Foundation in Difficult Subsoil

2Foundation in Difficult Subsoil Conditions

15th Southeast Asian Geotechnical Society Conference, 22 to 26 November 2004, Bangkok, Thailand

165

Prediction of Differential Settlement of Buildings Induced by Land Subsidence from Deep Well Pumping

N. Phienwej, S. Thepparak & P.H. GiaoSchool of Civil Eng�g., Asian Institute of Technology,P.O. Box 4, Kong Luang, Pathumthani 12120, Thailand [email protected]

Abstract: An FDA code, ConSett, was developed for prediction of time-dependent differential settlement between buildings founded at different depths caused by consolidation of clay layers from the effect of deep well pumping as well as land fill. The numerical solution follows the Crank-Nicholson scheme based on Terzaghi �s one dimensional consolidation theory. The code was tested for its applic abil-ity against land subsidence monitoring data at selected locations in Bangkok and the result was promising. The analysis also indicated that the present piezometric drawdown in the first sand layer of Bangkok was induced by recent deep well pumping not by past geo-logic evolution. The code was further useful to predict long term differential settlement at connection between tunnel and station wall of the Bangkok MRT subway.

1 INTRODUCTION

Bangkok, a metropolis with population over 10 million, has been seriously affected by land subsidence from deep well pumping since 1970s. In early 1980s, the land subsidence reached a criti-cal level as the land in the worst affected area was sinking at the rate as large as 120 mm/year. Till present the land subsidence has not yet been brought under control because of inability of the me-tropolis to be independent on groundwater use. The subsidence affected area has expanded significantly in all directions follow-ing the rapid growth of the city. Currently, the most affected area of 1980s still continues to subside at a rate of 15 to 30 mm/year. Even for the inner city area where deep well pumping no longer exists, subsidence at a small rate of 5 to 10mm/year still occurs. The amount of groundwater extraction increased from 1.2 million cu.m/day in early 1980s to 2.0 million cu.m/day at present, re-sulting in an increase in piezometric drawdown in the production aquifers in the critical area from about 40 m below ground sur-face in early 1980s to 60 m at present. Even in the inner city area, the drawdown has not been recovered.

2 AQUIFERS AND SUBSIDENCE PROBLEMS

Bangkok is located at the head of the Gulf of Thailand on the Lower Chao Phraya Plain. The plain was formed in a geo-logical depression resulted from block faulting during the Tertiary Period. The basin was later filled with clastic sedi-ments consisting of alluvial sand and gravel interbedded with flood plain silts and clays which progressed seaward into del-taic deposits and marine clays. Beneath Bangkok there exist eight principal aquifers with a combined thickness more than 550 m. They consist of sands and gravel separated by rela-tively impermeable stiff to hard clays deposited during sea transgressions. These aquifers constitute the Bangkok multi-layer aquifer system which covers the entire Lower Chao Phraya Plain which extends 250 km to the north of Bangkok city and 150 km from west to east. The production aquifers are the 2nd to 4th aquifer layers located at depths between 100 m and

200 m. Groundwater in the first aquifer (named Bangkok Aqui-fer), which is located at depth between 20 m and 50 m, is no longer potent because of high salinity and contamination.

From foundation engineering point of view, Bangkok is founded on thick soft marine clay, 12 to 18 m in thickness, fol-lowed by a thin layer of medium stiff clay of the same origin. Underlying it is a stiff clay layer known as the �first stiff clay�. Below it is a fine to medium dense silty sand layer known as �Bangkok first sand� which is in fact the upper sub-layer of the first aquifer layer. The top of the sand layer is typically found at a depth of 20 to 22 m. Underlying the first sand layer is a layer of stiff clay which in turn overlies the �Bangkok second sand layer� (the lower sub-layer of the first aquifer layer). The sand layer typically exists at a depth of 40 to 60 m which is the founding depth of the deepest foundation constructed in Bang-kok.

Owing to the low lying and flat topography and the close proximity to the sea, the land subsidence has intensified flood risk of the city. The city is kept dry by a massive flood protection system. Continuing land subsidence necessitates frequent upgrad-ing of the system as it continues to sink with the ground. Apart from the flood threat, deep well pumping causes various founda-tion problems. The situation is rather severe because of an exis-tence of the highly compressible soft marine clay at the ground surface covering the entire metropolis area. Large compression of the soft clay triggered by piezometric drawdown from deep well pumping can cause differential settlement problem of building and other structures. Moreover, the common practice of land fill-ing in most development projects in the low lying Bangkok areas further complicates the settlement problem. Fill height of 1 to 2 m is common and will inevitably cause large compression of the soft clay layer.

Although deep wells extract groundwater from aquifers at depths of 100 to 200 m, the piezometric drawdown is also felt in the shallow soil layers. Monitoring of groundwater condition in the inner city area during the last 30 years showed that the pie-zometric head in the first sand layer had been drawndown 20 to 24 m below the ground surface while the drawdowns in the pro-duction aquifers at deeper depth were at 40 to 50 m from the

166

ground surface. The decline in piezometric head in the First Sand layer triggered the decline in porewater pressure of the overlying clay layers at the surface. The drawdown gradually propagated upward in the clay layers, and by early 1980s it was well in the lower part of the soft clay layer. Consequently, large consolida-tion settlement of the clay occurred, which in long term it may constitute a considerable portion of the overall surface subsi-dence. Monitoring data indicated that approximately 30 to 50% of the total land subsidence in the inner city area was from the compression of the soft and stiff clay layers overlying the first sand layer (Duc, 1999 and Phienwej, 1999). The observation is supported by the results of numerical analyses made be Premchit (1978) and Giao (1997). This phenomenon results in differential settlements of buildings founded at different depths in the subsi-dence affected areas.

Because of the existence of the soft clay layer at the surface, pile foundation is necessary for buildings constructed in Bang-kok. Various depths of piles were used depending on type and size of structures. Small or old buildings are typically founded on short piles with tips extending within the soft clay layer (depth 6-12 m). Medium sized buildings are mostly founded on long piles extending into the stiff clay or first sand layers (depth 20 to 30 m). Tall buildings or large-sized structures usually have pile tips extending into the second sand layer (depth 40 to 60 m). Build-ings on deep piles would experience less subsidence than adjoin-ing structures founded at shallower depths. Distortion and cracks of structures from differential settlements may jeopardize the in-tended function of the structures. While the differential settle-ment induced by a land fill can be prevented or minimized by adopting a proper foundation design, the differential settlement from land subsidence can not. Unlike the land subsidence, ground settlement induced by a land fill is only confined to a small area and can be reliably predicted by common methods of analysis. The differential settlement from land subsidence will not stop as long as the piezometric drawdown in the clay layers from deep well pumping still continues to develop. In fact, this adverse phenomenon was one of the major design considerations in the recent underground infrastructure development projects in Bangkok, such as the MRT subway system. In the project, the an-ticipated long term differential settlement phenomenon carried significant impact on the design of shallow tunnels and tran-sition zones between running tunnels and station walls.

Prediction on magnitude of differential settlements between adjoining structures brought about by land subsidence was usu-ally skeptical due to lack of reliable prediction tools as well as inadequacy of information on soil conditions and past records on piezometric condition at the site. Typically, the predictions were made by extrapolation of available monitoring data using some empirical rules and assumptions. To avail a better prediction tool, the authors have developed a numerical code of the cou-pled flow-consolidation finite difference analysis (FDA) type (Thepparak, 2001).

3 THE NUMERICAL SOLUTION

The numerical solution is developed to predict the magnitude of differential settlement of foundations in Bangkok subsoils due to piezometric drawdown in shallow sand layers induced by deep well pumping well as due to surcharge load from land fill. The solution calculates time-dependent distribution of porewater pressure and settlements at various depths over a time period in a

multi clay-layer system representing the shallow Bangkok sub-soils. Drawdowns in piezometric pressure in the underlying and interbedding sand layers and surcharge load are input as bound-ary conditions. These values may be allowed to change with time for the time period analyzed.

The mechanism of land subsidence due to compression of clay layers from piezometric drawdowns in underlying or inter-bedding sand layers and surcharge load over a wide area can be explained by the Terzaghi�s one-dimensional consolidation the-ory. The basic differential equation of porewater pressure change in a homogeneous clay layer is:

2

2vu uct z

(1)

For a heterogeneous clay layer system the equation becomes:

svu uk S

z z t (2)

where: cv : coefficient of consolidation u : pore water pressure t : time z : depth k : coefficient of permeability Ssv : mv w

The numerical solution using the finite difference formulation of Crank-Nicholson scheme is adopted

i

2i i l i i l i l i i l in l n l n l

i l sv i i l

k k k k k k k kzu 2S u u2 2 2 t 2

i

2i i l i i l i l i i l in n n

i l sv i i l

k k k k k k k kzu 2S u u2 2 2 t 2

(3)

where: ki : permeability at grid point No. i un

i : pore water pressure at grid point No. i and time step n Ssv,i : mv w at grid point No. i

The settlement of each clay layer at a time step, St, is calcu-lated from:

1t

to

eS He

(4)

where: St : magnitude of clay settlement at t time step H : thickness of the clay layer

eo : initial void ratio et : change in void ratio at t time step

A computer program, ConSett, is written using Visual Basic language under Windows Operating System to permit graphical user-interface data inputting and computation. The program routes outputs to Microsoft Excel for graphical presentation.

The assumptions used in developing the code are as follows.

167

A system of a series of clay layers with underlying and interbedding sand layers may be considered. To reflect the heterogeneity in compressibility and permeability in the vertical direction, each clay layer may be repre-sented by a number of clay sub-layers. Permeability of the clay is considered to vary during the consolidation period. A function of permeability value with vertical effective stress is input for such a purpose. Boundary conditions are given in term of changes in piezometric head in the underlying and interbedding sand layers and in surcharge load at the ground surface.

The decrease in permeability of clay during consolidation has a significant effect on the time-rate long-term settlement as illus-trated by the results of a comparative analysis made for the typi-cal Bangkok subsoil profile. Three permeability conditions were compared, namely,

Case 1 (Test 2): Permeability of each clay sub-layer is a func-tion of the initial effective stress. The value remains constant with time. Case 2 (Test 3): Each clay sub-layer has a constant value of permeability which is equivalent to the values used in Case 1. Case 3 (Test 4): Initial values of permeability are the same as in Case 1, but they decrease during consolidation with a func-tion with the increasing effective stress.

The analysis shows that Cases 1 and 2 give similar predic-tions of porewater pressure distribution with depth in the clay layers, while Case 3 gives slightly higher values at large elapsed time (more than 40 years after the occurrence of the drawdown), Fig. 1. In terms of the predicted differential settlement, Case 3 gives significantly smaller value in long term, Fig. 1. The pre-dicted settlement considering permeability decreasing during consolidation gives one-third smaller final settlement than the case of constant permeability values throughout the consolidation period. Therefore, the change in permeability of consolidating clay layers should be considered in the settlement analysis.

y (cm/sec)

0

2

4

6

8

10

12

14

16

18

20

0.00 0.20 0.40 0.60

k-Test 2

k-Test 3

k-initial-Test 4

k-final-Test 4

0

2

4

6

8

10

12

14

16

18

20

0 50 100 150 200 250

Pore Water Pressure (kPa)

Weathered Clay

Soft Clay

Soft Clay

Soft Clay

Stiff Clay

Permeability, 10-6 (cm/sec)

y (cm/sec)

0

2

4

6

8

10

12

14

16

18

20

0.00 0.20 0.40 0.60

k-Test 2

k-Test 3

k-initial-Test 4

k-final-Test 4

0

2

4

6

8

10

12

14

16

18

20

0 50 100 150 200 250


Weathered Clay

Soft Clay

Soft Clay

Soft Clay

Stiff Clay

y (cm/sec)

0

2

4

6

8

10

12

14

16

18

20

0.00 0.20 0.40 0.60

k-Test 2

k-Test 3

k-initial-Test 4

k-final-Test 4

0

2

4

6

8

10

12

14

16

18

20

0 50 100 150 200 250


Weathered Clay

Soft Clay

Soft Clay

Soft Clay

Stiff Clay

Permeability, 10-6 (cm/sec)

Fig. 1. Predicted piezometric drawndowns for different condi-tions of input permeability.

4 VERIFICATION

Verification of the numerical code is made using monitoring data on piezometric pressure and ground settlement at some selected subsidence monitoring stations in Bangkok (AIT, 1981). Input parameters of clay layers were chosen and adjusted within the likely range of Bangkok clay. Monitoring stations at three moni-toring stations were selected for the purpose.

Fig. 2. Predicted time rate settlement using different conditions of input parameter.

4.1 Chulalongkorn University Site

The site is located in the center Bangkok. The first sand layer starts at 20 m depth. Monitoring data on peizometric pressure and settlement at various depths in clays and first sand have been col-lected since 1975. By early 1980s, the piezometric head in the first sand layer was drawndown by slightly over 20 m. Although the chronology of the drawdown at the site is not known, the re-cord showed that deep pumping in inner area of Bangkok began in early 1950s and did not cease until late 1980s. Therefore, two scenarios of the assumptions on the chronology of the drawdown are considered.

Scenario A: Drawdown of to present value occurred gradally from 1950 to 1995.

Scenario B: Drawdown to present value occurred suddenly in 1950.

Table 1. Soil properties at Chulalongkorn University campus. Soil Type Depth k x10-6 Cv Density OCR CR RR

m cm/sec m2 /yr kN/m3

Crust 0-3 0.500 3.00 17.50 5.00 0.10 0.012Soft clay 3-6.5 0.085 2.00 16.00 1.10 0.45 0.080Soft clay 6.5-10 0.062 2.50 16.00 1.20 0.45 0.080Soft clay 10-13 0.040 3.00 16.00 1.35 0.30 0.060Stiff Clay 13-20 0.026 3.00 17.20 1.50 0.10 0.012

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0 20 40 60 80 100 120Time (Years)

Test 2

Test 3

Test 4

Test 2

Test 3

Test 4

168

Chulalongkorn University

0

2

4

6

8

10

12

14

16

18

20

0 50 100 150 200 250

Pore water pressure (kPa)

Weathered Clay

Soft Clay

Soft Clay

Soft Clay

Fig. 3. Predicted versus measured piezometric drawndowns at Chulalongkorn University Campus.

Chulalongkorn University

0

2

4

6

8

10

12

14

16

18

20

0.0 0.2 0.4 0.6 0.8Settlement (m)

Weathered Clay

Soft Clay

Soft Clay

Soft Clay

Fig. 4. Calculated differential settlement with depth at Chu-lalongkorn University campus.

Table 1 shows input soil properties used in the analysis. The computation results in term of the distributions of porewater pressure and settlement with depth in the clay layers using the gradual drawdown (modeled as incremental drawdowns) are shown in Figs. 3 and 4, respectively. The predicted time-dependent differential settlements correspond closely with the observed values, Fig. 5. The assumption of a sudden drawdown in 1950, gave very unrealistic settlement condition compared with the measured values, Fig. 6.

The analysis sheds light on the reason for the drawdown in the first sand layer of Bangkok. Because there were no records of piezometric pressure in the sand layer prior to late 1970s, there has been argument on the reason of the drawdown. Although

0.000.050.100.150.200.250.300.350.400.45

1978 1988 1998 2008 2018 2028 2038 2048 2058

Time-Year

Monitoring Settlement, m (0-27.1 m.)

Calculated Settlement, m (0-20 m.)

Monitoring Settlement, m (1-10 m.)


Fig. 5. Comparison of predicted and measured differential settle-ments at Chulalongkorn University Campus, Scenario A draw-down.

0.000.050.100.150.200.250.300.350.400.45

1978 1988 1998 2008 2018 2028 2038 2048 2058Time-Year


Monitoring Settlement, m (0-27.1 m.)


Monitoring Settlement, m (1-10 m.)

Fig. 6. Comparison of predicted and measured differential settle-ments at Chulalongkorn University Campus, Scenario B draw-down.

most people believe that it was caused by deep well pumping, some argue that it was the result of some geo-hydrological evolution occurring during the past geologic time not related to the recent deep well pumping. The cause of the piezometric drawdowns carries an implication to the design criteria on future load carrying capacity of piles and uplift pressure against under-ground structures constructed in Bangkok. If one considers that the drawdown is the result of recent deep well pumping, a re-bound should be expected in the near future when the pumping is terminated or significantly reduced. Therefore, foundations need to be designed for an anticipated increase in uplift pressure and a decrease in load capacity of piles. On the other hand, if the drawdown was a result of a past geological process the present piezometric conditions in the sand layer would not be much af-fected by any future reduction in deep well pumping and there was no need for consideration on the future adverse foundation condition in the design.

The result of the analysis of the Chulalongkorn University site clearly indicates that the drawdown was a recent phenome-non. The analysis with the assumption of a gradual drawdown since 1950s yields comparable results with the observed values in term of piezometric pressure and differential settlement in the clay layers while that of the sudden drawdown in 1950s does not. Moreover, the sudden drawdown analysis shows a steady state of piezometric pressure in the clay layers more or less 50 years after the drawdown occurred and there would been insignificant con-

169

solidation settlement afterward. This contradicts the actual con-dition in recent time.

4.2 Ramkhamhaeng University Site

The site is located in the area that experienced the highest rate of subsidence in early 1980s. The rate reached 120 mm/year in 1981. From mid 1960s to early 1980s, the area was the newly developed and rapidly expanding residential and industrial zone in the eastern outskirt of the city. Thus it relied solely on groundwater supply. The rate of subsidence remained high dur-ing 1978 to 1986 with the average of 45 to 50 mm/year (Duc, 1999). The assumption that a piezometric drawdown in the first sand layer of 22 m in 1960 gave a good match with the moni-tored piezometric heads in the overlying clay layers in 1980. The calculated pore water pressure distribution with depth is shown on Fig. 7. The predicted differential settlements compare well with the observed data, Fig. 8.

4.3 King Mongkutt Institute of Technology-Lad Krabang Site

Subsidence monitoring station was installed in 1978 at the site which was located in the vicinity of the New Bangkok Interna-tional Airport in the remote eastern outskirt of Bangkok. The present piezometric head in the first sand layer was at 19 m. be-low ground surface in 1980. A good match between the calcu-lated piezometric pressure and differential settlements with the monitoring data is obtained when assuming incremental drops of the piezometric pressure in the sand layer of that amount occur-

0

5

10

15

20

0 50 100 150 200 250Pore Water Pressure (kPa)

Weathered Clay

Soft Clay

Soft Clay

Soft Clay

Fig. 7. Predicted versus measured piezometric drawndowns at Ramkhamhaeng University Campus.

g y

0.00.1

0.20.3

0.4

0.50.60.7

1978 1998 2018 2038 2058 2078

Time-Year

Calculation settlement, m (0-23.25 m.) Monitoring settlement, m (1-24 m.)

Calculated Settlement, m (0-10 m.) Monitoring Settlement, m (1-10 m.)

Calculated Settlement, m (0-12.75 m.) Monitoring Settlement, m (1-12.8 m.)

Fig. 8. Comparison of predicted and measured differential settle-ments at Ramkhamhaeng University Campus.

Fig. 9. Predicted versus measured piezometric drawndowns at KMIT-Ladkrabang Campus.

Fig. 10. Comparison of predicted and measured differential set-tlements at KMIT-Ladkrabang Campus.

ring between 1960 and 1990. The comparisons are shown in Figs. 9 and 10.

The good agreement between the calculated differential set-tlements by the developed numerical code to the monitoring data at these subsidence monitoring stations suggests applicability of the code for prediction on future differential settlements between structures founded on different foundation depths in Bangkok area once the properties of the clay layers and the history and magnitude of the drawdown in the shallow sand layers are known.

5 DIFFERENTIAL SETTLEMENTS IN MRT SUBWAY

In the design of the first Bangkok MRT subway project which has just been completed in 2003, running tunnels constructed us-ing EPB shield tunneling were within the depth of the first stiff clay layer and station boxes were founded on diaphragm walls with tips extending to the second stiff clay or second sand layer. Therefore, allowance for long-term differential settlement to be-induced by future land subsidence at the connections of tunnels and station walls was one of the major design consideration. Pre-diction on the values was rather doubtful due to lack of under-standing on the mechanisms of subsidence in the shallow zone and lack of records on the drawdown. The prediction methods adopted in the design were empirical extrapolation of the moni-tored data available in nearby areas and analysis assuming secon-dary compression effect of the clays.

KMIT- Lad krabang Campus

0

5

10

15

20

25

0 50 100 150 200 250 300Pore Water Pressure (kPa)

Weathered Clay

Soft Clay

Soft Clay

Soft Clay

Stiff Clay

Stiff Clay

-0.2

0.0

0.2

0.4

0.6

0.8

1.0

1978 1998 2018 2038 2058 2078

Time-Year

Calculation settlement, m (0-25 m.) Monitoring settlement, m (1-27.4 m.)Calculation settlement, m (0-10 m.) Monitoring settlement, m (1-27.4 m.)

170

Table 2. Soil conditions and differential settlements between tun-nels and station walls of Bangkok MRTA subway.

Hua Lampong Station

Silom Station

Lad Phrao Staion

Bang Su Station

A. Soil Stratigrahy (Depth in m) Fill/Crust 0-1.5 m 0-4 m 0-2 m 0-2 m Soft clay 1.5-12 m 4-12 m 2-15 m 2-15 m 1st stiff clay 12-22 m 12-25 m 15-25 m 15-25 m 1st sand 22+ m 25-37 m 25 + m 25 + m 2nd stiff clay 37-44 m 2nd sand 44 + m B. Depth of structure (m) Tunnel 14.3 18.0 20.5 15.0 Station wall 18.8 38.5 35.1 18.8 C. Long-term differential settlement from land subsidence in 50 years (mm) 10 15 30 50

Deep well pumping in Inner Bangkok city area where the MRT subway was constructed was terminated in early 1990s but the land subsidence still continues till present at the annual rate of 5-10 mm. The phenomenon is induced by the delayed effect of the piezometric drawdown caused by deep well pumping in the past. Therefore, it is relevant to make prediction of the differential set-tlement using the developed code.

Subsoil stratigraphy and depths of the tunnel and the station wall at these stations are given in Table 2. The predicted differ-ential settlements over the next 50 years assuming no further changes in the piezometric pressure in the first and second sand layers at the first two stations (located in central area) and addi-tional 10-m drawdown in the following 10 years at the other two stations (located in northern area) are given in Table 2. The cal-culated time-rate settlement at Silom Station, the deepest station, is shown in Fig. 10 and the simulated piezometric drawdown in the clay layers is as in Fig. 11. The predicted settlements for the first two stations are small because the tunnels were placed in stiff clay layers not the shallower soft clay and it is assumed no further significant drawdown in the sand layers in the future.

Fig. 10. Predicted differential settlement from future land subsi-dence between running tunnel and station wall, Silom Station.

0

5

10

15

20

25

30

35

40

0 200 400 600Pore Water Pressure (kPa)

Clay

Sand

Clay

Fig. 11. Predicted piezometric drawdowns at Silom Station.

When considering additional 10 m drawdown in the first sand layer, the predicted settlements of the other two stations are 3 to 5 times larger. The values predicted based on the extrapolation of monitoring data and the calculation assuming primary and secon-dary consolidation of the clays adopted in the design of the pro-ject were 10 to 55 mm of all stations over 120 years period. Movement joints that could accommodate 100 mm of differential settlements were provided.

6 CONCLUSIONS

An FDA computer code developed for prediction of differential settlements in shallow clay layers of Bangkok from the effect of deep well pumping shows good comparison of the results with the monitoring data. The analysis simulates time-rate piezomet-ric drawdown and consolidation of the clay layer, thus it is a more rationale approach than other methods such as empirical ex-trapolation of subsidence monitoring data, etc.

REFERENCES

Duc, N. A. 1999. Updating and analysis of Bangkok land subsi-dence caused by deep well pumping with emphasis on shal-low soil settlement, M.Eng. Thesis No. GE-98-1, AIT, Bang-kok, Thailand

Giao, P. H. 1997. Artificial recharge of the Bangkok aquifer sys-tem for the mitigation of land subsidence, Doctoral Disserta-tion No. GE-96-2, Asian Institute of Technology, Bangkok, Thailand

Phienwej, N. 1999. Bangkok land subsidence and its problems in foundation engineering. Proceedings Seminar of the Engi-neering Institute of Thailand , Bangkok, Thailand.

Premchit, J. 1978. Analysis and simulation of land subsidence with special reference to Bangkok, Doctoral Dissertation No.D37, Asian Institute of Technology, Bangkok, Thailand.

Thepparak, S. 2001. Analysis of settlement and compression of shallow soil strata due to drawdown of groundwater in an underlying aquifer from well pumping in Bangkok area, M.Eng. Thesis No. GE-00-01, Asian Institute of Technology, Bangkok, Thailand.

Silom Station

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

0 20 40 60 80 100 120Time (years)

Differential Settlement 0-18.00 m.

Differential Settlement 0-44.00 m.

Differential Settlement between 44.00 m. and 18.00 m.

Differential settlement from 2000

MRT Completion

ïë¬¸ Í±«¬¸»¿¬ ß·¿² Ù»±¬»½¸²·½¿´ Í±½·»¬§ Ý±²º»®»²½»ô îî ¬± îê Ò±ª»³¾»® îððìô Þ¿²¹µ±µô Ì¸¿·´¿²¼

ïéï

ß Ü»·¹² ß°°®±¿½¸ º±® Ð·´»¼ Î¿º¬ ©·¬¸ Í¸±®¬ Ú®·½¬·±² Ð·´» º±® Ô±© Î·» Þ«·´¼·²¹ ±² Ê»®§ Í±º¬ Ý´¿§

ÇòÝò Ì¿²ô ÝòÓò Ý¸±© ú ÍòÍò Ù«» Ù«» ú Ð¿®¬²»® Í¼² Þ¸¼ô Õ«¿´¿ Ô«³°«®ô Ó¿´¿§·¿ ©©©ò¹«»¿²¼°¿®¬²»®ò½±³ò³§

ß¾¬®¿½¬æ Ù»±¬»½¸²·½¿´ ©±®µ ·² ¼»»° ¼»°±·¬ ±º ¸·¹¸´§ ½±³°®»·¾´» ±º¬ ½´¿§ · ±º¬»² ¿±½·¿¬»¼ ©·¬¸ °®±¾´»³ «½¸ ¿ »¨½»·ª» ¼·ºóº»®»²¬·¿´ »¬¬´»³»²¬ô ²»¹¿¬·ª» µ·² º®·½¬·±² ¿²¼ ¾»¿®·²¹ ½¿°¿½·¬§ º¿·´«®»ò Ì®¿¼·¬·±²¿´´§ô °·´» ¿®» ·²¬®±¼«½»¼ ¬± ¿¼¼®» ¬¸» ·«» ±º ¾»¿®ó·²¹ ½¿°¿½·¬§ ¿²¼ »¨½»·ª» ¼·ºº»®»²¬·¿´ »¬¬´»³»²¬ò Ð·´» ¿®» ±º¬»² ·²¬¿´´»¼ ·²¬± ½±³°»¬»²¬ ¬®¿¬«³ ±® �»¬� ·² ±®¼»® ¬± ´·³·¬ ¬¸» ¼·ºº»®»²ó¬·¿´ »¬¬´»³»²¬ ¾§ ®»¼«½·²¹ ¬¸» ±ª»®¿´´ »¬¬´»³»²¬ ±º ¿ ¬®«½¬«®»ò Ø±©»ª»®ô ¬¸· ±´«¬·±² ¹»²»®¿´´§ ±²´§ ¿¼¼®»» ¸±®¬ó¬»®³ °®±¾´»³ ¿±½·¿¬»¼ ©·¬¸ ±º¬ ½´¿§ ¿ °·´» ½¿°¿½·¬§ · ¿´± ·¹²·º·½¿²¬´§ ®»¼«½»¼ ¼«» ¬± ²»¹¿¬·ª» µ·² º®·½¬·±²ò Ì¸· ±º¬»² ®»¼«½» ¬¸» ½±¬ó»ºº»½¬·ª»²» ±º «½¸ �½±²ª»²¬·±²¿´ ±´«¬·±²�ò ×² ¬¸· °¿°»®ô ¿ ¼»·¹² ¿°°®±¿½¸ · °®»»²¬»¼ ·² ©¸·½¸ ¿ �º´±¿¬·²¹� °·´»¼ ®¿º¬ º±«²¼¿¬·±² §¬»³ · ¿¼±°¬»¼ º±® ¿ °®±°±»¼ ¸±«·²¹ ¼»ª»´±°³»²¬ ±² ±º¬ ½´¿§ò Ì¸» �º´±¿¬·²¹� °·´»¼ ®¿º¬ º±«²¼¿¬·±² · ¼»·¹²»¼ ¬± ´·³·¬ ¼·ºº»®»²¬·¿´ »¬¬´»³»²¬ ¿²¼ ·¬ ½±²·¬ ±º ¸±®¬ °·´» ¬®¿¬»¹·½¿´´§ ´±½¿¬»¼ ¿¬ ¿®»¿ ±º ½±²½»²¬®¿¬»¼ ´±¿¼·²¹ ¿²¼ ·²¬»®½±²²»½¬»¼ ©·¬¸ ¿ ®·¹·¼ §¬»³ ±º ¬®·°ó®¿º¬ ¬± ½±²¬®±´ ¼·ºº»®»²¬·¿´ »¬¬´»³»²¬ò Ì¸· º±«²¼¿¬·±² §¬»³ ½±«°´»¼ ©·¬¸ ¿ °®±°»®´§ °´¿²²»¼ ¬»³°±®¿®§ «®½¸¿®¹·²¹ ±º ¬¸» »¿®¬¸ °´¿¬º±®³ ¸¿ ¸±©² ¬± ¾» ª»®§ »ºº»½¬·ª» ¿ ¼»³±²¬®¿¬»¼ ¾§ ³±²·¬±®·²¹ ®»«´¬ ±² ¬¸» ½±³°´»¬»¼ ¬®«½¬«®»ò

ï ×ÒÌÎÑÜËÝÌ×ÑÒ

ß ®»·¼»²¬·¿´ ¿²¼ ½±³³»®½·¿´ ¼»ª»´±°³»²¬ ©¿ ½¿®»º«´´§ °´¿²²»¼ ¿²¼ »¨»½«¬»¼ ¿¬ ¿ ·¬» ±º ¿¾±«¬ ïîðð ¿½®» ¿¬ Þ«µ·¬ Ì·²¹¹·ô Õ´¿²¹ô Ó¿´¿§·¿ô ©¸·½¸ · ¿¾±«¬ ìð µ³ ¬±©¿®¼ ±«¬¸ ©»¬ ±º Õ«¿´¿ Ô«³°«®ò Ì¸· ¼»ª»´±°³»²¬ ©¿ ½±²¬®«½¬»¼ ±ª»® ±º¬ ·´¬§ ½´¿§ô ¬»®³»¼ ¿ Õ´¿²¹ Ý´¿§ øÌ¿² »¬ ¿´òô îððì÷ò Ì¸» °®±°±»¼ ¿²¼ ½±²ó¬®«½¬»¼ ¾«·´¼·²¹ ³¿·²´§ ½±²·¬ ±º ¬©±ó¬±®»§ ¬»®®¿½» ¸±«»ô »³·ó¼»¬¿½¸»¼ ¸±«»ô ½±³³»®½·¿´ «²·¬ô º·ª»ó¬±®»§ ¿°¿®¬³»²¬ ¿²¼ ¿´± ±¬¸»® ¿³»²·¬·» ¾«·´¼·²¹ò

Ý±²¬®«½¬·±² ¿²¼ ³¿·²¬¿·²·²¹ ¾«·´¼·²¹ ±ª»® ¼»»° ¼»°±·¬ ±º ¸·¹¸´§ ½±³°®»·¾´» ±º¬ ½´¿§ · ±º¬»² ¿±½·¿¬»¼ ©·¬¸ °®±¾´»³ «½¸ ¿ »¨½»·ª» ¼·ºº»®»²¬·¿´ »¬¬´»³»²¬ô ²»¹¿¬·ª» µ·² º®·½¬·±² ¿²¼ ¾»¿®·²¹ ½¿°¿½·¬§ º¿·´«®»ò Ì®¿¼·¬·±²¿´´§ô °·´» ¿®» ·²¬®±¼«½»¼ ¬± ¿¼¼®» ¬¸» ·«» ±º ¾»¿®·²¹ ½¿°¿½·¬§ ¿²¼ »¨½»·ª» ¼·ºº»®»²¬·¿´ »¬¬´»³»²¬ò Ì¸» °·´» ¿®» ±º¬»² ·²¬¿´´»¼ ·²¬± ½±³°»¬»²¬ ¬®¿¬«³ ±® �»¬� ·² ±®¼»® ¬± ´·³·¬ ¬¸» ¼·ºº»®»²¬·¿´ »¬¬´»³»²¬ ¾§ ®»¼«½·²¹ ¬¸» ±ª»®¿´´ »¬¬´»³»²¬ ±º ¬¸» ¬®«½¬«®»ò Ø±©»ª»®ô ¬¸· ±´«¬·±² ±²´§ ¿¼¼®»» ¸±®¬ó¬»®³ °®±¾´»³ ¿±½·¿¬»¼ ©·¬¸ ±º¬ ½´¿§ ¿ °·´» ½¿ó°¿½·¬§ · ¿´± ·¹²·º·½¿²¬´§ ®»¼«½»¼ ¼«» ¬± ²»¹¿¬·ª» µ·² º®·½¬·±²ò Ì¸· ±º¬»² ®»¼«½» ¬¸» ½±¬ó»ºº»½¬·ª»²» ±º «½¸ �½±²ª»²¬·±²¿´ ±´«¬·±²�ò ×² ¬¸· °¿°»®ô ¿ ¼»·¹² ¿°°®±¿½¸ · °®»»²¬»¼ ·² ©¸·½¸ ¿ �º´±¿¬·²¹� °·´»¼ ®¿º¬ º±«²¼¿¬·±² §¬»³ · ¿¼±°¬»¼ º±® ¬©±ó¬±®»§ ¬»®®¿½» ¿²¼ »³·ó¼»¬¿½¸»¼ ¸±«» ±² ±º¬ ½´¿§ò ß¬ ¬¸» ¬·³» ±º °®»°¿®¿¬·±² ±º ¬¸· °¿°»®ô ³±®» ¬¸¿² ïððð «²·¬ ±º ¬©±ó¬±®»§ ¿²¼ »³·ó¼»¬¿½¸»¼ ¸±«» «°°±®¬»¼ ¾§ ¬¸» �º´±¿¬·²¹� °·´»¼ ®¿º¬ ¸¿ª» ¾»»² ½±²¬®«½¬»¼ ¿²¼ ®»¿¼§ º±® ±½½«°¿²½§ò Í±³» »¬¬´»³»²¬ ³±²·¬±®·²¹ ®»«´¬ ±² ¬¸» ¾«·´¼·²¹ ¿®» °®»»²¬»¼ ¿²¼ ¼·½«»¼ ·² ¬¸· °¿°»®ò

î ÍËÞÍÑ×Ô ÝÑÒÜ×Ì×ÑÒ

Ì¸» ¿´´«ª·¿´ ¼»°±·¬ ¿¬ ¬¸» ·¬» ¹»²»®¿´´§ ½±²·¬ ±º ª»®§ ±º¬ ¬± º·®³ ·´¬§ ÝÔßÇ «° ¬± ¿ ¼»°¬¸ ±º îë ³ ¬± íð ³ ©·¬¸ °®»»²½» ±º ·²¬»®³»¼·¿¬» ¿²¼§ ´¿§»®ò Ì¸» ·´¬§ ÝÔßÇ ¬®¿¬«³ · ¹»²»®¿´´§ «²¼»®´¿·² ¾§ ·´¬§ ÍßÒÜò Õ´¿²¹ Ý´¿§ ½¿² ¾» ¼·ª·¼»¼ ·²¬± ¬©±

¼·¬·²½¬ ´¿§»® ¿¬ ¼»°¬¸ ±º ïë ³ò Í±³» ±º ¬¸» ½±³°®»·¾·´·¬§ °¿ó®¿³»¬»® ±º Õ´¿²¹ Ý´¿§ ¿®» °®»»²¬»¼ ·² Ú·¹ò ï ¿²¼ ¬¸»» °¿®¿³»ó¬»® °´¿§ ¿ ª·¬¿´ ®±´» ·² »¬¬´»³»²¬ ¿²¿´§» º±® ¬¸» º±«²¼¿¬·±² ¼»ó·¹²ò Ì¸» «²¼®¿·²»¼ ¸»¿® ¬®»²¹¬¸ °®±º·´» ¿²¼ »²·¬·ª·¬§ ±º ¬¸» Õ´¿²¹ Ý´¿§ ¿ ±¾¬¿·²»¼ º®±³ ·²ó·¬« º·»´¼ ª¿²» ¸»¿® ¬»¬ ¿®» ¸±©² ·² Ú·¹ò îò

Ì¸» «²¼®¿·²»¼ ¸»¿® ¬®»²¹¬¸ ±º Õ´¿²¹ Ý´¿§ ·²½®»¿» ¿´³±¬ ´·²»¿®´§ ©·¬¸ ¼»°¬¸ ¿²¼ ¸±© ®»´¿¬·ª»´§ ¸·¹¸ ª¿´«» º±® ¬¸» º·®¬ í ³ ©·¬¸ ¬¸» »¨·¬»²½» ±º ±ª»®½±²±´·¼¿¬»¼ ½®«¬ò Ì¸» «øºª÷ñÐ�½ ®¿ó¬·± ±º Õ´¿²¹ Ý´¿§ · ®»´¿¬·ª»´§ ¸·¹¸ ©·¬¸ ®¿¬·± ±º «øºª÷ñÐ�½ ã ðòì ¿²¼ · ·²¼»°»²¼»²¬ ±º °´¿¬·½·¬§ ·²¼»¨ øÐ×÷ò Û²¹·²»»®·²¹ °®±°»®ó¬·» ±º ¬¸» Õ´¿²¹ Ý´¿§ ¿²¼ ®»´¿¬»¼ ½±®®»´¿¬·±² ¿®» ®»°±®¬»¼ ·² ¬¸» °¿°»® ¾§ Ì¿² »¬ ¿´ò øîððì÷ò

í ÜÛÍ×ÙÒ ßÐÐÎÑßÝØ ÚÑÎ ÐÔßÌÚÑÎÓ ÛßÎÌØÉÑÎÕÍ

Ì¸» ¼»·¹² ¿°°®±¿½¸» º±® º±«²¼¿¬·±² ±º ¬¸» ¾«·´¼·²¹ ±² ª»®§ ±º¬ ±·´ ¸¿ª» ¬± ·²¬»¹®¿¬» ©·¬¸ ¹®±«²¼ ¬®»¿¬³»²¬ ¼»·¹² º±® ¬¸» »¿®¬¸©±®µ ± ¬¸¿¬ ¾±¬¸ ¼»·¹² ¿®» ¬»½¸²·½¿´´§ ½±³°¿¬·¾´» ¿²¼ »ºº·½·»²¬ò ×² Ó¿´¿§·¿ô ³¿²§ º±«²¼¿¬·±² º¿·´«®» ·²ª»¬·¹¿¬»¼ ¾§ ¬¸» ß«¬¸±® øÙ«» »¬ ¿´òô îððì÷ ¸±© ¬¸¿¬ ¬¸»» º¿·´«®» ¿®» »·¬¸»® ½¿«»¼ ¾§ »¬¬´»³»²¬ ±º ¬¸» «¾±·´ ²±¬ ¾»·²¹ ¬¿µ»² ·²¬± ½±²·¼ó»®¿¬·±² ·² ¬¸» º±«²¼¿¬·±² ¼»·¹² ±® ´¿½µ ±º ¹»±¬»½¸²·½¿´ »²¹·²»»®ó·²¹ µ²±©´»¼¹» ±² »¬¬´»³»²¬ °®±¾´»³ò

Ú±® ¬¸» ½«®®»²¬ °®±¶»½¬ô ¾±¬¸ ¬»³°±®¿®§ «®½¸¿®¹·²¹ ¿²¼ °®»ó´±¿¼·²¹ ¬»½¸²·¯«» · ¿¼±°¬»¼ ¬± ½±²¬®±´ ´±²¹ó¬»®³ »¬¬´»³»²¬ ±º ¬¸» «¾±·´ «²¼»® ¬¸» ´±¿¼ º®±³ ¬¸» º·´´ ¿²¼ ¾«·´¼·²¹ ¬± ¾» °´¿½»¼ ±² ¬±° ±º ·¬ò Ù»²»®¿´´§ô ¬¸» ²»¬ º·´´ ¸»·¹¸¬ ¿¬ ¬¸» ·¬» · ¿¾±«¬ ðòë ³ ¬± ïòð ³ ¿²¼ ¬¸» ¿ª»®¿¹» «²·º±®³ ´±¿¼ º®±³ ¬¸» ¬©±ó¬±®»§ ¬»®®¿½» ¸±«» · ¿¾±«¬ îë µÐ¿ò Ì¸» ¬»³°±®¿®§ «®½¸¿®¹·²¹ ¸»·¹¸¬ ®¿²¹» º®±³ î ³ ¬± ë ³ ¼»°»²¼·²¹ ±² ¬¸» ¿ª¿·´¿¾´» ©¿·¬ó·²¹ °»®·±¼ò

ßº¬»® ¬¸» «¾±·´ ¸¿ ¿½¸·»ª»¼ ¬¸» ®»¯«·®»¼ °»®½»²¬¿¹» ±º »¬ó¬´»³»²¬ ¿²¼ ª»®·º·»¼ «·²¹ ß¿±µ¿� ³»¬¸±¼ øß¿±µ¿ô ïçéè÷ô ¬¸» ¬»³°±®¿®§ »¿®¬¸ º·´´ ¿®» ®»³±ª»¼ ¿²¼ ¬¸» ½±²¬®«½¬·±² ±º ¬¸» º±«²¼¿¬·±² ¾»¹·²ò

ïéî

ð ï î í ì»±

íë

íð

îë

îð

ïë

ïð

ë

ð

ïðð îðð íðð ìððÐ½ù øµÐ¿÷

î í ì ë ê é è çÑÝÎ

ðòë ï ïòëÝÎãÝ½ñïõ»½

ðòðë ðòï ðòïëÎÎãÝ®ñïõ»±

ïð îð íð

Ýªø³îñ§®÷

Ú·¹ò îò Ë²¼®¿·²»¼ ¸»¿® ¬®»²¹¬¸ ¿²¼ »²·¬·ª·¬§ ±º Õ´¿²¹ Ý´¿§ øº®±³ Ì¿² »¬ ¿´ò îððì÷ò

ì ÜÛÍ×ÙÒ ßÐÐÎÑßÝØ ÚÑÎ ÚÑËÒÜßÌ×ÑÒÍ ÑÚ ÌÉÑóÍÌÑÎÛÇ ÌÛÎÎßÝÛ ßÒÜ ÍÛÓ×óÜÛÌßÝØÛÜ ØÑËÍÛÍ

ìòï Ô±¿¼·²¹ ±º ¬©±ó¬±®»§ Ø±«» ¿²¼ Ü»·¹² Ý®·¬»®·¿

Ù»²»®¿´´§ô ¬¸» ´±¿¼·²¹ ±º îó¬±®»§ ¸±«» ¿®» ¸·¹¸»¬ ¿¬ ¬¸» ½±´ó«³² ¿²¼ ®¿²¹» º®±³ ïð µÒ ¬± íêð µÒò Ì¸» ´·²» ´±¿¼ º®±³ ¬¸» ¾®·½µ ©¿´´ ®¿²¹» º®±³ ç µÒñ³ øìòë� ¾®·½µ ©¿´´÷ ¬± ïê µÒñ³ øç� ¾®·½µ ©¿´´÷ ¿²¼ ¬¸» «²·º±®³ ´·ª» ´±¿¼ ¿½¬·²¹ ±² ¬¸» ¹®±«²¼ º´±±® ®¿º¬ ®¿²¹» º®±³ îòë µÒñ³î ¬± íòð µÒñ³î ¿ °»® ®»½±³³»²¼»¼ ª¿´«» ¹·ª»² ¾§ ÞÍêíççæ Ð¿®¬ ïæ ïççêò Ì¸» ³¿·² ¼»·¹² ½®·¬»®·±² º±® ¬¸» ¬©±ó¬±®»§ ¸±«» · ¬± ´·³·¬ ¬¸» ®»´¿¬·ª» ®±¬¿¬·±² ø¿²¹«´¿® ¼·¬±®¬·±²÷ ¬± ïñíëð øÍµ»³°¬±² ú Ó¿½Ü±²¿´¼ô ïçëê÷ ¬± °®»ª»²¬ ½®¿½µ·²¹ ·² ©¿´´ ¿²¼ °¿®¬·¬·±²ò Ì¸» ´±²¹ó¬»®³ ¬±¬¿´ »¬¬´»³»²¬

øîð §»¿®÷ ¸±«´¼ ²±¬ ¾» ³±®» ¬¸¿² éë ³³ ¿º¬»® ½±³°´»¬·±² ±º ¬¸» ¾«·´¼·²¹ò Ûª»² ©¸»² ¬¸» ¬±¬¿´ »¬¬´»³»²¬ · ´¿®¹»® ¬¸¿² ¬¸» ¼»ó·¹²»¼ ª¿´«»ô ¬¸» ¾«·´¼·²¹ ©·´´ ¬·´´ ²±¬ ½®¿½µ ¿ ¼·ºº»®»²¬·¿´ »¬ó¬´»³»²¬ · ½±²¬®±´´»¼ò

ìòî Ü»·¹² Ó»¬¸±¼±´±¹§

Ó¿²§ ½¿» ±² ¬¸» «» ±º »¬¬´»³»²¬ ®»¼«½·²¹ °·´» ¬± «°°±®¬ ¿ ¬®«½¬«®» ¸¿ª» ¾»»² ®»°±®¬»¼ ø»ò¹ò Ô±ª»ô îððíô Ç¿³¿¸·¬¿ »¬ ¿´òïççì ¿²¼ Þ«®´¿²¼ ú Õ¿´®¿ô ïçèê÷ò Î»½»²¬ «²¼»®¬¿²¼·²¹ ¿²¼ ¿¼óª¿²½» ·² ¬¸» ¼»·¹² ¿²¼ ¿²¿´§· ±º °·´»¼ ®¿º¬ º±«²¼¿¬·±² ø»ò¹ò Ð±«´±ô îððï ¿²¼ Î¿²¼±´°¸ô ïççì÷ ¸¿ª» ¿´± »²¿¾´»¼ °®¿½¬·½·²¹ »²¹·²»»® ¬± ¿¼±°¬ ¬¸» ½±²½»°¬ ±º »¬¬´»³»²¬ ®»¼«½·²¹ °·´» ©·¬¸ ¹®»¿¬»® ½±²º·¼»²½»ò

