Three-Dimensional Interaction PDF

Three-dimensional interaction of multiple crossing
tunnels: centrifuge and numerical modeling
TITLE
by
BOONYARAK Thayanan
A Thesis Submitted to
The Hong Kong University of Science and Technology
in Partial Fulfillment of the Requirements for
the Degree of Doctor of Philosophy
in Civil Engineering
August 2014, Hong Kong
HKUST Library
Reproduction is prohibited without the authors prior written consent
ABSTRACT
Due to the growing demand for underground space in densely populated cities worldwide, an
increasing number of closely spaced multiple tunnels are being constructed. The effects of
new tunnel excavation cause ground movements and stress changes which in turn result in an
adverse impact on the adjacent tunnels. Adverse effects of tunnel driving on adjacent existing
tunnels, such as excessive settlement, large angular distortion and cracking of tunnel linings,
have been reported in a number of case histories. However, the crossing-tunnel interaction is
complex and is influenced by many factors. The main objectives of this research are to
improve the fundamental understanding of the interaction of multiple crossing tunnels and to
provide high quality physical data for numerical modelers and engineers for checking their
designs.
A total of eight centrifuge tests were carried out in dry Toyoura sand. Factors influencing the
interaction of crossing tunnels, namely the effects of modeling technique, construction
sequence, cover depth, pillar depth, shielding and twin new tunnel excavation in side-by-side
and vertically stacked arrangements, were investigated. Tunnel excavation was simulated
three-dimensionally in-flight using a novel device called a Donut, which modeled the
effects of tunnel volume loss equaling 2% and weight loss. The measured results were
back-analyzed using the finite element method to enhance understanding of the stress transfer
mechanism, strain induced and mobilization of stiffness in crossing-tunnel interaction. An
advanced hypoplasticity constitutive model with small strain stiffness was adopted. In
addition, numerical parametric studies were also performed to examine the influence of
volume loss, relative density, tunneling in saturated sand and tunnel diameter in regards to the
interaction of crossing tunnels.
One of factors that strongly influenced the crossing-tunnel interaction is the pillar
depth-to-diameter ratio (P/D). Note that the pillar depth is the vertical clear distance between
tunnels. In the excavation of a new tunnel underneath at P/D of 0.5, the maximum settlement,
tensile strain and shear stress induced in the existing tunnel exceeded the permissible limits
given by LTA (2000), ACI (2001) and ACI (2011), respectively. By increasing P/D from 0.5 to
2, the tunnel settlement was reduced by 50%. This is attributed to a larger shear modulus and
a smaller reduction in confining stress of soil in the case for P/D of 2 along the invert of the
existing tunnel than for P/D of 0.5. The existing tunnel was elongated horizontally when P/D
equaled to 0.5. This is because the stress reduction in the horizontal direction was greater than
that in the vertical direction. The stress relief caused by the new tunnel not only led to a
reduction in the vertical stress at the invert but also resulted in substantial stress reduction at
the springline of the existing tunnel. On the contrary, the existing tunnel was elongated
vertically as the new tunnel was excavated at P/D of 2.0 since the reduction in stress in the
vertical direction dominated.
For multiple crossing-tunnel interaction, the settlement of the existing tunnel caused by the
vertically stacked tunnel arrangement was smaller than that due to the side-by-side tunnel
case. This is attributed to larger P/D of the lower new tunnel in the vertically stacked tunnel
case than the new side-by-side tunnels. In addition, the shielding effects provided by the
upper new tunnel minimized the effects of the new tunnel excavation on the existing tunnel.
ii
ACKNOWLEDGEMENTS
First of all I would like to sincerely thank my supervisor, Professor Charles W. W. Ng, for his
constructive guidance and supervision throughout the period of my research. His expert
knowledge in this field is the key for the successful completion of this research. I am deeply
impressed by his critical and creative thinking that benefited me not only for this research but
also for my future career.
I acknowledge, with many thanks, to my committee members Prof. Chun Fai Leung, Prof.
Gang Wang, Prof. Jui-Pin Wang and Prof. Christopher Y. H. Chao for their valuable comments
and suggestions on my research. I also appreciate help from Prof. David Man (Charles
University in Prague) for providing information and recommendation on the hypoplasticity
constitutive model.
I am grateful for all the support provided by Dr Paul van Laak, Peter Wong, Ka Shing Poon
Michael Lai, Michael Wu and Shirley Tse, the staff of the geotechnical centrifuge facility at
HKUST. Without their assistance, my centrifuge tests would not have been successfully
completed.
I appreciate support from all my friends at HKUST. It is always very pleasant and enjoyable
to work in our wonderful research group. Their assistance during my test and comments on
my work help me to improve my research.
Special thank to my parent and younger brother for their love, understanding and caring
throughout these years. They are the one who make my life meaningful and delightful.
Most importantly, I would like to thank my wife, Police Lieutenant Colonel Kongkwan
Boonyarak. Her love, inspiration and encouragement always give me strength to overcome all
the difficulties.
v
TABLE OF CONTENTS
TITLE .................................................................................................................................. i
ABSTRACT .............................................................................................................................. ii
AUTHORIZATION ................................................................................................................ iii
ACKNOWLEDGEMENTS ..................................................................................................... v
NOMENCLATURE .............................................................................................................. xiv
LIST OF TABLES ................................................................................................................. xvi
LIST OF FIGURES ............................................................................................................. xviii
LIST OF PLATES ................................................................................................................ xxv
CHAPTER 1 INTRODUCTION ............................................................................................ 1
1.1 Research background ............................................................................................... 1

1.2 Objectives of the research ........................................................................................ 3
1.3 Thesis outline ........................................................................................................... 3
CHAPTER 2 LITERATURE REVIEW ................................................................................ 5
2.1 Introduction .............................................................................................................. 5

2.2 Ground movement due to tunneling ........................................................................ 5
2.2.1 Surface settlement ......................................................................................... 5
2.2.2 Subsurface settlement ................................................................................... 6
2.2.3 Horizontal displacement ............................................................................... 7
2.2.4 Ground displacement due to multiple tunnel excavation .............................. 8
2.3 Change in stress, strain and stiffness of soil due to tunnel excavation .................... 9
2.3.1 Stress transfer due to tunneling ..................................................................... 9
2.3.2 Strain induced caused by tunnel construction ............................................. 10
2.3.3 Mobilization of stiffness due to tunnel excavation ..................................... 10
2.4 Effects of tunneling on adjacent existing tunnels .................................................. 11
2.4.1 Observed response of existing tunnels due to new tunnel excavation ........ 11
vi
2.4.2 Physical modeling of tunnel excavation ..................................................... 13
2.4.3 Analytical study and numerical analyses .................................................... 14
2.5 Summary ................................................................................................................ 17
CHAPTER 3 THREE-DIMENSIONAL CENTRIFUGE MODELING ........................... 34
3.1 Introduction ............................................................................................................ 34

3.2 Modeling principles and scaling laws .................................................................... 34
3.3 Test setup ............................................................................................................... 35
3.3.1 The geotechnical centrifuge facility............................................................ 35
3.3.2 Typical setup of centrifuge model package................................................. 36
3.4 Experimental program ........................................................................................... 37
3.5 In-flight tunneling simulation technique ................................................................ 39
3.6 Instrumentation ...................................................................................................... 41
3.6.1 Types and location of instrumentation ........................................................ 41
3.6.2 Calibration of transducers ........................................................................... 43
3.7 Model preparation .................................................................................................. 43
3.8 Test procedure ........................................................................................................ 45
CHAPTER 4 THREE-DIMENSIONAL NUMERICAL ANALYSIS................................ 75
4.1 Introduction ............................................................................................................ 75

4.2 Detail of numerical runs......................................................................................... 75
4.2.1 Finite element program ............................................................................... 75
4.2.2 Numerical back-analysis ............................................................................. 76
4.2.3 Numerical parametric study ........................................................................ 76
4.3 Finite element mesh and boundary conditions....................................................... 77
4.4 Constitutive model ................................................................................................. 78
4.4.1 Description of hypoplasticity constitutive model ....................................... 78
4.4.2 Advantage of hypoplasticity model ............................................................ 80
4.5 Determination of model parameters....................................................................... 82
vii
4.5.1 Parameter controlling behavior in large strain ............................................ 82
4.5.2 Small strain stiffness parameters................................................................. 84
4.5.3 Numerical calibration result ........................................................................ 84
4.6 Numerical modeling procedure.............................................................................. 85
4.6.1 Numerical back-analysis ............................................................................. 85
4.6.2 Numerical parametric study ........................................................................ 87
CHAPTER 5 EFFECTS OF VOLUME AND WEIGHT LOSSES ON

PERPENDICULARLY CROSSING TUNNELS ............................................ 100
5.1 Introduction .......................................................................................................... 100

5.2 Effects of volume and weight losses on ground and existing tunnel ................... 101
5.2.1 Ground surface settlement above the existing tunnel ............................... 101
5.2.2 Settlement of the existing tunnel and tunnel gradient ............................... 102
5.2.3 Induced strain and shear stress in the longitudinal direction of the existing
tunnel ................................................................................................................... 103
5.2.4 Tunnel deformation ................................................................................... 105
5.2.5 Induced strain in the transverse direction of the existing tunnel .............. 106
5.3 Effects of three-dimensional stress transfer on an existing tunnel....................... 107
5.3.1 Incremental normal stress in the transverse direction of the existing tunnel107
5.3.2 Vertical stress in the longitudinal direction of the existing tunnel ............ 108
5.4 Soil arching in perpendicularly crossing tunnels ................................................. 109
5.5 Summary and conclusions ................................................................................... 110
CHAPTER 6 INFLUENCE OF CONSTRUCTION SEQUENCE AND COVER DEPTH

ON CROSSING-TUNNEL INTERACTION .................................................. 122
6.1 Introduction .......................................................................................................... 122

6.2 Response of an existing tunnel due to new tunnel excavation............................. 123
6.2.1 Vertical displacement and gradient of the existing tunnel ........................ 123
viii
tunnel ................................................................................................................... 125
6.2.3 Deformation of the existing tunnel ........................................................... 126
6.3 Change in soil stiffness with construction sequence and cover depth ................. 129
6.4 Stress acting on the existing tunnel ...................................................................... 130
6.4.1 Incremental normal stress acting on the existing tunnel ........................... 130
6.4.2 Mobilized vertical stress of soil acting on the existing tunnel .................. 133
6.4.3 Arch length of new tunnel excavation ...................................................... 134
CHAPTER 7 EFFECTS OF PILLAR DEPTH AND SHIELDING ON CROSSING

MULTI-TUNNEL INTERACTION ................................................................ 147
7.1 Introduction .......................................................................................................... 147

7.2 Existing tunnel responses due to new tunnel excavation with effects of pillar depth
and shielding .................................................................................................................. 148
7.2.1 Settlement of the existing tunnel............................................................... 148
tunnel ................................................................................................................... 149
7.3 Influence of pillar depth and shielding on soil stiffness ...................................... 153
7.4 Stress and strain of soil caused by new tunnel excavation .................................. 154
7.4.2 Incremental normal stress in the longitudinal direction of the existing tunnel156
7.4.3 Induced deviatoric strain of soil ................................................................ 157
7.4.4 Induced deviatoric strain of soil ................................................................ 158
CHAPTER 8 RESPONSES OF AN EXISTING TUNNEL DUE TO TWIN TUNNEL

EXCAVATION ACROSS UNDERNEATH ..................................................... 175
ix
8.1 Introduction .......................................................................................................... 175
8.2 Impact of twin tunnel excavation on an existing tunnel ...................................... 176
8.2.1 Settlement and gradient of the existing tunnel .......................................... 176
tunnel ................................................................................................................... 177
8.3 Change in stress and shear strain of soil due to twin tunnel excavation .............. 181
8.3.2 Induced shear strain of soil ....................................................................... 183
8.4 Effects of arrangement and excavation sequence of twin tunnels on multi-tunnel
interaction ....................................................................................................................... 184
8.4.1 Effects of excavation sequence on the settlement of the existing tunnel .. 184
8.4.2 Incremental vertical stress caused by different tunneling sequences ........ 185
CHAPTER 9 NUMERICAL PARAMETRIC STUDY OF INFLUENCE OF VOLUME

LOSS, RELATIVE DENSITY, TUNNELING IN SATURATED SAND AND
TUNNEL DIAMETER ON CROSSING-TUNNEL INTERACTION ......... 200
9.1 Introduction .......................................................................................................... 200

9.2 Effects of volume loss on crossing-tunnel interaction ......................................... 201
9.2.1 Ground surface settlement ........................................................................ 201
9.2.2 Settlement of the existing tunnel............................................................... 202
9.2.3 Maximum induced tensile strain on the tunnel lining ............................... 203
9.2.4 Incremental vertical stress acting on the invert of the existing tunnel ...... 204
9.2.5 Mobilized shear modulus along the invert of the existing tunnel at different
volume loss ............................................................................................................. 205
9.3 Effects of soil relative density on the interaction of crossing-tunnels ................. 206
9.3.1 Magnitude and shape of tunnel settlement at different soil densities ....... 206
x
9.3.2 Induced tensile strain on the lining of the existing tunnel at different
densities .................................................................................................................. 207
9.3.3 Mobilized vertical stress acting on the invert of the existing tunnel at
different densities ................................................................................................... 208
9.3.4 Mobilized shear modulus of soil at different densities ............................. 209
9.4 Response of existing tunnel due to the new tunnel excavation in saturated sand 210
9.4.1 Comparison of settlement of the existing tunnel in dry and saturated sands210
9.4.2 Induced strain in the lining in the longitudinal direction of the existing
tunnel in dry and saturated sands ........................................................................... 211
9.4.3 Vertical stress acting on the invert of the existing tunnel in dry and
saturated sands........................................................................................................ 211
9.4.4 Mobilized stiffness of soil along the invert of the existing tunnel in dry and
saturated sands........................................................................................................ 212
9.5 Effects of tunnel diameter on crossing-tunnel interaction ................................... 213
9.5.1 Comparison of settlement of the existing tunnel with different diameters 213
9.5.2 Induced strain in tunnel lining with different tunnel diameters ................ 214
9.5.3 Distribution of vertical stress on the existing tunnel with different diameters214
CHAPTER 10 CONCLUSIONS AND FURTHER WORK ............................................. 235
10.1 Summary of this research..................................................................................... 235

10.2 Effects of volume and weight losses on the interaction of crossing tunnels........ 236
10.2.1 Behavior of the existing tunnel due to the new tunnel excavation
underneath .............................................................................................................. 237
10.2.2 Response of the existing tunnel due to modeling volume and weight losses
separately ................................................................................................................ 238
10.3 Influence of construction sequence and cover depth on perpendicularly crossing
tunnels ............................................................................................................................ 238
10.3.1 Behavior of the existing tunnel due to different construction sequences . 238
xi
10.3.2 Effects of cover depth on the tunnel-tunnel interaction ............................ 239
10.4 Effects of pillar depth and shielding on crossing multi-tunnel interaction .......... 240
10.4.1 Behavior of the existing tunnel due to new tunnel excavation beneath with
different pillar depths ............................................................................................. 240
10.4.2 Influence of shielding on crossing multi-tunnel interaction ..................... 241
10.5 Response of an existing tunnel due to new twin tunnel excavation underneath .. 242
10.5.1 Behavior of the existing tunnel due to side-by-side new twin tunnel
excavation............................................................................................................... 242
10.5.2 Response of the existing tunnel due to vertically stacked new twin tunnel
excavation............................................................................................................... 243
10.5.3 Influence of tunnel arrangement and tunneling sequence on the responses of
the existing tunnel due to new twin tunnel excavation .......................................... 243
10.6 Influence of volume loss, relative density and tunneling in saturated sand on
crossing-tunnel interaction ............................................................................................. 244
10.6.1 Volume loss effects on the interaction of crossing tunnels ....................... 244
10.6.2 Response of the existing tunnel due to new tunnel excavation in different
relative densities ..................................................................................................... 245
10.6.3 Interaction of crossing tunnels in saturated sand ...................................... 245
10.7 Discussion of factors to be considered in crossing-tunnel interaction ................. 246
10.7.1 Effects of vibration on multi-tunnel interaction........................................ 246
10.7.2 Optimization between potential damage on the existing tunnel and financial
aspect of tunnel construction .................................................................................. 246
10.8 Recommendations for further research ................................................................ 247
10.8.1 Behavior of the existing tunnel due to new tunnel construction in clay ... 247
10.8.2 Influence of the shape of tunnels on crossing-tunnel interaction ............. 248
10.8.3 Effects of joint in the lining of the existing tunnel on the interaction of
crossing tunnels ...................................................................................................... 248
10.8.4 Effects of boundary condition of the new tunnel ...................................... 249
REFERENCES ..................................................................................................................... 250
xii
APPENDIX A: GROUND SURFACE SETTLEMENT ABOVE THE EXISTING
TUNNEL ............................................................................................................ 259
APPENDIX B: CALIBRATION OF STRAIN GAGE ON THE EXISTING TUNNEL 265
APPENDIX C: AN EXAMPLE OF PLAXIS INPUT FILE............................................. 269
APPENDIX D: LIST OF PUBLICATIONS ...................................................................... 276
xiii
NOMENCLATURE
C Cover depth of tunnel

D Diameter of tunnel
Dr Relative density of soil
D50 Mean particle diameter of soil
E Youngs modulus
e Void ratio
eco Critical void ratio at zero pressure
edo Minimum void ratio at zero pressure
eio Maximum void ratio at zero pressure
Gm Mobilized shear modulus of soil

g gravitational acceleration
hs Granulates hardness
I Moment of inertia of tunnel

K0 Coefficient of lateral earth pressure at rest
mR Parameter controlling the initial shear modulus upon 180 strain path reversal and in
the initial loading
mT Parameter controlling the initial shear modulus upon 90 strain path reversal
N an integer by which gravitational acceleration is to be multiplied

n Exponent n
P Pillar depth (clear vertical distance) between tunnels
p Mean effective stress in soil
q Deviatoric stress in soil
R The size of the elastic range
VL Volume loss of tunnel
xiv
X Distance along the longitudinal direction of the existing tunnel
Y Distance along the transverse direction of the existing tunnel
Z Vertical distance
Exponent
Exponent
r Parameter controlling the rate of degradation of stiffness with strain
n Incremental normal stress acting on a tunnel

v Incremental effective vertical stress in soil
s Ground surface settlement
Tunnel settlement
max Maximum tunnel settlement
s deviatoric strain of soil
d Unit weight of soil in dry condition
t Total unit weight of soil
Parameter controlling the rate of degradation of stiffness with strain
Density of soil
v Effective vertical stress in soil
Friction angle of soil
c Critical state angle of friction
Angle of dilation
xv
LIST OF TABLES
CHAPTER 2
Table 2.1 Summary of case studies .......................................................................................... 19
CHAPTER 3
Table 3.1 Scaling laws applied in the centrifuge tests.............................................................. 46

Table 3.2 Summary of centrifuge tests ..................................................................................... 47
Table 3.3 Modeling sequences of new tunnel advancement to separately investigate effects of
volume (V) and weight (W) losses ............................................................................... 48
Table 3.4 Properties of Toyoura sand ....................................................................................... 49
CHAPTER 4
Table 4.1 Summary of numerical analysis of tunneling sequence of new twin tunnels........... 89
Table 4.2 Summary of numerical parametric study to investigate the effects of volume loss,
relative density and tunneling in saturated sand on the interaction of crossing tunnels
...................................................................................................................................... 90
Table 4.3 Summary of material parameters adopted in finite element analysis ....................... 91
CHAPTER 6
Table 6.1 Summary of tests interpreted to investigate the influence of construction sequence
and cover depth on crossing-tunnel interaction.......................................................... 137
Table 6.2 Summary of maximum induced tunnel gradient .................................................... 138
Table 6.3 Summary of maximum shear stress induced on tunnel lining................................ 138
CHAPTER 7
Table 7.1 Summary of tests interpreted to investigate the effects of pillar depth and shielding
on crossing multi-tunnel interaction ........................................................................... 162
xvi
CHAPTER 8
Table 8.1 Summary of tests interpreted to investigate the response of an existing tunnel due to
new twin tunnel excavation across underneath .......................................................... 188
Table 8.2 Summary of maximum induced tunnel gradient .................................................... 188
Table 8.3 Summary of maximum induced shear stress on tunnel lining................................ 189
CHAPTER 9
Table 9.1 Summary of maximum induced tunnel gradient at different volume losses .......... 218
Table 9.2 Summary of maximum induced tunnel gradient at different soil densities ............ 218
xvii
LIST OF FIGURES
CHAPTER 2
Figure 2.1 Relationship between settlement trough width parameter (i) and depth of tunnel (z0)
for different ground conditions (Peck, 1969) ............................................................... 20
Figure 2.2 Twin tunnel excavation in side-by-side and vertically stacked arrangement in
Bangkok (Sirivachiraporn and Phienwej, 2012) .......................................................... 21
Figure 2.3 Finite element mesh to investigate stress transfer mechanism in an open face tunnel
(Ng and Lee, 2005)....................................................................................................... 22
Figure 2.4 Shear stiffness degradation curve with shear strain and typical shear strain range
for retaining walls, foundations and tunnels (Mair, 1993) ........................................... 23
Figure 2.5 Shear strain induced due to tunnel excavation with and without presence of a
pipeline above in centrifuge test in plane strain condition (Marshall et al., 2010) ...... 24
Figure 2.6 Heathrow Express Tunnels underneath Piccadilly Line Tunnels (a) plan; (b) section
(Cooper et al., 2002) ..................................................................................................... 25
Figure 2.7 Northbound Thameslink 2000 Tunnel underneath Thameslink Tunnel (a) plan; (b)
section (Mohamad et al., 2010) .................................................................................... 26
Figure 2.8 Cut-and-cover tunnel above Metro Line 1 Tunnels (a) plan; (b) section (Liu et al.,
2011) ............................................................................................................................. 27
Figure 2.9 Shekou Line Tunnels underneath Luobao Line Tunnel (a) plan; (b) section (Li and
Yuan, 2012) .................................................................................................................. 28
Figure 2.10 Laboratory 1g model test of crossing tunnels in clay (Kim et al., 1998).............. 29
Figure 2.11 Tunneling underneath a pipeline in sand (Vorster et al., 2005)............................. 30
Figure 2.12 Effects of pillar depth on (a) displacement (b) deformation of the previously
constructed tunnel (Addenbrooke and Potts, 2001) ..................................................... 31
Figure 2.13 Finite element mesh to investigate the effects of construction sequence of
crossing tunnel in rock (Liu et al., 2009) ..................................................................... 32
Figure 2.14 (a) Finite element mesh to investigate the effects of spacing of new twin tunnels
xviii
on an existing tunnel; (b) details of a new tunnel (Chakeri et al., 2011) ..................... 33
CHAPTER 3
Figure 3.1 Comparison of major geotechnical centrifuge facilities in the world (after Ng, 2014)
...................................................................................................................................... 50
Figure 3.2 Typical setup of centrifuge model package in a reference test (a) plan view; (b)
elevation view .............................................................................................................. 51
Figure 3.3 (a) Plan view and; (b) elevation view of Test E2N3 (reference test) ...................... 52
Figure 3.4 (a) Plan view and; (b) elevation view of Test E3N2 ............................................... 53
Figure 3.7 (a) Plan view and; (b) elevation view of Test E2,3N5 ............................................ 56
Figure 3.8 (a) Plan view and; (b) elevation view of Test E2N3,3 ............................................ 57
Figure 3.9 (a) Plan view and; (b) elevation view of Test E2N3,5 ............................................ 58
Figure 3.10 Details of Donut simulating effects of volume and weight losses in (a)
transverse section; (b) longitudinal section view ......................................................... 59
Figure 3.11 (a) Instrumentation in the transverse cross-section view and; (b) the longitudinal
sectional view of the existing tunnel ............................................................................ 60
Figure 3.12 Relative density of sand with height of pluviation in each test ............................ 61
CHAPTER 4
Figure 4.1 Three-dimensional finite element mesh for (a) the reference case (E2N3); (b)
side-by-side twin tunnel excavation (E2N3,3) ............................................................. 92
Figure 4.2 (a) Some details of the perpendicularly crossing tunnels; (b) sequence of between
wished-in-place and non-wished-in-place lining of the new tunnel ..................... 93
Figure 4.3 Limiting curves for void ratio ( Herle and Gudehus, 1999) ................................... 94
Figure 4.4 (a) Stress path; (b) shear stiffness of reconstituted London Clay with different
loading directions (modified from Atkinson et al., 1990) ............................................ 95
Figure 4.5 Response envelope of small strain stiffness in hypoplasticity model (after
xix
Niemunis and Herle, 1997) .......................................................................................... 96
Figure 4.6 Determining of parameter n from limiting curves of void ratio (Herle and
Gudehus, 1999) ............................................................................................................ 97
Figure 4.7 Controlling void ratio curves for Toyoura sand ...................................................... 98
Figure 4.8 Calibration of parameters with test results with bender element and local strain
measurement................................................................................................................. 99
CHAPTER 5
Figure 5.1 Comparison of measured and computed ground surface settlement at the end of
tunnel excavation........................................................................................................ 113
Figure 5.2 Comparison of measured and computed settlement of the existing tunnel at the end
of tunnel excavation ................................................................................................... 114
Figure 5.3 Induced strain along the invert in the longitudinal direction of the existing tunnel at
the end of tunnel excavation....................................................................................... 115
Figure 5.4 Deformations of the existing tunnel in Test E2N3 and the new tunnel in Test
E2N3-VW................................................................................................................... 116
Figure 5.5 Induced strain at the outer surface of the lining in the transverse direction of the
existing tunnel in Test E2N3 ...................................................................................... 117
Figure 5.6 Computed incremental normal stress acting on the existing tunnel at different
excavation stages in Test E2N3 .................................................................................. 118
Figure 5.7 Computed vertical stress in the longitudinal direction of the existing tunnel at the
end of tunnel excavation in Test E2N3 ...................................................................... 119
Figure 5.8 Computed directions of principal stress in case E2N3 in (a) the transverse direction
before tunneling; (b) the transverse direction when the new tunnel reached Ex4; (c) the
longitudinal direction before tunneling; (d) the longitudinal direction when the new
tunnel reached Ex4 ..................................................................................................... 121
xx
CHAPTER 6
Figure 6.1 Vertical displacement along the crown of the existing tunnel at the end of tunnel
excavation................................................................................................................... 139
the end of tunnel excavation....................................................................................... 140
Figure 6.3 Deformation of the existing tunnel (a) when the excavated section reached the
centerline of the existing tunnel (Y/D = 0.3); and (b) at the end of the tunnel
excavation (Y/D = 1.5) ............................................................................................... 141
Figure 6.4 Induced strain measured on the outer surface of the existing tunnel in the
transverse direction (a) when the excavated section reached the centerline of the
existing tunnel (Y/D = 0.3); (b) at the end of tunnel excavation (Y/D = 1.5)............ 142
Figure 6.5 Mobilized secant shear modulus computed along (a) the crown; (b) the invert of
the existing tunnel before and after the excavation of the new tunnel ....................... 143
Figure 6.6 Computed incremental normal stress acting on the existing tunnel (a) when the
excavated section reached the centerline of the existing tunnel (Y/D = 0.3); and (b) at
the end of the tunnel excavation (Y/D = 1.5) ............................................................. 144
Figure 6.7 Mobilized vertical stress ratio of soil located along the side of the existing tunnel
facing the new tunnel after the completion of the new tunnel construction .............. 145
Figure 6.8 Computed direction of principal stress in cases (a) E2N3; (b) E3N2 ................... 146
CHAPTER 7
Figure 7.1 Settlement of the existing tunnel .......................................................................... 163

Figure 7.2 Induced strain in the longitudinal direction of the existing tunnel ....................... 164
Figure 7.3 Measured strain induced on the outer surface of the existing tunnel in the
transverse direction in Tests (a) E2N3; (b) E2N5; (c) E2,3N5 upper tunnel; (d) E2,3N5
lower ........................................................................................................................... 166
Figure 7.4 Deformation of the existing tunnel in (a) the vertical direction; (b) the horizontal
direction ...................................................................................................................... 167
Figure 7.5 Computed normalized stiffness of soil (Gafter / Gbefore) along the invert of the
xxi
existing tunnel ............................................................................................................ 168
Figure 7.6 Computed incremental normal stress of the existing tunnel in cases (a) E2N3; (b)
E2N5; (c) E2,3N5 upper tunnel; (d) E2,3N5 lower tunnel ........................................ 170
Figure 7.7 Computed incremental normal stress at the end of tunnel excavation in the
longitudinal direction along (a) the crown and (b) the invert of the existing tunnel.. 171
Figure 7.8 Contours of computed deviatoric strain induced by new tunnel advancement in
cases (a) E2N3; (b) E2N5; (c) E2,3N5 ....................................................................... 173
Figure 7.9 Computed direction of principal stress in cases (a) E2N3; (b) E2N5 ................... 174
CHAPTER 8
Figure 8.1 Settlement along the crown of the existing tunnel obtained from (a) E2N3,3; (b)
E2N3,5........................................................................................................................ 190
Figure 8.2 Induced strain in the longitudinal direction along the invert of the existing tunnel in
Tests (a) E2N3,3; (b) E2N3,5 ..................................................................................... 191
Figure 8.3 Deformation of the existing tunnel in Tests (a) E2N3,3; (b) E2N3,5 ................... 192
Figure 8.4 Induced strain measured in the transverse direction on the outer face of the existing
tunnel .......................................................................................................................... 193
Figure 8.5 Incremental normal stress around the centre of the existing tunnel after completion
of (a) left tunnel [E2N3,3]; (b) right tunnel [E2N3,3]; (c) upper tunnel [E2N3,5]; (d)
lower tunnel [E2N3,5] ................................................................................................ 195
Figure 8.6 Contour of shear strain after completion of (a) left new tunnel [E2N3,3]; (b) right
new tunnel [E2N3,3]; (c) upper new tunnel [E2N3,5]; (d) lower new tunnel [E2N3,5]
.................................................................................................................................... 197
Figure 8.7 Effects of non-wished-in-place new twin tunnel excavation on settlement of the
existing tunnel ............................................................................................................ 198
Figure 8.8 Incremental vertical stress along the invert of the existing tunnel caused by
different tunneling sequences ..................................................................................... 199
xxii
CHAPTER 9
Figure 9.1 Ground surface settlement above the existing tunnel at different volume losses . 219
Figure 9.2 Settlement of the existing tunnel normalized with (a) diameter of the tunnel; (b)
maximum tunnel settlement at the end of the new tunnel excavation at different
volume losses ............................................................................................................. 220
Figure 9.3 Maximum tensile strain induced at the invert of the existing tunnel at the end of the
new tunnel excavation at different volume losses ...................................................... 221
Figure 9.4 Incremental vertical stress along the invert of the existing tunnel at the end of new
tunnel excavation at different volume losses ............................................................. 222
Figure 9.5 Mobilized shear modulus ratio of soil located along the invert of the existing tunnel
at different volume losses ........................................................................................... 223
maximum tunnel settlement at the end of the new tunnel excavation at different
densities ...................................................................................................................... 224
Figure 9.7 Maximum tensile strain induced at the invert of the existing tunnel at the end of the
new tunnel excavation at different volume losses at different relative densities ....... 225
tunnel excavation at different relative densities ......................................................... 226
Figure 9.9 Mobilized shear modulus of soil located at the invert at the center of the existing
tunnel at the end of new tunnel excavation at different relative densities ................. 227
Figure 9.10 Comparison of settlement of the existing tunnel between dry and saturated sands
at the end of tunnel excavation normalized by (a) tunnel diameter; (b) maximum
tunnel settlement ........................................................................................................ 228
Figure 9.11 Induced tensile strain in the lining in the longitudinal direction of the existing
tunnel in dry and saturated sands at the end of tunnel excavation ............................. 229
Figure 9.12 (a) Vertical stress distribution; (b) incremental vertical stress along the invert of
the existing tunnel for dry and saturated sands .......................................................... 230
Figure 9.13 Mobilized secant shear modulus of soil along the invert of the existing tunnel for
dry and saturated sands .............................................................................................. 231
xxiii
Figure 9.14 Comparison of settlement of the existing tunnel at different tunnel diameters at
the end of tunnel excavation normalized by (a) tunnel diameter; (b) maximum tunnel
settlement ................................................................................................................... 232
tunnel in dry and saturated sands at the end of tunnel excavation ............................. 233
Figure 9.16 (a) Incremental vertical stress; (b) mobilized vertical stress ratio along the invert
of the existing tunnel with different tunnel diameters ................................................ 234
APPENDIX A
Figure A.1 Ground surface settlement due to different modeling techniques ........................ 260
Figure A.2 Ground surface settlement due to different construction sequences and cover
depths ......................................................................................................................... 261
Figure A.3 Ground surface settlement due to different pillar depths and number of existing
tunnels ........................................................................................................................ 262
Figure A.4 Ground surface settlement due to side-by-side new twin tunnel excavation ....... 263
Figure A.5 Ground surface settlement due to vertically stacked vertical twin tunnel excavation
.................................................................................................................................... 264
APPENDIX B
Figure B.1 Calibration curves for strain gages in the longitudinal direction of the existing
tunnel (a) from 0 to -5D (b) from 0.5D to 5D ............................................................ 266
Figure B.2 Calibration curves for strain gages in the transverse direction of the existing tunnel
.................................................................................................................................... 267
Figure B.3 Calibration curves for (a) potentiometers to measure tunnel deformation (b)
LVDTs measuring tunnel settlement .......................................................................... 268
xxiv
LIST OF PLATES
CHAPTER 3
Plate 3.1 Liquid silicone rubber and catalyst ........................................................................... 62

Plate 3.2 Mold for fabrication of (a) outer rubber membrane; (b) inner rubber bag ................ 63
Plate 3.3 Details of (a) outer rubber membrane and inner rubber bag; (b) Donut simulating
effects of volume and weight losses ............................................................................. 64
Plate 3.4 Simulation of tunnel excavation in-flight (a) new tunnel with tubes for heavy fluid;
(b) valves controlling outflow of heavy fluid............................................................... 65
Plate 3.5 A new tunnel (a) excavation underneath an existing tunnel; (b) construction above an
existing tunnel .............................................................................................................. 66
Plate 3.6 Twin new tunnel excavation in (a) side-by-side arrangement; (b) vertically stacked
arrangement .................................................................................................................. 67
Plate 3.7 Typical instrumentation on the existing tunnel where the new tunnel excavate (a)
underneath and; (b) above ............................................................................................ 68
Plate 3.8 (a) Potentiometer and strain gage in the transverse direction of the existing tunnel; (b)
frame and plate for potentiometers............................................................................... 69
Plate 3.9 Measurement of (a) subsurface settlement in Test E2N5; (b) settlement of the lower
existing tunnel in Test E2,3N5 ..................................................................................... 70
Plate 3.10 Calibration of (a) strain gage on the existing tunnel; (b) LVDT; (c) potentiometer 71
Plate 3.11 (a) Placing new tunnel in the strong box; (b) drop height of sand using pluviation
technique ...................................................................................................................... 72
Plate 3.12 (a) LVDT measuring ground surface and tunnel displacement; (b) control panel for
valves and data logger for transducer ........................................................................... 73
Plate 3.13 (a) A typical setup of a centrifuge model package; (b) a model package on the
centrifuge platform ....................................................................................................... 74
xxv
Chapter 1 Introduction
1.1 Research background
Ground movements and changes of stresses caused by tunneling are increasingly important as
more and more tunnels are constructed in urban areas. As a result of a new tunnel excavation
adjacent to an existing tunnel, the existing tunnel may experience excessive deformation and
cracks induced on the tunnel lining.
Some case studies have observed large differential tunnel settlement, along with cracks on
tunnel linings, caused by the excavation of adjacent tunnels (Cooper et al., 2002; Mohamad et
al., 2010; Liu et al., 2011; Li and Yuan, 2012). However, interpreting data from the field is
particularly difficult due to variations in the soil properties, the in-situ stress conditions and
tunneling workmanship.
Three-dimensional numerical analyses of perpendicularly crossing tunnels constructed in rock,

according to different sequences, were conducted by Liu et al. (2009). Compressive failure of
the concrete lining at the crown of the existing tunnel was induced by the construction of a
new tunnel above. In contrast, the tensile strain caused cracks to appear in the lining of the
existing tunnel at the springlines when the new tunnel was excavated underneath. Even
though the effects of tunnel construction sequence on tunnel-tunnel interaction were
investigated in their study, the results may not be applicable to soil as the stiffness
dependency on the stress, strain and path was not modeled. In addition, as the cover depth of
the existing tunnel varies from one construction sequence to another, the influence of cover
depth on the interaction of crossing tunnels should also be further examined. Marshall et al.
(2012) carried out centrifuge tests of a single tunnel excavation in sand, in plane strain
conditions. They reported that the settlement trough width increased with increasing cover
1
depth-to-diameter ratio (C/D) of the tunnel.
Kim et al. (1998) carried out a series of tunnel-tunnel interaction tests using a 1-g model in
clay. Two different pillar depths, defined as the clear vertical distance between two tunnels,
between the existing and new tunnel were tested. They reported that the existing tunnel was
compressed vertically because a jacking force was applied to install the liners of the new
tunnels. Nevertheless, the effects of the second new tunnel excavation on the existing tunnel
were likely affected by the presence of the first new tunnel, as the two new tunnels were
driven one after the other in the same model box.
Plane strain numerical analysis has been carried out to study the effects of parallel twin tunnel
construction in side-by-side and vertically stacked arrangements on the ground movement
(Addenbrooke and Potts, 2001). They suggested that the ground surface settlement, in a
vertically stacked tunnel case, when the lower tunnel was excavated first was larger than
when the upper tunnel was excavated first. In the case of the latter, the upper tunnel
experienced increasing settlement and elongation in the vertical direction with decreasing
pillar depth. Cherhade and Shahrour (2008) reported that vertically stacked tunnels caused a
larger surface settlement than in the side-by-side case. However, the results from both studies
may not be applicable in crossing tunnels as the effects of stress transfer on ground movement
and stress change in the longitudinal direction of the tunnels were not simulated.
The behavior of a pipeline due to a tunnel excavation beneath in sand has been investigated in
a centrifuge (Vorster et al., 2005; Marshall et al., 2010). The effects of volume loss caused by
tunneling were simulated in plane strain conditions. They reported that soil-pipe stiffness was
a major factor influencing the longitudinal bending moment of the pipeline. In addition, the
presence of the pipeline significantly reduced the amount of shear strain induced above the
pipeline. Addenbrooke and Potts (2001) also reported that the size and shape of the ground
surface settlement trough in the case of vertically stacked twin tunnels differed from that in
the greenfield case.
2
1.2 Objectives of the research
This research aims to improve fundamental understanding of crossing multi-tunnel interaction

and to investigate the factors that influence the interaction. The objectives of this research are
as follows:
(1) To understand the responses of an existing tunnel due to a new tunnel excavation
perpendicularly across underneath
(2) To investigate separately the effects of modeling volume and weight losses on crossing
tunnel interaction
(3) To improve understanding of the effects of construction sequence and cover depth on
the interaction of crossing tunnels
(4) To study the influence of pillar depth and shielding on multi-tunnel interaction
(5) To investigate the responses of an existing tunnel due to new twin tunnel excavation with
different arrangements
(6) To examine the effects of volume loss, relative density and tunneling in saturated sand on
the interaction of crossing tunnels
These objectives are achieved by adopting two major methodologies, three-dimensional

centrifuge modeling and numerical analysis.
1.3 Thesis outline
This chapter introduces the research background, objectives and structure of the research.
Chapter 2 reviews the literature related to ground movement, stress transfer, induced strain
and mobilization of stiffness caused by tunneling. Current understanding of the effects of
tunnel excavation on adjacent tunnels is summarized. Key conclusions and limitations of the
literatures are reported.
Chapter 3 presents three-dimensional centrifuge testing. Details of the experimental program,

the objective of each test, a novel tunneling technique, instrumentation, model preparation
3
and testing procedures are given.
Chapter 4 describes three-dimensional numerical analysis. Details of the numerical plan,

