration in Soil Slope”. Electronic
surface (Figure 9). J. of Geotechnical Engineering. 13 (E), pp.1-13
Readings of pore-water pressure and soil water content recorded Ng, C.W.W., Springman, S.M. and Alonso E.E. (2008). “Monitoring
by TM1-SM1, TM2-SM2, TM3-SM3 are representative of the SWCC the Performance of Unsaturated Soil Slopes”. Geotechnical
of the residual soil. Thus, the readings are representative of the soil and Geological Engineering. 26(6), pp.799-816.
where the instrumentations were installed. However, the plot for Ng. C.W.W., Zhan, L.T., Bao, C.G., Fredlund, D.G. and Gong, B.W.,
TM4-SM4 located at vertical distance of 23 cm from ground surface (2003) “Performance of an Unsaturated Expansive Soil Slope
was more representative of top soil which has lower density and Subjected to Artificial Rainfall Infiltration”. Géotechnique.
higher coefficient of permeability. This showed that the readings of 53(2), pp. 143-157.
instrumentation closer to the ground surface was affected by the flux Penman, H.L (1948) “Natural evapotranspiration from open water,
boundary conditions and the consistency of surface soil as well as bare soil and grass”. Proc. Royal Society London Ser. A.193,
vegetative cover. pp.120-145.
Rahardjo, H., Amalia, N., Leong, E.C. Harnas, F.R. Lee, T.T. and
4. REFERENCES Fong, Y.K. (2017) “Flux Boundary Measurements for the
study of Tree Stability”. Landscape Ecol Eng. 13:81–92
Rahardjo, H., Satyanaga, A. Harnas, F.R. and Leong, E.C. (2014)
ASTM D5084-10 Standard Test Method for Measurement of “Comprehensive Instrumentation for Real Time Monitoring of
Hydraulic Conductivity of Saturated Porous Material using a Flux Boundary Conditions in Slope” Proc. 3rd Italian
Flexible Wall Permeameter, ASTM International, West Workshop on Landslides, Italy.
Conshohoken. PA. Rahardjo, H., Satyanaga, A. and Leong, E.C. (2013). “Effects of Flux
ASTM D6838-02 Standard Test Method for Determination of SWCC Boundary Conditions on Pore-water Pressure Distribution in
for Desorption using Hanging Column, Pressure Extractor, Slope”. Engineering Geology, 165, October, pp. 133–142.
Chilled Mirror Hygrometer or Centrifuge, ASTM Rahardjo, H., Lee, T.T. Leong, E.C. and Rezaur, R.B. (2005).
International, West Conshohoken. PA. “Response of a Residual Soil Slope to Rainfall”. Canadian
Geotechnical Journal, 42(2), pp. 340–351.
ASTM D 2487-00 Standard Practice for Classification of Soils for Rahardjo, H., Leong, E.C., Gasmo, J.M. and Tang, S.K. (1998)
Engineering Purposes (Unified Soil Classification System) , “Assessment of rainfall effects on stability of residual soil
ASTM International, West Conshohoken. PA.
601
slopes”. Proc. 2nd Intl. Conf. Unsaturated Soils, China, pp. 280-
285.
Sattler, P.J. & Fredlund, D.G. (1991). “Modelling vertical ground
movements using surface climatic flux”. Proc. Geotechnical
Engineering Congress, US., pp. 1292-1306.
Yunusa, G.H. Kassim, A. & Gofar, N. (2014) “Effect of Surface Flux
Boundary Conditions on Transient Suction Distribution in
Homogeneous Slope”. Indian J. Science & Technology, 7(12),
2064–2075
Zotarelli L, Dukes MD, Romero CC, Migliaccio KW, Morgan KT
(2009) Step by step calculation of Penman-Monteith
evapotranspiration (FAO-56 Method), University of Florida.
602
Efek EPS Geofoam Sebagai Material Pengisi Terhadap

Nilai CBR Laboratorium pada Tanah Kohesif Daerah Bandung
A.N. Lestari1 and C. Julian2
1
Dosen Universitas Katolik Parahyangan, Bandung, Indonesia
2
Mahasiswa Universitas Katolik Parahyangan, Bandung, Indonesia
E-mail: sri@unpar.ac.id
ABSTRACT: Expanded polystyrene (EPS) yang disebut Geofoam merupakan material pengisi yang sangat ringan dengan berat kurang dari
satu persen dari berat tanah. Pada daerah yang mengalami kesulitan untuk tanah timbunan dengan volume cukup besar, maka material
pengisi dapat digunakan sebagai alternatif tanah timbunan. Penggunaan Geofoam sebagai material pengisi yang ringan perlu dikaji
kesetaraanya sebagai material pengganti tanah timbunan, dalam hal ini akan diteliti efek Geofoam sebagai material pengisi terhadap nilai
CBR Laboratorium pada tanah kohesif di Bandung. Pada Penelitian ini dilakukan uji California Bearing Ratio (CBR) dengan rendaman
(soaked) dan tanpa rendaman (unsoaked) pada lokasi di Rancaengkek, Gedebage dan Setiabudi sebagai sampel uji. EPS 29 digunakan
sebagai material pengisi dalam penelitian ini dengan spesifikasi kepadatan 28.8 kg/m3. Dari hasil Uji CBR dengan standard Proctor nilai
CBR tanah asli kondisi unsoak menggunakan material pengisi EPS 29 mengindikasikan peningkatan 50 – 70% dan peningkatan 15 – 150%
pada kondisi soak, hal ini terjadi pada tanah kohesif dengan plastisitas tinggi. Tanah dengan plastisitas sedang dan berpasir tidak memberikan
peningkatan nilai CBR, tapi dapat disimpulkan bahan EPS Geofoam dapat digunakan sebagai bahan pengisi .
Keywords: EPS Geofoam, CBR , material pengisi.
1 PENDAHULUAN Tabel 1 merupakan spesifikasi dari beberapa tipe Geofoam.

Ditinjau dari segi teknis, Geofoam dapat mengurangi penurunan,
Pada daerah yang mengalami kesulitan untuk tanah timbunan meskipun ditambah beban bangunan lagi diatas nya tidak akan
dengan volume cukup besar dan transportasi yang mahal, maka mengalami penurunan yang cukup banyak, disebabkan pengurangan
material pengganti dapat digunakan sebagai alternative timbunan. beban tanah asli sebelum nya yang digantikan oleh Geofoam.
Perhitungan waktu untuk melakukan timbunan juga merupakan Geofoam juga memilki keuntungan lain yaitu resistensi terhadap
salah satu alasan untuk mencari material lain sebagai pengganti tekan. Material EPS Geofoam ± 98% adalah udara. Dikarenakan
tanah urug. material EPS Geofoam mengikat udara layaknya ban kendaraan
Inovasi dalam bidang teknologi terutama yang terkait dengan bermotor, sehingga EPS Geofoam memiliki daya dukung tersendiri.
bidang konstruksi terus berkembang sangat pesat. Kecepatan dan Disain EPS Geofoam dianjurkan tidak melebihi 1% kapasitas
ketepatan menjadi fokus utama dalam inovasi, karena akan sangat resistensi tekan (ASTM Test Method D2126).
terkait dengan kualitas dan biaya dalam setiap pekerjaan. Expanded
polystyrene (EPS) merupakan suatu inovasi yang bisa diterapkan di
lapangan sebagai material pengisi, karena beratnya sangat ringan.
EPS Geofoam diterapkan pada proyek timbunan untuk jalan raya
(gambar 1) atau timbunan dibelakang abutment jembatan untuk
mencapai tinggi yang ditentukan.
EPS geofoam merupakan bahan material dengan harga tinggi.
walaupun Geofoam cukup mahal tapi dari segi waktu bisa
dipertimbangkan untuk daerah yang mengalami kesulitan dengan
proyek tanah timbunan dengan volume besar dan mobilisasi
pengambilan tanah cukup jauh sehingga membutuhkan waktu lama. Gambar 2 Contoh EPS 29 Geofoam
Contoh proyek di Lambeusoe, Aceh Indonesia (Cahyana Strofoam),
pengiriman EPS Geofoam lebih terjangkau dan proses pemasangan Batas ini mengontrol jumlah defleksi jangka panjang yang
jauh lebih mudah dibandingkan memperoleh material tanah dihasilkan dari beban terus menerus. EPS Geofoam tidak mudah
timbunan. membusuk yang digunakan untuk material tahan api. Penggunaan
Geofoam sebagai material pengisi memiliki kepadatan kurang dari
tanah dan batuan, sehingga gravitasi yang dihasilkan dan kekuatan
gempa jauh lebih berkurang daripada material pengisi lainnya.
Tabel 1 Spesifikasi EPS Geofoam (ASTM D6817-07)