Ì¸» ¿°°®±¿½¸ ¿¼±°¬»¼ ·² ¬¸· °¿°»® ¹»²»®¿´´§ º±´´±© ¬¸» ®»½ó±³³»²¼¿¬·±² ±º Ð±«´± øîððï÷ ©¸»®» º±«® ½·®½«³¬¿²½» ·² ©¸·½¸ ¿ °·´» · °®±ª·¼»¼ ¾»²»¿¬¸ ¿ ½±²½»²¬®¿¬»¼ ´±¿¼ ø·ò»ò ½±´ó«³²÷æ

¿÷ Ý±²¼·¬·±² ïæ ·º ¬¸» ³¿¨·³«³ ³±³»²¬ ·² ¬¸» ¬®«½¬«®¿´ ³»³¾»® ¾»´±© ¬¸» ½±´«³² »¨½»»¼ ¬¸» ¿´´±©¿¾´» ª¿´«» º±® ¬¸» ¬®«½¬«®¿´ ³»³¾»®ò

¾÷ Ý±²¼·¬·±² îæ ·º ¬¸» ³¿¨·³«³ ¸»¿® ·² ¬¸» ¬®«½¬«®¿´ ³»³¾»® ¾»´±© ¬¸» ½±´«³² »¨½»»¼ ¬¸» ¿´´±©¿¾´» ª¿´«» º±® ¬¸» ¬®«½¬«®¿´ ³»³¾»®ò

½÷ Ý±²¼·¬·±² íæ ·º ¬¸» ³¿¨·³«³ ½±²¬¿½¬ °®»«®» ¾»´±© ¬¸» º±«²¼¿¬·±² »¨½»»¼ ¬¸» ¿´´±©¿¾´» ¼»·¹² ª¿´«» º±® ¬¸» ±·´ò

¼÷ Ý±²¼·¬·±² ìæ ·º ¬¸» ´±½¿´ »¬¬´»³»²¬ ¾»´±© ¬¸» ½±´«³² »¨½»»¼ ¬¸» ¿´´±©¿¾´» ª¿´«»ò

Ì¸» ±²´§ ¼·ºº»®»²½» ¾»¬©»»² Ð±«´± øîððï÷ ®»½±³³»²¼¿¬·±² ¿²¼ ¬¸» °®±°±»¼ ¿°°®±¿½¸ ¾§ ¬¸» ß«¬¸±® · ¬¸¿¬ Ð±«´± øîððï÷ ®»½±³³»²¼¿¬·±² ¿®» ±²´§ º±® ¬·ººñ¼»²» ±·´ò Ì¸» ¿°°®±¿½¸ ¼»ó½®·¾»¼ ·² ¬¸· °¿°»® ¸±©»ª»®ô ¸¿ ¾»»² »¨¬»²¼»¼ º±® ª»®§ ±º¬ ±·´ò

×² ±®¼»® ¬± ¿¼±°¬ ¬¸» ½±²½»°¬ ±º »¬¬´»³»²¬ ®»¼«½·²¹ °·´»ô ¬¸» º±«²¼¿¬·±² ®¿º¬ ³«¬ ¾» ¿¾´» ¬± °®±ª·¼» ¿¼»¯«¿¬» ¾»¿®·²¹ ½¿°¿½·¬§ ·² ¬¸» º·®¬ °´¿½» ¿²¼ ¬¸» °·´» ¿®» ±´»´§ ·²¬®±¼«½»¼ ¬± ½±²¬®±´ ¼·ºóº»®»²¬·¿´ »¬¬´»³»²¬ ©·¬¸·² ¿´´±©¿¾´» ª¿´«» ø»ò¹ò ïñíëð÷ô ¿²¼

ð íð êð çð«øºª÷øµÐ¿÷

îë

îð

ïë

ïð

ë

ð

Ì±¬¿´ ²± ±º º·»´¼ ª¿²» ¸»¿®¬»¬ ´±½¿¬·±² ã çç

Ì±¬¿´ ²± ±º ¬»¬ ã ïîêð

î ì ê è ïðÍ»²·¬·ª·¬§

Ú·¹ò ïò Ý±³°®»·¾·´·¬§ °¿®¿³»¬»® º±® Õ´¿²¹ Ý´¿§ øº®±³ Ì¿² »¬ ¿´òô îððì÷ò

ïéí

¿´± ®»¼«½» ¬¸» ¬®»» ±² ¬¸» ¬®«½¬«®¿´ ³»³¾»®ò ß ·³°´» ½¿´ó½«´¿¬·±² ¾¿»¼ ±² ¾»¿®·²¹ ½¿°¿½·¬§ ¬¸»±®§ ©±«´¼ ·²¼·½¿¬» ¬¸¿¬ ¬¸» ®¿º¬ º±® ±²» ¾´±½µ ±º ïî «²·¬ ±º ¬»®®¿½» ¸±«» ¬§°·½¿´´§ ³»¿«®ó·²¹ ïë ³ ¨ èð ³ ¿²¼ ±²» ¾´±½µ ±º »³·ó¼»¬¿½¸»¼ ¸±«» ¬§°·½¿´´§ ³»¿«®·²¹ îð ³ ¨ îð ³ ©±«´¼ °®±ª·¼» ¿¼»¯«¿¬» ¾»¿®·²¹ ½¿°¿½ó·¬§ò Ø±©»ª»®ô ¬¸» ½±²½»²¬®¿¬»¼ ½±´«³² ´±¿¼ ©±«´¼ ·³°±» ´¿®¹» ½±²¬¿½¬ °®»«®» ¾»´±© ¬¸» º±«²¼¿¬·±² ¿²¼ ©±«´¼ ´»¿¼ ¬± »¨½»ó·ª» ´±½¿´ »¬¬´»³»²¬ ´»¿¼·²¹ ¬± ½®¿½µ ±² ¾«·´¼·²¹ò Ì¸»®»º±®»ô Ý±²¼·¬·±² í ¿²¼ Ý±²¼·¬·±² ì ¿®·» ©¸·½¸ ²»½»·¬¿¬» ¬¸» ·²¬®±¼«½ó¬·±² ±º »¬¬´»³»²¬ ®»¼«½·²¹ °·´»ò Ì¸» »¬¬´»³»²¬ ®»¼«½·²¹ °·´» ¿®» ¼»·¹²»¼ ¿ º®·½¬·±² °·´» ¿²¼ ¬¸· »´·³·²¿¬» ¬¸» ®·µ ±º ¬®«½ó¬«®¿´ º¿·´«®» ±® ·²¿¼»¯«¿½§ ±º °·´» ¼«» ¬± ²»¹¿¬·ª» µ·² º®·½¬·±²ò

Ì¸» ¬®«½¬«®¿´ ³»³¾»® ±º ¬¸» º±«²¼¿¬·±² §¬»³ ½±²·¬ ±º ½±³¾·²¿¬·±² ±º ¬®·° ¿²¼ ®¿º¬ò Ì¸· §¬»³ · ¿¼±°¬»¼ ¬± ³·²·ó³·¦» ¬¸» ¬¸·½µ²» ±º ¬¸» ®¿º¬ º±® ³¿¨·³«³ »½±²±³·½ ¾»²»º·¬ ©¸·´» ²±¬ ¿½®·º·½·²¹ ¬¸» ®»¯«·®»¼ ®·¹·¼·¬§ò Ì¸»®»º±®»ô ¬¸» ¬®·° »®ª» ¬¸» ¼«¿´ °«®°±» ±º °®±ª·¼·²¹ ¬¸» ®»¯«·®»¼ ®·¹·¼·¬§ ¬± ¬¸» º±«²¼¿¬·±² §¬»³ ¿²¼ ¿´± ¿ °·´» ½¿° ¬± ¼·¬®·¾«¬» ¬¸» ½±´«³² ´±¿¼ ¬± ¬¸» °·´»ò É·¬¸ ¬¸» ¬®·° ´±½¿¬»¼ ¼·®»½¬´§ ¾»²»¿¬¸ ¬¸» ½±´«³²ô Ý±²¼·¬·±² ï ¿²¼ îô ©¸·½¸ ¿®» ¹±ª»®²»¼ ¾§ ¬®«½¬«®¿´ ½±²·¼»®¿¬·±²ô ¿®» ²± ´±²¹»® ½®·¬·½¿´ò

ìòí ß²¿´§·

Ì¸» ´±½¿¬·±² ±º ¬¸» »¬¬´»³»²¬ ®»¼«½·²¹ °·´» ¿ ¼»¬»®³·²»¼ ¾¿»¼ ±² Ý±²¼·¬·±² í ¿²¼ ì ³¿·²´§ ½±²½»²¬®¿¬» ¿¬ ½±´«³² ´±½¿ó¬·±² ¿²¼ ¿´±²¹ ¬¸» ´±²¹ °¿² ±º ´·²» ´±¿¼ò É·¬¸ ¬¸» °·´·²¹ ´¿§±«¬ ½±²º·®³»¼ ¿²¼ ¬¸» º®¿³·²¹ ±º ¬¸» ¬®·°ó®¿º¬ ½±³°´»¬»¼ô ¼»¬¿·´»¼ ¿²¿´§· ±º ¬¸» º±«²¼¿¬·±² §¬»³ ¬± ¼»¬»®³·²» ¬¸» ¬®»» ·²ó¼«½»¼ ±² ¬¸» ¬®«½¬«®¿´ ³»³¾»® ¿®» ½¿®®·»¼ ±«¬ º±® «¾»¯«»²¬ ¬®«½¬«®¿´ ¼»·¹²ò Ì¸· ½¿² ¾» ½¿®®·»¼ ±«¬ «·²¹ ½±³³»®½·¿´´§ ¿ª¿·´¿¾´» ¬®«½¬«®¿´ ¿²¿´§· ±º¬©¿®»ò

Ø±©»ª»®ô ¼«» ¬± ¬¸» ´·³·¬¿¬·±² ±º ¬®«½¬«®¿´ ¿²¿´§· ±º¬ó©¿®» ©¸»®» «°°±®¬ ¿®» ««¿´´§ ³±¼»´´»¼ «·²¹ «²½±«°´»¼ °®·²¹ ½±²¬¿²¬ ±® É·²µ´»® º±«²¼¿¬·±²ô ·¬ · ²»½»¿®§ ¬± ¼»¬»®ó³·²» ¬¸» ¿°°®±°®·¿¬» °®·²¹ ½±²¬¿²¬ ¬± ¿½½±«²¬ º±® ¬¸» ¿½¬«¿´ ¾»¸¿ª·±«® ±º ¬¸» º±«²¼¿¬·±² §¬»³ò Ì¸» ´·³·¬¿¬·±² ±º ¬¸» É·²µ´»® º±«²¼¿¬·±² ¿ ¸·¹¸´·¹¸¬»¼ ¾§ Ð±«´± øîððð÷ ³«¬ ¾» ½´»¿®´§ «²¼»®¬±±¼ ·² ±®¼»® ¬± °®±¼«½» ³»¿²·²¹º«´ ¿²¿´§· ®»ó«´¬ò Ì¸» ¼»¬¿·´»¼ ¿²¿´§· ½¿®®·»¼ ±«¬ ½¿² ¾» ¾®±¿¼´§ ¼·ª·¼»¼ ·²¬± ¬©± ½¿¬»¹±®·»æ

¿÷ Ô±½¿´ ¬®»» ¿¬ ´±½¿¬·±² ±º ½±²½»²¬®¿¬»¼ ´±¿¼ ¾÷ Ñª»®¿´´ ¬®»» º±® ¬¸» ©¸±´» ¾´±½µ ±º ¬¸» ¸±«» ß²¿´§· ¬± ¼»¬»®³·²» ´±½¿´ ¬®»» ¿®» º«®¬¸»® ¼·ª·¼»¼ ·²¬±

¬¸®»» ¼·ºº»®»²¬ ½¿»æ ¿÷ Ý¿» ïæ Ð·´» °»®º±®³¿²½» ¿ °»® °®»¼·½¬·±² ¾÷ Ý¿» îæ Ð·´» °»®º±®³¿²½» · ´±©»® ¬¸¿² °®»¼·½¬·±² ø«²¼»®ó

½¿°¿½·¬§÷ ½÷ Ý¿» íæ Ð·´» °»®º±®³¿²½» · ¾»¬¬»® ¬¸¿² °®»¼·½¬·±² ø±ª»®ó

½¿°¿½·¬§÷ Ì¸» ¬¸®»» ½¿» ¿®» ¬± ½¿¬»® º±® °±·¾´» ª¿®·¿¬·±² ·² ¬¸» «¾ó

±·´ °®±°»®¬·» ¿²¼ °·´» ·²¬¿´´¿¬·±² °®±½»¼«®» ®»«´¬·²¹ ·² ¼·ºº»®ó»²¬ ª¿´«» ±º ®»´¿¬·ª» °·´» ¬·ºº²» ¿²¼ ±·´ ¬·ºº²» ¾»²»¿¬¸ ¬¸» ®¿º¬ò Ì¸» ª¿®·¿¬·±² ±º ¬¸» ¬·ºº²» ©±«´¼ ¿ºº»½¬ ¬¸» ¬®»» ¹»²ó»®¿¬»¼ ·² ¬¸» ¬®«½¬«®¿´ ³»³¾»® ¿²¼ ²»»¼ ¬± ¾» ¬¿µ»² ·²¬± ½±²ó·¼»®¿¬·±²ò Í·³·´¿® ¼»·¹² ¿°°®±¿½¸ «·²¹ »¬¬´»³»²¬ ®»¼«½·²¹ °·´» ¸¿ ¿´± ¾»»² ¿¼±°¬»¼ ¾§ Ô±ª» øîððí÷ò

ß ¸·¹¸´·¹¸¬»¼ ¾§ Ì»®¦¿¹¸· øïçëë÷ ¿²¼ Ð±«´± øîððð÷ô ¬¸» É·²µ´»® §¬»³ ¸¿ ·¬ ´·³·¬¿¬·±² ·² ¬¸¿¬ ·¬ · ±²´§ ¿¾´» ¬± º«®²·¸ ª¿´«» ±º ´±½¿´ ¬®»»ò Ì¸»®»º±®»ô ·² ±®¼»® ¬± ½¿¬»® º±® ±ª»®¿´´ ¬®»» ±² ¬¸» ©¸±´» ¾´±½µ ±º ¬¸» ¸±«»ô ¿¼¼·¬·±²¿´ »¬¬´»³»²¬ ¿²¿´§» ¿®» ½¿®®·»¼ ±«¬ ¬± ¼»¬»®³·²» ¬¸» »¬¬´»³»²¬ °®±º·´» º±® «¾»¯«»²¬ ¼»¬»®³·²¿¬·±² ±º °®·²¹ ¬·ºº²» º±® ¬¸» °·´» ¿²¼ ±·´ò

Ì¸» »¬¬´»³»²¬ ¿²¿´§· · ½¿®®·»¼ ±«¬ ¾¿»¼ ±² Ì»®¦¿¹¸·� ïó¼·³»²·±²¿´ ½±²±´·¼¿¬·±² ¬¸»±®§ ¿²¼ ¬¸» ¬®» ¼·¬®·¾«¬·±² · ¾¿»¼ ±² Þ±«·²»¯� ¬¸»±®§ò Ì¸» ®¿º¬ · ¿«³»¼ ¬± ¾» ¬®«´§ º´»¨·¾´» ©¸·½¸ · ±² ¬¸» ½±²»®ª¿¬·ª» ·¼» º±® ¼»·¹²ò Ì¸» »¬¬´»ó³»²¬ ¿²¿´§· ¿´± ¬¿µ» ·²¬± ½±²·¼»®¿¬·±² ¬¸» »ºº»½¬ ±º ¿¼¶¿½»²¬ ®±© ±º ¸±«» ¿ ¸±©² ·² Ú·¹ò í ¿²¼ ìò

Ú·¹ò íò Ûºº»½¬ ±º ¿¼¶¿½»²¬ ¸±«» ±² »¬¬´»³»²¬ ø¬»®®¿½» ¸±«»÷ò

Ú·¹ò ìò Ûºº»½¬ ±º ¿¼¶¿½»²¬ ¸±«» ±² »¬¬´»³»²¬ ø»³·ó¼»¬¿½¸»¼ ¸±«»÷ò

Ì§°·½¿´ »¬¬´»³»²¬ °®±º·´» ±¾¬¿·²»¼ º®±³ ¬¸» »¬¬´»³»²¬ ¿²¿´§» ¿®» ¸±©² ·² Ú·¹ò ëô ê ¿²¼ é ®»°»½¬·ª»´§ò Ì¸» »¬¬´»ó³»²¬ °®±º·´» ©·´´ «¾»¯«»²¬´§ ¾» «»¼ ¬± ¼»¬»®³·²» ¬¸» °®·²¹ ¬·ºº²» øª¿´«» ±º ´±¿¼ñ»¬¬´»³»²¬÷ º±® ¬¸» °·´» ¿²¼ ±·´ «°°±®¬ ·² ±®¼»® ¬± ·³«´¿¬» ¿ ·³·´¿® »¬¬´»³»²¬ °®±º·´» ¹·ª·²¹ ¬¸» ±ª»®¿´´ ¬®»» ±² ¬¸» ©¸±´» ¾´±½µ ±º ¸±«»ò Ì¸» º·¹«®» ¸±© ¬¸» »ºóº»½¬ ±º ´±¿¼ º®±³ ¬¸» ¿¼¶¿½»²¬ ¾´±½µ ±º ¸±«» ·² ·²½®»¿·²¹ ¬¸» »¬¬´»³»²¬ ¿²¼ ¸»²½» ¼·ºº»®»²¬·¿´ »¬¬´»³»²¬ò Ì¸»®»º±®»ô °¿®¬·½«´¿® ½¿®» ¸±«´¼ ¾» ¹·ª»² ¬± ¬¸» ¼»·¹² ±º ¬¸» º±«²¼¿¬·±² §¬»³ »°»ó½·¿´´§ ¿¬ ¬¸» ½±®²»® ±º ¬¸» ¾´±½µ ¿²¼ ¿¬ ¿®»¿ º¿½·²¹ ¿²±¬¸»® ¿¼¶¿ó½»²¬ ¾´±½µ ±º ¸±«»ô ¿ ¬¸» ¼·ºº»®»²¬·¿´ »¬¬´»³»²¬ ¿²¼ ¸»²½» ¾»²¼·²¹ ³±³»²¬ ¿®» ¬¸» ´¿®¹»¬ ¿¬ ¬¸±» ¿®»¿ò Ì¸» º·²¿´ º±«²¼¿ó¬·±² §¬»³ ¿¼±°¬»¼ ½±²·¬ ±º ïëð ³³ ¨ ïëð ³³ ¨ ç ³ ´»²¹¬¸ ®»·²º±®½»¼ ½±²½®»¬» øÎÝ÷ ¯«¿®» °·´» ·²¬»®½±²²»½¬»¼ ©·¬¸ íëð ³³ ¨ êðð ³³ ¬®·° ¿²¼ ïëð ³³ ¬¸·½µ ®¿º¬ò Ú·¹ò è ¿²¼ ç ¸±© ¬§°·½¿´ ´¿§±«¬ ±º º±«²¼¿¬·±² §¬»³ ¿¼±°¬»¼ º±® ¬»®®¿½»¸±«» ¿²¼ ¬§°·½¿´ ½®±ó»½¬·±² ±º ¬¸» ¬®·°ó®¿º¬ º±«²¼¿¬·±² §ó¬»³ ®»°»½¬·ª»´§ò

Ú·¹ò ëò Í»¬¬´»³»²¬ °®±º·´» º±® ¬»®®¿½» ¸±«»ò

Í¬®» ¾«´¾ º®±³ ¿¼¶¿ó½»²¬ ®±© ±º ¸±«»

Î±© ±º ¸±«» «²¼»® ½±²·¼»®¿¬·±²

Í»½¬·±²¿´ Ê·»©

Ð´¿² Ê·»©

Ý±®²»® «²·¬ ×²¬»®³»¼·¿¬» «²·¬

Í¬®» º·»´¼ º®±³ ¿¼¶¿½»²¬ ¸±«»

Ø±«» «²¼»® ½±²·¼»®¿¬·±²

Ò±¬»æ Ê»®¬·½¿´ ½¿´» »¨¿¹¹»®¿¬»¼

ÞßÝÕ

ÚÎÑÒÌÎ»´¿¬·ª» Í»¬¬´»³»²¬ ø³³÷

Ü·¬¿²½» ø³÷

ïéì

Ú·¹ò êò Í»¬¬´»³»²¬ °®±º·´» º±® »³·ó¼»¬¿½¸»¼ ¸±«» ø½±®²»® «²·¬÷ò

Ú·¹ò éò Í»¬¬´»³»²¬ °®±º·´» º±® »³·ó¼»¬¿½¸»¼ ¸±«» ø·²¬»®³»¼·¿¬» «²·¬÷ò

Ú·¹ò èò Ì§°·½¿´ ´¿§±«¬ ±º º±«²¼¿¬·±² §¬»³ º±® ¬»®®¿½» ¸±«»ò

Ú·¹ò çò Ì§°·½¿´ ½®±ó»½¬·±² ±º ¬®·° ®¿º¬ º±«²¼¿¬·±² §¬»³ò

ë ÍÛÌÌÔÛÓÛÒÌ ÓÑÒ×ÌÑÎ×ÒÙ

ß »¬¬´»³»²¬ ³±²·¬±®·²¹ °®±¹®¿³³» ©¿ ½¿®®·»¼ ±«¬ «·²¹ °®»ó½·» »¬¬´»³»²¬ ³¿®µ»® ·²¬¿´´»¼ ±² ½±´«³² ±º ¬¸» ¬®«½¬«®»ò Ú·¹ò ïð ¸±© ¬¸» ´±½¿¬·±² ±º ¬¸» »¬¬´»³»²¬ ³¿®µ»® º±® ¬¸» ¬©± ¾´±½µ ±º ¬»®®¿½» ¸±«» ©¸»®» ¿ ¬±¬¿´ ±º îî ²±ò ±º »¬¬´»³»²¬ ³¿®µ»® ©»®» ·²¬¿´´»¼ò Ó±²·¬±®·²¹ ©±®µ ©»®» ½¿®®·»¼ ±«¬ ¬¿®¬ó·²¹ º®±³ Ò±ª»³¾»® îððî ©¸»² ¬¸» ¹®±«²¼ º´±±® ½±´«³² ¸¿ª» ¾»»² ½±²¬®«½¬»¼ ¬± Ó¿®½¸ñß°®·´ îððí ©¸»² ·¬ ©¿ ¬±°°»¼ °®»ó³¿¬«®»´§ ¬± º¿½·´·¬¿¬» ·²¬¿´´¿¬·±² ±º ¿®½¸·¬»½¬«®¿´ º·²·¸»ò Ì¸»®»óº±®»ô ¬¸» ³±²·¬±®·²¹ · ±²´§ ½¿®®·»¼ ±«¬ ¼«®·²¹ ¬¸» ½±²¬®«½¬·±² °»®·±¼ ©¸»®» ¿¼¼·¬·±²¿´ ´±¿¼ º®±³ ¬¸» º·®¬ º´±±® ©»®» «¾»ó¯«»²¬´§ ·³°±»¼ò

Ú·¹«®» ïï ¿²¼ ïî ¸±© ¬¸» »¬¬´»³»²¬ ³±²·¬±®·²¹ ®»«´¬ º±® ¬¸» ¬©± ¾´±½µ ±º ¬»®®¿½» ¸±«»ò ×¬ · ©±®¬¸ ²±¬·²¹ ¬¸¿¬ ¬¸» »¬ó¬´»³»²¬ ·²½®»¿» ®¿¬¸»® ¬»»°´§ ©·¬¸ ¬·³»ò Ø±©»ª»®ô ¬¸· · »¨ó°»½¬»¼ ¿ ´±¿¼ º®±³ ¬¸» º·®¬ º´±±® ø¿¼¼·¬·±²¿´ ´±¿¼÷ ©»®» ·³ó°±»¼ ¼«®·²¹ ¬¸» ³±²·¬±®·²¹ °»®·±¼ò Ì¸» ³¿¨·³«³ ®»½±®¼»¼ ¼·ºº»®»²¬·¿´ »¬¬´»³»²¬ ¿®» èòçç ³³ ø¾»¬©»»² ÝÍðé ó ÝÍðç÷ ¿²¼ ïíòîë ³³ øÝÍïé ó ÝÍîï÷ º±® Þ´±½µ î ¿²¼ í ®»°»½¬·ª»´§ò Ì¸» ³¿¨·³«³ ¿²¹«´¿® ¼·¬±®¬·±² ®»½±®¼»¼ ø»²¼ ±º ½±²¬®«½¬·±²÷ ¿®» ïñîèëð ¿²¼ ïñïééë º±® Þ´±½µ î ¿²¼ í ®»°»½¬·ª»´§ò

Ì¸» »¬¬´»³»²¬ ³±²·¬±®·²¹ ®»«´¬ ¿´± ½±²º·®³ ¬¸» »¬¬´»³»²¬ ½¸¿®¿½¬»®·¬·½ ±º ¬¸» ¬®«½¬«®» ©¸»®» ·¬ ½¿² ¾» »»² ¬¸¿¬ ¬¸» »¬ó¬´»³»²¬ ¿¬ ¬¸» ½±®²»® ±º ¬¸» ¬®«½¬«®» ¿®» ¬¸» ³¿´´»¬ ©¸·½¸ · ½¸¿®¿½¬»®·¬·½ º±® ¿ º´»¨·¾´» º±«²¼¿¬·±² ©¸»®» ¬¸» »¬¬´»³»²¬ °®±óº·´» · ±º ¿ ½«®ª»¼ ±® �¾±©´� ¸¿°»ò ×² ¿¼¼·¬·±²ô ¿ ¬¸» ¾¿½µ °±®ó¬·±² ±º ¬¸» ¾´±½µ ±º ¸±«» · ª»®§ ½´±» ø ïð ³÷ ¬± »¿½¸ ±¬¸»® ¿ ½±³°¿®»¼ ¬± ¬¸» º®±²¬ ©¸»®» ¬¸» ¬®«½¬«®» · »°¿®¿¬»¼ ¾§ ¿°ó°®±¨·³¿¬»´§ îð³ ¾§ ¿ ®±¿¼ ¿²¼ ¬¸» º®±²¬ §¿®¼ô ¬¸» »¬¬´»³»²¬ ¿¬ ¬¸» ¾¿½µ · ¹®»¿¬»® ¼«» ¬± ¬¸» ·²º´«»²½» ±º ´±¿¼·²¹ º®±³ ¿¼¶¿½»²¬ ¾´±½µ ±º ¸±«» ¿ ¸±©² ·² Ú·¹ò íò Í«½¸ º·²¼·²¹ ¿¹®»» ©»´´ ©·¬¸ ¬¸» °®»¼·½¬»¼ »¬¬´»³»²¬ ¬®»²¼ ¸±©² ·² Ú·¹ò ëô ©¸·½¸ ¸¿ ¬¿µ»² ·²¬± ½±²·¼»®¿¬·±² ¬¸» »ºº»½¬ ±º ´±¿¼·²¹ º®±³ ¿¼¶¿½»²¬ ¸±«»ò Ì¸· ¼»³±²¬®¿¬» ¬¸» ·³°±®¬¿²½» ±º ¼»¬»®³·²·²¹ ¬¸» ±ª»®¿´´ ¬®»» ±º ¬¸» º±«²¼¿¬·±² ¾¿»¼ ±² ·¬ »¬¬´»³»²¬ °®±º·´» ®¿¬¸»® ¬¸¿² ®»´§·²¹ ±´»´§ ±² ´±½¿´ ¬®»» ±¾¬¿·²»¼ º®±³ ½±²ª»²¬·±²¿´ É·²µ´»® º±«²¼¿¬·±² ¿²¿´§·ò

Ò±¬»æ Ê»®¬·½¿´ ½¿´» »¨¿¹¹»®¿¬»¼

Ò±¬»æ Ê»®¬·½¿´ ½¿´» »¨¿¹¹»®¿¬»¼

Ô»¹»²¼æïëð³³ ¨ ïëð³³ ÎòÝò Í¯«¿®» Ð·´» ±º ç³ ´»²¹¬¸

Ý±´«³² ´±½¿¬·±²

ÞßÝÕ

ÚÎÑÒÌ

Î»´¿¬·ª» Í»¬¬´»³»²¬ ø³³÷

Î»´¿¬·ª» Í»¬¬´»³»²¬ ø³³÷

Ü·¬¿²½» ø³÷

Ü·¬¿²½» ø³÷

Î»·²º±®½»³»²¬ ¬± ¾» ¼»·¹²»¼

ïéë

Ú·¹ò ïðò Ô±½¿¬·±² ±º »¬¬´»³»²¬ ³¿®µ»®ò

Ú·¹ò ïïò Í»¬¬´»³»²¬ ³±²·¬±®·²¹ ®»«´¬ º±® Þ´±½µ îò Ú·¹ò ïîò Í»¬¬´»³»²¬ ³±²·¬±®·²¹ ®»«´¬ º±® Þ´±½µ íò

ïçóÒ±ªóðî îìóÜ»½óðî ïêóÖ¿²óðí ïéóÚ»¾óðí ïèóÓ¿®óðí îîóß°®óðíÜ¿¬»

îë

îð

ïë

ïð

ë

ð

îì

îí

îî

îï

ïç

ïè

ïé

ïê

ïì

ïí

ïî

ïï

ç

è

é

ê

ì

í

î

ï

Í»¬¬´»³»²¬ Ó¿®µ»®ÝÍïîô ÝÍïéô ÝÍïèô ÝÍîî

ÝÍïíô ÝÍïìô ÝÍïëô ÝÍïê

ÝÍïçô ÝÍîðô ÝÍîï

îë

îð

ïë

ïð

ë

ð

îì

îí

îî

îï

ïç

ïè

ïé

ïê

ïì

ïí

ïî

ïï

ç

è

é

ê

ì

í

î

ï

ð íë ëè çð ïïç ïëì

Ü¿§

Ý±®²»®

Ú®±²¬

Þ¿½µ

Ý±®²»®

Ú®±²¬

Þ¿½µ

Ô»¹»²¼æ ÝÍ � Í»¬¬´»³»²¬ ³¿®µ»® øîî Ò±ò÷

ïçóÒ±ªóðî îìóÜ»½óðî ïêóÖ¿²óðí ïéóÚ»¾óðí ïèóÓ¿®óðíÜ¿¬»

îð

ïë

ïð

ë

ð

ïç

ïè

ïé

ïê

ïì

ïí

ïî

ïï

ç

è

é

ê

ì

í

î

ï

Í»¬¬´»³»²¬ Ó¿®µ»®ÝÍðïô ÝÍðéô ÝÍïï

ÝÍðîô ÝÍðíô ÝÍðìô ÝÍðëô ÝÍðê

ÝÍðèô ÝÍðçô ÝÍïð

îð

ïë

ïð

ë

ð

ïç

ïè

ïé

ïê

ïì

ïí

ïî

ïï

ç

è

é

ê

ì

í

î

ï

ð íë ëè çð ïïç

Ü¿§

ïéê

ê ÝÑÒÝÔËÍ×ÑÒÍ

ß ¼»·¹² ¿°°®±¿½¸ º±® �º´±¿¬·²¹� °·´»¼ ®¿º¬ º±«²¼¿¬·±² §¬»³ º±® ¬©±ó¬±®»§ ¬»®®¿½» ¿²¼ »³·ó¼»¬¿½¸»¼ ¸±«» ±² ±º¬ ½´¿§ ¿®» °®»ó»²¬»¼ò Ì¸» º±«²¼¿¬·±² ½±²·¬ ±º ¿ §¬»³ ±º »¬¬´»³»²¬ ®»¼«½ó·²¹ °·´» ·²¬»®½±²²»½¬»¼ ©·¬¸ ¿ ¬®·°ó®¿º¬ º±«²¼¿¬·±² §¬»³ò

Ì¸» «» ±º »¬¬´»³»²¬ ®»¼«½·²¹ °·´» · ±´»´§ ¬± ½±²¬®±´ ¼·ºº»®ó»²¬·¿´ »¬¬´»³»²¬ ©·¬¸·² ¿´´±©¿¾´» ª¿´«» ¬± °®»ª»²¬ ½®¿½µ·²¹ ·² ©¿´´ ¿²¼ °¿®¬·¬·±² ¿²¼ ¿´± ¬± ®»¼«½» ¬®»» ±² ¬¸» ¬®«½¬«®¿´ ³»³¾»®ò ×² ±®¼»® ¬± ¿¼±°¬ ¬¸» ½±²½»°¬ ±º »¬¬´»³»²¬ ®»¼«½·²¹ °·´»ô ¬¸» º±«²¼¿¬·±² ®¿º¬ ³«¬ º·®¬ ¾» ¿¾´» ¬± °®±ª·¼» ¿¼»¯«¿¬» ¾»¿®·²¹ ½¿°¿½·¬§ò

Ì¸» ¼»¬¿·´»¼ ¿²¿´§· ±º ¬¸» º±«²¼¿¬·±² §¬»³ ¬± ¼»¬»®³·²» ¬¸» ¬®»» ¿½¬·²¹ ±² ¬¸» ¬®«½¬«®¿´ ³»³¾»® ®»¯«·®» ½¿®»º«´ ½±²ó·¼»®¿¬·±² ±º ¬¸» º±´´±©·²¹ ½¿»æ

¿÷ Ô±½¿´ ¬®»» ¿¬ ´±½¿¬·±² ±º ½±²½»²¬®¿¬»¼ ´±¿¼ ¾÷ Ñª»®¿´´ ¬®»» º±® ¬¸» ©¸±´» ¾´±½µ ±º ¸±«»ò ß²¿´§· ¬± ¼»¬»®³·²» ´±½¿´ ¬®»» ¿®» º«®¬¸»® ¼·ª·¼»¼ ·²¬±

¬¸®»» ¼·ºº»®»²¬ ½¿»æ ¿÷ Ý¿» ïæ Ð·´» °»®º±®³¿²½» ¿ °»® °®»¼·½¬·±² ¾÷ Ý¿» îæ Ð·´» °»®º±®³¿²½» · ´±©»® ¬¸¿² °®»¼·½¬·±² ø«²¼»®ó

½¿°¿½·¬§÷ ½÷ Ý¿» íæ Ð·´» °»®º±®³¿²½» · ¾»¬¬»® ¬¸¿² °®»¼·½¬·±² ø±ª»®ó

½¿°¿½·¬§÷ ß²¿´§· ¬± ¼»¬»®³·²» ±ª»®¿´´ ¬®»» ·²ª±´ª» ¬¸» ¼»¬»®³·²¿ó

¬·±² ±º »¬¬´»³»²¬ °®±º·´» º±® ¬¸» ©¸±´» ¾´±½µ ±º ¸±«»ò Ì¸· · ¬± º¿½·´·¬¿¬» ¬¸» ¿¼¶«¬³»²¬ ±º °®·²¹ ¬·ºº²» ª¿´«» º±® ¬¸» °·´» ·² ½±²ª»²¬·±²¿´ É·²µ´»® º±«²¼¿¬·±² ¿²¿´§· ¿²¼ ¬± ·³«´¿¬» ·³·´¿® »¬¬´»³»²¬ °®±º·´» ¹·ª·²¹ ¬¸» ±ª»®¿´´ ¬®»»ò ×²º´«»²½» ±º ¿¼¶¿ó½»²¬ ®±© ±º ¸±«» ±² ¬¸» »¬¬´»³»²¬ °®±º·´» · ¿´± ·³°±®¬¿²¬ô ¿ ·¬ ©±«´¼ ¼·®»½¬´§ ¿ºº»½¬ ¬¸» ±ª»®¿´´ ¬®»»ò

Í»¬¬´»³»²¬ ³±²·¬±®·²¹ ®»«´¬ ¸¿ª» ¼»³±²¬®¿¬»¼ ¬¸» »ºº»½ó¬·ª»²» ±º ¬¸» °®±°±»¼ º±«²¼¿¬·±² §¬»³ ·² ½±²¬®±´´·²¹ ¼·ºº»®ó»²¬·¿´ »¬¬´»³»²¬ ¿²¼ ¬¸» ·³°±®¬¿²½» ±º ¬¸» ·²º´«»²½» º®±³ ¿¼¶¿ó½»²¬ ²»¿®¾§ ¸±«» ±² ¬¸» »¬¬´»³»²¬ °®±º·´» ¿²¼ ¸»²½» ¬®»» ±² ¬¸» ¬®«½¬«®¿´ ³»³¾»®ò

ÎÛÚÛÎÛÒÝÛÍ

ß¿±µ¿ô ßò ïçéèò Ñ¾»®ª¿¬·±²¿´ °®±½»¼«®» ±º »¬¬´»³»²¬ °®»¼·½ó¬·±²ò Í±·´ ¿²¼ Ú±«²¼¿¬·±² îèøì÷æ èéóïðïò

Þ®·¬·¸ Í¬¿²¼¿®¼ ×²¬·¬«¬·±²ò ïççêò ÞÍêíççæ Ð¿®¬ ïæ ïççêæ Ý±¼» ±º °®¿½¬·½» º±® ¼»¿¼ ¿²¼ ·³°±»¼ ´±¿¼ò

Þ«®´¿²¼ô ÖòÞò ú Õ¿´®¿ô ÖòÝò ïçèêò Ï«»»² Û´·¦¿¾»¬¸ ×× Ý±²º»®»²½» Ý»²¬®»æ ¹»±¬»½¸²·½¿´ ¿°»½¬ò Ð®±½»»¼·²¹ ×²¬·¬«¬·±² Ý·ª·´ Û²¹·²»»® ïøèð÷æ ïìéçóïëðíò

Ù«»ô ÍòÍò ú Ì¿²ô ÇòÝò îððìò Ð®»ª»²¬·±² ±º º¿·´«®» ®»´¿¬»¼ ¬± ¹»±¬»½¸²·½¿´ ©±®µ ±² ±º¬ ¹®±«²¼ øÍ°»½·¿´ Ô»½¬«®»÷ò Ð®±ó½»»¼·²¹ Ó¿´¿§·¿² Ù»±¬»½¸²·½¿´ Ý±²º»®»²½»æ ëçóéí Õ«¿´¿ Ô«³°«®ò

Ô±ª»ô ÖòÐò îððíò Ë» ±º »¬¬´»³»²¬ ®»¼«½·²¹ °·´» ¬± «°°±®¬ ¿ ®¿º¬ ¬®«½¬«®»ò Ð®±½»»¼·²¹ ×²¬·¬«¬·±² Ý·ª·´ Û²¹·²»»® ïëêøì÷æ ïééóïèïò

Ð±«´±ô ØòÙò îðððò Ú±«²¼¿¬·±² »¬¬´»³»²¬ ¿²¿´§· � °®¿½¬·½» ª»®« ®»»¿®½¸ò Ì¸» Û·¹¸¬ Í°»²½»® Öò Þ«½¸¿²¿² Ô»½¬«®» ôÌ»¨¿ô ËòÍòßò

Ð±«´±ô ØòÙò îððïò Ð·´»¼ ®¿º¬ º±«²¼¿¬·±²æ ¼»·¹² ¿²¼ ¿°°´·½¿ó¬·±²ò Ù»±¬»½¸²·¯«» ëïøî÷æ çëóïïíò

Î¿²¼±´°¸ô Óò Úò ïççìò Ü»·¹² ³»¬¸±¼ º±® °·´» ¹®±«° ¿²¼ °·´»¼ ®¿º¬ò Ð®±½»»¼·²¹ ïí¬¸ ×ÝÍÓÚÛæ êïóèîô Ò»© Ü»´¸·ô ×²¼·¿ò

Íµ»³°¬±²ô ßòÉò ú Ó¿½Ü±²¿´¼ô ÜòØò ïçëêò Ì¸» ¿´´±©¿¾´» »¬ó¬´»³»²¬ ±º ¾«·´¼·²¹ò Ð®±½»»¼·²¹ ×²¬·¬«¬·±² Ý·ª·´ Û²¹·²»»® íøë÷æ éîéóéèìò

Ì¿²ô ÇòÝòô Ù«»ô ÍòÍòô Ò¹ô ØòÞò ú Ô»»ô ÐòÌò îððìò Í±³» ¹»±¬»½¸ó²·½¿´ °®±°»®¬·» ±º Õ´¿²¹ Ý´¿§ò Ð®±½»»¼·²¹ Ó¿´¿§·¿² Ù»±ó¬»½¸²·½¿´ Ý±²º»®»²½»æ ïéçóïèëô Õ«¿´¿ Ô«³°«®ò

Ì»®¦¿¹¸·ô Õò ïçëëò Ûª¿´«¿¬·±² ±º ½±»ºº·½·»²¬ ±º «¾¹®¿¼» ®»¿½ó¬·±²ò Ù»±¬»½¸²·¯«» ëøì÷æ îçéóíîêò

Ç¿³¿¸·¬¿ô Õòô Õ¿µ«®¿·ô Óò ú Ç¿³¿¼¿ô Ìò ïççìò ×²ª»¬·¹¿¬·±² ±º ¿ °·´»¼ ®¿º¬ º±«²¼¿¬·±² ±² ¬·ºº ½´¿§ò Ð®±½»»¼·²¹ ïí¬¸

×ÝÍÓÚÛæ ëìíóëìêô Ò»© Ü»´¸·ô ×²¼·¿ò


177

Earth Pressures of Anisotropic Soft Clay on Rigid Rough Retaining Walls at Failure

A. El Nahas Division of Civil Engineering, University of Dundee, Dundee DD1 4HN, UK [email protected]

J. Takemura Department of Civil Engineering, Tokyo Institute of Technology, Meguro-Ku, Tokyo 152-8552, Japan. [email protected]

Abstract: Upper bound solutions for active and passive earth pressures on rough self-supported retaining wall in anisotropic soft clay were developed, and tested together with the design earth pressures of the current design method in Japan and BS8002 against centri-fuge test results. The upper bounds matched the measured earth pressures at failure assuming full mobilization of undrained triaxial compression soil strength behind the wall, partial mobilization of the triaxial extension soil strength in front of the wall, and ignoring the soil adhesion on the wall front. The current design method overestimated the active and passive earth pressures as it ignores the soil strength anisotropy and soil adhesion on the wall surfaces, and uses the unconfined compressive soil strength . The BS8002 overesti-mated the passive earth pressures, as it overestimated the soil adhesion on the wall front and the mobilized soil strength in front of the wall. Soil adhesion factors were calculated and presented to be used, instead of the approximate factors, to calculate the design active earth pressures.

1 INTRODUCTION

The current design method in Japan of the DM on self-supported walls, supporting open excavations in anisotropic normally con-solidated (NC) clay, uses Rankine�s theory to estimate the design earth pressures on the wall (Shiomi et al., 1996). The undrained soil strength (cu) is estimated by conducting Unconfined Com-pression Tests on undisturbed field soil samples, and is consid-ered fully mobilized everywhere in the soil mass. In order to as-sess the validity of the method for a safe and economic design verification of its design earth pressures is necessary, which in turn requires a theoretical approach to predict the mobilized soil shear strengths on the wall surfaces at failure, considering the soil characteristics (e.g. strength anisotropy & profile), and boundary conditions. This work is yet to be done. However, the BS8002 empirical design earth pressure relationships conservatively con-sider the soil adhesion (cw) on the wall surfaces (0 cw/cu 0.75), and allows using the Triaxial extension test to estimate the undrained soil strength in front of the wall.

In this study, the external wall stability and efficiency of the current design method are being investigated using upper bound calculations and results of centrifuge model tests. El Nahas & Takemura (2002a,b) proved that the main design criterion in the design of self-supported type DMM wall floating in clay, is the wall stability against sliding failure in the undrained condition. This paper is concerned with upper bound solutions to predict the earth pressures on the wall at failure, considering the soil strength anisotropy and soil adhesion on the wall surfaces. The solutions are examined by comparing them, together with the design earth pressures of the current design method in Japan, and BS8002, against the centrifuge model test results. The mobilized soil shear strengths and adhesion at failure on both wall sides are also discussed using total stress analysis.

2 UPPER BOUND ANALYSIS

2.1 Failure Mechanisms

Four compatible failure Mechanisms (Figs. 1- 4) were used to es-timate upper bounds for the active and passive earth pressure forces on the wall (Na & Np). Mechanism A is the Coulomb�s mechanism in the undrained condition ( u=0 & 1,2=45°). Mechanism B, on each wall side, consisted of a slip fan besides the wall, and two triangular failure blocks, and is fully described by three angles ( 1,2,3 & 4,5,6). Mechanism C consisted of a trian-gular failure block, a slip fan, and a Rankine failure block, on each wall side, and is fully described by one angle ( 1,2). Mecha-nism D, on each wall side, consisted of two triangular blocks, and is fully described by three angles ( 1,2, 1,2& 1,2)

El Nahas & Takemura (2002a & 2004) specified the parame-ters that affect the external wall stability against sliding type fail-ure as shown in Fig.5. These are the wall embedded height into clay, Hclay, & breadth, B, soil adhesion on the wall surfaces, sur-charge on the clay surface, q, excavated height into clay at fail-ure, zcf, clay bulk unit weight, , coefficient of soil strength ani-sotropy, m, clay compressive strength at its surface c0, and gradient of the clay strength increase with depth, k. From these parameters, seven non-dimensional parameters were derived: m, cw/cu, B/c0, Hu= Hclay/c0, Q = q/c0, k/ & Zu= zcf/c0. Using six parameters from them excluding B/Hclay, the active & passive earth pressure forces in nondimensional form are given. Eqs. 1 & 2 are these non-dimensional form, in which another non-dimensional parameter Du (= D/c0) is used instead of Zu. Where D is the embedded wall height into clay under the excavation bot-tom (D=Hclay-zcf). WAct & WPas in the equations are the sums of the rates of internal energy dissipation on the failure planes behind and in front of the wall, respectively, and wwall is the wall velocity. WAct & WPas were functions in the defining angles for the failure mechanism, k/ , Hu, cw/cu, m & Du,. The soil shear

178

wallclayActuclaya wHcWHQHcN 00 2 (1)

wallclayPasuuclayp wHcWHDHcN 02

0 2 (2)

Table 1. Sums of the rates of energy dissipation on the failure planes in the active side, for Mechanisms A, B, C & D.