finite element mesh, boundary condition, hypoplasticity constitutive model with small strain
stiffness, determination of parameters and numerical procedures are reported.
Chapter 5 presents effects of volume and weight losses on crossing-tunnel interaction. Two
tests with different simulation sequences of tunnel volume and weight losses were
investigated.
Chapter 6 discusses the effects of construction sequence and cover depth on crossing-tunnel
interaction. Results from three tests with different construction sequences and cover depths
for the existing and new tunnels were interpreted.
Chapter 7 explains the effects of pillar depth and shielding on multi-tunnel interaction. Three
tests with different pillar depths and number of existing tunnel are reported and discussed.
Chapter 8 discusses the effects of new twin tunnel excavation with different arrangements on
the interaction of multiple tunnels. The responses of an existing tunnel due to side-by-side and
vertically stacked twin tunnel excavation from two tests were interpreted.
Chapter 9 presents numerical parametric studies to examine the effects of volume loss,
relative density, tunneling in saturated sand and tunnel diameter on the interaction of crossing
tunnels.
Chapter 10 summarizes the research work carried out and the major conclusions reached.
Recommendations for further research are also given.
4
Chapter 2 Literature review
2.1 Introduction
To meet the growing traffic demands and reduce environmental impacts, there have been an
increasing number of tunnels constructed in urban cities. Consequently, the clear distance
between tunnels has become closer and closer. Tunnel driving inevitably induces changes in
stress and deformation in the ground, which could cause ultimate and serviceability problems
to an adjacent tunnel. However, the effects of ground movement, induced stress and change in
stiffness on crossing-tunnel interaction are still not fully understood.
This chapter reviews ground movement due to tunnel construction. Change in stress, induced
strain and mobilization of stiffness caused by excavation of a tunnel are also explained. The
current knowledge of the effects of tunneling on underground structures such as existing
tunnels and pipelines are also discussed. Limitations of previous studies are summarized.
2.2 Ground movement due to tunneling
2.2.1 Surface settlement
Peck (1969) suggested that a Gaussian distribution curve can be adopted to estimate the shape
of ground surface settlement trough caused by tunnel excavation. By using following
expression, ground surface settlement (S) can be calculated by using eq. 2-1 and 2-2.
S = Smax exp(-x2/2i2) 2-1
Vs
S max 2-2
2 i
5
where Smax is the maximum ground surface settlement above the tunnel centerline, i and x are
the lateral distances from the tunnel centerline to the point of inflection and any arbitrary
point on the settlement trough, respectively. Vs is the volume of surface settlement trough. A
relationship between i and the tunnel depth and diameter for various material was proposed as
shown in Figure 2.1. An empirical relationship of 2.5i can also be adopted as the influence
zone caused by tunnel excavation on ground surface.
Oreilly and New (1982) collected ground surface settlement data from shallow tunnels in
United Kingdom. They suggested that the width of surface settlement trough (i) increases
almost proportionally with depth of tunnel axis according to following equation:
i=K.z0 2-3
where K is a settlement trough width parameter which varies for different materials and Z is
the vertical distance from ground surface to the centre of tunnel. K typically equals to 0.5 for
clay and in 0.25 for sand and gravel, respectively.
Mair and Taylor (1997) interpreted surface settlement trough from more than 50 case histories
and suggested the value of K was from 0.4 to 0.6 for clay and from 0.25 to 0.45 for gravel and
sand.
Marshall et al. (2012) investigated effects of tunnel diameter, tunnel depth and volume loss on
greenfield ground surface settlement trough in dry sand by using centrifuge test. They found
that a curve with three degree of freedom can provide a better fit compared with conventional
Gaussian distribution curve. The key parameters controlling ground surface settlement trough
shape are zt-z0, C/D and VL.
2.2.2 Subsurface settlement
Mair et al. (1993) investigated subsurface settlement from both measured data in centrifuge
and field monitoring in soft and stiff clay. They suggested that the tunneling induced
6
subsurface settlement trough in clay can also be represented by using Gaussian distribution
curve. The trough width parameter i at a depth z can be described by the following
relationship:
i = K (z0 - z) 2-4
where z0 is a depth of tunnel axis. It was illustrated that the settlement trough width parameter
K increases with depth and K can be calculated by using the following relationship;
0.175 0.325(1 z / z 0 )
K 2-5
1 z / z0
Moh et al. (1996) also suggested that K of subsurface settlement in silty sand below the water
level in Taipei increases with depth.
Dyer et al. (1996) confirmed that K of subsurface settlement in loose sand overlain by a firm
to stiff clay layer have a similar pattern of K increasing with depth.
Boonyarak et al. (2014) reported field data collected from tunneling in stiff clay. It is reported
that subsurface settlement increased with increasing depth for soil located above the tunnel
and heave was observed in soil located below the tunnel. An average volume loss determined
from ground surface settlement was found to be about 0.7%.
2.2.3 Horizontal displacement
Attewell (1978) and OReilly and New (1982) suggested that ground displacement vector of
tunnel in clay are directed towards the tunnel axis. The following equation can be used to
estimate horizontal displacement
Sh = (x/z0).Sv 2-6
Mair and Taylor (1993) proposed a relationship to estimate horizontal ground movement at
the depth of tunnel axis in London Clay by considering distance from the tunnel axis and
7
tunnel radius.
Deane and Bassett (1995) reported case studies in London Clay. They found that ground
movement was generally found to be directed towards point near the tunnel invert.
Taylor (1995b) suggested that the variation of the trough width parameter (K) with depth
causes soil movements around the tunnel in clay to have subsurface settlement vectors
towards a point that is located at a distance 0.54Z below the tunnel axis. This results in about
65% smaller horizontal movements compared that suggested by Attawell (1978) and OReilly
and New (1982).
Cording (1991) suggested that, for tunnel in sand, the assumption that ground movement is
towards the tunnel axis significantly underestimates the horizontal ground movement at the
ground surface. Hong and Bae (1995) confirmed the finding of Cording (1991) from a case
study of tunneling predominantly in sandy strata in Korea.
2.2.4 Ground displacement due to multiple tunnel excavation
Peck (1969) reported case histories of surface settlement trough caused by closely spaced
twin tunnel excavation. It was suggested that the settlement induced by the excavation of the
second tunnel was often larger than the settlement induced by the first tunnel due to plastic
strain induced by the first tunnel.
Shirlaw et al. (1988) reported a twin tunnel construction in stiff to hard clay in Singapore and
found that volume loss observed from ground surface settlement from the second tunnel was
much larger than that caused by the first one.
Mair and Taylor (1997) summarized several field case studies reporting evidence of
asymmetrical settlement profiles induced by twin tunneling in a variety of materials. The
settlement profiles normally shifts to the second tunnel because of shear strain associated with
the construction of the first tunnel, resulting in reduced stiffness. As a result, a higher volume
8
loss is likely to occur in the second tunnel compared with the first one. They reported typical
volume losses due to tunnel excavation using earth pressure balance shields in sand and soft
clay of up to 1% and 2%, respectively.
Sirivachiraporn and Phienwej (2012) presents a summary tunneling parameters, ground
movement and building settlement induced by tunnel construction in the first Bangkok
subway project. Twin tunnels were constructed with side-by-side and vertically stacked
arrangement (see Figure 2.2), depended on the allowable underground space. Measured data
consists of ground surface settlement, face pressure, grouting pressure, settlement trough
width parameter (i), horizontal ground displacement and building settlement. Most of ground
surface settlement ranges from 20 to 40 mm, corresponding to 0.5-2.0% of tunnel volume loss.
Tunneling in sand or mixed face ground causes larger ground surface settlement which may
be up to 100 mm. There is a slight trend between face pressure and ground surface settlement.
Predominant factor to ground surface settlement appears to be workmanship.
2.3 Change in stress, strain and stiffness of soil due to tunnel excavation
2.3.1 Stress transfer due to tunneling
Ng and Lee (2005) presented a series of three-dimensional numerical analysis of an open-face
tunnel as shown in Figure 2.3. They reported that due to tunnel excavation, the maximum
normal stress reduction in the lateral and vertical directions took place at the tunnel heading,
while the normal stress reduction in the longitudinal direction occurred at a distance of 0.5
times the tunnel diameter ahead of the tunnel face. Stress transfer mechanism in
three-dimension and stress redistribution due to soil arching was discussed in their paper.
Lee et al. (2006) studied the effects of soil arching due to tunneling on the stability of the
tunnel. An arching ratio is defined in this study to describe arching effect in soil mass around
the tunnel. The boundary of arching zone and boundary of positive and negative arching were
proposed.
9
Zhuang et al. (2012) investigated the effects of soil arching on movement of soil and vertical
stress on the area without rigid support using three-dimensional finite element analysis. They
suggested that the minimum ratio of vertical stress at the area without rigid support and the
calculated overburden pressure was about 0.2, which is defined as the maximum arching.
2.3.2 Strain induced caused by tunnel construction
Mair (1993) suggested a non-linear curve of shear modulus reduction with shear strain along
with typical ranges of shear strain in retaining walls, foundations and tunnels (refer to Figure
2.4). For tunneling, shear strain induced was ranging from 0.1% to 1%. Local strain
measurement is required to measured change of stiffness in this region.
Effects of advancing open face tunneling on an existing pile was investigated and reported by
Lee and Ng (2005). They reported that significant increase plastic strain induced around a pile
located within an influence zone of tunneling occurred when tunnel face advance at distance
from -1D to 1D from the centerline of the pile.
Profile of vertical, horizontal, volumetric and shear strain induced by tunnel-pile interaction
was presented by Marshall et al. (2011). The tunnel excavation was simulated using
centrifuge testing in plane strain conditions. Contractive and dilative behavior of soil was
illustrated in their study.
Shear strain induced by a single tunnel excavation was reported by Marshall et al. (2012).
They simulated tunnel excavation in very dense sand in plane strain condition using
centrifuge test. At volume loss of 2.5%, shear strain of up to 1% above the crown of the
tunnel.
2.3.3 Mobilization of stiffness due to tunnel excavation
As discussed in Mair (1993) in the previous section, shear strain induced due to tunneling
resulted in reduction in shear stiffness. It suggests that stiffness at the small strain is very
10
important and should be taken into account in numerical analysis.
Man (2009) adopted an advance hypoplastic model with small strain stiffness to predict
ground movement caused by the Heathrow express trial tunnel. It is found that the hypoplastic
model which modeled stiffness dependency on state, strain and recent stress history gave a
better prediction of ground displacement than the modified Cam-clay model.
Svoboda et al. (2010) presented a comparison between class A prediction of displacement
induced by tunnel excavation and field monitoring. The tunnel was constructed using NATM
in stiff clay. They demonstrated that tunnel simulation using a non-linear constitutive model
with small strain stiffness such as hypoplasticity resulted in a good prediction of ground
movement compared with the measured one. This is because soil stiffness dependency on
state, strain and recent stress history was simulated in the hypoplasticity model.
Marshall et al. (2010) investigated the interaction of tunneling and a pipeline above using
centrifuge test in plane strain conditions. A method considering shear strain level of greenfield
and out-of-plane shear strain around the pipe was proposed to improve the prediction of
bending moment on pipeline. Shear stiffness was estimated from shear strain induced at each
volume loss of tunnel. In addition, the amount of shear strain induced above the pipeline was
smaller than that at the same depth in the greenfield test as shown in Figure 2.5. It suggests
that the reduction of mobilized shear modulus of the soil located above the pipeline would be
smaller than that without a pipeline.
2.4 Effects of tunneling on adjacent existing tunnels
2.4.1 Observed response of existing tunnels due to new tunnel excavation
Yamaguchi et al. (1998) presented observed data from four parallel tunnels constructed in
close proximity in Kyoto. The tunnels were constructed by using earth pressure balanced
shield in diluvial sand and gravel. They found that, when two shields are driven in
vertical-stacked, the existing tunnel which located lower than the new one experiences a
11
decrease in vertical stress and an increase in horizontal stress. For side-by-side tunnels, there
was increment of stress in both vertical and horizontal direction of the existing tunnel due to
stress redistribution. As a result, the increase of horizontal stress is larger than that in vertical
direction.
Cooper et al. (2002) reported a case history of triple side-by-side tunnels of the Heathrow
Express driven underneath across the existing Piccadilly Line twin tunnels (see Figure 2.6).
The new triple tunnels were constructed by using a pilot shield with tunnel enlargement. After
the crossing of the new tunnel, numbers of cracks per ring of about 10-40 per ring were found
on the existing tunnel. The maximum tunnel settlement was 10-30 mm and the maximum
gradient was 1:900. The distortion of the existing tunnel was found to be dependent to the
direction of principle stress around the existing tunnel.
Mohamad et al. (2010) investigated the application of optical fiber cables to measure
additional strain in transverse and longitudinal direction of the existing tunnel due to an
excavation of a new tunnel (refers to Figure 2.7). Field monitoring data was collected during
the construction of new twin Thameslink 2000 Tunnel underneath Thameslink Tunnel. The
new tunnel was driven by using an open face shield in London Clay. The maximum tensile
strain on inner face of tunnel lining of 0.25% was developed in transverse direction on the
side wall and minor cracks were observed. In longitudinal direction, the maximum tensile
strain in longitudinal direction of the existing tunnel was 0.10%.
Liu et al. (2011) reported a case study of cut-and-cover tunnel construction above existing
twin side-by-side circular tunnels in Nanjing as shown in Figure 2.8. The new tunnel was
excavated in soft silty clay with instrumentation on the existing tunnel and surrounding soil.
Heave of the existing tunnel up to 7 mm and horizontal displacement of about 4 mm was
observed. Several measures were used to mitigate impact of excavation on tunnels namely,
sequential excavation, jet grouting and a pile-slab retaining system.
Li and Yuan (2012) collected data from the construction of twin tunnels of Shekou Line
12
driven underneath an existing double-deck tunnel of Luobao Line in Shenzhen (see Figure
2.9). The new twin tunnels were constructed by EPB shield in highly decomposed
granite-gneiss. The existing tunnel settlement caused by the second tunnel crossing was about
two times larger than that caused by the first tunnel. The maximum settlement of the existing
tunnel was 9 mm, the maximum slope in longitudinal direction was 1:3000 and minimum
curvature was 10,730 m.
Mohamad et al. (2012) reported a field trial measurement of strain in transverse direction by
using optical fiber cable of tunnel from side-by-side twin tunnels in Singapore. Due to new
tunnel excavation, maximum compressive strain inside the existing tunnel lining was recorded
just below the tunnel springline on the side of the new tunnel. The compressive strain at the
springline was larger than that at the crown suggesting a distorted vertical compression of the
existing tunnel.
Some details related to crossing-tunnel interaction such as tunnel geometry, soil type, tunnel
excavation method and volume loss from some of the above case histories are summarized in
Table 2.1.
2.4.2 Physical modeling of tunnel excavation
Kim et al. (1998) carried out a series of parallel tunnels and perpendicular crossing tunnels
interaction tests using a 1-g model in clay as shown in Figure 2.10. The new tunnel was
installed by using a model shield tunneling machine that produced a ground loss of about 6%.
The model shield machine, developed based on earth pressure balance shield, was made of
rotating cutter blades, soil removal system and tunnel lining. The result illustrated that in case
of perpendicular crossing tunnel, interaction effects are predominantly caused by jacking
forces applied to the liner and installation of the tunnel lining. As a result, the existing tunnel
was compressed vertically. The effects of pillar depth-to-diameter ratio (P/D) and relative
stiffness of tunnel lining were investigated. Nevertheless, the effects of the second new tunnel
excavation on the existing tunnel were likely affected by the presence of the first new tunnel
13
as the two new tunnels were driven one after the other in the same model box.
Vorster et al. (2005) investigated the behavior of a pipeline caused by a tunnel excavation by
using centrifuge testing in sand as shown in Figure 2.11. Effects of volume loss were adopted
to simulate tunnel excavation by the application of controlled water extraction from annulus
fitted around a hollow mandrel. Tunnel excavation was simulated in plane strain condition by
varying the magnitude of volume loss in-flight. A method to derive deviatoric strain on the
pipeline was proposed and the equation was verified by centrifuge test.
Chapman et al. (2007) presented results from a 1-g laboratory model testing to investigate
effects of multiple side-by-side tunnel excavation on ground settlement. Auger type cutter was
adopted to simulate tunnel driving, then a shield and a liner tube attached to the tail of the
shield was pushed forward into the soil as the auger rotates. Some behavior such as greater
surface settlement above the second excavated tunnel and the translation of the maximum
surface settlement above the second tunnel towards the first tunnel was simulated.
2.4.3 Analytical study and numerical analyses
Verruijt and Brooker (1996) proposed an approximate analytical solution to calculate ground
deformation caused by tunnel volume loss and tunnel ovalization. They suggested that when a
tunnel was vertically compressed, additional ground surface settlement above the tunnel and
narrower surface settlement trough width was observed.
Addenbrooke and Potts (2001) carried out parametric study by using plane strain, coupled
consolidation, non-linear finite element analyses to investigate the effects of twin tunnel
construction. A non-linear elastic constitutive model with Mohr-Coulomb failure criterion was
adopted to simulate tunneling in London Clay. Stiffness decay with strain, stiffness variation
with mean stress and stress reversal was considered. For a case of vertical twin tunnel,
reduction of maximum ground surface settlement and wider ground surface settlement trough
compared with a greenfield case was observed. Influence of pillar depth of displacement and
14
deformation of the previously constructed tunnel is shown in Figure 2.12. In addition, the
existing tunnel was vertically elongated due to a new tunnel excavation in a vertically stacked
tunnel arrangement.
Interaction between twin NATM parallel tunnels was investigated by Ng et al. (2004).
Three-dimensional coupled finite element analysis was carried out. An elastic-perfectly plastic
model with a modified DruckerPrager yield criterion was adopted. The influence of lagged
distance between the first tunnel and second tunnel on the interaction of parallel tunnel was
examined. Effects of new tunnel excavation on displacement, deformation and stress transfer
on the lining of the previously constructed tunnel were reported.
Cherhade and Shahrour (2008) investigated effects of relative position of parallel twin tunnels
and construction sequences on soil movement and internal forces in the tunnel lining. Plane
strain numerical analyses were conducted using an elastic perfectly plastic constitutive model
with Mohr-Coulomb failure criterion. For the case of vertical twin tunnel, when the upper
tunnel is constructed before the lower one, a larger ground surface settlement and bending
moment on the first tunnel was induced compared with when the lower tunnel was
constructed first. The vertical twin tunnel excavation induced the largest ground surface
settlement while the side-by-side twin tunnel excavation induced the smallest ground surface
settlement.
Liu et al. (2008) carried out three-dimensional numerical analysis to investigate interaction
between a new tunnel and an existing parallel tunnel. The soil condition in the analyses was
based on a region with high horizontal pressure such as Sydney by adopting NATM for tunnel
simulation. They suggested that the impact of the new tunnel construction on the existing
tunnel became significant when the distance between tunnels was about 2D in both vertical
and horizontal tunnel alignment.
Verruijt and Strack (2008) investigated the effects of soil weight removal inside the tunnel by
using plane strain numerical analysis with an elastic soil model. They found that a net weight
15
reduction due to removal of soil inside the tunnel causes smaller and narrower ground surface
settlement. The effects of soil weight removal should also be considered to determine
tunneling induce ground settlement.
Liu et al. (2009) adopted three-dimensional numerical modeling of perpendicularly crossing
tunnels for the ground condition based on Sydney region (refer to Figure 2.13). An elastic
perfectly plastic constitutive model with Mohr-Coulomb failure criterion was adopted to
model rock. They found that the interaction between perpendicularly crossing tunnels during
tunnel advancing process was larger than those after at the end of tunnel excavation. When the
new tunnel was excavated underneath the existing tunnel, tensile cracks occurred at the
springline of the existing tunnel. For a case of the new tunnel advanced above the existing
tunnel, compressive failure of the shotcrete lining at the crown was computed.
Chakeri et al. (2011) conducted three-dimensional numerical analysis to investigate the effects
of horizontal spacing of twin side-by-side tunnel excavation underneath an existing tunnel (as
shown in Figure 2.14). Tunnel geometry and soil conditions were based on crossing tunnels of
Tehran Metro. Due to the excavation of the underlying tunnels, the existing tunnel was
vertically compressed. When a horizontal spacing between two new tunnels increased,
maximum settlement of the existing tunnel decreased.
Liu et al. (2011) conducted three-dimensional numerical analyses to investigate effect of
cut-and-cover tunnel construction on underlying existing twin tunnels based on a case study
in Nanjing. An elastic perfectly plastic constitutive model with Mohr-Coulomb failure was
adopted to model silty clay. Sequences of staggered excavation from both sides to the center
of tunnel was the more effective to minimize heave compared with that when the excavation
from the middle to both sides. The influence of cut-and-cover tunnel on a side-by-side parallel
tunnel was smaller than that on the twin perpendicular tunnels located underneath.
16
2.5 Summary
Despite a number of study on the effects of tunnel excavation on another tunnel were carried
out, the interaction of crossing tunnels were still not fully understood. The limitations of the
previous studies are summarized below.
(a) From case histories, large settlement of the existing tunnels, tunnel distortion and cracks
in the tunnel lining caused by the effects of tunnel driving have been reported. However,
variation in ground conditions, tunneling technique, geometry of tunnels and volume loss
among each case history caused difficulty in data interpretation.
(b) Only limited number of research related to tunnel-tunnel interaction were carried out
using physical model test. In addition, crossing-tunnel interaction in some studies was
carried using 1g physical model test. Although some aspects of tunnel-tunnel interaction
in 1g test can be simulated, the behavior of soil which is stress-dependent was not
correctly modeled.
(c) In previous studies, tunnel excavation in centrifuge test was simulated by modeling
volume loss in plane strain condition. However, stress transfer in the longitudinal
direction due to advancing tunnel face was not simulated. In addition, the influence zone
of tunnel advancing and critical conditions on the existing tunnel with respect to the
location of tunnel face cannot be modeled in plane strain condition.
(d) The effects of volume loss only were simulated. Nevertheless, the effects of soil removal
inside a tunnel (i.e., the effects of weight loss) also influenced the shape and magnitude
of ground displacement above the tunnel. In addition, stress change due to weight loss
should also be considered and modeled.
(e) Numerical analyses of tunnel-tunnel interaction from some previous studies were carried
out using a simple elastic-perfectly plastic constitutive model. As a result, important
aspects of soil behavior for tunneling such as change in shear modulus due to incremental
17
stress, induced shear strain and mobilization of stiffness were not modeled.
To improve the fundamental understanding of crossing-tunnel interaction, centrifuge
modeling was adopted to investigate the effects of new tunnel excavation on an existing
tunnel. Three-dimensional tunnel advancement in-flight was carried out in centrifuge to
model stress transfer in the longitudinal direction of the advancing tunnel. Both effects of
volume and weight losses were simulated.
Three-dimensional numerical back-analyses were performed using an advance constitutive
model such as hypoplasticity, which is capable of simulating stiffness dependency on state,
strain and recent stress history. The purpose of using numerical back-analysis was to further
understand the stress transfer, induced strain and mobilization of stiffness in the interaction of
crossing tunnels.
18
Table 2.1 Summary of case studies

Heathrow Northbound Cut-and-cover Shekou Line
Express Tunnels Thameslink tunnel above Tunnels
underneath 2000 Tunnel Metro Line 1 underneath
Piccadilly Line underneath Tunnels Luobao Line
Tunnels Thameslink Tunnel
Tunnel
(Cooper et al., (Mohamad et (Liu et al., (Li and Yuan,
2002) al., 2010) 2011) 2012)
Location London London Nanjing Shenzhen
Soil type London Clay London Clay Silty clay Highly
decomposed
granite
Estimated K0 at the 1.70(a) 1.90(a) 0.53 0.40(b)
depth of the axis of
existing tunnel
Outer dimension of D = 4.1 D = 8.5 D = 6.8 W = 6.8 x
existing tunnel, H = 13.6
(m)
Outer dimension of D = 9.1 D = 6.5 W = 11.2 x D = 6.3
new tunnel, (m) H = 6.2
Cover depth of 11.0 5.5 10.2 15.0
existing tunnel,
(m)
Cover depth of new 21.5 17.0 2.0 30
tunnel, (m)
Pillar depth, (m) 7.0 3.6 2.0 2.0
Angle of tunnel 69 21 70 55
crossing, (degrees)
Tunnel excavation Pilot shield with Open shield Cut-and-cover EPB shield
method tunnel
enlargement
Volume loss reported 1.3 2.5 Not available Not available Not available
(%)
Note: (a) Estimated from Hight et al. (2007)
(b) Adopted from Viana da Fonseca et al. (1997)
(d) D is diameter of tunnel; W is diameter of tunnel; H is height of tunnel
19
Figure 2.1 Relationship between settlement trough width parameter (i) and depth of tunnel (z0)
for different ground conditions (Peck, 1969)
20
Figure 2.2 Twin tunnel excavation in side-by-side and vertically stacked arrangement in
Bangkok (Sirivachiraporn and Phienwej, 2012)
21
Figure 2.3 Finite element mesh to investigate stress transfer mechanism in an open face
tunnel (Ng and Lee, 2005)
22
Figure 2.4 Shear stiffness degradation curve with shear strain and typical shear strain range
for retaining walls, foundations and tunnels (Mair, 1993)
23
Figure 2.5 Shear strain induced due to tunnel excavation with and without presence of a
pipeline above in centrifuge test in plane strain condition (Marshall et al., 2010)
24
(a)
(b)
Figure 2.6 Heathrow Express Tunnels underneath Piccadilly Line Tunnels (a) plan; (b)
section (Cooper et al., 2002)
25
(a)
(b)
Figure 2.7 Northbound Thameslink 2000 Tunnel underneath Thameslink Tunnel (a) plan; (b)
section (Mohamad et al., 2010)
26
Figure 2.8 Cut-and-cover tunnel above Metro Line 1 Tunnels (a) plan; (b) section (Liu et al.,
2011)
27
(a)
(b)
Figure 2.9 Shekou Line Tunnels underneath Luobao Line Tunnel (a) plan; (b) section (Li and
Yuan, 2012)
28
Figure 2.10 Laboratory 1g model test of crossing tunnels in clay (Kim et al., 1998)
29
Figure 2.11 Tunneling underneath a pipeline in sand (Vorster et al., 2005)
30
(a)
(b)
Figure 2.12 Effects of pillar depth on (a) displacement (b) deformation of the previously
constructed tunnel (Addenbrooke and Potts, 2001)
31
Figure 2.13 Finite element mesh to investigate the effects of construction sequence of
crossing tunnel in rock (Liu et al., 2009)
32
(a)
(b)
Figure 2.14 (a) Finite element mesh to investigate the effects of spacing of new twin tunnels
on an existing tunnel; (b) details of a new tunnel (Chakeri et al., 2011)
33
Chapter 3 Three-dimensional centrifuge modeling
3.1 Introduction
From the literature review in Chapter 2, the short coming of knowledge of multi
crossing-tunnel interaction is demonstrated. Thus, the main objective of this research is to
improve the understanding of the interaction of crossing-tunnels. Two methodologies,
centrifuge modeling and numerical analysis, were adopted.
This chapter presents details of centrifuge testing in this research. Objective of each test to
investigate each factor influencing the interaction of crossing tunnels is given. A novel
modeling technique called Donut to simulate tunnel excavation in centrifuge is introduced.
Details of instrumentation, model preparation and testing procedures are described.
3.2 Modeling principles and scaling laws
As soil behavior is governed by effective stress, a prototype stress condition can be recreated
by enhancing a body force of a model which greatly reduced in size. This enhancing body
force of the model can be achieved by applying a centrifugal acceleration on the model, using
a geotechnical centrifuge. When a centrifuge rotates at a constant angular velocity () with
radius (r) about the axis of the centrifuge, centripetal acceleration (r2) is provided on the
model. The acceleration can be represented by Ng, where N is an integer to describe the
acceleration and g is gravitational acceleration of the Earth (9.81 m/s2). Following is an
example of a scaling law for length (depth).
When a soil in a model with a density of is subjected to an acceleration of Ng, the vertical
34
stress of the soil model vm at a depth hm (in model scale) is equal to:
vm=Ng hm 3-1
For the prototype scale, the vertical stress vp at depth hp equals:
vp=g hp 3-2
As the vertical stress of soil in prototype scale equals to that in the model scale (vm =
vpfollowing scale law for length can be derived as follows:
Ng hm = g hp 3-3
or hp / hm = N 3-4
The centrifugal acceleration used in this study was 60 times that of the earth. Further details
of scaling laws and centrifuge applications are also reported by Taylor (1995a) and Ng (2014).
The relevant centrifuge scaling laws are summarized in Table 3.1.
3.3 Test setup
3.3.1 The geotechnical centrifuge facility
The centrifuge tests were carried out in a geotechnical centrifuge located at the Hong Kong
University of Science and Technology (Ng et al., 2001, 2002). The capacity of the centrifuge
in HKUST compared with other major centrifuge in the world is shown in Figure 3.1. This 8.4
m diameter centrifuge is a beam type with the maximum capacity of 400 g-ton. For a test in
static conditions, the centrifuge is able to accommodate a model with dimension up to 1.5 m
(width) x 1.5 (length) x 1.0 m (height). The centrifuge is driven by a 350 hp vector drive
variable speed AC motor and the maximum acceleration is up to 150 g.
35
3.3.2 Typical setup of centrifuge model package
Figure 3.2a shows a typical plan view of centrifuge tests to investigate multi-tunnel
interaction. A soil model with the width, length and height of 1250, 930 and 750 mm,
respectively was prepared for each test. Each model tunnel was made from an aluminum alloy
tube. The outer diameter (D) and the lining thickness were 100 mm and 3 mm, respectively,
equivalent to 6 m and 180 mm in prototype scale. The scaling laws for the flexural stiffness of
the lining per unit width and the flexural stiffness of the whole model tunnel are 1/N3 and
1/N4, respectively (see Table 3.1). For tunnel diameter, the scaling law is 1/N. By assuming
that the compressive strength of concrete (fc) is 50 MPa, Youngs modulus (Ec) is estimated
to be 33 GPa (ACI, 2011). Thus, the tunnel lining thicknesses are equivalent to 230 mm and
420 mm in prototype scale in the transverse and longitudinal directions of the existing tunnel,
respectively. The existing tunnel was modeled as wished-in-place and each end was closed
to keep soil out. The two ends of the existing tunnels were not connected to the model box
and no additional fixity was imposed. Thus, the existing tunnel was not modeled as a
continuous tunnel.
Ye et al. (2014) carried out model testing by applying a point load on tunnel linings with and
without radial (transverse) joints. They reported that the tunnel linings with joints deformed
larger but induced smaller transverse bending moment than that without joints. By adopting
their finding, it is possible that the deformation of the existing tunnel in this study (without
joints) may be underestimated that in the field. On the other hand, induced strains in the
transverse direction of the existing tunnel in this study should be larger than that of jointed
tunnel linings and hence more conservative.
The study of pipe-soil interaction (Klar et al., 2007) suggested that when R (EtI / Esri3) is
larger than 5, the pipe is considered rigid. By adopting their finding, the tunnel in this study is
considered as a rigid tunnel as the ratio of tunnel flexural stiffness to soil stiffness (R) is from
50 to 60.
36
Figure 3.2b shows a typical elevation view of a reference test (E2N3). The test identity
denotes an approximate cover-to-diameter ratio (C/D) of each tunnel while E and N
denotes existing and new tunnels, respectively. In this test, the C/D of the existing tunnel, the
C/D of the new tunnel and the pillar depth-to-diameter ratio (P/D) were 2, 3.5 and 0.5,
respectively. These correspond respectively to the cover depths of the existing tunnel and the
new tunnel of 12 m and 21 m in prototype, and the pillar distance of 3 m.
According to Jacobsz et al. (2004), the influence zone of tunneling in sand was found to be in
parabolic shape projecting 45 from the invert of the tunnel to the ground surface. By
adopting this finding, the influence zone of the new tunnel excavation was estimated to be
located within the boundary of the existing tunnel.
3.4 Experimental program
A total eight centrifuge tests were carried out in this research. The C/D of the existing tunnel,
C/D of the new tunnel, P/D between each tunnel and a brief description of each test are
summarized in Table 3.2.
The major objective of Test E2N3 (as shown in Figure 3.3) was to provide a reference test to
investigate the effects of a new tunnel excavation perpendicularly underneath an existing
tunnel. The geometry of each tunnel in this test was described in the previous section.
In Test E2N3-VW, the C/D of the existing tunnel, C/D of the new tunnel, P/D between each
tunnel were identical to that in Test E2N3 (refer to Figure 3.3). The objective of Test
E2N3-VW was to separately investigate to influence of volume and weight losses on
crossing-tunnel interaction. Unlike other tests where the effects of volume and weight losses
were model simultaneously, the effects of volume loss only was simulated first followed by
the effects of weight loss in Test E2N3-VW. Modeling sequence of this test is shown in Table
3.3.
37
Figure 3.4 shows plan and elevation views of Test E3N2 where the new tunnel was excavated
above the existing tunnel. The purpose of this test was to investigate the effects of
construction sequences of each tunnel. The result of Test E2N2 was compared with Test E2N3
where the new tunnel was excavated underneath the existing tunnel with the same P/D of 0.5.
In Test E3N5 (refers to Figure 3.5), the C/Ds of existing and new tunnels were increased by
1.5, compared with Test E2N3, while the P/D between the two tunnel was 0.5. The major
objective of Test E3N5 was to investigate the effects of increasing C/Ds of both tunnel
compared with the reference test (i.e. Test E2N3).
By comparing P/D of Test E2N5 with that of Test E2N3, the P/D of Test E2N5 (see Figure 3.6)
was increased from 0.5 to 2. The major objective of Test E2N5 was to investigate the effects
of pillar-to-depth ratio on the response of the existing tunnel.
Figure 3.7 shows details of Test E2,3N5 where a parallel existing tunnel was located between
the perpendicular tunnel and the new tunnel with P/D equaled to 0.5 between each tunnel. The
effects of shielding was studied by comparing Test E2N5 (Fig. 3.2b) and Test E2,3N5. In Test
E2,3N5, the lower existing tunnel was located above parallel to the new tunnel providing the
shielding effects to the upper existing tunnel.
A typical setup of Test E2N3,3 where model side-by-side twin tunnels advanced underneath
perpendicularly to an existing model tunnel is shown in Figure 3.8. The major objective of
this test was to investigate responses of the existing tunnel due to excavation of side-by-side
twin tunnels underneath. First, the left tunnel advanced in six excavation stages by 0.6D at a
time, followed by the right tunnel.
In Test E2N3,5 (refer to Figure 3.9), vertically stacked new twin tunnels were excavated
underneath an existing tunnel. The purpose of this study is the study the response of the
existing tunnel due to twin tunnel excavation in vertically stacked arrangement underneath. In
addition, the results in this test was also compared with that in side-by-side new twin tunnel
38
excavation beneath an existing tunnel (Test E2N3,3). For vertical new twin tunnels, the upper
new tunnel was excavated first followed by the lower new tunnel. The reason for this
excavation sequence is the lining of the two new tunnels were already embedded in the model
during model preparation (i.e., wished-in-place tunnel lining). Based on the literature, if the
lower new tunnel was excavated first, the effects of the lower new tunnel excavation on the
ground above is likely to be shielded be the lining of the upper new tunnel.
3.5 In-flight tunneling simulation technique
Figure 3.10a illustrates a novel modeling device, which is called the Donut, to simulate
tunnel advancement in a centrifuge test. A pair of rubber bags, one mounted outside (outer
rubber membrane) and the other mounted inside a model tunnel (inner rubber bag), were used
to simulate the effects of both volume and weight losses at each stage of excavation in the
centrifuge. The tunnel lining of 100 mm in diameter (or 6 m in prototype scale) was made of
aluminum alloy and its lining bending stiffness per unit width of 0.16 kN.m2/m (or 33.5
MN.m2/m in prototype scale) and thickness of 3 mm (or 180 mm in prototype scale) was
scaled properly.
The material of an outer rubber membrane and an inner rubber bag were silicone rubber. To
fabricate the outer and inner rubber bags, liquid silicone was prepare by mixing the base
silicone material with the catalyst as shown in Plate 3.1. The liquid mixture was placed inside
the vacuum chamber for about 20 minutes to minimize air bubbles. After that, the liquid
silicone was injected to a mold of each rubber bag (see Plate 3.2). The silicone was in the
mold for one day to harden, and then the mold was removed. Plate 3.3a shows each rubber
bag after removal from each mold. A total 12 pairs of rubber bags (6 pairs for each new tunnel)
were used in this research. A cross-section of the Donut consists of the outer rubber
membrane and inner rubber bag is shown in Plate 3.3b.
During the centrifuge model preparation, each rubber bag was filled with a heavy fluid (ZnCl2)
having a density similar to that of the soil sample or about 1530 kg/m3 to simulate the
39
presence of soil. Each outer rubber membrane was filled with a known amount of the heavy
fluid representing an equivalent percentage of volume loss, which in this study was 2%.
Volume loss was simulated by controlling the outflow of the heavy fluid from the outer rubber
membrane. This volume loss was considered as tunnel volume loss imposed by each Donut.
Likewise, each inner rubber bag was filled with the heavy fluid which was drained away at
different stages to simulate weight loss due to tunnel excavation in the centrifuge.
Tunnel simulation in this study was intended to mimic the effects of closed-face shield
tunneling. Mair and Taylor (1997) reported typical volume losses due to tunnel excavation
using earth pressure balance shields in sand and soft clay of up to 1% and 2%, respectively.
Shirlaw et al. (2003) and Abrams (2007) reported volume losses in mixed face tunneling
involving clay and sand of between 1 and 4%. Based on these reports, a volume loss of 2%
was adopted in this study.
Mair and Taylor (1997) reported ground movement caused by tunnel excavation consists of
ground movement due to tunnel face, passage of the shield, tail void, deflection of the lining
and consolidation. For a design purpose, it is common that the effects of ground movement
caused by the above components are defined as the tunnel volume loss. Standing and Burland
(2006) reported a volume loss of 2% was adopted in the design for tunnel.
Figure 3.10b and Plate 3.4a show the advancing sequence of the new model tunnel. Excavated
sections 1 to 6 were assembled to form the new tunnel. Both ends of the new tunnel were
closed to prevent the displacement of soil into the tunnel. The six advancing sections, each
representing a length of 0.6D or 3.6 m in prototype scale, were controlled independently
in-flight in a centrifuge test. Each rubber bag was connected to an outlet valve (see Plate 3.4b)
by a drainage tube. Each valve could be opened in-flight allowing outflow of the heavy fluid
which was collected in a reservoir.
To simulate effects of both volume and weight losses simultaneously, the two valves to which
the inner and outer bags in each section were connected were regulated to simulate the effects
40
of tunneling in-flight. To simulate only volume loss, only the valve to which the outer bag in
each section was connected was regulated, whereas the valve to which the inner bag in the
same section was connected was closed. The modeling sequence of effects volume loss and
weight loss separately is summarized in Table 3.3.
A new tunnel excavation beneath an existing tunnel is shown in Plate 3.5a while the new
tunnel construction above an existing tunnel is shown Plate 3.5b. The fourth excavation stage
was where the excavated section was closest to the existing tunnel. New twin tunnel
excavation in side-by-side and vertically stacked arrangements underneath the existing tunnel
are shown in Plates 3.6a and 3.6b, respectively.
3.6 Instrumentation
3.6.1 Types and location of instrumentation
Plate 3.7 illustrates the types and locations of instrumentation installed on the existing tunnel
to investigate responses of the existing tunnel in the longitudinal and transverse directions.
In the longitudinal direction of the existing tunnel, vertical displacement of the existing tunnel
was measured using linear variable differential transformers (LVDTs) connected to extension
rods fixed along the crown of the existing tunnel. The extension rods were encased in hollow
tubes to minimize friction with the surrounding soil. During the dry pluviation of sand, the
extension rods were temporary supported by a temporary beam which was removed after the
sand sample reached the desired height of 750 mm).
In the transverse direction of the existing tunnel, Figure 3.11a and Plate 3.8a show a section
view inside the section of the existing tunnel directly above the new tunnel. Tunnel
deformation was measured using four potentiometers installed at the crown, at each springline
and at the invert to record changes in the horizontal and vertical diameters of the existing
tunnel. A linear potentiometer is a variable resistor connected to three leads. The first two
leads are connected to the ends of the resistor, so the resistance between them is fixed. The
41
third lead is connected to a slider that travels along the resistor varying the resistance between
it and the other two connections. Changes in resistance in a linear potentiometer are linearly
proportional to the travel distance of the slider (Todd, 1975). In this study, the accuracy of a
potentiometer about 1 mm in prototype scale, achieved by considering the fluctuation of data
before the start of the new tunnel excavation. The plate for potentiometers was fixed on a
frame that was connected to the tunnel lining at a distance of 3D and 5D away from the center
of the existing tunnel as shown in Plate 3.8b.
Figure 3.11b shows longitudinal section view of the existing tunnel. To measure strain in the
longitudinal direction of the existing tunnel, 19 sets of strain gages or longitudinal bending
moment transducers were installed along the crown and invert of the existing tunnel. Full
Wheatstone bridge semiconductor strain gages having a gage factor of 140 were used to
compensate for temperature effects. In addition to measuring the deformation of the existing
tunnel, eight sets of strain gages were installed evenly at an interval of 45 around the tunnel
circumference to measure transverse bending moment. Full Wheatstone bridge foil strain
gages having a gage factor of 2 were used instead of the semiconductor type simply because it
was not possible to mount the latter inside the model tunnel.
In Tests E2N5 and E2,3N5 (refer to Plate 3.9), the subsurface ground settlements were
analyzed along with settlement of the lower existing tunnel. Subsurface ground settlement in
Test E2N5 was measured using extension rod for LVDT connected with a 20 mm diameter
subsurface settlement plate. The location of extension rods in Test E2,3N5 placed on the
lower existing tunnel was the same as that in Test E2N5. The purpose of this analysis of
subsurface and tunnel settlement is to investigate the effects of shielding provided by the
lower existing tunnel on soil displacement.
In addition, in Test E2,3N5 (see Plate 3.9b), potentiometers and strain gages in the transverse
direction were also installed in the lower existing tunnel at the location directly below the
upper existing tunnel. Arrangement of the potentiometers and strain gages in the lower
existing tunnel was the same as that in the upper existing tunnel (refer to Plate 3.8a).
42
3.6.2 Calibration of transducers
After the strain gage were attached on the existing tunnel, calibration of the strain gages was
carried out as shown in Plate 3.10a. To calibrate the strain gage, the existing tunnel was
placed on two supports, which were considered as a roller. A dead load was applied using
circular steel plates with known weight at the center of the existing tunnel. Reading of each
transducer was recorded by data logger. To obtain the calibration factor of each transducer, the
reading was correlated with the computed bending moment on the tunnel lining. The
geometry of the tunnel, roller support and loading conditions in numerical analysis were
identical to that in the calibration. The tunnel lining was modeled using four-node shell
element. Calibration factor of each transducer is summarized in Appendix B.
Plate 3.10b shows calibration of an LVDT using a digital vernier caliper. The accuracy of the
caliper is within 0.01 mm. The calibration factor of each LVDT was obtained by relating the
reading of LVDT from data logger to travel distance from the caliper. Likewise, calibration of
potentiometer used the same method as the LVDT (see Plate 3.10c).
3.7 Model preparation
Heavy fluid was filled in the Donuts in the new tunnel in each test as shown in Plate 3.11a.
Temporary support (refer to Plates 3.6b and 3.7b) was used to place the new tunnel in the
specified location.
Dry silica Toyoura sand was used in each centrifuge test. The average particle size (D50),
maximum void ratio (emax), minimum void ratio (emin), specific gravity (Gs) and critical state
internal friction angle (c) of Toyoura sand are 0.17 mm, 0.977, 0.597, 2.64 and 30,
respectively (Ishihara, 1993). Details of properties of Toyoura sand is summarized in Table
3.4. Grain size effects on soil-tunnel interaction were considered insignificant when the ratio
of tunnel diameter to average particle size was larger than 175 (Garnier et al., 2007). In this
study, the ratio of model tunnel diameter (100 mm) to average particle size (0.17 mm) was
43
588.
A dry pluviation technique was adopted to prepare the soil sample in each test as shown in
Plate 3.11b. A drop height of 500 mm and a pluviation rate of about 100 kg/hour were used to
control the density of the soil sample. The density achieved in every test was ranging between
1529 and 1535 kg/m3, equivalent to relative density of 64% to 66%. The relative density of
sand in each test is summarized in Table 3.2. According to a study on the homogeneity of
pluviated sand samples in centrifuge tests, variations in dry density was estimated to be within
0.5% or 8 kg/m3 (Garnier, 2001). It should be noted that the density of sand around each
tunnel could be slightly looser than that located further away from the tunnel. However, the
density variation with depth was still within the recommended range given by Garnier (2001)
as shown in Figure 3.12.
Each existing tunnel was placed after the level of pluviated sand reached the designed height
(refer to Plates 3.7 and 3.9b). By using some thin wires and a temporary structural beam
above, the new tunnel was wished-in-place in position (see Plate 3.11b). These wires and
the beam were removed after the pluvial deposition reached the bottom of the new tunnel to
support it.
After completion of sand pluviation, the model package was transferred to the centrifuge
platform. LVDTs were calibrated in each test, using the procedure as discussed previously.
The LVDTs were mounted on a structural frame as shown in Plate 3.12a. The connector for
each transducer was plugged in each channel in the data logger as shown in Plate 3.12b.
Appropriate gain factor was applied to amplify the reading from each transducer. A connector
for each valve was also plugged in the control panel for valves (see Plate 3.4b) to control the
outflow of heavy fluid in the Donut in-flight (see Plates 3.5 and 3.6).
In each test, video camera was installed on the model package as shown in Plate 3.13a to
observe the overall view during the test in-flight. To balance the centrifugal force from each
arm of the centrifuge, counter weight made of steel was installed on the other platform as
44
shown in Plate 3.13b. The selection of counter weight was based on the calculation of the
moment caused by the model package on the axis.
3.8 Test procedure
After the centrifuge model package was prepared and all transducers calibrated in 1g as
discussed previously. The centrifuge was gradually spun up to a nominal gravitational
acceleration of 60g. Before commencing new tunnel advancement, sufficient time was
allowed to ensure that there was no further ground surface settlement. An initial reading was
taken from each transducer at this stage.
Subsequently, the six stages of tunnel excavation was carried out in-flight according to the
corresponding modeling sequence (only for Test E2N3-VW refer to Table 3.3) and tunneling
sequence (see Figure 3.8 and Figure 3.9 for Tests E2N3,3 and E2N3,5, respectively).
Sufficient time was provided to allow all the transducer readings had stabilized before each
excavated section was advanced to the next stage. After completion of tunnel advancement,
the centrifuge was spun down. Controlling of the centrifuge, in-flight tunnel excavation and
data acquisition was carried out using web-based application developed at the Geotechnical
Centrifuge Facility at the Hong Kong University of Science and Technology.
45
Table 3.1 Scaling laws applied in the centrifuge tests

