FOUNDATIONS ON SOFT SOILS FOR KHULNA MEDICAL COLLEGE BUILDINGS IN BANGLADESH

Muhammed H. Kabir1, Mohammad J. Alam2, Arif M. Hamid3 and Abul K.M. Akhtaruzzaman4

ABSTRACT This paper reports on the design and constructional aspects of geotextile included granular mattress foundations of Khulna Medical College in Bangladesh. These foundations were envisaged, designed and constructed on soft soils for four and five storied buildings. Four buildings, including the Academic Building, Nurses Training Centre and two Hostel Buildings were constructed. The granular mattresses were constructed by laying a nonwoven needle punched geotextile on top of the soft soil at the base of excavation, about two metres in depth. A graded mix of crushed brick aggregates and coarse sand was placed on the top of the geotextile. The aggregates were vibro compacted, in layers, to desired density, by using twin drum vibratory rollers. A layer of locally available fine river sand was placed by vibro compaction on top of the coarse aggregate layer. The total thickness of the mattress, comprising the geotextile, coarse and fine aggregates, varied between 1.8 and 2 m depending on the loading condition. Continuous footings were designed for masonry as well as framed structures, which were cast on the mattress. A brief description of the functions of the mattress is presented. The deflection and distortion behaviour of the continuous footings is also presented as comparative studies of cases with and without granular mattresses. INTRODUCTION
88 E 89 90 91 92 E

Khulna City, situated in the south east of Bangladesh, is generally 26 MAP OF underlain by soft Recent swamp deposits containing layers of decomposed BANGLADESH mangrove vegetation. The soft soil areas of Bangladesh showing the location 25 H of Khulna area are shown in Figure 1. Non existence of bearing layer at moderate depths poses problems for construction of buildings between 4 and 24 23 22 21 Map Showing Soft Soil E N G A L B A Y O F B Areas. Fig. 1 H 6 stories, which are the more desired types in this area. Buildings above 6 stories are founded on piles whose length often exceed 30 m. 2 to 3 story buildings are normally founded on shallow foundations. In the year 1994 the senior author, in collaboration with Public Works Department (PWD) of Bangladesh, decided to investigate on the suitability of use of granular mattress foundations for 3 to 5 storied buildings of the Khulna Medical College (KMC). The philosophical and analytical bases of the project came Figure 1 : Map showing soft from previous research under the supervision of the senior author (Kabir et soil areas al. 1992), a brief description of which is presented in this paper. Initially four buildings were taken up. These are the Academic Building (AB), Nurses Training Centre (NTC), Girls Hostel (GH) and Boys Hostel (BH). The AB complex has a framed structure with wings 3 to 5 storied in height. The other three buildings are load bearing wall types with height between 4 and 5 stories. Photographs of the AB and GH buildings are presented in Figures 2 and 3 respectively. The PWD of Bangladesh, the executing authority, has now adopted the construction method in general, and is constructing a dozen more residential buildings by using this technology. The granular mattresses were constructed up to 2 m thick by remove and replacement technique. The construction of the buildings was completed in 1997. Settlements at gauge points were recorded time to time, but not on a regular basis.
Soft Soil ZONE 1 High ZONE 2 ****Soil** *Khulna ****** **3 *** **** * ** ***** * *Organic ** * * * ** Coast Chittagong ** Moderate********* *******Chittagong*** ZONE Low

1 2

Professor, CE Department, Bangladesh University of Engineering and Technology, Dhaka-1000, Bangladesh Lecturer, CE Department, Bangladesh University of Engineering and Technology, Dhaka-1000, Bangladesh 3 PG Student, CE Department, Bangladesh University of Engineering and Technology, Dhaka-1000, Bangladesh 4 Senior Engineer, Beximco Engineering Ltd., Bell Tower, Dhanmondi, Dhaka-1207, Bangladesh