Adhesion ratio=1.0c Adhesion ratio=0.0d

k/ =0.0 k/ =0.1 k/ =0.0 k/ =0.1 m 0.5e 1.0 f 0.5 1.0 1.0 1.0 WAct_MechA

a 2.75 3.0 11.14 12.15 2.0 8.1 WAct_MechB

a 2.43 2.57 10.42 11.20 2.57 11.20 WAct_MechC

a 2.43 2.57 10.24 10.95 2.0 8.1 WAct_MechD

a 2.46 2.61 10.27 10.99 2.0 8.1

Table 2. Sums of the rates of energy dissipation on the failure planes in the passive side, for Mechanisms A, B, C & D. Adhesion ratio=1.0 Adhesion ratio=0.0

k/ =0.0 k/ =0.1 k/ =0.0 k/ =0.1mM 0.5 1.0 0.5 1.0 1.0 1.0 WPas_MechA

b 1.75 3.0 7.09 12.15 2.0 8.1 WPas_MechB

b 1.43 2.57 6.37 11.2 2.57 11.2 WPas_MechC

b 1.43 2.57 6.19 10.95 2.0 8.1 WPas_MechD

b 1.46 2.61 6.21 10.99 2.0 8.1 a , b WAct/c0Hclay wwall & WPas/c0Hclay wwall). c , d cw/cu=1 & 0 e , f anisotropic and isotropic soils.

strength on each failure plane was estimated using Casagrande & Carillo�s (1944) formula (Eq. (3)). Where cuv, cuh & cu are the undrained soil shear strengths when the major principal stresses are vertical, horizontal & inclined with an angle to the vertical.Details of Eqs 1 &2 are given in El Nahas & Takemura (2004).

2cosuhuvuhu cccc (3)

2.2 Comparison of Failure Mechanism Solutions

Tables 1 & 2 show the normalized sums of the rates of internal energy dissipation on the failure planes behind and in front of the wall, estimated with the four mechanisms on a 6.1 m high rigid wall, embedded in a clay layer ( =15.0 kN/m3 & c0=1.5kN/m3)with a horizontal surface and without any surcharge on its sur-face, or excavation (zcf=0). The non-dimensional active and pas-sive earth pressure forces could be estimated by substituting with the sums of energy dissipation on the failure planes, shown in the tables, together with the rates of work done by the soil self weights (Hu/2 = Du

2/2Hu =30.5) in Eqs. 1 & 2. The analysis was conducted for isotropic and anisotropic clay

soil (m = 1.0 & 0.5), with and without increasing strength pro-files (k/ = 0.1 & 0) and soil adhesion on the vertical soil-wall in-terfaces (cw/cu = 1.0 & 0). Mechanism C yielded the lowest sum of the rates of energy dissipation on the failure surfaces on each wall side.

By using Mechanism C and assuming the soil to be isotropic with a uniform strength profile (k/ =0) and the soil adhesion on the wall surfaces to be fully mobilized, the rate of energy dissipa-tion was found to be (2+ )/2 and the mechanism was reduced to a slip fan and a Rankine failure block outside it on each wall side ( 1,2=0, and 1,2=45 ). In this case, the mechanism became iden-tical to Terzaghi�s mechanism (1996) for rough rigid wall in undrained condition ( u=0), and its solution was identical to the exact solution. However, when the soil had a strength increasing with depth (k/ >0), the lowest upper bounds were achieved with the blocks B and C between the slip fans and the wall ( 1,2>0, and 1,2<45 ). Hence, for excavations in typical NC clay, for which the strength increases with depth, Mechanism C led to bet-ter solutions than Terzaghi�s mechanism. Therefore, the Mecha-nism upper bounds were considered the closest to the exact solu-tions, and used in the analytical study.

3 CENTRIFUGE MODELING TESTS

3.1 Test Conditions and Used Soils

El Nahas & Takemura (2002a) conducted six Centrifuge mod-eling tests in Tokyo Institute of Technology. The details and specifications of the test systems, and setup, as well as the used soils are mentioned by Takemura et al. (1999), and El Nahas & Takemura (2002a). The test setup is illustrated in Fig. 6. The

B

q

cw

cw

cw

c0

k

clay strength profile

1

compucextuc

m

B

q

cw

cw

cw

c0

k

clay strength profile

1

compucextuc

m

Fig. 5. Parameters affect ting the external wall stability.

Fig. 3. Failure mechanism C. Sb

wall

2

1A

B C

O

zcf

21

45°

Slip Fan 2Slip Fan 1

NaNp

D

q

45°

Sb

wall

2

1A

B C

O

zcf

21

45°

Slip Fan 2Slip Fan 1

NaNp

D

q

45°

Sb

wallq

2

2

1

1

1 2

ACB

D

O1 2

zcf

NaNp

Sb

wallq

2

2

1

1

1 2

ACB

D

O1 2

zcf

NaNp

Fig. 1. Failure mechanism A.

2=45°

q

1= 45°

AC

O

zcf

NaNp

Sb

wall

2=45°

q

1= 45°

AC

O

zcf

NaNp

Sb

wall

Fig. 2. Failure mechanism B. Sb

wall

q6

1

1

2

A

B C

D

Slip Fan 1 Slip Fan 2

O

Na

zcf

Np

Sb

wall

q6

1

1

2

A

B C

D

Slip Fan 1 Slip Fan 2

O

Na

zcf

Np

Fig. 4. Failure mechanism D.

179

model ground consisted of upper loose sand layer ( = 14.2 kN/m3 & 14 mm thick) and lower dense sand ( = 19.6 kN/m3 & 35 mm thick) and NC Kaolin clay ( = 15.5 kN/m3 2.5% & 153 mm thick) between them. Speswhite Kaolin clay and Toyoura sand were used to form the model ground. The clay has liquid and plastic limits of 77.5% & 30.3%, and normalized undrained shear strengths from Triaxial CK0UC and CK0UE tests, cu-

comp/ vc & cuext/ vc, of 0.24 & 0.144, respectively. The undrained soil strength was estimated from the soil stress history, in the plain strain condition. The undrained soil shear strengths in the plain strain compression and extension tests (PSC & PSE) are 1.087 and 1.22 times those estimated from the CK 0UC & CK0UETriaxial tests, respectively (Ladd et al., 1977). No account was taken of differences in soil properties between a model and a pro-totype.

Table 3. Test conditions and excavated height at failure. Test Surf *a

Cond. V. Suprt Cond. *b.

H*c

(mm) B*d

(mm) (ze)f

*e

(mm) SSW-0 smooth floating 123 (8.6) *f 82 (5.7)*e 32 (2.2)*e

SSW-3 rough floating 123 (8.6) 82 (5.7) 43 (3.0) SSW-4 rough floating 163(11.4) 82 (5.7) 64 (4.5) SSW-5 rough floating 124(8.7) 123 (8.6) 67 (4.7) SSW-6 rough resting 165 (11.6) 82 (5.7) 100 (7.0) SSW-7 rough floating 122 (8.5) 61 (4.3) 50 (3.5) a & b wall surface and vertical support conditions.

c, d & e wall height and breadth & failure excavation height. f dimensions in prototype scale (m).

Table 4. Model wall displacements and tiltings before failure.

SSW-3 SSW-4 SSW-5 SSW-6 SSW-7ze

a (m) 2.1 2.8 2.2 3.5 2.1 wall at ze (m) 0.007 0.022 0.0 0.010 0.007 wall at ze (°) 0.06 -0.003 0.0 -0.02 0.03

(ze)fb (m) 2.9 4.3 4.3 7.0 2.8

wall at (ze)f (m) 0.004 0.023 0.011 0.035 0.016 wall at (ze)f (°) 0.18 0.35 0.0 0.47 0.12

b,a excavation depths just before failure, and earlier during excavation, re-spectively. Dimensions are in prototype scale

The model wall consisted of Aluminum and Perspex plates,instrumented with pressure cells on its back, front, and base sur-faces. The wall height (H) and breadth (B) could be varied by adding Perspex horizontal and vertical plates to the wall.

Table 3 shows the test conditions. The wall bulk density was 16.4 kN/m3 in SSW-0, 3, 4 & 5, and 17.6 & 19 kN/m3 in SSW-6 & 7. Except for SSW-0, the wall sides and bottom were covered by sandpaper sheets (No. 100) to create a rough surface condi-tion. The wall was used in test SSW-0 with its polished surfaces not covered by sand paper sheets, in order to create a smooth sur-face condition. In SSW-6, the wall rested on the bottom dense sand layer. In the other tests, the wall floated in the clay layer, as shown in Fig. 6. The test parameters were the wall dimensions, surface roughness, and vertical support condition at its base.

3.2 Model Preparation and Test Procedures

The submerged dense sand layer was made by compaction. Then, de-aired clay slurry with a water content of 1.5 times the liquid limit, was poured into the container, and stepwise consolidated on the Lab floor until a pressure level of 15kPa. Thereafter, seven pore pressure transducers were inserted in the clay and bottom sand layers, and the clay layer was subjected to centrifugal con-solidation at 70g to form a normally consolidated clay layer with strength increasing with depth under brass rods surcharge of 15

kPa. Thereafter, the front window of the strong box was de-tached, the model wall was placed in its location, and the box was closed back. Then the clay was submerged, and the sand was laid inside the water on the clay surface to form the upper loose sand layer. After installing LVDTs, laser displacement transduc-ers (LDTs) and the excavator as shown in Fig. 6, the test setup was mounted on the centrifuge and then the model ground was re-consolidated under 70g until excess pore pressure dissipated. During the reconsolidation, ground water table was kept at the level of the clay surface. At the end of re-consolidation, excava-tion was conducted step-by-step using the in-flight excavator, un-til clear failure occurred. In each step, about 10mm thickness of the soil (0.7m in the prototype scale) was cut from the front of the wall every five minutes. The water level in the excavated area was lowered by draining water from the box to the drainage tank, so that it could be kept at the same level of the excavation bot-tom. However, in SSW-6, the maximum excavation depth (100 mm) was achieved without observing any failure. So, water was poured from the water supply line on the upper loose sand layer behind the wall till failure took place. El Nahas & Takemura (2002a) described in details the model preparations and test pro-cedures. Wall and ground displacements, pore water pressures as well as earth pressures on the wall surfaces were measured during excavation. Test results are given in prototype scale in the fol-lowing section.

4 TEST RESULTS AND DISCUSSIONS

4.1 Wall Movements and Ground Deformations

El Nahas & Takemura(2002a,b) showed that excavations with floating type walls with H/B 2 in SSW-0, 3, 4, 5 & 7 had a sudden slip failure without marked prefailure soil and wall dis-placements; the excavation supported by a wall resting on a rigid strong base in SSW-6 failed due to overturning mechanism with-out marked lateral wall base displacements; and the prefailure wall displacements and tiltings didn�t exceed 0.035 m & 0.5°, re-spectively (Table 4). Figure 7 shows the ground and wall dis-placements at failure in SSW-0 & 4. On the active side, clear failure planes bound the failed zones with inclinations to the horizontal of 59, 51.5, 50, 64, 49.5& 53 in tests SSW-0, 3, 4, 5, 6 & 7, respectively. Outside these failure planes and below the wall no marked displacements were observed even after failure. On the passive side, the soil deformed laterally with some vertical displacements without showing any clear failure planes.

Fig. 6. Centrifuge tests setup.

in-flight excavator

160 400-BB140700

wall

LVDTsLDTs

soil retaining gate

upper sand layer

kaolin clay

lower sand layer

acrylic plates

gate motor

(1-7 ) pore pressure transducers No. (1) to No. (7)

tank

solenoid valve

unit: mm.

surface markersdumped soil

(1)(4)

(6)(5)

(2)(3)

(7)

in-flight excavator

160 400-BB140700

wall

LVDTsLDTs

soil retaining gate

upper sand layer

kaolin clay

lower sand layer

acrylic plates

gate motor

(1-7 ) pore pressure transducers No. (1) to No. (7)(1-7 ) pore pressure transducers No. (1) to No. (7)

tank

solenoid valve

unit: mm.

surface markersdumped soil

(1)(4)

(6)(5)

(2)(3)

(7)

180

Table 5. Measured and calculated active earth pressure forces. SSW-3 SSW-4h SSW-5 SSW-6h SSW-7

(Na)Measureda (kN/m) 386.7 619.9 364.8 641.3 395.4

(Na)Mech_C_PSRb(kN/m) 373.0 621.6 397.4 642.6 363.6

(Na)Mech_C_PSSc(kN/m) 412.0 679.0 438.4 679.2 402.2

(Na)Mech_C_TARd (kN/m) 388.7 644.3 413.8 667.3 379.0

(Na)Mech_C_TASe(kN/m) 423.0 694.9 449.8 722.1 413.0

(Na)Shiomi et alf (kN/m) 448.1 731.3 476.1 761.5 437.9

(Na) BS8002g (kN/m) 382.6 636.1 407.4 656.8 373.0

Table 6. Measured and calculated passive earth pressure forces. SSW-3 SSW-4 SSW-5 SSW-6 SSW-7

(Np)Measureda (kN/m) 302.3 464.4 213.5 219.2 308.6

(Np)Mech_C_PSRb(kN/m) 347.7 565.2 253.1 300.4 361.2

(Np)Mech_C_PSSc(kN/m) 316.1 513.7 225.7 261.9 329.0

(Np)Mech_C_TASe(kN/m) 308.5 490.8 213.2 243.9 314.9

(Np)Shiomi et alf (kN/m) 329.9 525.5 232.2 271.2 336.5

(Np) BS8002g (kN/m) 328.9 524.5 231.6 256.0 335.8

a calculated from the measured earth pressures on the model wall.b,c calculated using Mechanism C with the soil strength in the plain strain condition and cw/cu= 1.0 & 0, respectively. d,e calculated using Mechanism C with the soil strength based on the Triaxial test re-sults (cu = cucomp & cuext, respectively), and cw/cu=0.f,g calculated using the current design method (cw/cu=0 & cu/cucomp 0.8) & BS8002

(cw/cu= 0.75 & m= 0.6), respectively. h calculated over the wall height that is installed with pressure cells (7.7 m).

4.2 Earth Pressures and Mobilized Soil Strength and Adhesion

Figures 8-12 present profiles of the measured and calculated earth pressures using the upper bound failure mechanism on both wall vertical surfaces, just before failure, together with the meas-ured earth pressures on the wall back after earlier excavation steps. The wall base displacements and tiltings after these excava-tion steps are shown in Table 4. Tables 5 and 6 show the horizon-tal active and passive earth pressure forces on the wall that were: estimated from the measured earth pressures just before failure, (Na,p)Measured, and calculated using the upper bound method, (Na,p)Mech C, the current design method, (N a,p)Shiomi et al, and BS8002 (Na,p)BS8002. The upper bound solutions were estimated using soil strength profiles in the plain strain condition (PS), and also based on the Triaxial test results (TA), with and without considering the soil adhesion on the vertical wall surfaces. For the earth pressures of the current design method, the soil strength everywhere in the soil mass was considered identical to 80% of its peak level in the CK0UC Triaxial tests, so as to consider the strength reduction due to the soil sample disturbance and the use of the unconfined compression test rather than undrained Triaxial compression test. The BS8002 empirical relationships of the de-sign active ( an) and passive ( pn) earth pressures on the wall at depth (z) from the upper clay surface are:

3.5 7.0 10.514.017.521.024.528.00-1.6

3.4

8.4

13.4

18.4

-10 -5 0 5 10 15 20 25 30

(m)

(m)

14.0

10.5

7.0

3.5

0.0

SSW-4 (ze= 5.6 m)

(m)3.50 7.0 28.021.024.517.514.010.5

(m)

SSW-0 ( ze= 2.8 m)

-6-1

4

9

14

-10 -5 0 5 10 15 20 25 30

3.5

7.0

14.0

10.5

0.0

3.5 7.0 10.514.017.521.024.528.00-1.6

3.4

8.4

13.4

18.4

-10 -5 0 5 10 15 20 25 30

(m)

(m)

14.0

10.5

7.0

3.5

0.0

SSW-4 (ze= 5.6 m)

3.5 7.0 10.514.017.521.024.528.00 3.5 7.0 10.514.017.521.024.528.00-1.6

3.4

8.4

13.4

18.4

-10 -5 0 5 10 15 20 25 30

(m)

(m)

14.0

10.5

7.0

3.5

0.0

-1.6

3.4

8.4

13.4

18.4

-10 -5 0 5 10 15 20 25 30

(m)

(m)

14.0

10.5

7.0

3.5

0.0

SSW-4 (ze= 5.6 m)

(m)3.50 7.0 28.021.024.517.514.010.5

(m)

SSW-0 ( ze= 2.8 m)

-6-1

4

9

14

-10 -5 0 5 10 15 20 25 30

3.5

7.0

14.0

10.5

0.0

(m)3.50 7.0 28.021.024.517.514.010.5

(m)

SSW-0 ( ze= 2.8 m)

-6-1

4

9

14

-10 -5 0 5 10 15 20 25 30

3.5

7.0

14.0

10.5

0.0

0 40 80 120

back side

SSW-3

earth pressure (kPa)

ze=2.1 m ze=2.9 m Mech C (cw/cu=1.0)Mech C (cw/cu=0.0)

04080120

0

1

2

34

5

6

78

front side


Fig. 8. Earth pressures on both wall sides before failure in SSW-3.


0 40 80 120

back sideSSW-4



04080120

3

4

5

6

7

8

910

11

front side


0 40 80 120earth pressure (kPa)

SSW-5

back side


04080120

01

2

34

56

78


front side



SSW-6 back side


04080120

345

678

910

11


front side



SSW-7

back side


04080120

0

1

2

3

4

5

6

7

8


front side

Fig. 7. Observed ground displacements at failure in SSW-0 &


181

uzcacKzan2 (4)

uzcpcKzpn2 (5)

where: cucwpcKacK 1 , and z & cuz are the total vertical

stress & undrained clay shear strength at depth (z) & cw/cu is the design soil adhesion ratio, and was considered in this study to be 0.75. The soil strength (cuz) behind and in front of the wall was considered identical to its peak and maximum levels in the Triax-ial CK0UC and CK0UE, respectively.

The measured active earth pressure forces in tests SSW-3, 4 & 6 matched their upper bounds, assuming: full mobilization of the soil peak shear strength in the Triaxial test behind the wall in SSW-3, and in the plain strain condition in tests SSW-4 & 6, as well as full mobilization of the soil adhesion on the wall back in the three tests. In SSW-7, the measured active earth pressure force was more than its upper bound assuming full mobilization of the Triaxial compressive soil strength behind the wall and of the soil adhesion on the wall back. In SSW-5, however, the measured active earth pressure force was less than its upper bounds. This probably resulted from having initial stress profile on the wall back before excavation in SSW-5 that was less than the calculated at-rest stress distribution (El Nahas & Takemura, 2002a). Hence, increasing the wall height in SSW-4 & 6, in-creased the mobilized soil strength behind the wall, as the prefailure soil displacements became more ductile (El Nahas & Takemura, 2002a). The measured passive earth pressures on the wall front in SSW-3, 4, 6 & 7 were less than their upper bounds. However, assuming the mobilized soil strengths in front of the wall in these tests to be 0.91, 0.75, 0.34 & 0.91 of the maximum soil shear strength in the CK0UE Triaxial test and ignoring the soil adhesion on the wall front, led to upper bounds that were identical to the measured passive earth pressure forces on the wall, before failure. The small wall lateral displacements and tilt-ings at early excavation stages before failure (Table 4) were suffi-cient for the retained soil strength to reach its peak level in the CK0UC Triaxial test. Hence, the BS8002�s mobilization factor (M 1.5) to assess the mobilized undrained soil shear strength behind the wall when the wall displacement, in service, is to be restricted to 0.5% of its height, is a conservative design assump-tion. The prefailure soil displacements were sufficient for the ac-tive earth pressure distributions on the walls that failed by sliding mechanism, to become close to the linear distribution, proposed by Terzaghi et al. (1996). However, when the failure mode be-came overturning in SSW-6, the prefailure wall tilting (0.47 )and retained soil displacements were not sufficient to form linear stress distribution on the wall back. Further, the lowest mobilized strength in front of the wall was in SSW-6, with an overturning failure mechanism and the maximum prefailure wall tilting (Ta-ble 4). Hence, it can be implied that the soil in front of the rotat-ing wall around its tip required more displacements to mobilize its passive earth pressure than the soil in front of sliding walls. These experimental observations agreed with the numerical analysis prediction by Potts (1991). The mobilized earth pres-sures on the wall in all tests were close to linear distribution, al-though the maximum soil strength was not fully mobilized.

The BS8002 predicted reasonably the active earth pressure forces on the wall before failure, as for the used kaolin NC clay, the BS8002 underestimation of the soil adhesion ratio canceled its ignorance of the influence of the soil strength anisotropy. However, the BS8002 overestimated the mobilized earth pressure on the wall front before failure, as it overestimated the mobilized soil adhesion on the wall front and soil strength in front of the wall. The current design method overestimated both of the active

and passive earth pressure forces, as it ignored the soil strength anisotropy and soil adhesion on the wall vertical surfaces. Using the soil unconfined compressive strength did not improve the method�s estimate of the active earth pressures on the wall, than Rankine�s solution (Na)Mech_C_TAS. In this method, the design critical wall cross section is assumed to be at the level of excava-tion bottom. Hence, when compared with the upper bound solu-tion assuming full mobilization of the soil adhesion and Triaxial compressive soil shear strength behind the wall, this method overestimated both of the design bending moment and shear force on the wall in SSW-3, 4, 5, 6 & 7 by 15.3, 13.5, 15.1, 14.1 & 15.5%, respectively, as it overestimated the active earth pres-sure forces with these percentages, besides overestimating the maximum wall bearing stresses on the foundation soil, as shown by El Nahas & Takemura (2002a).

The slide resisting forces due to the soil adhesion on the model wall base were backcalculated from the measured earth pressure forces (measured base slide resisting force), (Sb)Measured,and presented in Table 7 with the calculated slide resisting forces for the tests SSW-3, 5, and 7, in which the earth pressures were measured along the whole embedded wall height into clay, and the underestimation of the base slide resisting forces, Sb, by the calculated forces. Assuming full mobilization of the soil adhesion on the wall base, the calculated base slide resisting forces, using the undrained Triaxial (TA) soil shear strength and the undrained soil strength in the plain strain (PS) condition, were estimated from the relationship: (Sb)TA,PS=0.5 (cw/cu) { wall Hwall- water

Hclay} (cu/ z) TA,PS (1+m) B; where wall, water & Hwall are the unit weight of the wall and water & the wall height, respec-tively. Using a soil strength profile that was based on Triaxial test results underestimated the wall base slide resisting forces. How-ever, using the undrained soil shear strength profiles in the plain strain condition led to reasonable agreement between the calcu-lated and measured wall base slide resisting forces in SSW-3 and 7. In SSW-5, the measured base slide resisting force was re-markably higher than the calculated forces, even with the undrained soil strength in the plain strain condition. If the active earth pressure force were within the range of the similar meas-ured forces in SSW-3 and 7 with the same wall height ( 391.1kN/m2), the minimum wall base slide resisting force to keep a safety factor against sliding failure of unity should have been 177.5 kN/m2, which is much more than the measured force and the maximum possible force (Sb)PS, assuming full mobilization of the soil adhesion on the wall base. Hence, the excavation in SSW-5 continued till an excavation depth (z e) of 4.3 m, because the active earth pressure force on the wall was exceptionally small. This relatively deep excavation caused marked prefailure soil displacements (El Nahas & Takemura, 2002a). Subsequently, the earth pressure on the wall front before failure, reached Rankine�s solution with the soil strength identical to its maxi-mum level in the CK0UE Triaxial test, probably because the mo-bilized more soil strength in front of the wall at failure was more than the mobilized strengths in the other tests (Table 6). Hence, the designer of the floating type wall may estimate the earth pres-sures on the wall when the excavation reach the ultimate limit state using: 1. The peak CK0UC Triaxial soil shear strength everywhere behind the wall, and assuming full mobilization of the soil adhe-sion on the wall back. 2. A mobilization factor of about 1.1 1.4 to estimate the mobi-lized soil shear strength in front of the wall, from the maximum CK0UE Triaxial soil strength, without considering any soil adhe-sion on the wall front.

182

Table 7. Measured and calculated base slide resistance forces.

SSW-3 SSW-5 SSW-7 (Sb)Measured

a (kN/m) 84.5 151.3 86.8 (Sb) TA

b (kN/m) 73. 3 103.4 71.80 (Sb)PS

c (kN/m) 83.3 117.6 81.7 ( Sb) TA

d (%) 13.25 8.69 31.64 ( Sb) PS (%) 1.40 31.64 22.30 a (Na)Measured-(Np)Measured. d Sb (%)={1-[Sb/( Sb)Measured]} 100b,c calculated assuming cw/cu=1.0 and using undrained Triaxial soil shear strengths, and undrained soil shear strengths in the plain strain condition.

Table 8. Soil adhesion factors. k/ =0.076 k/ =0.14

m = 0.5 m = 1.0 m = 0.5 m = 1.0 Cnd 1b Cnd 2c Cnd 1 Cnd 2 Cnd 1 Cnd 2 Cnd 1 Cnd 2 Kac Kac Kac Kac Kac Kac Kac Kac

Hua = 60 1.157 1.262 1.210 1.349 1.158 1.266 1.211 1.355

Hu =300 1.159 1.271 1.213 1.361 1.160 1.272 1.213 1.363 averaged 1.158 1.266 1.211 1.355 1.159 1.269 1.212 1.359 a,b,c Hclay/c0, cw/cu = 0.5& 1.0, respectively. d Average (Kac)values for Hu=60~300

3. The maximum soil shear strength in the plain strain condi-tion to estimate the wall base slide resisting force, assuming full mobilization of the soil adhesion on the wall base. Using these assumptions, the passive earth pressure forces on the wall in SSW-3 & 7 were 302.0 & 308.4 kN/m, and the safety fac-tors against sliding failure were 0.99 & 1.029, respectively, which implies that these excavations were on the verge of sliding failure.

4.3 Soil Adhesion Factor in the Active Side

The upper bound analysis results were used to define Eq. 4 �s soil adhesion factor (Kac), in order to use that equation to estimate the design active earth pressures on the wall, assuming the undrained soil strength (cuz) to be identical to its peak level in the CK0UC Triaxial test (cucomp). Table 8 presents the soil adhesion factor (Kac), which is defined as:

RankineActAct

WW

acK (6)

where [ WAct]Rankine is the sum of the rates of internal energy dis-sipation on the failure planes behind the wall, without consider-ing the soil adhesion on the wall back (c w=0). The analysis con-sidered the minimum and maximum limits for: the total unit weight of the clay to be 15 and 20 kN/m3, the embedded wall height into clay to be 6 and 22.5 m, the ratio c ucomp/ zc to be 0.22 and 0.27, the coefficient of soil strength anisotropy to be 0.5 and 1.0, and the soil adhesion ratio (c w/cu) to be 0.5 and 1.0, respec-tively, and assuming the undrained compressive soil strength at the clay surface, c0, to be 1.5 kN/m2 (without any surcharge on the clay surface). The soil strength anisotropy and soil adhesion ratio had marked influence on the soil adhesion factor, and the influences of the other parameters were marginal. Thus, average soil adhesion factors for Hclay/c0=60 ~ 300 are also presented in the Table. For design purposes, K ac can be considered to range between 1.16 and 1.21 for c w/cu = 0.5, and between 1.26 and 1.36 for cw/cu = 1.0, when the coefficient of soil strength anisotropy ranges between 0.5 and unity, and Kac for a specific site can be estimated from these limits by interpolation.

CONCLUSIONS

Using Mechanism C in the framework of the upper bound

method, the active and passive earth pressures on the rough DM self-supported walls were successfully calculated and matched the measured earth pressures in the Centrifuge tests at failure. In addition, the mobilized soil shear strengths and soil adhesion on the wall surfaces, at failure, were assessed, and a range of average values for the soil adhesion factor behind the wall were pre-sented, in order to be used in the wall design. The numerical re-sults led to the following conclusions: 1. The current design method overestimated the active and pas-sive earth pressures on both wall sides, the design bending mo-ment and shear force as well as the maximum bearing stress on the foundation soil, as it use the unconfined compressive soil shear strength and ignore both of the soil strength anisotropy and soil adhesion on the vertical wall surfaces. 2. The BS8002 empirical design earth pressure relationships pre-dicted reasonably the active earth pressures on the wall, but over-estimated the passive earth pressures, as it overestimated the mo-bilized soil adhesion on the wall front and soil strength in front of the wall. 3. The small prefailure soil and wall displacements fully mobi-lized the undrained Triaxial peak soil shear strengths behind the wall and the soil adhesion on the wall back, in early excavation steps before failure, as well as the maximum soil shear strength in the plain strain condition on the soil-wall base interface before failure. However, they were not sufficient to mobilize the maxi-mum undrained extension Triaxial soil shear strengths in front of the wall, even without considering the soil adhesion on the wall front. So, the wall designers should use a mobilization factor to estimate the mobilized soil strengths in front of the wall at fail-ure, and ignore the soil adhesion on the wall front.

REFERENCES

British Standards Institution 1994. Code of Practice for Earth Retaining Structures-BS8002 , London, UK.

Casagrande, A. & Carillo, N. 1944. Shear failure of Anisotropic Soils. Contributions to Soil Mech anics, ASCE: 122-135.

El Nahas, A. & Takemura, J. 2002a. External stability of vertical excavations in soft clay with self-supported DMM walls. Soilsand Foundations 42: 53-69.

El Nahas, A. & Takemura, J. 2002b. External stability of open excavations in soft clay with DM self-supported walls. Intl.Conf. Physical Model in Geotechnics -ICPMG �02: 835-840.

El Nahas, A. & Takemura, J. 2004. Upper bound solution and Centrifuge modeling for excavations in anisotropic soft clay with rough retaining walls at failure. Intl. Journal of Physical Modelling in Geotechnics-IJPMG. (Submitted)

Ladd, D.D., et al. 1977. Stress-deformation and strength characteristics: State-of-the-Art Report. Proceedings of 9th International Conference Soil Mechanics Foundation Engineering 2:421-482. Tokyo.

Potts, D.M. 1991. Finite element simulation of embedded retaining walls. In Banerjee & Butterfield (ed). Developments in Soil Mechanics & Foundation Engineering : 131-167.Elsevier Applied Science.

Shiomi, M. et al. 1996. Slope stability using the admixture method. Proceedings IS-Tokyo 96 1: 563-568.

Takemura, J. et al. 1999. Centrifuge model tests on double propped wall excavation in soft clay. Soils and Foundations39: 75-87.

Terzaghi, K., et al. 1996. Soil Mechanics in Engineering PracticeNew York: Wiley & Sons.


183

Analysis of Piled Raft Foundation Subjected to Ground Movement Induced by Tunnelling

P. Kitiyodomi), T. Matsumotoii) & K. Kawaguchiiii)

Kanazawa University, 2-40-20 Kodatsuno, Kanazawa 920-8667, Japan i) [email protected], ii) [email protected], iii) [email protected]

Abstract: In this paper, a three-dimensional simplified deformation analytical method is presented for the analysis of piled raft founda-tions subjected to ground movement induced by tunnelling. In the method, a hybrid model is employed in which the flexible raft ismodelled as thin plates, the piles as elastic beams, and the soil is treated as springs. The interactions between structural members, pile-soil-pile, pile-soil-raft and raft-soil-raft interactions, are modelled based on Mindlin�s solutions for both vertical and lateral forces. Reasonable agreement is found between some existing published solutions and those developed herein. The method is then used for a parametric study of single piles, pile groups and piled rafts subjected to ground movements induced by tunnelling.

1 INTRODUCTION

Piled raft foundations have been widely recognized as one of the most economical foundation systems since Burland et al. (1977) presented the concept of �settlement reducers�. In recent years, there has been an increasing recognition that the inclusion of the resistance of the raft in pile foundation design can lead to a con-siderable economy without compromising the safety or the per-formance of the foundation. Considering current trends toward the limit state design or performance based design in the area of foundation engineering, precise estimation of the deformation of a pile foundation and of the stresses of their structural members is a vital issue in the framework of these new design criteria. In the preliminary design stage, a number of alternative calculations may be required, varying the number of piles, the pile length, the pile spacing, the locations of the piles, and so on. Hence, a feasi-ble but reliable deformation analysis method for piled raft foun-dations would be useful. As a preliminary routine design tool of a piled raft foundation subjected to external loading (vertical, lat-eral and moment loads), a computer program PRAB (Piled Raft Analysis with Batter piles) has been developed by the authors (Kitiyodom & Matsumoto 2002 & 2003a). In Kitiyodom & Ma-tsumoto (2003b), the program was extended to accommodate three-dimensional simplified analysis of piled raft foundations subjected to ground movements.

In an urban environment, many buildings are supported by pile foundations. Due to a lack of available space, a lot of under-ground constructions such as tunnelling are increasing. Tunnel-ling may cause ground movements invariably, which in turn im-pose axial and lateral forces on the existing pile foundations resulting in extra deformations of the foundations. Some research on the analysis of pile foundations subjected to ground move-ments induced by tunnelling has been done. Chen et al. (1999) analyzed the response of single piles by de-coupled loadings in two dimensions. Xu & Poulos (2001) and Loganathan et al.(2001) employed a three-dimensional coupled boundary element approach to analyze the response of vertical piles subjected to ground movements induced by tunnelling. However, in their works only single piles and pile groups were considered, and the method can give only the elastic solutions of the problems. A complete three-dimensional analysis of a foundation system sub-jected to ground movements induced by tunnelling can be carried

out by a finite element analysis, for instance, the work of Mroueh & Shahrour (2002). However, a finite element analysis is more suited to obtaining benchmark solutions against which to com-pare simpler analysis methods, or to obtaining solutions of a de-tailed analysis for the final design of a foundation, rather than as a preliminary routine design tool.

This paper presents an extension of the computer program PRAB, in order to incorporate the problem of piled raft founda-tions subjected to ground movements induced by tunnelling. With an aim to examine the validity of the extended PRAB, the results calculated using PRAB are compared with the solutions available from previous research. The proposed method is then used as a parametric study tool for the analyses of single piles, pile groups and piled rafts subjected to ground movements in-duced by tunnelling.

2 METHOD OF ANALYSIS

The problem considered in this work is shown in Fig. 1 where an existing pile foundation is located adjacent to a tunnel under con-struction. Tunnelling generally will induce ground movements in both vertical and lateral directions. These ground movements will cause vertical settlements and lateral deflections in the piles. The analysis of such a problem may be carried out in two stages; first, estimation of the free-field ground movements induced by tunnel-ling; second, the imposition of these ground movements on the foundation and the computation of the consequent foundation re-sponses (e.g. Chen et al. 1999; Xu & Poulos; 2001; Loganathan et al.; 2001).

Tunnel L

R

H

d

z

x Pile foundation

Fig. 1. Pile foundation adjacent to tunnelling.

184

2.1 Estimation of Tunnelling-Induced Ground Movements

Methods for estimating ground movements induced by tunnelling may be classified broadly into three categories, empirical method, finite element method, and analytical method. However in the case that there is inadequate detailed sited information to warrant the use of either the empirical method or a complex finite element method, simple closed form analytical solutions, such as those given by Sagaseta (1987), Verruijt & Booker (1996), and Loga-nathan & Poulos (1998), may be useful. In this work, surface set-tlement, subsurface settlements and lateral deformations of the ground induced by tunnelling are calculated by means of Eqs. (1) to (3) based on the closed-form analytical solutions presented by Loganathan & Poulos (1998).

2s2

0 0 2 2 24 1 1.38expz

H xU RH x H R

(1)

22s2

0 2 2 22 2 22

2 2

2 2

23

1.38 0.69exp (2)

z

z x z Hz H z HU R

x z H x z H x z H

x zHH R

s20 2 2 22 2 22

2 2

2 2

3 4 41

1.38 0.69exp (3)

xz z H

U R xx z H x H z x H z

x zHH R

where Uz = 0 is the ground surface settlement, Uz the subsurface settlement, Ux the lateral soil movement, R the tunnel radius, zthe depth below the ground surface, H the depth of tunnel hori-zontal axis level, s the soil Poisson�s ratio, 0 the average ground loss ratio, and x is the lateral distance from the tunnel centreline.

The equivalent average ground loss ratio, 0, is defined as

22

2

0 2 242 100% 100%

4

gR RgR g

R R (4)

where g is the gap parameter and can be estimated as follows (Lee et al., 1992):

*p 3Dg G U (5)

where Gp is the physical gap (usually the difference between the maximum outside diameter of the tunnelling machine and the outside diameter of the lining for a circular tunnel), U*

3D the elas-toplastic deformation into the tunnel face, and is the work-manship factor.

2.2 Analysis of Pile Foundation Response

Figure 2 illustrates the analytical model for piled raft foundations employed in this study. The flexible raft is modelled as thin plates, the piles as elastic beams, and the soil is treated as springs. The interaction between structural members, pile-soil-pile, pile-soil-raft and raft-soil-raft interactions are calculated based on Mindlin�s solutions (Mindlin, 1936) for both vertical and lateral forces.

y

x

Fig. 2. Plate-beam-spring modelling of a piled raft.

The load-displacement relationship of the group piles and the raft can be written as

pK w P (6)

rK w P (7)

where [Kp] is the pile stiffness matrix, [Kr] the raft stiffness ma-trix, {w} the displacement vector, and {P} the internal force vec-tor.

The relative displacement of the soil at the structure member-soil interface, wi-w0i, at a particular node i due to interaction forces acting on itself and at other nodes in the piled raft system can be written in the following discrete form:

01

n

i i ij jj

w w a P (8)

where wi is the soil deformation at the structure member-soil in-terface at node i, w0i is the soil deformation at node i due to the ground movement, aij is the soil flexibility coefficient denoting the deformation at node i due to a unit load acting at node j, and n is the total number nodes in the piled raft system. Eq. (8) is re-written in the following matrix form considering all degrees of freedom at each node.

0w w A P (9)

where [A] is the soil flexibility matrix. The diagonal coefficients of [A] are determined by inverting

the soil spring stiffness matrix [Ks]. The details of the soil spring stiffness values can be found in Kitiyodom & Matsumoto (2002, 2003a). The off-diagonal non-zero coefficients in the matrix [ A]represent structural member-soil-structural member interactions and are calculated based on Mindlin �s solutions for both vertical and lateral forces.

For further use, Eq. (9) is rewritten as

0[ ]C w w P (10)

where [C] = [A]-1.Finally, from Eqs. (6), (7) and (10), we get

r p 0[ ]{ } [ ]{ } [ ]C K K w K w C w (11)

where [K] is the global stiffness of the piled raft system. The vector [C]{w0} represents the forces acting on the piled

raft induced by the ground movements. This set of equations can be solved for the vertical and lateral ground movements to give the pile settlements, deflections and rotations from which the ax-ial forces, the shear forces and the bending moments can be ob-tained.

185

The bilinear (elastic perfectly plastic) responses of the pile shaft and the pile base resistances, and also the raft base resis-tance in the case of piled rafts can be taken into account in the analysis.

3 ACCURACY OF THE PROPOSED METHOD

In order to ensure the validity of the proposed method, the results calculated using PRAB were compared with the results from re-lated previous research.

3.1 Tunnelling-Induced Single Pile Responses

Xu & Poulos (2001) demonstrates the elastic response of a single pile in the problem as shown in Fig. 3. Figure 4 shows the com-puted free-field vertical and lateral ground movement profiles in-duced by tunnelling, corresponding to three ground loss ratios 0of 1, 2.5 and 5% and for a distance x = 4.5 m. It can be seen that the ground movements increase with increasing ground loss ratio.

Figure 5 shows the single pile responses calculated using the proposed method compared with the responses from the bound-ary element method by Xu & Poulos (2001). It can be seen that the results obtained from PRAB, except for the axial forces, match very well with the results from the boundary element method. It can be seen from Fig. 5(d) that although the shape of the profiles are identical, the results calculated by PRAB under-estimated the axial forces about 50 percents compared with those of the boundary element method. This was thought to be due to the inherent assumption in the load-transfer method for model-ling vertical soil behaviour in single pile in which the influence of vertical forces acting at other nodes associated with the same pile on the considered node are neglected.

L = 25 m

R = 3 m

H = 20 m d = 0.5 m

z

x = 4.5 m

Ep = 30000 MPaEs = 24 MPa

s = 0.5

Fig. 3. Problem analyzed (single pile).

30

25

20

15

10

5

0-20 0 20 40 60

0 = 1%

0 = 2.5%

0 = 5%

Vertical Ground Movement (mm)

30

25

20

15

10

5

0-50 -40 -30 -20 -10 0

0 = 1%

0 = 2.5%

0 = 5%

Lateral Ground Movement (mm)

(a) Vertical ground movement (b) Lateral ground movement

Fig. 4. Computed free-field ground movement at x = 4.5 m.

30

25

20

15

10

5

0-75 -50 -25 0

0 = 1%

0 = 2.5%

0 = 5%

PRAB Xu & Poulos, 2001

Lateral Deflection of Pile (mm)

30

25

20

15

10

5

0-100 0 100 200 300 400

0 = 5%0 = 2.5%0 = 1%


Bending Moment (kN.m)

(a) Lateral deflection (b) Bending moment

30

25

20

15

10

5

00 10 20 30 40 50

0 = 1%

0 = 2.5%0 = 5%


Vertical Movement of Pile (mm)

30

25

20

15

10

5

00 1000 2000 3000 4000

0 = 1%

0 = 2.5%0 = 5%


Axial Force (kN)

(c) Vertical movement (d) Axial force

Fig. 5. Typical single pile response at x = 4.5 m.

3.2 Tunnelling-Induced Pile Group Responses

Loganathan et al. (2001) demonstrates the elastic response of a pile group to the ground movement induced by tunnelling with the ground loss ratio 0 of 1% as shown in Fig. 6.

Figure 7 shows pile group responses calculated using the pro-posed method compared with the responses from the boundary element method by Loganathan et al. (2001). Both the front piles and back piles responses are shown in the figure. It can be seen again that the results calculated using PRAB, except for the axial forces, match very well with the results from the boundary ele-ment method. Again the results calculated by PRAB underesti-mated the axial forces about 50 percents compared with those of the boundary element method.

L = 25 m

R = 3 m

H = 20 md = 0.8 m

z

x = 4.5 m


p = 0.25 s = 0.5

s = 2.4 m

front back

Rigid raft

Fig. 6. Problem analyzed (pile group).

186

30

25

20

15

10

5

0-12 -10 -8 -6 -4 -2 0

PRAB (front) PRAB (back) Loganathan et al., 2001


30

25

20

15

10

5

0-150 0 150 300 450




30

25

20

15

10

5

00 2 4 6 8



30

25

20

15

10

5

00 200 400 600 800

frontback


Axial Force (kN)


Fig. 7. Typical pile group response.

4 PARAMETRIC STUDY

In the previous section, the validity of the proposed method was examined. It was shown that the proposed method is a valid ap-proach to analyze deformation of and load distribution in pile foundations subjected to vertical and lateral ground movements induced by tunnelling. In this section, a parametric study of sin-gle piles, pile groups and piled rafts subjected to ground move-ments induced by tunnelling is conducted.

4.1 Single Piles

Analyses were conducted for single piles subjected to ground movements induced by tunnelling. The ranges of the parameters were set as 4.5 to 13.5 m for the lateral distance of pile from tun-nel centreline x, 5 to 25 for the pile slenderness ratio L/d, and 0.5 to 2.0 for the ratio of the depth of tunnel horizontal axis level, H,to the pile embedment length, L. The pile diameter, d, was set at 1 m, the tunnel radius, R, at 3 m, the equivalent average undrained ground loss, 0, at 1 %, the soil Young�s modulus, Es,at 24 MN/m2, the pile Young�s modulus, Ep, at 30 GN/m2, the soil Poisson�s ratio, s, at 0.5, and the pile Poisson�s ratio, p,was set at 0.25 throughout.

Figure 8 shows the responses of single piles located at differ-ent distances away from the tunnel centerline. The pile slender-ness ratio, L/d, was set at 20, and the ratio H/L was set constant at 1. It can be seen that the calculated values of the maximum lateral deflection and the maximum vertical movement of the pile, and the maximum bending moment and the maximum axial force along the pile decreases as the lateral distance of the pile from the tunnel centreline increases.

Figure 9 shows the response of single piles with different pile slenderness ratios, L/d, for the case where the ratio H/L = 1 and the lateral distance of the pile from the tunnel centreline x = 4.5 m. It can be seen from Fig. 9(a) that in the case of L/d < 10, the pile deformed like a short pile in which all parts of the pile

leaned, while in the other cases, the bottom parts of the pile near the tunnel axis largely deformed compared to the upper parts. When the pile deformed like a short pile the vertical movement of the pile (Fig. 9(c)) and the axial force along the pile (Fig. 9(d)) increase as the pile slenderness ratio increases. On the other hand, when the pile deformed like a long pile the vertical movement of the pile and the axial force along the pile decreases as the pile slenderness ratio increases. As shown in Fig. 9(b), the value of bending moment and the depth where the maximum bending moment occurred increase as the pile slenderness ratio increases. However, when L/d > 15, the values of the maximum bending moment are almost the same.

1.0

0.8

0.6

0.4

0.2

0.0-10 -8 -6 -4 -2 0Lateral deflection of pile(mm)

x 4.50m 6.75m 9.00m 11.25m 13.00m

1.0

0.8

0.6

0.4

0.2

0.0-100 0 100 200

Bending moment(kN.m)

x 4.50m 6.75m 9.00m 11.25m 13.00m


1.0

0.8

0.6

0.4

0.2

0.00 5 10 15

Vertical movement of pile(mm)

x 4.50m 6.75m 9.00m 11.25m 13.00m

1.0

0.8

0.6

0.4

0.2

0.00 100 200 300 400 500

Axial force(kN)

x 4.50m 6.75m 9.00m 11.25m 13.00m


Fig. 8. Single pile responses located at different distances from tunnel.