Type EPS12 EPS15 EPS19 EPS22 EPS29
Density, min. 11.2 14.4 (0.90) 18.4 (1.15) 21.6 28.8
Kg/m3(lb/ft3) (0.70) (1.35) (1.80)
Gambar 1 Aplikasi Geofoam pada timbunan Compressive 15 (2.2) 25 (3.6) 40 (5.8) 50 (7.3) 75
Resistance, min. (10.9)
kPa (psi) at 1%
2 EXPANDED POLYSTYRENE ( EPS) GEOFOAM Compressive 35 (5.1) 55 (80) 90 (13.1) 115 170
Resistance, min. (16.7) (24.7)
EPS (Expanded Polystyrene) sejenis bahan Styrofoam yang tidak kPa (psi) at 5%
asing dalam kehidupan sehari-hari, sebagai bahan pengepakan atau Compressive 40 (5.8) 70 (10.2) 110 (16.0) 135 200
dekorasi. Expanded Polystyrene atau Geofoam sebagai material Resistance, min. (19.6) (29.0)
produk rekayasa, dapat diproduksi dengan berbagai kepadatan kPa (psi) at 10%
sesuai resistensi tekanan yang diperlukan (gambar 2). Flexural 69 172 (25.0) 207 (30.0) 276 345
Inovasi material terus berkembang dalam bidang Geoteknik. Strength, min. (10.0) (40.0) (50.0)
kPa (psi)
Sebagai bahan material pengisi yang ringan hanya sekitar satu
Oxygen index, 24.0 24.0 24.0 24.0 24.0
persen dari berat tanah atau batu, dapat mengurangi tekanan pada min. Volume %
dasar subgrade.
603
Di lapangan geofoam tidak langsung menerima beban, karena Berdasarkan dari grafik plastisitas Casagrande dan ukuran butir
hanya merupakan material pengisi, umumnya diatas geofoam (gambar 4), seluruh sampel termasuk katagori tanah lanau dengan
diberikan lapisan perkerasan berupa beton atau aspal dan plastisitas tinggi.
dibawahnya merupakan lapisan pasir dan geotextile.
3 LOKASI SAMPEL
Ketiga sampel diambil dari daerah Bandung yaitu dari Setiabudi,
Gedebage dan Rancaengkek (Gambar 3). Ketiga sampel merupakan
tanah kohesif. Klasifikasi berdasarkan ukuran butir sampel tanah
dapat dilihat pada tabel 3.
Gambar 4 Klasifikasi tanah berdasarkan grafik Plastisitas
5.3 Karakteristik Mekanis

Gambar 3 Lokasi pengambilan di area Bandung, daerah Setiabudi,
Rancaengkek dan Gedebage Perilaku pemadatan dengan metode Standard Proctor sebagai uji
kompaksi untuk menentukan berat isi kering dan kadar air optimum
4 METODOLIGI PENELITIAN sebagai acuan uji CBR, sampel pada mold dilakukan dengan tiga
lapis sesuai uji Standard proctor. Bahan Geofoam dicoba diletakkan
Pengujian yang dilakukan adalah pengujian sifat fisis, pengujian pada lapis yang berbeda diantara tiga lapis tanah pada Standard
berdasarkan ukuran butir, sifat plastis dan mekanis. Pengujian Proctor, diukur nilai CBR nya sesuai variasi posisi Geofoam.
perilaku fisik meliputi kadar air, berat isi, berat jenis dan keplastisan
ketiga sampel adalah batas cair dan batas plastis dapat dilihat pada
tabel 2. Pengujian berdasarkan ukuran butir ketiga sampel
merupakan tanah lanau dapat dilihat disribusi ukuran butir pada
tabel 3 dan gambar 4. Perilaku mekanis meliputi pengujian
kompaksi dan pengujian CBR kondisi soak dan unsoak pada tanah
asli dan dengan material pengisi EPS Geofoam. Bahan Geofoam
dicoba diletakan pada beberapa posisi sebagai lapis pengganti tanah,
pada mold uji kemudian dilakukan uji CBR.
5 HASIL UJI DAN PEMBAHASAN

5.1 Sifat fisik dan plastisitas tanah
Hasil pengujian sifat fisik dan kadar air dapat dilihat pada tabel 2 Gambar 5 Kurva kompaksi sampel Setiabudi
sesuai sampel masing masing
Tabel 2 Sifat fisik dan keplastisan tanah

Sample Water Specific Liquid Plastic IP
content, Grafity, limit Limit
ω (%) Gs
Setiabudi 54 2.696 47.20 31.11 16.09
Rancaekek 32.85 2.546 54.20 36.57 17.73

Gedebage 80.18 2.740 76.20 56.23 19.97
5.2 Kalsifikasi berdasarkan ukuran butir

Gambar 6 Kurva kompaksi sampel Rancaengkek
Hasil pengujian berdasarkan uji saringan dan uji hydrometer
diperoleh klasifikasi berdasarkan ukuran butir pada tabel 3.
Tabel 3 Klasifikasi sampel berdasarkan ukuran butir