SITE GEOLOGY AND GEOTECHNICS The surffacial landform of Bangladesh is composed of Recent deltaic deposits. There are stretches of areas where soft subsoils occur. These, mostly situated in the coastal areas, are composed of Recent deposits of soft under consolidated to normally consolidated silty clays and clayey silts. Pockets of marshland deposits of soft soils occur also in the interior of the count ry. A m a p of Bangladesh showing the soft soil areas are presented in Figure 1, with location of Khulna City where the project site is situated. The subsoil in the area is composed of Recent soft Figure 2: Academic building soils with interlayers of Figure 3: Girls hostel building decomposed and partially decomposed vegetative organic matters originating from subsidence of mangrove forest. A generalised soil profile of the project site is presented in Figure 4. This figure shows the SPT, DPL, screw plate modulus, moisture condition, undrained shear strength and compressibility parameters. The DPL (Dynamic Probing Light) tests were conducted to ascertain the depth of soft top layer more precisely than that revealed by SPT. A typical consolidation test result and a screw plate load test result is presented in Figures 5 and 6 respectively, showing the consistency of the soft soils.
Plate Modulus Moisture (MPa) Condition

SPT

DPL

Su (kPa)

Cc/(1+ eo)

Soft clay Organic clay

Void Ratio

-10 0 0 4

10 20 0

10 20

4 25

50

25

50

0.1

0.2

0.9 0.8 0.7 0.6 10 100 1000

Depth (m)

8
Soft clay

Pressure (kPa)

12 16 20

PL NMC

LL

Figure 5: Consolidation behavior of Khulna soil

0 5 10 15 20 25 0 (kP ) 100 150 50

Pressure

Figure 4: Typical bore log of Khulna sub-soil (Site: Khulna Medical College) PREVIOUS RESEARCH

Settlement (mm)

200

Figure 6: Screw plate test result in Khulna soil

As a primary study to investigate the deformation and bending behaviour of mattress foundation on soft subgrades a test programme was carried out. Model footing tests were performed under plain strain conditions in a test tank having thick glass sides (Kabir et. al. 1992). A rigid metal footing having width B=100 mm and length L=325 mm was used. Polyurethane foam was used to simulate a thick layer of compressible media. Gravel matress of thickness H equal to 0.5B, 1B and 1.5B were tested. In an effort to simulate the effect of in soil confinement, suction was applied in the gravel layer. This was achieved by encapsulating the gravel layer by a thin polyethylene membrane. The magnitude of suction or confining pressure was variable in these tests. Four tests were conducted for each matress thickness conditions.

These are at suction values of 0 mm, 127 mm, 254 mm and 381 mm of Hg, corresponding to confining pressures of 0, 17, 34 and 51 kPa. The suction was applied from a vacuum source, which was maintained at high vacuum through a vacuum pump. Load was applied to the footing through a proving ring and surface deformations were measured at the footing location and at distances, R = 1.5B, 4.5B and 7.5B from the centre line on either side of the footing.
500 mm Footing-1 B=100 mm H/B =0.5 1.0 25 mm 1.5 Gravel Mat 12.5 mm Footing-2 B=100 mm

15

0* 17*
* Suction in kPa

f (mm)

0.024

10 5

/L

0.016 0.008 0

34*

15 30 45 60 Pressure (kPa) Figure 7: Plot of /L vs. confining pressure

Soft Rubber Settlement Profile of surroundings due Foam to induced settlement of footings (Not to Scale) Rigid Base