1.0

0.8

0.6

0.4

0.2

0.0-20 -15 -10 -5 0

Lateral deflection of pile(mm)

L/d 5 10 15 20 25

1.0

0.8

0.6

0.4

0.2

0.0-200 -100 0 100 200


L/d 5 10 15 20 25


1.0

0.8

0.6

0.4

0.2

0.00 2 4 6 8 10 12


L/d 5 10 15 20 25

1.0

0.8

0.6

0.4

0.2

0.0-100 0 100 200 300

Axial force(kN)

L/d 5 10 15 20 25


Fig. 9. Single pile responses with different pile slenderness ratio.

187

1.0

0.8

0.6

0.4

0.2

0.0-14 -12 -10 -8 -6 -4 -2 0Lateral deflection of pile(mm)

H/L 0.50 0.75 1.00 1.50 2.00

1.0

0.8

0.6

0.4

0.2

0.0-200 0 200 400 600 800


H/L 0.50 0.75 1.00 1.50 2.00


1.0

0.8

0.6

0.4

0.2

0.00 2 4 6 8 10


H/L 0.50 0.75 1.00 1.50 2.00

1.0

0.8

0.6

0.4

0.2

0.0 0 300 600 900 1200 Axial force(kN)

H/L 0.50 0.75 1.00 1.50 2.00


Fig. 10. Influence of pile tip location on the single pile responses.

Figure 10 shows the influence of the position the pile tip re-garding the tunnel horizontal axis on the responses of single piles. The slenderness ratio, L/d, was set at 20, and the lateral distance of pile from the tunnel centreline x = 4.5 m. As shown in Fig. 10(a), the maximum lateral deflection of pile is observed near the position of the tunnel horizontal axis. It can be also seen from Fig. 10(b) that the position of the pile tip regarding the tunnel axis largely affects the bending moment. When the pile tip is above the tunnel horizontal axis, tunnelling induces low pile bending moment compared with that obtained when the pile tip is under the tunnel horizontal axis.

Moreover, it can be seen also from Fig. 10 (d) that the position of the pile tip affects the distribution of axial force along the pile. When the pile tip is located above the tunnel horizontal axis, the axial force is negative. When the pile tip is located below the tunnel horizontal axis, the axial force increases up to its maxi-mum value at the depth of the tunnel horizontal axis followed by a decrease due to the upward movement of the soil situated under the tunnel horizontal axis, and the maximum axial force is ob-tained near the depth of the tunnel horizontal axis level.

4.2 Piled Rafts and Pile Groups

Figure 11 illustrates the problem of existing 2 2 squared piled raft foundations subjected to ground movements induced by tun-nelling. The square rigid raft has dimensions of 6 m in breadth and length, and 2 m in thickness. Piles beneath the raft are spaced at 3d. The pile Young�s modulus, Ep = 30 GN/m2, the soil Young�s modulus, Es = 24 MN/m2, the pile Poisson�s ratio, p = 0.25 and the soil Poisson �s ratio, s = 0.5. The ratio H/L = 1, the tunnel radius, R = 3 m, the pile diameter, d = 1 m, and the lateral distance of the front pile from the tunnel centreline, x = 4.5 m.

The analyses of piled raft and pile group responses have been carried out using the proposed method varying the pile slender-ness ratio, L/d. Only elastic behaviour of the pile and the soil has been considered in this analysis.

Figures 12 & 13 show comparisons of the computed responses

of piles in the piled raft (PR) and piles in the pile group (PG). The responses of the single piles (SP) located at the lateral dis-tances from the tunnel centreline x = 4.5 and 7.5 m are also shown in the figures.

Figure 12 shows comparisons of the responses of piles in the piled raft, piles in the pile group and single piles with the slen-derness ratio, L/d, of 25. A comparison of the induced lateral de-formation profiles of piles is shown in Fig 12(a). It can be seen that the maximum lateral deflection of the front piles is higher than that of the back piles, and occurs at the depth of the pile tip, which equals to the depth of the tunnel horizontal axis. The lat-eral deformation profiles of single piles are almost identical to those of the corresponding piles in the pile group and in the piled raft.

Figure 12(b) shows a comparison of the induced bending mo-ment profiles of piles in the piled raft, piles in the pile group and single piles. It can be seen that the bending moment profiles of single piles are almost the same with those of the corresponding piles in the pile group and in the piled raft, except for a difference near the pile head due to the difference in the pile head connec-tion conditions between single pile and pile in the piled raft and the pile group.

Figure 12(c) shows a comparison of the induced vertical movement profiles of piles in the piled raft, piles in the pile group and single piles. It can be seen that the vertical movement of the front pile is higher than that of the back pile. The vertical movement profiles of single piles are almost identical to those of the corresponding piles in the pile group and in the piled raft.

Figure 12(d) shows a comparison of the induced axial force profiles of piles in the piled raft, piles in the pile group and single piles. The axial force profiles of single piles are different from the axial force profiles of the corresponding piles in the pile group and in the piled raft.

Loganathan et al. (2001) suggests that the results of a rela-tively simple single pile analysis can be used to predict the in-duced bending moment, lateral deflection and vertical movement in a pile group at identical distances from the tunnel, except for the axial forces in the pile. The above calculation results support this suggestion. However, it will be shown in Fig. 13 that the sin-gle pile responses cannot be used to predict the responses of the corresponding piles in the pile group as well as those in the piled raft, if the value of the pile slenderness ratio is small.

Figure 13 shows comparisons of the responses of piles in the piled raft, piles in the pile group and single piles with the slen-derness ratio, L/d, of 5. It can be seen that the responses of the single piles are totally different from those of the corresponding piles in the piled raft and in the pile group.

L

R = 3 m

Hd = 1 m

z

x = 4.5 m


p = 0.25

s = 0.5 H/L = 1

s = 3 m

front back

Br = 6 m

tr = 2 m

x = 7.5 m

Fig. 11. Problem analyze (piled raft).

188

1.0

0.8

0.6

0.4

0.2

0.0-15 -12 -9 -6 -3 0

L/d = 25

PR (front) PR (back) PG (front) PG (back) SP (x = 4.5m) SP (x = 7.5m)


1.0

0.8

0.6

0.4

0.2

0.0-600 -400 -200 0 200

L/d = 25




1.0

0.8

0.6

0.4

0.2

0.00 4 8 12 16 20 24

L/d = 25



1.0

0.8

0.6

0.4

0.2

0.0-450-300-150 0 150 300 450

L/d = 25


Axial Force (kN)


Fig. 12. Piled raft and pile group responses ( L/d = 25).

1.0

0.8

0.6

0.4

0.2

0.0-15 -12 -9 -6 -3 0

L/d = 5



1.0

0.8

0.6

0.4

0.2

0.0-1200 -800 -400 0 400

L/d = 5




1.0

0.8

0.6

0.4

0.2

0.00 4 8 12 16 20 24

L/d = 5



1.0

0.8

0.6

0.4

0.2

0.0-450-300-150 0 150 300 450


L/d = 5

Axial Force (kN)


Fig. 13. Piled raft and pile group responses ( L/d = 5).

Moreover, it can be seen from Fig. 12 that the responses of the piles in the pile group and the piles in the piled raft are almost identical when the pile slenderness ratio is high ( L/d = 25). How-ever, for a low pile slenderness ratio (L/d = 5) as shown in Fig. 13, the piles in the pile group deform less than those in the piled raft, and the induced axial force and bending moment are higher in the case of the piles in the piled raft. Note that in the analysis, the foundations are subjected to the same vertical and lateral free field ground movements induced by tunnelling at the pile-soil in-terface, but the passive loads induced by these ground move-ments acting on two types of the foundations are different.

5 CONCLUSIONS

A simplified analytical method has been proposed for the analysis of the deformation and the load distribution of piled raft founda-tions subjected to ground movements induced by tunnelling. The proposed method was verified through comparisons with the re-sults from previous research.

The proposed method was employed for a parametric study of single piles, pile groups and piled rafts. It was suggested from the calculation results that in the case of piled rafts in which the short piles are commonly used, the resistance of the raft and interac-tions between the structural members should be considered in the analysis of the foundation.

REFERENCES

Burland, J.B., Broms, B.B. & De Mello, V.F.B. 1977. Behaviour of foundations and structures. Proceeding of the 9th ICSMFE2: 496-546, Tokyo, Japan.

Chen, L.T., Poulos, H.G. & Loganathan, N. 1999. Pile responses caused by tunneling. Journal of Geotechnical and Geoenvi-ronmental Engineering, ASCE 125(3): 802-811.

Kitiyodom, P. & Matsumoto, T. 2002. A simplified analysis method for piled raft and pile group foundations with batter piles. International Journal for Numerical and Analytical Methods in Geomechanics 26: 1349-1369.

Kitiyodom, P. & Matsumoto, T. 2003a. A simplified analysis method for piled raft foundations in non-homogeneous soils. International Journal for Numerical and Analytical Methods in Geomechanics 27: 85-109.

Kitiyodom, P. & Matsumoto, T. 2003b. Extension of a computer program PRAB for deformation analysis of piled rafts sub-jected to ground movements. Proceedings of the Sino-Japanese Symposium on Geotechnical Engineering : 74-79.

Lee, K.M., Rowe, R.K. & Lo, K.Y. 1992. Subsidence owing to tunnelling I: Estimating the gap parameter. Canadian Geo-technical Journal 29: 929-940.

Loganathan, N. & Poulos, H.G. 1998. Analytical prediction for tunelling-induced ground movements in clays. Journal of Geotechnical and Geoenvironmental Engineering ASCE124(9): 846-856.

Loganathan, N., Poulos, H.G. & Xu K.J. 2001. Ground and pile-group responses due to tunnelling. Soils and Foundations41(1): 57-67.

Mindlin, R.D. 1936. Force at a point interior of a semi-infinite solid. Physics 7: 245-256.

Mroueh, H. & Shahrour, I. 2002. Three-dimensional finite ele-ment analysis of the interaction between tunneling and pile foundations. International Journal for Numerical and Ana-lytical Methods in Geomechanics 26: 217-230.

Sagaseta, C. 1987. Analysis of undrained soil deformation due to ground loss. Géotechnique 37: 301-320.

Verruijt, A. & Booker, J.R. 1996. Surface settlements due to de-formation of a tunnel in an elastic half plane Géotechnique46(4): 753-756.

Xu, K.J. & Poulos, H.G. 2001. 3-D elastic analysis of vertical pile subjected to �passive� loadings. Computers and Geo-technics 28: 349-375.


189

Experimental Verification of Vertical Support Systems for Long Steel-Pipe Piles and Sheet Piles

K. Tomisawa & S. Nishimoto Civil Engineering Research Institute of Hokkaido(CERI) Hiragishi 1-3-1-34,Toyohira-ku,Sapporo, Hokkaido 062-8602 [email protected] & [email protected]

Abstract: When installing long steel-pipe piles and sheet piles, construction management that is more precise and accurate than that for ordinary piling work is considered to be necessary for ensuring the required design bearing capacity of foundations. In this study, im-pact, vertical loading and dynamic loading tests for long steel-pipe piles (65 m) and sheet piles (61.5 m) were conducted at a site with deep peaty soft ground as part of construction management, and a vertical support system for deep foundations was examined to de-velop appropriate construction management methods for the future .

1 INTRODUCTION

In this study, impact and vertical loading tests for long steel-pipe piles (L1 = 65 m) and driving management, impact and dynamic loading tests for long steel-pipe sheet piles (L2 = 61.5 m) were conducted at a site with deep peaty soft ground as part of con-struction management, and a vertical support system for deep foundations was examined to develop appropriate construction management methods for the future. During these tests, the driv-ing process had a tendency to be difficult due to the applicability of rebound management through the wave equation and residual stress, especially in the case of the driving of long steel-pipe piles, as well as because of the general existence of coupling pipes in the case of long steel-pipe sheet piles in comparison with single piles. Focus was therefore placed on the estimation of ap-propriate coupling pipe resistance, stress distribution and other technical issues concerning construction management including percussion for finish.

2 OVERVIEW OF THE SITE AND FOUNDATION FORM

A series of management tests for long steel-pipe and long steel-pile sheet piles were conducted at the steel-pipe pile foundation (steel-pipe pile diameter: 600mm, pile thickness: t=13mm, pile length: L= 65m) of P-2 pier and the steel-pipe sheet pile founda-tion (steel-pipe sheet pile diameter: 1000mm, sheet pile thick-ness: t=12 mm, sheet pile length: L = 61.5 m) of P-4 pier of the Shin-Kushirogawa Bridge on Hokkaido in Japan.

Figure 1 shows the type of the steel-pipe pile foundation of P-2 pier, as well as the structural type, planar shape and soil boring log of the steel-pipe sheet pile foundation of P-4 pier The P-2pier was a group of pile-type foundations with seven rows of seven piles, and the installation of steel-pipe piles was conducted using the driving method with a hydraulic hammer of W = 10 tons. While on-site coupling of six rods (1 rod = 11m to 12 m ) is required for the installation of steel-pipe piles, friction cut was not applied to the pile heads to ensure the design skin friction. As

Fig. 1. Forms of P-2 steel-pipe pile and P-4 steel-pipe sheet and soil log.

190

a result, the design ultimate bearing capacity per steel-pipe pile of P-2pier was set at Rup=5800kN, and the de sign end-bearing ca-pacity as a bearing pile was qdp=7640 kN/m2.

Since installation of steel-pipe sheet piles requires on-site cou-pling of five rods (lower sheet = 13.5 m, middle sheets 1, 2 and 3 = 12.0 m, upper sheet = 12.0 m), the lower sheet was erected and closed by taking the workability and residual stress associated with sheet driving into account. Subsequently, pre-placement of sheets was carried out in between middle sheet piles 1, 2 and 3 and upper sheet piles, and post- placement of sheets was also conducted in between them with a hydraulic hammer of W = 11.5tons, followed by the closure of each rod. Regarding the vertical bearing capacity of the entire steel-pipe sheet pile foundation, the bearing capacity of each single sheet pile was first calculated us-ing the formula for vertical bearing capacity of single piles by ig-noring the cross section of coupling pipes. This vertical bearing capacity was then evaluated as the total vertical bearing capacity for the number of piles 1),2). As a result, the design ultimate bear-ing capacity of each single sheet pile was Rus=9300kN per pile for P4 steel-pipe sheet piles.

The bearing stratum was found in a gravel bed 60 m or deeper since soft quarternary alluvial silt is distributed deeply to the depth of 40 to 50 m in the ground of the site. As a result, instal-lation of a very long foundation was required to place the steel-pipe piles and sheet piles of this bridge in the bearing stratum and ensure the required bearing capacity.

The coupling pipes of steel-pipe sheet piles are divided into P-P, P-T, L-T1 and L-T2 types depending on their used areas, such as curves. The commonly used P-P type was employed for P4steel-pipe sheet piles in this study.

3 DYNAMIC LOADING TEST METHOD FOR STEEL-PIPE SHEET PILES

The dynamic loading test is a test used to apply axial impact on vertically installed single piles, measure the dynamic strain and acceleration and estimate the dynamic resistance from the meas-ured values using analysis based on the kinematic wave theory. In Japan, it was compiled in manual form as a practical test method in 2002 3).

If load is applied in the axial direction of a pile body for a short period of time, the stress wave is transmitted in the axial di-rection of the pile body. This wave phenomenon can be ex-pressed as a one-dimensional wave equation (Eq. (1)).

2 u(x ,t) / t2 = c2 u(x ,t) / x2 (1)

where c: stress wave propagation velocity (m/sec2), u=dis-placement (m), x=position (m) and t=time (sec)

From the measured values of dynamic strain and acceleration, the general solution for this one-dimensional wave theory was used, and the total resistance Rt (= static resistance component Rs+ dynamic resistance component Rd) was calculated using the CASE method 5), which can be expressed by Eq. (2) on the as-sumption that the total of the progressive wave generated by the impact on the pile head and the backward wave that returned af-ter moving around the pile body balances out with the ground re-sistance.

Rt(xo,t)=Fd(xo,t Lm/c) Fu(xo,t+Lm/c) (2)

Fig. 2. Conceptual drawing of the dynamic loading test for steel-pipe sheet piles.

Rs(xo,t)=Rt(xo,t) Rd(xo,t)Rd(xo,t)=Jc Z b(t)

b(t)= Fd(xo,t Lm/c) Fu(xo,t+Lm/c) /Zwhere Rt(xo,t)=total resistance (kN), Rs(xo,t)=static resistance component (kN), Rd(xo,t)=dynamic resistance component (kN), Lm=total pile length (m), Fd(xo,t Lm/c)=progressive wave of impact (kN), Fu(xo,t +Lm/c)=longitudinal backward wave (kN); Jc=CASE damping, Z=pile body impedance and b(t)=pile head velocity (m/sec).

Here, the static resistance component of the driving pile equivalent to the design ultimate bearing capacity can be found by removing the dynamic resistance component from the total re-sistance using CASE damping. In this case, because CASE damp-ing for estimation of the dynamic resistance component varies according to ground conditions and pile specifications, it is de-termined by waveform matching analysis using a characteristic curve in which the resistance components are converted into a dash-pot model.

The dynamic loading test can therefore generally be conducted for single steel and existing concrete piles. For this study, how-ever, a new application method for the dynamic loading test was developed to calculate the vertical bearing capacity of steel-pipe sheet pile foundations. Figure 2 shows the conceptual drawing of this dynamic loading test. The vertical bearing capacity of steel-pipe sheet piles differs from that of single piles. Vertical bearing capacity P1 is thought to consist of end-bearing capacity A, skin resistance B and resistance C, which is generated at the joint. In other words, it can be described by the relationship P1 = A+B+C. This means that the vertical bearing capacity of inter-locking sheet piles is considered to be P2 = 2A+2B+C. This is because the competition of coupling pipes at the center of inter-locking can be ignored if load is applied on two interlocking steel-pipe sheet piles.

As a result, the sheet pile bearing capacity P after removing re-sistance C of the coupling pipes can in principle be calculated by subtracting P1 from P2 in Eq. (3) if dynamic resistance P1 of one steel-pipe sheet pile and P2 of the interlocking steel-pipe sheet piles can be estimated using the dynamic loading test.

P = P2 - P1 = ( 2A + 2B + C ) - ( A + B +C ) = A + B (3)

Since two steel-pipe sheet piles (No. 4 and 5) must be put in place simultaneously while being interlocking to each other in addition to the driving of ordinary steel-pipe sheet pile No. 3 at the time of the test, a special cap was produced to connect and cover the heads of two steel-pipe sheet piles and installed at the head of the steel-pipe sheet piles, and driving was conducted with one hammer.

191

Fig. 3. Impact wave forms of steel-pipe piles.

The dynamic loading test was conducted at the position of per-cussion for finish in the same manner as in the impact test to es-timate the final bearing capacity.

4 VERIFICATION OF IMPACT TEST RESULTS

4.1 Impact Waveform and Stress of Long Steel-Pipe Piles

While the dropping height of the hydraulic hammer ( W = 10tons) was fixed at 1 m in the impact test for test steel-pipe pile No. 1, piles almost sank by themselves in the soft ground during the early stage, and the measured impact stress reached its maximum in the final rod pile after intrusion into the bearing stratum. the early stage, and the measured impact stress reached its maxi-mum in the final rod pile after intrusion into the bearing stratum.

Figure 3 illustrates the impact stress waveform measured at the pile head at the position of percussion for finish of test piles (rep-resentative waveform after elimination of noise) overlapping the basic waveform of the St. Venant solution 5) provided in Eq. (4).

= o e-( p Ap/WH Cp t ) (4)o=Ep/Cp (2g h)

where Ap=pile cross section (m2), WH=hammer weight (tons), p=unit weight of the pile (kN/m3), Cp=elastic wave velocity

in the pipe (m/s), Ep=Elastic modulus of the pile (kN/m2),t=elapsed time (t), g=gravitational acceleration (m/s2) and h=hammer dropping height (m).

Moreover, Fig. 3 shows that the measured waveform nearly corresponded with the theoretical waveform, although the peak value was slightly lower due to the effect of energy loss, indicat-ing that the standard construction of single piles was verified. It was also considered that there were no problems concerning workability since the impact stress measured at the pile head was approximately 22000 kN/m2 and the buckling stress of the steel-pipe pile of SKK400, t=13 mm was approximately 30000 kN/m 2.

Figure 4 presents the relationship between the measured im-pact stress in the depth direction at the position of percussion for-finish andthe residual stress on the pile body after being left un-touched for a fixed period of (15 days). It also shows the standard construction state in which the impact stress is transmit ted almost linearly in the pile body after impact was applied to the pile.

Fig. 4. Impact steel of steel-pipe piles.

In addition, the residual stress on the pile body, which was at first a concern, was generally small, although there was a ten-dency for slight tensile stress to be generated at the pile heads against the initial value at the time of the lining of piles. It was presumed that this phenomenon occurred due to elastic deforma-tion because stress was gradually released from the condition in which the pile heads were set in the bearing stratum with the compressive action of impacts on the pile bodies. However, since bending prevailed only at the pile heads in the case of the pile body stress generated by design external force, the body stress of the entire system found by integrating those values was within the range of allowable stress.

4.2 Impact Waveform and Stress Distribution of Long Steel-Pipe Sheet Piles

Figure 5 shows the impact waveform at the head of test sheet pile No. 1 (post-placement) at the time of percussion for finish, which was obtained as a result of the impact test for steel-pipe sheet piles, together with the results of the above-mentioned impact test of long steel-pipe piles of P2 pier. According to the figure, the time phase of the peak after impact, rebound at the cushion (peak of the second wave) and the attenuation conditions of im-pact stress nearly corresponded between the steel-pipe sheet piles and single piles, although the maximum stress values differed due to the difference in specifications and hydraulic hammer weight. It was therefore determined that the construction and intrusion properties were similar to those of steel-pipe single piles under buckling stress.

Figure 6 shows stress transmission in the depth direction at the time of installation of test steel-pipe sheet piles and the stress dis-tribution measured for adjoining sheet pile No. 2. The impact stress on the bodies of test steel-pipe sheet piles were attenuated in the depth direction along the soil boring log, and workability for steel-pipe sheet piles was similar to that of single piles in the same manner as in the impact waveform at the head. Stress dis-tribution of approximately 200 to 400 kN/m 2 to adjoining piles was, however, found to develop nearly linearly in the depth di-rection, and these distribution values for both of the adjoining sheet piles were approximately 5 to 10% of the stress on the bod-ies of the test steel-pipe sheet piles. Although it is difficult to calculate the vertical bearing capacity of these test steel-pipe sheet piles and adjoining sheet piles from this phenomenon, the considerable effect of coupling pipes at the time of installation of steel-pipe sheet piles was confirmed.

192

Fig. 5. Head impact waveforms of steel-pipe sheet piles and sin-gle steel-pipe piles.

Fig. 6. Impact stress of test steel-pipe sheet piles and adjoining sheet piles.

5 STATIC VERTICAL BEARING OF LONG STEEL-PIPE PILES

Figure 7 shows the axial force distribution on the pile bodies in the depth direction against vertical load P, which was obtained by the vertical loading test of test steel-pipe pile No. 1, and Fig. 8 illustrates the relationship between load Log P and vertical set-tlement at the pile head LogS. The axial force was vertically transmitted to the soft stratum approximately 40 m in depth with-out significant development of skin friction. Following this, rela-tively high skin friction was secured linearly toward the pile head. The relationship between load and settlement was nearly linear without a clear point of inflection equivalent to yield and ultimate bearing capacity, although the displacement had a ten-dency to slightly incline at the final point of planned maximum load P = 6600 kN.

The evaluation method is not necessarily appropriate for this case. The ultimate bearing capacity Ru2 of the steel-pipe piles in this vertical loading test was therefore determined using Uzu�s method 6) shown in the exponent model of Eq. (5).

P = Ru2 1 exp( S / So)m (5)

where Ru2: ultimate bearing capacity (kN), P=load on pile head (kN), m=displacement exponent, S=settlement (mm) and So=standard settlement (mm).

Fig. 7. Axial force distribution of test piles in the vertical Load-ing test.

Fig. 8. Relationship between the test load and settlement in the vertical loading test.

As a result, the ultimate bearing capacity was estimated to be Rup2 = 9100 kN, and it was believed that design vertical ultimate bearing capacity Rus was secured in a non-excessive range in the same manner as for dynamic bearing capacity Rupl. Regarding bearing capacity at the pile head Rp2, Rp2 = 2920 kN was calcu-lated using a similar method. Since end-bearing capacity qd2 = Rp2/A was qd2 = 10390 kN/m2, design value qd = 7640 kN/m2

was secured. In terms of skin friction, the maximum skin friction to the depth of approximately 40 m in the soft stratum was as low as f2' = 685 kN, which is equivalent to the present design friction f with N value = approx. 2 (f2/L U). However, a relatively high friction force exceeding the design value was displayed at depths of 40 m or deeper, making the total skin friction f2 = 6,180 kN.

The spring constant in the axial direction of piles Kv = 150000 kN/m, which was obtained from the secant gradient of loaded weight and settlement (10-mm settlement load), was nearly equivalent to the design value.

6 COUPLING PILE RESISTANCE OF LONG STEEL-PIPE SHEET PILES

In the dynamic loading test of long steel-pipe sheet piles, it was necessary to clarify the difference in impact force and total resis-tance of interlocking sheet piles when only one steel-pipe pile (No. 3) was installed and two sheet piles (No. 4 and 5 ) were

193

Fig. 9. Dynamic resistance as determined using the CASE method.

Fig. 10. Upward waves of single and interlocking sheet piles.

installed concurrently to calculate the resistance of coupling pipes at the time of construction of steel-pipe sheet piles.

Figure 9 presents the relationship between the impact force generated at the heads of two single sheet piles and one interlock-ing sheet pile by changing the dropping height of the hydraulic hammer and the dynamic resistance obtained using the CASEmethod. According to the figure, the dynamic maximum resis-tance or total resistance at the head, which is the total for the dy-namic and static resistance components of the value of the CASE method in the impact test, was approximately 8000 kN even for single sheet piles and was lower than design ultimate bearing ca-pacity Rus = 9300 kN. It was therefore presumed that the impact force was insufficient for obtaining the maximum development of the end ground resistance. As can be seen in the figure, however, the dynamic resistance of single sheet piles was approximately 2000 kN higher than that of interlocking sheet piles at the percus-sion for finish. The same tendency was also found in measured values of maximum acceleration and strain, and it was presumed that this phenomenon was caused by greater vibrations generated on single sheet piles due to the difference in body weight and coupling pipe resistance even when the same impact force was applied.

The difference of 2000 kN in maximum resistance between single and interlocking sheet piles in this study was thought to imply the resistance of coupling pipes between interlocking sheet piles. This value was equivalent to approximately 25% of the to-tal resistance (2000 kN / 8000 kN), and was greater than the

stress distribution through coupling pipes, which was obtained through the impact test. This also implies that the competition of coupling pipes or friction resistance should not be ignored when installing steel-pipe sheet piles.

Figure 10 illustrates the relationship between the upward waves of single sheet pile No. 3 and interlocking sheet piles No. 4and 5. In the dynamic loading test, downward waves are simply impact input waves generated by the hammer, and calculation of the vertical bearing capacity of steel-pipe sheet piles is made pos-sible by examining the upward waves, which were found as a re-sult of the dynamic loading test, including the elements of end-bearing capacity, skin resistance and resistance generated at joints. The figure shows that the impact of steel-pipe sheet piles began at approximately 7 ms and that the upward waves of the pile ends returned to the pile heads at approximately 30 ms. The length of the underlined portions during this time shows the re-turn time (2L/c) for the impact wave to be transmitted through the pile body. If these underlined portions are regarded as the body of a steel-pipe sheet pile lying sideways, it means that the upward wave shown above it indicates the changes of the sheet pile in the depth direction.

The time period between 7 and 30 ms is therefore applied to the sheet pile head and refers to upward waves caused by sheet pile skin resistance B and resistance C generated at the joint, and the time at 30 ms or later is applied to upward waves generated by end-bearing capacity A. According to the figure, the upward wave obtained in this dynamic loading test was attenuated before the wave motion of impact at the head reached the end of the sheet pile because the peak of the upward wave (approx. 25 ms) occurs before 30 ms, which is at the end of the sheet pile. This phenomenon probably occurred because the majority of the im-pact force on the sheet pile head was dispersed to the coupling pipe resistance and transmission of the impact force became in-sufficient, although driving itself was possible, as the competition length of coupling pipes gradually increased with the driving of steel-pipe sheet piles.

In the case of dynamic resistance P1 of single sheet piles and P2 of interlocking sheet piles, it is therefore difficult to calculate the true end-bearing capacity itself with a small impact force on long sheet piles by setting end-bearing capacity A, skin resistance B and resistance generated at joint C of steel-pipe sheet piles ac-cording to the initial principal of examination. Thus, assuming that the transmission of the impact force to the end is very small and the end-bearing capacity against the impact force in this test is A 0, the relationship between Eqs. (6) and (7) can be ob-tained, respectively.

P1 B C (6) P2 2B C (7)

Furthermore, Fig. 10 shows that the upward waves of single sheet pile No. 3 and interlocking sheet piles No. 4 and 5 nearly corresponded with each other, indicating the relationship P1 P2 within the time of the return wave generated in steel-pipe sheet piles. In other words, the skin resistance of the ground B

0 (from 2B = B), or the skin resistance of the sheet piles, was not developed during installation of steel-pipe sheet piles, and this was nearly consistent with the soft soil boring log of the ground. Although the dynamic maximum resistance (approx. 8000 kN) of single steel-pipe sheet piles obtained using the CASE method in this study was not necessarily the ultimate bear-ing capacity of steel-pipe sheet piles under this system condition, the maximum resistance of 2000 kN calculated from Fig. 9 could

194

be considered as resistance C generated at the joint. This was for the most part consistent with the difference in dynamic resistance between single and interlocking sheet piles found from Fig. 10. This result shows a relatively high resistance of coupling pipes at the time of installation of steel-pipe sheet piles, in the same way as the tendency of stress distribution to adjoining sheet piles ob-tained in the impact test. Although this resistance of coupling pipes would be very disadvantageous for the management of in-stallation of steel-pipe sheet piles, it is thought to be an effective element of closing to establish an integrated closed foundation and ensure the required rigidity in the end.

7 CONCLUSIONS

From the results of a series of construction management tests conducted for long steel-pipe piles (L1 = 65 m) and sheet piles (L2 = 61.5 m), including, impact, vertical loading and dynamic loading tests, the following conclusions were obtained as assess-ments of the vertical support system of deep foundations in rela-tion to workability:

1. From the impact test for long steel-pipe piles, it was found that verification of workability of long steel-pipe piles was possible, the normal impact basic waveform was developed even for long piles and residual impact stress in the depth direction was insufficient.

2. From the vertical loading test of long steel-pipe piles, the pile skin friction, end-bearing capacity and axial spring constant were directly evaluated and the vertical support system of long piles was verified.

3. From the impact waveform and stress obtained as a result of the impact test of long steel-pipe sheet piles, it was con-firmed that steel-pipe sheet piles exhibited workability equivalent to that of single piles. As a result, stress distri-bution of 5 to 10% through coupling piles was observed in the depth direction of adjoining sheet piles at the time of installation of steel-pipe sheet piles.

4. As a result of the dynamic loading test of single steel-pipe sheet piles and interlocking steel-pipe sheet piles using special fixtures, the resistance properties of the bodies of coupling pipes were clarified through the analysis of up-

ward and downward waves using the CASE method. In the dynamic loading test for the long steel-pipe sheet piles in question, a relatively large resistance of coupling pipes was found, probably due to the effect of competition of cou-pling pipes at the time of construction.

From the above conclusions, it is believed that the support system including construction management of steel-pipe sheet piles was roughly verified, although it was previously known that there were many unclear matters concerning the evaluation of vertical bearing capacity related to workability, such as driving methods for long piles and selection of hammers, as well as the resistance of coupling pipes. The use of the dynamic loading test for steel-pipe sheet piles in this study, in particular, led to new possible methods of calculating the bearing capacity of steel-pipe sheet piles 7), 8) taking the effect of coupling pipes into considera-tion. This also indicated the possibility of ensuring the required bearing capacity through appropriate construction management even for long foundations.

REFERENCES

Japan Road Association 2002. Specification for Highway Bridges and Instruction Manual IV : 434-465.

Japan Road Association. 1997. Guidebook for Design and Con-struction of Steel-Pipe Sheet Pile Foundations : 177-257.

Japanese Geotechnical Society 2002. Method for Horizontal Loading Test of Piles and Instruction Manual : 223-244.

Sakai T. 1990. Japan Society for Civil Engineers, No. 424/III-14 :75 � 83.

Japanese Geotechnical Society 1989. Proceedings of the Sympo-sium on Driving Performance of Piles and Application of the Kinematic Wave Theory to Piles.

Uzu, F. 1978. Pile loading test arrangement method, Collection of lectures . Proceedings 31st Soil Engineering Society Meeting.

Kimura, Isobe, J. & Arop Too, K. 2002. Proceedings of the 37th

Geotechnical Research Presentation : 1407 � 1408. Kazama, Oki, Okubo & Nambu. 2002. Proceedings of the 37th

Geotechnical Research Presentation : 714 � 720.


195

Analysis of Ultimate Bearing Capacity on Composite Layered Soil

S. OhtsukaDept. of Civil and Environmental Eng., Nagaoka University of Tech., Nagaoka, Niigata 940-2188, Japan [email protected]

A. Husna Dept. of Energy and Environmental Science, Nagaoka University of Tech., Nagaoka, Niigata 940-2188, Japan [email protected]

Abstract: A study has been carried out for ultimate bearing capacity of footing on composite layered soil where sandy soil overlies clayey soil. It is, however, difficult to estimate the ultimate bearing capacity due to complicate failure mode. This paper took into ac-count two aspects of difficulty in assessment. One was a computational method and the other, a determination on soil constants. The rigid-plastic finite element method (RPFEM) was employed to estimate the ultimate bearing capacity of composite layered soil bytaking into account the dilation property. The soil constant as angle of shear resistance for sand was obtained by the inversed analysis of centrifuge tests on ultimate bearing capacity for uniform Toyoura sand. Using inversed constants, the direct analysis of ultimate bearing capacity was conducted for composite layered soil on two-dimensional ( 2-D) and three-dimensional (3-D) computations. The obtained results were compared with those of centrifuge tests and conventional methods. It exhibited RPFEM favorably simulated the ultimate bearing capacity and the complicated failure modes of composite layered soil even with difference in boundary condition.

1 INTRODUCTION

In ultimate bearing capacity assessment, the soil profile is gener-ally not homogeneous and it affects the result greatly. Numbers of study have conducted for ultimate bearing capacity of compos-ite layered soil (Yamaguchi, 1963; Hanna & Meyerhof, 1980; Kraft & Helfrich, 1982; Michalowski & Shi, 1995; Burd & Frydman, 1997; Kenny & Andrawes, 1997; etc) and experimental model tests (Craig & Chua, 1990; Okamura et al., 1993, 1997; etc). Those may give a proper solution for real construction pro-jects; however, the conventional methods inevitably introduce assumptions on failure mechanism of soil prior to analysis. Since the failure mode of composite layered soil is more complicated and different from that of uniform soil, the applicability of con-ventional methods naturally becomes limited. On the other hand, Tamura (1990) developed rigid-plastic analysis (RPFEM) to solve various problems in geotechnical engineering. The advan-tages of this method are the failure mode of earth structure is determined as a result of analysis and pre-information on feasible failure mode is not necessary different from conventional formu-las. Ohtsuka & Husna (2003) carried out an investigation on the applicability of RPFEM for ultimate bearing capacity assessment and gave good estimation for sandy soil.

This paper is based on the advanced computation of RPFEM to investigate the inversed angle of shear resistance for founda-tion designs on composite layered soil, sand overlying soft clay. An inverse analysis of centrifuge test is conducted for ultimate bearing capacity of Toyoura sand to obtain the soil constant as angle of shear resistance. The applicability of inversed constants is led to compute ultimate bearing capacity on composite layered soil for various conditions and it is compared not only with cen-trifuge tests but also those well-known conventional methods. A three-dimensional (3-D) calculation of ultimate bearing capacity for square footing is also given to demonstrate the relevance and the wide applicability of the employed method.

2 INVERSE ANALYSIS FOR SOIL CONSTANT

It is generally known there are various factors affect the shear strength of sandy soil, such as stress level, anisotropy, and strain

softening. These effects express the material constant of sandy soil as angle of shear resistance widely varies in the ground de-pending on stress, shearing pattern and strain softening. On the contrary, the conventional design of ultimate bearing capacity formula assumes a uniform material constant for soil. It is of course a simple assumption, however the applicability of this assumption has not been made clear yet. The question is, fur-thermore, the way to determine the design constant for conven-tional method.

Okamura et al. (1993, 1997) performed centrifuge tests for shallow footing on dense Toyoura sand and composite sand over-lying soft clay. Toyoura sand is the standard sand in Japan and its physical property has been studied in detail as listed in Table 1. An empirical equation to express the angles of shear resistance and dilation angle d of Toyoura sand was given by Tatsuoka et al. (1986) as d = - 30o. It means the dilation angle is naturally obtained with the angle of shear resistance.

Table 1. Physical property of Toyoura sand. Gs D50 (mm) Uc emax emin

2.640 0.190 1.560 0.973 0.609 (Source: Okamura et al., 1993)

As a most realistic approach to real problems, inverse analysis is carried out for axis-symmetric condition. The angle of shear resistance for Toyoura sand is computed with the assumption of Tatsuoka�s equation and uniform soil constant in the ground. In computation, the unit weight and the void ratio of soil are 15.6kN/m3 and 0.65, respectively. The radius of footing is set in the range of B=0.03m to 2.264m from centrifuge test. The inverse analysis is conducted only for peak state since there was small range in bearing capacity between peak and residual loads of Fig. 1 except for case of 1g-model test (Okamura et al., 1993). Figure 2 presents the calculated inversed constant with dif-ference in footing radius. The angle of shear resistance is ob-tained in the range of 34.5o - 36.0o. It drops with the increase in footing radius and lower about 5.0o � 10.0o than those obtained in triaxial tests ( =40o-45o). The difference in angle of shear resis-tance evidently express the influence of stress level, anisotropy and progressive failure caused by strain softening in the analysis. The detail discussion of the inversed analysis for soil constant

196

Fig. 1. Normalized load intensity of axis-symmetric footing.

0 0.5 1 1.5 2 2.510

20

30

40

Radius of footing (m)

Fig. 2. Inverse analysis for angle of shear resistance.

can be found in Ohtsuka & Husna (2003). Some parts are re-viewed only for easy understanding of current paper. The ob-tained angle of shear resistance differs for footing radius and geometry, which attributes to the scale-effect and due to the stress distribution in the ground different by failure mode. It is obvious that the mobilized angle of6 shear resistance is affected by various factors as stress level, anisotropy and progressive failure caused by strain softening. However, the degree in effect of each factor on inversed angle of shear resistance is not made clear in this paper. The angle of shear resistance decreases with the increase in footing radius and it is lower than those obtained in testing methods. Taking into consideration the effects of ani-sotropy and strain softening are nearly same for footing radius, the change in angle of shear resistance for footing radius might express the stress level effect mainly. In fact, the inversed angle of shear resistance does not have physical meaning directly, but the mean value to match the ultimate bearing capacity exactly.

These results are based on advanced computation to employ the non-associated flow rule and different from conventional formulas, but give valuable information for design especially on the relationship between the inversed and original soil constants. By accumulating the case studies on other soils, a general method to determine the design constant for ultimate bearing capacity might be constituted.

3 2-D ANALYSIS OF ULTIMATE BEARING CAPACITY ON COMPOSITE LAYERED SOIL

3.1 Analysis Procedures and Result

The direct analysis of inverse constant for ultimate bearing ca-pacity on two-layers foundation soil system is conducted for the cases where the thickness of sand layer, H, is comparable to the footing radius, B, and in all cases the ground surface and inter-

face between the two soil layers are horizontal. In computation, the fully drainage condition is set for sand and the undrained condition for clay. Footing radius is taken as 1.5m and 3.0m from centrifuge test and the effective unit weight of soil is 9.74kN/m3.The angle of shear resistance of sand layer is put as 34.5o from the inversed constant by taking into account the footing radius. The thickness of sand layer is changed by the ratio of sand layer thickness to footing radius from 0.5 to 2.0. The shear strength of clay is predicted with the strength increase ratio cu/p�. The mean effective stress p� is estimated by employing Jaky�s formula for earth pressure coefficient.

In general, conventional methods to calculate ultimate bearing capacity of composite layered soil are divided by assuming fail-ure mechanism beneath the footing. First group is called Load-spread model. It assumes that the load spreads in sand layer and works on the surface of clay foundation. Ultimate bearing capac-ity is assessed for spread load by estimating the shear resistance of clay layer only. The load-spread mechanism within the sand layer is modelled by assuming the zone defined by lines inclined at angle of side surface of the sand block to the vertical, as shown in Fig. 3(a). Load from footing is assumed to be distrib-uted uniformly over the width B� at the base of sand layer, where

. The chosen value of has an important in-fluence on the ultimate bearing capacity calculation, however, it is not clear how this value should be selected. Although the side surface angle of the sand block to the vertical is influenced by the strength of the sand, in practice, (Yamaguchi, 1963) and (Kraft & Helfrich, 1983) are generally adopted. The ultimate bearing capacity of axis-symmetric footing qu is estimated using the expression :

(1)

The second group, Punching-shear model assumes the direct shear of sand along a vertical plane beneath each edge of the footing, as shown in Fig. 3(b). Sandy soil beneath the footing is pushed downwards into clay and ultimate bearing capacity is exerted with shear resistance in sand and clay layers. The charts of Hanna & Meyerhof (1980) were not given in non-dimensional form, however, and so are proper only for the certain values of sand unit weight and layer thickness. First chart was given as function of the punching shear coefficient Ks and the angle of shear resistance �. While another one was function of the clay shear strength cu and the ratio of / �. The ultimate bearing ca-pacity of axis-symmetric footing qu is computed as :

(2)

The obtained ultimate bearing capacity of each method is drawn in Fig. 4. Footing radius is taken as B=1.5m and 3.0m. The ratio of footing radius to thickness of sand layer is arranged from H/B=0.5 to 2.0. Generally, the calculated ultimate bearing capac-ity with RPFEM is in good conformity with those of centrifuge tests even the boundary condition was different from the inverse analysis. It strongly implies the applicability of direct analysis with the inversed soil constant for ultimate bearing capacity analysis. Since the failure mode of composite layered soil is more complicate and different from that of uniform sandy soil, it seems difficult to assess the ultimate bearing capacity with the inversed angle of shear resistance prior to computation. However, the obtained result suggests the selection of soil constant is insensi-tive for ultimate bearing capacity assessment of composite lay-ered soil by taking into account the footing radius, and the wide applicability of inversed constant could be expected. In conventional methods, Load-spread model gives good estimation for centrifuge test than that of Punching-shear model,however underestimates the ultimate bearing capacity for com-

axis-symmetric footing =36.0o - 34.5o

197

B

Q

po'

HSand : ' '

Clay : cu

cu Nc + po'

B'

B

Q

po'

H

Clay : cu

cu Nc + po'

Ks : punching shear coefficient

zSand : ' '

(p0' + z') Ks

a. Load-spread model b. Punching-shear model Fig. 3. Assumptions on failure mechanisms in conventional methods.

posite layered soil in case of H/B>1.5. Conventional methods usually employ the assumption on failure mode, it causes the accuracy of conventional method highly depends on the applica-bility of assumed failure mode to the actual problem. Notably, RPFEM does not require the failure mode prior to computation.

3.2 Failure Modes and Soil Movement

The patterns of soil movement at failure and the development of plastic zones in soil under the footing are in interest. Then, the failure mechanism with the variation in ratio of sand layer thick-ness to footing radius H/B is discussed as shown in Fig. 5. A punching-shear mode of sand layer is observed for thin sand layer and it gradually changes to a general-shearing mode for thick sand layer on Fig. 5(d). It clearly exhibits that the failure pattern is largely affected by the ratio of H/B. It is apparent that the sand block beneath the footing is pushed into clay layer up to H/B=1 and the shear zone in sand layer is restricted along the side of sand block as obtained in centrifuge test. For H/B=2.0,failure occurs entirely within sand layer and a general-shearing mode takes place. It is totally same as obtained failure mode for

uniform sand of Fig. 5(a). This denotes that for the greater thick-ness of top layer (H/B 2.0), the computation of ultimate bearing capacity on composite layered soil, sand overlying soft clay, could be generated as uniform sandy ground. The failure mode of RPFEM computation is determined as a consequence of computation and it is contrary to the conventional methods. In case of Fig. 4(a), Load-spread model gives relatively good estimation on the ultimate bearing capacity, however, it only takes into account the failure of clay layer. It implies incon-sistency in the assumption and the result of Load-spread model in Fig. 5 and it naturally suggests RPFEM results are more rational. The obtained results express both the applicability of the inversed soil constant and the employed numerical procedure to assess ultimate bearing capacity for composite layered soil. Although various factors as stress level, anisotropy and strain softening greatly affect the angle of shear resistance for case of sandy soil, RPFEM with the inversed angle of shear resistance could well estimate the ultimate bearing capacity. The detailed discussion on the relationship between the physical property and the inversed one is necessary to develop a general method for determining the design constants.

0 1 2 3 40

1000

2000

Thickness of sand layer (m) Thickness of sand layer (m)

a. B = 1.5m

0 2 4 6 8

2-D RPFEM Punching-shear model Load-spread model Centrifuge test

b. B = 3.0m

Fig. 4. Computational result of direct analysis on composite layered soil.

Y

0 .18

YY

Y

a. Uniform sand d. H/B=2.0c. H/B=1.0b. H/B=0.5

Sand

Sand

Clay

B/2 B/2 B/2

Sand

Clay

B/2

Sand

Clay

Fig. 5. Failure modes dependent of sand layer thickness in two-dimensional computation.