Sampel Clay Silt Fine Coarse Gravel
(%) (%) Sand to (%)
(%) Medium
Sand(%)
Setiabudi 24.58 64.93 6.84 3.65 0
Rancaekek 28.00 63.70 6.36 1.95 0
Gedebage 35.59 56.92 3.44 4.05 0
Gambar 7 Kurva kompaksi sampel Gedebage
604
Hasil kompaksi dari tiga sampel dengan nilai berat isi kering Berdasarkan persamaan 2 untuk energi pada kondisi tanah dan
maksimum dan kadar air optimum pada ketiga sampel tanah kondisi tanah dan geofoam sebagai material pengisi, maka:
tersebut dapat dilihat pada gambar 5, 6, 7.  tumbukan hammer perlapis pada tanah saja 10 tumbukan
menjadi 15 tumbukan tanah dan geofoam;
Tabel 4 Hasil Kompaksi Tiga Sampel Dengan Standard Proctor  tumbukan hammer perlapis pada tanah saja 25 tumbukan
menjadi 38 tumbukan tanah dan geofoam;
Sampel tanah ɣd maximum (gr/cm3) Woptimum (%)
 tumbukan hammer perlapis pada tanah saja 56 tumbukan
Setiabudi 1.295 36.4 menjadi 84 tumbukan tanah dan geofoam.
Rancaekek 1.360 29.8
Gedebage 1.275 34.8 Pada kondisi A dengan meletakkan geofoam pada lapis tengah,
nilai CBR meningkat namun tetap jauh lebih rendah dibandingkan
dengan nilai CBR tanah asli. Oleh karena itu dilakukan revisi
5.4 CBR (California Bearing Ratio) peletakan EPS Geofoam diletakkan pada lapis bawah kondisi B.
Dengan acuan kadar air optimum dari hasil kompaksi tanah Geofoam dapat merendam energy tumbukan yang diberikan pada
dilakukan uji CBR untuk masing masing sampel tanpa geofoam saat proses pemadatan dan hasil lebih tinggi dari nilai CBR tanah
dengan kondisi unsoak dan soak. tanpa geofoam. Pada kondisi B lokasi geofoam berada di lapis
bawah, karena geofoam tidak bisa langsung menerima beban
sebuah ketentuan Schmertmann yang sering dijadikan pegangan
adalah bahwa distribusi beban pada tanah sampai kedalaman 2 x
diameter bidang tekan yang mengalami tegangan. Jika luas bidang
tekan (piston CBR) 3 in2, maka diameter piston sebesar 1.732 in ≈
4.4 cm. Sehingga sampel tanah dalam mold CBR yang mengalami
tegangan sampai pada kedalaman 8.8 cm, posisi lapis 2 terdalam
pada 8.6 cm pada mold. Perletakan EPS Geofoam pada lapisan ke-3
hanya bersinggungan 2 mm dari zona tegangan.
Gambar 8 Uji CBR tanah metode Standard Proctor 5.4.1 Uji CBR dengan geofoam kondisi B
5.4.1.a Tanah Setiabudi
Pada uji tanah tanpa geofoam, metode Standard Proctor adalah
sebagai berikut :
- Tumbukan dengan 10 kali tumbukan /lapis,
- Tumbukan dengan 25 kali tumbukan/lapis,
- Tumbukan dengan 56 kali tumbukan/lapis.
Energi Standard atau Modified Proctor tergantung pada berat
hammer, tinggi jatuh Hammer, jumlah lapis pemadatan dan jumlah
tumbukan tiap lapis.
N  n W  H
Energy  (1)
V
Dimana :
N = jumlah tumbukan hammer perlapis dalam mold;
(a) (b)
n = jumlah lapis dalam mold;
W = berat hammer;
Gambar 10 (a) Nilai CBR tanah tanpa geofoam (Setiabudi), (b)
H = tinggi jatuh hammer;
Nilai CBR tanah + Geofoam (Setiabudi)
V = volume mold = volume sample.
Nilai CBR sampel tanah Setiabudi tanpa geofoam kondisi unsoak
Karena Geofoam sebagai material pengisi, sehingga geofoam
pada 95% ɣ dry maximum dari hasil uji kompaksi diperoleh 15.8%
tidak terkena beban langsung, jadi geofoam diletakkan coba coba
dan nilai CBR soak diperoleh 7.6%. Nilai CBR tanah Setiabudi +
pada lapis kedua dan ketiga.
Geofoam kondisi unsoak pada 95% ɣ dry maximum diperoleh 14.2
% dan nilai CBR kondisi soak 7.3 %.
Nta  n  W  H Ntag  n  W  H
 (2)
V V 5.4.1.b Tanah Rancaengkek
Dimana :
Nta = jumlah tumbukan hammer pada tanah asli;
Ntag= jumlah tumbukan hammer pada tanah dan geofoam.
Kondisi A Kondisi B (a) (b)

A: Geofoam pada lapis 2 B: Geofoam pada lapis 3 Gambar 11 (a) Nilai CBR Tanah tanpa geofoam (Rancaengkek),
Gambar 9 Posisi Geofoam di Mold CBR (b) Nilai CBR Tanah + Geofoam (Rancaengkek)
605
Nilai CBR Tanah Rancaengkek unsoak dengan kondisi 95% ɣ dry Hidrostatis uplift :
maximum dari uji kompaksi sebesar 7.9% dan kondisi CBR soak Karena EPS geofoam mempunyai berat sangat ringan dan
adalah 1.1 %. Nilai CBR tanah Rancaengkek + Geofoam unsoak kepadatan rendah. Timbunan tanah harus dihitung untuk gaya
sebesar 11.8 % dan nilai CBR kondisi soak adalah 2.75 %. hidrostatis uplift.
5.4.1.c Tanah Gedebage

Nilai CBR Tanah Gedebage unsoak dengan kondisi 95% ɣ dry
maximum dari uji kompaksi sebesar 7 % dan kondisi CBR soak
adalah 1.9 %. Nilai CBR tanah Gedebage + Geofoam unsoak
sebesar 12.4 % dan nilai CBR kondisi soak adalah 2.2 %.
Gambar 13. Aplikasi geofoam di lapangan (Cahyana Styrofoam

PT. Cahyana Ekspansindo)
7 KESIMPULAN
 EPS Geofoam dengan berat dan kepadatan kecil tidak bisa
langsung menerima beban,utama, hanya sebagai material
(a) (b) pengisi.
Gambar 11 (a) Nilai CBR tanah (Gedebage), (b) Nilai CBR tanah +  EPS Geofoam akan effektif jika diletakkan pada kedalaman
Geofoam (Gedebage) kurang lebih dua kali zona tegangan akibat beban utama.
 Dari hasil nilai CBR unsoak dan soak, EPS Geofoam dapat
Tabel 5 Nilai CBR Tanah Tanpa Geofoam dan Tanah + Geofoam digunakan sebagai alternatif material pengisi pada tanah
kohesif, hal ini sesuai nilai CBR dengan EPS Geofoam
Sampel Nilai CBR mendekati atau lebih besar nilai CBR tanah tanpa geofoam.
Tanah Unsoaked Soaked  Pada kondisi muka air tinggi, harus diperhitungkan tinggi
Tanah Tanah + Tanah Tanah + tanah diatas geofoam agar dapat mengimbangi adanya gaya
Geofoam Geofoam uplift akibat tekanan hidrostatik.
Setiabudi 15.8 14.2 7.6 7.3
Rancaengkek 7.9 11.8 1.1 2.75
Gedebage 7 12.4 1.9 2.2
Dari Table 5, terlihat Tanah Setiabudi nilai CBR tanah + 8 DAFTAR PUSTAKA
Geofoam kondisi unsoak dan soak agak mengecil, jika diperhatikan
Tanah Setiabudi mempunyai persentase pasir lebih banyak diantara American Society for Testing Materials, “ Standard Test Method for
dua sampel yang lain. CBR ( California Bearing Ratio) of Laboratory-compacted
Soils”, ASTM designation: D1883-07e2. Philadelphia, Pa.,
6 APLIKASI GEOFOAM MATERIAL PENGISI 309-310, 2007.
TIMBUNAN DI LAPANGAN Beton Elemindo Perkasa ( BEP ), “Calculation B-Foam Geofoam-
Grade 29 as fill material Embankment“ , Bandung Indonesia,
2015.
Reynaud, David, ”Guidelines for Geofoam Applications in
Embankment Projects”, NCHRP Project 24-11(02), January
2013.
Stark, Timothy, “Design Procedure for Geofoam Applications in
Embankment Projects”, Illinois , August 2003.
Stark, Timothy, “Guidelines and Recommended Standard for
Gambar 12 Aplikasi Geofoam sebagai timbunan Geofoam Applications in Highway Embankments”, NCHRP
(BEP, Bandung-Indonesia) Report 529, July 2004.
606
Prediction of Excess Pore Pressure Due to Pile Driving Based on CPTu

A. Arafianto1 and P.P. Rahardjo2
1
Engineering Faculty, Universitas Katolik Parahyangan, Bandung, Indonesia
2
Engineering Faculty, Universitas Katolik Parahyangan, Bandung, Indonesia
E-mail: aflizalarafi@gmail.com
ABSTRACT: Piezocone test (CPTu) is one of the most versatile tools for geotechnical investigation. It is commonly used for soil type
identification, soil stratification, and geotechnical parameters determination. In soft soils, the CPTu test is very important because of its
capability of pore pressure measurement. This paper presents the result of CPTu test in the soft soil after pile driving works. The dissipation
test results are evaluated to distinguish between the excess pore pressure induced by cone penetration and by pile driving works. The results
show that the generated excess pore pressure induced by pile driving is up to 1.27 to 1.40 times the effective overburden pressure. In
addition, the dissipation of the excess pore water pressure is observed by conducting another CPTu test five months later after the first test.
The result shows that the excess pore pressure due to pile driving still exists, with the remaining value of 0.25 times the effective overburden
pressure.
KEYWORDS: CPTu, pile driving, soft soil, excess pore water pressure
1. INTRODUCTION 4. SOIL CONDITION