Scaling law
Parameter Unit (model/prototype)
Length m 1/N
Area m2 1/N2
Volume m3 1/N3
Density kg/m3 1
Unit weight N/m3 N
Flexural stiffness per unit width Nm2/m 1/N3
Flexural stiffness Nm2 1/N4
Stress N/m2 1
Strain - 1
46
Table 3.2 Summary of centrifuge tests

Test ID C/D C/D P/D Density Dr(a) Brief description of each test
(existing) (new) (kg/m3) (%)
E2N3 2.0 3.0 0.5 1529 64.3 A new tunnel excavation

underneath an existing tunnel
(Reference test)
E2N3-VW 2.0 3.0 0.5 1531 64.6 Simulates effects of volume and
weight losses separately
E3N2 3.0 2.0 0.5 1534 65.7 Change a construction sequence
(A new tunnel construction
above an existing tunnel)
E3N5 3.5 5.0 0.5 1529 64.3 Increase C/D of both tunnels
E2N5 2.0 5.0 2.0 1532 65.0 Increase P/D from 0.5 to 2
E2,3N5 2, 3.5 5.0 0.5 1535 66.0 Provide effects of shielding by

the lower existing tunnel
E2N3,3 2.0 3.5, 0.5 1529 64.3 Side-by-side twin tunnel
3.5 excavation underneath an
existing tunnel
E2N3,5 2.0 3.5, 0.5 1535 66.0 Vertically stacked twin tunnel
5.0 excavation underneath an
existing tunnel
Note: (a) Dr denotes relative density
47
Table 3.3 Modeling sequences of new tunnel advancement to separately investigate effects of
volume (V) and weight (W) losses
Modeling sequences
VL WL E2N3 E2N3-VW
V1+W1 V1
V6 W6 V2+W2 V2
V5 W5 V3+W3 V3
Existing
V4 W4 tunnel V4+W4 V4 E2N3-V
V3 W3 V5+W5 V5
V2 W2 V6+W6 V6
V1 W1 W1
W2
New W3
tunnel W4 E2N3-VW
W5
W6
48
Table 3.4 Properties of Toyoura sand

Parameter Value
Mean diameter, D50 (mm) 0.17
Uniformity coefficient, Uc 1.7
Maximum void ratio, emax 0.977
Minimum void ratio, emin 0.597
Specific gravity, Gs 2.65
Effective angle of friction at critical (') 30o
49
Figure 3.1 Comparison of major geotechnical centrifuge facilities in the world (after Ng, 2014)
50
1250
Drainage tubes for heavy fluid

Reservoir
Valves
Tunnel
Y
excavation
sequence
6
5 X
4
930 3
Existing tunnel 2
diameter (D) 1
= 100 mm
(6.0 m at 60g) New tunnel
D = 100 mm
(6.0 m at 60g) Note: Dimension in mm
(model scale)
(a)
C = Cover depth
Existing tunnel
750 P = Pillar depth
New tunnel
Test ID: E2N3
(b)
Figure 3.2 Typical setup of centrifuge model package in a reference test (a) plan view; (b)
elevation view

51
1250
New tunnel
excavation Location of LVDT
sequence on existing tunnel
6
5
4
930 3
Existing tunnel 2
diameter (D) 1
= 100 mm
New tunnel
D = 100 mm
Note: Dimension in mm
(model scale)
(a)
LVDT
C = 200
(C/D = 2)
Existing tunnel
750 P = 50
Extension rod
New tunnel
for LVDT
(P/D = 0.5)
E2N3
(b)
Figure 3.3 (a) Plan view and; (b) elevation view of Test E2N3 (reference test)
52
1250
New tunnel
6
5
4
930 3
Existing tunnel 2
diameter (D) 1
= 100 mm
New tunnel
D = 100 mm
(model scale)
(a)
LVDT
C = 200
(C/D = 2)
P = 50
750 (P/D = 0.5)
Existing tunnel Extension rod
New tunnel
E3N2
(b)
Figure 3.4 (a) Plan view and; (b) elevation view of Test E3N2
53
1250
New tunnel
6
5
4
930 3
Existing tunnel 2
diameter (D) 1
= 100 mm
New tunnel
D = 100 mm
(model scale)
(a)
LVDT
C = 350
(C/D = 3.5)
750
P = 50
(P/D = 0.5)
New tunnel
E3N5
(b)
54
1250
New tunnel
6
5
4
930 3
Existing tunnel 2
diameter (D) 1
= 100 mm
New tunnel
D = 100 mm
(model scale)
(a)
LVDT
C = 200
(C/D = 2)
Existing tunnel
750
Extension rod P = 200
(P/D = 2)
New tunnel
E2N5
(b)
55
1250
New tunnel
6
5
4
930 3
Existing tunnel 2
diameter (D) 1
= 100 mm
New tunnel
D = 100 mm
(model scale)
(a)
LVDT
C = 200
(C/D = 2)
Upper existing tunnel
750
Lower existing Extension rod P = 200
tunnel (P/D = 2)
New tunnel
E2,3N5
(b)
Figure 3.7 (a) Plan view and; (b) elevation view of Test E2,3N5
56
1250
Tunnel diameter
100 mm Location of LVDT
Tunnel
on existing tunnel
advancing
sequence 6L 6R
5L 5R
4L 4R
930 3L 3R
Existing 2L 2R
tunnel 1L 1R
Left new Right new

tunnel (1 st ) tunnel (2 nd ) Note: Dimension in mm
(model scale)
(a)
LVDT
C = 200
(C/D = 2)
750 P = 50
Left new Right new (P/D = 0.5)
tunnel (1st ) tunnel (2 nd )
0.5D
(b)
Figure 3.8 (a) Plan view and; (b) elevation view of Test E2N3,3
57
1250
New tunnel
6
5
4
930 3
Existing tunnel 2
diameter (D) 1
= 100 mm Upper and lower
new tunnel
D = 100 mm
(model scale)
(a)
LVDT
C = 200
(C/D = 2)
750 P = 50
Upper new (P/D = 0.5)
tunnel (1st )
0.5D
Lower new
tunnel (2 nd )
E2N3,5
(b)
Figure 3.9 (a) Plan view and; (b) elevation view of Test E2N3,5
58
Outer rubber membrane Excavation

(controlling volume loss, VL = 2%) length = 60 mm
Aluminum alloy tube

(tunnel lining)
outer diameter = 100 mm
Heavy fluid thickness =3 mm
(ZnCl2)
Inner rubber bag
(controlling weight loss,WL )
Drainage tubes for heavy fluid
Transverse section Longitudinal section

(a) (b)
Figure 3.10 Details of Donut simulating effects of volume and weight losses in (a)
transverse section; (b) longitudinal section view
59
Strain gage in the transverse

direction of the existing tunnel
Potentiometer
Plate for potentiometer
Tunnel lining
(a)
Center of the Strain gage in the longitudinal

existing tunnel direction of the existing tunnel
Frame for potentiometer
Potentiometer
3D 2D
(b)
Figure 3.11 (a) Instrumentation in the transverse cross-section view and; (b) the longitudinal
sectional view of the existing tunnel
60
Top of model box

750
Height above the base of model box (mm)
E2N3
600 E2N5
Existing tunnel E2,3N5
450
New tunnel [E2N3]

Lower existing tunnel
300 [E2,3N5]
New tunnel
[E2N5, E2,3N5]
150
d < 8 kg/m3
(Garnier et al., 2007)
0
60 62 64 66 68 70
Relative density (%)
Figure 3.12 Relative density of sand with height of pluviation in each test
61
Liquid Catalyst
silicone
rubber
Plate 3.1 Liquid silicone rubber and catalyst
62
Injection of liquid
silicone rubber
(a)
Injection of liquid
silicone rubber
(b)
Plate 3.2 Mold for fabrication of (a) outer rubber membrane; (b) inner rubber bag
63
Outer rubber membrane Inner rubber bag
(a)
Outer membrane
(Volume loss of 2%)
Heavy fluid
Inner rubber bag (ZnCl2),
(Weight loss) = 1500 kg/m3
Tunnel lining
(Aluminum alloy)
Tunnel diameter = 100 mm
lining thickness = 3 mm
in model scale
WL VL
(b)
Plate 3.3 Details of (a) outer rubber membrane and inner rubber bag; (b) Donut simulating
effects of volume and weight losses
64
Partition
New tunnel
Drainage tube
for heavy fluid
(a)
Reservoir for Drainage tube

heavy fluid for heavy fluid
Tube for compressed air Valve controlling

to open and close a valve outflow of heavy fluid
(b)
Plate 3.4 Simulation of tunnel excavation in-flight (a) new tunnel with tubes for heavy fluid;
(b) valves controlling outflow of heavy fluid
65
Excavation
Donut sequence
6
6 Excavation
5 sequence
Location of 5
existing tunnel 4
4
3 3
2 2
1 Existing 1
tunnel
0.6D or 3.6 m
in prototype
scale Outer diameter
= 100 mm
New Lining thickness New
tunnel = 3 mm tunnel
(a) (b)
Plate 3.5 A new tunnel (a) excavation underneath an existing tunnel; (b) construction above
an existing tunnel
66
Donut
6L 6R
5L 5R
Location of the
4L 4R existing tunnel
3L 3R
2L 2R
Length of
1L 1R excavation = 0.6D
(3.6 m in prototype
scale)
Left tunnel 0.5D Right tunnel
(1st) (2nd)
(a)
1U 2U 3U 4U 5U 6U
Upper new
tunnel (1st)
0.5D
Lower new
tunnel (2nd)
1L 2L 3L 4L 5L 6L
Location of the
existing tunnel
(b)
Plate 3.6 Twin new tunnel excavation in (a) side-by-side arrangement; (b) vertically stacked
arrangement
67
Chapter 3 Centrifuge modeling
Temporary beam to support extension rods
Existing tunnel
Extension rod
Strain gage in the longitudinal Location of for LVDT
direction of the existing tunnel new tunnel
(a)
Strain gage in the longitudinal

Existing tunnel
Extension rod
for LVDT New tunnel
(b)
Plate 3.7 Typical instrumentation on the existing tunnel where the new tunnel excavate (a)
underneath and; (b) above
68
Strain gage in the transverse

Potentiometer
(a)
Plate for Support for frame

potentiometer connecting to tunnel lining
(b)
Plate 3.8 (a) Potentiometer and strain gage in the transverse direction of the existing tunnel;
(b) frame and plate for potentiometers
69
Existing tunnel
Extension for New tunnel

LVDT
E2N5
Subsurface
settlement plate
(a)
Upper existing
tunnel
Extension for Lower existing
LVDT tunnel
New tunnel
E2,3N5
Lower existing
tunnel Location of strain gage
and potentiometer
(b)
Plate 3.9 Measurement of (a) subsurface settlement in Test E2N5; (b) settlement of the lower
existing tunnel in Test E2,3N5
70
Existing tunnel Applied dead load
(a)
LVDT
Digital vernier
caliper
(b)
Digital vernier
caliper
Potentiometer
(c)
Plate 3.10 Calibration of (a) strain gage on the existing tunnel; (b) LVDT; (c) potentiometer
71
New tunnel
(a)
Drop height
= 500 mm
(b)
Plate 3.11 (a) Placing new tunnel in the strong box; (b) drop height of sand using pluviation
technique
72
LVDT measuring ground

surface and tunnel settlement
Frame supporting LVDT
(a)
Data logger for

transducers
Control panel
for valves
(b)
Plate 3.12 (a) LVDT measuring ground surface and tunnel displacement; (b) control panel for
valves and data logger for transducer
73
VDO camera
Existing tunnel
New tunnel
Centrifuge platform
(a)
Counter Model
weight package
(b)
Plate 3.13 (a) A typical setup of a centrifuge model package; (b) a model package on the
centrifuge platform
74
Chapter 4 Three-dimensional numerical analysis
4.1 Introduction
Details of centrifuge modeling of crossing-tunnel interaction were described in Chapter 3.

Three-dimensional numerical back-analysis of each centrifuge test was carried out to improve
the understanding of stress transfer, strain induced and stiffness mobilized in the interaction of
crossing tunnels. In addition, numerical parametric study was conducted to investigate the
sensitivity of some factors influencing the interaction of crossing-tunnels such as excavation
sequence in new twin tunnels, volume loss, relative density and tunneling in saturated sand.
In this chapter, details of numerical runs, finite element mesh, boundary conditions,
constitutive model, determination of model parameters and modeling procedures are
presented.
4.2 Detail of numerical runs
Numerical back-analyses and parametric studies were carried out in this research. Brief
information of the finite element program and details of each type of analysis are given as
follows:
4.2.1 Finite element program
A commercial finite element program PLAXIS 3D 2013 (Brinkgreve et al., 2013) was
adopted to back-analyze centrifuge test results. PLAXIS 3D has been developed for analysis
of deformation and stability in geotechnical problem. Some features in the program allow for
computation of tunnel excavation and increasing centrifugal acceleration in soil-structural
75
interaction problem. In addition, user defined constitutive model can be implemented in the
program providing user to use an advance model.
4.2.2 Numerical back-analysis
A total eight numerical runs were carried out to back-analyze the centrifuge tests. The
dimension of the mesh, relative density of sand and geometry of tunnels were identical to that
in each centrifuge test. The numerical modeling procedures basically followed that in the
centrifuge test.
4.2.3 Numerical parametric study
Apart from numerical back-analysis, numerical parametric study was conducted to further
investigate some factors that influence the interaction of crossing tunnels. Summary of
numerical parametric study is given as follows:
4.2.3.1. Tunnel excavation in greenfield conditions
The objective of numerical parametric study of tunnel excavation in greenfield conditions is

to improve the understanding of soil displacement and stress change without the presence of
the existing tunnel. The results were interpreted along with the measured results of crossing
tunnels to explain the shielding or stiffening effects provided by the existing tunnel on the
surrounding soil. The results of tunnel in greenfield conditions are discussed in Chapter 5.
4.2.3.2. Influence of tunnel arrangement and excavation sequence in new twin tunnel
excavation on the response of an existing tunnel
Study of non-wished-in-place lining and construction sequence of new twin tunnels was
carried out. The purpose of this analysis is to identify the influence of tunnel lining and
excavation of new twin tunnels on an existing tunnel, which located above. Details of
numerical runs to study tunneling sequence are summarized in Table 4.1. The results of
76
numerical parametric study on arrangement and excavation sequence of tunnel using

non-wished-in-place lining activation of the new tunnel are given in Chapter 8.
4.2.3.3. Influence of volume loss, relative density and tunneling in saturated sand on
crossing-tunnel interaction
Numerical parametric study of effects of volume loss, relative density and tunneling in
saturated sand on tunnel-tunnel interaction was carried out. The objective of numerical
parametric study is to examine the sensitivity of these three factors on the response of the
existing tunnel due to the new tunnel excavation. Summary of numerical run in the parametric
study in is given in Table 4.2. The results of the numerical investigation of these three factors
are discussed and explained in Chapter 9.
4.3 Finite element mesh and boundary conditions
Figure 4.1a shows the three-dimensional finite element mesh for a reference case (E2N3). The
mesh had dimensions of 625 mm x 930 mm x 750 mm in model scale. Owing to symmetry,
only half of the entire mesh was required except for the test of side-by-side new twin tunnel
excavation beneath an existing tunnel (case E2N3,3). Thus, a plane of symmetry was defined
at X/D = 0 in each case except in case E2N3,3.
In case E2N3,3 (see Figure 4.1b), symmetry conditions was not applicable as each new tunnel
was excavated one after the other. Unlike other numerical run, the entire mesh in case E2N3,3
was generated. Boundary conditions adopted in the finite element analysis in every case were
roller support on the four vertical sides and pin support at the base of the mesh. The soil was
modeled using a 10-node tetrahedral element.
Figure 4.2a shows some details of the two perpendicularly crossing tunnels in a reference case
(E2N3). The existing tunnel and the lining of the new tunnel were modeled as
wished-in-place by activating the tunnel lining and deactivating soil elements inside the
77
tunnel in the initial stage. An additional constraint was adopted for the tunnel lining at the
plane of symmetry. Each edge of the tunnel lining at the plane of symmetry was allowed
neither translational movement in the X direction nor rotation around the Y and Z axes
(i.e., ux, y and z = 0). For case E2N3,3, the plane of symmetry was not applied, thus,
additional constraint was not required. The tunnel lining in every case was modeled using a
6-node elastic plate element.
Activating sequence of tunnel lining of the existing and new tunnels is shown in Figure 4.2b.
For back-analysis case, lining of the existing and new tunnels was activated since the initial
stage. In parametric study of excavation sequence of new twin tunnels on an existing tunnel,
only the lining of the existing tunnel was activated in the initial stage. The lining of the new
tunnel was activated in each excavation stage one after another.
4.4 Constitutive model
4.4.1 Description of hypoplasticity constitutive model
Toyoura sand was represented using an open-source hypoplasticity model implementation

with intergranular strain or small strain stiffness (Gudehus et al., 2008). The constitutive
model models have been developed to describe the non-linear response of coarse-grained soil
(Kolymbas, 1991; Gudehus, 1996; von Wolffersdorff, 1996; Wu et al., 1996; Gudehus and
Man, 2009). The hypoplasticity model differs from elasto-plasticity in that it does not
decompose strain into elastic and plastic parts. Nevertheless, the model can still predict
fundamental features of soil behaviour such as non-linearity, irreversibility and failure. To
model effects of small-strain-stiffness in hypoplastic constitutive model, intergranular strain
concept (small strain stiffness) was proposed by Niemunis and Herle (1997). In summary,
hypoplasticity constitutive model with intergranular strain concept is a non-linear constitutive
model that considered stiffness dependent on state, strain level and recent stress history.
The governing equation of hypoplastic constitutive model is given in Equation 4-1 based von
Wolffersdorff (1996) and Niemunis and Herle (1997):
78
0
f s ( L : D f d N || D ||) 4-1
Linear response Non-linear response

where
0
: Stress rate tensor (i.e. equivalent to )
D : Deformation (strain) rate tensor (i.e. equivalent to )
L : Linear part of hypoplastic stiffness matrix
N : Non-linear part of hypoplastic stiffness matrix
fs : Factor controlling responses on mean stress level
fd : Factor controlling response on void ratio
The model parameters are divided into two major groups, which are parameters for large
strain behavior and intergranular strain concept (small strain stiffness) parameters.
Parameters for large strain behavior mainly control stiffness dependency on stress level and
void ratio (von Wolffersdorff, 1996). Figure 4.3 shows limiting curves for void ratio and mean
stress. For parameter in large strain behavior, the prediction of stiffness is dependent on state
and direction of loading. However, if only these parameters are adopted without small strain
stiffness parameters, the model appears to overestimate strain at the small strain range and
during change of loading direction. As a result, the model tends to predict excessive soil
displacement (Niemunis and Herle, 1997).
Figure 4.4 shows the effects stiffness degradation and recent stress history from element test
in reconstituted London Clay (Atkinson et al., 1990). Four stress paths were adopted in their
study (refer to Figure 4.4a). It is illustrated in Figure 4.4b that for every stress path, shear
stiffness reduced with increasing shear strain. The largest stiffness occurred in the test
following the path DOX, where 180 degrees stress reversal was carried out. With a 90 degrees
changing of loading direction (path AOX and COX), stiffness was smaller than 180 degrees
79
stress reversal. When loading path continued in the same direction (i.e. path BOX), the
stiffness was the smallest.
By considering small strain stiffness, Niemunis and Herle (1997) proposed intergranular strain
concept that considered stiffness dependent on strain and recent stress history. The
intergranular strain concept was incorporated into basic hypoplastic constitutive model to
improve the prediction related to small strain stiffness.
Idealized response envelope of intergranular strain concept proposed by Niemunis and Herle
(1997) is shown in Figure 4.5. The abscissa and ordinate denotes strain or stress rate in
horizontal and vertical direction, respectively. Parameter R denotes size of elastic range. For
normalized length of intergranular strain tensor (= 0 or = 1) with strain path reversal,
deformation is elastic (i.e., mRL). For = 1 with 90 rotation of strain path, deformation is
elastic, but with lower stiffness (i.e., mTL). For = 1 without strain path reversal, deformation
is followed large strain behavior in hypoplasticity model (LN).
4.4.2 Advantage of hypoplasticity model
Like many other advanced soil models, the hypoplasticity constitutive model is able to
capture soil behavior related to stress state and history. More importantly, it is capable of
modelling strain-dependency and path-dependency of soil stiffness at small strains (Niemunis
and Herle, 1997).
Man (2009) and Svoboda et al. (2010) adopted hypoplasticity model with small strain
stiffness to predict ground movement due to a single tunnel excavation in two case histories.
They found that the computed ground displacement were in a reasonable agreement with that
from field monitoring in each case.
Stiffness of soil mobilized by the effects of tunneling includes change in stresses, induced
strain and change of stress path. Thus, it is necessary to adopt a constitutive model that
reasonably simulated the stiffness dependency on stress, strain and path. Details of each
80
component that influence stiffness of soil due to tunneling are explained below.
Stress dependent stiffness: In centrifuge test, stress increases with increasing body weight of
soil as centrifugal acceleration rises. This rise in acceleration resulted in increase in stiffness
of soil. However, due to tunnel excavation, reduction in stress occurred in the soil located
around the new tunnel. As a result, stiffness decreased as confining stress reduced. On the
other hand, for soil located at a further distance away from the new tunnel, increase in stress
caused by stress redistribution occurred to maintain equilibrium. Thus, soil stiffness increased
at a further distance away from the new tunnel.
Strain dependent stiffness: To simulate the stiffness of soil in centrifuge test, the process of
increasing of centrifugal acceleration was modeled. Before the excavation of the tunnel, rise
in centrifugal acceleration shear strain was already induced and should be taken into account
to consider mobilization of stiffness. During tunnel excavation, induced shear strain was
estimated to be from 0.1 to 1% (Mair, 1993) depended on volume loss imposed. For this
reason, constitutive model should be able to capture the reduction in stiffness with increasing
shear strain.
Path dependent stiffness: Due to increase of centrifugal acceleration, the soil in the model
package was subjected to 1D consolidation and path of the soil followed K0 conditions.
During tunnel excavation, change of stress path or loading direction on the soil element took
place. Tunnel excavation caused reduction in effective mean stress (p) due to effects of stress
relief on soil located around the tunnel. For soil located further away from the tunnel, increase
in effective mean stress occurred due to stress redistribution. Shear stress (q) also induced
during increase in centrifugal acceleration and excavation of the new tunnel. The constitutive
model should be able to simulate path dependent stiffness in order to reasonably back-analyze
the data from centrifuge tests.
81
4.5 Determination of model parameters
Hypoplasticity model parameters for large strain behavior (c, hs, n, ec0, ed0, ei0,) for Toyoura
sand were adopted based on calibration resulted proposed by Herle and Gudehus (1999).
Exponent , exponent and small strain stiffness or intergranular strain concept parameters
(mR, mT, R, r and ) were calibrated using the curve fitting method. In this research, the
computed results were fitted to triaxial test with local strain measurement results from
Yamashita et al. (2000) and bender element tests from Yamashita el al. (2009). Model
parameters were summarized in Table 4.3. Detail calibration of each parameter is given
below.
4.5.1 Parameter controlling behavior in large strain
Following details show determination criteria for parameters controlling behavior in large
strain range, which consist of c, hs, n, ec0, ed0, ei0, , .
c: Critical state friction angle

Calibration: perform angle of repose test or shear test
hs: Parameter controlling the overall slope of limiting void ratio curves
n: Parameter controlling curvature of limiting void ratio curves
ec0: Parameter controlling position of initial critical state void ratio
ei0: Parameter controlling isotropic compression line or the theoretical loosest possible
state
ed0: Parameter controlling minimum void ratio or the densest state void ratio
Equation 4-2 shows the relationship of hs, n, ec0, ed0 and ei0 and Figure 4.3 shows limiting
void ratio curves
ei ec ed 3 p' n
exp 4-2
ei 0 ec 0 ed 0 hs
82
Calibration: carry out oedometric test or isotropic triaxial test

n can be calibrated by oedometric loading test of an initially loose sample as shown in
Equation 4-3 and parameter n can be determined using Figure 4.6.
ln(e p1Cc 2 / e p 2Cc1 )

n 4-3
ln( ps 2 / ps1 )
hs can be calculated based on relationship between hs and n as shown in Equation 4-4:
1/ n
ne
hs 3 ps p 4-4
Cc
Calibration: perform shear test, index test to determine minimum and maximum density and
cyclic shearing test
ec0 can be obtained from a shear test on a soil element at the same time as the calibration of
critical state friction angle (c) or from a relationship between ec0 and emax from index test
Empirical relationship ec0 = emax (Herle and Gudehus, 1999)
ed0 can be obtained by cyclic shearing test with small amplitude under constant pressure or
adopted from an empirical relationship with ec0 or from an index test
Empirical relationship ed0 = 0.5 ec0 (Herle and Gudehus, 1999)
Empirical relationship ed0 = emin from index test
ei0 can be calculated from idealized packing of spheres and cubes. It is not practical to obtain
ei0 from experimental and it is often related to an empirical relationship with ec0.
Empirical relationship ei0 = 1.2 ec0 (Herle and Gudehus, 1999)
The controlling void ratio curves for Toyoura sand using parameters in Table 4.3 are shown in
83
Figure 4.7.
: Parameter controlling the dependency of peak friction angle on relative density

: Parameter controlling the dependency of soil stiffness on relative density
Calibration: conduct drained triaxial shear test

Parameter can be calculated based on relationship between p, c, e, ed and ec.
Parameter can be calculated based on relationship between e, ec0, ed0, Edense and Eloose.
Alternative calibration: curve fitting and to results of drained triaxial shear test using
numerical soil testing software such as PLAXIS 3D Soil test utility.
4.5.2 Small strain stiffness parameters
Small strain stiffness or intergranular strain parameters consist of following details.

mR: Parameter controlling the shear modulus of initial loading (very small strain) or upon
180 strain path reversal (i.e., mR controls G0)
m T: Parameter controlling the shear modulus upon 90 strain path reversal
R: Parameter controlling the size of the elastic range
r: Parameter controlling the intergranular strain evolution rate
Parameter controlling the tangent stiffness degradation with normalized length of
intergranular strain tensor ()
Calibration: perform drained triaxial shear test on sample using local strain measurement then
calibrate mR, mT, R, r and by means of fittings to the element test results
Empirical relationship mT = 0.5 mR (Atkinson et al., 1990)
4.5.3 Numerical calibration result
Figure 4.8 shows a comparison between measured and computed secant modulus Youngs
modulus of Toyoura sand (Yamashita et al., 2000, 2009). Element tests from Yamashita et al.
84
(2000) were carried out by using drained triaxial compression test with local strain
measurement. In addition, a result of secant shear modulus (G) of Toyoura sand obtained from
Yamashita et al. (2009) was converted into Es as a comparison in the small strain range. The
soil sample tested in this figure was prepared by anisotropic consolidation (K0 = 0.46) on
reconstituted dry pluvial deposition method, which is the same method adopted to prepare soil
sample in the centrifuge test. The void ratio of sand and confining pressure in Yamashita et al.
(2000) were 0.799 and 46 kPa, respectively. The computed result was carried out by using
Soil test utility in PLAXIS 3D 2013 (Brinkgreve et al, 2013). The parameters controlling
behavior in large strain (i.e., c, hs, n, ec0, ed0, ei0) were adopted from Herle and Gudehus
(1999). Exponents and were fitted to triaxial test result with local strain measurement
presented by Yamashita et al. (2000). Initial secant modulus or very small strain modulus was
fitted by controlling mR. Parameter mT was assumed to be 0.5mR based on Atkinson et al.
(1990). Parameters R and were assumed to be material independent constants as 3x10-5 and
1, respectively. Thus, secant modulus degradation curve was fitted by controlling r.
Intergranular strain concept parameters adopted in this study were within the range reported in
previous studies (Niemunis and Herle 1997).
4.6 Numerical modeling procedure
4.6.1 Numerical back-analysis
The procedure of numerical modeling basically followed that in the centrifuge tests. Drained
effective stress analysis was adopted as every test was carried out in dry sand. The numerical
back-analysis followed the procedure below:
Create the initial conditions with following details:

- Specify the void ratio in 1g conditions
- Set the intergranular strain of soil elements to zero (i.e., no deformation at the
initial stage)
- Activate the lining of the existing and new tunnels to simulate wished-in-place
85
tunnel lining
- Deactivate the soil elements inside the existing tunnel and some parts inside the
new tunnel (see Figure 4.1)
- Initialize stress under 1g conditions with K0 equaling 0.5
Increase the centrifugal acceleration by increasing gravity by 60 times (Mweight = 60)
Excavate new tunnel by simulating the effects of both volume loss and weight loss as
follows:
- To simulate the effects of 2% tunnel volume loss, the surface contraction
technique (a utility available in the PLAXIS software) was used. This technique
applies a uniform radial contraction on the tunnel lining. It should be noted that
this numerical simulation technique does not represent a perfect match to that of
centrifuge model tests. However, the discrepancy between the numerical and
centrifuge simulation techniques should not affect any key conclusion obtained
from this study significantly since the volume loss simulated in both numerical
and physical modeling is identical.
The surface contraction technique simulated a uniform radial contraction of
the tunnel lining in each section. At the common nodes or interface of each
section, the magnitude of contraction was half of that in the middle of the
section. The contraction gradually increased from the interface to the middle
part of each section. Another half of the contraction at the interface of the
section was simulated in the following excavation stage. By doing so, there
is no double contraction simulated during the numerical modelling
procedures.
- Simulate the effects of weight loss by removing (i.e., deactivating) the soil
elements (with the same unit weight as the heavy fluid used in the centrifuge
test) inside the tunnel.
- The face of the new tunnel was supported using rollers. These rollers were
applied simultaneously with soil removal (deactivation of soil elements) at each
86
stage. This means that end conditions simulated by numerical and centrifuge
modelling are consistent.
Advance the new tunnel by a distance of 0.6D in each excavation stage by repeating
above steps for a total distance of 3.6D in six stages.
4.6.2 Numerical parametric study
The following details describe the differences in procedures between that in numerical
back-analysis and parametric study.
4.6.2.1. Procedures for tunneling in greenfield conditions
All the procedures of tunneling in greenfield conditions are the same as that in the numerical
back-analysis except there is no the existing tunnel in numerical parametric study. The other
configurations such as mesh density, boundary conditions and initial conditions remained
unchanged from those in back-analysis.
4.6.2.2. Procedure of non-wished-in-place lining activation to investigate the influence
of tunneling sequence of new twin tunnels
Unlike the numerical back-analysis, non-wished-in-place activation of lining for new twin
tunnels was adopted (see Figure 4.2b). For the side-by-side new twin tunnel case, tunnel
excavation started from the left new tunnel, followed by the right tunnel. In the case of the
vertically stacked tunnel case, the upper new tunnel was excavated first followed by the lower
new tunnel in case E2N3,5N while the lower new tunnel started first in case E2N5,3N. In
each excavation stage, the lining of the new tunnel was activated sequentially. Apart from this
difference in tunnel lining activation sequence, other procedures in parametric study were the
same as that in the back-analysis.
87
4.6.2.3. Procedures to investigate influence of volume loss, relative density and tunneling
in saturated sand on the interaction of crossing tunnels
Following details are the difference between procedures adopted in numerical back-analysis
and parametric study. To investigate the effects of volume loss on tunnel-tunnel interaction,
the imposed volume loss of the new tunnel varied from 0.5% to 4% by changing the specified
surface contraction, which is a uniform radial contraction applied on the soil around the
new tunnel.
In order to study the influence of relative density on crossing-tunnel interaction, total density
and void ratio of sand was modified to achieve the relative density ranging from 30% for
loose sand to 80% for dense sand.
For tunneling in saturated sand, hydrostatic pore water pressure distribution was specified in
1g condition. As the centrifugal acceleration increased, pore water pressure also increased as
rise in unit weight of water.
88
Table 4.1 Summary of numerical analysis of tunneling sequence of new twin tunnels
Test C/D C/D C/D Centrifuge Numerical Remark

(existing (first (second test analysis
tunnel) new new
tunnel) tunnel)
E2N3,3 2.0 3.5 3.5 Back-analysis of

centrifuge tests
E2N3,5 2.0 3.5 5.0
E2N3,3N 2.0 3.5 3.5 Simulation of non

wished-in-place
E2N3,5N 2.0 3.5 5.0
lining of new tunnels
E2N5,3N 2.0 5.0 3.5
89
Table 4.2 Summary of numerical parametric study to investigate the effects of volume loss,
relative density and tunneling in saturated sand on the interaction of crossing tunnels
Series Run ID VL(a) Dr(b) Water Remark
(%) (%) condition
Reference case E2N3 2.0 64 Dry sand Reference case based
on back-analysis of Test
E2N3
E2N3-V0.5 0.5 64 Dry sand

Volume loss
E2N3-D30 2.0 30 Dry sand

Density E2N3-D50 2.0 50 Dry sand
E2N3-D80 2.0 80 Dry sand
Saturated sand E2N3-SS 2.0 64 Saturated sand
Note: (a) VL denotes volume loss
(b) Dr denotes relative density
90
Table 4.3 Summary of material parameters adopted in finite element analysis

30
Critical state friction angle(a),c
2.6 GPa
Granulates hardness(a), hs
Exponent n(a), n 0.27

0.61
Minimum void ratio at zero pressure(a), edo
0.98
Critical void ratio at zero pressure(a), eco
1.10
Maximum void ratio at zero pressure(a), eio
Exponent (b) 0.5
Exponent (b) 3
Parameter controlling the initial shear modulus upon 180 strain path 8
reversal and in the initial loading(b), mR
Parameter controlling the initial shear modulus upon 90 strain path 4

reversal(b), mT
The size of the elastic range(b), R 0.00003

0.2
Parameter controlling rate of degradation of stiffness with strain(b)r
Parameter controlling rate of degradation of stiffness with strain(b) 1.0

The coefficient of at-rest earth pressure, K0 0.5
Note: (a) Herle and Gudehus, 1999
(b) Calibrated based on previous literatures (Yamashita et al., 2000, 2009)
91
625
930
Z
Y X
Existing tunnel
750
New tunnel
Plane of symmetry
(X/D = 0)
Note: Dimension in mm (model scale)

(a)
930
1250
Z
Y X
750
Location of the
existing tunnel
Right new tunnel

Left new tunnel
(b)
Figure 4.1 Three-dimensional finite element mesh for (a) the reference case (E2N3); (b)
side-by-side twin tunnel excavation (E2N3,3)
92
Z
Y X
New tunnel centerline

Existing tunnel
(Plane of symmetry,
=100
X/D = 0)
P/D = 0.5
New tunnel
=100
Dimension in mm
(model scale)
(a)
1 1 8 15
1 1 8 15
1 1 7 14
1 1 1 1 6 13
1 1 5 12
Existing 1 1 Existing 4 11
tunnel 1 1 tunnel 3 10
1 1 2 9
Left new Right new Left new Right new
tunnel (1st ) tunnel (2 nd ) tunnel (1st ) tunnel (2 nd )
Note: Number denotes the sequence of each tunnel lining is activated
Numerical back-analysis Numerical parametric study
(b)
Figure 4.2 (a) Some details of the perpendicularly crossing tunnels; (b) sequence of between
wished-in-place and non-wished-in-place lining of the new tunnel
93
Figure 4.3 Limiting curves for void ratio ( Herle and Gudehus, 1999)
94
q: kPa X
100
D
C 0 A
200 400 P: kPa
-100
B
(a)
(b)
Figure 4.4 (a) Stress path; (b) shear stiffness of reconstituted London Clay with different
loading directions (modified from Atkinson et al., 1990)
95
Size of the elastic range
Figure 4.5 Response envelope of small strain stiffness in hypoplasticity model (after Niemunis
and Herle, 1997)
96
Figure 4.6 Determining of parameter n from limiting curves of void ratio (Herle and
Gudehus, 1999)
97
1.2 Average Dr = 64%,

ei e0 = 0.733
1.0 ec
0.8 Existing tunnel-

Void ratio, e
Crown
ed
0.6
Existing
0.4 tunnel Invert
0.2
Normalized mean stress - void ratio curve
Reference test (E2N3)
0.0
1.E-07 1.E-05 1.E-03 1.E-01 1.E+01
p'/hs
Figure 4.7 Controlling void ratio curves for Toyoura sand
98
400 Measured [Yamashita et al., 2000]

Measured [Yamashita et al., 2009]
350
Secant Young's modulus,
Computed [this study]

300
250
Es (MPa)
200
150
100
50
0
0.0001 0.001 0.01 0.1 1
Axial strain, a (%)
Figure 4.8 Calibration of parameters with test results with bender element and local strain
measurement
99
Chapter 5 Effects of volume and weight losses on perpendicular crossing tunnels
Chapter 5 Effects of volume and weight losses on