198

0 2 4 6 80

1000

2000

Thickness of sand layer (m)

Centrifuge test 2-D RPFEM 3-D RPFEM

Fig. 6. Ultimate bearing capacity of 3-D computation.

4 3-D ANALYSIS OF ULTIMATE BEARING CAPACITY ON COMPOSITE LAYERED SOIL

Direct analysis of inversed constant is conducted for 3-D (three-dimensional) RPFEM of square footing resting on sand layer overlying soft clay. The condition of computation as soil con-stants and footing in case of 3-D computation is set similar with that in case of two-dimensional (2-D). A direct comparison of the obtained results could be done. It is noted that the angle of shear resistance for sand layer is determined by considering the footing width. Figure 6 illustrates the obtained ultimate bearing capacity of 3-D computation. It is given in comparison with 2-D computa-tion and centrifuge test.

The 3-D ultimate bearing capacity matches well with those of centrifuge tests and 2-D results even if the mesh employed in 3-DRPFEM is coarser than those of 2-D cases. This indicates the effect of geometry in footing does not affect the ultimate bearing capacity so much. The obtained failure mode of square footing is presented in Fig. 7. The generated mesh of 3-D analysis shows the realistic failure modes for square footing. It is different from that of 2-D RPFEM and it indicates a good idea for employing the 3-D computation, especially to assess the ultimate bearing capacity of rectangular footing.

5 CONCLUSIONS

The two- and three-dimensional computations of axis-symmetric and square footings for composite layered soil, sand overlying soft clay, were carried out to investigate the ultimate bearing capacity and its failure mechanism of foundation with the use of inversed soil constant. The result of numerical analysis was com-pared with centrifuge test and some of the available theoretical solutions to calculate ultimate bearing capacity for composite layered soil. The followings were obtained in this study. 1. Ultimate bearing capacity was well simulated by direct

analysis with RPFEM for various thickness of sand layer un-derlain clay. In addition, RPFEM also properly modeled the complicated failure modes of composite layered soil. It obvi-ously exhibited the wide applicability of the inversed pa-rameter to ultimate bearing capacity assessment of composite layered soil for various boundary value problems.

2. The applicability of conventional methods was found to de-pend highly on agreement of the assumed failure mode with the actual behavior. No need of assumption for feasible fail-ure modes is great advantage of RPFEM to conventional methods. Employment of non-associated flow rule for consti-tutive equation was also noted. The dilation angle also seems to have a significant influence on the magnitude of ultimate bearing capacity for composite layered soil.

X

Y

0.5

1.

1.5

2.

2.5

3.

3.5

4.

4.5

5.

5.5

6.

6.5

7.

X

Y

0. 5

1.

1. 5

2.

2.5

3.

3. 5

4.

4. 5

5.

5.5

6.

6. 5

7.

b. H/B = 1.0

a. H/B = 0.5

Sand

Clay

Sand

Clay

B/2

B/2

Fig. 7. Failure modes of 3-D computation.

3. The applicability of 3-D computation should be examined in comparison with the experimental results. It was especially in the assessment of ultimate bearing capacity for rectangular footing.

REFERENCES

Hanna, A.M. & Meyerhof, G.G. 1980. Design Charts for Ulti-mate Bearing Capacity of Foundations on Sand Overlying Soft Clay. Canadian Geotechnical Journal 17: 300-303.

Ohtsuka, S. & Husna, A. 2003. Inverse Analysis of Material Constants for Ultimate Bearing Capacity of Sandy Soil. Proc. of International Symposium on Shallow Foundation 1: 397-404.

Okamura, M., Takemura, J. & Kimura, T. 1993. A Study on Bearing Capacities of Shallow Footings on Sand. Proc. of JSCE 463(3): 85-94 (in Japanese).

Okamura, M., Takemura, J. & Kimura, T. 1997. Centrifuge Model Tests on Bearing Capacity and Deformation of Sand Layer Overlying Clay. Soils and Foundations 37(1): 73-87.

Tamura, T. 1990. Rigid-Plastic Finite Element Method in Geo-technical Engineering. Computational Plasticity, Current Japanese Material Research 7: 135-164.

Tatsuoka, M., Sakamoto, M., Kawamura T. & Fukushima, S. 1986. Strength and Deformation Characteristics of Sand in Plane Strain Compression at Extremely Low Pressure. Soils and Foundations 26(1): 65-84.

Yamaguchi, H. 1963. Practical Formula of Bearing Value for Two Layered Ground. Proceedings of the 2nd ARCSMFE 1: 176-180.


199

Fig. 1. Outline of sheet-pile foundation.

Table 1. Conditions of modeled ground.

Container size (W*H*D) 2000mm*1000mm*600mm Ground size (W*H*D) 2000mm*580mm*600mm

Material of ground Dry Toyoura sand Relative density Dr Dr = 90% or 60% Dry unit weight d d = 16.2 kN/m3 or 15.1 kN/m3

Lubricated layer Rubber membrane(t=0.2mm) with Grease(10 m)

A Series of Static Loading Tests of Modeled Sheet-Pile Foundation Combining Footing with Sheet-Piles on Sand

H. Nishioka, M. Koda & O. Murata Railway Technical Research Institute, 2-8-38, Hikari-cho, Kokubunji-shi, Tokyo, 185-8540, Japan [email protected]

Abstract: The authors have proposed the sheet-pile foundation as a new reasonable foundation structure. This foundation can be con-structed at a lower cost than the pile foundation, and be used more widely than the shallow foundation. In this research, in order to grasp the fundamental characteristics of the sheet-pile foundation, a series of static loading tests are conducted on a model of the shal-low foundation and sheet-pile foundation. It is shown that the sheet-pile foundation has excellent performance against seismic force. In addition, the resistance mechanism of the sheet-pile foundation is discussed based on the displacement of the ground which is com-puted by an Image Processing System, to show that this mechanism restrains deformation of the ground, prevents local failure of the ground, and resists horizontal force in a large area of the ground.

1 INTRODUCTION

Recently, construction work is increasing in densely populated urban areas in Japan. For example, in order to ease traffic con-gestion, railroads running through urban areas are re-laid on ele-vated bridges, which are constructed close to existing structures. Moreover, the space for construction work is often very small. On the other hand, in such construction work in urban areas, it is required to cut down costs and reduce noise and vibration. The disposal of surplus soil on construction work must also be taken into consideration.

Furthermore, shallow foundations and pile foundations are mainly used for structures in urban areas. Although the shallow foundation is one of the cheapest forms of foundation structure, it is constructed only on the ground of good quality, such as the dense sand whose N-value is higher than 30 (Railway Technical Research Institute 1999, 2000). Since the quantity to dig is com-paratively small, however noise, vibration or the surplus soil does not pose a serious problem. On the other hand, the pile founda-tion applies to various ground conditions and can also be con-structed on the soft ground. However, the pile foundation has the problems of noise, vibration and surplus soil, when boreholes for piles are excavated.

With the above in background, "the sheet-pile foundation" which combines the footing and sheet-piles has been proposed as a new foundation form (Punrattanasin et al., 2002, 2003a, b, Koda et al., 2003). Since the sheet-pile foundation reinforces the ground with sheet-piles, its bearing capacity and horizontal seis-mic resistance are higher than those of the shallow foundation. Since its applicability is higher than that of the shallow founda-tion, it can be used on the loose sandy ground to which the pile foundation has been applied until now, for example. Its con-struction cost is almost the same as that of the shallow foundation and lower than that of the pile foundation. On the other hand, since pile excavation is not necessary, it can avoid various prob-lems of pile foundation mentioned above. Figure 1 shows an out-line of the sheet-pile foundation compared with the shallow foundation and the pile foundation.

In this research, a series of static loading tests were carried out on a model of shallow foundation and sheet-pile foundation for the purpose of grasping the performance of the sheet-pile founda-tion.

2 OUTLINE OF STATIC LOADING TESTS

This chapter explains an outline of static loading tests including the size of the model and loading conditions. A modeled ground was made from dry Toyoura sand in a sand container shown in Fig. 2. It was a two-dimensional model in plane strain conditions. The sand container was a box whose height, depth and width were 1,000mm, 600mm and 2,000mm, respectively. Planes of the front and backside of the sand container were made of a transparent acrylic plate to allow observation of the deformation

200

Fig. 2. Picture of container and ground.

Fig. 3. Picture of Sheet-Pile Model C (concavo-convex form).

Table 2. Specifications of modeled sheet-piles.

Prototype Model A Model B Model C

Material Steel Plate Copper Plate

Thickness t (mm) 15.5 1.0 0.2 0.2

Height of concavo-convex h (mm) 340 Not concavo-convex

form 1.4

Width W (m) 4.8 0.6

Area As (m2) 0.11 6.00*10 -4 1.20*10-4 1.35*10-4

Young's modulus E (kN/mm2) 200 110

Geometrical moment of inertia I (m4) 1.23*10-3 5.00*10-11 4.00*10-13 4.29*10-11

Characteristic value of sheet-pile

0.689 33.41 111.7 34.72

Length L (mm) 4800 100

L 3.31 3.34 11.2 3.47

Table 3. Cases of vertical loading tests.

Case Density of

ground Foundation Form

Sheet-Pile models

V-D-1 Shallow foundation V-D-2

Dr=90% Sheet-pile foundation Model A

V-L-1 Shallow foundation V-L-2

Dr=60% Sheet-pile foundation Model C

of the ground in it. The sand container was reinforced with steel frames in order to hold plane strain conditions. The relative den-sity Dr of the modeled ground was controlled by the height of the sand hopper to 90% or 60%. The height of the ground was 580mm. In order to reduce friction between the acrylic plate and sand, before making the modeled ground, rubber membranes were pasted on the acrylic plates with grease. Moreover, target points were marked on the rubber membrane, and their displace-ment was computed from photographs taken with a digital cam-era through the acrylic plate by an Image Processing System (Watanabe & Tateyama, 2003). Table 1 summarizes the condi-tions of the modeled ground.

The modeled footing was made of aluminum with a width of 100mm. In order to raise the rigidity of the footing, the height of the aluminum block was set to 100mm, but its ratio of width to height was larger than that of actual footings.

The model of sheet-pile was made of copper plates. Its length L that was penetrating into the sand was 100mm, the same as the width of a footing model. There were three sheet-pile models whose cross-sections were different as shown in Table 2. Models A and B were produced with flat plates with thicknesses of 1mm and 0.2mm. Model C was made of flat plates with a thickness of 0.2mm and manufactured to a concavo-convex form whose height is 1.4mm. Figure 3 is a picture of the sheet-pile Model C. The value of L shown in Eq. 1 of Models A and C was the same grade as that of an actual sheet-pile. In addition, the sheet-pile model was put into the model ground.

LEIDkL h4 4 (1)

where : characteristic value of pile (1/m), kh: coefficient of hori-zontal subgrade reaction (kN/m3), D: width of sheet-pile (m), EI:flexural rigidity of sheet-pile (kNm2) and L: length of sheet-pile (m).

3 VERTICAL LOADING TESTS

3.1 Outline of Vertical Loading Tests

The purpose of vertical loading tests was to check the ground conditions and the improvement effect of the bearing capacity of the sheet-pile foundation. Tests were carried out for four cases of different ground densities and foundation forms shown in Table 2. A screw jack gave vertical displacement to the model of the shal-low foundation and the sheet-pile foundation placed on the dense ground whose relative density Dr was 90% and the loose ground whose relative density Dr was 60%. The screw jack was jointed to the center on the upper surface of the footing model (alumi-num block) with a pin in the direction of the long axis of the footing model so that the moment could be removed. An outline of loading equipment is shown in Fig. 4. The displacement was increased monotonically at the rate of 1 mm/min. In addition, tests of the sheet-pile foundation were started from the position where the sheet-pile tip had not touched the sands. The dis-placement was also increased and the sheet-pile was put into the sands, until the bottom of footing touched the sands. Further-more, the experiment on the modeled sheet-pile foundation was carried out continuously.

201

0

2

4

6

8

10

12

14

16

18

20

-10 0 10 20 30 40

Displacement (mm)

Case-V-D-1

Case-V-D-2

a) Cases on Dr=90%

0

1

2

3

4

5

6

7

8

9

10

-10 -5 0 5 10 15 20Displacement (mm)

Case-V-L-1

Case-V-L-2

b) Cases on Dr=60% Fig. 5. Relationship between vertical load and displacement.

Table 4. Bearing Capacity.

Case Dr Foundation Form Bearing Capacity V-D-1 Shallow foundation 13.1kN V-D-2 90% Sheet-pile foundation 15.4kN V-L-1 Shallow foundation 4.7kN V-L-2 60% Sheet-pile foundation 8.0kN

Fig. 4. Equipment for vertical loading tests.

3.2 Results and Discussions

The relation between vertical load and displacement of all cases is shown in Fig. 5. However, the displacement when the footing bottom contacts the sands was set to zero. Moreover, the bearing capacity was calculated as the maximum load measured at the time when the displacement increased to 10mm which was 10% of the footing width as summarized in Table 4.

First, the difference in the density of ground and the relation of bearing capacity of the shallow foundation were considered. Although it was natural, it was clear that the bearing capacity of the Dr = 90% case was about three times higher than that of the Dr = 60% case. Moreover, the rigidity at early stages of the Dr = 90% case was larger than that of the Dr = 60% case. Moreover, in the Dr = 90% case, the load became the maximum by about7mm of displacement, fell sharply after that and became about70% of the maximum. On the other hand, in the Dr = 60% case, the load did not fall clearly. It turned out that the Dr = 90% case corresponded to typical dense sands and the Dr = 60% case to loose sands.

Next, the effect of the improvement on the shallow foundation of the sheet-pile foundation was considered. Regardless of the density, although the rigidity at early stages did not change, the bearing capacity increased. From the viewpoint of the incre-mental ratio, the effect of improvement of Dr = 60% case was higher, namely, the sheet-pile foundation was more effective on the loose ground. On the other hand, the increment of the bear-ing capacity of the sheet-pile foundation against that of the shal-low foundation was about 2.0 to 3.0 kN, which was larger than 0.5~1.0 kN. This was the penetration resistance when pushing a sheet-pile into the sands. This penetration resistance was the load that was measured when the displacement was zero mm just before the bottom of the footing touched the sand. Therefore, sheet-piles reinforced the ground. Consequently, it was thought that the reaction force at the bottom of footing became large. Moreover, the residual plastic deformation of the sheet-pile model was not observed after the experiment.

Figure 6 shows the displacement computed by the Image Processing System. The line in the figure is the locus of each target point until the vertical displacement of the footing model reaches 10mm.

In all cases, the target points of the ground in a certain domain under the footing model were moved in the downward vertical di-rection, and the target points of the ground of outside that domain were moved in the outward horizontal direction. However, the sizes of that domain differed in different cases, and in the case of the sheet-pile foundation, that domain spread to a deeper area. This means that sheet-piles restricted the horizontal dilation to-ward the outside of the ground under the footing. Therefore, as a result of image processing, it became clear that the mechanism to increase the bearing capacity of sheet-pile foundation depended on the reinforcement effect of the ground.

4 HORIZONTAL RECIPROCAL LOADING TESTS

4.1 Outline of Horizontal Reciprocal Loading Tests

An outline of horizontal reciprocal loading tests is shown in Fig. 7. The repetition displacement in the horizontal direction was given at the position similar to the bridge pier top, by keeping the vertical load constant.

202

Fig. 7. Outline of horizontal reciprocal loading test.

-20

-15

-10

-5

0

5

10

15

20

0 20 40 60 80 100 120 140 160Time (min)

History AHistory B

Fig. 8. Reciprocal history of Horizontal displacement .

Table 5. Cases of horizontal reciprocal loading tests.

Case Dr Foundation Form Sheet-Pile models History

H-D-1 Shallow foundation History A

H-D-2 90% Sheet-pile founda-tion Model B

H-L-1 Shallow foundation

H-L-2 60% Sheet-pile founda-tion Model B

History B

Table 6. Settlement of foundation. Case Dr Foundation Form Settlement

H-D-1 Shallow foundation 4.1mm H-D-2

90% Sheet-pile foundation 4.2mm

H-L-1 Shallow foundation 12.6mm H-L-2

60% Sheet-pile foundation 7.8mm

The vertical load was kept at 1.2kN, which was about 10% of the bearing capacity of the shallow foundation in the Dr = 90% ground by an air cylinder, because the ratio of the dead load to the bearing capacity of the shallow foundation was 10% in the situation of the design for railway structures in Japan. The hori-zontal displacement was given with a screw jack reciprocally at a height of 230mm from the footing model bottom that was similar to the bridge pier top. In this research, we named this position "the top of pier." Figure 8 shows the reciprocal history of hori-zontal displacement.

Four cases shown in Table 5 were tested for the purpose of grasping the seismic-resistance of the sheet-pile foundation. Mainly, the horizontal load at the top of pier, displacement of the footing (in the vertical, horizontal and rotational directions), re-action stress of the bottom of footing, and strain of the sheet-pile model were measured. a) Case-V-D-1 (The shallow foundation on Dr=90%)

b) Case-V-D-2 (The sheet-pile foundation on Dr=90%)

c) Case-V-L-1 (The shallow foundation on Dr=60%)

d) Case-V-L-2 (The sheet-pile foundation on Dr=60%)

Fig. 6. Displacement computed by an Image Processing Systemwhen vertical displacement reaches 10mm.

203

-0.4

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

0.4

-25 -20 -15 -10 -5 0 5 10 15 20 25

Horizontal displacement (mm)

Case-H-D-1

Case-H-D-2

a) Cases on Dr=90%

-0.4

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

0.4

-25 -20 -15 -10 -5 0 5 10 15 20 25

Horizontal displacement at top of pier (mm)

Case-H-L-1

Case-H-L-2

b) Cases on Dr=60%

Fig. 9. Historical curves of P- relationship.

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0 5 10 15 20


Case-H-D-1

Case-H-D-2

Case-H-L-1

Case-H-L-2

Fig. 10. Skeleton curves of P- relationship.

4.2 Results and Discussions

In this section, the settlement characteristic, which is the impor-tant function of a foundation structure, is described first. Table 6 summarizes the vertical displacement of footing, i.e. , the quan-tity of the settlement, at the time when the horizontal loading fin-ished. However, in this displacement, a part to be generated by the vertical load was removed. In the Dr = 60% case of loose sand, it was clear that the sheet-pile foundation reduced the quan-tity of settlement.

Next, as the resistance characteristic against the inertia force by earthquakes, the relations between the horizontal load P and the horizontal displacement at the top of pier are described. About the P- relation, the historical curve of each case is shown in Fig. 9, and a skeleton curve, which connects the turning point on each cycle, is shown in Fig. 10. Figures 9 & 10 show that the sheet-pile foundation has horizontal resistance higher than that of the shallow foundation, and the yield capacity and the secondary slope after the yield of sheet-pile foundation rose conspicuously. Additionally, since the loop of the history curve of the sheet-pile foundation was larger than that of the shallow foundation in Fig. 9, it was clear that the hysteresis damping of the sheet-pile foun-dation was larger than that of the shallow foundation.

Moreover, the residual plasticity deformation of the sheet-pile model was unobservable after the experiment. This is the same phenomenon as the one seen in vertical loading tests.

Figure 11 shows the displacement computed by an Image Processing System. The line in the Fig. 11 shows the locus of each target point until the turning point of the cycle whose hori-zontal displacement at the top of pier reaches 20mm.

First, two cases on the Dr = 90% dense sandy ground is de-scribed. About the shallow foundation, the ground deformation was observed only in a small local area under the edge of footing on the compressed side. However, since the floating of footing on the opposite side was observed, it was thought that intensity depended on the floating of footing, but not on the ground under the footing. On the other hand, about the sheet-pile foundation, the displacement of the domain enclosed with the sheet-pile oc-curred integrally. The center of the rotation was deeper than the footing bottom. Therefore, the ground of the outside of sheet-piles was also pushed by the sheet-pile, and moved outside and a large area of the ground around the foundation model deformed as a whole.

Next, in two cases of the Dr = 60% loose sandy ground, the displacement of the ground was observed in a large area around the footing model. However, the measurement and direction of displacement differed from each other on the shallow foundation and the sheet-pile foundation. In the case of shallow foundation, the ground had failure like a circular slip in the comparatively limited domain shallower than 100mm on both sides. The dis-placement of the ground around the edges of the footing model turned in the outward horizontal direction. On the other hand, about the sheet-pile foundation, sheet-piles restrained the hori-zontal deformation of the ground. Since deeper areas of the ground were deformed, as a whole, local failure of the ground had not occurred.

As mentioned above, it was thought that the resistance of the sheet-pile foundation was based on the following mechanisms. Since sheet-piles restrained the horizontal displacement of the ground, local failure of ground was prevented and horizontal re-sistance increased.

204

5 CONCLUSIONS

The following knowledge is acquired in this research. 1. The vertical bearing capacity of the sheet-pile foundation

is larger than that of the shallow foundation. 2. On the loose ground, the sheet-pile foundation can reduce

the settlement when repeated horizontal force acts on the structure.

3. The horizontal resistance characteristic of the sheet-pile foundation is higher than that of the shallow foundation.

4. The damping characteristic of sheet-pile foundation is higher than that of the shallow foundation.

5. The resistance mechanism of the sheet-pile foundation is to restrain the deformation of the ground, prevent local failure of the ground and resist horizontal force in a large area of ground.

Based on the above, it is shown that the sheet-pile foundation which is a new form of foundation structure is excellent in the performance for seismic force. From now on, construction tests of the sheet-pile foundation and full scale loading tests will be carried out and a design method will established, aiming at the utilization of sheet-pile foundation into railway structure.

REFERENCES

Koda, M., Murata, O., Nishioka, H., Punrattanasin, P., & Kusakabe. O. 2003. The proposal of sheet-pile foundation combining a footing with sheet-piles (in Japanese) : TSUCHI-TO-KISO JGS Ser. No.550, Tokyo: JGS. 51(11): 8-10.

Punrattanasin, P., Kusakabe, O., Murata, O., Koda, M. & Nishi-oka, H. 2002. Sheet pile foundation on sand under combined loading- A literature review and preliminary investigation. Technical Report No. 65, Department of Civil Engineering, Tokyo Institute of Technology, Tokyo: 57-85

Punrattanasin, P., Kusakabe, O., Murata, O., Koda, M. & Nishi-oka, H. 2003a. The behavior of sheet pile foundation on sand. Proceeding of BGA International Conference on FoundationsLondon: Thomas Telford.

Punrattanasin, P., Nishioka, H. Murata, O. & Kusakabe, O. 2003b. Combined loading apparatus for centrifuge test. Inter-national Journal of Physical Modeling in Geotechnics 3(4): 1-14.

Railway Technical Research Institute. 2000. Design Standard for Railway Structures (Foundation Structures) (in Japanese) To-kyo: Maruzen.

Railway Technical Research Institute 1999. Design Standard for Railway Structures (Seismic Design) (in Japanese) Tokyo: Maruzen.

Watanabe. K. & Tateyama. M. 2003. Shaking Table Tests on Seismic Behavior of Retaining Walls Using Image Processing System (in Japanese). RTRI REPORT 17 Railway Technical Research Institute, Tokyo. (3): 19-24.

a) Case-H-D-1 (The shallow foundation on Dr=90%)

b) Case-H-D-2 (The sheet-pile foundation on Dr=90%)

c) Case-H-L-1 (The shallow foundation on Dr=60%)

d) Case-V-L-2 (The sheet-pile foundation on Dr=60%)

Fig. 11. Displacement computed by an Image Processing Sys-tem when horizontal displacement reaches 20mm.


205

Pile Load Test Data Interpretation and Design Verification for HSR Project in Taiwan

S. P. Corbet FaberMaunsell Ltd, 160 Croydon Road, Beckenham, Kent, BR3 4DE, UK [email protected]

B. C. B. Hsiung Department of Civil Engineering, National Kaohsiung University of Applied Sciences, Kaohsiung, 807, Taiwan (formerly FaberMaunsell) [email protected]

F. Huppert BilfingerBerger AG, Germany [email protected]

Abstract: Construction of the southern section of Taiwan High Speed Rail Link will involve the construction of a large number of piles. The behaviour of the piles must be understood. At an early stage in the project, some static pile load tests were conducted at Taipao, Chiayi to investigate the ultimate capacity of piles when formed in the alluvium materials of the southwest plain in Taiwan. At later stages in the project, a number of pile load tests were carried out with the loads applied using Osterberg load cells (O-cells). The pile O-cell tests were performed in Contract C270 through the length of the contract. Some of the pile tests were carried out at a location about 2km north of Taipao. Using data from these pile load tests, the paper will explore and discuss the pile capacities of piles in-stalled in soft ground using the construction methods used during the construction of C270. The effects of pile size, base grouting and the variations possible in the elastic modulus of concrete will be considered. Based on the pile load test results at Taipao and in Con-tract C270, it was found that the percentage of the load carried in end bearing approximately 22% to 24% of the ultimate capacity. The variation in the value of the elastic modulus of the concrete can give a 16% to 25% variation in the calculated shaft friction force along the pile. For the piles tested using the O-cell system, it is concluded that some of the inconsistent calculated values of the unit shaft friction results close to the O-cells may be a result of soil disturbance resulting from soil movements as the O-cell is opened during the tests.

1 INTRODUCTION

To enable the rapid development of urban areas in the western part of Taiwan, a high-speed rail link is being constructed from Panchiao, Taipei County, in the north to Tzoying Kaohsiung City in the south. Total route length of the Taiwan High Speed Rail Link (THSRL) is 345 km. It will have six stations including Pan-chiao, Taoyuan, Hsinchu, Taichung, Chiayi, Tainan and Tzoying. The railway runs on viaducts for much of the route south of Paguashan tunnel. Along the southern section of the route, the main type of soil is Alluvium. To support the structures, a large number of piles have been installed. Pile load tests are an essen-tial part of the design process to confirm estimates of the ultimate pile capacity. This paper will consider the results from pile tests carried out by BOTHSR prior to inviting tenders for the construc-tion and some of the pile tests carried out by the Contractor to validate the designs. The early pile tests were carried out at the depot site at Taipao, which is close to the route in contract C270.

2 THE SITE

Contract C270 (THSRL chainage of 207km+015 to 249km+ 814) is located between Hsichou, Changhwa County and Taipao, Chiayi County and is 42.799 km long. The geology along this

section is alluvial deposits, consisting of sands, silts and clay with occasional layers of gravel. In some sections, there is a thin covering of made ground. The whole of the area is seismically very active with the Meishan Fault close to the route between ch 241+606 to 245+415.

The depth of the alluvium exceeds 100m. Underlying solid basement was not penetrated in any of the boreholes sunk during the ground investigations.

The soil types encountered in contract C270 are predomi-nantly either loose to medium dense deposits of soft to firm silty sands or firm to stiff silty clays to depths of about 30m below ground level. From 30 m to 50 m, the density of the soils in-creases and the soils are generally medium dense, with a few re-sults recorded as dense, cohesionless material. In this depth zone, clays would be described as being generally firm to stiff, occasionally very stiff. At depths greater than 50m below groundlevel the soils are mainly medium dense to dense with a greater proportion being described as very dense below 60m.

The sequence of the soil layers along the route is variable. The soils can be described collectively as alluvium, of generally low plasticity (where cohesive), and generally with strength and density increasing linearly with depth.

Groundwater was encountered between 2.7 and 4.3m below ground level during the ground investigation. The ground water level can vary locally due to abstractions for irrigation.

206

3 EARLY PILE LOAD TESTS (BOTHSR)

In 1996, to provide information and an understanding of the op-timum design for structural foundations for the THSRL, the Tai-wan High Speed Rail Bureau (BOTHSR), initiated a research contract before the tenders were invited for design and construc-tion . Piles were constructed using both reverse circulation drill-ing (RCD) and the casing oscillator method (OCM) at the provi-sional site of the Taipao Depot (THSRL chainage 251km+200). Three key aspects were considered in this study (Taiwan High Speed Rail Bureau, 1997):

(1) A review of the design criteria (2) Identification of problems of foundation analysis and con-struction (3) The effects of soil liquefaction and soil improvement

The Taipao Depot is located next to the southern end of con-tract C270; the ground profile is similar to the soils along Con-tract C270. The test results from Taipao Depot will be compared with pile tests carried out by the contractor for section C270.

4 C270 TEST PILES LOADED USING OSTERBERG CELL

The primary objectives of the C270 pile load tests were to verify design assumptions given in the Design Manual (FaberMaunsell Ltd, 2000) and where possible to optimise pile design by analys-ing the distribution of forces within the pile during loading. De-tails of the test piles constructed during this programme are given in the reports prepared by the sub-contractor LOADTEST (2000a, 2000b, 2000c, 2000d & 2000e), all of the test piles were loaded using Osterberg in-shaft load cells (O-cells). A schematic section of a test pile is shown in Fig. 1. The tests were carried out with three loading stages. In the first stage, the lower O-cell is expanded to assess the combined end bearing and shaft friction below the O-cell. In the second stage, after unloading the lower O-cell, the upper O-cell is pressurised to assess the shaft friction characteristic of the pile between the two O-cell assemblies by using the upper shaft friction as the reaction, the lower O-cell is left free to drain. During the last stage, after closing the lower O-cell, the upper O-cell is loaded to assess the friction characteris-tics of the pile above the upper O-cell assembly by using the combined middle and lower shaft friction and the end bearing as the reaction.

5 EVALUATION OF PILE CAPACITY

After reviewing data from pile load tests at the Taipao Depot and the eight O-cell test piles constructed in Contract C270 during the design stage, data from four of the tests, which are considered representative of the soil conditions encountered in the area of C270, have been selected for this comparison. To eliminate the influence of different construction methods on the analysis, only test piles at Taipao Depot constructed by the RCD method have been considered. Some differences may, however, still pertain due to the use of different drilling muds and the different lengths of the piles. Although the piles are of different lengths, the soils at the pile toe and over upper portions of the shaft are similar. Results selected for this comparison are from Pile B4 and B7 at Taipao Depot and piles 270-05 and 270-07 from the contractor �s test pile programme. During the construction of test pile B7, the

base was grouted to maximise the base resistance. Details of the test piles are presented in Table 1.

Test pile number B4 B7 270-05 270-07

Location Taipao Main Site

Pile diameter (m) 1.5 1.5 2 2

Pile length (m) 34.7 34.7 57.1 63.4

In a conventional pile load test, skin friction force along the shaft (P) is defined by:

s s c cP E A E A (1)

: strain measured from strain gauge on pile shaft Es: elastic modulus of reinforcement Ec: elastic modulus of concrete Ac: area of cross section of concrete As: area of cross section of reinforcement

Es is taken as 2.04 105 MPa and Ec is interpreted in

15,000 'c cE f (2)

Table 1 . Details of test piles.

Fig. 1. Schematic section of a test pile.

207

in which fc� is the unconfined compressive strength of concrete. Ec is equal to 2.5 104 MPa for a value of fc� = 28 MPa. Figure 2 shows the friction force along the pile shaft interpreted from strain gauges when the maximum top load, (1375 tonnes) is ap-plied to pile B4 and B7 using a conventional top loading static system.

A strain gauge was installed in the pile 40 cm above the pile toe (34.3 m below ground level) to provide data for an estimate of the end bearing resistance of the pile. The results from pile test B4 show the measured load from this strain gauge to be 3319 kN when 13,750 kN was applied to the pile head. In contrast, the base load measured in pile B7 was only 613 kN when 13,000 kN is applied to the pile head.

Considering the results of the O-cell tests, the unit shaft fric-tion force (P) is determined by

mixP E A (3)

: strain measured from strain gauges in pile shaft at one level. Emix: weighted pile modulus A: area of cross section of pile

Figure 3 presents the measured unit shaft friction force along the pile taken from the strain gauges in piles P270-05 and P270-07 tested using the double O-cell test.

The load is applied to the lower O-cell directly in the first stage test. Lower side shear force can be calculated from the strain gauges and it is possible to separate end bearing and shaft friction. Test results show that load carried in end bearing is 6.25MN for Pile P270-05 and 8.65MN for Pile P270-07.

6 DISCUSSION

The Taiwanese Building Code (TBC, 2001) recommends that the unit shaft friction resistance (fs) of piles in sand can be estimated using the Standard Penetration Test results ( SPT N) as:

1.96sf N (units: kPa) (4)

Based on the data from the conventional top loaded pile load tests at Taipao Depot, BOTHSR suggested that the relationship between shaft friction force (fs) and SPT N value could be:

for sandy soils 4.7 105.84sf N (units: kPa) (5)

for silty and clayey soils 4.7sf N (units: kPa) (6)

Based on the pile tests in the soft alluvium in south Taiwan, FaberMaunsell (2000) suggested that fs could be

for sands and gravels 3.3sf N (units: kPa) (7)

with a maximum value of 165kPa or N<50

for silt and clay of N 4 6.25sf N (units: kPa) (8)

with a maximum value of 150kPa or N<24

and for silt and clay of N>4 1.31 26sf N (units: kPa) (9)

Taking a typical common ground profile, the friction force along the pile shaft is calculated from Eqs. (5) to (9). Figs. 4 & 5 show the shaft friction force interpreted from the different equa-tions. The estimation from Eqs. (7) to (9) are close to pile load test results carried out by the contractor.

Fig. 2. Friction force along the shaft Test B4 and B7.

0

5

10

15

20

25

30

35

40

0 100 200 300

Unit pile shaft friction force (kPa)

B4

B7

0

10

20

30

40

50

60

70

0 100 200 300

Unit shaft friction force (kPa)

270-05

270-07

Fig. 3. Friction force along the shaft from the O-cell tests.

208

Considering different sizes of test piles installed on site, a similar relationship between f s and SPT N in sandy soils is found by TBC, BOTHSR and FaberMaunsell .. In this study it is seen that the pile diameter, in the range 1.5m to 2.0m, does not affect the unit values of shaft friction ( fs). Variations noted are a func-tion of the strain gauges in the test piles and possibly the effects of different drilling muds used during construction.

End bearing and shaft friction of the piles has been estimated from the pile load tests. Table 2 lists the shaft friction, end bear-ing and the ratio of end bearing to ultimate pile capacity as a per-centage for the pile load tests.

Test pile B4 B7 P270-05 P270- 07 Shaft friction force

(MN) 13.75 13.00 21.50 30.00

End bearing (MN) 3.32 0.61 6.25 8.65 Ratio of end bear-ing to ultimate pile

capacity (%) 24.1 4.72 22.5 22.4

It can be seen that the shaft friction is the most significant element in the pile capacity for the piles installed in the local soft alluvial soils. The ratio of end bearing to ultimate pile capacity varies from 22.4% to 24.1%. The end bearing and the base grouting at test pile B7 does not seem to have induced a signifi-cant increase in the end bearing resistance, this is consistent with the suggestion of BOTHSR (1997). However, there may be other reasons, which are not obvious from the BOTHSR report to ex-plain this result.

During the interpretation of shaft friction along the pile us-ing the strain gauges, an elastic modulus of concrete must be as-sumed. For test pile P270-05, at the time of testing, the concrete unconfined compressive strength was reported to be 37.1 MPa, using the relationship for the elastic modulus of the concrete ( Ec)as defined:

57000 'c cE f (where fc� is in psi) (10)

Combined with the area of reinforcing steel, a weighted pile modulus of 28,800 MPa is determined. Hsiung (2002) reported that the elastic modulus of concrete may be expressed as (BS8110, 1985)

,28 0 ,280.2c ag cuE K f (11)

where Ec,28 is the elastic modulus of concrete at 28 days. fcu,28 isthe characteristic cube strength at 28 days (in MPa). K0ag is a constant related to the aggregate material used for concrete, vary-ing from 14 to 26 GPa. The test was carried out 27- 29 days after pile construction. Thus, 37.1MPa of fcu,28 is used in Eq. (11). From Eq. (11), the elastic modulus of concrete at 28 days ( Ec,28)was calculated with a possible range of Ec,28 from 21.4 to 33.4 GPa depending on aggregate type, as shown in Fig. 6. The weighted pile elastic modulus varies in the range 21.5 to 33.5 GPa.

Shaft friction along the pile can be re-calculated using the weighted pile elastic modulus and the results are shown in Figs. 7 & 8. The variation of elastic modulus of concrete induces 16- 25% change of skin friction force along the pile.

As indicated in Fig. 3, it is seen that shaft force measured by reference to the strain gauge in the pile close to an O-cell in Pile P270-05 is not in agreement with the range of values expected. The disturbance of the soils close to the O-cell caused by a flow of soil during the opening of the O-cell is considered to be one possible reason for this observation

Fig. 4. Shaft friction based on the ground profile at Pier 7-570 in contract C270.

Table 2. Shaft friction, end bearing and ratio of end bearing to ultimate pile capacity.

Fig. 5. Shaft friction force based on the ground profile at Pier 7-621 at contract C270.

0

10

20

30

40

50

60

70

0 100 200 300 400 500Unit shaft friction (kPa)

FaberMaunsellBOTHSRTBCO-cell pile test

0

10

20

30

40

50

60

70

0 100 200 300 400 500

Unit shaft friction (kPa)

FaberMaunsell

BOTHSR

TBC

O-cell plie test

209

P270-05 is not in agreement with the range of values expected. The disturbance of the soils close to the O-cell caused by a flow of soil during the opening of the O-cell is considered to be one possible reason for this observation.

7 CONCLUSIONS

From pile load tests carried out in soft alluvium in south Taiwan to validate the design criteria to be used in the design of the foundations for the Taiwan High Speed Railway, the following conclusions have been reached: 1. Pile diameter does not affect the unit shaft friction force along

the pile in the ranges considered, 1.5m to 2.0m. 2. Shaft friction along the pile plays a key role in the ultimate pile

capacity for the piles installed in soft alluvial soils. The per-centage of end bearing to pile ultimate capacity is approxi-mately 22% to 24.%.

3. Base grouting does not seem to contribute significantly to end bearing. If grouting to increase the base resistance were to be evaluated further tests would be appropriate.

4. Variation of elastic modulus of concrete may result in varia-tions of between 16 to 25% in the values predicted for the shaft friction force.

5. The in-flow of soil during the opening of O-cells may disturb test results close to the cells.

ACKNOWLEDGMENTS

The authors would like to appreciate the permission given by Taiwan High Speed Rail Corporation, Bilfinger Berger AG/CEC Joint Venture and FaberMaunsell Ltd. for publication of this pa-per.

Fig. 6. Variation of elastic modulus of concrete.

Fig. 7. Influence of elastic modulus of concrete on shaft friction force at Pier 7-570.

Fig. 8. Influence of elastic modulus of concrete on shaft force at Pier 7-621.

0

10

20

30

40

50

60

70

0.0 100.0 200.0 300.0 400.0 500.0


UL: O-cell pile testLL: O-cell pile testFaberMaunsellBOTHSR

0

10

20

30

40

50

60

70

0 100 200 300 400 500


LL: O-cell pile testUL: O-cell pile testFaberM aunsellBOTHSR

0

5

10

15

20

25

30

35

40

0 10 20 30 40 50f cu,28 (MPa)

Lower limit

Upper limit

Design value

210

REFERENCES

BS8110. 1985. Structural use of concrete . British Standards In-stitution

BOTHSR. 1997. Optimal Design for bridge foundation of HSR Project, Research Report No. 2-0-86-05-02-165

FaberMaunsell Ltd. 2002. Test Pile Interpretative Report- Con-tract C270, Doc ID: C270/P/3501/C04/0042/A8

FaberMaunsell 2000. Contract C270: Design Manual Part II: Standard Bridges and Viaducts, DOC ID: C270/P/3502/C01 /0002

Hsiung, B.C. 2002. Engineering Performance of Deep Excava-tions in Taipei, PhD thesis, University of Bristol, UK

LOADTEST. 2000a. Data Report On Pile Load Testing: Test Pile C270/05

LOADTEST. 2000b. Data Report On Pile Load Testing: Test Pile C270/06

LOADTEST. 2000c. Data Report On Pile Load Testing: Test Pile C270/07

LOADTEST. 2000d. Data Report On Pile Load Testing: Test Pile C270/08

LOADTEST. 2000e. Data Report On Pile Load Testing: Test Pile C270/09

Taiwanese Building Code. 2001, Ministry of Interior Affair, printed by Construction Magazine (in Chinese)


211

Load Transfer Characteristics of Bored Piles in Singapore�s Old Alluvium

M. F. Chang Associate Professor, School of Civil & Environmental Engineering, Nanyang Technological University, Block N1, Nanyang Avenue, Singapore 639798 [email protected]

P. Teo Design Engineer, Land Transport Authority, 1 Hampshire Road, Singapore 219428 [email protected]

Abstract: Previous investigations of bored piles in Singapore�s Old Alluvium have indicated that load transfer along bored pile s is predominantly governed by shaft resistance and the observed load transfer characteristics in relation to the standard penetrationresistance, or the N-value, vary significantly. This paper summarizes the results of load tests on six instrumented bored piles, five tested in compression and one tested in tension, in the Old Alluvium at two project sites in Singapore. Results indicate that there is no distinct difference in the ultimate shaft resistance f s versus N relationship among piles in different soil groups, cast in different conditions and tested in compression and in tension. The f s/N ratio ranges typically from 1.5 to 4.8, similar to those from previous investigations. A correlation exists between fs/ c ( c = critical shaft displacement) and N. The mobilized base resistance qb is typically 5000 to 7000 kPa and the qb/40N ratio is between 1.4 and 2.5.

1 INTRODUCTION

In conjunction with the development of transport infrastructures, a large number of large diameter bored piles are installed in the Old Alluvium that covers the eastern part of the Singapore Island. Safe and economic design of bored piles in Old Alluvium, which is traditionally based on static formula, estimates of shaft and base resistance from the standard penetration resistance and use of a global factor of safety, has been a prime concern.

A number of studies of load transfer along bored piles in Singapore�s Old Alluvium have been carried out by means of axial load tests of bored piles usually instrumented with strain gauges and tell-tale rods. A general observation is that bored piles of significant length function essentially as friction piles and that load transfer along bored piles is predominately governed by shaft resistance, similar to that of bored piles in the residual soils of the Bukit Timah Granite and the sedimentary Jurong Formation, two other major geological units in Singapore. The detailed load transfer characteristics, such as the unit shaft resistance, or the fs value, in relation to the prevailed standard penetration resistance, or the N value, however, vary significantly.

A number of factors could contribute to the variation in the back-calculated fs/N ratio. These include accuracy of strain measurements, correctness of interpretation, pile construction and load testing details, and natural variation in soil characteristics in the Old Alluvium. An investigation aiming at minimizing the impact of various key factors will be useful.

This paper presents the load transfer characteristics of bored piles in Singapore�s Old Alluvium as obtained from careful interpretation of results of compression and pull-out load tests on six (6) piles installed in the slightly weathered and unweathered Old Alluvium at two sites in the eastern part of Singapore. Load transfer parameters deduced from these tests are analyzed to explore possible effect of soil variability, construction details, and the loading condition and compared with results from previous investigations. Practical implications of results of the present are discussed and suggestions are made for improving the

design of bored piles in Singapore�s Old Alluvium and potentially also stiff and hard soils elsewhere.

2 GEOLOGY AND GENERAL SOIL CONDITIONS

The Old Alluvium, which covers the eastern part of Singapore Island, is a major geological Formation of lightly cemented Pleistocene sediments (Tan, et al. 1980). It consists predominantly of silty to clayey sand or sand-clay mixtures of different consistency that varies with the degree of weathering. The majority of soils (over 70%) can be classified as SC or SM, 20% can be classified as CL or CH, and less than 10% can be classified as SP or SW (Li & Wong, 2001).

The shear strength of the Old alluvium is affected by cementation in the matrix. In some cases, sand particles are weakly cemented. In others involving highly weathered material, low plasticity clay provides a weak binding. As the N-value from the standard penetration test (SPT) provides a good measure of the degree of weathering and somewhat the extent of cementation, Chiam et al. (2003) suggested a classification of the Old Alluvium on the basis of the range of value of N, as follows:

Unweathered N > 100 Partially weathered 50 < N < 100 Distinctively weathered 30 < N < 50 Destructed 10 < N < 30 Residual N < 10 blows/0.3m

Two project sites of Land Transport Authority, covered by Old Alluvium, were investigated in conjunction with major infrastructure developments for transportation. The first site, Contract C504, is in Changi area while the second, Contract C821, is in Kim Chuan. The Old Alluvium at Changi site consists primarily of silty sand (SM). The Old Alluvium at Kim Chuan site consists predominantly of clayey to silty sand (SC) with varying plasticity and fines content. As indicated by the SPT resistance, the Old Alluvium is generally partially weathered at

212

the Changi site and partially to unweathered at the Kim Chuan site within the depth of pile embedment.

3 PREVIOUS INVESTIGATIONS

The load transfer of bored piles in Singapore �s Old Alluvium has been reported by a number of investigators (Yong et al., 1982; Chin, et al., 1985; Chan & Lee, 1990; Ho & Lim, 1994; Chin, 1996; Ong et al., 2001, Wei et al., 2002) based on results of axial compression tests on instrumented piles. While the shaft resistance has been consistently found to play a dominant role in the load transfer along bored piles, the deduced load transfer parameters in relation to the SPT resistance have been found to vary significantly.

Because the Old Alluvium is relatively stiff to hard, bored piles in the Old Alluvium are often cast in dry holes. Chin, et al.(1985), from four test piles located at three well-apart sites, found that the ratio of the limiting or ultimate shaft resistance f s over the SPT resistance, or the fs /N ratio, varied from 3.8 and 4.7 in SC (30 < N < 60) material to 6.0 in SM (40 < N < 60) material. The maximum mobilized base resistance qb was found to range typically from 19N to 35N (in kN/m2), and the shaft resistance was found to mobilize fully at a pile head settlement of less than 15 mm. Chan & Lee (1990) reported a f s /N ratio of 3.1 for one bored pile in slightly cemented Old Alluvium with N > 50. Chin (1996) found that the back-calculated fs /N ratio varied from 1.3 to 5.8 in primarily SM material with 10 < N < 94. The value of N was 120 near the base and the maximum observed qb was only 130 kPa, or less than 1.3N, possibly due to the presence of soft toe. Ong et al. (1999) reported, based on four piles along the North-East Rapid Transit Line, that the fs /N ratios were between 2 and 6. The maximum qb observed varied drastically from 170 to 7900 kPa, and the equivalent qb /N ratio was 3 to 66. Wei et al. (2002) found that the fs /N ratio generally ranged from 1.4 to 6.8 at one site (27 < N < 88) and 2.8 to 5.0 at the other (9 < N < 47). The mobilized base resistance was found to be at 4550 kPa, or close to 90N, for one pile, but less than 12N for the other pile.