Piezocone test (CPTu) is one of the most versatile tools for From the CPTu test result, the soil stratification consists of the
geotechnical investigation. It is commonly used for soil type backfill layer with approximately 2.5 m thick in the upper layer.
identification, soil stratification, and geotechnical parameters Underneath this layer, silty clay layer with soft consistency found
determination. The measured data is generally more reliable because with a thickness of 17.5 m. At the bottom layer, a sand layer was
it measured in the in-situ stress condition. Furthermore, the found and then the test terminated at the depth of 21 m below the
piezocone test gives continuous data and it is fast and economical. existing ground level. The CPTu (PC-01) result is shown in Figure
In soft soils, the CPTu test is very important because it is 1.
capable to measure the pore water pressure. Various geotechnical It has to be highlighted that the pore pressure measurement
parameters and information can be obtained from the pore water shows high values, compared to hydrostatic pressure. The difference
pressure measurement, including the value of generated excess pore between the two values is up to 0.3 MPa. This is because the excess
water pressure due to pile driving. Evaluation and interpretation of pore pressure induced by pile driving still exists. Figure 2 shows the
the dissipation tests are to be conducted in order to distinguish testing location, which is very close to the hydraulic pile driver.
between excess pore water pressure induced by cone penetration and
by pile driving. In addition, the dissipation of the excess pore water
pressure is also observed by conducting another CPTu test, five
months later after the first test. In this particular case study, the
piezocone (CPTu) test is chosen because the cone penetration has
the same mechanism as the pile injection process in the field.
2. PROJECT INFORMATION AND PILING WORKS

The case study of the paper is an office and apartment building
project located in Surabaya, Indonesia. The foundation type of the
building is concrete square piles of 50×50 cm2, with an embedment
depth varies from 20 – 25 m. As for the piling works, the piles are
driven by the hydraulic pile driving method. The total number of
piles driven is 755 piles.
3. PIEZOCONE AND DISSIPATION TESTS

A number of CPTu and dissipation tests have been conducted to Figure 2 CPTu Testing Near Hydraulic Pile Driver
identify the soil condition at the project site. The first series of the
test was conducted in October 2012 (one day after pile driving 5. GENERATED EXCESS PORE PRESSURE DUE TO
works finished), while the second test was conducted in March PILE DRIVING
2013. Both tests were conducted after the pile driving works. As for The generated excess pore pressure due to pile driving has been an
the total number of tests, there are both five CPTu and dissipation interest in research for a very long time. A number of excess pore
tests, with details of test presented in Table 1. pressure measurements due to pile driving have been made by
numerous researcher. Results of the measurement have revealed that
Table 1 Series of CPTu and Dissipation Tests Conducted at Site the excess pore pressure may become equal to or even greater than
CPTu No Testing Date* Dissipation Test
the effective overburden stress. A summary of some measurements
Depth (m) of the variation with radial distance of excess pore pressures around
PC-01 8.93 a single driven pile is shown in Figure 3.
18 October 2012 Around the pile surface, very high excess pore pressures are
PC-02 14.57
CPTu-01 8.97 generated, in some cases approaching 1.5 to 2.0 times the effective
CPTu-02 18 March 2013 14.5 overburden pressure. However, the generated excess pore pressure
CPTu-03 6.84 decreases rapidly with radial distance from the pile and generally
*conducted after pile driving works dissipate rapidly.
607
Figure 1 CPTu (PC-01) Test Result (October 2012)
6. PREDICTION OF EXCESS PORE PRESSURE USING

DISSIPATION TEST
The dissipation test can be used for estimating the excess pore water
pressure induced by pile driving. The interpretation method is
suggested by Rahardjo et al (2008), which can be done by
extrapolating the dissipation test curve. This method assumes that
when the dissipation test carried out ultimately, the curve follows
the hyperbolic function.
When the pile driving works conducted in saturated clays,
excess pore pressure will be generated due to the pile is driven into
the soil mass. Therefore, when conducting a dissipation test
ultimately in such condition, the dissipation curve will not reach the
value of the hydrostatic pressure but the value of final pore pressure
(uf). The difference between uf and uo called the residual excess pore
pressure. Figure 4 shows the scheme of the method.
6.1 Dissipation Test Results

Five dissipation tests have been conducted at a certain depth, as
tabulated in Table 1. These dissipation curves are then extrapolated
using the suggested method above to obtain the final excess pore
pressure (uf). Figure 5 shows the dissipation test curves.
Figure 3 Summary of Measured Pore Pressure (Poulos and Davis, Figure 4 Extrapolation of Dissipation Test Curve (Rahardjo et al.,
1980) 2008)
608
Figure 5 Dissipation Test Results
Table 2 Estimation of Excess Pore Pressure Ratio After Pile Driving

Calculated Effective Normalized
Dissipation Hydrostatic
CPTu Residual Excess Overburden Excess Pore
Test Depth Pressure, uo
No Pore Pressure, uf Pressure, σv’ Pressure,
(m) (kPa)
(kPa) (kPa)* (uf-uo)/σv’
PC-01 8.93 64.3 172 85.01 1.27
PC-02 14.57 130.7 293 115.5 1.40
CPTu-01 8.97 79.7 97 71.8 0.24
CPTu-02 14.5 115 207 128.5 0.71
CPTu-03 6.84 58.4 73 56.9 0.25
*assuming the soil unit weight, γ = 16 kN/m3, and γsat= 17 kN/m3
Figure 6 Comparison of Pore Pressure Measurement and Normalized Excess Pore Pressure (Recalculated from Rahardjo et al, 2014)
The value of excess pore pressure due to pile driving then can be From the results, it can be concluded that the excess pore
calculated by subtracting with the value of hydrostatic pressure. pressure ratio in October 2012 (CPTu PC-01 and PC-02) is as high
Furthermore, the ratio between existing pore pressure and effective as 1.40 σ’v, while in March 2013 (CPTu-01 – CPTu-03) is much
overburden pressure is also calculated. Table 2 summarizes the less, with the value of 0.25 σ’v. These obtained values agree well
calculated value of excess pore pressure ratio after pile driving. with the previous publication from Poulos and Davis (1980), where
the generated excess pore pressure due to pile driving can be as high
as 2.0 σ’v.
609
6.2 Observation of Excess Pore Pressure Dissipation