0 0.50 0.75

1.00 1.25 H/B

1.50

51*

Figure 8: Diagram of model footing design

Figure 9: Influence chart for estimating differential settlement

The main objective of this research was to investigate the deformation and bending behaviour of granular mattresses rather than their load carrying capacities. The parameter defined as distortion ratio, /L, which is a measure of the curvature of the deflected shape of the granular mattress, was plotted as a function of confining pressure for different H/B ratios and are presented in Figure 7. Where is the maximum deflection (at mid point) and L is the span length of the deflected granular mattress, H is the thickness of the granular mattress and B is the width of the footing.(Kabir et al.1992). These show that the distortion ratio reduces drastically from that under no confining condition even for low increase in confining pressure. For small thickness of gravel layer (H/B=0.5) the distortion ratio remains unaffected due to confining pressure increase between 17 kPa to 51 kPa. Increase in layer thickness from H/B=0.5 to H/B=1 has small affect on the distortion ratio but between values of H/B=1 to 1.5 the effect is profound, which is true for all confining pressures. From distortion ratio point of view it is seen that a thin layer under confined condition may behave like a thick layer under unconfined condition. From the results it can be seen that a gravel layer having H/B=0.5 confined at a low pressure of 17 kPa is almost showing the same value of distortion ratio (/L=0.018) as a gravel layer with H/B=1.5 under unconfined condition. This shows the beneficial effects of the concept of prestressing gravel mattresses. One way of achieving this is by burying the gravel mattress deeper. In an effort to analyze the settlement and distortion behaviour of a pair of footing on gravel mattress foundation on compressible media, an analysis was performed for a pair of model footings, each having width of 100 mm with centre to centre distance of 500 mm. This model is a 9:1 representation of a prototype structure having continuous footings, 900 mm wide, with centre to centre spacing of 4.5 meters. This is a common grid structure in residential constructions in Bangladesh. These model footings, shown in Figure 8, were loaded unequally to such degrees that Footing-1 will produce a deflection of 25 mm in every case when loaded individually. For example for layer thickness ratio H/B=1 and confining pressure of 34 kPa Footing-1 will be subjected to a pressure of 22.5 kN/m2, whereas for H/B=1.5 and under zero confining condition the footing will be subjected to a pressure of 14.0 kN/m2 (Kabir et al. 1992). Similarly Footing-2 is subjected to pressures of such magnitudes that it would produce a deflection of 12.5 mm, in all the individual cases. This means that if the influence of deformation due to Footing-1 does not affect Footing-2 and vice-versa then their differential settlement will equal to 12.5 mm. The influence line approach of deformation analysis using superposition principle was used to analyze the differential settlement of these two footings for different values of layer thickness ratio (H/B) and confining pressure (P). The results of the differential settlement as a function of H/B ratio for different confining pressures are presented in Figure 9. These show the following. The differential settlement remains at its maximum value (12.5 mm) for H/B between 0.5 and 1.5 under no confining condition. That is under these conditions the deformation bowl of one footing does not spread far enough laterally, to affect the other. This is also the case for H/B=0.5 for all confining pressures (between 0

and 51 kPa). But for H/B=1 and 1.5 under confining conditions, deformation bowls spread laterally so far that the differential settlement values are reduced. The reduction of differential settlement with increase in confining pressure is more profound for higher value of thickness ratio (H/B=1.5) than for lower value ((H/B=1). At H/B=1.5 and p=51 kPa the deformation bowls of the two footings interacts to such degree that the differential settlement approaches equal to zero. FUNCTIONS OF GRANULAR MATTRESS AND GEOTEXTILE The functions of the granular mattress for the KMC building foundations may be described as the following. (a) Increase in bearing capacity. There will be an increase under undrained condition, which will increase further due to increase in shear strength of the clay layer as drainage sets in. The highly permeable mattress will allow very fast dissipation of pore pressure, especially from the region immediately underneath the geotextile layer where the pore pressure will be the maximum. (b) Distribution of stress over a large area. (c) Minimize total and differential settlement. An elaborate discussion on these will be presented elsewhere. The geotextile layer acted as separator, filter, containment layer and reinforcement. As separator the geotextile prevented mixing of clean aggregate layer with the clay underneath. This function was crucial during the vibro compaction stage but also quite dominant during the full loading of the footings. On loading of the footings, as the consolidation of the underlying clay sets in, the flow of water will be mainly through the highly permeable geotextile into the mattress layer. At this stage and also during vibro compaction stage the geotextile will work as a filter allowing the pore water to pass freely without allowing soil particles to contaminate the overlaying aggregates. As contaminant layer the frictional geotextile (>3/4) will not allow the aggregates to slip away from the loaded zone during vibro compaction as well as full loading. Without geotextile, would result in a loose boundary layer of aggregates and on loading, larger settlement due to poor stress distribution and slippage of aggregates away from loaded area. Tension membrane reinforcing effect by mobilizing in-plane tension in the geotextile which will result in small shear strains around the geotextile allowing large bearing stresses at low deformations. FOUNDATION FOR KMC BUILDINGS Granular mattress foundations were envisaged, designed and constructed for KMC buildings for a number of reasons. These are, (a) cost and time savings, (b) ease of construction by available local technical skill with the prospect of easy adaptability and (c) sound load carrying capacity, settlement and distortion. The philosophical bases of the project emanated from a case study (Wu and Scheessele, 1982) and previous research in this area on physical models and analytical studies (Kabir et. al. 1992). The mattress system resembles inverted road macadam with the coarsest layer buried as deep as possible. This made the system, used in this study, unique. The effect of confining pressure on stress strain modulus of brick aggregates were investigated using suction triaxial tests (Kabir et. al. 1992). This showed the effect of confining on the stress strain modulus of brick aggregates. To achieve the highest confining effect the coarse aggregate layer is placed as the deepest layer of the mattress. The beneficial effect of confining on settlement and distortion has been discussed earlier. Typical cross sections for the mattress for load bearing wall footing (type A) and continuous column footing for the framed structure (type B) are presented in Figures 10 and 11 respectively.
250