Results of these investigations indicate that there is a significant scatter in the reported fs /N ratio even for piles constructed in dry holes. There is no clear trend between f s/N ratio and N or soil grouping (SM or SC), although a vague indication of slightly higher ratio fs/N for piles in SM materials exists. The base resistance is usually not fully mobilized at the maximum test load and the qb/N ratio varies significantly mainly due to the large variation in pile construction details and the extent of base cleaning.

Ho & Lim (1994), on the other hand, reported results of two test piles that were constructed in slurry stabilized wet holes. They found that, similar to the common finding, the load-settlement behaviour of such piles was largely governed by the load transfer characteristics along the shaft. The limiting shaft resistance fs was found to be 1.5N to 2N for one pile in material with 74 < N < 116 and 2N to 2.4N for the other in material with 65 < N < 126. These ratios were much reduced when compared with similar piles constructed in dry holes, as a result of wetting. Nevertheless, the maximum observed mobilized based resistance were found to be 5470 kPa, or 47N, and 4230 kPa, or 33N, respectively for the above two piles, indicating the insignificant effect of slurry on the mobilized base resistance.

4 PILES INVESTIGATED AND TYPICAL RESULTS

A total of six (6) bored piles constructed in the Old Alluvium were investigated. Three piles in Changi, namely TP-1, TP-2 and TP-3, and two piles in Kim Chuan, namely KC-1 and KC-2, were subject to compression tests, and one additional pile in Kim Chuan, namely KC-3, was subject to tension or pull-out test.

These piles were load tested 18 to 28 days after construction. The details of these piles and the load tests are shown in Table 1.

Basically maintained loads provided by kentledge were used in compression tests and reaction pile systems were used in tension tests. The three piles in Changi piles were debonded in the upper 10 or 20 m, whereas, the piles at Kim chuan were installed from an excavated level averaged 15.5 m below the existing ground surface. All the piles were embedded in either partially weathered or unweathered Old Alluvium. In addition to two to three telltales, the piles were instrumented with vibrating wire strain gauges (VWSGs) at different levels along the pile axis to measure the strain distribution during axial load testing.

As all the piles were instrumented, axial strain distributions suitable for the derivation of the load distributions and shaft displacements at various levels were obtained in addition to the load-settlement curve. Because of uncertainties that may exist in the measured strain readings, one often needs to carefully screen through the raw data and select a valid workable strain distribution after applying certain process of simplification. Subsequently, a standard procedure such as that described in Chang (2001) can be followed to calculate the unit mobilized resistance for each step of loading and the corresponding pile displacement relative to the surrounding soil along the shaft as well as for the base. A set of load transfer curves which describes the gradual mobilization of unit resistance with an increase in relative displacement between the pile and the soil can then be constructed.

To illustrate this process of interpretation of load test results from instrumented piles, the typical test results from the first compression pile at Kim Chuan, KC-1, is selected as an example. Figure 1(a) shows the typical original strain distribution and for Pile KC-1 along with the relevant soil stratification. The upper most pile segment appeared to have attracted relatively high unit friction probably due to the presence of desiccated crust. Figure 1(b) shows the modified strain distribution with the friction in the upper most pile segment ignored that was selected for the subsequent analysis. As the pile penetration is relatively small, there were significantly large registered strains at the pile base, signifying a significant mobilization of resistance at the base at the end.

Figure 2 shows the observed load-settlement relationship at the pile head. At the working load, the pile settlement reached 4.1 mm. The pile exhibits significant displacement as the applied load exceeds 1.5 times the design load of 1500 tons (14710 kN). The pile plunged by over 50 mm after the applied load was increased to the maximum value of 2250 tons (22066 kN).

The assumed elastic modulus, E, of the pile has an important effect on the deduced load transfer curves. Very often, one back-calculates the E-value from strain measurements in gauges close to the pile top where the axial load is known to be equal to the applied load. Figure 3 shows the variation of elastic pile modulus back-calculated from the applied load and the corresponding measured strains at 0.4 m as well as at 0.7 m for Pile KC-1. The elastic modulus clearly decreases with an increase in applied load or strain level and the rate of decrease is rather rapid in the first few steps of loading. The two sets of modulus values were similar and the simplified modulus variation as indicated by the dotted line in the figure was selected for the subsequent calculations of load distribution and mobilized unit shaft resistance. The selected modulus is similar to a value of 36.9 kN/mm2 determined from a laboratory compression test on a concrete cylinder specimen prepared from a cube sample collected at the site.

It is important to incorporate the degradation of modulus in the analysis, especially for piles subjected to tension in a pull-out test where the decrease in E could be very drastic in the first few steps of loading.

213

Table 1. Details of test piles.

Pile No. TP-1 TP2 TP3 KC-1 KC-2 KC-3

Diameter, D (mm) 1000 1000 1000 1200 1200 800

Length, L (m) 20.8 38.2 35.7 17 17 12.2

De-bonded Length 10 m 20 m 20 m - - -

Concrete Grade & Casting Method

Grade 45 by tremie in dry

hole

Grade 45 by tremie in dry

hole

Grade 45 by tremie in

slurry

Grade 40 in dry hole



Working Load (WL) 510 tons 786 tons 786 tons 1000 tons 1000 tons 290 tons

Maximum Test Load 1110 tons (2.15WL)

1470 tons (1.87WL)

1325 tons (1.69WL)

2250 tons (2.25WL)

2550 tons (2.55WL)

520 tons (1.8WL)

Instrumentation 32 - VWSGs 3 - Telltales

32 � VWSGs 3 - Telltales

32 - VWSGs 3 - Telltales




(a) (b)

Fig. 1. Distributions of axial strain for Pile KC-1 : (a) measured (b) modified.

Figure 4 shows the deduced load transfer curves for the distinctive pile sections in various soil strata and for the pile base. It is seen that typically the unit shaft resistance is mobilized practically fully at a relative shaft displacement of between 5 and 6 mm, although the limiting fs values in various strata vary. On the other hand, the base resistance has not reached full mobilization even at the maximum test load when the base displacement reached 46 mm or 3.8% of the pile diameter.

5 LOAD TRANSFER PARAMETERS

Although load transfer curves similar to those shown in Fig. 4 can be directly used in load transfer analysis in the prediction of load settlement relationship for the design of bored piles at a

specific site, load transfer characteristics can be described alternatively by load transfer parameters, namely the limiting or ultimate resistance and the corresponding critical pile displacement beyond which the increase in mobilized resistance is drastically reduced with further increase in displacement, for general applications in practice.

Table 2 summarizes the key test results from the load tests for all the piles and the deduced load transfer parameters. As the shaft resistance has fully mobilized practically in all the piles investigated, both the limiting shaft resistance fs and the interpreted critical shaft displacement, c, can be reasonably interpreted or estimated from the load transfer curves for the pile shaft. For the base resistance, which did not mobilize fully at the end of the test, only the maximum mobilized base resistance q band the corresponding base displacement b are available.

0

2

4

6

8

10

12

14

16

18

0 200 400 600 800Axial Strain, , 10-6

250T

500T

750T

1000T

1250T

1500T

1750T

2000T

2168T

2218T

2250T

0

2

4

6

8

10

12

14

16

18

0 200 400 600 800

Clayey to Silty Sand (N=70)




0

2

4

6

8

10

12

14

16

18

0 200 400 600 800Axial Strain, 10-6

500T

1000T

1500T

2000T

2250T

214

0

10

20

30

40

50

60

0 5000 10000 15000 20000 25000

Applied Load, Po (kN)

Fig. 2. Load-settlement relationship for Pile KC-1.

2030405060708090

100110

0 5000 10000 15000 20000 25000

Applied Load, Po (kN)

0.4m VWSG

0.7m VWSG

E= 62 at Po = 4904kN

E =44-0.00065(Po-9807)

Fig. 3. Variation of pile modulus with applied load for Pile KC-1.

It is seen that, with the exception of Changi TP-1, generally the critical shaft displacement c is between 4 and 10 mm for the compression piles, similar to those commonly reported. For the tension pile, the c is much smaller, at 1 to 3 mm. With the exception of Changi TP-2 pile, which was cast in a slurry-filled borehole where base cleaning was probably not thorough, the maximum mobilized base restance ranged from 4920 to 6970 kPa, with the corresponding base displacement ranged from 2.4 to 6.9 % of the pile diameter. The mobilized q b values are similar to those reported by Ho and Lim (1994) and Wei et al. (2002) for piles that were cast in a carefully drilled borehole with a clean base.

6 PRACTICAL IMPLICATIONS AND RECOMMENDA-TIONS

In the traditional method of design of piles in stiff soils in Singapore, the ultimate pile capacity Qu is usually calculated using the static formula, as follows:

Qu = fsAs +qbAb = (KsN)As + (40KbN)Ab (1)

0

100

200

300

400

500

600

0 10 20 30 40 50 60Shaft Displacement, s (mm)

0.7-3.7m(N=71)3.7-12m(N=94)12-14.5m(N=100) 14.5-17m(N=100)

(a)

0

1000

2000

3000

4000

5000

6000

7000

0 10 20 30 40 50

Base Displacement, b (mm)

(b)

Fig. 4. Load transfer curves for Pile KC-1: (a) shaft; (b) base.

where, respectively, fs and qb are the ultimate shaft and base resistance, and As and Ab are the shaft area and base area. Note that both fs and qb, usually in kN/m2, are commonly estimated from the N-value. A design engineer is always concerned with what values of Ks and Kb are to use in the design of bored piles.

Notwithstanding the great diversity in the deduced resistance values in previous investigations, SPRING (2003) recommended the following ranges of values: Ks = 2 3 (fs 300 kPa) and Kb = 1 3 (qb 10, 000 kPa) for the design of bored piles in the Old Alluvium in Singapore.

Figure 5 shows the unit shaft resistance versus N-value relationship from the bored piles investigated, as summarized in Table 2. Although similar to results from others, there is a large scatter, the common observed trend of increase in shaft resistance with an increase in N-value is evident. Specifically, it is seen that (1) for a soil with a given range of N- value, the fs/N ratio or the Ks value is not drastically different between the compression test and the tension test and between piles cast in dry hole and cast in slurry-filled hole and (2) the majority of data points fall in the range where Ks is between 1.5 and 4.8, similar to those reported earlier.

215

Table 2. Summary of test results and deduced load transfer parameters.

Pile No. TP-1 TP-2 TP-3 KC-1 KC-2 KC-3 Settlement at WL (mm)

7.0 9.1 19.6 4.1 3.4 2.9

Max. Settlement (mm)

78.2 53.5 74.2 52.7 38.9 12.1

1 10-12m SM(N=48)

20-25m SM(N=57)

20-26m SM(N=56)

0.7-3.7m SC(N=70)

0.7-9.5m SC(N=105)

1.5-2m SC(N=70)

2 12-16m SM(N=47)

25-31m SM(N=80)

26-32m SM(N=50)

3.7-12m SC(N=102)

9.5-12m SC(N=105)

2-3.5m SC(N=80)

3 16-20.8m SM(N=53)

31-34m SM(N=85)

32-35.7m SM(N=95)

12-14.5m SC(N=118)

12-17m SC(N=105)

3.5-6.5m SC(N=92)

4 34-38.2m SM(N=85)

14.5-17m SC(N=116)

6.5-9m SC(N=120)

Major Soil Stratification

5 9-12m SC(N=111)

1 138 88 217 157 368 237 2 194 352 141 282 584 260 3 262 310 111 379 209 143 4 271 385 218

Limiting Shaft Resistance, fs(kN/m2)

5 85 1 22.0 10.0 9.5 6.0 4.0 3.0 2 14.5 7.6 7.5 5.0 10.0 3.3 3 14.0 6.5 7.5 5.0 4.0 1.7 4 6.0 6.0 1.0

Critical Shaft Displacement,

c (mm)

5 1.0 Max. qb(kN/m2)

4920 800 6970 6360 6600 -

Max. Base Movement (mm)

69 (6.9%D)

37.5 (3.8%D)

63.5 (6.4%D)

46(3.8%D)

29(2.4%D)

-

0.0

100.0

200.0

300.0

400.0

500.0

600.0

700.0

800.0

0.0 50.0 100.0 150.0 200.0

N Value (blows/0.3m)

TP-1(C)

TP-2(C)

TP-3(C)

KC-1(C)

KC-2(C)

KC-3(T)

fs/N=4.8

fs/N=1.5

Fig. 5. Limiting shaft resistance versus N-value.

Figure 6 shows the qb values from the five compression piles plotted against N. Except for one of the piles that was cast in slurry-filled borehole, the qb values were found to range from 5000 to 7000 kPa and the corresponding Kb is between 1.4 and 2.5, similar to that recommended in SPRING (2003). These values can be considered as the practical limiting values as they correspond to relatively large base displacements.

In an improved design procedure based on the load transfer method (Chang & Broms, 1991), one needs another load transfer parameter, the critical displacement for the shaft as well as for the base. Based on Chang & Goh (1988), a strong correlation might

exist between fs/ c, which represents the secant stiffness, and the N-value. Figure 7 shows the relationship between fs/ c and N based on data presented in Table 2. It is interesting that (1) the fs/ c values are distinctively higher for the tension pile than for the compression piles and (2) there is a good correlation between fs/ c and N for compression piles in the Old Alluvium with 50 < N < 120 as follows: fs/ c = 0.73 (N-30) in kPa/mm.

0.0

2000.0

4000.0

6000.0

8000.0

10000.0

12000.0

0.0 50.0 100.0 150.0

N Value (blows/0.3m)

Kb/N =2.5

Kb/N =1.4

Fig. 6. Mobilized base resistance versus N-value.

216

0

50

100

150

200

250

0.0 50.0 100.0 150.0

N Value (blow s/0.3m)

TP-1(C) TP-2(C)TP-3(C) KC-1(C)KC-2(C) KC-3(T)

For compression piles:f s/ c= 0.73 (N-30)

Fig. 7. Relationship between fs/ c and N.

For design applications of bored piles in major projects in Singapore�s Old Alluvium using the improved load transfer procedure, a site-specific verification of fs from load tests of instrumented piles is advisable because of the diversity of the field observed Ks value and construction effects. However, for preliminary design purposes, an average f s/N ratio of between 2 and 4 appears reasonable for compression piles. The use of f s/ c= 0.73 (N-30), combined with appropriate K s values, will provide engineers a means of estimating both f s and c and consequently a complete load transfer curve for the pile shaft, based on for example Vijayverjiya (1977) and others. As to the end bearing, a Kb value of 1.4 or 1.5 appears to provide a conservative estimate of qb for a properly constructed pile. The critical base displacement corresponding to the recommended qb value can be taken as 6% of the pile diameter.

Although a lowering of the fs/N ratio may not be necessary for tension piles, further verification is recommended. An improvement should be made on bonding between the main reinforcement and the concrete to allow more effective transfer of loads and to avoid severe cracking of the pile concrete under the working loads that may affect the pile integrity in the long run.

7 CONCLUSIONS

An investigation of five compression tests and one tension test on bored piles in Singapore�s Old Alluvium has indicated load transfer characteristics that are similar to those observed previously. The major conclusions are as follows:

(1) The relationship between fs and N is highly scattered and the fs/N ratio, or the Ks, is typically between 1.5 and 4.8, similar to those from previous investigations. There is no distinct difference in Ks values for compression piles constructed in different soil groups and in different borehole conditions. Site specific verification by means of load tests of instrumented piles is recommended for major projects, although a typical K s value of between 2 and 4 may appear reasonable for preliminary designs.

(2) There is no distinctive difference in the deduced fs values between compression piles and tension piles, although further verification may be necessary.

(3) The maximum observed qb varies due to variation in soil condition and construction details. The q b values are typically 5000 to 7000 kPa and the corresponding Kb values are 1.4 to 2.5 for piles in material with 50 N 120. These values of qbcorrespond to relative large base displacements and can be taken as the practical limiting base resistance in the design of properly constructed bored piles in partially to unweathered material.

(4) For an improved design of bored piles subject to axial compression in Singapore�s Old Alluvium using the load transfer approach, the critical shaft displacement c can be estimated using fs/ c = 0.73 (N-30) in kPa/mm and the corresponding fsvalues estimated from appropriate fs/N ratios.

ACKNOWLEDGEMENT

The authors would like to express their appreciation to the Nanyang Technological University and the Land Transport Authority of Singapore for supporting the relevant research on which the paper is based.

REFERENCES

Chan, S. F. & Lee, S. L. 1990. The design of foundations for Suntec City, Singapore. Proceedings of Conference on Deep Foundation Practice 27-32, Singapore.

Chang, M. F., 2001. Interpretation and use of axial load tests on instrumented bored piles. Journal of the Institution of Engineers, Singapore 41: 36-48.

Chang, M. F. & Goh, A. T. C. 1988. Behaviour of bored piles in residual soils and weathered rocks of Singapore. Research Project Report for RP 1/84 & RP 2/84, Nanyang Technological University, Singapore.

Chang, M. F. & Goh, A.T. C. 1989. Design of bored piles considering load transfer. Geotechnical Engineering 20(1): 1- 18.

Chang, M. F. & Broms, B. B. 1991. Design of bored piles in residual soils based on field-performance data. Canadian Geotechnical Journal 28(2): 200-209.

Chiam, et. al. 2003. The Old Alluvium. Proceedings of Underground Singapore 2003: 408-427. Singapore.

Chin, Y.K. et al. 1985. Ultimate load test on instrumented bored piles In Singapore Old Alluvium, Proceedings of the 8th

Southeast Asian Geotechnical Conference 1: 2.54-2.65, Kuala Lumpur.

Chin, J.T., 1996. Back-analysis of instrumented bored piles In Singapore Old Alluvium. Proceedings of the 12th Southeast Asian Geotechnical Conference I: 441-446, Kuala Lumpur.

Ho C. H. & Lim, C. H. 1994. Bearing capacity and settlement of slurry bored piles in Singapore Old Alluvium. Proceedings of the 3rd International Conference on Deep Foundation Practice: 125-132, Singapore.

Li, W.W. & Wong, K.S. 2001 Geotechnical properties of Old Alluvium in Singapore. Journal of the Institution of Engineers 41:10-20. Singapore.

Ong et al., 1999. A summary of preliminary pile load test results for North East Line. Proceedings of the Rapid Transit Conference: 689-702. Singapore.

SPRING. 2003. Singapore Standard CP4:2003- Code of Practice for Foundations. Spring Singapore.

Tan, S. B., et. al. 1980. Engineering geology of the Old Alluvium in Singapore, Proceedings of the 6th Southeast Asian Conference on Soil Engineering 1: 673-684. Taipei.

Wei, J. et al. 2002. Utilization of instrumented pile testing for cost effective foundations, Proceedings of Conference on Case Studies in Geotechnical Engineering 271-287. Singapore.

Vijayvergiya, V. N. 1977. Load-movement characteristics of piles. Proceedings of the ASCE Ports'77 Conference, Long Beach, California: 269-284.

Yong, K. Y. et al. 1982. Ultimate load test of an instrumented cast-in-place bored pile installed in stiff silty clay. Proeedings of the 7th Southeast Asian Geotechnical Conference 1: 453-463. Hong Kong.


217

Load Transfer Characteristics of Bored Piles Embedded in Weak Rocks

L. W. Wong Civil Engineering and Development Department, The Government of the HKSAR, Hong [email protected] R. Barsby Hyder Consulting Limited, Guildford, England [email protected] A. HoughtonTaiwan High Speed Rail Corporation, Taipei, [email protected]

Abstract: Based on results of 10 preliminary test piles, mobilization of shaft and base resistances of piles in weak rocks of sandstone and mudstone are interpreted. The load-displacement behaviour of the pile/rock interfaces can be described by a power function. The shaft and base resistance have the tendency to increase with depth but decrease with increasing displacement. With the empiricalparameters obtained from the preliminary tests performance of 5 working piles are analyzed by the load-transfer method. Results of analyses indicate good agreement between the observed and calculated pile head settlements.

1 INTRODUCTION

The use of the load transfer method, or the t-z curve method, in the analysis of pile load test results and the calculation of settlement of a single pile has been widely adopted in recent decades. The key parameters, the mobilized shaft friction and base resistance and the corresponding pile displacements are interpreted from the pile load tests. However, there were very limited literatures reporting the load transfer data on bored piles in slightly weathered to fresh weak rocks. Chang & Broms (1991) reviewed the unit shaft resistance and the critical displacement of bored piles in the residual soil of the Jurong Formation in Singapore. These piles were socketed into the highly weathered siltstone or the highly weathered shale with depth ranging from 1.5 m to 12 m. Moh et al. (1995) presented the load transfer results for a 1.5 m diameter bored pile in Taipei that embedded in the fresh shale and sandstone by 3.3 m. The data on shaft resistance versus displacement are so limited that the t-z curve method can hardly be a readily design tool until preliminary load tests have been conducted and the key parameters are determined.

During construction of the Hsinchu to Tungshiao section of the Taiwan High Speed Rail (HSR), a pile test programme that comprised 10 axially loaded compression instrumentation test piles and 5 working pile load tests were conducted. Majority of the test piles were embedded predominant 12 m to 36 m in fresh to slightly weathered rocks of the sandstone, the mudstone or the interbedded sandstone and mudstone. The first part of this paper interprets of the shaft and base resistances and their corresponding displacements from the preliminary tests. The abundance data enable the development of empirical resistance versus displacement envelops for various types of rocks so that the load transfer characteristics could be applied to piles in similar geological conditions. The second part of the paper applies the back-calculated load transfer parameters and the load transfer method to perform a Class 2 prediction on those working test piles. Results of this study provide an insight into the load displacement characteristics of piles in weak rocks, and the

significance of a pile with a soft toe. Topics for further research are suggested.

2 PRELIMINARY TEST PILES

The viaducts for the Hsinchu to Tungshaio section of the HSR is approximately 20 km in length. The viaduct foundations were supported on bored piles of 1.5 m in diameter with lengths varying from 24 m to 36 m. The piles are embedded in sandstone or the interbedded sandstone and mudstone of the Cholan formation or the Toukoshan formation. Belong to the late Pliocene and early Pleistocene epoch, the rocks are extremely weak to weak, with the uniaxial compressive strengths generally varying from 0.1 MPa to 2.3 MPa. The standard penetration test N300 values, extrapolated to 300 mm as defined by Stroud (1974), for the weak rocks range from 70 to 300, with an average of 210.

The piles were constructed by the cased bored piling method. Approximately 2,000 working piles of 1.5 m in diameter were constructed. The Casagrande type oscillator for advancing and withdrawal of casing and the clamshell bucket excavation system were adopted. Although groundwater levels are located at 2 m to 4 m below the ground surface, the hole of the pile is essentially dry during excavation due to the sealing effect of the casing.

Ten preliminary compression load tests were conducted. Lengths of these test piles varied from 12 m to 24 m. Vibrating wire strain gauges were installed at multi-levels to measure the distribution of strains along the pile shaft. A pair of extensometers was installed between the head and the toe of each of the test pile to measure pile toe deformations. Three strain gauges at each level and 3 to 6 levels were installed in each of the test piles. The vertical spacing between each level ranged from 3 m to 5 m. In total, there were 56 numbers of instrumented segments for these 10 preliminary test piles.

The test piles were embedded in the interbeds of sandstone and mudstone (Ss&Ms), sandstone (Ss), mudstone (Ms) and in the alluvial gravel deposits (GM). As summarized in Table 1, except for the test pile 3PC2, the preliminary test piles were loaded to the maximum load ranging from 20 MN to 38 MN.

218

Two to 4 unload-reload cycles were applied. Residual settlements were measured for each unloading stage. It is noted that the lengths of the preliminary test piles were one half to two third of the working piles and they were loaded up to 3 times of the normal design load. This arrangement enables sufficient movements of the piles so that significant base loads are mobilized. Due to page limit load-displacement results for the gravel deposits are beyond the scope of this paper.

3 ANALYSIS OF PRELIMINARY PILE LOAD TESTS

In the analysis of the preliminary pile load test results, load- transfer curves representing the relationships between the mobilized resistances and the corresponding absolute displacements of the pile segments relative the supporting ground are deduced from the load settlement data and the measured strain distributions.

The load-transfer curves for shaft resistances, t-z curves, for the interbedded sandstone and mudstone and for the sandstone materials are presented in Figs. 1 and 2 respectively. Most of the t-z curves have the characteristics that prior to reaching their ultimate resistances the curves are bilinear and the critical displacements could not readily be determined.

Development of base resistances against the pile toe displacements, the q-z curves, for pile bases founding on the interbedded sandstone and mudstone and on the predominant sandstone materials are presented in Figs. 3 and 4. An ultimate base resistance has not been mobilized even at pile base displacements exceeding 100 mm. The displacement of 145 mm that observed at the base of the pile 3PC2 for mobilization of a base resistance of 2.4 MN as shown in Fig. 4 would be an indication of a pile with a soft toe.

Table 1. Summary of preliminary pile load tests.

4 GROUND STIFFNESS

In previous literatures such as Coyle & Reese (1966) and Vijayvergiya (1977), the load-transfer curves in the residual soil or in clay materials could be idealized as elastic-plastic and a set of parameters of the ultimate resistance and the corresponding critical displacement could be interpreted. However, as presented in Figs. 1 to 4, it would be difficult to interpret the set of parameters for bored piles in weak rocks. In order to analyze the pile performance by the load-transfer method, alternatively the ground stiffnesses are interpreted from the slopes of the t-z and the q-z curves. Unload-reload stiffnesses are also determined from unload-reload cycles.

Pile 4PC5

0

100

200

300

0 100 200 300

Absolute displacement of pile shaft mm

5-8.5 m8.5-12 m12-15.5 m15.5-19.5 m

Fig. 1. Development of shaft resistance in sandstone and mudstone.

Pile 3PC4

0

100

200

300

0 50 100 150Absolute displacement of pile shaft mm

1-6 m6-11 m11-17 m17-23 m

Fig. 2. Development of shaft resistance in sandstone.

4.1 Ground Stiffness for Loading

The ground stiffness, Ks , for the pile-rock interface along the shaft is defined as:

Ks = fs Ds / Ss (1)

The Ks is basically the slope of the shaft resistance versus pile displacement curve extending from the origin and the shaft displacement is normalized with the shaft diameter.

The secant modulus of the ground below the pile base, Eb , is calculated by the formula suggested by the AASHTO (1996):

Eb = P / Db Sb (2)

where: fs the resistance of the pile-rock interface mobilized along the

shaft of the pile, P the load at the base of the pile,Ss the absolute displacement of the shaft relative to the ground, Sb the absolute displacement of the base of the pile, Ds the diameter of the pile shaft, Db the diameter of the pile base.

Pile length

Maximum test load

Maximumdisplace.

Residual displace.Pile

No.

Type of rock socket m MN mm mm

3PC1 GM, Ss&Ms 12.0 27.5 49.2 38.0 3PC2 Ss 18.4 7.0 142.5 136.6 3PC3 GM, Ss&Ms 22.0 35.0 52.2 37.5 3PC4 Ss, Ms 24.3 38.0 109.1 92.1 3PC5 Ss 24.1 33.0 184.5 171.0 4PC1 GM, Ss&Ms 14.0 21.8 71.5 58.4 4PC2 Ss, Ms 12.0 22.5 192.7 76.2 4PC3 Ss&Ms 12.3 25.0 90.1 72.5 4PC4 Ss&Ms 24.5 33.0 137.2 115.5 4PC5 Ss&Ms 20.5 30.3 238.6 225.7

219

Ss&Ms

0

10

20

0 25 50 75 100 125Absolute dipslacement of pile base mm

4PC44PC34PC23PC33PC1

Fig. 3. Development of base resistance in sandstone and mudstone.

Ss

0

5

10

15

0 50 100 150 200 250Absolute displacement of pile base mm

4PC53PC23PC5

Fig. 4. Development of base resistance in sandstone.

The ground stiffness for the pile-rock interface along the shaft and the secant modulus for the ground beneath the base are calculated respectively from the slopes of the t-z curves and the q-z curves. Figure 5 shows a typical resistance-displacement relationship for the supporting ground and the absolute displacements are normalized with the diameter of the pile. The diagram in Fig. 5 is actually interpreted from the q-z curve for Pile 3PC1 presented in Fig. 3. There are 3 unload-reload cycles. The resistance versus normalized displacement diagram exhibits irrecoverable displacements after unloading. The slopes for unload-reload cycles are somewhat parallel with each other.

Results of data processing on those t-z curves for Pile 4PC5 are presented in Fig. 6. The computed ground stiffnesses along the shaft are plotted against the corresponding normalized displacements. The stiffness versus displacement relationships are family of power function curves that straightened to linear curves when plotted in the log-log scale. The ground stiffness decreases with increasing displacement and increases with depth. Normalized with the in-situ total vertical stresses, the ground stiffness versus displacement diagrams for all pile segments merge into common envelops that are shown in Figs. 7 to 10.

The average values for the normalized ground stiffnesses along the pile-rock interface and the normalized secant moduli

0

1

2

3

0 0.01 0.02 0.03 0.04Absolute displacement / Diameter of pile

Shaf

t or b

ase

resi

stanc

e M

Pa

Fig. 5. Typical resistance versus displacement relationship for the supporting ground.

Pile 4PC5

1

10

100

1000

0.0001 0.001 0.01 0.1 1Absolute displacement of shaft / Diameter of shaft

5-8.5 m8.5-12 m12-15.5 m15.5-19.5 m

Fig. 6. Variation of secant modulus with displacement in sandstone and mudstone.

beneath the base, Ks / v and Eb / v , respectively, could be expressed as:

Ks / v = a (Ss /Ds)-m (3)

Eb / v = a (Sb /Db)-m (4)

where: v the in-situ total vertical stress at the center of the segment,

a the stiffness coefficient, m the stiffness power factor.

The unit weight of 22 kN/m3 for the weak rocks is adopted for calculating the vertical stresses. Empirical equations for ground stiffness or secant modulus versus displacement relationships, determined by regression analysis, are presented in Figs. 7 to 10. The stiffness parameters a and m are summarized in Table 2.

Stiffness coefficients for the base of the pile in a normal toe and in a soft toe are interpreted for the sandstone. As shown in Fig. 4, at displacements exceeding 40 mm, two of the test piles show stiffer base resistances. At displacements less than 40 mm, all three test piles exhibit softer moduli that are about 1/4 of that for the stiffer base. It appears that the ground stiffness beneath the pile base in sandstone would be construction dependant.

Origin for unloading

Eb-urKs-ur

1

EbKs

1

Unloading curve

Unload-reload cycle

Origin for reloading

220

Ss&Ms

1

10

100

1000

10000

0.0001 0.001 0.01 0.1 1Shaft displacement / Diameter of shaft

Gro

und

stiff

ness

/ V

ertic

al st

ress

LoadingUnload-reload

Fig. 7. Ground stiffness along pile shaft in sandstone and mudstone.

Ss

1

10

100

1000

0.0001 0.001 0.01 0.1 1Shaft displacement / Diameter of shaft

Gro

und

Stiff

ness

/ V

ertic

al S

tress

Loading

Fig. 8. Ground stiffness along pile shaft in sandstone.

Table 2. Summary of ground stiffness empirical coefficients.

4.2 Ground Stiffness for Unload-reload Cycles

As shown in Fig. 5, the ground stiffness or the secant modulus for an unload-reload cycle is interpreted from the unloading curve of the resistance versus displacement diagram with the origin shifted to the maximum load prior to unloading, or from the reloading curve up to the maximum load of the preceding cycle with the origin corresponding to the displacement when the load was totally removed.

Figures 7 & 9 indicate that ground stiffnesses or secant moduli for unload-reload cycles are approximately 2 to 3 times of that for the loading. For the sandstone and mudstone, the

Ss&Ms10

100

1000

10000

0.0001 0.001 0.01 0.1Pile base displacemnt / Diameter of base

Seca

nt M

odul

us /

Ver

tical

Stre

ss

LoadingUnload-reload

Fig. 9. Secant modulus below pile base in sandstone and mudstone.

Ss

10

100

1000

0.0001 0.001 0.01 0.1 1Pile base displacement / Diameter of base

Seca

nt m

odul

us /

Ver

tical

stre

ss

Normal toeSoft toe

Fig. 10. Secant modulus below pile base in sandstone.

unload-reload stiffnesses of the pile-rock interface along the shaft, Ks-ur , and that beneath the base, Eb-ur , could be expressed by Equations 5 & 6:

Ks-ur = 2 Ks (5)

Eb-ur = 2 Eb (6)

Although there are less unload-reload data for piles in the sandstone, it is considered that Equations 5 and 6 are also applicable to the sandstone.

5 ANALYSIS OF WORKING PILES

Results for 5 working pile load tests are available for verification of the empirical stiffness parameters a and m that interpreted from the preliminary pile load tests. The working piles were loaded to 2 times of the design normal loads. The observed pile head settlements for the test piles are summarized in Table 3.

The load transfer method, or the t-z curve method that reported by Coyle & Reese (1966) is adopted for the prediction of the working pile load tests. A simple computer programme is

Stiffness parameter Rock type

Resistance a m

Shaft 1.43 0.759 Sandstone and mudstone Base 85.5 0.312

Shaft 1.12 0.749 Base- Normal toe 6.36 0.560

Sandstone

Base- Soft toe 1.64 0.560

Eb / v = 85.5 (Sb /Db)-0.312

Ks / v = 1.43 (Ss /Ds)-0.759

Ks / v = 1.12 (Ss /Ds)-0.749

Eb / v = 6.36 (Sb /Db)-0.560

221

Pile 406

0

5

10

15

20

0 2 4 6 8 10Pile head settlement mm

Load

at p

ile h

ead

MN

ObservedCalculated- Normal toe

Fig. 11. Prediction of pile head settlement for Pile 406.

Pile 411

0

10

20

30

0 5 10 15Pile head settlement mm

ObservedCalculated-Normal toe


developed to facilitate rapid calculation. Basically the pile is idealized by a series of elastic discrete elements supported by a number of nonlinear side springs and a base spring, which represent the soil-structure interaction.

The load displacement behaviour expressed in Equations 1 to 6 together with empirical coefficients that summarized in Table 2 are initially adopted for estimating the pile head settlements. The 28-day strength of 28 MPa and the elastic modulus of 2.5 x 104

MPa are adopted for the concrete material. Piles 406 and 411 are embedded in the interbedded sandstone

and mudstone. It is noted that using the average stiffness coefficient a value of 1.43 would overestimate the pile head settlements. A stiffer a value of 2.15 has been adopted in the analysis of Piles 406 and 411 and results of analysis are shown in Figs. 11 & 12.

Table 3. Summary of working pile load tests.

Pile 407

0

10

20

30

0 5 10 15 20Pile head settlement mm

Load

at p

ile h

ead

MN

ObservedCalculated- Normal toeCalculated- Soft toe


Pile 314

0

10

20

30

0 5 10 15 20 25 30Pile head settlement mm

ObservedCalculated- Normal toeCalculated- Soft toe


As shown in Fig. 7, a factor of 1.5 for the average stiffness coefficient a is within the upper bound of the normalized ground stiffness for the interbedded sandstone and mudstone. The reason for greater ground stiffness along the shaft of the working piles would probably due to their longer lengths. The working piles in sandstone and mudstone are 27 m to 36 m while the preliminary test piles are 12 m to 24 m in length. The ground stiffness for deeper rock strata is stiffer.

Piles 407 and 314 are embedded in the sandstone. The secant moduli for a normal toe and for a soft toe that summarized in Table 2 are adopted in the analysis. Results of analysis are shown in Figs. 13 & 14, indicating that Pile 407 would have a normal toe and Pile 314 would encounter a soft toe.

Sets of stiffness parameters that are actually adopted for analyzing the working pile load tests are summarized in Table 4.

Results of analysis of working test piles show that at the maximum test load, the load transferred from the top to the base of the pile is 14 % and 20 %, respectively, for Piles 407 and 314. For Piles 406, 411 and 413, virtually zero percentage of the loads at the pile heads were transferred to the bases. Due to their longer lengths and higher ground stiffnesses, Piles 406, 411 and 413 behave essentially as friction piles.

Pile length

Maximum test load

Maximum displace.

Residualdisplace.Pile

no.

Type of Rock socket m MN mm mm

406 Ss&Ms 27.57 15.36 6.16 0.71 407 Ss 33.11 20.40 15.30 6.79 411 Ss&Ms 36.37 22.66 9.90 2.51 413 Ss&Ms 36.33 21.30 9.15 2.35 314 Ss 27.02 18.30 25.60 19.80

222

Table 4. Stiffness coefficients adopted for working pile load test analysis.

6 DISCUSSIONS

6.1 Load Transfer Parameters

As the results of analysis for the 5 working pile load tests closely agree with those observed pile head settlements, the sets of stiffness parameters interpreted from the preliminary pile load tests are verified and refined. The parameters shall generally be applicable to pile head displacements not exceeding 240 mm, the maximum displacement observed at the preliminary tests.

Since pile load tests are expensive, the ground stiffnesses for various rock materials could be estimated from a set of in-situ tests that would comprise vibration test, pressuremeter test and plate bearing test. This set of in-situ tests would cover ground strains ranging from 10-6 to 10-2. The pressuremeter test result could be analogue to the ground stiffness of the pile-rock interface along the shaft. The plate bearing test could be used for estimating the modulus of the rock beneath the base of the pile. Correlation of in-situ tests with pile load test results would be worthwhile for future research.

6.2 Measures for Soft Toe Effect

Based on results presented in Figs. 3 and 4, it appears that those preliminary test piles that embedded pre-dominantly in sandstones would be susceptible to the soft toe effect. This soft toe phenomenon is well recognized and is reported in the literature. Contingency measures such as pile base grouting or contact grouting could be considered in the design stage in order to achieve satisfactory pile performance. With the load-transfer method as an analytical tool, sensitivity studies on the performance of the pile with or without a soft toe can be conducted. The cost-effectiveness of a pile with a longer shaft or the beneficial effect for base grouting can be assessed.

Mitigation measures to overcome the potential soft toe effect that actually adopted in one of the HSR project was to provide additional lengths of the piles. Piles with as-constructed lengths of 15 to 30 percent longer than those specified in the working drawings were installed. Results of working pile load tests demonstrate that good performance has been achieved. As diagnosed in Figs. 11 to 14, piles socketed in the interbedded sandstone and mudstone have exceptionally low residual settlements of less than 3 mm after un-loaded from the 2 times design normal loads. For piles embedded in the sandstone and for which the soft toe effect is dominant, the residual pile head

settlements after releasing from 2 times design normal loads do not exceed 20 mm.

In view of the fact that 3 preliminary test piles and 1 working test pile experienced the soft toe phenomenon, it is suggested that larger factors of safety for assessing allowable pile base capacity could be considered in future pile foundation design when the conventional static approach is adopted.

7 CONCLUSIONS

Based on results of studies of 10 preliminary and 5 working pile load tests on bored piles embedded in weak rocks of the sandstone and the interbedded sandstone and mudstone, the following concluding remarks could be drawn: (1) The load-displacement characteristics of weak rocks of

sandstone and the interbedded sandstone and mudstone could be described by the power function relationship.

(2) In addition to the widely recognized characteristics of the decreasing stiffness with increasing displacement, the ground stiffness along the pile-rock interface of the shaft and the secant modulus beneath the base of the pile have the tendency to increase with depth.

(3) With reliable ground stiffness parameters that derived from the preliminary pile load tests, the load-transfer method is a useful tool for analyzing pile performance.

ACKNOWLEDGMENTS

The authors express thanks to Hyundai-Chung Lin-Zen Pacific Joint Venture and Hyundai-Chung Lin Joint Venture for providing pile test results that are presented in this paper. Sincere gratitude should be given to Dr. R.N. Hwang of Moh and Associates, Inc. for his valuable comments on the manuscript.

REFERENCES

AASHTO 1996. Standard Specifications for Highway Bridges. 16th Edition.

Chang, M.F. & Broms, B.B. 1991. Design of bored piles in residual soils based on field performance data. Canadian Geotechnical Journal 28: 200-209.

Coyle, H.M. & Reese, L.C. 1966. Load transfer for axially loaded piles in clay. Journal of the Soil Mechanics and Foundations Division, ASCE 92(2): 1-26.

Moh, Z.C., Chang, M.F. & Hwang, R.N. 1995. Load transfer in piles during load reversals. Proceedings of the 10th Asian Regional Conference on Soil Mechanics and Foundation Engineering, Aug 29-Sept 2, Beijing, China.

Stroud, M.A. 1974. The standard penetration test in sensitive clays and soft rock. Proceedings of the 1st European Seminar on Penetration Testing 2(2): 366-375. Stockholm.

Vijayvergiya, V.N. 1977. Load settlement characteristics of piles. Proceedings of Ports�77 Conference 269-284. Long Beach, CA.

Shaft Base Pile no.

Type of rock socket a m a m

406, 411 & 413

Ss&Ms 2.15 0.759 85.5 0.312

407 & 314 Ss- Normal toe Ss- Soft toe

1.12 0.749 6.36 1.64

0.560 0.560


223

Optimization of Pile Foundation Design Through Full-Scale Pile Load Tests in Taiwan High Speed Rail Project

S.W. Duann Moh and Associates, Inc., 11F, No.3, Tunhwa South Road, Section 1, Taipei 105, Taiwan [email protected]

M.S. Chen Moh and Associates, Inc., 11F, No.3, Tunhwa South Road, Section 1, Taipei 105, Taiwan [email protected]

T.H. Seah MAA Geotechnics Co., Ltd., 39/165-7 Moo 13, Lat Phrao Road, Lat Phrao, Bangkok, 10230, Thailand [email protected]

M. Fujita Evergreen/Shimizu Joint Venture of Contract Lot C291

Abstract: The Taiwan High Speed Rail (THSR) project is one of the largest BOT projects in terms of scale and construction cost in the World. The rail was constructed to link two main cities in Taiwan from the north to the south. The terrain along route varies enor-mously from the northern mountainous zone to flat plain in the south. Approximately three-quarters of the route are carried on via-ducts and bridges with over 30,000 piles supporting the structures. Having large number of piles to be constructed, there is a great benefit to optimize the pile design. The full-scale pile load tests were executed to determine the actual shearing resistance of the local soil. The test piles were instrumented with rebar stress transducers installed at different depths. The results of the pile load tests in compression, tension and lateral direction were evaluated. Some correlations between the unit skin friction and the SPT N values were established for various types of soil.

1 INTRODUCTION

1.1 General

For Contract Lot C291 of the Taiwan High Speed Rail (THSR) project stretching from Chainage TK284 to TK312 (totaling 28 km) constructed by the Joint Venture Shimizu-Evergreen and de-signed by Moh and Associates, Inc., the foundation of the struc-tures was supported by large bored piles due to the poor subsoil conditions. The diameters of bored pile used were 1.8 and 2 m, depending on loading requirements and subsoil conditions. Fac-tors, including environment impact, economic assessment and engineering issues, were taken into consideration in the design of pile foundation. Due to variation in the depth of sound bearing strata for the piles over the entire route, the uncertainty in con-struction quality together with economical consideration, an ex-tensive pile load test program was adopted to determine appro-priate design correlation.

The preliminary static pile load tests were performed at four (4) locations; namely Chainage TK287, TK299, TK307 and TK311. The interpretation and evaluation of static compression pile load tests, tension pile load tests as well as lateral pile load tests are summarized in this paper. The results were compared with estimates made based on published data, with focus on the correlation used.

During the course of pile load testing, it was observed that the measured pile capacities at some locations were much lower than the initial estimates, leading to re-evaluation of pile installation

method. After further investigation, it was concluded that low measured pile capacities were primarily due to caking effect of the sand layer causing large reduction in the skin friction of the sandy soil. Two (2) additional compression pile tests were im-mediately performed with improvement in construction methods, such as introduction of polymer to reduce caking, shortening un-necessary construction time during pile installation and introduc-tion of multistage toe grouting etc. The improvement had led to a drastic increase in the pile capacity.

1.2 Objectives

The primary objectives of the pile load tests were to optimize the pile design by establishing appropriate correlation based on the pile load tests through analyzing the distribution of skin friction along the pile and the end bearing during loading. The pile load test program was summarized in Table 1, including six (6) com-pression in conjunction with four (4) tension and two (2) lateral pile load tests. The piles tested in compression and in tension were instrumented with rebar stress transducers at seven (7) to eight (8) different depths together with surface movement meas-urements via displacement transducers. These instrumentations enable one to determine the skin friction along the pile between the adjacent rebar stress transducers as well as the end bearing.

In the initial plan, four (4) compression tests were proposed, due to unsatisfactory results of compression tests, two (2) addi-tional compression tests were conducted with improvement in the

224

construction method, including the use of polymer, shorter con-struction period and better toe grouting technique etc. Two (2) lateral pile load tests were conducted to determine the lateral re-sistance of the tested piles at selected locations (TK299 and TK311); the results of the tests were also compared with the pre-dictions based on design soil parameters, in particular, the hori-zontal modulus of subsoil reaction.

Table 1. Pile load test program.

Test No.

Static Pile Load Test

Tested Load (ton)

Pile Di-ameter (m)

Pile Length (m)

Remarks

Compression 2,209 Poor resultsTP-1

Tension 2,1041.8 65.0

TP-1B Compression 4,238 2.0 64.0 Additional

Compression 2,833

Tension 2,069TP-2

Lateral 351

2.0 62.0

Compression 2,807 Poor resultsTP-3

Tension 2,1072.0 56.5

TP-3B Compression 4,206 1.8 60.0 Additional

Compression 4,118

Tension 1,633TP-4

Lateral 452

1.8 54.5

2 GEOLOGICAL CONDITIONS AND PILE LOAD TEST PROGRAM

2.1 Subsoil Conditions

The subsoil conditions along the project route from Chainage TK284 to TK312, consist of mainly alluvial deposits except at some locations with presence of mudstone at deeper depth. The THSR alignment of Lot C291 lies in the Chianan coastal plain with elevations ranging from 5 m to 30 m MSL. The slope of ex-isting ground surface is approximately between 1/250 and 1/350. This flat alluvial plain can be divided into a number of distinct topographical zones based on the ground relief. They are, from north to south; Shanhua Terrace, Tawan lowland and Chungchou terrace. The top layer of the subsoil along the alignment is the Holocene alluvium consisting of lagoon and marsh deposits and Tainan formation. Both are composed of uncemented sand, silt and clay. The lagoon and marsh deposits contain high clay con-tents. Underlying bedrocks are the Liushuang formation, the Er-chungchi formation of Pleistocene, and the Gutingkeng formation of Pliocene. Since the area along the alignment is covered by al-luvium, the underlying bedrock does not have any outcrop.