As mentioned before in the previous section, another CPTu test is
conducted five months later after the first test. The purpose of the
second test is to observe the dissipation of the excess pore pressure.
In addition, the normalized excess pore pressure distribution along
depth is also plotted.
The normalized excess pore pressure distribution (Δu2/σv’) can
be calculated by subtracting the measured pore pressure (u2) with
the hydrostatic value (uo). The purpose of the calculation is to
investigate how high the total excess pore pressure (due to cone
penetration and pile driving) generated along the depth.
It has been observed that when the pile is long compared to its Figure 7 Determining the Degree of Dissipation
diameter, the excess pore pressure generated due to pile driving are
essentially constant with depth (Reese and Seed, 1955 and 7. DISCUSSION
Soderberg, 1962). By multiplying the proportion with the The generated excess pore pressure due to pile driving can be
normalized excess pore pressure, the excess pore pressure due to estimated by dissipation test of the CPTu test. In order to increase
pile driving (Δumt) can be calculated by the following equation: the confidence and the reliability of this method, it is interesting to
compare the obtained results both with piezometer measurements
‫ ݂ݑ‬− ‫݋ݑ‬ ߂‫ݑ‬2 (1)
߂‫ = ݐ ݉ݑ‬൬ ൰× ൬ ൰ and numerical simulations. Furthermore, the dissipation degree
‫ݑ‬2 − ‫݋ݑ‬ ߪ‫'ݒ‬ value towards the time is also interesting to be investigated.
where: 8. CONCLUSIONS
uf = residual excess pore pressure (obtained from
CPTu test has proven its ability to measure the pore pressure in the
extrapolation dissipation test)
soil strata, as well as the generated excess pore pressure due to pile
uo = hydrostatic pressure
Δu2/σ’v = normalized excess pore pressure driving by conducting the dissipation test. The dissipation test
during CPTu can be conducted easily and generally fast. In addition,
the dissipation test is so flexible that can be conducted at several
Figure 6 shows the pore pressure measurement of CPTu test in
depths during a test.
October 2012 and in March 2013, along with the calculated
normalized excess pore pressure. Furthermore, the extrapolation of the dissipation curve can be
done easily to calculate the proportion of the excess pore pressure
From the results, it can be seen that one day after the pile driving
due to cone penetration and due to pile driving.
works finished (October 2012), the value of the excess pore pressure
due to pile driving is as high as 2.0 times the effective overburden
pressure. While from the CPTu test conducted five months later 9. REFERENCES
(March 2013), it can be seen that the excess pore pressure due to
pile driving is not fully dissipated. The remaining excess pore Geotechnical Engineering Consultant. 2012. Laporan Hasil Uji
pressure due to pile driving is relatively low, with the value of 0.20 Geoteknik Lapangan: Uji Piezocone/CPTu Proyek Tunjungan
– 0.25 σv’. Plaza 5 Surabaya (In-situ Testing Result Report: Piezocone
Test/CPTu of Project Tunjungan Plaza 5 Surabaya).
6.3 Degree of Dissipation Bandung: GEC (In Indonesian)
Lo, K.Y., and Stermac, A.G. 1965. “Induced Pore Pressures During
The degree of excess pore pressure dissipation (U) can be estimated Pile-Driving Operations”. Proceeding of the International
using the dissipation test results, following the equation as follows: Conference of Soil Mechanics and Foundation Engineering.
߂‫ݐݑ‬ (2) Poulos, H. G., and Davis, E. H. (1980). Pile Foundation Analysis
ܷ =1− and Design. John Wiley & Sons, Inc.
߂‫݅ݑ‬
Rahardjo, P.P., Anggoro, B.W., Yakin, A.Y., and Darmawan, H.
2008. “Determination of the Degree of Consolidation of
where: Δut = excess pore pressure due to pile driving at time t Reclaimed Site on Deep Soft Mahakam Deltaic Soils using
Δui = excess pore pressure due to pile driving at the initial CPTu”. Proceeding of the Fourth International Symposium on
time after pile driving finished Deformation Characteristics of Geomaterials, Atlanta, GA,
883-889.
In this particular study, the CPTu test is not conducted directly Rahardjo, P.P., Wirawan, A., and Setiawan, R. 2014. “An Analytical
after the pile driving works finished, but conducted one day after. and Empirical Study on Ground Movement and Excess Pore
Therefore, the calculated degree of dissipation will be based on this Pressure Generation Due to Pile Driving in Soft Soils”.
available data, which will produce an overestimated value. Figure 7 Proceeding of the International Conference on the Soft Soils
shows the calculation steps to determine the degree of dissipation, and Ground Improvement, Bandung, Indonesia, A8-1 – A8-
obtaining 80.6% dissipation degree in October 2013. 13.
610
Increasing geotechnical challenges in the design and construction planning of the

Third Phase Bangkok MRT Underground
N. Phienwej1, A. Asanprakit2, P. Kittiyodom2 and S.Timpong2

1
School of Engineering and Technology, Asian Institution of Technology, Pathumthani, Thailand
2
Geotechnical and Foundation Engineering Co. Ltd., Bangkok, Thailand
E-mail: noppadol@ait.ac.th
ABSTRACT: The 3rd phase MRT underground project of Bangkok that is being implemented has faced increased difficulties in design and
construction owing to the congested corridor along the alignment and the phenomenon of groundwater rebound in Bangkok subsoils in the
aftermath of the end of land subsidence era from deep well pumping. The rebound is more than 10 meters from the situation during the
initial phase construction and thus it necessitates careful soil investigation and design approach to prevent instability of groundwater ingress
and uplift in bore tunnelling and excavation of station boxes. The tight space along the MRT underground alignment resulted in the twin
bored tunnels being placed at very small clearance in a few locations. It creates high risk in the shield driving, particularly in the saturated
sand layer. The works also involve tunnel driving at close proximity or under a number of existing buildings or road viaducts. In such cases,
various schemes of underpinning are employed, and one of them is strengthening of founding soil layers by cement jet grouting. An
intensive instrumentation program according to the observational design method is employed to safe guard against occurrence of any
undesirable instability in the excavations and impacts to the third party properties.
Keywords: MRT Underground, shield tunneling, deep excavation, underpinning, groundwater rebound, the observational method
1. INTRODUCTION 2. BANGKOK MRT SYSTEM

The 1st phase MRT underground project, i.e. Blue Line (Initial) was The MRT system of Bangkok, when fully implemented, is shown in
constructed in 1999-2003. It consisted of 18 stations and 22 km of Figure 1. As it has been planned, the underground portion
twin bored tunnels running along inner city roads. The tunnel was constitutes about 60 km in combined length, of which about one half
successfully bored with 4 EPB shields and the station box was have been completed. Besides the underground works of the
smoothly constructed using concrete diaphragm wall with the top Orange Line East that are being constructed, the Orange Line West
down excavation method. For the 2nd phase project, Blue Line and Purple Line South projects, which are to start the
Extension, the works began in 2011 and completed 2017. The 5.4- implementation by the end of the year, have the remaining
km underground section consisting of 4 stations and 4.8-km-long underground portion of the Bangkok MRT system. Works of the
twin bored tunnels were placed along narrow roads in the old city three underground MRT projects that are being or to be soon carried
quarter and crossed the main river of the city. In this phase, the out are challenging because of the very tight work spaces and the
shield tunnelling and station excavation experienced incidents of risen groundwater head in the subsoils.
severe difficulties associated with the impact of groundwater
rebound in the subsoils of Bangkok. The rebound was the
consequence of banning deep well pumping to solve land
subsidence problem in Bangkok area. At a section of shield
tunnelling, problematic groundwater ingress occurred through
segmented lining just installed while driving through a mixed face
of clay and sand layers; and in one of the station box excavation a
hydraulic uplift of the excavation base occurred that resulted in
flooding and significant complication in finishing the excavation
and casting the concrete base slab.
Currently, the 3rd phase underground project, Orange Line East, is
under construction. The underground section consists of 12.1 km of
twin bored tunnels and 10 stations placed along two of the most
traffic congested roads of the city where there exist elevated road
and expressway viaducts along the alignment. The piled foundation
of the viaduct complicates the alignment design of the MRT
underground tunnel and positioning the station boxes that result in
necessity in close proximity placement of the twin bored tunnels to
each other and to neighbouring structures at numerous locations. In
addition, the soil investigation reveals that the subsoil stratigraphy
consists of sand layer of larger thickness and higher rebounded
Figure 1 Bangkok MRT System
groundwater levels than that in the earlier phase in other zones of
the city. The design approaches of the underground works of the
3. SUBSOILS AND GROUNDWATER
project are outlined below. In addition, presentation is made on the
experiences in the design and difficulties in dealing with the
Bangkok is situated at the head of the Gulf of Thailand on the
situation of risen groundwater pressure in the sand layers in the
southern part of the low lying Chao Phraya plain. The vast flat plain
tunnelling and deep excavations just made in the 2nd phase – Blue
is carpeted with a thick soft marine clay layer which overlies a very
Line Extension.
thick series of alluvial soils of alternating layers of stiff to hard clay
and dense to very dense sands. In the upper 50 m zone within
which most foundation and excavation works are placed, the subsoil
profile is relatively uniform throughout the city area, which is an
611
ideal condition for tunnelling. However, the soft soil condition Variation of Piezometric Pressure Profile with
necessitates the use of shield method for tunnel construction. The Elevation
(Standpipe Piezometer)
shallow zone consists of a 12-15 m thick layer of soft to medium
stiff clay (Bangkok clay) that is followed by a layer of stiff to hard Short Term Design Piezometric Profile
clay and a layer of sand (the First Stiff Clay and the First Sand). 100 Long Term Design Piezometric Profile
Below them, alternating layers of stiff to hard clay and dense to very Existing Tender Orange Line (2005)
dense sand exist to a great depth. The sand layers are aquifers of Additional Tender Orange Line (2015)
Bangkok groundwater system. The subsoil stratigraphy along the