perpendicularly crossing tunnels
5.1 Introduction
Existing tunnels in the ground may experience excessive deformation and their linings may
show signs of cracking when new tunnels are excavated close to them. It is thus important to
consider ground movements and stress changes when constructing new tunnels close to
existing ones, especially in urban areas where more and more tunnels are being built with
greater proximity to each other. As demonstrated in Chapter 2, the interaction of crossing
tunnels is complex and required further investigation the improve the understanding.
This chapter presents responses of an existing tunnel due to effects of volume and weight
losses induced by a new tunnel excavation underneath. The major objective of the centrifuge
tests in this chapter is to examine the behavior of the existing tunnel and stress transfer
mechanism in crossing-tunnel interaction. The results are mainly interpreted from Test E2N3
(refer to Figure 3.3 and Table 3.2). In this test, the cover depth-to-diameter ratio (C/D) of the
existing tunnel, C/D of the new tunnel and pillar depth-to-diameter ratio (P/D) were 2, 3.5 and
0.5, respectively. Note that the pillar depth is the clear vertical distance between the two
tunnels
In order to study the effects of modeling volume and weight losses separately on
crossing-tunnel, comparison of results from Tests E2N3 and E2N3-VW was interpreted. The
tunnel configuration in both tests was identical. In Test E2N3, the effects of volume and
weight losses due to the excavation of a new tunnel were modeled simultaneously. On the
other hand, in Test E2N3-VW, the effects of volume loss were simulated first followed by the
100
effects of weight loss. When only the effects of volume loss are interpreted, the first part of
Test E2N3-VW is called Test E2N3-V. A summary of the modeling sequences carried out in
both tests is given in Table 3.3.
5.2 Effects of volume and weight losses on ground and existing tunnel
5.2.1 Ground surface settlement above the existing tunnel
Figure 5.1 compares measured and computed surface settlements normalized by tunnel
diameter (s/D) for different modeling sequences at the end of tunneling. The imposed volume
loss was 2% in each test. In Test E2N3, where the effects of volume and weight losses were
modeled simultaneously, the measured maximum normalized surface settlement was about
0.34% (20 mm). When only the effects of volume loss were simulated (Test E2N3-V), the
maximum normalized surface settlement was about 15% larger than that in Test E2N3. This is
because soil heave due to weight loss (or stress relief) was not simulated in Test E2N3-V,
resulting in the larger ground surface settlement. On the other hand, when the effects of
weight loss were simulated after the simulation of volume loss (Test E2N3-VW), additional
surface settlement was induced. The maximum surface settlement in Test E2N3-VW was
about 10% larger than that in Test E2N3-V. This finding was somewhat unexpected initially,
but after detailed investigation it was revealed that when the heavy fluid inside the rubber
bags mounted inside the tunnel lining was drained away, the supporting pressure exerted by
the heavy fluid on the new tunnel lining was removed. Consequently, the new tunnel was
compressed vertically by overburden pressure (to be further discussed later) causing the
additional surface settlement. Although the removal of soil from inside the new tunnel in Test
E2N3-VW led to stress relief and hence soil heave, the effects of vertical compression of the
new tunnel on ground settlement were more pronounced. Verruijt and Brooker (1996)
investigated the effects of vertical compression of a tunnel on ground surface displacements
by an analytical elastic solution and they reported that surface settlement occurs directly
above the tunnel whereas heave takes place at some distance away. In the physical model tests
carried out in this study, however, the vertical compression of the new tunnel only caused
101
surface settlement but not heave. This is expected since soil is not elastic, as assumed in the
analytical elastic solution.
Although the overall trends between measured and computed results were comparable, minor
discrepancy between the measured and computed surface settlements of the two tests was
observed. One of the possible reasons for the discrepancies is that the stress-induced
anisotropy computed implicitly by the hypoplastic constitutive model could not describe the
induced soil anisotropic responses in centrifuge tests exactly. Ng and Lee (2005) have
illustrated that the magnitude and profile of computed ground surface settlements are strongly
influenced by the degrees of stiffness anisotropy assumed in their numerical simulations.
To investigate the effects of the existing tunnel on the surface settlement induced by the
advancing perpendicularly crossing tunnel underneath, computed results of the greenfield
case N3 (refer to Section 4.2.3.1) were also compared with the computed surface settlements
above the existing tunnel for Test E2N3 considered. The computed greenfield maximum
surface settlement was larger than that due to the presence of the existing tunnel. Thus,
stiffening effects due to the presence of an existing tunnel should not be ignored in design
analysis.
5.2.2 Settlement of the existing tunnel and tunnel gradient
Figure 5.2 shows the measured and computed settlements of the existing tunnel in the
longitudinal direction at the end of tunnel excavation. Maximum measured normalized tunnel
settlement (/D) in Test E2N3 was about 0.3% (18 mm) which exceeded one allowable limit
of 15 mm (LTA, 2000) but was still within another allowable limit of 20 mm (BD, 2009).
Settlement of the existing tunnel for different modeling sequences had the same overall trend
as the measured ground surface settlement above the existing tunnel (refer to Figure 5.1). The
tunnel settlement measured in Test E2N3-V and Test E2N3-VW were larger than those
measured in Test E2N3 and exceeded the permissible limits set by LTA (2000) and BD (2009).
The measured and computed tunnel settlements in Test E2N3 were comparable, suggesting
102
that the stress transfer mechanism on the existing tunnel may be investigated using numerical
analysis.
The gradient of the existing tunnel was calculated from the slope of measured settlement of
the existing tunnel. The maximum tunnel gradient in Test E2N3 of 1:1600 exceeded one limit
of 1:2500 (Li and Yuan, 2012) but was still within another limit of 1:1000 (LTA, 2000; BD,
2009). The maximum gradient was located at a distance of about 2.5D from the centerline of
the new tunnel (i.e., X/D = 2.5).
In addition, settlements of the existing tunnel and gradients in Test E2N3 were compared with
data from two case histories where the settlement of an existing tunnel was induced by a new
tunnel excavation underneath. Given the potential differences between field monitoring and
centrifuge tests in terms of ground conditions, tunneling methods and the flexural rigidity of
the tunnel, the field monitoring data and centrifuge test results cannot be compared
quantitatively but it is possible to illustrate qualitatively the general trend of settlement of the
existing tunnel.
5.2.3 Induced strain and shear stress in the longitudinal direction of the existing tunnel
Figure 5.3 illustrates the induced strains measured along the invert in the longitudinal
direction of the existing tunnel at the end of tunnel excavation. Induced strain in the
longitudinal direction of the existing tunnel was measured by strain gages installed at the
crown and the invert of the existing tunnel (refer to Figure 3.11). Due to the new tunnel
excavation, sagging moment was induced at the location directly above the new tunnel (i.e.,
X/D = 0), resulting in tensile (+ve) and compressive (-ve) strain induced at the invert and the
crown of the existing tunnel, respectively. The cracking tensile strain of unreinforced concrete
is 150 (ACI, 2001).
In Test E2N3, the maximum tensile strain of about 151 was induced at the location directly
above the new tunnel. Hence cracks might appear on the lining of the existing tunnel.
103
Although most of the tunnel lining was made of reinforced concrete, induced tensile strain
can widen the gap in the circumferential joint and cause water leakage. The maximum
induced tensile strain was larger in Tests E2N3-V and E2N3-VW than in Test E2N3 within a
distance of 2D from the centerline of the new tunnel (i.e., from X/D = 0 to 2). This is because
the maximum settlement of the existing tunnel was larger in Tests E2N3-V and E2N3-VW
than in Test E2N3 (refer to Figure 5.2).
The shear stress on the tunnel lining was deduced from the slope of the induced strain in the
longitudinal direction of the existing tunnel. For comparison purposes, an allowable shear
stress of 660 kPa was estimated according to an assumed concrete compressive strength (fc)
of 50 MPa and a reduction factor of 0.55 (ACI, 2011). In Test E2N3, the maximum shear
stress was 780 kPa, which exceeded the allowable shear stress, suggesting that cracks might
have appeared on the tunnel lining. There was large shear stress on the lining of the existing
tunnel at a distance between 2D and 3D from the centerline of the new tunnel.
Liu (1990; cited by Liao et al., 2008) reported a case history from Shanghai in which diagonal
cracks were observed on tunnel linings when differential settlement occurred on a water
transmission tunnel. Liao et al. (2008) suggested that shear stress in the tunnel lining was one
of the key factors influencing tunnel deformation. The cracks in their study were located in an
area between the location of maximum tunnel settlement and the inflection point of the tunnel.
In this study, the inflection point was estimated to be at a distance between 2D and 3D from
the centerline of the new tunnel.
Given the effects of volume and weight losses on cross-tunnel interaction were investigated
separately, it is evident that the trends of surface and tunnel settlements and induced strains in
the longitudinal direction of the existing tunnel in the two tests are all similar but differ only
in magnitude. Thus, it suffices to report results mainly from Test E2N3 only from now on.
104
5.2.4 Tunnel deformation
Figure 5.4 shows deformations of the existing tunnel during the advancement of the new
tunnel. The section of the existing tunnel directly above the new tunnel centerline (X/D) in
Test E2N3 is illustrated to observe change in tunnel diameter of the existing tunnel. It can be
seen that the existing tunnel was vertically compressed and horizontally elongated as the new
tunnel advanced. The measured maximum normalized vertical compression and horizontal
elongation of the existing tunnel were 0.04% and of 0.07%, respectively. The measured
maximum normalized vertical compression and horizontal elongation of the existing tunnel
occurred when the new tunnel face was at -0.3D and -0.9D away from the centerline of the
existing tunnel, respectively. When the excavated section of the new tunnel was directly
underneath the existing tunnel (i.e., at Y/D = 0.3), a significant reduction in both vertical
compression and horizontal elongation of the existing tunnel was observed. As the new tunnel
passed the existing tunnel, the existing tunnel continued to deform but at a reduced rate.
On the other hand, the computed results show almost the same magnitude (or symmetrical) of
vertical compression and horizontal elongation of the existing tunnel due to the advancement
of the new tunnel. This is because uniform soil displacement around the new tunnel was
imposed in the numerical analysis. In the centrifuge test, however, soil displacement around
the new tunnel was unlikely to be uniform, resulting in the unsymmetrical measured vertical
compression and horizontal elongation of the existing tunnel. The computed maximum
vertical compression of the existing tunnel is smaller than the measured one when the
advancing tunnel face was located at -0.9D away from the centerline of the existing tunnel.
Both measured and computed results suggest that the most critical vertical compression and
horizontal elongation of the existing tunnel occurred when the approaching new tunnel face
was at -0.9D away from the existing one. At the end of new tunnel excavation, measured and
computed deformations of the existing tunnel were consistent with each other. This increase
the confidence in the conclusions derived from the test.
According to one code of practice (BTS, 2000), the minimum and maximum diameters of a
105
tunnel should not differ by more than 2% (i.e., (Dmax Dmin)/D0 2%), where D0 is the initial
diameter of the tunnel which equals to 6 m in this study. This allowable limit was not
exceeded. But because the existing tunnel was vertically compressed even before the new
tunnel excavation due to the vertical stress being larger than the horizontal stress (i.e., K0 < 1),
induced deformation may enlarge the gap in the radial joint and cause water leakage.
Kim et al. (1998) carried out a 1-g physical model test of crossing tunnels in clay. They
reported that the existing tunnel was compressed vertically by the large jacking forces from
the miniature tunneling machine when the new tunnel liner was driven. The lining of the new
tunnel in this study was wished-in-place before tunnel excavation. As the new tunnel
advanced, the existing tunnel was compressed vertically. This is because stress transfer due to
the new tunnel excavation caused a reduction in the horizontal stress larger than the vertical
stress acting on the existing tunnel. More explanations are given in Section 5.3.1.
In Figure 5.4, the computed deformation of the new tunnel at the location directly underneath
the existing tunnel (i.e., Y/D = 0) in Test E2N3-VW is also given to explain the effects of
different modeling sequences on ground surface settlement (Figure 5.1) and settlement of the
existing tunnel (Figure 5.2). After the soil inside the new tunnel was removed (case
E2N3-VW), which effectively meant that the supporting pressure inside the tunnel was also
removed causing additional ground settlement above the new tunnel, the new tunnel became
vertically compressed. Consequently, in case E2N3-VW the vertical compression of the new
tunnel caused additional settlement of soil above the new tunnel, which in turn increased
ground surface settlement and settlement of the existing tunnel.
5.2.5 Induced strain in the transverse direction of the existing tunnel
Figure 5.5 shows the measured and computed strains induced at the outer surface of the
existing tunnel at the end of tunnel excavation in Test E2N3. Induced strains at the outer face
of the existing tunnel were measured by strain gages fixed to the tunnel lining in the
transverse direction at the location directly above the new tunnel. The positive and negative
106
signs denote induced tensile and induced compressive strain, respectively. According to the
measured results, there was induced compressive strain at the crown, shoulders, knees and
invert while there was induced tensile strain at both springlines. By considering strain in the
transverse direction, it was confirmed that the existing tunnel was vertically compressed and
horizontally elongated (see Figure 5.4). Computed results were comparable to measured
results with minor discrepancy. The discrepancy is discussed previously in Section 5.2.2.
From measured results of the two tests, the maximum induced compressive strain and induced
tensile strain of 56 and 67 occurred at the invert and at the left springline, respectively. The
maximum tensile strain on the tunnel lining was still below the cracking tensile strain limit of
150 (ACI, 2001). However, if strain in the transverse direction was large even before the
start of the new tunnel excavation, tunneling may cause cracks on the lining of the existing
tunnel. It should be noted that induced strain was more significant in the vertical and
horizontal directions (i.e., at the crown, springlines and invert) than in the diagonal direction
(i.e., at the shoulders and knees). However, this observation may only be applicable for the
soil type and in-situ stress conditions adopted in this study.
5.3 Effects of three-dimensional stress transfer on an existing tunnel
5.3.1 Incremental normal stress in the transverse direction of the existing tunnel
Figure 5.6 shows the computed incremental normal stress in the transverse direction of the
section of the existing tunnel directly above the new tunnel in case E2N3. The effects of the
changes in normal stress on the responses of the existing tunnel in the transverse direction
were investigated at eight chosen locationsthe crown, shoulders, springlines, knees and the
invert. The positive and negative signs denote increases and decreases in normal stress
relative to that before tunneling, respectively. Note that change in normal stress due to the first
two excavation stages (i.e., from Y/D = -1.5 to -0.9) was significantly smaller than the
subsequent four stages (i.e., from Y/D = 0.3 to 1.5) and is not shown for clarity.
107
When the excavated section was at -0.3D away from the centerline of the existing tunnel, the
maximum normal stress reduction occurred at the left knee as it is closest to the advancing
tunnel face. As excavated section of the new tunnel reached the centerline of the existing
tunnel (Y/D = 0.3), largest stress reduction took place at the invert as it was the closest to the
excavated section. In addition, normal stress decreased at the springlines, knees and invert but
increased at the crown and shoulders. As the tunnel face advanced further at 0.9D beyond the
centerline of the existing tunnel, there was a significant decrease in normal stress at the right
knee in and a slight drop of stress reduction at the invert due to stress redistribution,
comparing with when Y/D = 0.3. At the completion of tunnel excavation (Y/D = 1.5), change
in normal stress at every location on the existing tunnel at the end of excavation is small
comparing with when Y/D = 0.9. It suggests that significant change in normal stress occurred
when tunnel face located at a distance within 1D (Y/D = 1) from the centerline of the
existing tunnel.
To investigate tunnel deformations, change in normal stress in the horizontal and vertical
direction was considered. At the end of new tunnel excavation, the reduction in stress in the
horizontal direction was larger than that in the vertical direction, resulting in tunnel
compression in the vertical direction. This is consistent with the measured and computed
deformation of the existing tunnel shown in Figure 5.4.
5.3.2 Vertical stress in the longitudinal direction of the existing tunnel
Figure 5.7 illustrates the computed vertical stress distribution along the crown and invert in
the longitudinal direction of the existing tunnel at the end of excavation in Test E2N3. At the
location directly above the new tunnel (i.e., from X/D = 0 to 0.5), stress increased
substantially at the crown whereas it decreased significantly at the invert of the existing tunnel.
Along the crown, normal stress decreased as the distance away from the centerline of the new
tunnel increased. On the other hand, normal stress along the invert increased with distance
until it reached a peak at 2D away from the new tunnels centerline.
108
The large tunnel settlement (Figure 5.2), large induced strain in the longitudinal direction and
large shear stress (Figure 5.3) are mainly caused by two factors. First, soil arching caused a
sharp reduction in vertical stress above the centerline of the new tunnel and an increase in
vertical stress at some distance away due to stress redistribution along the invert of the
existing tunnel. Second, overburden stress transfer along the crown of the existing tunnel
caused vertical stress to increase substantially. Soil arching is explained in the next section.
The changes in normal stress acting on both the crown and the invert of the existing tunnel
exceeded the limits defined in two codes of practice (i.e., +15 kPa for LTA, 2000; 20 kPa for
BD, 2009). Up to an offset distance of 1.5D from the centerline of the new tunnel, changes in
normal stress along the crown of the existing tunnel also exceeded the allowable limit. Along
the invert, normal stress reduced by more than the codes of practice would allow in the area
between the centerline and a distance of 1D away from the centerline of the new tunnel. At a
distance of 1.5D to 5D away from the centerline of the new tunnel, the increase in normal
stress along the invert exceeded the recommended limits as well. Thus, the structural capacity
of the existing tunnel should be reviewed based on changes in the loading condition around it.
5.4 Soil arching in perpendicularly crossing tunnels
Figure 5.8a and Figure 5.8b show the computed directions of principal stress in case E2N3 in
the transverse direction of the existing tunnel before tunnel excavation and when the new
tunnel reached the fourth section (Ex4 in the figures), respectively. There was a slight
decrease in the magnitude of principal stress above each of section 1 to 3 (Ex1 to Ex3) as a
result of tunnel excavation in each previous stage. Directly underneath the invert of the
existing tunnel (i.e., above Ex4), both minor and major principal stresses reduced sharply.
They did so because the soil above the existing tunnel tended to settle due to the new tunnel
excavation but was prevented from doing so by the existing tunnel. Subsequently, overburden
stress was transferred to the crown of the existing tunnel as a result of stress redistribution in
the longitudinal direction of the new tunnel causing an increase in the major principal stress.
109
The stress transfer around the existing tunnel resulted in a decrease in normal stress at the
invert and both springlines and an increase in normal stress at the tunnel crown when the
section of the new tunnel directly underneath the existing tunnel was being excavated (refer to
Figure 5.6; when Y/D = 0.3).
Figure 5.8c and Figure 5.8d show the computed directions and magnitudes of principal
stresses in the longitudinal direction of the existing tunnel, before tunnel advancement and
after the new tunnel reached the fourth section (Ex4), respectively. As expected, the
magnitudes (i.e., sizes of vectors) of both major and minor principal stresses near the new
tunnel reduced substantially due to the effects of volume loss (or shearing) and stress relief,
which in turn were due to the advancement of the new tunnel. As illustrated by the rotation of
principal stresses, shear stress was induced due to the excavation of the new tunnel. Since the
existing tunnel and the soil further away from the new tunnel (i.e., that directly above Ex5 and
Ex6 in Figure 5.8b and at X/D greater than 1 in Figure 5.8d) should have larger shear strength
and stiffness than the soil closer to the new tunnel due to stress relief and shearing, stress
redistribution (or soil arching) took place to maintain the overall equilibrium, as revealed by
the rotations and the increases in magnitude of principal stresses of the soil above the existing
tunnel in Figure 5.8b and Figure 5.8d. Also soil arching caused principal stress to rotate in
direction in the soil located at X/D greater than 1 and below the invert of the existing tunnel
(see Figure 5.8d).
5.5 Summary and conclusions
Three-dimensional centrifuge and numerical investigations of the interaction between two

perpendicularly crossing tunnels were carried out in dry sand. In order to simulate the effects
of both volume and weight losses on an existing tunnel due to the construction of a new
tunnel underneath, a novel donut was developed to control volume loss and to mimic soil
removal in-flight. Based on the measured and computed results, the following conclusions
may be drawn:
110
(a) The measured maximum ground surface settlement was the smallest when the effects
of both volume and weight losses were modeled simultaneously (i.e., Test E2N3). On
the other hand, the surface settlement induced when the effects of weight loss were
simulated after modeling volume loss (i.e., Test E2N3-VW) was 10% larger than that
induced when only volume loss was simulated (i.e., Test E2N3-V). This is because
when the heavy fluid inside the rubber bags mounted inside the tunnel lining was
drained away, the supporting pressure exerted by the heavy fluid on the lining of the
new tunnel was removed. Consequently, the new tunnel was compressed vertically by
overburden pressure, causing the additional surface settlement. Numerical simulations
show that the presence of an existing tunnel can stiffen the ground and reduce ground
surface settlement due to new tunnel excavation significantly.
(b) The measured settlement of the existing tunnel was 15% larger in Test E2N3-V than
that in Test E2N3. This is because the removal of soil mass in Test E2N3 led to stress
relief resulting in ground heave which reduced the settlement induced by volume loss.
However, there was about 10% more tunnel settlement in Test E2N3-VW than in Test
E2N3-V. This is because the removal of soil from inside the new tunnel resulted in a
reduction in supporting pressure on the tunnel lining, leading to the vertical
compression of the new tunnel. This in turn induced settlement of the existing tunnel
above it. The measured ground surface settlements were consistent with the observed
tunnel settlements in all tests.
(c) Due to the excavation of a new tunnel underneath the existing tunnel, the maximum
measured settlement of the existing tunnel in Test E2N3 was 0.3%D, where D is
tunnel diameter. This settlement exceeded the permissible limits of serviceability (e.g.
LTA, 2000). Moreover, the measured tensile strain and shear stress induced in the
existing tunnel exceeded the cracking tensile strain (ACI, 2001) and allowable shear
stress limit (ACI, 2011), respectively.
(d) The section of the existing tunnel immediately above the new tunnel was vertically
111
compressed at every stage of excavation of the new tunnel in Test E2N3. This is
because stress reduction acting on the existing tunnel was larger in the horizontal
direction than in the vertical direction.
(e) At the end of the tunnel excavation, computed vertical stress increased substantially at
the crown of the existing tunnel located directly above the new tunnel. This is because
of stress transfer in the longitudinal direction of the new tunnel during the tunnel
advancement. On the other hand, there was a sharp reduction in the computed vertical
stress at the invert of the section of the existing tunnel immediately above the new
tunnel. As a result of soil arching and stress redistribution, however, the computed
vertical stress acting on the invert of the existing tunnel increased significantly to
reach a peak at an offset distance of about 2D from the centerline of the new tunnel.
112
Normalized off-set distance from

centerline of new tunnel (X/D)
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
Normalized surface settlement (/D, %)
0.0
0.1
0.2
0.3
0.4
0.5 Measured [E2N3] Computed [E2N3]

Measured [E2N3-V] Computed [E2N3-V]
Measured [E2N3-VW] Computed [E2N3-VW]
Ref-boundary Computed [N3]
Z
X
Existing tunnel
New tunnel
Figure 5.1 Comparison of measured and computed ground surface settlement at the end of
tunnel excavation
113
Normalized longitudinal distance

along existing tunnel (X/D)
-6 -4 -2 0 2 4 6
Normalized tunnel settlement (/D, %)
0.0
0.1
Max. gradient
= 1: 1600
0.2 [E2N3]
15 mm
(LTA, 2000)
0.3 20 mm
Max. gradient (BD, 2009)
0.4 = 1: 400
(Cooper et al., 2002)
Measured [E2N3] Computed [E2N3]
0.5
Measured [E2N3-V] Field [Cooper et al., 2002]
Measured [E2N3-VW] Field [Li & Yuan, 2012]
Z
X
Existing tunnel
New tunnel
Figure 5.2 Comparison of measured and computed settlement of the existing tunnel at the end
of tunnel excavation
114

along invert of existing tunnel (X/D)
-6 -4 -2 0 2 4 6
-100
-50
Induced strain ()
0
Allowable shear
50 stress = 660 kPa
(ACI, 2011)
t, crack of
100 unreinforced concrete Max. shear stress =
780 kPa [E2N3]
150
200 Measured [E2N3] Computed [E2N3]

Measured [E2N3-V] ref1
Measured [E2N3-VW] ref
Z
X
Sign convention
+ Tensile strain
Existing tunnel - Compressive strain
New tunnel
the end of tunnel excavation
115
0.10
Change of normalized diameter
of tunnel (D/D0, %)
0.05
0.00
-0.05
-0.10
-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5
Normalized location of advancing tunnel face (Y/D)
Measured [E2N3] Vert Computed [E2N3] Vert Existing
Measured [E2N3] Hor Computed [E2N3] Hor tunnel
ref Computed [E2N3-VW] Vert New tunnel
Existing tunnel Existing tunnel

at X/D = 0
Y/D -1.5 1.5 Y/D -1.5 1.5

New tunnel New tunnel at Y/D = 0
DV Sign convention of D/D0

+ Tunnel diameter elongation
DH
- Tunnel diameter compression
Figure 5.4 Deformations of the existing tunnel in Test E2N3 and the new tunnel in Test
E2N3-VW
116
Induced strain ()
Crown
150
100
L-shoulder Initial R-shoulder
50
t of
0 unreinforced
-50 concrete
-100 (ACI, 2001)
L-springline -150 R-springline
Measured
Computed
L-knee R-knee
Invert
Existing
tunnel
Sign convention
+ Tensile strain
- Compressive strain
Y/D = 1.5
New tunnel
Figure 5.5 Induced strain at the outer surface of the lining in the transverse direction of the
existing tunnel in Test E2N3
117
Incremental normal Crown

stress n (kPa) -300
-250
L-Shoulder -200 R-Shoulder
-150
-100 Initial
-50
= 20 kPa
0
(BD, 2009)
50
L-Springline 100 R-Springline
Sign convention for

+ Increase in normal stress
- Decrease in normal stress
L-Knee R-Knee
Y/D
-0.3
Invert
0.3
0.9
1.5
Existing Existing Existing Existing

tunnel tunnel tunnel tunnel
Y/D = -0.3 Y/D = 0.3 Y/D = 0.9 Y/D = 1.5

New tunnel New tunnel New tunnel New tunnel
Figure 5.6 Computed incremental normal stress acting on the existing tunnel at different
excavation stages in Test E2N3
118
50
Incremental vertical stress,
0
v (kPa)
-50 = 20 kPa (BD, 2009)
-100
-150 Invert
Crown
-200
0 1 2 3 4 5 6
Normalized longitudinal distance along
invert of existing tunnel (X/D)
X/D = 0
Longitudinal Existing tunnel

stress transfer Transverse
stress transfer New tunnel
New tunnel
Figure 5.7 Computed vertical stress in the longitudinal direction of the existing tunnel at the
end of tunnel excavation in Test E2N3
119
-1
Normalized depth (Z/D)

-2
Existing
tunnel
-3
Ex1 Ex2 Ex3 Ex4 Ex5 Ex6

-4
-2 -1 0 1 2
Normalized offset distance from
existing tunnel centerline (Y/D)
(a)
-1
-2
Existing
tunnel
-3
Ex1 Ex2 Ex3 Ex4 Ex5 Ex6

-4
-2 -1 0 1 2
existing tunnel centerline (Y/D)
(b)
120
-1

-2
Existing tunnel
-3
-4
0 1 2 3 4
(c)
0
-1
-2
Existing tunnel
-3
-4
0 1 2 3 4
(d)
Figure 5.8 Computed directions of principal stress in case E2N3 in (a) the transverse direction
before tunneling; (b) the transverse direction when the new tunnel reached Ex4; (c) the
longitudinal direction before tunneling; (d) the longitudinal direction when the new tunnel
reached Ex4
121
Chapter 6 Influence of construction sequence and cover depth on crossing-tunnel interaction
Chapter 6 Influence of construction sequence and

cover depth on crossing-tunnel interaction
6.1 Introduction
When two tunnels are excavated across each other, a suitable construction sequence should be
adopted to minimize the adverse impact on the tunnel that is constructed first (considered as
the existing tunnel). This adverse impact on the existing tunnel is caused by ground
displacement and a change in stress induced by the construction of the second tunnel (i.e., the
new tunnel). It is thus important to improve the understanding of the effects of a new tunnel
excavation on an existing tunnel for different construction sequences. In addition, as the cover
depth of the existing tunnel varies from one construction sequence to another, the influence of
cover depth on the interaction of crossing tunnels should also be further examined.
This chapter presents interpretation of results from two pairs of tests to improve the
understanding of the effects of a new tunnel excavation on an existing tunnel with different
construction sequences and cover depths. For Test E2N3 (refer to Figure 3.3), the cover
depth-to-diameter ratios (C/Ds) of the existing and new tunnels were 2 and 3.5, respectively.
The pillar depth-to-diameter ratio (P/D) was 0.5 in all tests. To study the impact of
construction sequence on crossing-tunnel interaction, the new tunnel was constructed above
the existing tunnel in Test E3N2 (see Figure 3.4). Influence of construction sequence on
crossing-tunnel interaction can be considered as the effects of stress relief at different C/Ds.
To study the effects of cover depth on construction sequence, results from Tests E3N5 (as
shown in Figure 3.5) and E5N3, in which the existing and new tunnels had larger C/Ds than
122
those in Tests E2N3 and E3N2, were also analyzed. In addition, the results from Tests E2N3
and E3N5 were compared to examine the influence of C/D on the behavior of the existing
tunnel due to a new tunnel construction beneath. Note that for Test E5N3, only numerical
analysis was carried out to study the response of the existing tunnel due to the new tunnel
excavation above with different C/Ds by interpreting the results along with those from Test
E3N2. The C/Ds of the existing and new tunnels in all tests are summarized in Table 6.1.
6.2 Response of an existing tunnel due to new tunnel excavation
6.2.1 Vertical displacement and gradient of the existing tunnel
Figure 6.1 shows vertical displacement of the existing tunnel at the end of excavation. The
vertical displacement is normalized by the diameter of the new tunnel to observe the general
trend along with those in the case histories. The positive and negative signs denote heave and
settlement of the existing tunnel, respectively.
In Test E2N3, where the new tunnel was excavated beneath the existing tunnel, the maximum
measured tunnel settlement was 0.3%D (18 mm). On the other hand, heave of the existing
tunnel occurred due to the new tunnel construction above in Test E3N2. The magnitude of the
measured vertical displacement of the existing tunnel caused by the new tunnel excavation
beneath (Test E2N3) was much larger than that when the new tunnel advanced above (Test
E3N2). This is because the reduction in vertical stress acting on the existing tunnel was larger
in the former test than in the latter. In addition, the stiffness of the soil around the existing
tunnel was smaller in Test E2N3 than Test E3N2 (to be discussed later in Section 6.4.2).
At offset distances from 1D to 6D from the centerline of the new tunnel, settlement of the
existing tunnel occurred. This is because additional stress induced due to stress redistribution
at these distances. It is evident that stress redistribution took place in soil to maintain
equilibrium. Detailed explanations of stress re-distribution are given in Section 6.4.1).
The computed results were consistent with the measured ones in both tests with minor
123
discrepancies due likely to the use of model parameters adopted from the literature and
empirical relationships.
The behavior of the existing tunnel in Tests E3N5 and E5N3 was similar to that in Tests,
E2N3 and E3N2 even though the C/Ds of the existing and new tunnels in the former two tests
were 1.5 larger than those in the latter two tests. In the tests where the new tunnel was
excavated underneath the existing tunnel, tunnel settlement decreased as the C/Ds of tunnels
increased (i.e., settlement in Test E3N5 was smaller than that in Test E2N3). This is because
the stiffness of soil around the existing tunnel was larger in the deeper tunnel test than in the
shallower tunnel test, even though the effects of stress relief on the existing tunnel were larger
in the former case than in the latter. However, when the new tunnel advanced above the
existing tunnel, heave of the existing tunnel increased with increasing C/D (i.e., heave in Test
E5N3 was larger than that in Test E3N2). In these two tests, the effects of stress relief, which
increased with depth, dominated the rise in the stiffness of soil around the existing tunnel.
Further explanations are given later in Sections 6.3 and 6.4.
Field monitoring data from two case histories are also shown in Figure 6.1. It can be seen that
the vertical displacement of the existing tunnel caused by the new tunnel excavation beneath
was much more severe than that due to the new tunnel construction above.
Allowable tunnel displacements less than 15 mm and 20 mm were recommended by LTA

(2000) and BD (2009), respectively. Settlement of the existing tunnel measured in Tests E2N3
and E3N5 exceeded the limit suggested by LTA (2000) while heave of the existing tunnel in
Tests E3N2 and E5N3 was still within the allowable limits.
Table 6.2 summarizes the maximum induced tunnel gradient in each test. Similar to the tunnel
vertical displacement, the gradient of the existing tunnel induced by the new tunnel
excavation beneath (Tests E2N3 and E3N5) was significantly larger than that due to the new
tunnel advancement above (Tests E3N2 and E5N3). The induced tunnel gradient in each test
did not exceed the recommended limit of 1:1000 (LTA, 2000; BD, 2009).
124
Induced strain along the invert in the longitudinal direction of the existing tunnel at the end of
tunnel excavation is shown in Figure 6.2. The positive and negative signs denote induced
tensile and induced compressive strain, respectively.
In Test E2N3, the maximum tensile strain was induced within a distance of 0.5D from the
center of the existing tunnel (X/D = 0.5) while compressive strain was found at an offset
distance greater than 2.5D (X/D > 2.5). In Test E3N2, however, compressive strain was
induced at the center of the existing tunnel while the induced tensile strain was observed at
some distance away. The magnitude of maximum induced strain in Test E2N3 was much
larger than that in Test E3N2. This induced strain in the two tests was consistent with the
tunnel vertical displacement (refer to Figure 6.1). The computed results agreed reasonably
well with the measured ones. The possible reason for the minor discrepancies is as mentioned
above.
The direction and shape of the induced strain in the tunnel lining in Tests E3N5 and E5N3
were similar to those in Tests E2N3 and E3N2, respectively, even though the C/Ds of the
existing and new tunnels were larger in the latter two tests than in the former two. The
computed maximum induced tensile strain and compressive strain in Test E3N5 were slightly
larger than those in Test E2N3. Although settlement of the existing tunnel was larger at a
shallower depth (i.e., in Test E2N3) than at a greater depth (i.e., in Test E3N5) (see Figure
6.1), the induced strain was nevertheless slightly larger in the latter case. Similarly, for the
new tunnel excavation above the existing tunnel, the induced maximum compressive strain at
a greater depth (i.e., in Test E5N3) was larger than that at a shallower depth (i.e., in Test
E3N2). This is because of the greater vertical stress reduction of soil around the existing
tunnel in Tests E3N5 and E5N3 (with the larger C/Ds) than in Tests E2N3 and E3N2 (with the
smaller C/Ds). Further details are given later in Section 6.4.1.
According to ACI (2001), the cracking tensile strain of unreinforced concrete is 150 . The
125
maximum tensile strain induced in the existing tunnel by the construction of a new tunnel
underneath (Tests E2N3 and E3N5) exceeded the cracking tensile strain.
Table 6.3 gives the maximum shear stress induced on the tunnel lining in each test. The values
were deduced from the slope of the strain induced in the longitudinal direction of the existing
tunnel using beam theory. The allowable shear stress of concrete is estimated to be 660 kPa if
the concrete compressive strength (fc) is 50 MPa and a reduction factor of 0.55 is adopted
(ACI, 2011). The maximum shear stress induced on the existing tunnel by the new tunnel
advancement beneath (Tests E2N3 and E3N5) exceeded the allowable shear stress. Judging
from the maximum induced tensile strain and the maximum induced shear stress observed,
cracks may appear in the lining of the existing tunnel due to the new tunnel construction
beneath.
6.2.3 Deformation of the existing tunnel
Figure 6.3 shows the deformation of the existing tunnel in terms of the change in the
normalized tunnel radius (R/R0, where R is the change in the tunnel radius and R0 is the
undeformed tunnel radius, which is 3 m). The deformations at the center of the existing tunnel
(X/D = 0) when the excavated section of the new tunnel reached the centerline of the existing
tunnel (Y/D = 0.3) and at the end of tunnel excavation (Y/D = 1.5) are reported. The positive
and negative signs denote an increase and a decrease in the normalized tunnel radius,
respectively.
Figure 6.3a shows a decrease in radius at the left springline and invert but an increase in
radius at the crown and right springline of the existing tunnel as measured in Test E2N3. The
combined results of changes in radius at the crown, springlines and invert revealed that the
existing tunnel was elongated horizontally. This is because the reduction in stress caused by
the new tunnel excavation, not only resulted in a reduction in vertical stress but it also caused
a significant stress relief at each springline of the existing tunnel. By comparing the locations
of stress relief on the existing tunnel, it is found that stress reduction in the horizontal
126
direction was larger than that in the vertical direction. This means that the exiting tunnel
elongated horizontally. Further explanation of stress changes is given in Section 6.4.1. In Test
E3N2, however, the tunnel radius at the crown and left springline increased while the tunnel
radius at the right springline and invert decreased according to measurements taken. The
combined result was elongation in the vertical direction and compression in the horizontal
direction. In this test, stress decrease was larger in the vertical direction than in the horizontal
direction. The computed results were in an agreement with the measured ones with only
minor discrepancies, the reason for which has been mentioned above.
The existing tunnel experienced vertical compression in Test E3N5 as it did in Test E2N3, but
vertical elongation in Test E5N3 as it did in E3N2. This shows that an increase in the C/Ds of
the existing and new tunnels does not affect how the existing tunnel would deform.
Tunnel deformation at the end of the new tunnel excavation is shown in Figure 6.3b. As
measured in Test E2N3, the reduction in tunnel diameter in the vertical direction and the
increase in tunnel diameter in the horizontal direction were larger than those when Y/D = 0.3
(refer to Figure 6.3a). This suggests that the vertical compression of the existing tunnel at the
end of tunnel excavation was larger than that when the excavated section of the new tunnel
was directly underneath the existing tunnel. Similarly, the elongation of the existing tunnel in
the vertical direction in Test E3N2 was larger at the completion of the new tunnel than when
the excavated section of the new tunnel was directly above the existing tunnel (Y/D = 0.3).
According to BTS (2000), the maximum difference between the maximum and minimum
diameters of the tunnel should be within 2% [i.e., (Dmax Dmin)/D0 2%, where Dmax, Dmin
and D0 are the maximum, minimum and undeformed tunnel diameters, respectively]. The
deformation of the existing tunnel induced by the new tunnel excavation did not exceed this
allowable limit in any of the tests. However, given that the at-rest earth pressure coefficient
(K0) was estimated to be 0.5, the existing tunnel was vertically compressed due to overburden
pressure even before the new tunnel excavation. The construction of the new tunnel beneath
only served to compress the existing tunnel even more in Tests E2N3 and E3N5.
127
Induced strain measured at the outer surface at the center of the existing tunnel (X/D = 0) in
the transverse direction is shown in Figure 6.4. The measurements were taken when the
excavated section reached the existing tunnel (Y/D = 0.3) and at the end of the new tunnel
excavation (Y/D = 1.5). The positive and negative signs denote induced tensile and induced
compressive strain, respectively.
Figure 6.4a shows that in Test E2N3, compressive strain was induced at the left shoulder,
right knee and invert while the crown, right shoulder, springlines and left knee experienced
tensile strain. In Test E3N2, tensile strain was induced at the crown, right shoulder, knees and
invert while the left shoulder and springlines experienced compressive strain. In Test E3N5,
compressive strain was induced at the crown, knees and invert while tensile strain was
observed at the shoulders and springlines. The induced strains in all three tests were consistent
with the deformation of the existing tunnel (refer to Figure 6.3a).
The induced strain at the end of the new tunnel excavation is shown in Figure 6.4b. In the
tests where the new tunnel was excavated beneath the existing tunnel (Tests E2N3 and E3N5),
a larger induced tensile strain at the springlines and a larger compressive strain at the crown
and invert than when Y/D = 0.3 (see Figure 6.4a) were observed. In Test E3N2, induced
compressive strain at the springlines of the existing tunnel was larger at the end of tunnel
excavation than when the new tunnel was directly above the existing tunnel (Y/D = 0.3). The
induced strains confirm that vertical compression in Test E2N3 was larger at the end of tunnel
excavation than when the excavated section of the new tunnel reached the existing tunnel
(refer to Figure 6.3). Similarly, the vertical elongation of the existing tunnel in Test E3N2 was
larger at the completion of the new tunnel than at Y/D = 0.3.
The induced tensile strain in the transverse direction of the existing tunnel did not exceed the
cracking tensile strain of unreinforced concrete of 150 (ACI, 2001). However, in Tests
E2N3 and E3N5, tensile strain was present at both springlines due to overburden pressure of
128
soil even before tunnel excavation, as K0 was estimated to be 0.5. Thus, the tensile strain
induced by the new tunnel excavation may still cause cracks to appear in the lining of the
existing tunnel.
6.3 Change in soil stiffness with construction sequence and cover depth
Figure 6.5 shows the mobilized secant shear modulus of soil along the crown and invert of the
existing tunnel before and after the new tunnel excavation. The mobilized secant shear
modulus (Gm) was obtained from the computed deviatoric stress (q) and the computed
deviatoric strain (s) according to the equation Gm = q/(3s).
The mobilized shear modulus of soil along the crown of the existing tunnel is shown in Figure
6.5a. In Test E2N3, Gm was about 11 MPa before the tunnel excavation. After the tunnel
excavation, Gm only changed slightly as the influence of the new tunnel excavation on soil
stiffness was minimized by the presence of the existing tunnel. The increase in Gm after tunnel
excavation within a distance of 1D from the centerline of the new tunnel is mainly because of
increased confining stress resulting from stress redistribution in soil. This increase of stress at
the crown is discussed later in the next section.
Gm before tunnel excavation was larger in Test E3N2 than in Test E2N3 as the Gm increased
with confining pressure (which in turn increased with depth). However, after the construction
of the new tunnel above (Test E3N2), Gm reduced sharply at the center of the existing tunnel
(X/D = 0).
This reduction in Gm is because a substantial decrease in confining stress. It is well

understood that Gm is influenced by maximum shear modulus (G0) and shear strain (). In
addition, G0 is a function of void ratio and stress. The general equation of maximum shear
modulus G0 = 14.1f(e)v0.4 (MPa), (Yamashita et al., 2009). The void ratio function can be
expressed as f(e) = (2.17 e)2 / (1+e), (Iwasaki and Tatsuoka, 1997).
129
As expected, Gm before tunnel excavation was larger in Test E3N5 (with the larger C/Ds) than
in Test E2N3 (with the smaller C/Ds). After the excavation of the new tunnel underneath,
there was a slight reduction in Gm along the crown. In Test E5N3, Gm before and after tunnel
excavation were larger than those in Test E3N2.
In Figure 6.5b the mobilized shear modulus of soil along the invert of the existing tunnel is
illustrated. To compare the vertical displacement of the existing tunnel across the tests, the
mobilized shear modulus of soil on the side of the existing tunnel facing the new tunnel is
considered. It can be seen that before tunnel excavation, Gm at the crown was larger than that
at the invert of the existing tunnel in every test. This is because of the presence of the existing
tunnel, causing reduction in Gm at the invert of the existing tunnel. At any location further
away from the existing tunnel, Gm increased with depth.
Gm both before and after tunnel excavation along the crown in Test E3N2 (refer to Figure 6.5a)
were larger than those along the invert in Test E2N3. As a result, the magnitude of tunnel
heave in Test E3N2 was significantly smaller than the magnitude of tunnel settlement in Test
E2N3 (see Figure 6.1). Gm along the invert of the existing tunnel in Test E3N5 (with the larger
C/Ds) was larger than that in Test E2N3 (with the smaller C/Ds). This resulted in a smaller
settlement of the existing tunnel in Test E3N5 than in Test E2N3. However, heave of the
existing tunnel in Test E5N3 was larger than that in Test E3N2 as stress relief was a
dominating factor (to be discussed later in the next section).
6.4 Stress acting on the existing tunnel
6.4.1 Incremental normal stress acting on the existing tunnel
Figure 6.6 shows the incremental normal stress at the center of the existing tunnel (X/D = 0).
The positive and negative signs denote an increase and a decrease in normal stress,
respectively, compared with when the centrifugal acceleration reached 60g.
The incremental normal stress when the excavated section reached the centerline of the
130
existing tunnel (Y/D = 0.3) is shown in Figure 6.6a. In Test E2N3, a significant reduction in
normal stress occurred at the left knee and invert as the excavated section of the new tunnel
approached the existing tunnel. A decrease in stress was also observed at the springlines and
right knee while an increase in stress took place at the crown and shoulders. The stress
increase was due to a redistribution of stress to maintain equilibrium. In Test E3N2, normal
stress reduced at the crown, shoulders, springlines and left knee but increased slightly at the
invert of the existing tunnel. Even with the larger C/Ds of the existing and new tunnels in
Tests E3N5 and E5N3, normal stress exhibited trends similar to those in Tests E3N5 and
E5N3, respectively, differing only in magnitude. The reduction in normal stress at the
springlines, knees and invert in Test E3N5 was larger than that in Test E2N3. Similarly, large
stress reductions at the crown, left shoulder, left springline and left knee in Tests E5N3 than
E3N2 were observed. Further explanations are given later in the next section.
The results at the end of tunnel excavation (Y/D = 1.5) are shown in Figure 6.6b. Comparing
these results with those obtained when Y/D = 0.3, there was a significant reduction in normal
stress at the right knee in Tests E2N3 and E3N5. Similarly, the decrease in normal stress in
Tests E3N2 and E5N3 at the right shoulder at the end of tunnel excavation was much larger
than that when Y/D = 0.3.
Stress reduction at the invert in Test E3N5 was greater than that in Test E2N3. By using beam
theory, induced tensile strain in the lining at the invert of the existing tunnel in the former was
larger than that in the latter test (see Figure 6.2).
For the case when the new tunnel was excavated underneath the existing tunnel, the reduction
in vertical stress at the invert was greater with an increasing C/D of the existing tunnel from 2
to 3.5. The reduction in vertical stress in Tests E2N3 and E3N5 were 147 kPa and 195 kPa,
respectively. As a result of stress reduction, the decrease in stiffness was larger with an
increasing C/D of tunnels (refer to Figure 6.5). For Tests E2N3 and E3N5, a decrease in
mobilized shear modulus due to tunneling was 7.0 MPa and 9.4 MPa, respectively. The effects
of reduction in stress and stiffness in Test E2N3 were greater than that in Test E3N5, causing
131
larger tunnel settlement in the former than the latter (see Figure 6.1).
The normal stress reduction at the crown and shoulders of the existing tunnel when the new
tunnel was constructed above was larger at a greater depth (E5N3) than at a shallower depth
(E3N2). As a result, induced strain in the longitudinal direction of the existing tunnel were
observed in Test E5N3 than in Test E3N2 (refer to Figure 6.2).
In the tests where the new tunnel was constructed above the existing tunnel, the decreases in
vertical stress at the crown of the existing tunnel in Tests E3N2 and E5N3 were 161 kPa and
202 kPa, respectively. The decreases in stress resulted in stiffness reduction at the crown in
Tests E3N2 and E5N3 by 13.0 MPa and 16.1 MPa, respectively (see Figure 6.5). The effects
of decrease in stress and stiffness in Test E3N2 was less than that in Test E5N3, resulting in
smaller tunnel heave in the former test than the latter (refer to Figure 6.1).
The difference between incremental stress in the horizontal and vertical directions is one of
the factors influencing the deformation and induced strain in the transverse direction of the
existing tunnel. At the end of tunnel excavation in Tests E2N3 and E3N5, the reduction in
stress in the horizontal direction was larger than that in the vertical direction. As a result, the
existing tunnel was compressed vertically in these two tests (refer to Figure 6.3b and Figure
6.4b). On the other hand, the decrease in stress in the vertical direction was larger than that in
the horizontal direction in Tests E3N2 and E5N3, resulting in the vertical elongation of the
existing tunnel.
BD (2009) suggested that the change in stress acting on a tunnel lining should be within 20
kPa. In Tests E2N3 and E3N5, the reduction in normal stress at the springlines, knees and
invert as well as the increase in normal stress at the crown all exceeded the recommended
limit. For the tests where the new tunnel advanced above the existing tunnel (E3N2 and
E5N3), the decrease in normal stress at the crown, shoulders, springlines and knees all
exceeded the allowable stress. This suggests that a structural analysis of the existing tunnel
considering changes in stress should be undertaken. In addition, as the maximum change in
132
normal stress (e.g., stress reduction at the invert in Tests E2N3 and E3N5 as well as at the
crown in Tests E3N2 and E5N3) occurred during the advancement of the new tunnel but not
at the end of tunnel excavation, analysis of crossing-tunnel interaction should be carried out
three-dimensionally.
6.4.2 Mobilized vertical stress of soil acting on the existing tunnel
The ratio of vertical stress after tunnel excavation (v,after) to that before construction of the
new tunnel (v0) is referred to as the mobilized vertical stress ratio and is shown in Figure 6.7.
The ratio is considered for soil on the side of the existing tunnel facing the new tunnel (along
the invert in Tests E2N3 and E3N5 and along the crown in Tests E3N2 and E5N3).
The minimum mobilized vertical stress ratio of 0.33 was found in Test E2N3 at the center of
the existing tunnel (X/D = 0). This ratio increased with distance from the center of the
existing tunnel. In Test E3N2, the reduction in the mobilized vertical stress ratio was smaller
than that in Test E2N3 within the distance of 1D from the center of the existing tunnel (X/D
1). This smaller reduction in the mobilized vertical stress ratio and larger soil stiffness (refer
to Figure 6.5) resulted in significantly smaller vertical displacement and strain induced in the
lining of the existing tunnel in Test E3N2 than in Test E2N3 (see Figure 6.1 and Figure 6.2).
Despite the larger C/Ds of the existing and new tunnels in Test E3N5, the mobilized vertical
stress ratio in this test behaved in a similar manner to that in E2N3. The minimum vertical
stress ratio in both tests was about 0.35. Similarly, the mobilized vertical stress ratio in Test
E5N3 was almost identical to that in Test E3N2 with a minimum stress ratio of about 0.45.
Thus, the reduction in stress acting on the existing tunnel in the tests with the larger C/Ds was
greater than that in the tests with the smaller C/Ds (refer to Figure 6.6b) as the confining
pressure increased with depth. This suggests that the mobilized vertical stress ratio is
influenced by the effects of construction sequence but is independent of the effects of cover
depth on the crossing-tunnel interaction.
133
6.4.3 Arch length of new tunnel excavation
Figure 6.8 shows direction of principal stress in cases E2N3 and E3N2. The results are
illustrated when the location of the new tunnel face reach -0.9D away from the centerline of
the existing tunnel. Above and below the section of the new tunnel to be excavated (from Y/D
= -1.5 to -0.9), the size of the principal stresses were smaller than those outside of these areas.
It can be seen that the principal stress rotated around the zone of stress reduction. The extent
of principal stress rotation can be considered as the influence zone of tunnel excavation or the
arch length.
For the case the new tunnel was excavated underneath the existing tunnel as shown in Figure
6.8a (case E2N3), the arch lengths above and below the section of the new tunnel were 1D
and 0.4D, respectively. This arch length above the new tunnel extended to springlines of the
existing tunnel, causing reduction in stress on the existing tunnel in the horizontal direction
larger than that in the vertical direction as shown in Figure 6.6. As a result, the existing tunnel
was elongated horizontally (refer to Figure 6.3).
When the new tunnel was constructed above the existing tunnel in case E3N2 (Figure 6.8b),
the arch lengths were estimated to be 1.1D and 0.6D, respectively above and below the new
tunnel. The arch length below the new tunnel did not extend to the springlines of the existing
tunnel, resulting in larger reduction in the vertical stress than the horizontal stress (see Figure
6.6). These changes in stress on the existing tunnel caused vertical elongation of the tunnel (as
shown in Figure 6.3).
Two pairs of centrifuge tests were carried out to examine crossing-tunnel interaction. The
effects of construction sequence on crossing-tunnel interaction were investigated in the first
pair of tests, where a new tunnel was excavated beneath an existing tunnel in a reference test,
while the new tunnel advanced above in the other test. The second pair of tests were
134
conducted with larger cover depth-to-tunnel diameter ratios (C/Ds) than in the first pair to
study the effects of cover depth on the construction sequence in crossing tunnels. In order to
improve the understanding of the interaction of crossing tunnels, a hypoplasticity constitutive
model with small strain stiffness was adopted to back-analyze the tests. Based on the
interpretation of measured and computed results, the following conclusions may be drawn:
(a) The settlement of the existing tunnel when the new tunnel was excavated beneath was
significantly larger than the heave of the existing tunnel when the new tunnel was
constructed above. This is because the reduction in vertical stress acting on the existing
tunnel was larger in the former case than in the latter. In addition, the mobilized shear
stiffness of soil around the existing tunnel was smaller in the former tests than in the
latter, regardless of the C/Ds of the crossing tunnels.
(b) When the new tunnel was excavated beneath, the existing tunnel was compressed
vertically. However, the existing tunnel was elongated in the vertical direction due to the
new tunnel construction above. This is because the reduction in stress acting on the
existing tunnel in the horizontal direction was larger than that in the vertical direction
when the new tunnel was excavated underneath. On the other hand, the decrease in
stress acting on the existing tunnel in the vertical direction was larger than that in the
horizontal direction due to the new tunnel advancement above, irrespective of the C/Ds
of tunnels.
(c) The differential vertical displacement of the existing tunnel induced by the new tunnel
excavation underneath was larger than that when the new tunnel was constructed above
it. As a result, induced tensile strain and shear stress in the exiting tunnel lining in the
former was larger than that in the latter tunnel excavation sequence. The tensile strain
and shear stress in the existing tunnel induced by the new tunnel construction beneath
exceeded the cracking tensile strain (ACI, 2001) and allowable shear stress (ACI, 2011),
respectively.
135
(d) Settlement of the existing tunnel due to the new tunnel excavation underneath in the test
with the larger C/Ds was less than that in the test with the smaller C/Ds. This is because
the mobilized shear stiffness of soil around the existing tunnel increased with depth,
even though the reduction in vertical stress acting on the existing tunnel in the test with
the larger C/Ds was greater than that in the test with the smaller C/Ds. On the other
hand, heave of the existing tunnel caused by the new tunnel construction above
increased with increasing C/Ds of the existing and new tunnels due to a larger relief of
stress acting on the existing tunnel at a greater depth than at a shallower depth. In
addition, the increase in stress relief dominated the increase in the mobilized shear
stiffness of soil.
136
Table 6.1 Summary of tests interpreted to investigate the influence of construction sequence
and cover depth on crossing-tunnel interaction
Test ID C/D C/D Remark
(existing) (new)
E2N3 2.0 3.5
Centrifuge test and
E3N2 3.5 2.0
numerical back-analysis
E3N5 3.5 5.0
E5N3 5.0 3.5 Numerical analysis only
137
Table 6.2 Summary of maximum induced tunnel gradient