Under these geological formations and structural design re-quirements, four (4) locations have been selected for the prelimi-nary pile load tests. Detailed soil investigation was carried out for estimating the initial load capacity of the pile as well as for determining appropriate pile length for testing at each test loca-tion. The soil profiles at the test locations are presented in Fig. 1.

Fig. 1. Soil profiles at pile load test location.

2.2 Testing Program

Efforts had been placed on the compression pile load tests be-cause of the importance in obtaining the skin frictions and end bearing of the piles through measurements made by stress rebar transducers

2.3 Pile Details and Construction Records

All piles were installed by reversed circulation method with summary of construction records given in Table 2. It should be noted that all test piles were constructed without any use of polymer as stabilizing agent, except for Test Pile Nos. TP-1B and TP-3B.

Toe grouting was carried out on all compression piles, by in-jecting grout under pressure through sleeved pipes uniformly spaced over the base of the pile in order to recompact any soil loosened during boring and to mobilize the working toe resis-

288K 296K 304K 312K

Chanage (km)

SCALE (km)

0 2 4SANDCLAY

Legend:

Note: Number next to borehole denotes SPT N value

MUDSTONE

Location 2TP-2

43325757930283026292929181121382014142217181717262626316188634840352932373746

Location 1TP-1 and TP-1B

447676671010141221222423393437

3229694546293765655926

57283144572539396363635932

Location 3TP-3 and TP-3B

10621824292233272833272442363944464851552817201936182629344134212421243638222023273034698184

Location 4TP-4

716515121918282026282521122224121411141212141513173829

100

100

100

100

+20

+10

0

-10

-20

-30

-40

-50

-60

225

tance of the pile. Two (2) grout line circuits, each permitting grout to return to the pile top and suitable for flushing were pro-vided. Each circuit had four sleeved sections below the pile base. The sleeved sections were positioned on a frame at the end of the reinforcement cage assembly so as to be placed in close contact with the soil surface. Each sleeved section had eight 4-mm di-ameter holes covered by tightly fitted rubber sleeves. An initial opening of the sleeves was carried out within 48 hours, but not less than 24 hours, of concreting. Pressure was released immedi-ately when cracking had achieved to limit water injection to a minimum. Grout was injected in doses through each grout line circuit. Lines were thoroughly flushed after each injection to al-low for later injections. All the lines were injected in turn with the specified dose or until a maximum pressure of 60 bars was maintained for 5 minutes. Second and third injections were con-ducted 6 hours after previous injection. Rounds of grouting were continued until a total of 1000 liters had been injected, or until all lines were sustaining the required pressure or until pile uplift exceeding 3 mm had been detected.

The first four (4) compression tests (TP-1, TP-2, TP-3 and TP-4) were conducted under single stage injection due to inappropri-ate rubber sleeves. Additional two (2) tests (TP-1B and TP-3B) were performed with multiple injections, resulting in stiffer end bearing response.

Table 2. Construction records of compression piles.

Location TP-1 TP-1B TP-2 TP-3 TP-3B TP-4

Diameter (m) 1.8 2.0 2.0 2.0 1.8 1.8

Length (m) 65.0 64.0 62.0 56.5 60.0 54.5

Use of Polymer No Yes No No Yes No

Toe Grouting Single Stage

Multistage

Single Stage

Single Stage

Multistage

Single Stage

Installation Stages Time Taken (Hours)

Drilling 15.5 9.0 9.0 8.5 7.5 9.0

Halting Time 33.0 0.0 38.0 14.0 0.0 14.0

Base Cleaning 2.0 0.5 0.5 0.5 0.5 0.5

Halting Time 1.0 1.0 1.0 0.0 0.5 0.5

Sonic Logging 1.0 0.5 0.5 1.0 0.5 0.5

Halting Time 0.0 0.0 1.0 0.0 0.5 0.0

Caging 11.0 6.0 15.0 6.0 4.0 6.0

Halting Time 0.0 0.0 0.5 0.0 0.0 0.0

Tremie Pipe In-stallation 0.5 1.0 0.5 1.5 0.5 2.0

Base Cleaning 0.5 1.0 0.5 1.0 0.5 2.0

Halting Time 0.0 0.0 0.5 0.0 0.5 0.5

Concreting 4.0 5.0 5.0 3.5 3.0 3.0

Total Time 68.5 24.0 72.0 36.0 18.0 38.0

2.4 Testing Conditions

At four (4) test locations, the compression test piles were sub-jected to external applied load at the top of the pile by means of

nine (9) 500-ton hydraulic jacks against a reaction frame. This frame was fixed in place by four (4) anchor piles. During the test, the pile was loaded incrementally. The loads and displace-ments were recorded periodically by means of nine (9) load cells and eight (8) displacement transducers with four (4) on each test pile and one (1) on each anchor pile. A typical pile testing ar-rangement is illustrated in Fig. 2.

The tension test piles were subjected to external applied load through extended rebars from top of the pile to a beam with four (4) 500-ton hydraulic jacks resting on top of the beam. The frame was placed on four (4) reaction piles. During the test, the pile was loaded incrementally. The loads and displacements were recorded periodically by means of four (4) load cells and displacement transducers on test pile.

For lateral pile load tests, a hydraulic jack was pushed against a test beam supported by two (2) supporting piles as reaction. The test piles were loaded incrementally up to 350 ton for TP-2 and 450 ton for TP-4. They were located in different areas with different upper soil conditions, that is, for test Pile No. TP-2L, the upper soil consists of mainly clay. Whereas for test Pile No. TP-4L, the upper 20 m of soil layer consists of sand with thin layer of clay. During each load increment, the load, the horizon-tal displacement at pile top and the horizontal deformation profile were measured.

Fig. 2. Typical layout of static pile load test.

3 PILE LOAD TEST RESULTS

3.1 General

The results of the pile load tests, both compression and tension, have been compiled, and analysis has been conducted to evaluate the performance of the piles as well as to establish appropriate correlation with soil parameters. Emphasis has been placed on compression pile load tests since the tests gave both skin friction and end bearing of the piles. Non-uniform strains were observed between the reinforcement and concrete during loading in the tension tests, resulting in less reliable data interpretation.

The results of the pile load tests have been analyzed and the process of analysis is summarized in Figure 3 in the form of a flowchart.

C B

D E

S8 S5

S7 S6

S4

S3

S1

S2A

C3 C1C2

11.3 m

Rebar Stress Transducer

Compressionor Tension Pile

Pile

Legend :

S - DisplacementTransducers

C - Telltale

A - Rebar StressTransducer

226

Fig. 3. Flowchart of instrumented pile load test analysis.

3.2 Load Settlement Relationships

The load settlement curves of the compression and tension tests are presented in Figs 4 and 5, respectively. The initial portion of the compression curves exhibit very similar behavior up to 1,200 ton with corresponding settlement of approximately 1 cm, except for test pile TP-3B with stiffer response. Test piles, TP-1 and TP-2, have very distinct gradients beyond the yield points, indi-cating very low end bearing resistance.

For tension tests, the piles were tested to failure or maximum capacity of the equipment. The maximum applied loads for all four (4) tension tests range from 1,633 ton to 2,107 ton. Two (2) test piles, TP-3T and TP-4T, exhibit large movement beyond the yield points as shown in Figu 5. The other two (2) tests were terminated at the maximum equipment capacity of 2,100 ton, having final movement of around 5 cm.

Fig. 4. Load settlement curves for compression pile load tests.

Fig. 5. Load heave curves for tension pile load tests.

3.3 Stress-Strain Behavior

The concrete modulus of the pile was estimated based on the re-bar stress transducer response at 1.5 m depth. At each location, the concrete modulus versus strain relationship is presented as shown in Figures 6 and 7. For all compression piles except for Test Pile No. TP-3B, the relationship between estimated concrete modulus and strain is considered reasonable. For Test Pile No. TP-3B, the estimated concrete modulus is relatively high in the range of 0.38 to 0.41x 106 kg/cm2.

The stress rebar transducers installed at various depths along the pile were used to measure the stresses in the reinforcement and the corresponding strains could be estimated. The force of the pile at that level was computed based on estimated sectional modulus of the pile, which was a function of sectional properties of pile. It should be noted that not all rebar stress transducers behave accordingly; therefore it is often necessary to eliminate the abnormal rebar stress transducers. The malfunction of the re-bar stress transducer could be caused by short-circuit, damage in the cable and transducer during caging to concreting. An exam-ple of the load distributions of Test Pile Nos. TP-1 and TP-1B is presented in Figure 8, showing decrease in load with increasing depth.

For the tension tests, several stress rebar transducers at shal-low depths had abrupt change in the stress during loading, indi-cating some separation of reinforcement and surrounding con-crete. This sudden change in the stresses of the rebar hampers the analysis because of difficulty in determining the effective cross sectional area of intact concrete at those depths, therefore assumptions with respect to the intact concrete area has to be made. It is also observed that the separation begins at axial strain

area of about 30% is applied for computation of section modulus whenever the stress rebar transducer had experienced a sudden increase. This 30% reduction was obtained by trial and error on all tension test results.

Fig. 6. Estimated concrete modulus with strain in compression.

START

Input Field Data and Section Properties

Trend of Ec with StrainAcceptable ?

Yes

No

Compute Estimated Skin FrictionSelect AppropriateRebar Stress Data

and Section Properties

Compute Concrete Modulus, Ecfrom Rebar Stress Data

@ 2m Depth

Compute Section Modulusat Various Depth

Compute Shorteningof Given Section and

Compare with that from Telltate

Compute Movementof Each Point

Compute Force at Given Depth

Compute Skin Frictionat Given Depth

Compare Estimated Skin Frictionwith Measured Skin Friction

Acceptable ?

END

Recompute Soil Parameter

Yes

No

0

50

100

150

200

250

0 1,000 2,000 3,000 4,000 5,000 Load (ton)

TP-1TP-1BTP-2TP-3TP-3BTP-4

0

50

100

150

200

0 500 1,000 1,500 2,000 2,500 Load (ton)

TP-1T

TP-2T

TP-3T

TP-4T

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0 100 200 300 400 500 600 700Strain (10-6)


227

Fig. 7. Estimated concrete modulus with strain in tension.

Fig. 8. Load distribution curves of TP-1 and TP-1B.

4 ENGINEERING EVALUATION

4.1 General

The measured unit skin frictions of the soil layers and unit end bearing for each compression test were estimated based on the test results. Increase in the pile capacity due to improvement in construction was also quantified and recommendations were made accordingly.

Results of the tension tests were compared with the compres-sion tests to determine the mobilized skin friction in upwards and downwards directions. Assumptions were made in estimating the force within the pile due to the problem of non-uniformity in the tension tests, hence the measured skin frictions in tension mode are considered less reliable. Since most of the piles were lifted to over 4 cm, significant portions of the piles had mobilized their skin frictions. Therefore, the pulling resistance of the pile could be compared with the friction resistance of the pile in compres-sion.

The results of two (2) lateral pile load tests were also com-pared with predictions made by LPILE computer software. A brief description of the software is provided in latter section.

4.2 Estimation of Skin Friction

From the load distribution of the pile, the average skin friction between two (2) adjacent rebar stress transducer depths can be estimated. When the load on the pile is increased, the skin fric-tion will gradually increase up to a certain limit. Strain softening or hardening behavior may be expected depending on the type of soil at those depths. Plots of mobilized skin friction versus movement of Test Pile Nos. TP-1 and TP-1B are shown in Figs 9a and 9b, respectively. It was observed that in most cases, the skin friction reaches a peak at about 1 to 2 cm movement, indi-cating that the skin friction of the pile will be fully developed un-der this movement. The maximum skin frictions obtained are presented in Fig 10.

Fig. 9a. Mobilized skin resistance versus movement of TP-1.

Fig. 9b. Mobilized skin resistance versus movement of TP-1B.

Fig. 10. Skin friction profiles at test location 1 (TP-1, TP-1B).

0.0

0.1

0.2

0.3

0.4

0 200 400 600 800 1000 1200Strain (10-6)

TP-1T

TP-2T

TP-3T

TP-4T

0

10

20

30

40

50

60

70

0 1,000 2,000 3,000Loading (ton)

239 ton

721 ton

956 ton

1258 ton

1599 ton

1905 ton

2001 ton

2155 ton

TP-1

0 2,000 4,000 6,000Loading (ton)

241 ton

720 ton

961 ton

1592 ton

1997 ton

2400 ton

3198 ton

4101 ton

TP-1B

0

5

10

15

20

0 1 2 3 4 5 6Movement (cm)

1.5 m-11 m 11 m-19 m 19 m -32.5 m 32.5 m-37.5 m37.5 m-48.5 m 48.5 m-58 m58 m-64.5m

0

5

10

15

20

25

0 1 2 3 4 5 6Movement (cm)

1.5 m-12 m12 m-19.4 m19.4 m-33.3 m33.3 m-38.3 m38.3 m-47.8 m47.8 m-58.8 m58.8 m-64m

0

10

20

30

40

50

60

70

0 10 20 30Friction Resistance (ton/m2)

TP-1TP-1BTP-1T

0

10

20

30

40

50

60

70

0 50 100SPT N value

SPT N

LowerLimit

228

From the results of compression tests, the following observa-tions have been made:

Clay: By assuming linear relationship between skin friction and SPT N value for clay layers as shown in Figure 11a, the adhesion factor versus undrained shear strength rela-tionship can be established as presented in Figures 11b. The results indicated slightly lower adhesion factor for TP-1 and TP-3, and greater adhesion factor for TP-2 and TP-4, compared with published adhesion factor. With improve-ment in construction method, the analysis shows much higher adhesion factor for TP-1B and TP-3B, and it had exceeded unity in low strength range, and the most likely reason for this high factor is due to under-estimation of undrained shear strength.

Sand: Plots of skin friction versus SPT N value are made and linear relationships are assumed as presented in Figure 12. For Test Pile Nos. TP-1 and TP-3, the ratio of skin friction to SPT N was only 0.15, compared with values of 0.3 to 0.33 for TP-2, TP-4, TP-1B and TP-3B. The caking effect seems to reduce the skin friction substantially; there-fore the use of polymer together with shorter construction time had improved the shaft friction considerably.

Mudstone: Two (2) sets of rebar stress transducers were used to measure the skin friction of the mudstone at loca-tion 4 (TK311). Both measurements give a unit skin fric-tion of 16.8 ton/m2. It should be emphasized that since significant variation in the properties of mudstone was usu-ally expected; therefore a skin friction of 16 ton/m2 was

adopted in the design for mudstone layer. Fig. 11a. Measured unit skin friction versus SPT N value for clay Layers.

Fig. 11b. Adhesion factor versus undrained shear strength for clay Layers.

Fig. 12. Measured unit skin friction versus SPT N value for sand layers.

4.3 Unit End Bearing Response

The response of the end bearing was determined from extrapola-tion of the last set of rebar stress transducers to the pile tip. Fig-ure 13 shows the increase in unit end bearing with tip movement for all six (6) compression piles. Poor end bearing responses were encountered for test Pile Nos. TP-1, TP-2 and TP-3, except for test Pile No. TP-4 with tip embedded in mudstone, reaching unit end bearing of over 700 ton/m2 at movement of 7 cm.

With better grouting technique (multi-stage injection) in TP-1B and TP-3B, stiffer end bearing response has been achieved compared with other piles. Unfortunately, the end bearing of these two (2) piles did not reach full capacity due to limitation in the maximum applied load. Nevertheless, the unit end bearing seems to increase significantly with increasing movement for TP-3B. The results clearly indicated the importance of base (or toe) cleaning and the effectiveness of toe grouting as well as the pres-ence of sound bearing stratum. Normally, less emphasis has been placed on the end bearing and higher factor of safety is adopted for the end bearing compared with the skin friction.

Fig. 13. Unit end bearing versus pile tip movement.

4.4 Improvement due to Better Construction Methods

The change in construction method gives rise to significant im-provement in the performance of pile as shown in the results of Pile Nos. TP-1B and TP-3B. Figures 11 and 12 compare the measured skin frictions of test piles (Test Pile Nos. TP-1, TP-3, TP-1B and TP-3B) under two different construction methods, in-dicating that most of the soil layers, both sand and clay, have

0

5

10

15

20

0 10 20 30 40 50 60SPT N Value

TP-1 TP-1BTP-2TP-3 TP-3BTP-4

Average for TP-1, TP-3Average for TP-1B, TP-3BAverage for TP-2, TP-4

0.0

0.5

1.0

1.5

2.0

0 10 20 30 40Undrained Shear Strength, cu (ton/m2)

TP-1TP1BTP-2TP-3TP3BTP-4


0

10

20

30

0 10 20 30 40 50 60SPT N Value


Slope=0.33

Slope=0.3

Slope=0.15


0

50

100

150

200

0 200 400 600 800Unit End Bearing (ton/m2)


229

significant increase in the unit skin frictions. The improvement in skin friction is about 1.5 to 2.5 times higher than values from TP-1 and TP-3, this is probably due to a combination of polymer usage, shorter construction time etc. Therefore, it should be em-phasized that the good construction technique and control in in-stallation are essential in bored pile construction.

Since Test Pile Nos. TP-1 and TP-3, were injected with single stage grout and test Pile Nos. TP-1B and TP-3B were subjected to multi-stage grouting, therefore the effect of toe grouting meth-ods could also be evaluated. Figure 14 shows the unit end bear-ing response with pile tip movement, it can be seen that only TP-3B has substantial increase in the bearing resistance. The im-provement ratio for TP-3B compared with TP-3 is much higher; whereas TP-1B did not show any increase in the end bearing up to 20 mm of pile tip movement. Due to the lack of soil data be-low the pile tip at TP-1 location, it is not possible to evaluate the full performance of TP-1B. From the results of TP-3B, it can be concluded that multi-stage toe grouting will improve the end bearing significantly.

Fig. 14. Effect of toe grouting type on unit end bearing.

4.5 Comparison between Compression and Tension Tests

The maximum mobilized skin friction in compression is also compared with the results from the tension piles as shown in Fig 15. As mentioned in earlier section, the test piles, TP-1T and TP-2T, were not tested to �failure� due to limitation in testing equipment, higher tension resistance can be expected from these tests as also indicated in the mobilized skin friction distribution. Figure 15 presents comparison between the compression and mobilized tension frictions; the results indicated that the mobi-lized skin friction of the piles in tension and the maximum skin friction in compression are in certain agreement.

From the results of ultimate pile capacities, the ratio of ulti-mate tension capacity to compression pile capacity ranges from 62% to 105%, having an average of 81%.

4.6 Lateral Pile Load Tests

The results of the lateral pile load tests were compared with the predicted results as shown in Fig 16 at locations 2 (TK299) and 4 (TK311). The lateral resistance of pile under lateral loading was estimated by means of �p-y� curve method. In this method, a non-linear load (p) versus horizontal displacement (y) of soil layer for given pile diameter is estimated based on certain input soil parameters, such as undrained shear strength, undrained modulus, friction angle etc. The p-y curves of various soil layers are then generated for a given problem. Once these load-displacement curves are established for the soil layers, Winkler

beam theory (beam on springs) is applied to analyze the problem. Since each spring has a non-linear behavior, numerical approach, such as finite difference method, has been adopted to solve this problem. It should be noted that the major difference in this method compared with other conventional approach is the as-sumption on stress-strain behavior. Conventional approach as-sumes that the soil behaves elastically with and without consid-eration of ultimate soil resistance; the basic input soil parameters are horizontal coefficient of subgrade reaction or modulus and soil resistance or strength. Generally, the non-linear p-y curve approach yields higher movement than elastic approach at high stress level and small difference at small stress level. Therefore, large difference in lateral displacement may be expected if the stress level is high. A commercial computer program, named LPILE, incorporated the above p-y method has been used for predicting the behavior of laterally loaded pile.

Figure 17 presents the predicted and measured lateral move-ment along the pile, TP-2L, the result indicated very good agree-ment.

For the actual pile foundation, the pile tops will be connected together by the pile cap, giving a fixed head condition which will behave differently as in the free head pile load test. Therefore, the test results are mainly used to ensure that the design soil pa-rameters and model used are acceptable.

Fig. 15. Comparison of mobilized skin friction between tension piles and compression piles.

Fig. 16. Horizontal load versus pile top movement of TP-2L and TP-4L.

0

100

200

300

400

0 10 20 30 40 50Pile Tip Movement (mm)

TP-1 (Single-Stage Grouting)TP-1B (Multi-Stage Grouting)TP-3 (Single-Stage Grouting)TP-3B (Multi-Stage Grouting)

0

5

10

15

20

0 5 10 15 20Mobilized Compression Skin Friction (ton/m2)

Clay TP-1 Sand TP-1Clay TP-2 Sand TP-2Clay TP-3 Sand TP-3Clay TP-4 Sand TP-4Mudstone TP-4

0

100

200

300

400

500

0 50 100 150 200Pile Top Movement (mm)

TP-2 MeasurementTP-2 PredictionTP-4 MeasurementTP-4 Prediction

230

Fig. 17. Lateral movement under horizontal loading of TP-2L.

5 RECOMMENDATIONS AND CONCLUSIONS

The results obtained from the compression, tension and lateral load tests have been analyzed, and the following recommenda-tions and conclusions are obtained:

Construction methods: In this series of pile load tests, the construction methods of bored pile installation were im-proved to a great extent through the use of proper stabilizing agent (polymer), shorter construction time, multi-stage toe grouting etc. The pile load test results clearly demonstrated the increase in pile capacity with these improvements. High skin friction was achieved through reduction in caking effect of the sand layer with the introduction of polymer as stabiliz-ing agent. Shorter construction duration reduces the soften-ing of clayey soil due to swelling. Multistage toe grouting ensures that any soft toe caused by inadequate toe cleaning or sedimentation of soil at the bottom of excavated holes can be improved via injection of high-pressure grout in stages to strengthen the weak zone around the pile toe area. The im-proved method of construction was adopted as a standard procedure for the installation of the working piles. Skin friction from compression tests: The stresses from the rebar stress transducers at different depths together with compression load and surface settlement, were used to com-pute the unit skin frictions and end bearings of the soil for the test piles. Based on analysis of the pile load tests, the rec-ommended formulae for estimation of pile capacity are sum-marized in Table 3. The followings can be concluded: End bearing response: The development of end bearing is relatively small for most of the compression piles (except for

TP-4 and TP-3B), that is, the gain in end bearing capacity is rather insignificant compared with the skin friction. The con-tribution of the end bearing at 3 to 4 cm movement ranges from 9 to 25% to the total capacity, depending on the pile length and embedded soil stratum. The performance of the end bearing is greatly affected by the construction methods as well, hence in current adopted construction method; it can be seen that stiffer response in the end bearing was achieved through multi-stage toe grouting provided that the bearing layer is sandy soil. The recommended correlation is summa-rized in Table 3 for different soil types. Skin frictions in tension and compression: From the test results, the ratio of skin friction in tension mode to the skin friction in compression mode ranges from 62% to 105%, hav-ing an average of 81%. Performance of lateral pile load tests: Two (2) lateral pile load tests had been conducted with different upper soil condi-tions. The measurements give relatively similar response as predicted by LPILE computer program, indicating reasonable input soil parameters as well as model used. Therefore, the method of analysis was adopted in design.

Table 3. Recommended correlation for estimation of shaft fric-tion and end bearing.

SoilType

Shaft Resistance, fs

End Bearing, qb

Cohesive Soil

fs = cu 12 ton/m2

where = Adhesion factor

= 0.21+2.6/cu1

qb = 9 cu 160 ton/m2

where cu = Undrained shear strength (ton/m2)

Non-cohesive Soil

fs = N/3 15 ton/m2

where N = SPT-N value

qb = 7.5 N ton/m2

N = Average SPT-N value at depths of 4D above and 1D below pile toe. D = pile diameter

Mudstone fs = qu ton/m2

where = 0.4

fs = 16 ton/m2

for N = 50~100 qu = Uniaxial com-pressive strength of intact rock

qb= tan2(45+ /2)+1 quwhere = Friction angle of rock

qb 160 ton/m2 for N = 50~100

ACKNOWLEDGMENTS

The authors acknowledge technical assistance and review pro-vided by colleagues. They are indebted to Mr. Norawat Rattana-rungsan for preparing the final manuscript. Appreciations are due to the Taiwan High Speed Rail Corporation and Moh and Associates, Inc. for their permission to publish this paper

0

5

10

15

20

25

30

-50 0 50 100 150 200Lateral Displacement (mm)

60 ton (Prediction)

120 ton (Prediction)





60 Ton (Measurement)

120 ton (Measurement)





CL

Su = 5 t/m2

= 2.0 t/m3

SM

= 32o

= 2.0 t/m3

SM, = 32o

= 1.9 t/m3

CL

Su = 20 t/m2

= 2.0 t/m3


231

Three-Dimensional Analysis of Load Distribution in Pile Group

B. Ukritchon & P. Rungbanaphan Department of Civil Engineering, Chulalongkorn University, Pathumwan Rd., Bangkok 10300, Thailand [email protected]

Abstract: The objective of this research is to study the pile load distribution in the pile group with rigid caps subjected to the static vertical load by means of a three-dimensional finite element model. A computer program, �PILE3D� was developed based on the basic program of Three-Dimensional Finite Element Method proposed by Smith & Griffiths (1999). The program improvements include additional capabilities to mesh generation process and high efficient fast solver algorithm. In this study, the soil is modeled as an elasto-plastic material under undrained condition (total stress analysis). The groups of 4 piles, 5 piles and 9 piles were selected as case studies. The results show that each pile in the group does not carry the load equally, which is contrast to the standard design practice. Major parameters affecting the load distribution include pile spacing, the length of pile, the number of piles, and the relative stiffness of soil above and below the tip. For the 9-pile group, the corner pile carries the largest load, followed by the mid-edge pile, and the center pile. Similarly, for the 5-pile group, the center pile carries the load much less than the corner pile. The pile group efficiency increases when the pile spacing increases, the number of pile decreases, and the stiffness of end bearing soil layer increases.

1 INTRODUCTION

Pile group is generally used as foundation for high-rise buildings or infrastructures. For design practice, it is generally assumed that each pile in the group shares the applied working load equally. However, in the real situation, the load on each pile of the pile group may not distribute equally because the cap of the pile group acts as a rigid structure. Under the rigid pile cap, the top of each pile settles uniformly but the load acting on each pile is not distributed equally. Important parameters affecting on the load distribution in the pile group include pile-pile interaction behavior, a number of piles, pile arrangement, and pile spacing.

This paper studies the load distribution characteristic of pile group foundations with rigid pile cap subjected to vertical loads. The study is carried out by means of three-dimensional finite element method of pile group. The analysis considers three types of pile group, include 4 pile, 5 piles and 9 piles. Pile spacing ratios and length ratios are also considered in the study. The three three-dimensional finite element analysis used in this research is described in the next section.

2 THREE DIMENSIONAL FINITE ELEMENT METHOD

A computer program, PILE3D, used in this research to study the load distribution of pile group is developed based on three-dimensional finite element techniques (Smith and Griffiths, 1998). Several capabilities of PILE3D are implemented, including the type of element, mesh generation, mesh refinement, boundary conditions, the type of loading, the constitutive material, post-processing of nodal stress recovery method.

The number of piles is one of important parameters controlling the characteristics of finite element mesh used to model the pile group. For example, Fig. 1 shows the horizontal and vertical cross sections of a typical mesh of the structure of 9 piles. It can be seen that finer elements are generated at each pile, forming the center core of the mesh. Elements far from this center core has larger size increasing towards external boundary edges. For vertical loadings, due to problem symmetry, the generation of the

full mesh is not necessary. Only the one-forth of the full mesh is modeled. the use of symmetry reduced the size of the model to a quarter (1/4) of the full mesh. For all meshes, the nodes on external boundaries are allowed to move only in the vertical direction while bottom nodes were fixed in all directions.

Fig. 1. Typical finite element meshes of group for 9 piles.

The initial stress condition through out the mesh is set up by the self-equilibrating geostatic stress field. Subsequent loading on the piles is either single-step or multi-step, axial or lateral, force or displacement and is applied only at the pile head. The piles

232

i=1

n

i=1

n

and soil is modeled by the solid three-dimensional linear isoparametric 4-node tetrahedron elements (three nodes per face). The mesh is generated by a noncommercial mesh generation package, �GMSH� version 1.37 (Christophe & Jean-François, 2003), which is an automatic three-dimensional finite element mesh generator based primarily on Delaunay algorithm. GMSH is effectively incorporated to the developed program, Pile 3D, in which several types of pile group with various soil layers can be modeled.

Mesh refinement is used in certain regions of the finite element model, as shown in Fig. 1, in order to reduce numerical error from finite element approximation and ensure satisfactory accuracy of result. In this problem, the mesh with suitable degrees of refinement, especially in the region surrounding the piles and near the pile head and tip where stress singularity may occur, is automatically generated to smooth stress distribution at those areas and to ensure convergence of the numerical solutions. Thin element along pile proposed by Desai and Zaman is also implemented in PILE3D.

The pile elements are assumed to behave elastically at all times while the soil is modeled as either a linear elastic material or elasto-plastic material, based on Mohr-Coulomb failure criterion. This model can represent soil dilatancy and directly related to the physical soil properties (cohesion intercept and angle of internal friction). The constant stiffness approach is used to model material nonlinearity in the iterative calculation. In this approach, convergence is accepted when stresses generated by the loads satisfy some stress-strain law within prescribed tolerances.

To minimize computational time, a high performance equation solving algorithm of sparse symmetric positive definite system of linear equations, based on Cholesky factorization of the coefficient matrix, is applied in the developed program.

At the end of each step, nodal stresses are recovered based on the Superconvergent Patch Recovery Method (Zienkiewicz & Zhu, 1992; Bathe, 1996). The concept of this stress smoothing method is that the stresses at each node are estimated from the Gauss point stresses of the elements within the area around it (local stress smoothing). The stresses is assumed as a polynomial expansion P

* of the same complete order of the basic shape function N over an element patch surrounding the particular node considered.

P* = <P> {a} (1)

where <P> is a based function in the form of <1, x, y, z> and {a} is a set of unknown parameter expressed as {a0, a1, a2, a3}. The determination of the unknown {a} is made by ensuring a least square fit of P

* to the high accuracy sampling points existing in the patch considered which, in this case, are the Gauss points. The equation can be expressed as follow,

[P(xi,yi,zi)]T[P(xi,yi,zi)]{a} = h(xi,yi,zi)[P(xi,yi,zi)]T (2)

where h represents the Gauss point stress and n is the number of Gauss point within the element patch considered.

The relative error is estimated based on the comparison of the energy norm in order to check numerical error of the finite element analysis. The error distribution can be used as the indicator for the mesh refinement pattern.

The capability of PILE3D is verified by comparing some computed elastic solutions of pile group with those from previous researches. Thus, the selected problem is the three dimensional

analysis of the single pile, with 7.5 m long, 0.3 m diameter circular pile (Young�s Modulus Ep = 4.5x105 ton/m2), embedded in an elastic soil layer of 50 m deep (Young�s Modulus Es = 450ton/m2) and loaded axially at its head. Solutions are compared with those provided by Poulos & Davis (1980). The comparison results show that the head settlement from the finite element analysis is very close to the analytical solution (the difference about 0.5%) and so is the shear stress distribution along the pile shaft, as shown in Fig. 2, where �p� represents the shear stresses and �K� is the pile stiffness factor (Ep/Es)

Fig. 2. Comparison between distribution of shear stresses along the single pile shaft.


Researches on single pile and pile group have been extensively carried out during the last 30 years. Examples include Coyle & Reese (1966), Poulos & Davis (1968), Randolph & Wroth (1978), and Muqtadir & Desai (1986).

For this research, a limited parametric study is carried out using PILE3D with the major objectives of gaining a better understanding of the role of various factors on load distribution characteristic of pile group under the rigid pile cap. The analyses consider the load distribution in the pile group assuming elastic behavior of soil and pile. No interface elements are used between soil and pile elements. Fig. 3 shows example of problem geometry used in this study. Figure 3a illustrates the diagram of 5-pile group of floating type and Fig. 3b represents the end bearing type of 9-pile group. The basic pile properties are Young�s modulus Ep = 2.6x106 ton/m2, Poisson�s ratio p = 0.3, concrete unit weight p = 2.4 ton/m3, pile length L = 20, 30 m, and diameter D = 0.6 m (round piles). The typical finite element mesh for a 9 pile group (one quarter mesh) is shown in Fig. 4.

The elastic soil properties, corresponding to the Bangkok soft to medium clay layer as shown in Fig. 3, are Young�s modulus Es1 = 1600 ton/m2, Poisson�s ratio s1 = 0.45, and soil unit weight

s1 = 1.8 ton/m3. For elastic-perfectly plastic model, the Tresca failure criterion is used with the undrained shear strength cu = 4 ton/m2 and angle of internal friction = 0.

p.(pi)dL/P0 0.5 1 1.5 2 2.5 3 3.5

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Elastic Solution, K = 503D FEM, K = 50Elastic Solution, K = 50003D FEM, K = 5000

L/d = 25Vs = 0.45

p dL/P

z/L

d

P

233

(a) (b)

Fig. 3. Diagrams of models and parameters used in FEM analysis.

Fig. 4. Typical 3D Finite Element Mesh for a 9-Pile Group (One Quarter Problem).

For the first sand layer shown in Fig. 3, the elastic properties are Young�s modulus Es2 = 4000 ton/m2, Poisson�s ratio s2 = 0.3, unit weight s2 = 2.0 ton/m3. The Mohr-Coulomb failure criterion is used with the angle of internal friction = 30o and cohesion intercept c = 0 ton/m2.

Three types of pile group are investigated in this study. The 4, 5, and 9 piles groups, arranged as the pattern shown in Fig. 5, are analyzed with various pile spacing, pile length, and types of layered soil. The pile head settlement of 1 cm is equally constrained to all of the piles in the group and the pile head loadings corresponding to this settlement are obtained from PILE3D. The results of analysis are concluded in the next section.

(a) 4 piles (b) 5 piles (c) 9 piles

Fig. 5. Pile arrangement of the studied groups of piles.

4 RESULTS OF ANALYSIS

It should be noted that with the settlement of 1 cm, the behavior of piles in the studies is still elastic. Figure 6 shows the load distribution on group of 9 piles. The floating pile type is considered with the soil properties of soft to medium clay as described in the previous section. The load on each pile (P) is normalized by the average load of all piles in the group (P m). It can be seen that the greatest loads is carried out at the corner piles, followed by the mid-edge pile and the center pile. However, the mid-edge pile and the center pile carry the load much less than the average value, but the corner pile carries more load than the average value. Figure 7 show the ratio of load distribution in the 9-pile group. In this figure, Pmin correspond to the minimum load of the pile in the group. For practical design s/D = 3, the corner pile and the mid-edge pile carry the load approximately 4.4 and 2.4 times of that of the center pile, respectively. As the ratio of pile spacing diameter ratio (s/D) increases, the difference between the loads on each pile decrease largely. A more uniform load distribution is mainly due to a decrease in effect of adjacent piles in the group. This influence is reduced further with larger spacing distance between the adjacent piles.

Fig. 6. Pile load distribution in the 9-Pile group.

Figure 6 also shows the influence of pile length on the load distribution characteristic. It is obvious that there is a small increase in the difference of the pile load distribution as the length to diameter ratio (L/D) of pile increases. This confirms the fact that the larger stress bulb is generated on the longer pile. The

s

s

D

s

s

D

s

s

D

S

Ep, p, p

First SandEs2, s2, s2, c,

Analysis AreaD

Soft/Medium ClayEs1, s1, s1,Cu

Constant Settlement

S

Ep, p, P

Analysis AreaD

Constant Settlement

Soft/Medium ClayEs1, s1, s1,Cu

XY

Z

0.00

2.00

4.00

6.00

8.00

10.00

12.00

14.00

16.00

18.00

20.00

0.0 2.0 4.0 6.0 8.0 10.0s/D

L/D = 33L/D = 50

center

mid-edge corner

Floating pile D = 0.6 m

s

s

D

234

stress interference of adjacent pile increases, which increase in the pile load sharing.

Fig. 7. Ratio of pile load distribution of 9-Pile group.

Fig. 8. Influence of stiffness of end bearing soil layer on load distribution in groups of 9 Piles.

The effects of bearing layer at the pile tip are investigated as illustrated in Fig. 8. The group of 9 piles is also considered in this study. It should be noted that the soil stiffness ratio, E s2/Es1 of 2.5 corresponds to the case of first sand layer underlain the stiff to medium clay layer as in Fig. 3 (b) while the value of 1.0 represents the case of floating piles. The pile length is 20 m, where the length to diameter ratio is 33. From the Figure, It can be seen that the load distribution tends to more uniform as the ratio of Es2/Es1 increases. This means that the stiffer stratum at tip of pile, the more uniform pile load distribution in the group. Figures 9 & 10 show the results of analyses of load distribution in the 5-pile group of the floating type. These results are very similar to those of the 9-pile group. For s/D = 3, the corner pile carries the large portion of the load much smaller than that of the center pile, approximately by 2.8 times. The corner

pile also takes the load greater than the average value about 10%, but the center pile shares the load less than the average value about 60%. The increase in pile spacing has small effect on the corner pile, but has significant effect on the center pile.

Comparisons of load distribution between groups of 5 piles and 9 piles are shown in Fig. 11. It is obvious that the load distribution of the group of 5 piles is more uniform than that of 9 piles. This result makes sense because as the number of piles in the group increases with the constant pile spacing, the interaction between adjacent piles certainly increases and the effects are more pronounced.

Fig. 9. Pile load distribution of 5-Pile group

Fig. 10. Ratio of pile load distribution of 5-Pile group

The influences of numbers of pile, soil layer types and pile spacing on the total load of pile groups are studied as illustrated in Fig. 11. The figure shows the ratio of the average load of the pile group to the single pile load at the same unit settlement of the pile top.

0.0

0.5

1.0

1.5

2.0

0.0 2.0 4.0 6.0 8.0 10.0

s/D

L/D = 33L/D = 50

center

mid-edge

corner


s

s

D

0.0

0.5

1.0

1.5

2.0

0.0 2.0 4.0 6.0 8.0 10.0

s/D

Es2/Es1 = 1Es2/Es1 = 2.5Es2/Es1 = 10

center

mid-edge

corner

D = 0.6 mL/D = 33

s

s

D

0.0

0.5

1.0

1.5

2.0

0.0 2.0 4.0 6.0 8.0 10.0s/D

L/D = 33

L/D = 50

center

corner


s

s

D

0.0

2.0

4.0

6.0

8.0

10.0

12.0

14.0

16.0

18.0

0.0 2.0 4.0 6.0 8.0 10.0s/D

min

L/D = 33

L/D = 50

center

corner


s

s

D

235

Fig. 11. Influence of number of piles on load distribution in end bearing pile groups.

Fig. 12. Ratio of average load of pile group to load of single pile.

This parameter can be viewed as the pile group efficiency. In this study, the pile length is 20 m and its diameter is 0.3 m. For the typical pile spacing ratio of 3, the pile group efficiency ranges from 25% to 55%, depending on the number of piles in the group. The smaller number of piles in the group, the higher pile group efficiency. As the spacing ratio increases, the pile group efficiency also improves significantly. For the spacing ratio of 10.0, the group efficiency increase to 45%-75%. However, such large spacing ratio is not practical in pile foundation since the pile cap size increase significantly, and thus such design is not economical in practice. Thus, the interference of adjacent piles in the group has a significant effect on the total load of the pile group or the group efficiency. In addition, this figure also shows the effect of relative stiffness of the upper soil above the tip to the lower soil. For all pile groups, the pile group efficiency of the floating pile type is lower than that of the end bearing pile type, approximately 10-20% at the typical spacing ratio of 3.

5 CONCLUSIONS

The main objectives of this research described herein is to study the characteristics of pile load distribution in the pile group with rigid cap subjected to the static vertical load by means of a three-dimensional finite element model. A limited parametric study of the response of the various type of pile group subjected to a uniform unit settlement was conducted in order to examine the effects of pile spacing, pile dimension and soil properties. Under the applied settlement, the analyses assume that the behaviors of the soil and the pile are in the linear elastic range. The computer program, called �PILE3D�, was developed as an analysis tool in this research. The powerful three-dimensional mesh generation was implemented in order to generate efficient and accurate finite element meshes. The high performance equation solving algorithm was included to minimize computation time. The Superconvergent Patch Recovery Method was applied in the post-processing step in order to recover more accurate nodal stresses for the output of the analysis.

The groups of 4 piles, 5 piles and 9 piles were selected in this study. The soil properties was selected to match the Bangkok subsoil conditions, where the soil profile consists of the soft to medium clay layer and first the sand layer. The results show that the load of each pile is not equally shared in the group. These findings are contrast to the standard design used in practice, assuming that each pile in the group take the load equally. The analyses show that the load distribution in the group depends on pile spacing, number of piles, and relative stiffness of soil above and below the pile tip. There is an increase in the load sharing when the pile spacing decreases, the length of pile increases and the number of pile increases. In addition, the load distribution of the end bearing pile groups is more uniform than that of the floating pile groups. The pile group efficiency also depend the pile spacing in the group, and the number of piles. The typical spacing ratio of 3 used in design practice give the group efficiency in the range of 25%-55%. It should be noted that in this research, because the analyses consider elastic behavior of soil and pile, interface elements are not introduced between soil and pile elements. In real situation, slippage between the soil and the pile can happen and thus shaft friction capacity can be fully mobilized. Future researches should be carried out to consider the effect of slippage between the soil and the pile to the characteristics of pile load distribution in the group.

REFERENCES

Bathe, N.J. 1996. Finite Element Procedures. Englewood Cliffs, New Jersey: Prentice-Hill.

Christophe, G. & Jean-François, R. 2003. Gmsh: A three-dimensional finite element mesh generator with built-in pre- and post-processing facilities [online]. Available from: http://www.geuz.org/gmsh/ [2003, September 7]

Coyle, H.M. & Reese, L.C. 1966. Load transfer for axially loaded piles in clay. Journal of The Soil Mechanics and Foundation Division, ASCE 92(SM2): 1-26.

Desai, C.S. & Zaman, M.M. 1984. Thin-layer element for interfaces and joints. International Journal for Numerical and Analytical Methods in Geomechanics 8: 19-43.

Muqtadir, A. & Desai, C. S. 1986. Three-dimensional analysis of a pile group foundation. International Journal for Numerical and Analytical Methods in Geomechanics 10(1): 41-58.

0.0

0.5

1.0

1.5

2.0

0.0 2.0 4.0 6.0 8.0 10.0

s/D

9 piles5 piles

center

mid-edge

corner

D = 0.6 mL/D = 33Es2/Es1 = 2.5

s

s

D

s

s

D

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

0.0 2.0 4.0 6.0 8.0 10.0 12.0

s/D

Es2/Es1 = 1

Es2/Es1 = 2.5

9 piles

4 piles5 pilesD = 0.6 m

L/D = 33

236

Poulos, H. G. & Davis, E. H. 1968. The settlement behavious of single axially-loaded incompressible piles and piers. Geotechnique 18: 351-371.

Poulos, H. G. & Davis, E. H. 1980. Pile Foundation Analysis and Design. Canada: John Wiley & Sons.

Randolph, M. F. & Wroth, C. P. 1978. Analysis of deformation of vertically loaded piles. Journal of Geotechnical Engineering Division, ASCE 104 (GT12): 1465-1448.

Zienkiewicz, O. C. & Zhu, J. Z. 1992. The superconvergent patch recovery and a posteriori error estimates. Part 1: the recovery technique. International Journal for Numerical Methods in Engineering 33: 1331-1364.

Smith, I. M., & Griffiths, D. V. 1998. Programming the Finite Element Method. 3rd Edition. England: John Wiley & Sons.


237

The New Load Settlement Equation and the Prediction of Pile Failure Load from the Results of Static Pile Loading Test

M. Arayasiri Senior Geotechnical Consultant, Professional Civil Engineer No. 856, Council of Engineers of Thailand Charter Member No. 2523, Engineering Institute of Thailand [email protected]

Abstract: The results of the study indicate that the load-settlement relationship of piles under loading can be represented by the loga-rithmic equation which is transformed into the linear equation as = A+B( ) ( = settlement, P = pile load). It is found that the ratio varies linearly with with the coefficient of linear correlation, r in the range of 0.9999+, which will give the minimum error of prediction. The study covers the interpretation of the pile ultimate load in both the case of testing a pile to plunging failure and the case of not to plunging failure. In case of a pile not tested to plunging failure, the relationships of pile load, settlement and slope of the load-settlement curve are established as and . The plunging failure criteria at _ is adopted in predicting of pile failure load and failure settlement. The results from a total of 34 pile tests in this study areclose to the experimental value with a high degree of confidence and minimum errors as small as 5%.

1 INTRODUCTION

1.1 Pile Behavior Under Loading

Pile can be considered as a structural member for transferring the footing load to the foundation soils. The allowable pile load for safety and cost effectiveness can be derived by applying a fac-tor of safety (F.S) to the ultimate load of the pile. The prediction of pile ultimate load can be carried out by applying soil pa-rameters obtained from soil investigation to the static pile ca-pacity formula. The pile ultimate load can also be obtained from the results of static pile load test. The value of pile ultimate load is dependant on the soil-pile behavior.

The soil-pile behavior as shown by the load-settlement curve can be divided 3 parts as shown in Fig. 1. The first part, before the yielding pile load (PE), is the elastic range in which the settlement increase linearly with the applied load on pile. The second part from pile yielding load to failure load (P U) is the elasto-plastic range in which the increase of settlement with the pile load is not linear. The part beyond the failure load in which the settlement increases with no or slight increment of pile load is the plastic range.