90
Post Tender Orange Line (2017)
Orange Line East that is under construction is shown in Figure 2.

80
Contract E1 Contract E2
70
Elevation, (m asl)
60
50
40
30
Figure 2 Soil profile along E1&E2 Contracts of Orange Line East

20
0 20 40 60 80
Piezometric Pressure (m)
It is well known that Bangkok had suffered land subsidence Figure 1 - Variation of Piezometric Profile with Depth
problem owing to excessive deep well pumping from the sand Figure 3 Piezometric pressure in the subsoils at different times
aquifers for water supply. The decline in groundwater head in
shallow aquifer layers was as much as 25-35 m from the ground
surface during the peak of the crisis in early 1980s, when the
subsidence occurred at a rate as high as 120 mm/year. After the
surface water supply was made adequate in most of the city area
around the turn of the century, the situation improved as deep well
pumping diminished. Since then the rebound of groundwater started
to develop and the subsidence ceased in most city areas. The
rebound continues till present and the accumulative amount is about
10-12 m during the last 15 years. At present the piezometric water
head in the 1st and 2nd sand layers of Bangkok where the MRT
underground works are involved with is about 13-14 m depth below
the ground surface that is 10-12 m higher than the condition when
the 1st Phase MRT Underground line was constructed (Figure 3). It
has been risen at a rate of approximately 0.8-1.0 m per year (Figure
4). Because the groundwater law that has been in effect since 2004
totally bans deep well pumping in the entire area of Bangkok and Figure 4 Changes in piezometric heads in Bangkok sand layers
surrounding suburbs, it is foreseen that the ground water heads in
the sand layers might rebound to the hydrostatic level in the near 4. INCIDENTS IN PAST MRT UNDERGROUND WORKS
future.
In the past the subsidence situation created complication of long Problematic incidents related to the impact of groundwater rebound
term settlement and differential settlement in the design of MRT in the underground works of the past MRT project are presented.
underground structures. On the other hand, the condition of the The recently completed underground structures of the Blue Line
lowered groundwater head in the sand layers resulted in ease in deep Extension were placed in the stiff clay and sand layers as shown in
excavations for station construction and shield tunnelling. Incidents Figure 5. The deepest bored tunnel section is at 37 m depth at tunnel
of hydraulic instability were rare in past deep excavations and invert at a location near the river crossing.
tunnelling. However, at present the situation has changed since the
groundwater head started to show significant rebound about 15
years ago. Difficult and problematic incidents were experienced in
a number of recent deep excavations of building construction as well
as in the works of the Blue Line Extension Project. In addition, the
rise of groundwater head was linked to incidents of difficulties in
EPB shield tunnelling in the city including a severe incident in the
2nd phase MRT Underground project that led to a six-month delay of
the stall and repair of the TBM.
Figure 5 Soil Profile along Blue Line Extension Underground
612
of the soil with the side of the diaphragm wall. The U-shaped
Facing the situation of higher water pressure in the 1st and 2nd grouted zone as shown in Figure 7 was adopted. The grouting work
sand layers than that indicated by early investigation, the detailed was launched underneath the ground surface from a platform at the
design of two of the three station boxes that were placed at large mid-height of the station height. The scheme was proven to be an
depths needed to be modified from those planned in the pre-tendered effective method to prevent the hydraulic uplift failure of the
design. This is to avoid problems of base instability during excavation base and it allowed a normal construction process of the
excavation from hydraulic uplift of the stiff clay at the excavation concrete base slab casting under a dry condition.
base. The excavation procedure adopted for the third station
remained the normal approach as the designer considered it was still
adequate. Unfortunately, after a portion of the base area was
excavated and concrete casting, the hydraulic uplift occurred in the
next stage of the base excavation that resulted in flooding and
difficulty in control of water ingress in the remaining stages of the
excavation. The work suffered almost 20 months delay in the
construction to fight with the persistent problem of groundwater
ingress despite extensive remedial grouting exercise. A brief outline
on the situations of hydraulic instability involved with the base
excavation design and construction of these stations is given in the
following sections.
4.1 Difficulty and problem in station excavations
The first station excavation of BLE that needed modification in the

design was Station BS12 which was 225 m long, 25.8 m wide and
32 m deep. The diaphragm wall was 1.2 m thick and 44.5 m deep.
The cross section is shown in Figure 6. The detailed soil profile
along the station excavation was derived from a comprehensive
subsurface borehole investigation made after the award of the
contract. A piezometric pressure was observed at approximately
15.0 m below the ground surface. Figure 7 U-shaped base grouted zone of BS12
The second station was Station BS13 which was 181 m long, 23.4
m wide and 30 m deep. The diaphragm wall was 1.2 m thick and 40
m deep. Cross section and subsoil is shown Figure 9. The
+101.5
piezometric head was observed at approximately 16.0 m below the
ground surface.
Soft Clay
GWL+86.3 +88.0
st
1 Stiff Clay
+77.5
1st Sand
+65.0
2nd Stiff Clay +63.0
2nd Sand
Figure 6 Cross section of BS12
Preliminary design analysis indicated that excavation to the base