Test ID Induced tunnel gradient
Measured Computed
E2N3 1:1600 1:1850
E3N2 1:4300 1:3400
E3N5 1:1000 1:1880
E5N3 N/A 1:2800
Note: Allowable induced gradient is 1:1000 (LTA, 2000; BD, 2009)
Table 6.3 Summary of maximum shear stress induced on tunnel lining

Test ID Induced shear stress (kPa)
Measured Computed
E2N3 780 750
E3N2 200 500
E3N5 N/A 880
E5N3 N/A 510
Note: Allowable shear stress is 660 kPa (ACI, 2011)
138

-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
0.1
Heave
Normalized tunnel vertical
0.0
displacment (/D, %)
-0.1
Settlement
-0.2
15 mm (LTA, 2000)
-0.3 20 mm (BD, 2009)
-0.4
-0.5 Measured [E3N2] Computed [E3N2]
ref0 Computed [E5N3]
Field [Cooper et al., 2002] ref15
Field [Liu et al., 2011] ref20
Z Z Z Z
X X X X
New tunnel
Existing tunnel
New tunnel
New tunnel
Existing tunnel
New tunnel
E2N3 E3N2 E3N5 E5N3
Figure 6.1 Vertical displacement along the crown of the existing tunnel at the end of tunnel
excavation
139

-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
-100
longitudinal direction ( )
-50
Induced strain in
50
100
150
200 t, crack of unreinforced

concrete (ACI, 2001)
Ref0 Computed [E3N5]
Sign convention:
+ Tensile strain
Z Z Z Z
X X X X
New tunnel
Existing tunnel
New tunnel
New tunnel
Existing tunnel
New tunnel
E2N3 E3N2 E3N5 E5N3
Figure 6.2 Induced strain along the invert in the longitudinal direction of the existing tunnel
at the end of tunnel excavation
140
Change in normalized Crown

0.10 Y Y
tunnel radius
(R/R0 , %) New tunnel
Existing Y/D = 0.3
0.05 tunnel
E2N3
Initial
0.00 Existing
Y/D = 0.3
New tunnel tunnel
E3N5
-0.05 E2N3 E3N2
Y Y
L-Springline -0.10 R-Springline
New tunnel
Existing Y/D = 0.3
tunnel
E3N2
Existing
Y/D = 0.3
E5N3 New tunnel tunnel
E3N5 E5N3
Invert
Measured [E3N2] Computed [E3N2] Sign convention
Ref0r Computed [E5N3] + Tunnel radius increase
(a) - Tunnel radius decrease
tunnel radius 0.10
Y Y
(R/R0, %)
0.05 New tunnel
Existing Y/D = 1.5
tunnel
E2N3
0.00 Initial
Existing
Y/D = 1.5
E3N5 -0.05 New tunnel tunnel
E2N3 E3N2
Y Y
New tunnel
Existing Y/D = 1.5
tunnel
E3N2
Existing
Y/D = 1.5
E5N3 New tunnel tunnel
E3N5 E5N3
Invert
Ref0r Computed [E5N3]
(b)
Figure 6.3 Deformation of the existing tunnel (a) when the excavated section reached the
centerline of the existing tunnel (Y/D = 0.3); and (b) at the end of the tunnel excavation (Y/D
= 1.5)
141
Induced strain ()
Crown Y Y
150
New tunnel
100 Existing Y/D = 0.3
L-shoulder Initial R-shoulder tunnel
50
t of
0 Existing
unreinforced Y/D = 0.3
New tunnel tunnel
-50 concrete E2N3 E3N2

-100 (ACI, 2001)
Y Y
New tunnel
Existing Y/D = 0.3
tunnel
E2N3
L-knee R-knee E3N2 Y/D = 0.3

New tunnel
Existing
tunnel
E3N5 E5N3
E3N5
Invert Sign convention:
(a) + Tensile strain
Induced strain ()
Crown Y Y
150
New tunnel
100 Existing
tunnel
Y/D = 1.5
L-shoulder Initial R-shoulder

50
t of
0 unreinforced
Y/D = 1.5
New tunnel
Existing
tunnel
-50 concrete E2N3 E3N2

-100 (ACI, 2001) Y Y
New tunnel
Existing Y/D = 1.5
tunnel
E2N3
Existing
L-knee R-knee E3N2 Y/D = 1.5

New tunnel tunnel
E3N5 E5N3
E3N5
Invert
(b)
Figure 6.4 Induced strain measured on the outer surface of the existing tunnel in the
transverse direction (a) when the excavated section reached the centerline of the existing
tunnel (Y/D = 0.3); (b) at the end of tunnel excavation (Y/D = 1.5)
142
25 Z Z
X X
Mobilized secant shear modulus,
E5N3
New tunnel
20 Existing tunnel
Existing tunnel
E3N2 New tunnel
Gm (MPa)
15
E3N5
E2N3 E3N2
E2N3
Z Z
10
X X
Before tunnel excavation After tunnel excavation
[E2N3] -Before [E2N3] -After
5
[E3N2] -Before [E3N2] -After New tunnel
[E3N5] -Before [E3N5] -After Existing tunnel
Existing tunnel
0 New tunnel
0 1 2 3 4 E3N5 E5N3
(a)
25 Before tunnel excavation After tunnel excavation

Z Z
X X

20 [E3N5] -Before [E3N5] -After New tunnel
Existing tunnel
E5N3 Existing tunnel
New tunnel
Gm (MPa)
15
E2N3 E3N2
E3N2 Z Z
10
X X
E3N5
5
New tunnel
Existing tunnel
E2N3
0 Existing tunnel
New tunnel
0 1 2 3 4 E5N3
E3N5
(b)
Figure 6.5 Mobilized secant shear modulus computed along (a) the crown; (b) the invert of
the existing tunnel before and after the excavation of the new tunnel
143

stress n (kPa) -300
-250 Y Y
L-Shoulder -200 R-Shoulder
Initial New tunnel
-150 Existing
tunnel
Y/D = 0.3
-100
-50 Existing
Y/D = 0.3
0 = 20 kPa New tunnel tunnel
(BD, 2009) E3N2
50 E2N3
Y Y
E2N3
E3N2
New tunnel
E3N5 Existing Y/D = 0.3
tunnel
E5N3
L-Knee R-Knee Existing
Y/D = 0.3
New tunnel tunnel
E3N5 E5N3
Invert
(a)
Sign convention
stress n (kPa) -300
-250 Y Y
L-Shoulder -200 R-Shoulder New tunnel
-150 Existing Y/D = 1.5
tunnel
-100 Initial
-50
= 20 kPa Y/D = 1.5
Existing
0 (BD, 2009) New tunnel tunnel
50 E2N3 E3N2
Y Y
E2N3
E3N2
New tunnel
E3N5 Existing
tunnel
Y/D = 1.5
E5N3
L-Knee R-Knee Existing
Y/D = 1.5
New tunnel tunnel
E3N5 E5N3
Invert
(b)
Figure 6.6 Computed incremental normal stress acting on the existing tunnel (a) when the
excavated section reached the centerline of the existing tunnel (Y/D = 0.3); and (b) at the end
of the tunnel excavation (Y/D = 1.5)
144
Mobilized vertical stress ratio 1.4
1.2
1.0
('v, after/'v0)
0.8
0.6 E2N3
E3N2
0.4 E3N5
E5N3
0.2
0 1 2 3 4
along existing tunnel ( X/D)
New tunnel
Existing tunnel
Location of vertical stress considered
Location of vertical stress considered

Existing tunnel
New tunnel
For Tests E2N3 and E3N5 For Tests E3N2 and E5N3
Figure 6.7 Mobilized vertical stress ratio of soil located along the side of the existing tunnel
facing the new tunnel after the completion of the new tunnel construction
145
0 0
-1 -1
-2 -2
1.1D
Existing New tunnel
tunnel 0.6D
-3 -3
1D
-4 New tunnel -4 Existing
tunnel
0.4D
-5 -5
-6 -6
-2 -1 0 1 -2 -1 0 1
Normalized offset distance from existing tunnel centerline (Y/D)
(a) (b)
Existing tunnel 1.0D New tunnel
1.0D
0.4D
0.4D New tunnel Existing tunnel
E2N3 E3N2
Estimated arch length of new tunnel excavation
Figure 6.8 Computed direction of principal stress in cases (a) E2N3; (b) E3N2
146
Chapter 7 Effects of pillar depth and shielding on crossing multi-tunnel interaction
Chapter 7 Effects of pillar depth and shielding on

crossing multi-tunnel interaction
7.1 Introduction
Existing tunnels in the ground may experience excessive deformation and their linings may
show signs of cracking when new tunnels are excavated close to them. It is thus important to
consider ground movements and stress changes when constructing new tunnels close to
existing ones, especially in urban areas where more and more tunnels are being built with
greater proximity to each other.
From previous studies in Chapter 2, one of factors that may influence the interaction of
crossing tunnels is the pillar depth. Note that the pillar depth is the clear vertical distance
between each tunnel. As demonstrated in Chapter 5, the new tunnel caused adverse impact on
the existing tunnel such as large tunnel settlement and possible crack in the tunnel lining due
to tensile strain and shear stress. Thus, one of possible measures to mitigate the adverse
impact in crossing-tunnel interaction is to increase the pillar of the tunnels. In addition, the
interaction of crossing tunnels becomes more complex when the effects of new tunnel
excavation on one of the existing tunnel were shielded by another existing tunnel. Thus, it is
necessary to improve the understanding on the influence of pillar depth and the shielding
effects on crossing multi-tunnel interaction.
This chapter investigates the influence of pillar depth and shielding on the interaction of
crossing tunnels. To study the effects of pillar depth, the results in Test E2N5 (refer to Figure
3.6) was interpreted along with that in Test E2N3 (as shown in Figure 3.3). The pillar
147
depth-to-diameter ratios (P/D) between the existing and new tunnels in Tests E2N3 and E2N5
were 0.5 and 2, respectively. The shielding effects were examined by comparing the results in
Test E2,3N5 (refer to Figure 3.7), which featured two (upper and lower) existing tunnels
above the new tunnel, with those from Test E2N5. The term shielding effects is used to
consider the presence of a lower existing tunnel that potentially reduces the impact of a new
tunnel excavation on the upper existing tunnel. The C/Ds of the existing and new tunnels are
summarized in Table 7.1.
7.2 Existing tunnel responses due to new tunnel excavation with effects of
pillar depth and shielding
7.2.1 Settlement of the existing tunnel
In order to compare results from this study with a case history, the tunnel settlement was
normalized by the diameter of the new tunnel. Figure 7.1 compares the measured and
computed normalized settlements of the existing tunnel at the end of tunnel advancement.
The maximum measured tunnel settlement in Reference Test E2N3 was about 0.3%D (i.e., 18
mm), which exceeds the recommended serviceability limit of 15 mm (LTA, 2000). For the
two-tunnel interaction, the maximum measured tunnel settlement in Test E2N3 (i.e., P/D = 0.5)
was about 50% larger than that in Test E2N5 (i.e., P/D = 2). The larger tunnel settlement in
Test E2N3 was mainly due to a larger reduction in vertical stress and lower normalized soil
stiffness along the invert of the existing tunnel. Detailed explanations are given later in
Section 7.4.2.
As for the three-tunnel interaction, the maximum measured settlement of the upper existing
tunnel in Test E2,3N5 was about 25% smaller than that in Test E2N5 (provided that P/D = 2
for both tests) due to the presence of the lower existing tunnel (i.e., shielding effects). Further
away from the centerline of the new tunnel, the difference in tunnel settlement between Tests
148
E2N5 and E2,3N5 narrowed.
The induced tunnel gradient can be deduced from the slope of measured tunnel settlements.
The largest induced tunnel gradient of 1:1600 was observed in Test E2N3, where the largest
tunnel settlement occurred. The maximum induced tunnel gradients of the three tests all fell
within the recommended limit of 1:1000 (LTA, 2000; BD, 2009).
In Test E2N3, the computed tunnel settlements were underestimated by 16 % at the location
directly above the new tunnel. On the other hand, they were overestimated in Tests E2N5 and
E2,3N5 by 8 % and 12 %, respectively, at the same location. This discrepancy may be due to
the fact that some model parameters were obtained from the literature and empirical
relationships. Although there were discrepancies between the measured and the computed
results, both results show the same trend.
Cooper et al. (2002) reported settlement of an existing tunnel due to the excavation of a new
tunnel with P/DNewTunnel of 0.8 in London Clay. Note that the centrifuge test results in the
present study cannot be quantitatively compared with the field monitoring data. The general
trend was that when a new tunnel was excavated at P/D less than 1, settlement of an existing
tunnel may exceed the allowable limit. The ground conditions, tunneling method and the
flexural rigidity of the existing tunnel in Cooper et al.s case study are given in Table 2.1.
Figure 7.2 compares the measured and computed strain induced along the invert of the
existing tunnel at the end of tunnel advancement. The positive and negative signs denote
induced tensile strain and induced compressive strain at the invert of the existing tunnel,
respectively.
Due to differential settlement of the existing tunnel, sagging moment was induced directly
above the new tunnel. As a result, tensile strain was induced at the invert of the existing
149
tunnel. The measured maximum induced tensile strain of 152 was found in Test E2N3,
exceeding the cracking tensile strain of 150 for unreinforced concrete (ACI, 2001). The
measured maximum induced tensile strain in Test E2N5 (127 ) was 16% smaller than that
in Test E2N3, where the former test had a larger P/D than the latter test. The measured
maximum induced tensile strain in the upper existing tunnel in Test E2,3N5 (86 ) was 36%
smaller than that in Test E2N5. This is due to shielding effects provided by the lower existing
tunnel.
The induced strain in the longitudinal direction of the existing tunnel was consistent with the
tunnel settlement in every test (refer to Figure 7.1). The possible reason for discrepancies
between the measured and computed results has been discussed in the previous section.
Shear stress acting on the tunnel lining was deduced by differentiating bending moment,
which was converted from induced strain along the invert of the existing tunnel. At a given
concrete compressive strength (fc) of 50 MPa and a reduction factor of 0.55, the allowable
shear stress was estimated to be 660 kPa (ACI, 2011). The maximum deduced shear stress in
Tests E2N3 (for both measured and computed results) exceeded the allowable limit,
suggesting that the tunnel lining may crack.
There was a high possibility that cracking would occur in Test E2N3 as the induced tensile
strain and shear stress exceeded the cracking tensile strain and the allowable shear stress,
respectively. These cracks are likely to appear if the tunnel lining is made of concrete within
an offset distance of 3.5D from the centerline of the new tunnel.
Figure 7.3 shows the measured strain induced at the outer surface of the existing tunnel in the
transverse direction. As the strain induced in the first three excavation stages (i.e., from Y/D =
-1.5 to -0.3) was smaller than that induced in the last three excavation stages (i.e., from Y/D =
150
0.3 to 1.5) and to simplify presentation, the results from the former are not shown.
In Test E2N3 (Figure 7.3a), induced compressive strain was at shoulders, the right knee and
the invert. Tensile strain was induced at both springlines. The induced strain in Test E2N3
indicates that the existing tunnel was compressed vertically. To verify this vertical
compression, tunnel deformation is shown and discussed later.
Strain induced by tunneling caused tunnel deformations in different directions in Test E2N3
and Test E2N5 (Figure 7.3b), where the former test had a smaller P/D than the latter test. In
Test E2N5 tensile strain was induced at the crown and invert while compressive strain was
induced at both springlines. Larger compressive strain and tensile strain were induced at both
springlines and the invert, respectively, when Y/D = 0.9D than at the end of tunneling. The
induced strain in Test E2N5 suggests that the existing tunnel was elongated vertically. The
tunnels deformed in different directions in Tests E2N3 and E2N5 because of the varying
incremental normal stresses acting on the existing tunnel with different P/Ds. Further
explanations are given later in Section 7.4.1.
In Test E2,3N5 the P/D between the new tunnel and the upper existing tunnel was 2, identical
to that in Test E2N5. Thus, the strain induced in the upper existing tunnel in Tests E2,3N5
(Figure 7.3c) and that induced in the existing tunnel in Test E2N5 (Figure 7.3b) were similar
in terms of both magnitude and trend. As expected, the two tunnels were both elongated
vertically.
Induced strain in the lower existing tunnel in Test E2,3N5 is shown in Figure 7.3d.
Compressive strain was induced at the crown and both springlines while induced tensile strain
was found at the shoulders, knees and invert. The induced strain suggests that the lower
existing tunnel was elongated vertically.
Given that the induced compressive strain equaled the induced tensile strain on opposite
151
surfaces of the tunnel lining, the tensile strain was at the maximum of 170 at the right
springline on the inner surface of the existing tunnel in Test E2,3N5 (refer to Figure 7.3d).
This induced tensile strain exceeded the cracking tensile strain of 150 (ACI, 2001),
suggesting that the inner surface of the tunnel lining may crack.
Figure 7.4 compares the measured and computed deformation of the existing tunnel. The
change in the normalized diameter of an existing tunnel (D/D0), where D0 is the initial
diameter of the tunnel, was measured at the end of tunnel advancement.
Figure 7.4a shows the change in the normalized vertical diameter of the existing tunnel.
According to the measured results, the existing tunnel in Test E2N3 (where P/D = 0.5) was
compressed vertically, while that in Test E2N5 and the upper existing tunnel in Test E2,3N5
(where P/D = 2 in both cases) were vertically elongated. The tunnel deformation differed
from one P/D to another because of the varying incremental stresses acting on the existing
tunnel. More explanations are given later in Section 7.4.1. In Test E2,3N5, the upper existing
tunnel was elongated in the vertical direction as the P/D was equal to that in Test E2N5. The
lower existing tunnel in Test E2,3N5 was also elongated vertically even though the P/D was
just 0.5. This is because its own shielding effects reduced the effect of stress transfer in the
longitudinal direction of the new tunnel. This are further discussed later. The result of tunnel
deformation was consistent with induced strain in the transverse direction of the existing
tunnel (see Figure 7.3). Some discrepancies between measured and computed deformation of
tunnel were observed. The reason for these discrepancies has been discussed previously in
Section 7.2.1.
Figure 7.4b shows the change in the normalized horizontal diameter of the existing tunnel.
Both measured and computed results show that the existing tunnel was elongated horizontally
in Test E2N3. On the other hand, the existing tunnels in Tests E2N5 and E2,3N5 were
152
compressed horizontally. This is expected as change in tunnel diameter in the horizontal
direction is normally in the opposite direction to that in the vertical direction.
BTS (2000) recommended that the difference between the maximum and minimum diameters
of a tunnel be within 2% [i.e., (Dmax - Dmin)/D0 2%]. Deformations of the existing tunnels in
this study were still within the allowable limit. Note that the results reported here were due to
tunneling only. Given that in-situ vertical stress was larger than horizontal stress (i.e., K0 < 1),
the existing tunnel was compressed vertically even before the advancement of the new tunnel.
Thus, the induced vertical compression of the existing tunnel in Test E2N3 may still cause
structural and serviceability problems.
7.3 Influence of pillar depth and shielding on soil stiffness
To explain the difference in tunnel settlement among the three cases, the normalized soil
stiffness along the invert of the existing tunnel was calculated as shown in Figure 7.5. The
normalized soil stiffness was considered before tunnel excavation (once the centrifugal
acceleration had reached 60g) and after tunnel completion. Secant shear moduli (Gbefore and
Gafter) were calculated from deviatoric stress and deviatoric strain [G = q/(3s)] at the end of
each stage. The normalized soil stiffness differed among the three cases because the
hypoplasticity model can simulate the dependence of stiffness on the state, strain and recent
stress history of the soil.
In terms of two-tunnel interaction, the normalized soil stiffness was lower in Reference case
E2N3 than in case E2N5 at the center of the existing tunnel (X/D = 0), resulting in larger
tunnel settlement (see Figure 7.1). This is because the reduction in confining stress and the
increase in deviatoric strain due to tunnel excavation were larger in case E2N3 than in case
E2N5. In addition, an increase in stiffness of soil was observed at an offset distance between
2D and 4D from the center of the existing tunnel. This is because vertical stress increase at
those locations due to stress redistribution, resulting in increase in stiffness. Further
153
explanations are given later in next section.
As for three-tunnel interaction, there was almost no change in normalized stiffness at the
centerline of the new tunnel in case E2,3N5 whereas the normalized stiffness was
significantly reduced in case E2N5 (given that P/D = 2 in both tests). The minimum
normalized stiffness in case E2,3N5 was found at an offset distance of 0.5D, which is the
offset distance of the springline of the lower existing tunnel, due to the shielding effects of the
lower existing tunnel.
7.4 Stress and strain of soil caused by new tunnel excavation
Figure 7.6 shows the computed incremental normal stress acting on the existing tunnel in the
transverse direction at the location directly above the new tunnel (i.e., X/D = 0). The
incremental stress in this study is defined as the difference between the stress at the end of a
tunnel excavation stage and that when centrifugal acceleration had reached 60g.
In case E2N3 (see Figure 7.6a), normal stress increased gradually at the crown as the new
tunnel advanced. This is because stress was transferred in the longitudinal direction of the
new tunnel to maintain stress equilibrium (Ng and Lee, 2005). At both springlines, there was a
reduction in normal stress. At the invert, once the new tunnel approached the existing tunnel
(i.e., Y/D = -0.3), a slight increase in normal stress was observed. As the new tunnel advanced
further (i.e., from Y/D = 0.3 to 1.5), normal stress dropped sharply.
In case E2N5 (see Figure 7.6b), there was an increase in normal stress at the crown. At both
springlines, the reduction in normal stress was smaller than that in case E2N3 as the P/D for
the latter was smaller. At the invert, normal stress reduced substantially as the new tunnel
advanced.
154
The fact that the existing tunnel deformed in different directions in cases E2N3 and E2N5
(refer to Figure 7.4a and Figure 7.4b) can be explained by the difference in stresses in the
horizontal and vertical directions. In case E2N3, the stress reduction on the existing tunnel in
the horizontal direction was larger than in the vertical direction, resulting in vertical
compression of the existing tunnel. On the other hand, the decrease in stress in the vertical
direction was larger than in the horizontal direction in case E2N5 and caused the existing
tunnel to elongate in the vertical direction.
One of reasons for the larger tunnel settlement in Test E2N3 than that in Test E2N5 (refer to
Figure 7.1), where the two tests had different P/Ds, was because the stress reduction at the
invert and the stress increase at the crown of the existing tunnel at the location directly above
the new tunnel were both larger in the former test (see Figure 7.6a and Figure 7.6b).
For three-tunnel interaction, incremental normal stress acting on the upper existing tunnel in
case E2,3N5 (see Figure 7.6c) was smaller than case E2N5. This is because the presence of
the lower existing tunnel in case E2,3N5 reduced the change in normal stress at every part of
the upper existing tunnel compared with that in case E2N5. The stress reduction on the upper
existing tunnel in case E2,3N5 in was larger in the vertical direction than the horizontal
direction at the end of tunnel excavation. It caused the existing tunnel to elongate vertically
(refer to Figure 7.4a).
Figure 7.6d shows the incremental normal stress acting on the lower existing tunnel in case
E2,3N5. Although this tunnel minimized stress redistribution in the longitudinal direction of
the new tunnel, stress was still transferred in the transverse direction of the new tunnel
causing stress to increase at the crown of this tunnel. At each springline, a decrease in normal
stress was observed. At the invert, a sharp reduction in normal stress occurred at the
completion of tunneling. The decrease in stress in the vertical direction was larger than in the
horizontal direction at the completion of new tunnel excavation. As a result, the lower existing
155
tunnel was elongated vertically (see Figure 7.4a).
The allowable limits of stress change for a tunnel lining suggested by BD (2009) are shown in
Figure 7.6. It can be seen that the change in normal stress exceeded the allowable limit of 20
kPa in all cases. This suggests that the structural analysis considering these changes of stress
acting on the tunnel linings should be reviewed.
7.4.2 Incremental normal stress in the longitudinal direction of the existing tunnel
Figure 7.7 shows the computed incremental normal stress along the crown and invert of the
existing tunnel at the end of tunnel excavation. The positive and negative signs denote
increases and decreases in stress, respectively, compared with that prior to tunnel
advancement.
Figure 7.7a shows incremental normal stress along the crown of the existing tunnel. The
maximum increase in normal stress was observed in case E2N3 at the location directly above
the new tunnel (i.e., X/D = 0). With increasing offset distance from the centerline of the new
tunnel, incremental normal stress in this test gradually decreased. A similar trend was seen in
case E2N5. The increase in normal stress at the crown of the upper existing tunnel directly
above the new tunnel in Test E2,3N5 was only half of that in case E2N5, given that P/D = 2 in
both tests. This may be due to the shielding effects which would have reduced the transfer of
stress to the crown of the upper existing tunnel in the three-tunnel interaction test.
Incremental normal stress along the invert of the existing tunnel is illustrated in Figure 7.7b.
As the P/D in case E2N3 was smaller than that in case E2N5, the maximum stress reduction
in the former case was larger than that in the latter case. Due to the shielding effects of the
lower existing tunnel in case E2,3N5, the maximum reduction in stress on the upper existing
tunnel was significantly smaller than that in case E2N5 (given the same P/D). At a distance
away from the centerline of the new tunnel, stress increase was observed in every case as a
156
result of stress redistribution. The maximum stress increases were found at offset distances of
about 2D, 2.5D and 3D from the centerline of the new tunnel in cases E2N3, E2N5 and
E2,3N5, respectively. The reduction in normal stress (i.e., confining stress) and the increase in
deviatoric strain (to be discussed later) on the existing tunnel at the centerline of the new
tunnel was the largest in case E2N3. As a result, the normalized soil stiffness was the smallest
(see Figure 7.5).
The major reason for the larger tunnel settlement in Test E2N3 than that in Test E2N5 (refer to
Figure 7.1), where the two tests had different P/Ds, was because the stress reduction at the
invert and the stress increase at the crown of the existing tunnel at the location directly above
the new tunnel were both larger in the former test. The larger tunnel settlement in Test E2N5
than that in Test E2,3N5 can also be thus explained.
The increase in stress along the crown of the existing tunnel in cases E2N3 and E2N5 slightly
exceeded the recommended limit of 20 kPa (BD, 2009) up to an offset distance of 1D from
the centerline of the new tunnel (see Figure 7.7a). Along the invert of the existing tunnel, the
reduction in normal stress exceeded the recommended limit up to a distance of 1D from the
centerline of the new tunnel in all three tests (refer to Figure 7.7b). Due to stress redistribution
along the invert of the existing tunnel, the increase in stress exceeded the allowable limit up to
a distance of X/D = 4 in all three cases. This suggests that before excavating a new tunnel
close to an existing one, the structural capacity of the existing tunnel should be reviewed up to
an offset distance of 4D from the centerline of the new tunnel.
7.4.3 Induced deviatoric strain of soil
Figure 7.8 shows the computed induced deviatoric strain of soil at the end of tunnel
advancement. The positive and negative signs denote increases and decreases in deviatoric
strain, respectively, compared with that prior to tunnel excavation (at 60g).
157
In case E2N3 (as shown in Figure 7.8a), the maximum induced deviatoric strain was found at
the invert of the existing tunnel directly above the new tunnel (i.e., at X/D = 0). This
maximum deviatoric strain resulted in the largest reduction in soil stiffness at the centerline of
the new tunnel (see Fig. 6), in addition to the largest normal stress reduction (refer to Figure
7.7b).
Figure 7.8b shows the induced deviatoric strain in case E2N5. The magnitude and distribution
of induced deviatoric strain along the invert of the existing tunnel in case E2N5 were similar
to those in case E2N3 (Figure 7.8b). This suggests that the smaller normalized soil stiffness in
case E2N3 than that in case E2N5 (refer to Figure 7.5) was mainly due to the larger reduction
in normal stress at the invert of the existing tunnel in the former case (as shown in Figure
7.7b).
Induced deviatoric strain in case E2,3N5 is illustrated in Figure 7.8c. The induced deviatoric
strain at the invert of the upper existing tunnel at the centerline of the new tunnel in case
E2,3N5 was smaller than that in case E2N5. This is because the lower existing tunnel
shielded the upper existing tunnel from the deviatoric strain induced by the new tunnel
excavation. This smaller induced deviatoric strain in case E2,3N5 was another reason for the
smaller reduction in normalized soil stiffness at the invert of the existing tunnel (see Figure
7.6).
7.4.4 Induced deviatoric strain of soil
Figure 7.9 shows the computed direction of principal stress of soil caused by new tunnel
excavation in cases E2N3 and E2N5. The results are shown when the face of the new tunnel
was -0.9D away from the centerline of the existing tunnel. It can be seen from Figure 7.9a and
b that above and below the section to be excavated (Y/D of -0.9), size of the principal stress
was smaller than outside of these zones. In addition, principal stresses above and below the
section to be excavated were rotated towards the middle of the section. The arch length of
158
new tunnel excavation can be estimated by considering zone of principal stress rotation. From
the estimation, the arch lengths above and below the section to be excavated were 1D and
0.4D, respectively.
In case E2N3 (Figure 7.9a), the arch length extended to the springline of the existing tunnel
resulting in substantial reduction on the existing tunnel in stress in the horizontal direction
(refer to Figure 7.6a). This reduction in stress in the horizontal direction larger than that in the
vertical direction caused horizontal elongation of the existing tunnel (see Figure 7.4).
When P/D increased to 2 for case E2N5 (Figure 7.9b), the arch length did not extended to the
springlines of the existing tunnel, causing decrease in stress in the vertical direction larger
than that in the horizontal direction (refer to Figure 7.6b). As a result, the existing tunnel was
elongated in the vertical direction (see Figure 7.4).
In Three-dimensional centrifuge model tests and numerical back-analyses were conducted to
investigate crossing multi-tunnel interaction. Test E2N3 was considered as a reference test in
which a new tunnel advanced perpendicularly beneath an existing tunnel. Tests E2N5 and
E2,3N5 were carried out to investigate the effects of the pillar depth-to-diameter ratio (P/D)
and shielding on multi-tunnel interaction, respectively. Based on the interpretation of the
measured and computed results, the following conclusions may be drawn:
(a) In the case of two perpendicularly crossing tunnels (one new and one existing), the
measured maximum tunnel settlement at P/D of 0.5 was about 50% larger than that at
P/D of 2.0. This is attributed to a smaller shear stiffness of soil in the case of P/D of 0.5
along the invert of the existing tunnel. The mobilized soil stiffness was the smallest at
the location directly above the new tunnel as a result of a reduction in confining stress
and an increase in deviatoric strain caused by the new tunnel excavation. The other
159
contributing factor to the larger tunnel settlement at P/D of 0.5 is the stress acting on the
tunnel lining at the location directly above the new tunnel. In the test with P/D of 0.5,
stress reduction at the invert of the existing tunnel was larger than that in the test with
P/D of 2.
(b) During the new tunnel excavation, induced tensile strains in the longitudinal direction of
the existing tunnel and deduced shear stress on the tunnel lining were larger at P/D of
0.5 than at P/D of 2. This is due to the larger differential settlement of the existing
tunnel at P/D of 0.5. These induced tensile strain and deduced shear stress at P/D of 0.5
exceeded the cracking tensile strain (ACI, 2001) and allowable shear stress (ACI, 2011),
respectively.
(c) Different tunnel deformation mechanisms were observed. The existing tunnel was
elongated horizontally at P/D of 0.5. This is because stress reduction on the existing
tunnel in the horizontal direction was larger than that in the vertical direction. It should
be pointed out that stress relief caused by the new tunnel excavation at P/D of 0.5 not
only caused a reduction in vertical stress but also it resulted in substantial stress
reduction at each springline of the existing tunnel. On the contrary, the existing tunnel
was elongated vertically as the new tunnel advanced at P/D of 2, because stress relief
was dominated in the vertical direction and it mainly affected the invert of the existing
tunnel.
(d) In the case of three tunnels (two existing perpendicularly crossing tunnels above a new
tunnel), the lower existing tunnel shielded the upper existing tunnel from the
influence of the advancing new tunnel underneath, such that the measured settlement of
the upper existing tunnel was 25% smaller than in the case without the shielding effects
(given that P/D of 2 in both cases). This is because the lower existing tunnel reduced the
effect of stress reduction and decreased deviatoric strain induced at the invert of the
160
upper existing tunnel. These two effects resulted in a larger mobilized soil stiffness in
the case of two existing tunnels than in the case of just one existing tunnel.
(e) The lower existing tunnel in the case of three tunnels was elongated vertically due to the
new tunnel excavation. This is because the invert of the lower existing tunnel was
closest to each section to be excavated of the new tunnel, resulting in a substantial
decrease in stress in the vertical direction on the lower existing tunnel.
161
Table 7.1 Summary of tests interpreted to investigate the effects of pillar depth and shielding
on crossing multi-tunnel interaction
Test ID C/D C/D Brief description of each test
(existing) (new)
E2N3 2.0 3.5 Reference test
E2N5 2.5 5.0 Increase P/D from 0.5 to 2
E2,3N5 2.0, 3.5 5.0 Shielding effects provided by the lower
existing tunnel on the upper existing tunnel
162

-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
0.0
0.1
settlement (/D, %)
Normalized tunnel
0.2
15 mm (LTA, 2000)
Max. induced
0.3 20 mm (BD, 2009) tunnel gradient
E2N3 = 1:1600
E2N5 = 1:2700
0.4 E2,3N5 = 1:2800
0.5 Measured [E2N5] Computed [E2N5]
Measured [E2,3N5] Computed [E2,3N5]
Field [Cooper et al., 2002]
Z Z Z
X X X
Note: In Test E2,3N5, results are
shown for the upper existing tunnel
Existing tunnel Existing tunnel Upper existing tunnel
only
Lower
existing
tunnel
New tunnel
New tunnel New tunnel
E2N3 E2N5 E2,3N5
Figure 7.1 Settlement of the existing tunnel
163

-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
-100
-50
Induced strain ()
0
Deduced Vmax
50 (kPa) from
measured and
computed strain
100 = 780, 690 [E2N3] t, crack of unreinforced
= 510, 560 [E2N5] concrete (ACI, 2001)
= 330, 375 [E2,3N5]
150
Vallow (kPa)
= 660 (ACI, 2011)
200
Z Z Z
X X X
Sign convention
+ Induced tensile strain
- Induced compressive strain
Lower
existing
tunnel
New tunnel
E2N3 E2N5 E2,3N5
Note: In Test E2,3N5, results are shown

for the upper existing tunnel only
Figure 7.2 Induced strain in the longitudinal direction of the existing tunnel
164
Crown Induced strain ()

100
50 Z
L-shoulder R-shoulder
0 Y
-50
Initial
-100 Existing
tunnel
-150
New tunnel
Location of new Y/D 0.3 1.5
0.9
tunnel face (Y/D)
0.3
0.9
1.5 E2N3
L-knee R-knee
Invert
(a)

100
50 Z
0
Y
-50
Initial
-100 Existing
-150 tunnel

Location of new New tunnel
tunnel face (Y/D)
0.3 Y/D 0.3 1.5
0.9 0.9
1.5 E2N5
L-knee R-knee
Invert
(b) Sign convention

+ Induced tensile strain
165

100
50 Z
0
Y
-50
-100 Initial
Upper existing
-150 tunnel
L-springline -200 R-springline Lower existing

Location of new tunnel
New tunnel
tunnel face (Y/D)
0.3
Y/D 0.3 1.5
0.9 0.9
1.5 E2,3N5
L-knee R-knee
Invert
(c)
100
Z
50 Y
0
Upper existing
-50 tunnel
Initial
-100
Lower existing
-150 tunnel
New tunnel
Location of new Y/D 0.3 1.5
0.9
tunnel face (Y/D)
0.3 E2,3N5
0.9
1.5
L-knee R-knee
Invert
Sign convention
(d) + Induced tensile strain
Figure 7.3 Measured strain induced on the outer surface of the existing tunnel in the
transverse direction in Tests (a) E2N3; (b) E2N5; (c) E2,3N5 upper tunnel; (d) E2,3N5 lower
tunnel
166
0.2
Change of normalized vertical
tunnel diameter (DV/D0, %)
0.1
DV
Z
0.0
Y
Sign convention
-0.1 Existing
+ Tunnel diameter elongation tunnel
- Tunnel diameter compression

-0.2
0.0 0.5 1.0 1.5 2.0 New tunnel
Pillar depth-to-diameter ratio (P/D)

E2N3
Measured [E2,3N5 UpperTunnel] Computed [E2,3N5 UpperTunnel]
Z
Measured [E2,3N5 LowerTunnel] Computed [E2,3N5 LowerTunnel]
Y
(a)
Existing
tunnel
Change of normalized horizontal
0.2
tunnel diameter (DH/D0, %)
0.1 New tunnel

DH E2N5
0.0
Z
Sign convention Y
-0.1
+ Tunnel diameter elongation
Upper existing
- Tunnel diameter compression tunnel
-0.2
0.0 0.5 1.0 1.5 2.0 Lower existing
tunnel
Pillar depth-to-diameter ratio (P/D)
Measured [E2N3] Computed [E2N3] New tunnel
Measured [E2,3N5 UpperTunnel] Computed [E2,3N5 UpperTunnel] E2,3N5
Measured [E2,3N5 LowerTunnel] Computed [E2,3N5 LowerTunnel]
(b)
Figure 7.4 Deformation of the existing tunnel in (a) the vertical direction; (b) the horizontal
direction
167
1.2
Normalized shear modulus,
1.0
0.8
Gafter/Gbefore
0.6
E2N3
E2N5
0.4
E2,3N5
0.2
0.0
0 1 2 3 4 5 6
along invert of existing tunnel (X/D)
Note: In Test E2,3N5, results are shown for
the upper existing tunnel only
Z Z Z
X X X
Lower
existing
tunnel
New tunnel
E2N3 E2N5 E2,3N5
Figure 7.5 Computed normalized stiffness of soil (Gafter / Gbefore) along the invert of the
existing tunnel
168
150 = 20 kPa (BD, 2009)

Incremental normal stress, n (kPa)
100
50 Z
Y
0
-50 Existing
tunnel
-100 New tunnel

Crown
Y/D 1.5 0.3 1.5
-150 L-Springline
R-Springline
-200
Invert E2N3
-250
-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5
(a)
150 = 20 kPa (BD, 2009)

100
50
Z
0 Y
-50
Existing
tunnel
-100
Crown
-150 L-Springline New tunnel
R-Springline
-200 Y/D 1.5 0.3 1.5
Invert
-250
E2N5
-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5
(b)
Sign convention for
169
150 = 20 kPa (BD, 2009)

100
50 Z
Y
0
Upper existing
-50 tunnel
-100 Lower existing

Crown tunnel
New tunnel
-150 L-Springline
Y/D 1.5 0.3 1.5
R-Springline
-200
Invert E2,3N5
-250
-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5
(c)
150 = 20 kPa (BD, 2009)