75 DPC 600 Filling Sand 600 or variable Filling Sand 300 375 Coarse Sand Fill 1450 300 300 900 Densified Sand Fill 900
75 CC ( 1:2:4)

PL FGL 1500 600 600

200 300 Sand Type B 380

300

5300

300
1000

1400 600

1500 300

2000

Densified Sand Fill Type A or Type B Densified Aggregate + Sand Fill

( All dimensions are in mm )

( All dimensions are in mm )

Figure 10: Details of wall footing on mattress

Figure 11: Details of column footing on mattress

Analysis and Design of Foundations Most of the literature in the subject of granular mattress foundations on soft soils relates to ultimate carrying capacity. A very small number deal with stress distribution, settlement and distortions of the mattress and structural members placed over such foundations. A brief account is presented here on the bearing capacities, settlements and distortions of footings for KMC buildings.

Bearing Capacity There are now quite established calculation methods for bearing capacity of foundations on granular mattress over soft soils. In the case under consideration, strip footings similar in types shown in Figures 10 and 11 were considered. The ultimate bearing capacity of a strip footing of type A, 1.45 m wide was calculated as 400 kN/m2 while that of type B, having 5.3 m in width was found to be 325 kN/m2. These were calculated in accordance with the formula presented by Lee, White and Ingles (1983). The design soil pressures were 90 and 65 kN/m2 respectively for types A and B. This giving a factor of safety of 4.4 and 5.0 respectively against bearing failure. In the cases under consideration factor of safety against bearing failure were not the dominant criteria of design rather allowable total and differential settlements were the dominant criteria. Modulus of Subgrade Reaction The modulii values for the aggregate layers were established from TRL penetrometer tests and its indirect correlation with modulii. The modulii value for the clay layers were calculated from the undrained shear strength, consolidation and screw plate load tests. For the clays both undrained and drained modulii were calculated. Typical values for sands, aggregates and clay layers are presented in Table 1. The modulii of subgrade reaction, K values, were calculated with appropriate correction for the strip footings. To calculate the modulus of subgrade reaction values the popularly used road design program CIRCLY, developed by Wardle(1977), was used. The program allows calculation of stress and deformation beneath circular loaded areas resting on layered elastic systems. For calculation of modulus of subgrade reaction for type A and type B footings the diameters of loaded areas were taken as 1.45 m and 5.3 m respectively. Three layer system comprising the sand, aggregate and natural soft soils were considered. In each of the case of type A and type B footings four cases were considered. These are, (a) under undrained condition for footing on clay, (b) under drained condition for footing on clay, (c) under undrained condition for footing on mattress and (d) under drained condition for footing on mattress. The modulus of subgrade reaction values are designated as Ku, Kd, Kum and Kd m

Type B

Type A

Materials Used The specifications of materials used are presented in the following. Aggregates: The aggregate layer consisted of 2 parts crushed brick aggregates and 1 part coarse sand. The crushed brick aggregate consisted of 25 mm down graded aggregate conforming to the ASTM grading for concrete aggregates. The coarse sand consisted of river sand from the north east of the country, called SYLHET SAND, having fineness modulus 2.5. Sands: The sand fill layer consisted local river sand having Fineness Modulus (FM) greater than 1.0. Fines passing number 200 sieve was limited to 5% for FM up to 1.5 and 10% for FM greater than 1.5. The sands and aggregates were densified in layers by using twin steel drum vibratory rollers. The densities of the materials were monitored by TRL dynamic penetrometer tests. Geotextile: A geotextile separator and filter layer was placed at the bottom of the excavation on the soft clay layer. A nonwoven needle punched geotextile was used. The weight, grab tensile strength and permeability was greater than or equal to 200 gsm, 750 Newton and 1 x 10-3 m/s respectively.