would experience hydraulic uplift failure owing to the high
piezometric head in the 2nd sand layer underlying the thin 2nd stiff
clay layer below the base of the excavation. Factor of safety against
the failure considering the weight balance equilibrium was only 0.59
while the Outline Design Specification specified a minimum factor
of 1.03. Thus the potential basal hydraulic uplift failure was the key
factor in the design and construction and it necessitated an auxiliary Figure 9 Cross section of BS13
measures. Among three possible measures, i.e. extending tips of the
diaphragm wall to the 3rd stiff clay layer, the dewatering by deep Similar to the condition of BS12 excavation, potential base
well pumping and base grouting, the third method was adopted for instability against hydraulic uplift was faced. However, the situation
reason of the most technical viable and cost effective one. The entire was not as critical. Factor of safety against the failure considering
area of the excavation base was grouted. Two methods of grouting the weight balance equilibriumn was marginally at 0.96. In addition,
were adopted, i.e. chemical grouting by Tube-A-Manchette (TAM) the 2nd stiff clay layer at BS13 was not as thin as in the case of
and cement jet grouting. TAM grouting was chosen for the main BS12, (i.e. 6 m). Therefore, a different measures was adopted. To
part of station box excluding the north and south ends. Both minimize time and cost impacts on the construction work, the
methods were to create a watertight barrier in the 2nd sand to thicken normal approach without implementation of auxiliary ground
an impervious plug below the excavation base to counteract the improvement measures was considered and finally adopted. The
hydraulic uplift pressure as well as to improve side shear resistance fundamental of the method is that if the base excavation is made in
613
stages of a small sized excavation zone bounded by high soil BLE project that the EPB shield tunnelling had to be made through
surcharge + cast concrete slab around the perimeter of the the sand layer as deep as 37 m below the ground surface with the
excavation, the stability of the excavated base against the hydraulic piezometric head 10 m higher than that existed during the time of
uplift will be significantly assisted by the side shear resistance of the the first MRT shield tunnelling, an incident of serious water ingress
soil along the perimeter as compared with that of large excavation occurred in August 2013 during the early drive of an EPB shield
area. The Outline Design Specification stipulated at that in case through the interface between the 1st stiff clay and the 1st sand layer.
shear strength of base soil was considered, a minimum factor of The consequence was flooding and severe damage to the TBM that
safety was required at 1.20. This requirement could be satisfied for led to an interruption of the work for over 7 months. The incident
the soil and groundwater condition below the excavation base of the was an uncontrolled water ingress into the tunnel in the tail area of
station. The accurate determination of the density of the soil layers the EPB shield machine. Flooding and excessive movements of
and the spatial variation of the thickness of the stiff clay layer were segmental rings in vicinity of the shield were experienced. It was
a must in the adaptation of this method. The staged excavation reported the ingress initially occurred through a 20 mm wide gap
sequence adopted for BS13 excavation is shown in Figure 10. The formed at the circumferential joint of the key segment placed near
work was successfully completed as planned. the invert of the ring. The groundwater ingress from the 1st Sand
layer shortly led to piping of fine sand into the tunnel, and
subsequently excessive movements and distortion of several
segmental rings behind the shield. During the incident various
remedial grouting measures were urgently carried out to control the
ingress and avoid excessive distortion of the concrete segmental
rings that might ultimately result in a tunnel collapse.
It was obvious that the incident was caused by difficulty of the
EPB shield in controlling stability of the excavated tunnel face while
the tunnel was being excavated in the mixed face stiff clay and sand
layers. At that position the groundwater head in the sand layer was
17 m. The sand was in the lower part of the tunnel and as the
tunnel was progressively advanced, the portion of sand in face area
progressively increased as the TBM drive pitched down. This
condition was one of the most difficult tunnel grounds to deal with
in soft ground tunnelling. It was apparently that with high ground
water pressure in the sand layer in the excavated tunnel face, the
shield machine experienced difficulty in maintaining adequate face
Fig. 10 Sequence of staged base excavation of BS13 pressures. Thus over excavation would significantly occur. After a
careful examination on the TBM operation data it was believed that
Successful base excavation of the BLE Underground station was the triggering factor of the incident was an excessive displacement
not the case at BS11 Station which was the deepest excavation of the key segment of the segmental ring resulted from the over
among all MRT stations in Bangkok. The base of the excavation excavation of the mixed face soil by the TBM under the high
was at 32 m below the ground surface. Prior to the start of the groundwater condition.
excavation it was judged by the designer that the subsoil and
piezometric head in the sand layers data indicated that the base 5. ORANGE LINE EAST
excavation could be with adequate factor of safety against hydraulic
uplift. Unfortunately, after the excavation was made to the lower The 3rd Phase MRT Underground work that is under construction at
level of the bracing of the D-wall and base excavation and concrete this stage is a part of the Orange Line East project (Cultural Center
base slab casting was progressed from one end of the station box to to Minburi) that is implemented as design-build contracts. The
about 15 percent of the length of station, significant water ingress underground portion constitutes 12.1 km length out of the total 21.1
occurred that seriously affected the work. The whole station was km length of the line. There are 10 underground stations along the
flooded by 2-m-depth of water within 12 hours. Extensive underground section that the works are administered in 3 contracts
dewatering pump system was installed and a comprehensive (E1-E3). The underground line is mainly constructed by bored
program of chemical and cement grouting subsequently employed. tunneling using 3 EPB shields except for a short cut-and-cover
The amount of ingress was as high as 1400 cubic meter/hour at one startup section. Owing to the tight space along the corridor of the
stage. Groundwater ingress repeatedly occurred in the subsequent alignment of the line that traverse along two most congested roads
stages of excavation for concrete casting that altogether delayed the of the city that also have foundation of elevated roadway along the
work by 20 months. The investigation pointed that there might be a right or way, the MRT underground twin tunnels are mostly placed
condition of lateral non-continuity of the 2nd stiff clay below the in the vertically stacked configurations and in many places they
excavation base that in the initial design was considered sufficiently need to be placed at very close proximity to each other, particularly
thick and covering the entire area of the excavation base. This when leaving or entering the station boxes (Figures 2 and 11). The
incident of severe and prolonged groundwater ingress also led to tunnels also pass very near to or cross foundations of existing
subsidence problem of the surrounding ground and buildings that buildings or other structures, some of which need underpinning
necessitated extensive stabilizing grouting works throughout the works. The close proximity shield tunneling works for the deeper
period. section of alignment in the sand layer of high groundwater pressure
for the shallow section in soft clay layer necessitate special
4.2 Water ingress in EPB shield tunnelling considerations in the design and construction planning to avert
instability problems during construction and to ensure stability of
Prior to the tunnelling in the Blue Line Extension project, bored the completed structures in long term. Because of the adoption of
tunnelling by means of EPB shields and concrete segmented ring the vertically stacked configuration of the running tunnels most of
support had been commonly used with great success including those the underground stations are excavated at large depth. And with the
in the Blue Line Initial System project, in which the tunnelling was very extraordinary thickness of the 1st sand layer in the subsoils in
made deep in the sand layer at depth as much as 25 m below the the considerable part of the alignment, careful considerations are
ground surface. There had been no incidents of groundwater needed for the design and planning of the station excavation to avert
problem in EPB shield driving in Bangkok sands. However, in the potential basal instability problems as earlier experienced in the
previous project.
614
grouting (Figure 12). In addition, an extensive instrumentation

C&C Tunnel Bored Tunnel Bored Tunnel program is employed to closely monitor the response of the viaducts
and ground movements at all stages of the station excavation.
1400 m (approx.) 1400 m (approx.) 1900 m (approx.)
OR14
OR15
Tunnel Profile-Contract E1
Figure 11 Vertically stacked alignment of tunnels in Contract E1
5.1 Close proximity tunnelling
The design of the bored tunnelling and concrete segment lining has
to deal with the situation of very close proximity placement of the
twin tunnels in various places. The clear distance between the
tunnels is as close as 1.6 m which is only 25% of the tunnel
diameter (e.g. Figure 12). For such conditions where the tunnels are
to be excavated in the 1st sand layer, special measures are provided
to avert potential instability problems to the lining of the firstly
excavated tunnel from the TBM weight and any unexpected
excessive over-excavation during the excavation of the second
tunnel. The protective measures generally consist of ground
improvement by jet cement grouting. Where the ground space does
not permit the launching of ground improvement, other measures are
used. They include installation of steel ring bracing inside the
concrete segmental lining of the first driven tunnel, the careful
driving of the shield, and the close monitoring of the movements of
the lining and ground.
1.6 m
Figure 13 Examples of close proximity tunnel to existing piled