100
50 Z
0 Y
-50 Upper existing

tunnel
-100
Lower existing
Crown tunnel
-150 New tunnel
L & R-Springline
Y/D 1.5 0.3 1.5
-200 Invert
E2,3N5
-250
-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5
(d) Sign convention

Figure 7.6 Computed incremental normal stress of the existing tunnel in cases (a) E2N3; (b)
E2N5; (c) E2,3N5 upper tunnel; (d) E2,3N5 lower tunnel
170
50 = 20 kPa (BD, 2009)

Incremental normal stress,
0
n (kPa)
-50
Z Z Z
X X X
-100
E2N3
Lower
existing
-150 New tunnel
tunnel
E2N5
E2,3N5
E2N3 E2N5 E2,3N5
-200
0 1 2 3 4 5 6
crown of existing tunnel (X/D)
(a)
50 = 20 kPa (BD, 2009)

Incremental normal stress,
0
n (kPa)
-50
Z Z Z
X X X
-100
Lower E2N3
existing
-150 New tunnel
tunnel
E2N5
E2N3 E2N5 E2,3N5 E2,3N5
-200
0 1 2 3 4 5 6
invert of existing tunnel (X/D)
(b)
Figure 7.7 Computed incremental normal stress at the end of tunnel excavation in the
longitudinal direction along (a) the crown and (b) the invert of the existing tunnel
171
Existing tunnel
New tunnel
(a)
Existing tunnel
New tunnel
(b)
Sign convention
+ Increase in deviatoric strain
- Decrease in deviatoric strain
172
Upper existing tunnel
Lower existing tunnel
New tunnel
(c)
Sign convention
+ Increase in deviatoric strain
- Decrease in deviatoric strain
Figure 7.8 Contours of computed deviatoric strain induced by new tunnel advancement in
cases (a) E2N3; (b) E2N5; (c) E2,3N5
173
0 0
-1 -1
-2 -2
Existing Existing
tunnel tunnel
-3 -3
1D
-4 New tunnel -4
0.4D
-5 -5
1D
New tunnel
-6 -6
-2 -1 0 1 -2 -1 0 1
Normalized offset distance from existing tunnel centerline (Y/D)
(a) (b)
1.0D
1.0D
0.4D New tunnel
0.4D New tunnel
E2N3 E2N5
Estimated arch length of new tunnel excavation
Figure 7.9 Computed direction of principal stress in cases (a) E2N3; (b) E2N5
174
Chapter 8 Responses of an existing tunnel due to twin tunnel excavation across underneath
Chapter 8 Responses of an existing tunnel due to

twin tunnel excavation across underneath
8.1 Introduction
To increase the use of underground space for infrastructural needs while minimizing
environmental impacts, an increasing number of closely spaced multiple tunnels are being
constructed in densely urban areas worldwide. Parallel twin tunnels are commonly driven in
side-by-side or vertically stacked arrangements, depend on the available underground space. It
is important to understand the response of an existing tunnel due new twin tunnel excavation.
Another factor that should be considered is the suitable arrangement of the new twin tunnels
and sequence of excavation that minimizes the adverse impact on the existing tunnel.
This chapter presents the behavior of an existing tunnel due to new twin tunnel excavation
with different arrangements across underneath. Responses of an existing tunnel due to
side-by-side new twin tunnel excavation (Test E2N3,3; as refer to Figure 3.8) and vertically
stacked new twin tunnel construction (Test E2N3,5; as shown in Figure 3.9) were interpreted.
The C/Ds of the existing and new twin tunnels are summarized in Table 8.1. In addition,
numerical parametric study was conducted to investigate the effects of arrangement and
excavation sequence of the new twin tunnels on the existing tunnel. The numerical procedure
to investigate the effects of arrangement and tunneling sequence on the existing tunnel is
given in Section 4.6.2.2 and detail of numerical run is summarized in Table 4.1.
175
8.2 Impact of twin tunnel excavation on an existing tunnel
8.2.1 Settlement and gradient of the existing tunnel
Figure 8.1 shows the settlement of the existing tunnel normalized with the new tunnel
diameter. The results were obtained from centrifuge tests and numerical back-analysis at the
end of each new tunnel excavation.
Settlement of the existing tunnel due to side-by-side new twin tunnel excavation underneath
the existing one (Test E2N3,3) is shown in Figure 8.1a. The maximum measured settlement of
the existing tunnel caused by the left (first) new tunnel excavation was within the
recommended limit. However, the maximum settlement of the existing tunnel after
completion of the right (second) new tunnel exceeded both limits recommended by LTA (2000)
and BD (2009). Note that the maximum tunnel settlement caused by the second tunnel only
was less than that caused by the first tunnel. This is because stress reduction underneath the
invert of the existing tunnel due to the second new tunnel was smaller than the first one.
Further explanation is given later in Section 8.3.1.
The numerical-back analysis results were qualitatively in agreement with the measured one
with minor discrepancies. These discrepancies may be because some model parameters were
adopted from the literature and empirical relationships.
Settlement of an existing tunnel caused by side-by-side new tunnels in London Clay was
reported by Cooper et al. (2002). The outer diameters of the existing and new tunnels were 4.1
m and 9.1 m, respectively. The pillar depth between the existing and new tunnels was 7 m.
The volume loss caused by the new tunnel was estimated to be 1.3-2.5%. Asymmetry of
settlement of the existing tunnel induced by the first tunnel (centerline of the tunnel at X/D =
1.37) and the second tunnel (centerline of the tunnel at X/D = -1.37) was observed.
Figure 8.1b shows settlement of the existing tunnel caused by excavation of vertically stacked
176
twin tunnel beneath. The magnitude of the maximum measured settlement of the existing
tunnel due to the upper (first) new tunnel excavation in Test E2N3,5 was similar to that in Test
E2N3,3. Also, the settlement of the existing tunnel caused by the excavation of the lower
(second) new tunnel was smaller than that due to the upper one. By comparing the measured
results from the two tests, the settlement of the existing tunnel caused by the side-by-side new
twin tunnels was larger than the vertically stacked tunnel case. This is because the distance
between the lower new tunnel and the existing tunnel or pillar depth-to-diameter ratio (P/D)
was greater compared with those in the side-by-side new tunnel case. In addition, the presence
of the upper new tunnel in Test E2N3,5 reduced the effects of the lower new tunnel
excavation on the existing tunnel (i.e., shielding effects, as discussed in Chapter 7).
The induced gradient of the existing tunnel was deduced from the tunnel settlement as
summarized in Table 8.2. Due to the first tunnel excavation, the tunnel gradient from the two
tests in this study was still within the allowable limit of 1:1000 (LTA, 2000; BD, 2009).
However, after completion of the second tunnel, the maximum measured gradient in both tests
exceeded the allowable limit. This observation of settlement and gradient suggests that twin
tunnel excavation could cause problems on an existing tunnel.
Figure 8.2 illustrates the induced strain measured along the invert in the longitudinal direction
of the existing tunnel. The positive and negative signs denote induced tensile and induced
compressive strain, respectively. The induced strain was compared with the cracking tensile
strain (150 ) of unreinforced concrete, recommended by ACI (2001).
Strain on the existing tunnel induced by the side-by-side new twin tunnels is shown in Figure
8.2a. The location of the induced tensile strain was above the crown of the new tunnel while
the compressive strain was induced at the further offset distance away. The magnitude of the
maximum measured tensile strain induced by the left (first) new tunnel was still within the
177
cracking tensile strain limit. After the completion of the right (second) new tunnel excavation,
the induced tensile strain exceed the cracking tensile strain and asymmetry of the induced
strain was observed. The discrepancy between the measured and computed results is
discussed previously in the tunnel settlement section.
Figure 8.2b shows the induced strain caused by excavation of the vertically stacked new twin
tunnels (Test E2N3,5). The magnitude of the maximum measured and computed tensile strain
induced by the upper new tunnel was similar to that caused by the left new tunnel in Test
E2N3,3. This is because the P/D between the existing and new tunnels in both tests was the
same. At the end of both tunnel excavation, the maximum tensile strain induced in Test
E2N3,3 was larger than that in Test E2N3,5. This observation is consistent with the settlement
of the existing tunnel (refer to Figure 8.1a). One of possible reasons is reduction of stress
caused at the invert of the existing tunnel caused by the lower new tunnel in Test E2N3,5 was
smaller than that caused by the right (second) new tunnel in Test E2N3,3. Further details are
discussed later in Section 8.3.1.
Table 8.3 summarizes shear stress acting on the tunnel lining. The shear stress was deduced
from induced strain in the longitudinal direction of the existing tunnel. Given that the
compressive strength of the concrete is 50 MPa and a reduction factor of 0.55 was adopted,
the allowable stress was estimated to be 660 kPa (ACI, 2011). From the measured results, the
shear stress induced by the first tunnel in the two tests was similar and still under the
allowable shear stress. After the completion of the second tunnel, the shear stress induced in
the case of the side-by-side new twin tunnels was significantly larger than that in the
vertically stacked tunnel case. The shear stress induced at the end of the twin tunnel
excavation in both tests exceeded the allowable shear stress.
Figure 8.3 shows the change in the radius at the center of the existing tunnel (X/D = 0) after
178
completion of the twin new tunnel excavation. The change in tunnel radius (R) was
normalized with the undeformed tunnel radius (R0 = 3 m). The positive and negative signs
denote the increase and decrease in tunnel radius, respectively.
Tunnel deformation caused by the side-by-side new tunnel twin excavation (Test E2N3,3) is
shown in Figure 8.3a. From the measured results at the end of left new tunnel excavation,
decrease in radius at the crown and invert but increase in radius at springlines was observed.
It suggests that the existing tunnel was compressed in the vertical direction. A smaller
reduction in tunnel radius at the crown and invert occurred as well as a decrease in elongation
of the radius at both springlines at the end of the right (second) new tunnel excavation
compared with that at the end of the left (first) tunnel. It indicates that the effects of the
second new tunnel only resulted in the reduction in compression of the existing tunnel in the
vertical direction. This is because of the reduction in stress in the vertical direction was larger
than that in the horizontal direction due to the second new tunnel excavation compared with
the first one (to be discussed later in Section 8.3.1). Discrepancies between measured and
computed results were observed and a possible reason is discussed previously in Section
8.2.1.
Figure 8.3b shows the deformation of the existing tunnel caused by the vertically stacked new
twin tunnel excavation (Test E2N3,5). Due to the excavation of the new twin tunnels, the
existing tunnel was compressed in the vertical direction and elongated in the horizontal
direction. In addition, a larger vertical compression of the existing tunnel caused by the upper
(first) new tunnel than that caused by the lower (second) new tunnel was observed.
BTS (2000) suggested that the maximum difference between the maximum and minimum
diameter should be within 2% [i.e., (Dmax Dmin)/D0 2%, where Dmax, Dmin and D0 are
maximum, minimum and undeformed tunnel diameter, respectively]. The deformation of the
existing tunnel in both tests caused by the twin tunnel excavation was still within the
179
recommended limit. Note that this deformation was due to the twin tunnel excavation only,
and the total deformation of the existing tunnel should also be considered as the direction of
the deformation caused by tunneling was the same as that before tunneling (given that the
at-rest earth pressure coefficient (K0) was less than 1).
Figure 8.4 shows induced strain measured at the outer surface at the center of the existing
tunnel (X/D = 0) in the transverse direction. The results illustrated conditions at the end of
each new tunnel excavation. The positive and negative signs denote induced tensile and
induced compressive strain, respectively.
At the completion of the left (first) new tunnel excavation in the side-by-side tunnel case (Test
E2N3,3), tensile strain was induced at both springlines while induced compressive strain
occurred at the crown, shoulders, knees and invert. It suggests that the existing tunnel was
vertically compressed. After the end of the right (second) new tunnel excavation in Test
E2N3,3, decrease in induced tensile strain at both springlines as well as reduction in induced
compressive strain at the crown and invert took place, compared with that at the completion
of the first tunnel. This observation of the induced strain was consistent with the reduction of
vertical compression in the existing tunnel (refer to Figure 8.3a).
At the end of the upper (first) new tunnel excavation in the vertically stacked tunnel case (Test
E2N3,5), a similar strain direction but larger in magnitude compared with that caused by the
left new tunnel in Test E2N3,3 was observed. This is consistent with the larger vertical
compression of the existing tunnel in Test E2N3,5 than that in Test E2N3,3 (see Figure 8.3). A
possible reason is that the new tunnel in the former was closer to the center of the existing
tunnel (X/D = 0) than the latter, resulting larger effects of tunneling on the existing tunnel
caused by stress change (to be discussed later). In addition, the induced strain caused by the
upper (first) new tunnel was larger than that caused by the lower (second) one.
180
8.3 Change in stress and shear strain of soil due to twin tunnel excavation
Figure 8.5 shows incremental normal stress acting at the center of the existing tunnel (X/D =
0) in each stage of excavation or the normalized location of the advancing tunnel face (Y/D).
Note that change in normal stress due to the first two excavation stages (i.e., from Y/D = -1.5
to -0.9) was much smaller than the subsequent four stages (i.e., from Y/D = -0.3 to 1.5) and is
not shown for clarity. The positive and negative signs denote the increase and decrease in
normal stress, respectively compared with those when the centrifugal acceleration reached
60g.
The incremental normal stress caused by the left (first) new tunnel in the side-by-side twin
tunnel case is shown in Figure 8.5a. When the new tunnel approached the existing tunnel
(Y/D = -0.3), the maximum reduction in stress occurred at the left knee, as it was located
closest to the advancing tunnel face. As the excavated section was directly underneath the
existing tunnel (Y/D = 0.3), the increase in normal stress took place at the crown and right
shoulder, while a reduction in normal stress was observed at the left shoulder, springlines,
knees and invert. At the end of the tunnel excavation (Y/D = 1.5), the increase in normal
stress at the crown and decrease in normal stress at the right springline, knees and invert was
the largest. However, the largest change in normal stress at the shoulders and left springline
occurred before at the end of tunnel advancement. It suggests that simulation of the
three-dimensional tunnel advancement is required as the critical condition of loading may not
be at the end of tunnel excavation.
Figure 8.5b shows incremental normal stress at the end of the right (second) new tunnel
excavation in the side-by-side case. Compared with at the end of the left new tunnel, there
was an increase in normal stress at the crown and shoulders while there was a reduction in
normal stress at the springlines, knees and invert. However, the amount of normal stress
181
change due to the right new tunnel only was smaller than that caused by the left new tunnel.
At the invert, the reduction in normal stress caused by the second new tunnel only was 65 kPa
while the reduction caused by the first one was 100 kPa. This is because the stress reduction
was mainly mobilized due to the construction of the first tunnel. As a result, settlement of the
existing tunnel caused by the left new tunnel was larger than the right one (see to Fig. 5a).
In order to understand the deformation of the existing tunnel, the change in stress in both the
horizontal and vertical directions was considered. Due to excavation of each side-by-side new
twin tunnel, the reduction in stress in the horizontal direction was larger than that in the
vertical direction. As a result, the existing tunnel was elongated in the horizontal direction and
compressed in the vertical direction (refer to Figure 8.3a). In addition, the difference between
the stresses in the horizontal and vertical directions caused by the first new tunnel was larger
than in the second new tunnel. As a result, vertical compression of the existing tunnel caused
by the first (left) new tunnel was larger than the second (right) new tunnel.
The incremental normal stress at the end of the upper (first) new tunnel excavation in case
E2N3,5 is shown in Figure 8.5c. The trend of the incremental stress caused by the upper new
tunnel in the vertically stacked tunnel case was similar to that in the left (first) new tunnel
excavation in case E2N3,3 (refer to Figure 8.5a). However, the magnitude of the change in
normal stress at every location on the existing tunnel in the former case was larger than that in
the latter. This is because the upper new tunnel was located closer to the center of the existing
tunnel (X/D = 0) than the left new tunnel in case E2N3,3.
Figure 8.5d shows the incremental normal stress at the end of the lower new tunnel
excavation in case E2N3,5. At the crown and shoulders, the increase in normal stress at the
end of the second tunnel was larger than that caused by the first one. However, there was a
decrease in normal stress reduction at the springlines, knees and invert. This decrease in
normal stress reduction at the invert is because of stress redistribution and larger P/D between
182
the lower (second) new tunnel and the existing tunnel.
BD (2009) suggested that the change in stress acting on a tunnel lining should be within 20
kPa. It can be seen that reduction in normal stress at the springlines, knees and invert, as well
as increase in normal stress at the crown and shoulders, exceeded the recommended limit.
This observation suggests that the structural analysis of the tunnel lining based on change of
loading should also be considered. In addition, the maximum change in normal stress at some
locations on the existing tunnel occurred during the advancement of the new tunnel, rather
than the end of tunnel excavation.
8.3.2 Induced shear strain of soil
Figure 8.6 shows the contours of shear strain (s) of soil due to twin tunnel excavation. The
induced shear strain was considered at the end of each new tunnel excavation, compared with
before the tunnel construction (at the acceleration of 60g).
Shear strain of soil at the end of the left (first) new tunnel excavation in the side-by-side
tunnel case (E2N3,3) is shown in Figure 8.6a. Maximum shear strain of 0.4% occurred above
the crown of the left new tunnel. The shear strain along the invert of the existing tunnel was
also the largest above the left new tunnel. It suggests that the maximum reduction in soil
stiffness occurred above the left new tunnel as shear modulus reduces with increasing shear
strain. Asymmetry of shear strain extending from both sides of the left new tunnel was
observed. This is due to so-called shielding effects or the presence of the lining of the right
new tunnel which was modeled as wished-in-place.
At the end of the right (second) new tunnel excavation (as shown in Figure 8.6b), maximum
shear strain of 0.6% occurred at the right new tunnel. In addition, there was an increase in
shear strain of soil above the right shoulder of the left new tunnel. The maximum shear strain
on the invert was observed at the center of the existing tunnel and above the right new tunnel.
183
This asymmetrical distribution of shear strain along the invert of the existing tunnel caused
soil stiffness to be asymmetry and subsequently resulted in asymmetrical settlement and strain
in the longitudinal direction of the existing tunnel (see Figure 8.1and Figure 8.2).
The induced shear strain of soil by the upper (first) new tunnel in the vertically stacked tunnel
case (E2N3,5) is illustrated in Figure 8.6c. The maximum shear strain of soil of 0.4%
occurred underneath the invert of the existing tunnel above the upper new tunnel. This
maximum shear strain in this case was the same as above the left new tunnel in case E2N3,3
(see Figure 8.6a). It suggests that the soil stiffness reduction along the invert of the existing
tunnel due to the first new tunnel excavation in the two cases should be similar. As a result,
settlement of the existing tunnel caused by the first new tunnel in both cases was almost the
same (refer to Figure 8.1).
The shear strain of soil after completion of the lower (second) new tunnel in case E2N3,5 is
shown in Figure 8.6d. The maximum shear strain underneath the invert of the existing tunnel
increased from 0.4% (in Figure 8.6c) to 0.5% at the end of the lower new tunnel excavation.
This increase in shear strain also contributed reduction in soil stiffness, which induced
settlement of the existing tunnel caused by the lower new tunnel excavation (see Figure 8.1b).
8.4 Effects of arrangement and excavation sequence of twin tunnels on

multi-tunnel interaction
8.4.1 Effects of excavation sequence on the settlement of the existing tunnel
Figure 8.7 shows the settlement of the existing tunnel obtained from the numerical parametric
study. As the lining of the new tunnel was wished-in-place in centrifuge tests, the effects of
the sequence of the new tunnel excavation on the existing tunnel was investigated. The
sequence of tunneling in each case is summarized in Table 4.1.
After completion of both new tunnels, settlement of the existing tunnel in case E2N3,3N was
184
larger than that in case E2N3,5N. This observation was consistent with that in the centrifuge
tests and numerical back analyses. In addition, even though the lower new tunnel was
excavated first followed by the upper one (case E2N5,3N), the settlement of the existing
tunnel was still smaller than the side-by-side tunnel case. The reason is explained in the
previous section.
Settlement of the existing tunnel when the upper new tunnel was excavated first followed by
the lower one (case E2N3,5N) was smaller than when the lower one was excavated first (case
E2N5,3N). This is because the upper new tunnel, which was excavated first, provided
shielding effects or minimized impact of the lower new tunnel excavation of the existing
tunnel. Further explanation is given later.
8.4.2 Incremental vertical stress caused by different tunneling sequences
Figure 8.8 shows incremental vertical stress along the invert in the longitudinal direction of
the existing tunnel caused by different sequences of tunnel excavation in the vertically stacked
twin tunnel case. After completion of the upper (first) new tunnel excavation in case E2N3,5N,
the maximum reduction in vertical stress of about 180 kPa occurred at the center of the
existing tunnel (X/D = 0). At the end of the lower (second) new tunnel excavation, there was
an increase in vertical stress at X/D = 0 whereas stress decreased between X/D = 0.5 and 2.
This is because the upper new tunnel minimized stress reduction at the invert at the center of
the existing tunnel. The reduction in vertical stress took place at a distance of 0.5D to 2D
away from the center of the existing tunnel instead. In addition, settlement of the existing
tunnel caused by the lower new tunnel was due to further reduction in stress at an offset
distance between 0.5 to 2 times the tunnel diameter from the center of the existing tunnel (see
Figure 8.7).
Opposite results were observed when the lower new tunnel was excavated first followed by
the upper one (case E2N5,3N). The maximum vertical stress reduction at the center of
185
existing tunnel (X/D = 0) was 110 kPa caused by excavation of the lower (first) new tunnel.
After completion of the upper (second) new tunnel excavation, the maximum vertical stress
reduction increased to 195 kPa. By comparing the results in these two cases, reduction in the
stress at the invert of the existing tunnel in case E2N5,3N was larger than that in case
E2N3,5N. As a result, the settlement of the existing tunnel was larger when the lower new
tunnel was excavated first than when the upper new tunnel was constructed before the lower
one (refer to Figure 8.7).
Two centrifuge tests and numerical back-analyses to investigate the effects of
three-dimensional new twin tunnel excavation below a perpendicular existing tunnel are
reported. New twin tunnels were excavated in side-by-side and vertically stacked
arrangements. In addition, a numerical parametric study to examine the effect of tunneling
sequence is discussed. Based on the interpretation of measured and computed results, the
following conclusions may be drawn:
(a) Settlement of the existing tunnel caused by the vertically stacked new twin tunnel
excavation was smaller than the side-by-side tunnel case. This is because the lower new
tunnel in vertically stacked tunnel case was located further away from the existing
tunnel than each new tunnel in side-by-side tunnel case. In addition, when the lower
new tunnel was excavated after the upper one in vertically stacked tunnel case, the
tunneling effects of the lower one on the existing tunnel were minimized by the
shielding effects of the presence of the upper new tunnel. Total settlement and gradient
of the existing tunnel caused twin tunnel excavation by in the two tests exceeded the
recommended limits (LTA, 2000; BD, 2009).
(b) Tunnel settlements caused by the second new tunnel only in both tests were smaller than
those induced by the first one. This is because the reduction of vertical stress caused by
186
the second new tunnel only was smaller than the first new tunnel for all cases. For
side-by-side new twin tunnel the stress reduction was mainly mobilized due to the
construction of the first tunnel. In case of vertically stacked twin tunnel, the P/D of
lower (second) new tunnel to the existing tunnel was larger than that in the upper (first)
new tunnel. In addition, when the upper new tunnel was excavated first, the presence of
the upper new tunnel provided shielding effects to reduce the effects of the lower new
tunnel excavation on the existing tunnel.
(c) The effects of the sequence of new twin tunnel excavation show that when the lower
new tunnel was excavated first followed by the upper new tunnel, a larger settlement of
the existing tunnel occurred than when the upper one was excavated first. This is
because when the lower new tunnel was excavated first, there was no shielding effect or
presence of the upper new tunnel, resulting in a larger stress reduction underneath the
invert of the existing tunnel than when upper new tunnel was excavated first.
(d) Larger tensile strain was induced along the invert of the existing tunnel for the
side-by-side tunnel case than that in the vertically stacked tunnel case was observed.
Induced strain and shear stress in both tests exceeded the cracking tensile strain (ACI,
2001) and the allowable shear stress (ACI, 2011), suggesting that cracks are likely to
occur if the tunnel lining is made of concrete.
(e) In both tests, the existing tunnel was vertically compressed due to the twin tunnel
excavation. This is because the stress reduction on the existing tunnel in the horizontal
direction was larger than that in the vertical direction. The effects of the second new
tunnel excavation only caused a reduction in the vertical compression of the existing
tunnel in both tests, compared with that induced by the first new tunnel. This is because
the difference between stresses in the horizontal and vertical directions caused by the
second new tunnel was smaller than the first new tunnel.
187
Table 8.1 Summary of tests interpreted to investigate the response of an existing tunnel due to
new twin tunnel excavation across underneath
Test ID C/D C/D Brief description of each test
(existing) (new)
E2N3,3 2.0 3.5, 3.5 New side-by-side twin tunnel excavation
E2N3,5 2.0 3.5, 5.0 New vertically stack twin tunnel excavation
Table 8.2 Summary of maximum induced tunnel gradient
Test ID Result after the Induced tunnel gradient
completion of Measured Computed
E2N3,3 Left new tunnel (1st) 1:1400 1:2000
Right new tunnel (2nd) 1:780 1:1210
E2N3,5 Upper new tunnel (1st) 1:1240 1:1990
Lower new tunnel (2nd) 1:800 1:1430
Cooper et al. (2002) Concourse tunnel (1st) 1:540 N/A
Upline tunnel (2nd) 1:360 N/A
Note: Allowable induced gradient 1:1000 (LTA, 2000; BD, 2009)
188
Table 8.3 Summary of maximum induced shear stress on tunnel lining
Test ID Result after the Induced shear stress (kPa)
completion of Measured Computed
E2N3,3 Left new tunnel (1st) 630 840
Right new tunnel (2nd) 1170 1040
E2N3,5 Upper new tunnel (1st) 590 650
Lower new tunnel (2nd) 690 780
Note: Allowable shear stress 660 kPa (ACI, 2011)
189

-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
0.0
Existing tunnel
0.1
settlement (/D, %)
Normalized tunnel
Left new Right new

0.2 15 mm E2N3,3
(LTA, 2000)
20 mm
0.3 (BD, 2009)
0.4
Centerline of Centerline of right
left new tunnel new tunnel
0.5
Measured [E2N3,3-Left (1st)] Computed [E2N3,3-Left (1st)]

Measured [E2N3,3-Right (2nd)] Computed [E2N3,3-Right (2nd)]
Field [Cooper et al., 2002 (1st)] ref0
Field [Cooper et al., 2002 (2nd)]
(a)

-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
0.0
Existing tunnel
0.1 Upper new
settlement (/D, %)
Normalized tunnel
st
tunnel (1 )
Lower new
tunnel (2 nd )
0.2 15 mm E2N3,5
(LTA, 2000)
20 mm
0.3 (BD, 2009)
0.4 Centerline of upper and

lower new tunnels
0.5
Measured [E2N3,5-Upper (1st)] Computed [E2N3,5-Upper (1st)]

Measured [E2N3,5-Lower (2nd)] Computed [E2N3,5-Lower (2nd)]
(b)
Figure 8.1 Settlement along the crown of the existing tunnel obtained from (a) E2N3,3; (b)
E2N3,5
190

-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
-100
longitudinal direction ()
Centerline of Centerline of
Left new tunnel Right newtunnel
-50
Induced strain in
50
100
Existing tunnel
150 Left new Right new
200 t, crack of unreinforced

concrete (ACI, 2001) E2N3,3
250
Measured [E2N3,3-Left (1st)] Computed [E2N3,3-Left (1st)]
(a) Sign convention:
+ Tensile strain

-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
-100
Centerline of
longitudinal direction ()
Upper and lower new tunnel

-50
Induced strain in
50
100
Existing tunnel
150 Upper new
st
tunnel (1 )
Lower new
200 t, crack of unreinforced tunnel (2 nd )
E2N3,5
250
(b)
Figure 8.2 Induced strain in the longitudinal direction along the invert of the existing tunnel
in Tests (a) E2N3,3; (b) E2N3,5
191

tunnel radius 0.10
(R/R 0 , %) X/D = 0
0.05 Existing
tunnel
0.00 Initial
nd
Left (1st ) Right (2 )
-0.05 new tunnel new tunnel
Existing
tunnel
Left (1st ) and right (2nd)

new tunnels
Invert
Measured [E2N3,3-Left (1st)] Computed [E2N3,3-Left (1st)] E2N3,3

(a) Sign convention

+ Tunnel radius increase
- Tunnel radius decrease

tunnel radius 0.10
(R/R0 , %) X/D = 0
0.05 Existing
tunnel
0.00 Initial Upper (1 )
st
new tunnel
-0.05 Lower (2nd )

new tunnel
Existing
tunnel
Upper new tunnel (1st )
Invert
Lower new tunnel (2nd)
E2N3,5
(b)
Figure 8.3 Deformation of the existing tunnel in Tests (a) E2N3,3; (b) E2N3,5
192
Induced strain () Crown

150
100
50
t of
0 unreinforced
Initial
-50 concrete
-100 (ACI, 2001)
L-knee R-knee
Invert
E2N3,3-Left (1st) E2N3,3-Right (2nd)
E2N3,5-Upper (1st) E2N3,5-Lower (2nd)
Z Z
Y Y
Sign convention:
Existing Existing
tunnel tunnel + Tensile strain
st
Upper new tunnel (1 )
Left (1st ) and right (2nd)

new tunnels
Lower new tunnel (2nd)
E2N3,3 E2N3,5
Figure 8.4 Induced strain measured in the transverse direction on the outer face of the
existing tunnel
193
Incremental normal
stress (kPa) Crown
-180
-150
-120 20 kPa
L-Shoulder (BD, 2009) R-Shoulder
-90
-60
-30
0 Initial
30
60
Location of new
tunnel face (Y/D)
Z
-0.3
0.3 Y
L-Knee R-Knee
0.9
E2N3,3
1.5 Existing
Invert tunnel
(a)
Y/D -0.9 0.3 1.5
-1.5 -0.3 0.9
Incremental normal
stress (kPa) Crown Left (1st ) and right (2nd )
-180 new tunnels
-150
-120 20 kPa Sign convention
-90 + Increase in normal stress
-60
-30 - Decrease in normal stress
0 Initial
30
60
Location of new
tunnel face (Y/D)
-0.3
0.3
L-Knee R-Knee
0.9
1.5
Invert
(b)
194
Incremental normal
stress (kPa) Crown
-180
-150
-120 20 kPa
-90
-60
-30
0 Initial
30
60
Location of new Z
tunnel face (Y/D)
-0.3 Y
0.3
L-Knee R-Knee E2N3,5
0.9 Existing
tunnel
1.5
Invert Upper new tunnel (1st )
(c)
Incremental normal
stress (kPa) Crown Y/D -0.9 0.3 1.5
-1.5 -0.3 0.9
-180 Lower new tunnel (2nd )
-150
-120 20 kPa
L-Shoulder (BD, 2009) R-Shoulder Sign convention
-90
-60 + Increase in normal stress
-30
0 - Decrease in normal stress
Initial
30
60
Location of new
tunnel face (Y/D)
-0.3
0.3
L-Knee R-Knee
0.9
1.5
Invert
(d)
Figure 8.5 Incremental normal stress around the centre of the existing tunnel after completion
of (a) left tunnel [E2N3,3]; (b) right tunnel [E2N3,3]; (c) upper tunnel [E2N3,5]; (d) lower
tunnel [E2N3,5]
195
Center of existing tunnel
Left
tunnel
(a)
Left Right
tunnel tunnel
(b)
196
Upper
tunnel
(c)
Upper
tunnel
Lower
tunnel
(d)
Figure 8.6 Contour of shear strain after completion of (a) left new tunnel [E2N3,3]; (b) right
new tunnel [E2N3,3]; (c) upper new tunnel [E2N3,5]; (d) lower new tunnel [E2N3,5]
197

-3 -2 -1 0 1 2 3
0.2 Existing tunnel
Left new Right new
tunnel (1st ) tunnel (2nd )
15 mm
E2N3,3 (N)
(LTA, 2000)
settlement (/D, %)
Normalized tunnel
E2N3,3 (N)
E2N5,3 (N)
Existing tunnel
E2N3,5 (N)
0.3 Upper new
st
tunnel (1 )
20 mm Lower new
(BD, 2009) tunnel (2 nd )
E2N3,5 (N)
Centerline of upper and

lower new tunnel
Centerline of Centerline of right Existing tunnel
left new tunnel new tunnel Upper new
0.4 tunnel (2nd )
Computed [E2N3,3 (N)] Lower new
tunnel (1st )
Computed [E2N3,5 (N)] E2N5,3 (N)
Computed [E2N5,3 (N)]
Figure 8.7 Effects of non-wished-in-place new twin tunnel excavation on settlement of the
existing tunnel
198
100
Incremental vertical stress,
50
0
v (kPa)
-50
= 20 kPa
(BD, 2009)
-100
-150
-200
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
Normalized longitudinal distance along invert of
existing tunnel (X/D)
E2N3,5 (N)-Upper (1st) E2N3,5 (N)-Lower (2nd)
E2N5,3 (N)-Lower (1st) E2N5,3 (N)-Upper (2nd)

Upper new Upper new
st
tunnel (1 ) tunnel (2nd )
Lower new Lower new
tunnel (2 nd ) tunnel (1st )
E2N3,5N E2N5,3N
Figure 8.8 Incremental vertical stress along the invert of the existing tunnel caused by
different tunneling sequences
199
Chapter 9 Numerical parametric study of influence of volume loss, relative density, tunneling
in saturated sand and tunnel diameter in crossing-tunnel interaction
Chapter 9 Numerical parametric study of influence

of volume loss, relative density, tunneling in
saturated sand and tunnel diameter on
crossing-tunnel interaction
9.1 Introduction
Ground movements and changes of stresses caused by tunneling are increasingly important as
more and more tunnels have been constructed in urban areas. As a result of a new tunnel
excavation adjacent to an existing tunnel, the existing tunnel may experience excessive
deformation and cracks induced on the tunnel lining. One of effective measures to minimize
the effects of the new tunnel excavation on the existing tunnel is to control the volume loss.
As discussed in Chapter 2, volume loss from some case histories (e.g., Mair and Taylor, 1997;
Boonyarak et al., 2014) can be controlled up to 0.5% to 1%. Although tunnel volume loss in
many cases were reported to be with 1%, in mixed-face tunneling involving clay and sand,
volume loss can be between 1% and 4% (Shirlaw et al., 2003; Abrams, 2007). As summarized
in Chapter 3, all the centrifuge tests in this research were carried in dry medium dense sand
(Dr = 64% - 66%). Thus, the understanding of the response of the existing tunnel due to the
new tunnel excavation in different relative densities and saturated sand should also be
improved.
In this chapter, the interpretation of numerical parametric study is reported. The objectives of
the parametric study are to investigate the influence of volume loss, relative density,
200
tunneling in saturated sand and tunnel diameter on the interaction of crossing tunnels.
To study the effects of volume loss on crossing-tunnel interaction, the imposed volume loss
was from 0.5% to 4%. To examine to influence of relative density on tunnel-tunnel interaction,
tunnel excavation in different relative densities ranging from loose sand (Dr = 30%) to dense
sand (Dr = 80%) were carried out. Behavior of the existing tunnel caused by new tunnel
excavation in saturated sand was also investigated. The effects of tunnel diameter were
investigated by decreasing diameter of both existing and new tunnels from 6 m to 3 m. To
reduce tunnel diameter, g-level was decreased from 60g (reference test) to 30g. By adopting
this method, the C/Ds of tunnels, P/D and lining thickness to diameter ratio (t/D) in both cases
remained unchanged. The results were validated and interpreted along with those measured
from Test E2N3 (refer to Figure 3.3). Summary of numerical run of parametric study of the
influence of volume loss, relative density and tunneling in saturated sand on crossing-tunnel
interaction is given in Table 4.2.
9.2 Effects of volume loss on crossing-tunnel interaction
9.2.1 Ground surface settlement
Ground surface settlement above the existing tunnel at different volume losses is shown in
Figure 9.1. The results are illustrated at the end of new tunnel excavation. The computed
surface settlement at 2% volume loss was validated with the measured one. It can be seen that
there was a good agreement between the measured and computed results.
At 0.5% volume loss, the maximum ground surface settlement was about 6.8 mm. After
increasing in volume loss from 0.5% to 1%, 2% and 4%, the maximum computed surface
settlement became 12.5 mm, 20.3 mm and 27.1 mm, respectively. Surface settlement
increased with volume loss at a decreasing rate. This is because shear modulus and vertical
stress mainly mobilized at a higher volume loss. Further explanation is given in the Sections
201
9.2.4 and 9.2.5.
9.2.2 Settlement of the existing tunnel
Settlement along the crown of the existing tunnel caused by excavation of the new tunnel at
different volume losses is shown in Figure 9.2a. The settlements are normalized by the
diameter of the new tunnel and the results are given at the end of the new tunnel excavation.
When volume loss of 0.5% was adopted, the maximum computed tunnel settlement was about
0.08%D or equivalent to 5 mm. If the volume loss was doubled from 0.5% to 1%, the
maximum computed tunnel settlement became 0.15% (9 mm). Note that if the volume loss
can be controlled within 1%, settlement of existing tunnel did not exceed the recommended
limit of 15 mm (LTA, 2000).
The maximum measured settlement of the existing tunnel in the reference test (VL = 2%) was
about 0.3%D (18 mm). This maximum tunnel settlement exceeded the recommended
serviceability limit of 15 mm given by LTA (2000). The computed tunnel settlement of 15 mm
at volume loss of 2% was in a good agreement with the measured one. Minor discrepancy
between the measured and computed results may be due to model parameters were adopted
from empirical relationship and previous studies. For tunnel excavation at 4% volume loss,
the maximum computed settlement of the existing tunnel was 0.33%D, equivalent to 19.7 mm
in the prototype scale.
From the results from four different volume losses, the settlement of the existing tunnel
increased with increasing in volume loss at a reducing rate. This is because shear modulus and
vertical stress ratios almost fully mobilized at a higher volume loss. Further explanation is
given in the Sections 9.2.4 and 9.2.5.
Table 9.1 summarize maximum induced gradient of the existing tunnel at different volume
202
losses. As expected, the induced gradient of the existing tunnel increased with higher volume
loss. The by comparing with the allowable limit of tunnel gradient of 1:1000 (LTA, 2000; BD,
2009), the maximum induced tunnel gradient did not exceed the limit.
To investigate the effects of volume loss on the shape of settlement of the existing tunnel,
tunnel settlement at each location along the crown was normalized by that at the center of the
existing tunnel (X/D = 0) as shown in Figure 9.2b. Note that the measured results were
excluded as the number of data points were limited. It can be seen that the normalized tunnel
settlements at four different volume losses were almost identical. It suggests that the shape of
settlement of the existing tunnel was not influenced by volume loss.
9.2.3 Maximum induced tensile strain on the tunnel lining
Maximum induced tensile strain in the lining at the invert of the existing tunnel at the end of
the new tunnel excavation is shown in Figure 9.3. The location of the maximum tensile strain
was at the center of the existing tunnel (X/D = 0).
The maximum measured strain induced on the tunnel lining at volume loss of 2% was 151 .
The measured and computed results were similar in magnitude with minor discrepancy as
discussed in the previous section.
The maximum computed tensile increased with higher volume loss with a decreasing rate.
This observation was consistent with settlement of the existing tunnel with different volume
loss (refer to Figure 9.2) as the induced strain was resulted from differential settlement of the
existing tunnel. When the volume loss was controlled within 1%, the maximum computed
tensile strain induced in the tunnel lining did not exceed the cracking tensile strain limit of
150 (ACI, 2001). On the other hand, the induced tensile strain was larger the cracking
tensile strain limit when the volume loss reached or exceeded 2% suggesting that cracks may
appear in the tunnel lining.
203
9.2.4 Incremental vertical stress acting on the invert of the existing tunnel
Incremental vertical stress acting on the invert in the longitudinal direction of the existing
tunnel is shown in Figure 9.4. The incremental stress was considered at the end of tunnel
excavation. The positive and negative signs denote increase and decrease in vertical stress,
respectively compared with that when the centrifugal acceleration reached 60g
For volume loss 0.5%, the maximum vertical stress reduction of 52 kPa took place at the
center of the existing tunnel (X/D = 0). The reduction of vertical stress reduced with
increasing offset distance until at distance of 1.25D from the center of the existing tunnel that
vertical stress increased on the invert. The reason of the stress increase was due to stress
redistribution to maintain equilibrium as discussed previously in Section 5.3.2.
With increasing volume loss, the reduction of vertical stress is larger with a decreasing rate.
Similarly, increase in vertical stress at a distance of X/D = 2 increased with larger volume loss.
This is because the stress redistribution at further distances away was proportional to the
vertical stress reduction at the center of the existing tunnel (X/D = 0) in order to the maintain
stress equilibrium.
The vertical stress along the invert of the existing tunnel before tunnel excavation was about
220 kPa (result is not shown for clarity). The maximum reduction in vertical stress at volume
loss of 2% and 4% were 147 kPa and 194 kPa, respectively resulting in residual vertical stress
of 37 kPa and 26 kPa. Thus, the ratio of mobilized vertical stress after and before the tunnel
excavation was 0.33 and 0.12 at volume loss of 2% and 4%, respectively.
Zhuang et al. (2012) suggested that stress reduction with soil movement could reach the
minimum limit at the point of maximum arching (defined as a point of small reduction in
vertical stress). In their study, the ratio of stress of soil at the yielding part normalized by the
overburden stress was estimated to be 0.2.
204
By comparing the finding in Zhuang et al. (2012) with this study, further reduction in
mobilized vertical stress ratio was relatively small when the volume loss increased from 2%
to 4%. This is because the effects of soil arching already significantly mobilized when the
volume loss was reached 2%.
The reduction in vertical stress at the center of the existing tunnel in every case exceeded the
recommended limit of 20 kPa (BD, 2009). Although the tunnel settlement and induced tensile
strain in the lining (refer to Figure 9.2and Figure 9.3) when volume loss is within 1% did not
exceed the recommended tunnel settlement and cracking tensile strain, the change in stress
still exceeded the allowable limit. At further distance away, vertical stress in case in all except
at volume loss = 0.5% exceeded the recommended limit. It suggested that, structural capacity
of the existing tunnel should be review when the new tunnel is excavated adjacent to it.
9.2.5 Mobilized shear modulus along the invert of the existing tunnel at different
volume loss
Figure 9.5 shows mobilized shear modulus along the invert of the existing tunnel. The
mobilized secant shear modulus (Gm) was calculated from the computed deviatoric stress (q)
and the computed deviatoric strain (s) using the equation Gm = q/(3s). The ratio of Gm was
considered at the end of the new tunnel excavation and before the tunnel construction (Gafter /
Gbefore).
At volume loss of 0.5%, the minimum mobilized shear modulus ratio was 0.68, occurred at
the center of the existing tunnel (X/D). The reduction of mobilized shear modulus ratio is
because decrease in confining pressure (refer to Figure 9.4). With increasing offset distance,
mobilized shear modulus ratio increased and became more than 1 at X/D more than 2. The
increase in mobilized shear modulus ratio is due to increase in vertical stress due to stress
redistribution (as discussed in the previous section).
205
The shear modulus at the center of the existing tunnel decreased with higher volume loss.
This is because the confining stress reduced with increasing volume loss (see Figure 9.4). In
similar manner, the maximum shear modulus ratio at further distance away increased with
increasing volume loss as stress increase became larger at higher volume loss.
The shear modulus ratio decrease with volume loss at a decreasing rate. The minimum
mobilized shear modulus ratios were 0.22 and 0.10 at volume loss of 2% and 4%, respectively.
It suggests that the effects of soil arching may be significantly mobilized when volume loss
reached 2%.
9.3 Effects of soil relative density on the interaction of crossing-tunnels
9.3.1 Magnitude and shape of tunnel settlement at different soil densities
Settlement of the existing tunnel induced by the new tunnel excavation at different relative
densities of soil is shown in Figure 9.6a. The results are given at the end of tunnel excavation
at volume loss of 2% in every case along the crown of the existing tunnel. The tunnel
settlements are normalized by the diameter of the new tunnel.
At the relative density of 30%, the maximum computed settlement of the existing tunnel was
0.35%D or equivalent to 21 mm in the prototype scale. When the relative density increased to
50%, the maximum computed tunnel settlement reduced to 0.28%D (i.e., 17 mm in the
prototype scale).
For the reference test at the relative density of 64%, the maximum measured and computed
tunnel settlement was 0.30%D and 0.25%D, respectively, equivalent to 18 mm and 15 mm in
the prototype scale. The discrepancy between the measured and computed results has been
discussed previously in the Section 9.2.2. At the relative density of 80%, the maximum
computed settlement of the existing tunnel significantly reduced to 0.16%D (i.e., 10 mm in
206
the prototype scale).
The magnitude of tunnel settlement in loose sand (Dr = 30%) exceeded the allowable limits of
15 mm and 20 mm given by LTA (2000) and BD (2009), respectively. For tunnel excavation
at the relative density of 64% or smaller, the settlement of the existing tunnel exceeded the
allowable limit of 15 mm (LTA, 2000). Only when the existing tunnel was in dense sand (Dr =
80%), the settlement of the existing tunnel was within the allowable limits.
Table 9.2 summarizes the induced gradient of the existing tunnel at different relative densities.
Similar to settlement of the existing tunnel, the induced gradient of the existing tunnel
reduced with increasing density of soil. By comparing with allowable limit given by LTA
(2000) and BD (2009), the gradient of the tunnel did not exceed the limit of 1:1000.
In Figure 9.6b, the shape of settlement trough of the existing tunnel is illustrated by
normalizing settlement at each offset distance by the maximum tunnel settlement. For loose
sand (Dr = 30%), the normalized tunnel settlement trough was the widest. With increase
relative density, the tunnel settlement trough became narrower. The narrowest tunnel
settlement trough occurred in the test in dense sand (Dr = 80%).
9.3.2 Induced tensile strain on the lining of the existing tunnel at different densities
Maximum tensile strain induced at the invert of the existing tunnel is shown in Figure 9.7. As
the maximum settlement of the existing tunnel was at the center of the existing tunnel (refer to
the previous section), the maximum tensile strain in the tunnel lining occurred at the center of
the existing tunnel (X/D = 0). The results are given at the end of tunnel excavation.
The maximum measured tensile strain induced in the tunnel lining was 151 at the relative
density of 64%. At the same relative density, the maximum computed tensile strain was 181
. The reason of the minor discrepancy was discussed previously in Section 9.2.2. As the
207
sand became looser, there was an increase in maximum tensile strain in the tunnel lining. This
is because the differential settlement increased with decreasing relative density as shown in
Figure 9.6 and Table 9.2.
The magnitude of the maximum induced tensile strain in every case exceeded the cracking
tensile strain of 150 (ACI, 2001). It suggests that cracks may appear in the tunnel lining
although settlement of the existing tunnel in dense sand (Dr = 80%) was still with the
recommended limit (LTA, 2000) as refer to Figure 9.6. This is because induced strain in the
tunnel lining is not only controlled by stiffness of soil but also influenced by change in
vertical stress of soil. Further explanation is given in the next section.
9.3.3 Mobilized vertical stress acting on the invert of the existing tunnel at different
densities
Figure 9.8 shows incremental vertical stress acting on the invert of the existing tunnel at the
end of tunnel excavation. The vertical stress was compared with that prior to the construction
of the new tunnel.
For tunnel excavation in loose sand (Dr = 30%), the maximum reduction in vertical stress was
133 kPa at the center of the existing tunnel (X/D = 0). The reduction in stress was smaller
with increasing offset distance and became zero at an offset distance of 1.3D from the center
of the existing tunnel. Increase in vertical stress was observed at X/D more than 1.3 as a result
of stress redistribution. The maximum increase in stress for tunneling in loose sand was about
39 kPa at X/D = 2.5. The maximum reduction in the vertical stress slightly increased with
increasing relative density. This is because at given volume loss, change in stress is
proportional to the stiffness of soil, which increased with reduction in relative density. Further
explanation is given in the next section. The minimum incremental vertical stresses at relative
densities of 50%, 64% and 80% were 142 kPa, 147 kPa and 160 kPa, respectively.
208
The reduction in vertical stress at the center of the existing tunnel up to an offset distance of
1D in every case exceeded the recommended limit of 20 kPa stress change on the tunnel
lining given by BD (2009). Similarly, increase in vertical stress in all cases exceeded the
allowable limit of 20 kPa between a distance of 1.5D and 4D from the center of the existing
tunnel. It suggested that the influence of the new tunnel excavation on stress induced on the
existing tunnel was not located only close to the new tunnel but also extended to further
distance away from the center of the existing tunnel. In addition, change in the vertical stress
in the denser sand was larger than that in the looser sand.
9.3.4 Mobilized shear modulus of soil at different densities
Mobilized shear modulus of soil before (Gm, before) and after (Gm, after) tunnel excavation is
shown in Figure 9.9. The mobilized shear modulus (Gm) was considered in the soil located at
the invert at the center of the existing tunnel (X/D = 0). As discussed previously in Section
9.2.5, the minimum Gm took place at the center of the existing tunnel as it is closest to the new
tunnel. Gm was calculated using the equation Gm = q/3s where q is the computed deviatoric
stress and s is the computed deviatoric strain.
For tunnel excavation in loose sand (Dr = 30%), Gm before and after tunnel excavation were
4.8 MPa and 1.7 MPa, respectively. Thus, the ratio of Gm, after/Gm, before for loose sand became
0.35. For tunnel in medium dense sand (Dr = 50%), Gm before tunnel excavation was larger
than that in loose sand. This is expected as Gm before tunnel excavation increased with
reducing void ratio. After tunnel excavation, Gm in dense sand decreased to 1.9, resulting in
Gm, after/Gm, before of 0.26. It can be seen that Gm before tunnel excavation increased with larger
relative density while Gm after tunnel excavation was almost constant for different densities. It
suggests that Gm, after in each case was mainly influenced by the effects of volume loss not the
soil relative density.
209
This larger Gm before tunnel excavation in denser soil caused smaller settlement of the
existing tunnel (refer to Figure 9.6). Although Gm after tunnel excavation was almost the same
in every case, this effect of reduction in Gm was localized only at the area close to the new
tunnel.
9.4 Response of existing tunnel due to the new tunnel excavation in