Table 1 : Elastic modulii for the aggregates and clay layers Aggregate/clay Modulii (MPa) Sand 50 Coarse aggregate 80 Clay Undrained 3.0 Drained 1.0

Table 2: Modulus of subgrade reaction values for wall (type A) and column (types) footings Modulus of subgrade reaction Value in 3 kN/m

Ku Kd Kum Kdm Ku Kd Kum Kdm

4240 1920 13180 8970 1310 600 2170 1250

respectively. Typical values for the wall footing type A and those for column footing type B were calculated and are presented in Table 2. Settlement and distortion under working load Continuous inverted Tee beam type footings were used for load bearing walls and continuous tapered footings were used for column foundations of framed structures. A beam on elastic foundation program based on finite element analysis developed by Hulse and Mosley (1986) was used to analyze both the types of footings. Settlement and distortion under working load dictated the proportioning and design of the footings. As described earlier footing types A and B were considered on the mattresses and clay layers. The deflection of the footings for the foundation cases for wall (type A) and column (type B) footings are presented in Figures 12 and 13 respectively. As mentioned earlier four cases of foundations were considered to provide a comparative representation of cases with and without mattress foundation as well as those under undrained and drained conditions. The results in Figure 12 are for an inverted Tee beam wall footing, 1.45 m wide, 33.2m long having 250 mm wide web and 300 mm thick flange. The results show that provision of the mattress will reduce the settlement to less than 1/3. This is also the order of values for differential settlement for this case. Figure 13 shows results of a typical strip footing for carrying column loads. The footing is 5.3 wide, 410 mm deep and 27.3 m long. It can be seen that provision of mattress foundation more than halved the total settlement as well as the differential settlement of the footing. The record of settlement measurement was not kept on a regular basis. The order of values for maximum settlement for both the cases agreed well with the calculated values. CONCLUDING REMARKS
0 10 20

Wall load = 128.3 kN/m

10 20 Undrained (with mattress)

Drained (with mattress)

30 m

Deflection (mm)

Undrained (without mattress)

30 40 50 60
Drained (without mattress)

Figure 12 : Deflection of wall footing

1106 1033 1062 1034 1008 982 960 936 617 (kN)

0 10 20 30 40 50 60 70 80 90 100 110

10

15

20

25 m

Deflection (mm) (mm)

Undrained (with mattress) Undrained (without mattress) Drained (with mattress)

Drained (without mattress)

Figure 13 : Deflection of column footing.

Geotextile included granular mattress systems were designed and constructed for Khulna Medical College buildings. The beneficial effects of such foundation on soft soils for one framed structure building and three load bearing wall buildings were established. The bearing capacities, settlements and distortions of strip footings for load bearing walls and columns were established. After construction in 1997 the buildings are showing performance compatible with predicted values. This type of foundation may be adopted for Khulna area and future research should take into consideration monitoring of performance using proper instrumentation. REFERENCES Hulse, R and Mosley, W.H. (1986). "Reinforced Concrete design by Computer". Macmillan Education Ltd., UK, 288p. Kabir, M.H., Abedin, M.Z., Siddique, A., Akhtaruzzaman, M. and Amin, M.N. (1992). "Foundations for Soft Soils in Bangladesh". Proceedings of International Seminar on Problems of Lowland Development, ILT, Saga University, Japan, November, N.Murura, M.R. Madav and K. Koga eds., pp. 225-230. Lee, I. K., White, W. and Ingles, O. G. (1983). "Geotechnical Engineering", Pitman, London.. Wardle, L.J. (1977). "Program Circly: A Computer Program for the Analysis of Multiple Complex Circular Loads on Layered Anisotropic Media". CSIRO, Australia. Wu, A.H and Scheessele, D.J. (1982). "A Cost-effective, shallow foundation accommodates three feet of settlement". Civil Engineering, ASCE, Vol. 53, No. 3, pp. 65-67.