foundation and underpinning by jet cement grouting
Figure 12 Example of close proximity shield tunnelling
Close proximity tunneling of the works also need to deal with the
situation of potential impacts to the piled foundation of the existing
structures and vice versa the impact of the load to the stability of the
tunnel lining. Extensive numerical analyses are employed to check
the situation of all cases in details and necessary preventive and
corrective measures are provided in the design. In some cases, the
excavation may be made without the need for auxiliary measures,
but in some cases, ground improvement by jet cement grouting is
adopted (Figure 13). In the worst case, underpinning with pile
replacement needed adopted.
Besides the cases of close proximity tunneling, there is a case of
the station box excavation placed very close to the piers of the
elevated expressway at a clear distance between the diaphragm wall
and the foundation piles of the pier as small as 2.7 m. In this case
the impact assessment with the aid of rigorous numerical analysis
(PLAXIS) suggests implementation of the mitigation measure to Figure 14 Pre-installed struts for station box excavation next to
control the lateral movement of the viaduct pier and induced flexural expressway viaducts
stress in the piles within the allowable limits by using preinstalled
struts for the station excavation at two levels by means of jet cement
615
projects. publication in the SEAGS & AGSSEA Journal. We look

forward to receiving your paper contribution.
5.2 Underpinning work at crossing with existing Depot Access
A difficult and challenging underpinning work of the project is at
the crossing of the cut-and-cover tunnel construction with the
existing cut-cover-tunnels of the 1st Phase Blue Line Initial to the
depot yard. The existing tunnels are at shallow depth and have piled
mat foundation. Because the existing depot structure permits only 6
mm of movement of the rail structure, a careful consideration in the
planning and design of the underpinning works in the limited access
space is made. Methods of underpinning initially considered
included the use of pipe roof jacking technique, jet cement grouting
over the entire area, etc. However, due to the presence of many
piles underneath the existing depot structure, the pipe jacking was
not used. The underpinning scheme needs to employ combination of
methods including barrette pile installation, soil improvement by jet
grouting of large diameter columns and soldier pile wall excavation
underneath the existing structures, etc. The section of the crossing
is shown in Figure 15.
Figure 15 Underpining work at Depot Access crossing
5.3 Measures to deal with potential hydraulic uplift in

excavation of station boxes
After thorough assessment on the alternative measures to deal with
potential hydraulic uplift and water ingress problems at the base of
the excavation of various deep station boxes of the project, it has
been decided to adopt the option of positive cut-off of the
diaphragm walls by extending the tips deeper below the base of the Figure 16 Examples of designed diaphragm wall tips for uplift
excavation to the 3rd stiff clay layer. Although this measures calls prevention at two station boxes
for a much larger depth of the diaphragm wall than the base grout
plug alternative, it is judged that the ease and time duration of 7. REFERENCES
construction are more favorable. Figure 16 show examples of the
soil conditions and design of the diaphragm wall tip for uplift Ch. Karnchang Public Co. Ltd. (2013), Assessment of Grout Plug
prevention. Remedial Works at Sanam Chai Station (BS12), The MRT
Blue Line Extension Project: Contract 2.
6. CONCLUSION Ch. Karnchang Public Co. Ltd. (2013), Assessment of Staged
The MRT Underground works in Bangkok have faced increasing Excavation Works at Itsaraphap Station (BS13), The MRT
difficulties and challenges owing to the situation of rebound of Blue Line Extension Project: Contract 2.
groundwater level in the subsoils and the works in the remaining Ou, Chang Yu. (2006), Deep Excavation, 1st edition, London :
projects are situated in congested narrow roads of the city where Taylor & Francis
there are numerous obstructions along the corridor. Therefore, the
design and construction planning need to be carefully made to avert
undesirable instability incidents in both deep station excavation and
shield tunneling similar to that had recently occurred in the previous
616

Proceedings of SEAGS-AGSSEA Conference 2 PDF

Caricato da

Informazioni sul documento

Titolo originale

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

Proceedings of SEAGS-AGSSEA Conference 2 PDF

Caricato da

Copyright:

Formati disponibili

Proceeding 20th SEAGC - 3rd AGSSEA Conference in conjunction with 22nd Annual Indonesian National

“Geotechnical Challenge for Mega

Editor : Masyhur Irsyam

HIMPUNAN AHLI TEKNIK TANAH INDONESIA

The t heme o f 20SEAGC-3AGSSEA Conference i s “Geotechnical C hallenge f or M ega

Wassalamu’alaikum Wr. Wb.

Dr. Pintor T. Simatupang

MESSAGE FROM PRESIDENT OF INDONESIAN SOCIETY

Dear our honorable guest,

Geotechnical in our heart!!!

Jakarta, 6 November 2018

Prof. Ir. Masyhur Irsyam, MSCE., Ph.D.

Advisor : Prof. Ir. Masyhur Irsyam, MSE, Ph.D

Event : Ir. Fauzi Buldan Y.

Sponsorship : Anthony Setiawan, MBA

Secretariat : Josephine Aristiti Setyarini, ST. MT.

Publisher : Himpunan Ahli Teknik Tanah Indonesia

Address : Basement Aldevco Octagon

ISBN No. : 978-602-17221-6-9

Preface Committee Chairman …………………………………………………….. i

Message from President of Indonesian Society for Geotechnical Engineering ii

Organizing Committee ............................................................................................ iii

Table of Contents ..................................................................................................... v

1. Wastes Cover Wastes: A Novel, Scientific and Environmentally Friendly

2. Damages Associated with Geotechnical Problems in 2018 Palu Earthquake,

3. Piling Construction and Testing of Megastructures on Problematical Soil

9. Challenges in Design and Construction of Deep Excavation and Case

10. Increasing geotechnical challenges in the design and construction planning

11. Mega Reclamation Projects: Challenges and Lessons Learned in Soil

Session I.A1 – Pile. (TC 212)

1. Design and Construction of Foundation System for Malaysia First

2. Advancement of Bearing Capacity and Settlement Analyses of Piled-Raft

3. Effect of Intermediary Weak Layer on The Behaviour of Piled Raft

4. Analysis of In Situ Laterally Loaded Tests on Caisson Foundations

6. Foundation Value Engineering & Underpinning Using Micropile for

7. Interpretation on Performance of Two Drilled Shafts Subjected to Tensile

9. Re-Evaluation of Pile Capacity due to Shallow Gas – A Malaysia Case

Session II.A2 - Pile & Instrumentation. (TC 212)

10. Application of Newly Developed Real- Time Website-base GIS

15. Estimating Pile Axial Bearing Capacity by c-phi' Derived from

17. Potentiality of Boehmeria Nivea as Alternative Material in the Production

18. A Multidisciplinary Ground Model Approach to Geotechnical and

Session III.B1- FEM. (TC 103)

20. A Case Study On Monitoring and Numerical Analysis of Groundwater

21. Numerical Simulations of an Excavation Case in Jakarta by Hypoplasticity

22. 3D Finite Element Analysis of Deep Excavation in Central Jakarta using

24. Assessment of Mechanical Behavior of Granular Soils Adopting Various

25. Remediation of Oil Tank Using PLAXIS 3D (Anthony Gunawan)……..…..….. 228-233

27. 3D Finite-Element Analysis on Behaviours of Pile Group and Piled Raft

28. Finite Difference Analysis of Raft Foundations under Vertically Static

29. Determining Unsaturated Soil Properties Through Parameter Estimation

30. Stochastic Estimation of Consolidation Settlement of The Upper

34. Investigation of Aqueous Phase Liquids Migration in Double-Porosity Soil

36. Correlations Between Gradation, Physical and Mechanical Parameters for

37. Characterization of Leachate Distributions at Ngipik Municipal Solid

Session V.C1- Earthquake. (TC 203)

38. Evaluation of Remedial Works for a Spillway on Landslide-dammed Lakes

39. A Comparison Between VS30 Based- and Natural Frequency Based-Site

40. Maps of Corner Period (Tc) of Response Spectra In City of Jakarta

41. Analysis of Bedrock Synthetic Ground Motion on Bandung City using

42. Influence of Cyclic Behaviour of Vibratory Pile Driving And Surging on

43. Dynamic Compaction at New Yogyakarta International Airport for

45. Seismic Analysis of El–Agrem Concrete Face Rockfill Dam (Merouane