saturated sand
9.4.1 Comparison of settlement of the existing tunnel in dry and saturated sands
Figure 9.10a shows the comparison of settlement of the existing tunnel between that in dry
and saturated sands. The results are given at the end of tunnel excavation. The settlement of
the existing tunnel due to excavation of the new tunnel in saturated sand was smaller than that
in dry sand. This is because the reduction in vertical stress along the invert of the existing
tunnel in saturated was smaller than that in dry sand. Further explanation is given in Section
9.4.3. Different from the reference test in dry sand, the maximum settlement of the existing
tunnel due to new tunnel excavation in saturated sand was still within the permissible limit of
15 mm (LTA, 2000). Similarly, gradient of the existing tunnel caused by new tunnel
excavation in saturated sand was smaller than that in dry sand and did not exceed the
allowable gradient of 1: 1000 as recommended by LTA (2000) and BD (2009).
Settlement of the existing tunnel normalized the maximum settlement at the centerline of the
new tunnel is shown in Figure 9.10 to compare the shape of tunnel settlement trough in dry
and saturated sands. It can be seen that, the shape of settlement trough of the existing tunnel
due to the new tunnel excavation in saturated sand was narrower than that in dry sand.
210
9.4.2 Induced strain in the lining in the longitudinal direction of the existing tunnel in
dry and saturated sands
Induced strain in the lining in the longitudinal direction of the existing tunnel at the end of
tunnel excavation is shown in Figure 9.11. The positive and negative signs denote tensile and
compressive strain, respectively induced in the tunnel lining. It can be seen that the maximum
tensile strain induced in the tunnel lining in saturated sand was smaller than that in dry sand.
This is expected as the settlement and gradient of the existing tunnel due to tunnel excavation
in saturated sand was smaller than that in dry sand. The maximum induced tensile strain in the
tunnel lining in saturated sand was 145 , which was close to the cracking tensile strain limit
of 150 (ACI, 2001).
Shear stress in the tunnel lining was deduced from the slope of the strain induced in the
longitudinal direction of the existing tunnel using beam theory. The allowable shear stress of
concrete is estimated to be 660 kPa, given that the concrete compressive strength (fc) is 50
MPa and a reduction factor of 0.55 is adopted (ACI, 2011). It can be seen that shear stress in
the tunnel lining in saturated did not exceed the allowable shear stress limit.
9.4.3 Vertical stress acting on the invert of the existing tunnel in dry and saturated
sands
The vertical stress distribution along the invert of the existing tunnel before and after the
tunnel excavation in dry and saturated sands is shown Figure 9.12a. Before the tunnel
excavation, the vertical stress along the invert of the existing tunnel in dry sand was larger
than that in saturated sand as effective overburden was higher in dry sand. After tunnel
excavation, there was a significant decrease in vertical stress at the center of the existing
tunnel (X/D = 0) in dry sand. At an offset distance more than 1.5D from the center of the
existing tunnel, the vertical stress increased as a result of stress redistribution to maintain the
211
equilibrium. Similar behavior of the vertical stress change due to tunnel excavation was
observed in saturated sand.
Incremental vertical stress along the invert of the existing tunnel at the end of tunnel
excavation is shown Figure 9.12b. The maximum reduction in vertical stress at the center of
the existing tunnel in dry sand was larger than that in saturated sand. This is because the ratio
of vertical stress after and before the tunnel excavation in dry sand was almost the same as
that in saturated sand (refer to Figure 9.12a). The reduction in the vertical stress in both dry
and saturated sands exceeded the recommended limit of 20 kPa (BD, 2009) from the center of
the existing tunnel up to an offset distance 1D away. In addition, the vertical stress increase in
both dry and saturated sand at an offset distance from 1.5D to 3.5D from the center of the
existing tunnel was larger than the permissible limit suggested by BD (2009). It suggests that
the influence zone of large stress change on the existing tunnel in both dry and saturated sand
is from the center of the existing tunnel to an offset distance of 3.5D away. Thus, review of
structural capacity based on change in stress on the existing tunnel should be focus in these
locations.
9.4.4 Mobilized stiffness of soil along the invert of the existing tunnel in dry and
saturated sands
Figure 9.13 shows mobilized shear modulus of soil before (Gm, before) and after (Gm, after) tunnel
excavation along the invert of the existing tunnel. Gm was calculated using the equation Gm =
q/(3s) as discussed previously in Section 9.3.4.
Before tunnel excavation Gm in dry sand was larger than that in saturated sand as the effective
confining pressure was larger in the former than the latter. In similar manner, Gm in dry sand
was larger than that in saturated sand as the ratios of mobilized shear modulus after and
before tunnel excavation in both conditions were almost the same (about 0.2). However, the
212
reduction in vertical stress acting on the invert of the existing tunnel for dry sand was larger
than that in saturated sand, as discussed in the previous section. The effects of vertical stress
reduction dominated the effect of decrease in stiffness of soil, resulting in larger tunnel
settlement in dry sand than that in saturated sand (refer to Figure 9.10).
9.5 Effects of tunnel diameter on crossing-tunnel interaction
9.5.1 Comparison of settlement of the existing tunnel with different diameters
Figure 9.14a shows the comparison of settlement of the existing tunnel with different
diameters. The results are considered at the end of tunnel excavation. In the reference test
(E2N5), the diameter of the existing and new tunnels was 6 m. To study the effects of
diameter, numerical simulation of 3 m tunnel diameter (both existing and new tunnels) was
carried out.
The maximum measured settlement of the 6 m diameter existing tunnel was 12 mm, which
was still within the allowable limit given by LTA (2000). The computed results of 6 m
diameter tunnel reasonably agree with the measured one. When the diameter of the tunnels
decreased to 3 m, settlement of the existing tunnel became smaller than that in 6 m diameter
case. This is because the reduction in vertical stress acting on the invert of the existing tunnel
in 3 m diameter case was smaller than that in 6 m. More explanation of stress is given in
Section 9.5.3.
Settlement of the existing tunnel normalized with maximum tunnel settlement is shown in
Figure 9.14b. It can be seen that the normalized tunnel settlement in the 3 m diameter case
was wider than that in the 6 m diameter case. This is because stress reduction at the location
above the new tunnel in the 6 m diameter case was larger than the 3 m diameter case. In
addition, stress increase due to stress redistribution was also greater in the 6 m diameter case
than that in the 3 m diameter case.
213
9.5.2 Induced strain in tunnel lining with different tunnel diameters
The comparison of strain induced in the tunnel lining with different tunnel diameters at the
end of tunnel excavation is illustrated in Figure 9.15. The positive and negative signs denote
tensile and compressive strain, respectively induced in the tunnel lining. It can be seen that
the maximum tensile strain induced in the tunnel lining caused by the 6 m diameter tunnel
case was larger than that in the 3 m diameter tunnel case. This is expected as the differential
settlement of the existing tunnel due to the former was larger than the latter. The maximum
induced tensile strains in both cases were still within the cracking tensile strain limit given by
ACI (2001).
9.5.3 Distribution of vertical stress on the existing tunnel with different diameters
Incremental vertical stress acting on the invert of the existing tunnel with different tunnel
diameters at the end of tunnel excavation is shown in Figure 9.16a. The positive and negative
signs denote increase and decrease in vertical stress, respectively compared with that prior to
tunnel excavation.
The reduction in vertical stress at the location directly above the new tunnel (X/D = 0) in 6 m
diameter tunnel was greater than that in 3 m diameter case. Moreover, stress increase at offset
distance about 2D from the centerline of the new tunnel in 6 m diameter tunnel case was also
larger than that in 3 m diameter tunnel case. This is because the initial vertical stress in the
former was larger than the latter. Although the C/Ds and P/D of the tunnels in both cases were
identical, the embedded depth in the larger tunnel diameter case is deeper than the smaller
tunnel case. As a result, settlement of the existing tunnel and induced strain in the tunnel
lining in the 6 m diameter case was greater than that in the 3 m diameter case (refer to Figure
9.14 and Figure 9.15).
Mobilized vertical stress ratio with different tunnel diameter is shown in Figure 9.16b. It can
214
be seen that distribution of mobilized in both cases with different tunnel diameters was very
similar. It suggests that effects of stress reduction on the existing tunnel due to new tunnel
excavation are proportional to tunnel diameter. Thus, the finding in this study, which was
carried out with 6 m diameter tunnel is applicable to other size of tunnels.
Three series of numerical parametric study were carried out. The first series investigated the
effects of volume loss on crossing-tunnel interaction. The simulated volume loss was ranging
between 0.5% and 4%. The influence of relative density on the interaction of crossing tunnels
was studied in the second series. The range of relative density was from 30% (loose sand) to
80% (dense sand). The last series examined the response of the existing tunnel due to
excavation of the new tunnel in saturated sand. Based on the computed results, following
conclusions may be drawn:
(a) The settlement and strain induced in the longitudinal direction of the existing tunnel
exceeded the recommended limit of 15 mm (LTA, 2000) and 150 (ACI, 2001),
respectively when the volume loss of tunnel reached 2%. If the volume loss was
controlled within 1%, tunnel settlement and strain in the lining did not exceed the
allowable limits.
(b) The increase in settlement and strain in the longitudinal direction of the existing tunnel
increased with larger volume loss with decreasing rate. This is because reduction in
vertical stress was mobilized to a limit that vertical stress only slightly decreased with
increasing in volume loss. In addition, mobilized shear modulus of soil at the invert at
the center of the existing tunnel reduced with increasing volume loss at a decreasing
rate.
(c) The settlement of the existing with relative density of not more than 64% exceeded the
215
allowable limit of 15 mm given by LTA (2000). With increasing in relative density,
settlement of the existing tunnel decreased to be within the allowable limit (at volume
loss of 2%). The shape of tunnel settlement trough was narrower with increasing in the
relative density.
(d) As the sand became looser, the maximum tensile strain in the tunnel lining increased.
This observation is due to the increase in differential settlement with decreasing relative
density. The magnitude of the maximum induced tensile strain in every case exceeded
the cracking tensile strain of 150 (ACI, 2001). This is because induced strain in the
longitudinal direction of the tunnel is not only controlled by stiffness of soil but also
influenced by reduction in vertical stress of soil.
(e) The mobilized shear modulus of soil before tunnel excavation increased with larger
relative density while the mobilized shear modulus after tunnel excavation was almost
constant at different relative densities. It suggests that mobilized shear modulus in each
case was mainly influenced by the effects of volume loss not the relative density. As a
result, the maximum reduction in the vertical stress slightly increased with increasing
relative density. This is because at given volume loss, change in stress is proportional to
the stiffness of soil.
(f) The maximum tunnel settlement and tensile strain induced in the tunnel lining in the
longitudinal direction was still within the allowable limit of 15 mm (LTA, 2000) and
150 (ACI, 2001), respectively. The shape of settlement trough of the existing tunnel
due to the new tunnel excavation in saturated sand was narrower than that in dry sand.
(g) The mobilized shear modulus in saturated sand was smaller than that in dry sand for
both before and after the excavation of the new tunnel. However, the reduction in
vertical stress acting on the invert of the existing tunnel for saturated sand was smaller
216
than that in dry sand. The effects of vertical stress reduction dominated the effect of
decrease in stiffness of soil, resulting in smaller tunnel settlement in saturated sand than
that in dry sand.
(h) Settlement and induced strain in the lining of the existing tunnel caused by larger
diameter tunnels were greater than that in the smaller diameter tunnels. This is because
the reduction in vertical stress acting on the invert of the existing tunnel in the smaller
diameter tunnels was less than that in the larger diameter tunnels.
217
Table 9.1 Summary of maximum induced tunnel gradient at different volume losses
Volume loss Induced tunnel gradient
(%) Measured Computed
0.5 N/A 1: 5700
1.0 N/A 1: 3020
2.0 1:1600 1:1850
4.0 N/A 1: 1400
Table 9.2 Summary of maximum induced tunnel gradient at different soil densities
Relative density Induced tunnel gradient
Dr, (%) Measured Computed
30 N/A 1: 1500
50 N/A 1: 1700
64 1:1600 1:1850
80 N/A 1: 2500
218

-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
0
Surface settlement (mm)
10
15
20
25
30 Ref0 Computed [VL = 0.5%]

Ref1 Computed [VL = 1%]
Measured [VL = 2%] Computed [VL = 2%]
Ref4 Computed [VL = 4%]
Z
X
Existing tunnel
New tunnel
Figure 9.1 Ground surface settlement above the existing tunnel at different volume losses
219
Normalized longitudinal distance along the crown of

the existing tunnel (X/D)
Z
0 1 2 3 4 5 6 X
0.0
Existing tunnel
settlement (/D,%)
Normalized tunnel
0.1
New tunnel
0.2
15 mm (LTA, 2000)
0.3
20 mm (BD, 2009)
Ref15 Computed [VL = 0.5%]

0.4 Ref20 Computed [VL = 1%]
Measured [VL = 2%] Computed [VL = 2%]
Ref15A Computed [VL = 4%]
(a)

Z
0 1 2 3 4 5 6
X
0.0
0.1
Existing tunnel
0.2
0.3 New tunnel
0.4
/ max
0.5
0.6
0.7
0.8 Computed [VL = 0.5%]
0.9 Computed [VL = 1%]
Computed [VL = 2%]
1.0
Computed [VL = 4%]
(b)
maximum tunnel settlement at the end of the new tunnel excavation at different volume losses
220
250
Induced maximum tensile strain at
the invert of existing tunnel ()
200
t, crack of unreinforced Z
concrete (ACI, 2001) X
150
Existing tunnel
100
New tunnel
50
Measured
Computed
0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Volume loss (%)
Figure 9.3 Maximum tensile strain induced at the invert of the existing tunnel at the end of
the new tunnel excavation at different volume losses
221
100
= 20 kPa
Incremental vertical stress (kPa)
(BD, 2009)
50 Z
X
0
Existing tunnel
-50
New tunnel
-100
Computed [VL = 0.5%]
Computed [VL = 1%]
-150
Computed [VL = 2%]
Computed [VL = 4%]
-200
0 1 2 3 4 5 6
Normalized longitudinal distance along the invert of
tunnel excavation at different volume losses
222
1.2
Z
Mobilized shear modulus ratio,
X
1.0
Existing tunnel
0.8
GAfter/GBefore
New tunnel
0.6
0.4 Computed [VL = 0.5%]

Computed [VL = 1%]
0.2 Computed [VL = 2%]
Computed [VL = 4%]
0.0
0 1 2 3 4 5 6
Figure 9.5 Mobilized shear modulus ratio of soil located along the invert of the existing
tunnel at different volume losses
223

Z
0 1 2 3 4 5 6
X
0.0
Existing tunnel
0.1
settlement (/D,%)
Normalized tunnel
New tunnel
0.2
15 mm (LTA, 2000)
0.3
20 mm (BD, 2009)
Ref15 Computed [Dr = 30%]

0.4 Ref20 Computed [Dr = 50%]
Measured [Dr = 64%] Computed [Dr = 64%]
Ref15A Computed [Dr = 80%]
(a)

Z
0 1 2 3 4 5 6
X
0.0
0.1
Existing tunnel
0.2
0.3 New tunnel
0.4
/ max
0.5
0.6
0.7
0.8 Computed [Dr = 30%]

Computed [Dr = 80%]
(b)
maximum tunnel settlement at the end of the new tunnel excavation at different densities
224
250
Induced maximum tensile strain at
the invert of existing tunnel ()
200
Z
X
150
t, crack of unreinforced
Existing tunnel
100
New tunnel
50
Measured
Computed
0
20 30 40 50 60 70 80
Relative density, Dr (%)
Figure 9.7 Maximum tensile strain induced at the invert of the existing tunnel at the end of
the new tunnel excavation at different volume losses at different relative densities
225
100
= 20 kPa
(BD, 2009)
50 Z
X
0
Existing tunnel
-50
New tunnel
-100 Computed [Dr = 30%]

Computed [Dr = 50%]
-150 Computed [Dr = 64%]
Computed [Dr = 80%]
-200
0 1 2 3 4 5 6
tunnel excavation at different relative densities
226
20
Gm, before
Gm, after
15
Gm (MPa)
Z
X
10 Gm, before
Existing tunnel
5
New tunnel
Gm, after
0
20 30 40 50 60 70 80
Relative density, Dr (%)
Figure 9.9 Mobilized shear modulus of soil located at the invert at the center of the existing
tunnel at the end of new tunnel excavation at different relative densities
227

the existing tunnel (X/D) Z
0 1 2 3 4 5 6 X
0.0
Maximum tunnel gradient
Measured [Dry sand] 1:1600 Existing tunnel
Computed [Dry sand] 1:1850
0.1 Computed [Saturated sand] 1: 2200
settlement (/D,%)
Normalized tunnel
New tunnel
0.2
15 mm (LTA, 2000)
0.3
20 mm (BD, 2009)
Measured [Dry sand] Computed [Dry sand]

0.4 Ref15 Computed [Saturated sand]
Ref20 (a)
Z
0 1 2 3 4 5 6
X
0.0
0.1
Existing tunnel
0.2
0.3 New tunnel
0.4
/ max
0.5
0.6
0.7
0.8
Computed [Dry sand
0.9
Computed [Saturated sand]
1.0
(b)
Figure 9.10 Comparison of settlement of the existing tunnel between dry and saturated sands
at the end of tunnel excavation normalized by (a) tunnel diameter; (b) maximum tunnel
settlement
228

Z
0 1 2 3 4 5 6
Induced strain in tunnel lining ( )
X
-100
-50 Existing tunnel
0 New tunnel
Maximum shear stress

50
Measured [Dry sand] 780 kPa
100 Computed [Dry sand] 750 kPa
Computed [Saturated sand] 540 kPa
150
t, crack of unreinforced
200 concrete 150 (ACI, 2001)
Measured [Dry sand] Computed [Dry sand]
250
Ref150 Computed [Saturated sand]
Ref0
tunnel in dry and saturated sands at the end of tunnel excavation
229
350 Dry sand: 'v, after / 'v0 = 0.33

At X/D = 0 Z
Saturated sand: 'v, after / 'v0 = 0.27 X
Vertical stress along invert of
300
existing tunnel (kPa)
250 Existing tunnel
200 New tunnel
150
100 Computed [Dry sand-before]

Computed [Dry sand-after]
50 Computed [Saturated sand-before]
Computed [Saturated sand-after]
0
0 1 2 3 4
(a)
100
= 20 kPa
(BD, 2009)
50 Z
X
0
Existing tunnel
-50
New tunnel
-100
Computed [Dry sand]

-150
Computed [Saturated sand]
-200
0 1 2 3 4
(b)
Figure 9.12 (a) Vertical stress distribution; (b) incremental vertical stress along the invert of
the existing tunnel for dry and saturated sands
230
10 Z
X
8 Existing tunnel
Gm (MPa)
New tunnel
6
Computed [Dry sand-before]

4 Computed [Dry sand-after]
Computed [Saturated sand-before]
Computed [Saturated sand-after]
2
Dry sand: Gm, after / Gm, before = 0.22 At X/D = 0
Saturated sand: Gm, after / Gm, before = 0.17
0
0 1 2 3 4
Figure 9.13 Mobilized secant shear modulus of soil along the invert of the existing tunnel for
dry and saturated sands
231

along existing tunnel (X/D) Z
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 X
0.0
settlement (/D, %)
Existing tunnel
Normalized tunnel
0.1
New tunnel
0.2
15 mm (LTA, 2000)
0.3
20 mm (BD, 2009)
0.4 Measured [D = 6m] Computed [D = 6m]

Measured [E2,3N5] Computed [D = 3m]
(a)

0 1 2 3 4 5 6 X
0.0
0.1
Existing tunnel
0.2
0.3
0.4
/ max
New tunnel
0.5
0.6
0.7
0.8
0.9
Computed [D = 6m]
1.0
Computed [D = 3m]
(b)
Figure 9.14 Comparison of settlement of the existing tunnel at different tunnel diameters at
the end of tunnel excavation normalized by (a) tunnel diameter; (b) maximum tunnel
settlement
232

Z
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
X
-50
Induced strain at the invert ( )
Existing tunnel
0
50 New tunnel
100
t = 150
150
(ACI, 2001)
200 Measured [D = 6m] Computed [D = 6m]

Measured [E2,3N5] Computed [D = 3m]
tunnel in dry and saturated sands at the end of tunnel excavation
233

0 1 2 3 4 5 6 X
50
Existing tunnel
New tunnel
-50 20 kPa (BD, 2009)
-100
-150
Computed [D = 6m] Computed [D = 3m]
(a)

0 1 2 3 4 5 6 X
1.4
Mobilized vertical stress ratio
1.2 Existing tunnel
1.0
0.8 New tunnel
0.6
0.4
0.2
0.0
Computed [D = 6m] Computed [D = 3m]
(b)
Figure 9.16 (a) Incremental vertical stress; (b) mobilized vertical stress ratio along the invert
of the existing tunnel with different tunnel diameters
234
Chapter 10 Conclusions and further works
Chapter 10 Conclusions and further work
10.1 Summary of this research
Rise in demand for underground space in urban areas has resulted in more constructed tunnels.
The interaction between tunnels becomes important as the clear spacing between adjacent
tunnels decreases. Tunnel driving inevitably induces changes in stress and deformation in the
ground, which can cause structural and serviceability problems in adjacent tunnels. The main
objective of this research is to improve the fundamental understanding of the interaction of
crossing tunnels.
A total eight centrifuge tests were carried out in dry Toyoura sand. The diameter (D) of the
existing and new tunnels was equivalent to 6 m in the prototype scale. The new tunnel was
perpendicularly excavated in-flight in six stages across the existing tunnel. Three-dimensional
tunnel advancement was simulated using a novel technique called a Donut, which can
model the effects of tunnel volume loss equaling 2% and weight loss. Settlement, strain
induced in the lining in the longitudinal and transverse directions and deformation of the
existing tunnel were measured.
The test results were back-analyzed three-dimensionally using the finite element method to
improve the fundamental understanding of stress transfer mechanism, strain induced and
mobilization of stiffness in crossing-tunnel interaction. In addition, numerical parametric
studies were performed to investigate the sensitivity of the parameters that influence the
interaction of crossing tunnels. An advanced hypoplasticity constitutive model with small
strain stiffness was adopted. The key advantage of the hypoplasticity was the capability to
235
simulate stiffness dependency on the state, strain and recent stress history.
Five series of tests were interpreted to study the factors influencing the interaction of crossing
tunnels. The first series investigated the responses of an existing tunnel due to a new tunnel
excavation underneath. In addition, the effects of modeling the tunnel volume loss and weight
loss separately on crossing-tunnel interaction were examined based on analysis from two
tests.
The second series studied the influence of construction sequence and cover depth on the
interaction of perpendicularly crossing tunnels. Two pairs of crossing-tunnel excavations with
different construction sequences and cover depths were analyzed.
In the third series, the effects of pillar depth and shielding on crossing-tunnel interaction were
investigated. Interpretation of the three tests is reported. Note that the pillar depth is the clear
vertical distance between tunnels.
The responses of an existing tunnel due to excavation of new twin tunnels in side-by-side and
vertically stacked arrangements were investigated in the fourth series. The influence of the
arrangement and excavation sequence of the new twin tunnels on the responses of the existing
tunnel was also examined.
Numerical parametric studies were carried out to examine the effects of volume loss, relative
density and tunneling in saturated sand on the interaction of crossing tunnels in the last series.
The results were validated and interpreted along with those from the centrifuge test.
10.2 Effects of volume and weight losses on the interaction of crossing
tunnels
Responses of an existing tunnel due to a new tunnel excavation across beneath were
investigated to improve the understanding of the stress transfer mechanism in the
crossing-tunnel interaction. In addition, the influence of the modeling technique on the
236
interaction of perpendicular crossing tunnels was examined. In one test, the effects of volume
and weight losses were modeled simultaneously. In the other test, the effects of volume loss
only were simulated first followed by weight loss.
10.2.1 Behavior of the existing tunnel due to the new tunnel excavation underneath
Due to the excavation of a new tunnel underneath, the maximum measured settlement of the
existing tunnel was 0.3%D (18 mm in the prototype scale). This settlement exceeded the
permissible limits of serviceability of 15 mm (LTA, 2000). Moreover, the measured tensile
strain and shear stress induced in the longitudinal direction of the existing tunnel exceeded the
cracking tensile strain of 150 (ACI, 2001) and the allowable shear stress limit of 660 kPa
(ACI, 2011), respectively. It suggests that cracks are likely to occur in the tunnel lining.
The reason for the large tunnel settlement and tensile strain induced in the lining was due to a
sharp reduction in the vertical stress at the invert of the existing tunnel at a location directly
above the new tunnel. In addition, the vertical stress increased substantially at the crown of
the existing tunnel as a result of stress transfer in the longitudinal direction of the new tunnel
during the tunnel advancement. The change in stress on the tunnel lining exceeded the
allowable limit of 20 kPa (BD, 2009). This suggests that the structural analysis used when
considering these changes in stress on the tunnel linings should be reviewed.
The existing tunnel was vertically compressed at every stage of excavation of the new tunnel.
This is because the reduction in stress acting on the existing tunnel in the horizontal direction
was larger than that in the vertical direction. The deformation and strain induced in the
transverse direction of the existing tunnel did not exceed the allowable limits suggested by
BTS (2000) and ACI (2001), respectively. Note that the deformation and strain in the tunnel
lining reported in this study were due to tunneling only. Given that in-situ vertical stress was
larger than the horizontal stress (i.e., K0 < 1), the existing tunnel was compressed vertically
even before the advancement of the new tunnel. Thus, the induced vertical compression of the
existing tunnel may still cause structural and serviceability problems.
237
10.2.2 Response of the existing tunnel due to modeling volume and weight losses
separately
The measured settlement of the existing tunnel was 15% larger when only the effects of
volume loss were simulated than that when both effects of volume and weight losses were
modeled simultaneously. This is because the removal of the soil mass inside the new tunnel or
the effects of weight loss led to stress relief, resulting in a reduction in the settlement induced
by volume loss. However, there was about 10% additional settlement of the existing tunnel
when the effects of weight loss were simulated after the effects of volume loss. This is
because the removal of soil from inside the new tunnel resulted in a reduction in the
supporting pressure on the tunnel lining, leading to the vertical compression of the new tunnel.
This in turn induced settlement of the existing tunnel above it.
10.3 Influence of construction sequence and cover depth on perpendicularly
crossing tunnels
Two pairs of tunnel excavations with different construction sequences and cover depths were
analyzed. To consider the construction sequence of crossing tunnels, the tunnel that is
excavated first is the existing tunnel while the tunnel that is constructed later is the new tunnel.
The effects of construction sequence on crossing-tunnel interaction were investigated in the
first pair of tests, where a new tunnel was excavated beneath an existing tunnel in one test,
while the new tunnel advanced above in the other test. The second pair of tests was conducted
with larger cover depth-to-tunnel diameter ratios (C/Ds) than in the first pair in order to study
the effects of cover depth on the construction sequence in crossing tunnels.
10.3.1 Behavior of the existing tunnel due to different construction sequences
The settlement of the existing tunnel when the new tunnel was excavated beneath was
significantly larger than the heave of the existing tunnel when the new tunnel was constructed
above. This is because the reduction in vertical stress acting on the existing tunnel was larger
238
in the former case than in the latter. In addition, the mobilized shear modulus of the soil
around the existing tunnel was smaller in the former tests than in the latter, regardless of the
C/Ds of the crossing tunnels.
The tensile strain along the invert and the shear stress in the longitudinal direction of the
existing tunnel caused by the new tunnel construction beneath was significantly larger than
when the new tunnel was excavated above. This is because the vertical differential
displacement of the existing tunnel was much larger in the former than in the latter. The
tensile strain and shear stress in the existing tunnel only when the new tunnel was constructed
underneath, exceeded the cracking tensile strain (ACI, 2001) and allowable shear stress (ACI,
2011), respectively.
When the new tunnel was excavated beneath, the existing tunnel was compressed vertically.
However, the existing tunnel was elongated in the vertical direction due to the new tunnel
construction above. This is because the reduction in stress acting on the existing tunnel in the
horizontal direction was larger than that in the vertical direction when the new tunnel was
excavated underneath. On the other hand, the decrease in stress acting on the existing tunnel
in the vertical direction was larger than that in the horizontal direction due to the new tunnel
advancement above, irrespective of the C/Ds of the tunnels.
10.3.2 Effects of cover depth on the tunnel-tunnel interaction
Settlement of the existing tunnel due to the new tunnel excavation underneath in the test with
the larger C/Ds was less than that in the test with the smaller C/Ds. This is because the
mobilized shear stiffness of soil around the existing tunnel increased with depth, even though
the reduction in vertical stress acting on the existing tunnel in the test with the larger C/Ds
was greater than that in the test with the smaller C/Ds. On the other hand, heave of the
existing tunnel caused by new tunnel construction above increased with increasing C/Ds of
the existing and new tunnels, due to a larger stress relief acting on the existing tunnel at
greater depths. In addition, the increase in stress relief dominated the increase in the
239
mobilized shear modulus of soil.
10.4 Effects of pillar depth and shielding on crossing multi-tunnel
interaction
To investigate the influence of pillar depth on the interaction of crossing tunnels, two tests
with pillar depth-to-diameter ratios (P/Ds) of 0.5 and 2 were considered. In order to examine
the effects of shielding in the interaction of crossing tunnels, a new tunnel was excavated
underneath two (upper and lower) perpendicularly crossing existing tunnels. The results were
interpreted along with those without the lower existing tunnel (given that P/D of 2 in both
tests).
10.4.1 Behavior of the existing tunnel due to new tunnel excavation beneath with
different pillar depths
In the case of two perpendicularly crossing tunnels (one new and one existing), the measured
maximum tunnel settlement at P/D of 0.5 was about 50% larger than that at P/D of 2. This is
attributed to a smaller shear modulus of the soil in the case of P/D of 0.5 along the invert of
the existing tunnel. The soil stiffness in both tests was the smallest at a location directly above
the new tunnel as a result of a reduction in confining stress and an increase in shear strain
caused by the new tunnel excavation. The other contributing factor to the larger tunnel
settlement at P/D of 0.5 is the vertical stress acting on the tunnel lining at a location directly
above the new tunnel. In the test with P/D of 0.5, the vertical stress reduction at the invert of
the existing tunnel was larger than that in the test with P/D of 2. The settlement of the existing
tunnel only in the case of P/D of 0.5 exceeded the permissible limit (LTA, 2000).
During the new tunnel excavation, the induced tensile strain in the longitudinal direction of
the existing tunnel and deduced shear stress on the tunnel lining were larger for P/D of 0.5
than for P/D of 2. This is due to the larger differential settlement of the existing tunnel when
P/D of 0.5. Differing from P/D of 2, the induced tensile strain and deduced shear stress for
240
P/D of 0.5 exceeded the cracking tensile strain (ACI, 2001) and allowable shear stress (ACI,
2011), respectively.
Different tunnel deformation mechanisms were observed with different P/D ratios. The
existing tunnel was elongated horizontally for P/D of 0.5. This is because stress reduction in
the horizontal direction is greater than that in the vertical direction. The stress relief caused by
the new tunnel not only led to a reduction in the vertical stress at the invert but it also resulted
in substantial stress reduction at the springline of the existing tunnel. On the contrary, the
existing tunnel was elongated vertically as the new tunnel was excavated at P/D of 2.0 since
the reduction in stress in the vertical direction dominated.
10.4.2 Influence of shielding on crossing multi-tunnel interaction
In the case of three tunnels (two existing perpendicularly crossing tunnels above a new
tunnel), the lower existing tunnel shielded the upper existing tunnel from the influence of
the advancing new tunnel underneath, the measured settlement of the upper existing tunnel
was 25% smaller than in the test without the lower existing tunnel. This is because the lower
existing tunnel reduced the effect of stress reduction and decreased the shear strain at the
invert of the upper existing tunnel. These two effects resulted in a larger mobilized shear
modulus at the invert of the upper existing tunnel in the case of two existing tunnels than in
the case without shielding effects (i.e., just one existing tunnel).
Due to the excavation of the new tunnel, the upper existing tunnel was elongated in the
vertical direction. This is because the P/D between the upper existing tunnel and the new
tunnel was 2 (as discussed in section 10.4.1). This resulted in reduced stress acting on the
existing tunnel in the vertical direction larger than that in the horizontal direction.
The lower existing tunnel was also elongated vertically due to the new tunnel excavation. This
is because the shielding effects of the lower existing tunnel minimized the stress redistribution
in the longitudinal direction caused by the advancing new tunnel. As a result, the effects of
241
stress reduction on the lower existing tunnel were dominant, leading to the stress decrease in
the vertical direction.
10.5 Response of an existing tunnel due to new twin tunnel excavation
underneath
The behavior of an existing tunnel due to new twin tunnel excavation in side-by-side and
vertically stacked arrangements was investigated in two centrifuge tests and numerical
back-analysis. The influence of tunnel arrangement and tunneling sequence on the interaction
of multiple crossing tunnels was also examined using a numerical parametric study.
10.5.1 Behavior of the existing tunnel due to side-by-side new twin tunnel excavation
The maximum measured settlement of the existing tunnel caused by the first new tunnel
excavation was still within the recommended limit, however, the maximum settlement and
gradient of the existing tunnel after completion of the second new tunnel exceeded the
permissible limits of 20 mm (BD, 2009) and 1:1000 (LTA, 2000; BD, 2009), respectively. In
addition, the maximum tunnel settlement caused by the second new tunnel excavation only,
was less than that caused by the first new tunnel. This is because the vertical stress reduction
underneath the invert of the existing tunnel due to the second new tunnel was smaller than the
first new tunnel as the decrease in the stress was mainly mobilized by the first new tunnel.
The induced tensile strain and shear stress in the longitudinal direction of the existing tunnel
exceeded the cracking tensile strain (ACI, 2001) and the allowable shear stress (ACI, 2011),
respectively, suggesting that cracks are likely to occur.
The existing tunnel was vertically compressed due to the new twin tunnel excavation. This is
because the P/D between the existing tunnel and the two new tunnels was 0.5 (as discussed
previously in Section 10.4.1). As a result, the stress reduction on the existing tunnel in the
horizontal direction was larger than that in the vertical direction.
242
10.5.2 Response of the existing tunnel due to vertically stacked new twin tunnel
excavation
The maximum measured settlement of the existing tunnel due to the upper (first) new tunnel
excavation did not exceed the allowable limit, however, after the lower (second) new tunnel
excavation, the maximum settlement and gradient of the existing tunnel exceeded the
permissible limits of 20 mm (BD, 2009) and 1:1000 (LTA, 2000; BD, 2009), respectively. The
settlement of the existing tunnel caused by the excavation of the lower (second) new tunnel
alone was smaller than that due to the upper new tunnel. This is because the P/D was greater
for the lower new tunnel than the upper new tunnel (refer to Section 10.4.1). In addition, the
presence of the upper new tunnel provided shielding effects (as discussed in Section 10.4.2)
for the existing tunnel above.
The upper (first) new tunnel (P/D of 0.5) caused the vertical compression of the existing
tunnel while the excavation of the lower (second) new tunnel (P/D of 2) resulted in reduction
in the vertical compression. This is because of the effects of the new tunnel excavation on the
deformation of the existing tunnel at different P/Ds (as discussed in Section 10.4.1). By
combining the deformation caused by the two new tunnels, the deformation induced by the
upper (first) new tunnel dominated over that of the lower (second) new tunnel. As a result, the
stress reduction on the existing tunnel in the horizontal direction was greater than that in the
vertical direction.
10.5.3 Influence of tunnel arrangement and tunneling sequence on the responses of the
existing tunnel due to new twin tunnel excavation
Settlement of the existing tunnel caused by the vertically stacked new twin tunnel excavation
was smaller than the side-by-side new twin tunnel case, irrespective of the tunneling sequence.
This is because the P/D of the lower new tunnel in the vertically stacked tunnel case was
greater than that in each new tunnel in the side-by-side tunnel case.
243
The effects of the sequence of the new twin tunnel excavation show that the maximum
settlement of the existing tunnel when the upper new tunnel advanced first was smaller than
when the lower new tunnel was constructed first. This is because when the upper new tunnel
was excavated first, the tunneling effects of the lower new tunnel on the existing tunnel were
shielded by the upper new tunnel. As a result, the stress reduction underneath the invert of the
existing tunnel when the upper new tunnel advanced first was smaller than when the lower
new tunnel was constructed first.
10.6 Influence of volume loss, relative density and tunneling in saturated
sand on crossing-tunnel interaction
To study the effects of volume loss on the interaction of crossing tunnels, the tunnel volume
loss imposed ranged from 0.5% to 4%. Numerical investigation of crossing-tunnel excavation
in different relative densities, ranging from loose sand (Dr of 30%) to dense sand (Dr of 80%),
were performed. The responses of an existing tunnel due to new tunnel excavation in
saturated sand were also examined. The results from numerical analysis were validated in a
centrifuge test that was carried out in dry medium dense sand (Dr of 64%) at tunnel volume
loss of 2%.
10.6.1 Volume loss effects on the interaction of crossing tunnels
The settlement and strain induced in the longitudinal direction of the existing tunnel exceeded
the recommended limit of 15 mm (LTA, 2000) and 150 (ACI, 2001), respectively, when
the volume loss reached 2%.
The settlement and strain in the lining in the longitudinal direction of the existing tunnel
increased with increasing volume loss at a decreasing rate. This is because the reduction in the
vertical stress was mobilized to a limit so that the vertical stress only decreased slightly with
increasing volume loss. Similarly, the mobilized shear modulus of the soil at the invert of the
existing tunnel located directly above the new tunnel became smaller as the volume loss
244
increased at a decreasing rate.
10.6.2 Response of the existing tunnel due to new tunnel excavation in different relative
densities
The maximum settlement of the existing tunnel due to the new tunnel excavation at a relative
density not more than 64% exceeded the allowable limit (LTA, 2000). In addition, the shape
of the tunnel settlement trough became narrower with increasing relative density.
As the sand became looser, the maximum tensile strain in the tunnel lining in the longitudinal
direction increased. This is due to the increase in differential settlement of the existing tunnel
with decreasing relative density. The magnitude of the maximum induced tensile strain in
every case exceeded the cracking tensile strain (ACI, 2001).
The mobilized shear modulus of the soil before tunnel excavation increased as the soil
became denser, while the mobilized shear modulus after tunnel excavation was almost
constant, irrespective of different relative densities. In addition, the maximum reduction in the
vertical stress increased as the soil became denser. This is because, at a given volume loss, the
change in stress is proportional to the change in stiffness of the soil, which is larger at a higher
relative density.
10.6.3 Interaction of crossing tunnels in saturated sand
The settlement of the existing tunnel caused by the new tunnel excavation in saturated sand
was smaller than that in dry sand. This is because the reduction in stress in the vertical
direction at the invert of the existing tunnel in saturated sand was smaller than that in dry sand,
even though the mobilized shear modulus in saturated sand was less than that in dry sand.
The maximum tunnel settlement and tensile strain induced in the lining in the longitudinal
direction for tunneling the saturated sand was still within the permissible limits suggested by
LTA (2000) and ACI (2001), respectively. In addition, the shape of the settlement trough of
245
the existing tunnel due to the new tunnel excavation in saturated sand was narrower than that
in dry sand.
10.7 Discussion of factors to be considered in crossing-tunnel interaction
In addition to technical issues investigated in this research, other factors such as effects of
vibration and financial aspect on the interaction of crossing tunnels are discussed in this
section.
10.7.1 Effects of vibration on multi-tunnel interaction
For tunneling in rock, the drill-and-blast method is commonly used. Xia et al. (2013) reported
a case history and numerical analysis of the effects of new tunnel blasting on a parallel
existing tunnel in rock. They suggested that the zone of high particle velocity was within 1.6
m from the springline of the new tunnel.
If the tunnel is constructed in dry coarse-grained soil, the drill-and-blast method causes
vibration, which in turn densifies the surrounding soil. As the soil becomes denser around the
advancing face of the tunnel, ground displacement and settlement of the existing tunnel
should be smaller than in the case without vibration. The effects of soil density are discussed
in Section 9.3. If the drill-and-blast method is adopted in saturated sand or clay, excess pore
water pressure may be induced as the loading rate is faster than the dissipation rate. As a
result, the effective stress is reduced and soil may reach a yield state causing larger soil
displacement and tunnel settlement than a case without adopting the drill-and-blast technique.
10.7.2 Optimization between potential damage on the existing tunnel and financial
aspect of tunnel construction
In addition to technical issues, the cost associate with crossing-tunnel interaction should be
considered. As discussed with Chapters 5, 6, 7, 8 and 9, tunnel construction could cause
damages on adjacent existing tunnels. If there are damage or serviceability problems on the
246
existing tunnel, cost of remedial work or service disruption may occur.
As discussed in Chapter 6, an increase in C/D ratio resulted in a reduction in settlement of the

existing tunnel and ground surface settlement. Another measure to mitigate the impact of the
new tunnel excavation on the existing tunnel is to increase the P/D ratio, as discussed in
Chapter 7. However, increasing C/D or P/D inevitably increases the construction cost. If the
tunnel serves as a subway tunnel, construction cost the shaft or subway station increases with
depth of the tunnel.
In order to address this issue, optimization between cost of tunneling impact on existing
tunnels and construction cost should also be carried out. Another factor that should also be
considered is the practicality of the measure adopted to mitigate tunnel excavation impact on
existing tunnels.
10.8 Recommendations for further research
10.8.1 Behavior of the existing tunnel due to new tunnel construction in clay
All centrifuge tests in this research were carried out in dry sand. Long-term ground settlement
caused by tunnel excavation has been reported (Hwang and Moh, 2005; Kao et al., 2009;
Sirivachiraporn and Phienwej, 2012; Boonyarak et al., 2014). For crossing-tunnel interaction
in clay, the time dependent behavior of soil that affects the existing tunnel is another factor
that is still not fully understood. Due to tunnel excavation, the effects of stress relief and
shearing occur. In undrained conditions, the effects of stress relief cause negative excess pore
water pressure. However, the excess pore water pressure induced by the effects of shearing
depends on the overconsolidation ratio. The effects of shearing in an undrained condition on
normally consolidated clay result the positive excess pore water pressure, while shearing on
heavily overconsolidated clay, induces negative excess pore water pressure. As the excess
pore water pressure dissipates, deformation of the clay takes place, resulting in further
displacement of the existing tunnel. To investigate the long-term behavior of crossing-tunnel
247
interaction, centrifuge modeling and numerical analysis of a new tunnel excavation across an
existing tunnel in clay should be carried out.
10.8.2 Influence of the shape of tunnels on crossing-tunnel interaction
The existing and new tunnels in this research were of circular shape. From some case histories,
tunnels have also been constructed in rectangular, cart or horseshoe shapes (Gonzlez-Nicieza
et al., 2008; Mohamad et al., 2010; Liu et al., 2011; Li and Yuan, 2012). Due to the
overburden stress, deformation in a horseshoe tunnel was larger than that in a circular tunnel
(Gonzlez-Nicieza et al., 2008). In addition, soil movement and the stress distribution around
a non-circular new tunnel is complex and still not fully understood. Thus, the effects of tunnel
shape on crossing-tunnel interaction should also be further investigated.
10.8.3 Effects of joint in the lining of the existing tunnel on the interaction of crossing
tunnels
The model existing tunnel tested in this study was a continuous pipe without joints. However,
tunnel linings reported in case histories were commonly constructed with radial and
circumferential joints (Bilotta and Russo, 2013; Standing and Selemetas, 2013).
Teachavorasinskun and Chub-uppakarn (2010) recommended ranges of radial (transverse)
joint stiffness. They also examined the effects of the number and orientation of the joints on
bending moment in the tunnel lining. Wang et al. (2014) adopted three-dimensional numerical
analysis to investigate the influence of a circumferential joint (i.e., longitudinal direction of
the existing tunnel) on the behavior of a tunnel due to differential settlement. It was reported
that deformation of the circumferential joints consists of opening and vertical displacement of
the joints, but the joint vertical displacement was a dominating factor. To date, physical model
test and three-dimensional numerical analysis to investigate crossing-tunnel interaction with
joint are not available. Thus, it is also important to improve understanding of the effects of
tunnel joints on the interaction of crossing tunnels.
248
10.8.4 Effects of boundary condition of the new tunnel
The excavation length of the section in the new tunnel in this study was 3.6 times the tunnel
diameter (refer to Figure 3.2). It can be seen from the figure that the boundary of the new
tunnel was about 0.6D away from the last section to be excavated. The lining of the new
tunnel should be extended to the boundary of the model box or at least 2.5D from the last
section to be excavated. This extended section should be filled with soil or heavy fluid that
has the same density as that outside of the new tunnel. Hence, the boundary effect on
crossing-tunnel interaction can be minimized.
249
References
References
Abrams, A. J. (2007). Earth pressure balance (EPB) tunneling induced settlements in the Tren
Urbano Project, Rio Piedras, Puerto Rico. Thesis (M. Eng.), Dept. of Civil and
Environmental Engineering, Massachusetts Institute of Technology.
Addenbrooke, T. I. and Potts, D. M. (2001). Twin tunnel interaction: Surface and subsurface
effects. The international journal of geomechanics (ASCE), 1 (2): 249-271.
ACI (2001). Control of cracking in concrete structures (ACI 224R-01). American Concrete
Institute (ACI). M.I.
ACI (2011). Building code requirements for structural concrete and commentary (ACI
318M-11). American Concrete Institute (ACI). M.I.
Atkinson, J. H., Richardson, D., Stallebrass, S. E. (1990). Effect of stress history on the
stiffness of overconsolidated soil. Gotechnique, 40 (4): 531-540.
Attewell, P. B. (1978). Ground movements caused by tunnelling in soil. Proc. Int. Conf. on
Large Movements and Structures (ed. J. D. Geddes), Pentech Press, London, 812-948
Bilotta, E. and Russo, G. (2013). Internal forces arising in the segmental lining of an earth
pressure balance-bored tunnel. Journal of Geotechnical and Geoenvironmental
Engineering (ASCE), 139 (10):1765-1780.
Boonyarak, T., Phisitkul, K., Ng, C. W. W., Teparaksa, W. and Aye, Z. Z. (2014). Observed
ground and pile group responses due to tunneling in Bangkok stiff clay. Canadian
Geotechnical Journal, 51 (5): 479495.
Brinkgreve, R. B. J., Engin, E. and Swolfs, W. M. (2013). PLAXIS 3D 2013 User manual.
Plaxis bv, Delft.
BTS (2000). Specification for tunnelling. British Tunnelling Society (BTS). Thomas Telford,
London.
BD (2009). Practice note for authorized persons APP-24. Technical notes for guidance in
250
References
assessing the effects of civil engineering construction / building development on railway

structures and operations. Building department of the government of HKSAR (BD).
Chakeri, H., Hasanpour, R., Hindistan, M. A. and nver, B. (2011). Analysis of interaction
between tunnels in soft ground by 3D numerical modeling. Bulletin of Engineering
Geology and the Environment, 70: 439-448.
Chambon, P., Couillaud, A., Munch, P., Schrmann, A. and Knig, D. (1995). Stabilit du
front de taille dun tunnel: tude de leffet dchelle. Geo 95, 3 p
Chapman, D. N., Ahn, S. K. and Hunt, D. V. L. (2007). Investigating ground movements

caused by the construction of multiple tunnels in soft ground using laboratory model
tests . Canadian Geotechnical Journal, 44 (6): 631643.
Chehade, F. H. and Shahrour, I. (2008). Numerical analysis of the interaction between

twin-tunnels: Influence of the relative position and construction procedure. Tunnelling
and Underground Space Technology, 23 (2): 210214.
Cooper, M. L., Chapman, D. N., Rogers, C. D. F. and Chan, A. H. C. (2002). Movements in

the Piccadilly Line tunnels due to the Heathrow Express construction. Gotechnique, 52
(4): 243-257.
Cording, E. J. (1991). Control of ground movements around tunnels in soil. General Report,
9th Pan-American Conference on Soil Mechanics and Foundation Engineering, Chile.
Deane, A. P. and Bassett, R. H. (1995). The Heathrow Express Trial Tunnel, Proceedings of
the ICE - Geotechnical Engineering, 113 (3): 144-156.
Dyer, M. R., Hutchison, M. T. and Evans, N. (1996). Sudden Valley Sewer: a case history.
Proc. Int. Symposium on Geotechnical Aspect of Underground Construction in Soft
Ground, London (eds. R. J. Mair and R. N. Taylor), Balkema, 671-676.
Garnier, J. (2001). Physical models in geotechnics: state of the art and recent advances. First
Coulomb Lecture, Paris, 3 October. CFMS (eds)
Garnier, J., Gaudin, C., Springman, S. M., Culligan, P. J., Goodings, D. Knig, D., Kutter, B.,
251
References
Phillips, R., Randolph, M. F. and Thorel, L. (2007). Catalogue of scaling laws and
similitude questions in geotechnical centrifuge modelling. IJPMG-International Journal of
Physical Modelling in Geotechnics, 3: 1-23.
Gonzlez-Nicieza, C., lvarez-Vigil, A. E., Menndez-Daz, A. and Gonzlez-Palacio, C.

(2008). Influence of the depth and shape of a tunnel in the application of the
convergenceconfinement method. Tunnelling and Underground Space Technology, 23
(1): 2537.
Gudehus, G. (1996). A comprehensive constitutive equation for granular materials. Soils and
foundations, 36(1): 1-12.
Gudehus, G., Amorosi, A., Gens, A., Herle, I., Kolymbas, D., Man, D., Muir Wood, D.,
Nova, R., Niemunis, A., Pastor, M., Tamagnini, C. and Viggiani, G. (2008). The
soilmodels.info project. International Journal for Numerical and Analytical Methods in
Geomechanics, 32 (12): 1571-1572.
Gudehus, G. and Man, D. (2009). Graphical representation of constitutive equations.

Gotechnique, 59 (2): 147151.
Herle, I. and Gudehus, G. (1999). Determination of parameters of a hypoplastic constitutive

model from properties of grain assemblies. Mechanics of cohesive-frictional materials,
4 (5): 461-486.
Hight, D. W., Gasparre, A., Nishimura, S., Minh, N. A., Jardine, R. J. and Coop, M. R. (2007).
Characteristics of the London Clay from the Terminal 5 site at Heathrow Airport.
Gotechnique, 57 (1): 318.
Hong, S. W. and Bae, G. J. (1995). Ground movements associated with subway tunnelling in
Korea. Underground Construction in Soft Ground (eds K. Fujita and O. Kusakabe),
Balkema, 229-232
Ishihara K. (1993). Liquefaction and flow failure during earthquakes. Gotechnique, 43(3):
351-415.
252
References
Iwasaki, T. and Tatsuoka, F. (1997). Effect of grain size and grading on dynamic shear moduli
of sands. Soils and foundations, 17 (3): 19-35
Klar, A., Vorster, T. E. B., Soga, K. and Mair, R. J. (2007). Elastoplastic solution for
soil-pipe-tunnel interaction. Journal of Geotechnical and Geoenvironmental
Engineering (ASCE), 133 (7): 782-792.
Kolymbas, D. (1991) An outline of hypoplasticity. Archive of Applied Mechanics, 61 (3):

143-151.
Kim, S. H., Burd, H. J. and Milligan, G. W. E. (1998). Model testing of closely spaced tunnels
in clay. Gotechnique, 48 (3): 375-388.
LTA (2000). Code of practice for railway protection. Development and Building Control
Department, Land Transport Authority (LTA), Singapore.
Lee C. J., Wu B. R., Chen H. T. and Chiang K. H. (2006). Tunnel stability and arching effects
during tunneling in soft clayey soil. Tunnelling and Underground Space Technology 21
(2): 119132.
Lee, G. K. T. and Ng, C. W. W. (2005). Effects of advancing open face tunneling on an

existing loaded pile. Journal of Geotechnical and Geoenvironmental Engineering (ASCE),
131 (2): 193-201.
Li, X. G. and Yuan, D. J. (2012). Responses of a double-decked metro tunnel to shield driving
of twin closely under-crossing tunnels. Tunneling and underground space technology,
28: 1830.
Liao, S. M., Peng, F. L. and Shen, S. L. (2008). Analysis of shearing effect on tunnel induced
by load transfer along longitudinal direction. Tunneling and underground space
technology, 23 (4): 421430.
Lim, K. S.G., Hong, C. Y., Wang, Y. and Ng, C. W.W. (2010). Soil-structure interaction of
tunnel excavation beneath existing buried pipeline. The 4th International Conference on
Geotechnical Engineering and Soil Mechanics, November 2-3, 2010, Tehran, Paper No.
253
References
587.
Liu, H. Y., Small, J. C. and Carter, J. P. (2008). Full 3D modeling for effects of tunnelling on
existing support systems in the Sydney region. Tunnelling and Underground Space
Technology, 23 (4): 399420.
Liu, H. Y., Small, J. C. and Carter, J. P. and Williams, D. J. (2009). Effects of tunnelling on
existing support systems of perpendicularly crossing tunnels. Computers and
Geotechnics, 36 (5): 880894.
Liu, J. H. (1990). Construction technical manual for municipal underground engineering in

soft ground. Shanghai (in Chinese).
Liu, H., Li, P. and Liu, J. (2011). Numerical investigation of underlying tunnel heave during a
new tunnel construction. Tunnelling and Underground Space Technology, 26 (2):
276283.
Maeda, K. and Miura, K. (1999). Relative density dependency of mechanical properties of

sands. Soils and Foundations, 39 (1): 69-79.
Mair, R. J. (1993). Developments in geotechnical engineering research: application to tunnels

and deep excavations. Proceedings of the ICE - Geotechnical Engineering, 97 (1), 27-41
Mair, R. J. and Taylor, R. N. (1997). Theme lecture: Bored tunneling in the urban
environment. Proc. 14th International Conference in Soil Mechanics and Foundation
Engineering, Hamburg, Balkema, pp. 2353-2385.
Marshall, A. M., Klar, A. and Mair, R. J. (2010). Tunneling beneath buried pipes: view of soil
strain and its effect on pipeline behavior. Journal of Geotechnical and
Geoenvironmental Engineering (ASCE), 136 (12): 1664-1672.
Marshall A.M. and Mair R.J. (2011). Tunneling beneath driven or jacked end-bearing piles in
sand. Canadian Geotechnical Journal, 48 (2011): 1757-1771.
Marshall, A. M., Farrell, R., Klar, A. and Mair, R. (2012). Tunnel in sands: the effect of size,
depth and volume loss on greenfield displacements. Gotechnique, 62(5): 385-399
254
References
Man, D. (2009). 3D modeling of an NATM tunnel in high K0 clay using two different
constitutive models. Journal of Geotechnical and Geoenvironmental Engineering
(ASCE), 135 (9): 1326-1335.
Moh, Z-C., Ju, D. H. and Hwang, R. N. (1996). Ground movements around tunnels in soft
ground. Proc. Int. Symposium on Geotechnical Aspect of Underground Construction in
Soft Ground, London (eds. R. J. Mair and R. N. Taylor), Balkema, pp. 725-730.
Mohamad, H., Bennett, P. J., Soga K., Mair R. J. and Bowers, K. (2010). Behaviour of an
old masonry tunnel due to tunneling-induced ground settlement. Gotechnique, 60(12):
927938.
Mohamad, H., Soga, K., Bennett, P. J., Mair, R. J. and Lim, C. S. (2012). Monitoring twin
tunnel interaction using distributed optical fiber strain measurements. Journal of
Geotechnical and Geoenvironmental Engineering (ASCE), 138 (8): 957-967.
Ng, C. W. W. (2014). The state-of-the-art centrifuge modelling of geotechnical problems at

HKUST. Journal of Zhejiang University Science A (Applied Physics and Engineering), 15
(1): 1-21.
Ng, C. W. W., Van Laak, P. Tang, W. H., Li, X. S. and Zhang, L. M. (2001). The Hong Kong
Geotechnical Centrifuge. Proc. 3rd Int. Conf. Soft Soil Engineering, Dec., Hong Kong. pp.
225-230.
Ng, C. W. W., Van Laak, P. A., Zhang, L. M., Tang, W. H., Zong, G. H., Wang, Z. L., Xu, G. M.
and Liu, S. H. (2002). Development of a four-axis robotic manipulator for centrifuge
modeling at HKUST. Proc. Int. Conf. on Physical Modelling in Geotechnics, St. Johns
Newfoundland, Canada, pp. 71-76.
Ng, C. W. W. and Lee, G. K. T. (2005). Three-dimensional ground settlements and stress

transfer mechanisms due to open-face tunneling. Canadian Geotechnical Journal, 42 (4):
10151029.
Niemunis, A. and Herle, I. (1997). Hypoplastic model for cohesionless soils with elastic strain
255
References
range. Mechanics of cohesive-frictional materials, 2 (4): 279-299.
OReilly, M. P. and New, B. M. (1982). Settlement above tunnels in the United Kingdom-their
magnitude and prediction. Proceeding of Tunnelling 82, Institution of Mining and
Metallurgy , London, pp. 173-181.
Peck, R. B. (1969). Deep excavations and tunneling in soft ground: state-of-the-art report.
Proceedings of the 7th International Conference on Soil Mechanics and Foundation
Engineering, Mexico City, 25-29 August. Rotterdam: Balkema, pp 225-290.
Remaud, D. (1999). Pieux sous charges laterals: tude exprimentale de leffet de groupe.
Thse de Doctorat de lUniversit de Nantes. 328 p
Shirlaw, J. N., Doran, S. and Benjamin, B. (1988). A case study two tunnels driven in
Singapore. Boulder Bed and in grouted coral sands. Engineering Geology of
Underground Movements (eds F. G. Bell et al), Geological Society Engineering Geology
Special Publication, 5: 93-103
Shirlaw, J. N., Ong, J. C. W., Rosser, H. B., Tan, C. G., Osborne, N. H. & Heslop, P. E.
(2003). Local settlements and sinkholes due to EPB tunnelling. Proceedings of the ICE -
Geotechnical Engineering, 156 (4): 193211.
Standing, J. R. and Burland, J. B. (2006). Unexpected tunnelling volume losses in the

Westminster area, London. Gotechnique, 56 (1): 11-26.
Standing, J. R. and Selemetas, D. (2013). Greenfield ground response to EPBM tunnelling in

London Clay. Gotechnique, 63 (12): 989-1007.
Svoboda, G. T., Man, D. and Bohc, J. (2010). Class A predictions of a NATM tunnel in stiff
clay. Computers and Geotechnics, 37 (6): 817825.
Taylor, R. N. (1995a). Geotechnical Centrifuge Technology. Blackie Academic and

Professional, London.
Taylor, R. N. (1995b). Tunnelling in soft ground in the UK. International Symposium on

Underground Construction in Soft Ground (eds. K. Fujita and O. Kusakabe), New Delhi,
256
References
Balkema, 123-126.
Teachavorasinskun, S. and Chub-uppakarn T. (2010). Influence of segmental joints on tunnel

lining. Tunnelling and Underground Space Technology, 25 (4): 490494.
Todd, C. D. (1975). The potentiometer handbook. McGraw-Hill, New York.
Verruijt, A. and Booker, J. R. (1996). Surface settlements due to deformation of a tunnel in an

elastic half plane. Gotechnique, 46 (4): 753-756.
Verruijt, A. and Strack, O. E. (2008). Buoyancy of tunnels in soft soils. Gotechnique, 58 (6):
513515.
Viana da Fonseca, A., Matos Fernandes, M. and Silva Cardoso, A. (1997). Interpretation of a
footing load test on a saprolitic soil from granite, Gotechnique 47 ( 3): 633-651.
von Wolffersdorff, P. A. (1996). A hypoplastic relation for granular materials with a

predefined limit state surface. Mechanics of cohesive-frictional materials, 1 (3):
251-271.
Vorster, T. E. B., Klar, A., Soga, K. and Mair, R. J. (2005). Estimating the effects of tunneling
on existing pipelines. Journal of Geotechnical and Geoenvironmental Engineering
(ASCE), 131 (11), 1399-1410.
Wang, Z., Wang,L., Li, L. and Wang, J. (2014). Failure mechanism of tunnel lining joints and
bolts with uneven longitudinal ground settlement. Tunnelling and Underground Space
Technology, 40: 300308.
Wu, W., Bauer, E. and Kolymbas, D. (1996). Hypoplastic constitutive model with critical state
for granular materials. Mechanics of materials, 23 (1): 45-69.
Yamaguchi, I., Yamazaki, I. and Kiritani, Y. (1998). Study of ground-tunnel interaction of four
shield tunnels driven in close proximity, in relation to design and construction of
parallel shield tunnels. Tunneling and underground space technology, 13 (2): 289-304.
Yamashita, S., Jamiolkowski, M. and Lo Presti, D.C.F. (2000). Stiffness nonlinearity of three
sands. Journal of Geotechnical and Geoenvironmental Engineering (ASCE), 126 (10):
257
References
929-938.
Yamashita, S., Kawaguchi, T., Nakata, Y., Mikami, T., Fujiwara, T. and Shibuya, S. (2009).
Interpretation of international parallel test on the measurement of Gmax using bender
elements. Soils and Foundations, 49 (4): 631650.
Ye, F., Gou, C. F., Sun, H. D., Liu Y. P., Xia Y. X. and Zhou, Z. (2014). Model test study on
effective ratio of segment transverse bending rigidity of shield tunnel. Tunnelling and
Underground Space Technology, 41: 193-205.
Zhuang, Y., Ellis, E. and Yu, H. S. (2012). Three-dimensional finite-element analysis of

arching in a piled embankment, Gotechnique 62 (12): 11271131.
258
Appendix
Appendix A: Ground surface settlement above the existing

tunnel
259
Appendix

-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
Ground surface settlement (mm)
10
15
20
25
30

Measured [E2N3-V] Computed [E2N3-V]
Measured [E2N3-VW] Computed [E2N3-VW]
Ref-boundary Computed [N3]
E2N3 E2N3-VW
Figure A.1 Ground surface settlement due to different modeling techniques
260
Appendix

-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
10
15
20
25
30

RefX0 Computed [E5N3]
Existing tunnel New tunnel
New tunnel Existing tunnel Existing tunnel New tunnel
New tunnel Existing tunnel
E2N3 E3N2 E3N5 E5N3
Figure A.2 Ground surface settlement due to different construction sequences and cover
depths
261
Appendix

-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
10
15
20
25
30
35
New tunnel Lower existing

tunnel
E2N3 E2N5 E2,3N5
Figure A.3 Ground surface settlement due to different pillar depths and number of existing
tunnels
262
Appendix

parallel to existing tunnel (X/D)
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
10
15
20
25
Centerline of Centerline of right
30
left new tunnel new tunnel
35
Measured [E2N3,3]-Left 1st Computed [E2N3,3]-Left 1st
Measured [E2N3,3]-Right 2nd Computed [E2N3,3]-Right 2nd
Existing tunnel
Left new Right new

tunnel (1st ) tunnel (2nd )
E2N3,3
Figure A.4 Ground surface settlement due to side-by-side new twin tunnel excavation
263
Appendix

-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
10
15
20
25
30
35
Measured [E2N3,5]-Upper 1st Computed [E2N3,5]-Upper 1st
Measured [E2N3,5]-Lower 2nd Computed [E2N3,5]-Lower 2nd
Existing tunnel
Upper new
tunnel (1st )
Lower new
E2N3,5 tunnel (2 nd )
Figure A.5 Ground surface settlement due to vertically stacked vertical twin tunnel
excavation
264
Appendix
Appendix B: Calibration of strain gage on the existing

tunnel
265
Appendix
180
160 -5.0D
Bending moment (kN.m)
140 -4.5D
120 -4.0D
-3.5D
100
-2.5D
80 -2.0D
60 -1.5D
40 -1.0D
-0.5D
20
Center
0
-2 -1 0 1 2
Output voltage (V)
(a)
180
160
5.0D
140 4.5D
120 4.0D
100 3.5D
80 3.0D
2.5D
60
1.5D
40
1.0D
20
0.5D
0
-2 -1 0 1 2
Output voltage (V)
(b)
Figure B.1 Calibration curves for strain gages in the longitudinal direction of the existing
tunnel (a) from 0 to -5D (b) from 0.5D to 5D
266
Appendix
Output voltage (V)

-1 0 1
0
-5
-10 Crown
R-Shoulder
-15 R-Springline
R-Knee
-20
Invert
L-Knee
-25
L-Springline
-30 L-Shoulder
Figure B.2 Calibration curves for strain gages in the transverse direction of the existing
tunnel
267
Appendix
12
Crown
10
L-Springline
Displacement (mm)
R-Springline
8 Invert
0
0 1 2 3 4 5
Output voltage (V)
(a)
15
10
Displacement (mm)
-5 Center Ref0
+1D -1D
-10 +2D -2D
+3D -3D
-15
-10 -8 -6 -4 -2 0 2 4 6 8 10
Output voltage (V)
(b)
Figure B.3 Calibration curves for (a) potentiometers to measure tunnel deformation (b)
LVDTs measuring tunnel settlement
268
Appendix
Appendix C: An example of PLAXIS input file
269
Appendix
_initializerectangular SoilContour 0 0 625 930

_setproperties "Title" "E2N3" "Comments" "" "UnitForce" "N" "UnitLength" "mm" "UnitTime" "day" "Gravity"
9810 "WaterWeight" 1E-5
_borehole 0 0
_soillayer 0
_setsoillayerlevel Borehole_1 1 -750
_set Soillayer_1.Conditions "User-defined"
_save "C:\Ohm\1-E2N3-HP\E2N3.P3D"
_gotostructures
_soilmat "Comments" "" "MaterialName" "ToyouraSand" "Colour" 15262369 "MaterialNumber" 0
"LoadCaseRef0" "" "LoadCaseRef1" "" "LoadCaseRef2" "" "LoadCaseRef3" "" "LoadCaseRef4" ""
"SoilModel" 100 "UserDefinedIndex" 0 "DrainageType" 0 "DilatancyCutOff" False "UndrainedBehaviour" 0
"InterfaceStrength" 0 "ConsiderGapClosure" True "K0PrimaryIsK0Secondary" True "K0Determination" 0
"DefaultValuesAdvanced" True "DataSetFlow" 0 "ModelFlow" 0 "UDModelFlow" 0 "SoilTypeFlow" 0
"LowerUpper" 0 "UsdaSoilType" 0 "StaringUpperSoilType" 0 "StaringLowerSoilType" 0 "M50" 0
"DefaultValuesFlow" False "TablePsiPermSat" "[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]" "SplinePsiPerm" "[0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]" "SplinePsiSat" "[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]" "UserDLLName"
"udsm_hps.dll" "UserModel" "Hypoplas.-sand" 1.81818181818182 0 20 0.0001 0 0 0 0 1.529E-5 1.529E-5 0 0 1
0 0 0 0 0 0 0.422965954991345 0 0.999 0 0 0 0 1.00000000000000000E+015 0.495 0.1 0 30 0.001 2600 0.27
0.61 0.98 1.1 0.5 3 8 4 3E-5 0.2 1 0 0.733 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0.5 0.5 1 0 0 -1060 -2370 0 0 10000000 1 0.0620347394540943 0.00383 1.3774
1.25 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
000000000000000000000000000
_platemat "Comments" "" "MaterialName" "TunnelLining" "Colour" 16711680 "MaterialNumber" 0
"LoadCaseRef5" "" "LoadCaseRef6" "" "LoadCaseRef7" "" "LoadCaseRef8" "" "LoadCaseRef9" "" "IsLinear"
True "IsIsotropic" True "IsEndBearing" False "NormalX" 0 "NormalY" 0 "NormalZ" 0
"XYListNXXEpsilonXX" "[-0.01, 0, 0, 0, 0.01, 0]" "XYListNYYEpsilonYY" "[-0.01, 0, 0, 0, 0.01, 0]"
"XYListQXYGammaXX" "[-0.01, 0, 0, 0, 0.01, 0]" "XYListQXZPsiXX" "[-0.01, 0, 0, 0, 0.01, 0]"
"XYListQYZPsiYX" "[-0.01, 0, 0, 0, 0.01, 0]" "XYListMXXKappaXX" "[-0.01, 0, 0, 0, 0.01, 0]"
"XYListMYYKappaYY" "[-0.01, 0, 0, 0, 0.01, 0]" "XYListMXYKappaXY" "[-0.01, 0, 0, 0, 0.01, 0]" "d" 3 "w"
2.7E-5 "E1" 69000 "E2" 69000 "Nu12" 0.33 "Nu22" 0.33 "Nu13" 0.33 "G12" 25939.8496240602 "G13"
25939.8496240602 "G23" 25939.8496240602 "RayleighAlpha" 0 "RayleighBeta" 0
_polycurve -323 96 0 1 0 0 0 1 0
_set Polycurve_1.OrientationAxis1X 0
270
Appendix
_set Polycurve_1.OrientationAxis1Y 1
_set Polycurve_1.OrientationAxis2Z 1
_add Polycurve_1
_set Segment_1.SegmentType "Arc"
_set ArcProperties_1.Radius 50
_set ArcProperties_1.CentralAngle 360
_set Polycurve_1.x 0
_set Polycurve_1.y 535
_set Polycurve_1.z -300
_extrude (Polycurve_1) 600 0 0
_plate Surface_1
_set Plate_1.Material TunnelLining
_surface Polycurve_1
_arrayr (Polygon_1) 2 600 0 0
_plate Polygon_2
_extrude (Polygon_1) 600 0 0
_delete (Polygon_1)
_polycurve -518 -74 0 1 0 0 0 1 0
_set Polycurve_2.OrientationAxis1X 0
_add Polycurve_2
_set Polycurve_2.z 325
_delete (Polycurve_2)
_polycurve -506 -1063 0 1 0 0 0 1 0
_add Polycurve_2
271
Appendix
_extrude (Polycurve_2) 0 750 0
_plate Surface_2
_polycurve -1199 -665 0 1 0 0 0 1 0
_add Polycurve_3
_surface Polycurve_3
_intersect (Surface_2 Polygon_3)
_intersect (Surface_4 Polygon_5 Polygon_6 Polygon_7 Polygon_8 Polygon_9 Polygon_10 Polygon_4)
_intersect (Surface_5 Polygon_1)
_delete (Surface_4)
_delete (Polycurve_3)
_delete (Surface_3 Surface_12 Surface_11 Surface_10 Surface_9 Surface_8 Surface_7 Surface_2)
_extrude (Surface_28) 0 325 0
_undo
_redo
_delete (Volume_2)
_extrude (Surface_28) 0 750 0
_intersect (Surface_22 Surface_21 Surface_14 Surface_23 Surface_15 Surface_24 Surface_16 Surface_26
Surface_25 Surface_18 Surface_17 Surface_19 Surface_6 Volume_2)
_rename Volume_1 "SoilExisting"
_rename Volume_3 "SoilBack"
_rename Volume_4 "SoilEx6"
272
Appendix
_rename Volume_10 "SoilFront"

_surfdispl Surface_31
_surfdispl (Surface_28 Surface_30)
_surfdispl (Surface_29 Surface_12 Surface_11 Surface_10 Surface_9 Surface_20)
_set SurfaceDisplacement_1.Displacement_y "Fixed"
_set SurfaceDisplacement_1.Displacement_z "Free"
_contraction (Surface_8 Surface_7 Surface_5 Surface_4 Surface_3 Surface_2)
_rename SurfaceContraction_1 "Ex1"
_set Ex1.C 1
_set Ex2.C 1
_set Ex3.C 1
_set Ex4.C 1
_set Ex5.C 1
_set Ex6.C 1
_gotomesh
_refine CS_Surface_1
_refine CS_Polygon_2
_refine CS_SoilExisting_GeneratedSoilVolume_1_Soillayer_1
_refine CS_SoilExisting_GeneratedSoilVolume_1_Soillayer_1
_refine CS_SoilFront_GeneratedSoilVolume_1_Soillayer_1
_refine CS_SoilFront_GeneratedSoilVolume_1_Soillayer_1
273
Appendix
_refine CS_SoilEx1_GeneratedSoilVolume_1_Soillayer_1
_refine CS_SoilBack_GeneratedSoilVolume_1_Soillayer_1
_refine CS_SoilBack_GeneratedSoilVolume_1_Soillayer_1
_mesh 0.05 30 15 256
_viewmesh
_gotowater
_gotostages
_set (Soil_10_1.Material Soil_9_1.Material Soil_8_1.Material Soil_7_1.Material Soil_6_1.Material
Soil_5_1.Material Soil_4_1.Material Soil_2_1.Material Soil_11_1.Material Soil_1_1.Material) InitialPhase
ToyouraSand
_activate Plates InitialPhase
_activate SurfaceDisplacements InitialPhase
_deactivate (CS_SoilExisting_GeneratedSoilVolume_1_Soillayer_1) InitialPhase
_deactivate (CS_SoilBack_GeneratedSoilVolume_1_Soillayer_1) InitialPhase
_deactivate (CS_SoilFront_GeneratedSoilVolume_1_Soillayer_1) InitialPhase
_phase InitialPhase
_setcurrentphase Phase_1
_set Phase_1.Identification "Rising-g-level"
_set Deform_1.SumMWeight 60
_set Deform_1.ResetDisplacementsToZero False
_set Phase_1.Solver "Pardiso (multicore direct)"
_phase Phase_1
274
Appendix
_set Phase_2.Identification "Ex1"

_deactivate (CS_SoilEx1_GeneratedSoilVolume_1_Soillayer_1) Phase_2
_activate Ex1 Phase_2
_phase Phase_2
_deactivate Ex1 Phase_3
_phase Phase_3
_phase Phase_4
_phase Phase_5
_phase Phase_6
275
Appendix
Appendix D: List of publications
276
Appendix
Refereed journal publications:

1. Ng, C. W. W., Boonyarak, T. and Man, D. (2013). Three-dimensional centrifuge and
numerical modeling of the interaction between perpendicularly crossing tunnels.
Canadian Geotechnical Journal. 50 (9): 935-946.
2. Boonyarak, T., Phisitkul, K., Ng, C. W. W., Teparaksa, W. and Aye, Z. Z. (2014).
Observed ground and pile group responses due to tunneling in Bangkok stiff clay.
Canadian Geotechnical Journal. 51 (5): 479495.
3. Ng, C. W. W., Boonyarak, T. and Man, D. (2014). Effects of pillar depth and shielding
on crossing multi-tunnel interaction. ASCE Journal of Geotechnical and
Geoenvironmental Engineering. Under review
4. Boonyarak, T. and Ng, C. W. W. (2014). Effects of construction sequence and cover
depth on crossing-tunnel interaction. Canadian Geotechnical Journal. Under review
5. Ng, C. W. W. and Boonyarak, T. (2014). Multiple tunnel interaction: three-dimensional
centrifuge and numerical modeling. In preparation
277
Appendix
Conference papers:
1. Boonyarak, T. and Ng, C. W. W. (2012). Tunneling effect on pile group response in
Bangkok. Geocongress - State of the Art and Practice in Geotechnical Engineering,
ASCE, Geotechnical Special Publication no. 225, 3119-3128.
2. Lim K. S. G., Boonyarak T. and Ng, C. W. W. (2012). Centrifuge modelling of tunnel
excavation over an existing perpendicular tunnel. Proc. of the 32th Geotechnical
Divisions Annual Seminar, 25 May, Hong Kong Institution of Engineers (HKIE).
319-324.
3. Boonyarak, T, Lim, K. S.G. and Ng, C. W. W. (2012). Three-dimensional centrifuge and
numerical modeling of perpendicularly crossing tunnel interaction. World Tunnelling
Congress, 18-23 May, Bangkok. 904-905. Full paper in CD.
4. Lim, K.S.G., Boonyarak T. and Ng, C.W.W. (2012). Significance of simulating weight
loss in numerical investigation of perpendicularly crossing tunnel interaction. World
Tunnel Congress, 18-23 May, Bangkok. 928-929. Full paper in CD.
5. Boonyarak, T., Ng, C. W. W. and Man, D. (2014). An investigation of perpendicularly
across twin-tunnel interaction. 8th International Conference on Physical Modelling in
Geotechnics. 14-17 January, Perth. 951-957
6. Boonyarak, T. and Ng, C. W. W. (2015). Influence zone of three-dimensional new
tunnel excavation beneath across an existing tunnel. 15th Asian Regional Conference on
Soil Mechanics and Geotechnical Engineering (15ARC), Fukuoka, Japan, 9-13
November. In preparation
278

Three-Dimensional Interaction 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

Three-Dimensional Interaction PDF

Caricato da

Copyright:

Formati disponibili

Three-dimensional interaction of multiple crossing

tunnels: centrifuge and numerical modeling

The Hong Kong University of Science and Technology

in Partial Fulfillment of the Requirements for

the Degree of Doctor of Philosophy

August 2014, Hong Kong

CHAPTER 1 INTRODUCTION ............................................................................................ 1

1.1 Research background ............................................................................................... 1

CHAPTER 2 LITERATURE REVIEW ................................................................................ 5

2.1 Introduction .............................................................................................................. 5

CHAPTER 3 THREE-DIMENSIONAL CENTRIFUGE MODELING ........................... 34

3.1 Introduction ............................................................................................................ 34

CHAPTER 4 THREE-DIMENSIONAL NUMERICAL ANALYSIS................................ 75

4.1 Introduction ............................................................................................................ 75

CHAPTER 5 EFFECTS OF VOLUME AND WEIGHT LOSSES ON

5.1 Introduction .......................................................................................................... 100

CHAPTER 6 INFLUENCE OF CONSTRUCTION SEQUENCE AND COVER DEPTH

6.1 Introduction .......................................................................................................... 122

CHAPTER 7 EFFECTS OF PILLAR DEPTH AND SHIELDING ON CROSSING

7.1 Introduction .......................................................................................................... 147

CHAPTER 8 RESPONSES OF AN EXISTING TUNNEL DUE TO TWIN TUNNEL

CHAPTER 9 NUMERICAL PARAMETRIC STUDY OF INFLUENCE OF VOLUME

9.1 Introduction .......................................................................................................... 200

CHAPTER 10 CONCLUSIONS AND FURTHER WORK ............................................. 235

10.1 Summary of this research..................................................................................... 235

REFERENCES ..................................................................................................................... 250

C Cover depth of tunnel

eco Critical void ratio at zero pressure

edo Minimum void ratio at zero pressure

eio Maximum void ratio at zero pressure

Gm Mobilized shear modulus of soil

I Moment of inertia of tunnel

the initial loading

N an integer by which gravitational acceleration is to be multiplied

r Parameter controlling the rate of degradation of stiffness with strain

n Incremental normal stress acting on a tunnel

Table 2.1 Summary of case studies .......................................................................................... 19

Table 3.1 Scaling laws applied in the centrifuge tests.............................................................. 46

Figure 7.1 Settlement of the existing tunnel .......................................................................... 163

Plate 3.1 Liquid silicone rubber and catalyst ........................................................................... 62

1.1 Research background

Three-dimensional numerical analyses of perpendicularly crossing tunnels constructed in rock,

depth-to-diameter ratio (C/D) of the tunnel.

1.2 Objectives of the research

This research aims to improve fundamental understanding of crossing multi-tunnel interaction

These objectives are achieved by adopting two major methodologies, three-dimensional

1.3 Thesis outline

Chapter 3 presents three-dimensional centrifuge testing. Details of the experimental program,

and testing procedures are given.

Chapter 4 describes three-dimensional numerical analysis. Details of the numerical plan,

Chapter 2 Literature review

stiffness on crossing-tunnel interaction are still not fully understood.

current knowledge of the effects of tunneling on underground structures such as existing

2.2 Ground movement due to tunneling

2.2.1 Surface settlement

of ground surface settlement trough caused by tunnel excavation. By using following

S = Smax exp(-x2/2i2) 2-1

zone caused by tunnel excavation on ground surface.

almost proportionally with depth of tunnel axis according to following equation:

clay and in 0.25 for sand and gravel, respectively.

shape are zt-z0, C/D and VL.

2.2.2 Subsurface settlement

level in Taipei increases with depth.