1.2 Interpretation of Failure Load

Fellenius (1980) concluded that the pile failure load or ultimate load has different meanings as follow:

1) For pile which is stronger than the foundation soils the failure load is generated when the settlement suddenly increases under the constant pile load or under small pile load increment. This phenomenon is called �pile plunge�

However, this definition is not sufficient because pile plunge requires large settlement and in several cases the failure load obtained is not dependent on the capacity of the soil-pile behavior, but it may depend on the capacity of man-pump system used in the test.

OA = Elastic range (linear) AB = Elasto-plastic range (non-linear) BC = Plastic range B = Point of plunging failure PE = Yielding pile load PU = Ultimate pile load

E = Settlement at yielding pile loadU = Settlement at ultimate pile load or plunging

failure

Fig. 1. Pile-load settlement.

2) In the past, the definition of pile failure load is defined as the pile load that causes the pile to settle more than 10 % of its diameter. This definition does not take into account the elastic deformation of the pile which is a large value in the case of long piles and low value for shorter piles.

3) Sometimes the pile ultimate load is defined as the pile load at the point of intersection of the two asymptotic lines, e.g. the extensions of the elastic range line and the plastic range line of the load settlement curve. The pile load interpreted by this definition is dependent on the graphical scale and the inter-preter�s judgment. Using different scales will effect the interpre-tation.

Plog

Plog

dPdbapdPd log/)/(

dPdbadPd ''log/)/(

dPd

A

B

PU

C

0 Pile Load

Settlement( ) U

PE

E

238

4) For practical use, the definition of the suitable failure load should be based on the mathematical concept and always give one answer. This means that the interpretation is not de-pendent on the graphical scale of the load-settlement curve and the interpreter�s judgment. In some cases, the shape of load-settlement curve may be taken into consideration, or the length of the pile which indirectly influences the shape of the load-settlement curve. Without such suitable definitions mentioned above the interpretation of the failure load of the pile may be mis-leading.

1.3 Load Settlement Relationship and Pile Failure Load

The information derived from the load-settlement curve obtained from the static pile load test is used to determine the value of subgrade reaction and the allowable load of the pile in foundation design. Several methods have been proposed for the determination of the pile ultimate load from such curve for establishing the factor of safety (F.S. = ultimate pile load / allowable pile load). Fellenius(1980) pointed out that two out of nine are scientific methods by which the result is derived by mathematical means. These two methods are;

1) Chin (1970) method, based on the assumption that the load-settlement relationship is a hyperbolic function, expressed as:

(1)

when, x = settlement, y = pile load. The ultimate pile load can be derived from:

uP 1/ b @ x (2)

2) Brinch Hansen (1963) method based on the assumption of a parabolic load-settlement relationship. The equation can be expressed as:

(3)

when, x = settlement, y = pile load In determining the ultimate pile load, Hansen set up the 80%

criterion where the settlement at 80% ultimate load (80% P U) is equal to ¼ of the settlement at ultimate load ( u @ 100% PU) So that:

(4)

/u a b (5)

Fellenius (1980) also pointed out that the pile ultimate load estimated from these two methods will be effective when the load-settlement is beyond the Davisson Limit Load or elastic limit of the load-settlement curve. The results of pile ultimate load from Chin (1970) will be the upper limit of all methods, and for Brinch Hansen (1963), the ultimate pile load will be close to the tested value. In the case of Maintained Load Test (ML-test) the co-ordinates of load settlement beyond elastic range is limited in numbers, so it may not have sufficient data to get accurate results by the Chin and Brinch Hansen methods. For Constant Rate Penetration test (CRP test), Chin and Brinch Hansen methods will give better results.

From the Author�s experience in Thailand, the pile ultimate load obtained by the Chin method from the result of ML-test is always higher than that of the test result by about 20 - 30 %. For the Brinch Hansen method, it is very difficult to get the load set-tlement co-ordinates beyond the elastic range, because the plung-ing failure loads always found close to the pile load at the yield point or about 5 % of the pile diameter. The Brinch Hansen method is therefore has limited use in determining the ultimate pile load in Thailand where the conventional load test is ML-test.

2 THE NEW LOAD SETTLEMENT EQUATION

In order to overcome the above mentioned limitations, Arayasiri (2002) presented the New Load Settlement Equation derived mathematically as follows:

(6)

r1/( )

P 10

where = settlement, P = pile loadIt is found that the coefficient of linear correlation, r, of the

ratio and is in the order of r = 0.9999+ for all load-settlement data obtained from pile load test. It is also found that the coefficient r increases with the pile load, and the constants A and B also vary accordingly with the load variation. The ultimate pile load can be derived as:

(7)

where u = Settlement at plunging failure and the theoretical pile ultimate load can be derived as

(8)

As mentioned above, the constants A and B vary with the maximum pile load. Therefore the theoretical pile ultimate load, PU(th), will also vary with the maximum pile load. It is clear that only in the case where the maximum test load is close to plunging failure load, will the estimated theoretical PU(th) be close to PU(th)at plunging failure. If the maximum test load is less than plunging failure load the estimated PU(th) will have an error. Anyway, the theoretical failure load PU(th) estimated at is not the plunging failure load in which the practical measurement is taken at .

In order to estimate the plunging failure load and settlement from load settlement curve, the relationship of pile load, pile settlement and the slope of the curve have been developed as follows:

(9)

and (10)

in which (11)

dPd

2log

.

e A Bd

dP A P

dPdba

PdP

d

log

dPdbadP

d''

log

dPd

( )x

a b xy

( )x a b xy

12

UPab

A+Blog P

log P

1/( )

uP 10

u

1/Bu(th) uP 10 @

239

It is found that the linear correlation coefficient of Eqs. (9) & (10) are as high as r = 0.9999+. From Eqs. (9) & (10) considering the plunging failure criteria, , the plunging failure load and failure settlement can be derived as

Pu = 10 1/b (12)

and u = 10 1/b (13)

3 VERIFICATION OF THE NEW LOAD SETTLEMENT EQUATION

3.1 Method of Study

The new load settlement equation was checked by the linear cor-relation coefficient, r, of the load settlement data of each test pile as shown in Tables 1 & 2. These data were obtained from 34 static pile load tests, of which 10 tests were performed on driven piles and 24 tests were those on bored piles. On both driven and bored piles, 24 tests were loaded to plunging failure and 10 tests were not loaded to plunging failure.

Table 1. Summary of results of piles tested to plunging failure.

Put = tested ultimate load Puc = calculated ultimate load Pua = analyzed ultimate load

ut = settlement at tested ultimate load ua = settlement at analyzed ultimate load

At plunging failure load (Put)Analysis of pile ultimate load and

failure settlement Ultimate pile

load Failure settlement Test pile No.

Diameter D

(m)

Length L

(m) L/D Put

(T) ut

(mm.) Puc(T)

d /dP (mm./T)

Linear cor-relation r Pua

(T) Pua/Putua

(mm.) ua/D

32-D-I 0.22X0.22 21.00 95.5 65.0 40.15 65.49 11.04 0.99988 65.44 1.007 44.65 0.203 35-D-I 0.30X0.30 21.00 70.0 95.6 39.40 96.07 8.19 0.99993 96.28 1.007 44.11 0.147 37-D-I 0.22X0.22 21.00 95.5 63.6 39.40 64.27 9.96 0.99981 64.17 1.009 44.42 0.202 41-D-I 0.26X0.26 21.00 80.8 76 21.51 77.08 2.90 0.99972 78.30 1.030 29.14 0.112 62-B 1.50 38.37 25.6 1147.1 81.70 1130.62 2.87 0.999994 1163.50 1.014 102.48 0.068

76-D- 0.60 25.00 41.7 300 46.12 301.93 2.55 0.99998 302.98 1.010 52.37 0.087 78-D- 0.40X0.40 13.00 32.5 250 31.61 241.50 2.65 0.99987 261.60 1.046 57.95 0.145 81-D-I 0.30X0.30 24.00 80.0 117 75.47 117.10 23.92 0.999979 117.53 1.005 82.27 0.274 82-D 0.40 25.00 62.5 150 73.93 149.70 10.53 0.999924 152.99 1.020 91.07 0.228 84-B 0.50 18.50 37.0 110 17.25 108.70 2.16 0.999969 110.14 1.001 20.02 0.040

85-D- 0.40X0.40 13.00 32.5 315 22.83 311.80 1.58 0.999988 317.03 1.006 28.06 0.070 88-B 0.60 21.50 35.8 175 32.18 175.29 4.64 0.999972 176.16 1.007 37.38 0.062 89-D 0.40 21.00 52.5 160 47.76 160.51 5.12 0.999957 161.16 1.007 53.29 0.133

91-D-I 0.22X0.22 18.00 81.8 25 71.29 25.00 126.4 0.999993 25.02 1.001 73.44 0.334 94-B 1.00 35.00 35.0 831 85.48 822.11 4.30 0.999988 830.96 0.999 87.16 0.087

TP1.1-B 1.50 60.00 40.00 3477 66.19 3472.1 0.1564 0.999990 3657.2 1.052 87.80 0.059 TP1.3-B 1.50 55.00 36.70 3451 112.17 3469.1 0.4626 0.999996 3509.6 1.017 124.56 0.083 TP1.4-B 1.50 52.00 34.70 3440 61.69 3440.5 0.1441 0.999985 3620.6 1.053 82.56 0.055 TP2.3-B 1.50 55.00 36.70 2932 88.78 2937.6 0.4528 0.999995 2971.9 1.014 102.07 0.068 TP2.4-B 1.50 60.00 40.00 3458 90.67 3475.9 0.2866 0.999993 3553.7 1.028 108.04 0.072 TP3.3-B 1.50 55.00 36.70 3215 50.12 3215.9 0.1444 0.999980 3382.5 1.052 67.84 0.0452 PP2.1-B 1.20 51.10 42.60 1800 52.46 1798.8 0.3688 0.999990 1848.2 1.027 64.88 0.0540 PP2.2-B 1.20 53.40 44.50 1800 42.61 1796.1 0.2259 0.999983 1882.0 1.046 56.51 0.0471 PP2.3-B 1.20 49.40 41.20 1757 117.63 1748.2 2.0040 0.999995 1755.5 0.999 112.26 0.0936

ddP

240

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35

ut/D

( ua/D) = 0.01215+1.0377( ut/D)r =0.9845

Table 2. Summary of results of piles not tested to failure.

Pmt = tested maximum load

Pmc = calculated maximum load

mt = settlement at tested maximum load

3.2 Interpretation of the Ultimate Pile Load

Case A: The test piles were loaded to plunging failure and the final settlement ( u) is recorded. The pile ultimate load can be obtained from Eq. (7).

Case B: The test piles were not loaded to plunging failure and the final settlement ( u) is not attained. The load settlement relationship up to the maximum tested load is governed by Eq. (6).

The slope of the load-settlemtent curve in all cases can be cal-culated by Eq. (11).

The criteria for determining the ultimate pile load can be set as follows:

1) The linear correlation, r, of the load settlement equation at the maximum test load should be r = 0.9999+.

2) Pile load should be beyond the elastic yielding load or Davisson limit load.

3) The linear correlation r P and r of the linear equations Eqs. (9) & (10) should be r = 0.999 + Min. At the plunging failure where , Eqs. (12) & (13) should be applied.

4 RESULTS AND DISCUSSION

1) All the values of the coefficient of linear correlation, r, at the plunging failure load and at maximum test load shown respectively in Tables 1 & 2. are in the order of 0.9999+ ,which indicates that the load settlement curve and the derived equations are nearly coincidental.

2) The values of the ultimate calculated pile load (P uc) by Eq. (7) as shown in Table 1 compared to the plunging pile load (P ut)observed in the field test show an error not exceeding + 2 %. This

indicates the precision of using Eq. (7) in calculating the ultimate load.

3) The values of the ultimate pile load (P ua) analyzed from the criterion that at plunging failure slope of load � settlement curve as presented in Table 1 also differ from the plunging failure load (Put) by not more than 5%. It can therefore be concluded that the plunging failure load observed in the field test does conform with the criterion

4) The values of the failure settlement( ua) analyzed from the criterion at plunging failure base on presented in Table 1 and Fig. 2 are slightly higher than the settlement observed at plunging failure load ( ut) which seems to be reasonable. The ratios of the settlement analyzed from and the test pile diameter are shown in Tables 1 & 2. It is seen that the ratios are in the range of 4 to 20% of pile diameter, and seem to depend on the value of slenderness ratio L/D and diameter D as shown in Figs. 3 & 4.

Fig. 2. Relationship of analyzed ultimate settlement, ua and the tested ultimate settlement, ut.

At maximum test load (Pmt)Analysis of pile ultimate

load and failure settlement Ultimate pile

load Failure

settlement Test pile

No.

DiameterD

(m)

Length L

(m) L/D Pmt

(T) mt

(mm.) Pmc(T)

d /dP (mm./T)

Linearcorrelation r Pua

(T) Pua/Pmt ua

(mm.) ua/D

TP1.5-B 1.50 55.00 36.7 3440 49.15 3449.4 0.1004 0.999976 3653.4 1.062 68.13 0.0454

TP2.1-B 1.50 58.00 38.7 3670 44.64 3606.6 0.0662 0.999944 4260.6 1.161 78.47 0.0523

TP2.5-B 1.50 60.00 40.0 3700 56.80 3692.3 0.1051 0.999977 4006.3 1.083 81.72 0.0545

TP2.6-B 1.50 60.00 40.0 3825 41.83 3746.6 0.0497 0.999903 4818.4 1.260 89.87 0.0600

PP1.1-B 1.20 53.50 44.6 1804 23.61 1739.6 0.0630 0.999906 2219.3 1.230 55.50 0.0463

PP1.2-B 1.20 53.30 44.4 1800 18.34 1746.3 0.0361 0.999795 2995.4 1.664 120.69 0.1000

PP1.3-B 1.20 51.40 42.8 1799 29.91 1740.6 0.0893 0.999913 2136.0 1.187 60.58 0.0505

PP1.4-B 1.20 52.20 43.5 1800 29.61 1736.0 0.0883 0.999889 2193.3 1.219 65.85 0.0550

PP1.5-B 1.00 51.60 51.6 1350 28.80 1281.1 0.0896 0.999876 1722.6 1.276 68.47 0.0685

PP2.4-B 1.20 51.10 42.6 1720 31.41 1713.5 0.1119 0.999941 1952.8 1.135 54.84 0.0457

ddP

ddP

.ddPddP

ddP

241

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

0 200 400 600 800 1000 1200 1400 1600

D (mm)

ua /D = - 0.00659 /( (1/log D) - 0.43796)

r = 0.99676 ,r2 = 0.9935 , n =34

u/ D = 0.10 (Terzaghi )

0.00

0.10

0.20

0.30

0.40

0.50

0 20 40 60 80 100 120

L/D

ua/D = 0.00792/ ((1/log(L/D))-0.5019)

r2 = 0.9868 , n = 34

ua/D = - 0.0477+0.00302(L/D)

r2 = 0.6106 , n = 34u / D = 0.10 (Terzaghi )

0

2

4

6

8

10

12

0 20 40 60 80 100 120 140

L/D

d /dP = 0.20683 / ((1/log(L/D)) - 0.5209r = 0.99989 , r2 = 0.99978 , n = 24

0

2

4

6

8

10

12

0 200 400 600 800 1000 1200 1400 1600

D(mm)

(d /dP)/ log D = A+B (d /dP) A = - 0.1452 , B = 0.42725

r = 0.99989 , r2 = 0.9998 , n = 24

Fig. 3. Relationship of analyzed ultimate settlement, ua and the pile diameter, D.

Fig. 4. Relationship of analyzed ultimate settlement, ua and the pile slenderness ratio, L/D.

It�s seen that the ratio does not decrease linearly with the increase in pile diameter, D and does not increase linearly with the increase in the slenderness ratio L/D. Such behavior is found to be different from the Terzaghi concept that the ultimate settlement is estimated to be 10% of the pile diameter.

5) The settlement rate at plunging failure load, is found to be in the range of 0.10 mm/T to more than 100 mm/T as shown in Table 1. This rate exhibits good correlation with log D and log (L/D) (Figs. 5 & 6). It is seen that the settlement rate decrease logarithmically with the increase in pile diameter and increases logarithmically with the slenderness ratio.

Fig. 5. Relationship of settlement rate, d dP, at plunging failure from pile load test and the pile diameter, D.

Fig.6. Relationship of settlement rate, d dP at plunging failure from pile load test and the pile slenderness ratio, L/D.

5 CONCLUSIONS AND RECOMMENDATIONS

1) The New Load Settlement Eq. (6) fits well with the load - set-tlement curve obtained from the static pile loading test. The lin-ear correlation coefficient, r, is found to increase with the in-crease in load, and varies in the range of r = 0.999+ to 0.9999+.

2) The values of ultimate load and settlement at ultimate load analyzed from the criterion that at plunging failure (Pua,

ua) show good correlation with the corresponding tested values (Put, ut). For Pua it is + 5% max. greater than Put. The ratio of ultimate settlement and pile diameter, ua/D is not constant for all pile diameters but increases logarithmically with the slenderness ratio L/D, and decreases logarithmically with pile diameter (D).

3) The values of the settlement rate at plunging failure are in the range of 0.10 to more than 100 mm/T, and vary with the pile diameter (D) and slenderness ratio L/D.

4) The criteria for predicting the failure load can be con-cluded as follows:

a. The linear correlation, r, of the load settlement equation up to the maximum test load should be r = 0.9999+.

b. The slope of load-settlement curve, at the maximum test load should be more than 0.15 mm/T depending on the pile diameter and slenderness ratio.

5) For further study, it is recommended to study the result of pile load test from wide ranging geotechnical areas in order to verify the global validity of the New Load Settlement Equation and the slope of load settlement curve, equations in predicting the ultimate pile load and settlement.

REFERENCES

Arayasiri, M. 2002. The new load settlement equation and the prediction of ultimate load from the results of static pile load-ing test. EIT Seminar on pile behavior resulted from pile load test and the application of pile load test data in foundation work, Bangkok, Thailand (in Thai).

Chin, F.K. 1970. Estimation of the ultimate load of test pile not carried to failure. Proceedings of the 2nd Southeast Asian Conference on Soil Engineering .

Davisson, M.T. 1972. High capacity piles . Proceedings of the Lecture Series, Innovations in Foundation Construction,ASCE, Illinois Section.

DU

dPd

dPd

ddP

242

Fellenius, B.H. 1980. The analysis of results from routine pile load test. Ground Engineering, 13(6): 19-31.

Terzaghi, K. & Peck, R.B. 1967. Soil Mechanic in Engineering Practice. New York: Wiley.


243

Improved Effect of Skin Friction of Pile Extended in Sand

N. Yasufuku, H. Ochiai, & K. Omine Department of Civil Engineering, Kyushu University, Fukuoka 812-8581,[email protected]

S. Babasaki Obayashi Co. Ltd., Tokyo, Japan.

Abstract: An improved effect of skin friction of pile expanded in sandy ground is discussed based on the experimental results from a series of the model pile load tests. The characteristics of the apparatus, which is newly developed in this study, is first explained, mainly focused on the function of the expanded system of the pile. The way of changes in horizontal stresses of a model ground sur-rounding the pile due to its expansion is then discussed with reference to the degree of the expansion. Finally the improved effect of the skin friction due to the expansion is clarified based on the experimental data. In addition, a simple idea to evaluate the improved pile skin friction is introduced in this study. Its applicability is verified by the experimental data. It can be found from the experimental results that the pile with an expanded function in practice is environmentally effective to improve the pile skin friction.

1 INTRODUCTION

When considering the mobilized mechanism of pile skin friction, it is essential to understand the horizontal stress of the ground surrounding the pile. Recently, several kinds of pile construction methods without generating the construction surplus soils during the pile setup are produced by Japanese construction and steel companies, aiming at reducing the environmental impact. Such pile construction methods are expected not only to reduce the construction surplus soils but also to increase the horizontal stresses acting on the pile due to horizontally pushing outward and compacting the soils surrounding the pile. In this study, fundamental characteristics of the improved effect of pile skin friction by the increments of the horizontal stresses acting on the pile are discussed through the results of model pile load tests and the brief theoretical considerations. In addition, the propagation characteristics of the incremental horizontal stresses in the ground surrounding the pile is considered, particular refer-ence to the degree of the model pile expansion which may di-rectly reflect the increments of the horizontal stresses in the ground close to the pile.

2 MODEL PILE LOAD TEST APPARATUS AND TEST PROCEDURE

2.1 Model Test Apparatus Produced

A model pile load apparatus newly made is schematically shown in Fig. 1, in which the model pile is characterized as being hori-zontally expanded. The cylindrical chamber is about 700mm in diameter and 1000mm in height. A constant overburden pressure up to 100kPa can be applied through a loading plate with two air cylinders. Small earth pressure sensors (PDA-200-500kPa, To-kyo Sokki Co., Ltd) shown in Fig. 2 are set up in the proper posi-tions of the model ground surrounding the pile, which are used to measure the changes of the horizontal stresses, when the model pile is horizontally expanded and then penetrated under a certain

overburden pressure. Figure 3 shows the structural system of mechanically expanded model pile used here, which is named as an expanded pile, where the diameter of the upper and lower

700mm

Air Cylinder

Model Pile

Load Cell

Sample

DisplacementTransducer

Loading Plate

1000mm

Fig. 1. Schematic view of model pile load apparatus.

5mm

Fig. 2. Earth pressure measurement sensor used.

244

Extension ratio= a / a0

After pile extension

Before pile extension

a0

r

Location of earth pressure sensor set up

9.0r / a = 1.0 3.0

a

v v

Df

Modeling a fixed depth by applying a corresponding surcharge load

Actual pile Model test pile

Fig. 4. Modeling of the pile and location of earth pressure sensor.

parts are around 60mm and 50mm, respectively, which indicates that the pile is cylindrically taper shaped one, and also the length is about 500mm. The model pile consists of the brass plates which are divided by 16 pieces. When a central shaft is rotated by a rod as shown in Fig. 3, each plate smoothly slides and then the diameter of the pile increases. Figures 3(b) & 3(c) shows the cross section of the pile and a piece of pile surface before and af-ter the pile diameter increases. The model pile is always set up to the initial penetration depth of 450 mm from the model ground surface and all the tests are conducted under various overburden pressures to simulate a stress condition in any depth of actual pile (see Fig.4).

2.2 Test Conditions and Definitions of Parameters Used

The parameter for defining the degree of changes in pile diameter, simply called as pile extension ratio here, and the position of small sensors for measuring the changes of the horizontal effec-tive stresses in the ground during testing are shown in Fig. 4. In

model tests, the diameter �a0� at the middle of the pile length D fis used as a representative initial diameter of the pile (see Fig.4) and then the pile extended ratio is simply defined as a/a 0 where �a� is the pile diameter after cylindrically extending the pile out-ward shown in Fig. 4. In addition, the normalized horizontal dis-tance from the center of the pile defined as �r/a� is used to char-acterize the distribution of the horizontal stresses in the ground due to cylindrically pile extension. Test conditions are summa-rized in Table 1. Four different initial horizontal stresses and three extended ratios are selected in this study to investigate the effect of the overburden pressure and the pile extension outwards on the characteristics of horizontal stress changes and the im-proved effect of the mobilized pile skin friction. After the model pile is set up to the fixed position in the chamber, Toyoura sand ground with the relative density of 75% is made as a representa-tive model ground, which is prepared by air �pluviation method.

3 CHARACTERISTICS OF HORIZONTAL STRESS CHANGES DUE TO CYLINDRICAL PILE EXTENSION

3.1 Horizontal Stress Changes with the Degree of Pile Extension

Figure 5 shows the horizontal stresses h acting on the pile sur-face just after pile extension, which is normalized by the initial horizontal stress h0, h/ h0, against the degree of pile extension a/a0 in terms of the initial horizontal stresses. In this case, the normalized horizontal stresses almost linearly increase with the increasing pile extension ratio a/a0, irrespective of the initial horizontal stresses. Figure 6(a) shows the characteristics of the normalized horizontal stresses under the initial horizontal stress of 12.5kPa, at which the normalized horizontal distance r/a is from 1.0 to 9.0, against the pile extension ratio. It is clear that the increasing rate of the normalized horizontal stresses with the pile

50mm

Before pile extension

After pile extension

Cross section

SteelBrass

Part of pile surface

SteelBrass

60mm

500mm

Cross section

Part of pile surface

(a) (b) (c)

Fig. 3. Model pile and its extension mechanism.

Table 1. Test conditions selected.

Initial horizontal stress h0 (kPa) 8.2, 12.5, 25.0, 50.0

Initial pile diameter a 0 (mm) 27.5, 28.7, 29.6

Pile extension ratio a / a 0 1.0, 1.042, 1.075Relative density D r (%) 75

Pile surface roughness R max ( m) 100

Penetration depth D f (mm) 450

0

1

2

3

4

5

6

7

1 1.01 1.02 1.03 1.04 1.05 1.06 1.07

a / a0

Dr = 75%Rmax = 100 mr / a = 1.0

h0 = 8.2kPa

h0 = 12.5kPa

h0 = 25.0kPa

Fig. 5. Increments of normalized horizontal stresses at pile surface with increasing pile extension ratio.

245

extension ratio decreases with the increasing distance from the center of the pile r/a. Such tendency can also be understood from the h/ h0-r/a relationship shown in Fig. 6(b). When r/a becomes greater, the changes of h/ h0 exponentially reduce to zero irre-spective of a/a0. There seems to exist a limited distance from the center of pile in which the stress changes do not appear even if the pile is cylindrically extended outward. In this case, the lim-ited distance become 9.0 in r/a. Similar results were observed in the cases conducted under the different overburden pressures. Figure 7 shows the relationship between the normalized horizon-tal stresses at r/a=3.0 and the initial horizontal stresses h0. It is clear that the h/ h0 for each pile extension ratio gradually de-

crease with the increasing initial horizontal stresses, which means that the dramatic effect on the stress changes due to pile exten-sion tends to be in more narrow ranges with the increasing over-burden pressure.

3.2 Characteristics of Horizontal Stress Relaxation after Pile Extension

The horizontal stresses in the ground due to pile extension may decrease with the increasing elapsed time because of the rear-rangement of soil particle structure and its visco-plastic proper-ties. Figure 8 shows the results in the cases that pile was ex-tended up to 7.5% against the initial diameter of pile a 0(a/a0=1.075). The horizontal stress changes with the increasing pile extension ratio and the elapsed time at an initial horizontal stress of 8.2kPa and at four different r/a are shown in Figs. 8(a) and (b). It is clear that the horizontal stresses at zero elapsed time, which increased with pile extension ratio, quickly decreased within several hours just after stopping the pile extension and then almost converge a constant value till 20 hours passed, which is called as the residual horizontal stress and in these cases, the horizontal stresses from 70% to 80% are remained. When com-paring the relaxation properties in terms of r/a, smaller r/a is, lar-ger the decreasing rate with time becomes. In other words, the re-laxation ratio of the horizontal stresses at a location relatively far from the center of the pile tends to be small The incremental

0

1

2

3

4

5

6

1 1.01 1.02 1.03 1.04 1.05 1.06 1.07

a / a0

h0 = 12.5kPa

Dr = 75%Rmax = 100 m

r / a = 1.0r / a = 2.0r / a = 3.0r / a = 4.0r / a = 6.0

0

1

2

3

4

5

6

1 2 3 4 5 6 7 8 9

r / a

h0 = 12.5kPa

Dr = 75%Rmax = 100 m

a / a0 = 1.02a / a0 = 1.04a / a0 = 1.06

Fig. 6. Changes of horizontal stresses in surrounding ground related to pile extension ratio.

0

1

2

3

4

5 10 15 20 25 30 35 40

h0kPa

a / a0 = 1.030

a / a0 = 1.055a / a0 = 1.042

r / a = 3.0

Fig. 7 Changes of initial stresses at r/a=3.0 with initial hori-zontal stresses.

0

5

10

15

20

25

30

35

40

0 5 10 15 20 25 30 35 40

t h

r / a = 1.0

r / a = 2.0

r / a = 4.0

r / a = 3.0

0

5

10

15

20

25

30

35

40

1 1.075

a / a0

h0 = 8.2kPaRmax = 100 mDr = 75%a / a0 = 1.075

Fig. 8. Stress relaxation properties at h0=8.2kPa after pile extension.

246

horizontal stresses at an elapsed time, ht, normalized by the initial incremental stresses h(at t=0) against the elapsed time are shown in Fig. 9, which are obtained under three different kinds of initial horizontal stresses. The stress relaxation ratio with time depends on the initial horizontal stresses, namely, the re-laxation ratio is just greater when increasing the initial horizontal stresses. It is however noted that the horizontal stresses around 70% for each case continues to remain as the residual stresses.

3.3 Evaluation of Horizontal Stress Changes in Surrounding Ground with Pile Extension

It must be convenient if the effect of pile extension on stress and deformation state in the surrounding ground can be easily evalu-ated. In this study, a cylindrical cavity expansion theory, which is derived by Yu & Houlsby (1991), is applied to predict the changes of the horizontal stresses in the ground surrounding the pile. Four soil constants are needed for calculation, which are Young modulus E, Poisson�s ratio , internal friction angle ,and dilatancy angle summarized in Table 2. All the parameters are determined by the results from triaxial compression tests. The initial and boundary condition which is linked with experimental condition are also shown in this table. Figure 10 shows the com-parison of the predicted results with the experimental ones, in which the changes of h/ h0 with r/a at two different pile exten-sions are shown in Figs. 10(a) & (b), respectively. Although the difference between the predicted and experimental results can be found in the case that the pile extension is relatively small (see Fig.10(a), the experimental tendency which h/ h0 exponentiallydecreases with the increasing r/a can be represented by the model used. Therefore, if more accurate prediction is requested, some modification of model may be necessary.

4 IMPROVED PILE SKIN FRICTION DUE TO PILE EXTENSION

Figure 11 shows the relationship between the pile skin friction and normalized settlement S/D under two different overburden pressures obtained by the model pile load tests, where S and D are defined as pile settlement and diameter, respectively. In both figures, black and white circles indicate the results without and with pile extension, respectively. It is confirmed from these fig-ures that the improved effect of the skin friction due to pile ex-tension is clear, which is presented as the increases of the initial stiffness in the fs-S/D relationship and also presented as the in-

creases of the maximum skin friction, irrespective of the over-burden pressure. The relationship between the degree of the im-proved effects fs/fs0 and the normalized settlements S/D is shown in Fig. 12, which is depicted by using the results in Fig. 11, where fs0 and fs are defined as the mobilized skin frictions with-out and with the pile extension, respectively. The degrees of the improved effects are gradually increased and are then reach the peak values around S/D from 0.05 to 0.1, which are generally within an allowable settlements in design. When increasing the normalized settlements further, the degree of the improved effect decreases and then approaches to a constant values. Figure 13 shows the normalized maximum pile skin friction fs/fs0 against the initial horizontal stress h0 directly related to the pile penetra-tion depth. It is noted that all the results are obtained for the cases that pile extension ratios are in the range around from 4% to 7%. Although the degree of the improved effect tends to decrease with the increasing initial horizontal stress, the resulting degree of the improved effect defined as the normalized maximum skin

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 5 10 15 20 25 30 35 40t h

a / a0 = 1.075r / a = 1.0

h0 = 12.5kPa

h0 = 25.0kPa

h0 = 8.2kPa

Fig. 9. Horizontal stress relaxation properties related to initial horizontal stresses.

1

h0 = 25.0kPaa / a0 = 1.021

= -10

0

1

2

3

4

5

6

2 4 6 8 10r / a

0

0

1

2

3

4

5

6

2 4 6 8 10r / a

0

1

h0 = 25.0kPaa / a0 = 1.042

= -10

Fig. 10. Comparison of predicted horizontal stresses against r/a 0with experimental ones.

Experiment Prediction

Experiment Prediction

Table 2. Analytical parameters used and initial condition.

Cohesion c (kPa) 0

Internal friction angle (degs.) 30

Poisson ratio 0.3Dilatancy angle (degs.) 10

Young's modulus E ( MPa) 200, 250, 300

Initial cavity inner pressure p 0 (kPa) 8.2, 12.5, 25.0Cavity radius a 0 (mm) 27.5

247

friction fs/fs0 is found to be from 140% to 200% when the pile ex-tension ratio is in the range from 4% to 7%.

5 EVALUATION OF IMPROVED PILE SKIN FRICTION RELATED TO HORIZONTAL STRESS CHANGES

Skin friction of a pile is generally determined as the sum of pile to soil adhesion and friction components as shown in the follow-ing equation:

''' tanhi cf (1)

c� and � are the adhesion and friction parameters between pile and soil, and 'h is the effective lateral stress acting on the pile. When assuming that the mobilized mechanism of skin friction between piles and soils is essentially based on the shear failure mechanism in the thin layer of soils, it is reasonable to use the strength parameters at the critical state corresponding to the suf-ficiently large displacements, such that:

0'c (2) ''cv (3)

where �cv is a friction angle at the critical state. If this friction angle is used for the soil, it is independent of density and is unique, irrespective of the confining pressure. In addition it is important to point out that, if this is used, the soil strengths minimum value is assured which is very useful (Yasufuku et al.,1997, 1998 and 2001). Thus, submitting Eqs.(2) and (3) into Eq.(1), the following simple equation is derived, which is con-sidered to be a basic equation in this study:

h0 = 8.2kPa

h0 = 25.0kPa

fs a / a0 = 1.042fs0 a / a0 = 1.0

1

1.1

1.2

1.3

1.4

1.5

1.6

1.7

0 0.05 0.1 0.15 0.2 0.25 0.3

S / DFig. 12 Improved effect of pile skin friction related to normal-ized settlements.

Dr = 75%Rmax = 100 m

a / a0 = 1.075a / a0 = 1.042

1

1.2

1.4

1.6

1.8

2

2.2

0 10 20 30 40 50 60h0

Fig. 13. Improved effect of pile skin friction against initial horizontal stresses related to pile extension ratio.

Dr = 75%Rmax = 100 ma = 57.3mm

a /a0 = 1.042

a /a0 = 1.0

h0 = 8.2kPav0 = 16.3kPa

0

5

10

15

20

25

30

0 0.05 0.1 0.15 0.2

S / D

Dr = 75%Rmax = 100 ma = 57.3mm

a /a0 = 1.042

a /a0 = 1.0

h0 = 25.0kPav0 = 50.0kPa

0

10

20

30

40

50

60

0 0.05 0.1 0.15 0.2

S / DFig. 11. Improved pile skin friction due to pile extension in the ground related to overburden pressure.

248

Fig. 14. Comparison of predicted skin frictions due to pile exten-sion with measured ones.

'' tan cvhif (4) Now, referring to Eq.(4), the skin friction related to the incre-

ment of horizontal stress due to pile extension can be expressed as :

'' tan cvhhif (5) where, h0 and h are defined as the initial horizontal stress and the changes of the horizontal stresses which include the stress re-laxation effect, respectively. Further, in order to introduce the stress relaxation effect into Eq.(5) based on the experimental re-sults shown in Figs. 8 and 9, h is divided by two parts, that is, the initial stress changes h1 due to pile extension and the amount of stress relaxation h2 such that :

'21

' tan cvhhhif (6)

A comparison of the experimental maximum skin frictions fs(mea) with the calculated values fs(cal) using Eq.(6) is shown in Fig. 14. Here, h1 and h2 in Eq.(6) are determined by intro-ducing the experimental results obtained as shown in Figs. 8 and 9. It is found that the calculated results give a relatively good agreement with the experimental results. Therefore, an important thing as a next step is to present a rational manner for estimating the horizontal stress changes due to pile extension.

6 CONCLUSIONS

The following conclusions are obtained from a series of model pile load tests;

1) Horizontal stresses in the ground surrounding the pile in-crease with the pile extension and then exponentially decrease with the elapsed time after pile extension. However, in this study, the horizontal increment stresses around 70-80% of the incre-mental ones just after pile extension remained as a residual hori-zontal stress, irrespective of overburden pressure and the distance from the pile.

2) The horizontal stress increments due to pile extension expo-nentially decrease with the increasing distance from the pile. They were not measured at the position more than 9 times far from the center of the pile in this study.

3) Improved effect of pile skin friction due to pile extension is clearly found, which appeared as an increment of the initial stiff-ness and an increases of the maximum skin friction. The im-proved effect is in the range of 1.4 to 2.0 times greater than the maximum skin friction without pile extension. 4) It is confirmed that the improved effect of pile skin friction due to pile extension can be evaluated by using the incremental horizontal stress in the ground close to the pile surface, properly considering the stress relaxation.

ACKNOWLEDGMENTS

This study is supported by Japanese Society for the promotion of science through the grant No: 14350160 �Evaluation of bearing capacity of taper type pile foundation taking care of soil com-pressibility and its application to the performance based design � The authors also wish their sincere thanks to Mr. S. Yamada of Fukuoka University and Mr. M. Nakashima of Kyushu Univer-sity for their advice and encouragements.

REFERENCES

Yasufuku, N., Ochiai, H. & Maeda,Y.1997. Geotechnical analysis of skin friction of cast-in-place piles. Proceedings of 14th International Conference Soil Mechanics and Geotechni-cal Engineering, Hamburg: 921-924.

Yasufuku, N, Ochiai, H, Kwag, J & Miyazaki, K. 1998. Effec-tiveness of critical state friction angle of volcanic ash soils in design applications. Proceedings of International Symposium on Problematic Soils 1: 235-239.

Yasufuku, N, Ochiai, H & Ohno, S. 2001. Pile end-bearing ca-pacity of sand related to soil compressibility. Soils and Foun-dations 41(4): 59-71.

Yu, H.S. & Houlsby, G.T. 1991. Finite cavity expansion in dila-tant soils: loading analysis. Geotechnique 41(2): 173-183.

h0 = 8.2kPah0 = 12.5kPah0 = 25.0kPa

a / a0 = 1.042

a / a0 = 1.075

0

10

20

30

40

50

60

70

0 10 20 30 40 50 60 70f s

(cal1)


249

Undrained Lateral Loading for Drilled Shafts

Y. J. ChenSinotech Engineering Consultants, 171 Nanking East Road, Sec.5, Taipei, [email protected]

F. H. Kulhawy School of Civil and Environmental Engineering, Cornell University, Hollister Hall, Ithaca, NY 14853, [email protected]

Abstract: Undrained lateral loading capacity models are evaluated for drilled shaft foundations. The basic models are reviewed briefly, and then the model capacity predictions are compared with two capacities interpreted from load test results, the lateral or moment limit and hyperbolic capacity. The relationships between the models and interpreted capacities are presented, noting the lower and upper bound values. The effect of shaft rigidity on lateral capacity also is discussed, citing differences for rigid, intermediate, and flexible shafts. Design recommendations are developed, supported by basic statistics to define the recommendation quality.

1 INTRODUCTION

The force system for a laterally loaded drilled shaft is complex and three-dimensional. Although these forces are dominated by the passive lateral soil resistance, there is shearing along the shaft tip (toe) and along the front and back faces and sides, and the axial force components can affect the lateral behavior. A rigorous analysis of these forces requires three-dimensional numerical methods, such as finite elements. However, the problem usually is condensed to a two-dimensional model.

The shaft response also is affected by many boundary condi-tions (Kulhawy & Chen, 1995), such as shaft rigidity and shaft butt (top or head) fixity. For shaft rigidity, no generally accepted standard definitions exist for rigid, intermediate, and flexible behavior, although several have been suggested (Broms, 1964; Bierschwale et al., 1981; Poulos & Davis, 1980; Poulos & Hull, 1989; Carter & Kulhawy, 1992). For butt fixity, nearly all load tests have been conducted on unrestrained shafts.

Many analytical models have been proposed for the undrained lateral capacity, based on a variety of theoretical assumptions. Also, various methods have been proposed for interpreting the shaft "capacity" or "failure" from load test results. Different models and interpretation methods could result in considerable differences. Some clarification is warranted.

In this paper, a comparison is made of representative analytical models, using a large database of undrained lateral load tests on drilled shafts, at both laboratory and field scales, and consistent methods of interpreting the load test results. The results lead to design recommendations for future use.

2 UNDRAINED LATERAL LOAD ANALYSIS

The total stress method of analysis is appropriate for undrained loading. For this method, the lateral capacity (H u) can be calcu-lated by limit equilibrium analysis of the horizontal forces and moment shown in Fig. 1. The lateral soil resistance or yield stress (pu) in Fig. 1 is the most critical parameter and usually is expressed in terms of a lateral bearing factor (Np), defined as Np = pu /su in which su = undrained shear strength. The appropriate undrained

strength test types to use are TE (triaxial extension) for the lateral soil resistance and DSS (direct simple shear) for the tip resistance (e.g., Kulhawy & Mayne, 1990).

In any case, the soil limit state at greater depths is characterized by flow around the shaft. At shallow depths, the soil in front of the shaft can move upward and laterally. Therefore, most analytical models include some variation of Np with depth. Some of the commonly-referenced models are illustrated in Fig. 2 and are described briefly below.

Reese (1958) utilized a wedge model for shallow failure and a plasticity model for lateral plastic flow around the shaft for deep failure. Hansen (1961) employed an earth pressure model at shallow depths, a Rankine wedge equilibrium model at moderate depths, and a plastic flow model at greater depths. Broms (1964) used a plastic flow model for deep failure and judgment for shal-low failure, based on the Reese & Hansen models and available load test data. Stevens & Audibert (1979) backfigured profiles of Np with depth from analysis of instrumented driven pile load test data. Randolph & Houlsby (1984) essentially modified the Reese shallow model to include surface roughness and expanded on the Broms deep plastic flow model.

Fig. 1. Profile for undrained equilibrium analysis.

250

Fig. 2. Lateral bearing factors from analytical models.

Finally, Davidson et al. (1982) proposed a more complete equilibrium system for the lateral or moment loads acting on a rigid drilled shaft. They considered four modes of load resistance, including lateral soil resistance, vertical side shear, base or tip shear, and tip moment.

3 DATABASE FOR LATERAL LOAD EVALUATION

To evaluate these undrained lateral capacity models, available case histories were compiled that included both laboratory model-scale and field full-scale load tests. Complete database details are given by Chen (1993) and Chen & Kulhawy (1994).

In the database, 50 laboratory tests and 43 field tests were evaluated. All were conducted in cohesive soil profiles. For the laboratory tests, the shaft diameter (B) ranged from 89 to 175 mm, D/B (depth/diameter) was between 3 and 8, and all shafts were rigid. For the field tests, B ranged from 0.1 to 2.0 meters, and D/B was between 2.5 and 31.6, with 27 rigid, 4 intermediate, and 12 flexible shafts.

For the laboratory tests, extensive property testing had been done, and therefore all of the required data were available to evaluate the necessary parameters. However, for the field tests, all of the necessary parameters were not available. Therefore, it was necessary to rely on available empirical correlations to obtain these parameters, as given by Kulhawy & Mayne (1990).

Both the lateral or moment limit and hyperbolic capacity were used to determine the "capacity" from the load test data. Hirany & Kulhawy (1989) discuss the actual load-displacement response and the modes of soil-shaft failure to develop the lateral or moment limit (HL). Basically, these modes become evident by plotting the applied load or moment versus the apparent point of rotation, which is the ratio of butt displacement to the tangent of the butt slope. HL corresponds approximately to the load at which initial failure or yield occurs, generally in the "knee" of the load-displacement curve, and is not the ultimate limit state.

For the ultimate limit state, a simple hyperbolic representation (e.g. Manoliu et al., 1985) can be used. By transforming the load-displacement curve to a displacement/load versus displace-ment plot, the slope of the data can be obtained, and then its re-ciprocal is the hyperbolic capacity (Hh). The capacity always lies above the measured data and represents an upper bound.

4 COMPARISON OF CAPACITY PREDICTIONS

Illustrative results of the analyses for undrained lateral loading are shown for the Reese model in Fig. 3 for the laboratory tests and in Fig. 4 for the field tests. Each of these figures is in two parts.

Fig. 3. Laboratory lateral capacity comparisons for Reese model.

251

Fig. 4. Field lateral capacity comparisons for Reese model.

The first is a plot of HL and Hh versus Hu and includes the re-gression lines through the origin for both data sets. The second is a histogram of the lateral capacity ratios, with mean and standard deviation (S.D.) shown for each data set. In these figures, all of the 50 laboratory tests are classified as rigid, and only the 21 rigid field tests are included. Note that 7 of the 27 rigid field tests were conducted at the same site. To minimize potential site bias, only the average of these 7 was used, resulting in 21 cases analyzed. Also, because of insuf-ficient data, HL and Hh could only be evaluated for 45 and 47 tests, respectively.

Table 1 summarizes the results of the regression analyses for all of the analytical models evaluated. This table is separated into three parts for the laboratory tests, field tests, and all tests together. As can be seen, the coefficient of determination (r 2) and the S.D. are comparable within each comparison group (e.g., for H L in the laboratory tests), and both suggest relatively high quality com-parisons. However, visual examination of the plotted laboratory test data indicates two data points at high capacity that may bias the regression lines somewhat. The same is true for one high capacity data point in the field test data.

Further examination of the results in Table 1 shows that the laboratory and field data may constitute two separate populations

Table 1. Regressions for undrained lateral load tests on rigid drilled shafts.

A. Laboratory tests (n = 45 for HL & 47 for Hh)

Model Lateral/Moment Limit, HL Hyperbolic Capacity, Hh

La r2 SD (kN) h

a r2 SD (kN)

Reese 0.90 0.981 0.065 1.46 0.979 0.108 Hansen 1.33 0.981 0.065 2.17 0.980 0.106 Broms 1.06 0.977 0.071 1.73 0.977 0.114 Stevens 0.76 0.981 0.065 1.24 0.979 0.107 Davidson 1.11 0.983 0.060 1.81 0.981 0.101 Randolph 0.86 0.981 0.064 1.40 0.980 0.106

B. Field tests (n = 21 for HL & Hh, except Davidson w. 20)


La r2 SD (kN) h

a r2 SD (kN)

Reese 0.65 0.982 54.7 0.91 0.945 131. Hansen 1.00 0.986 48.8 1.39 0.948 128. Broms 0.85 0.979 59.1 1.19 0.956 117. Stevens 0.57 0.986 48.9 0.79 0.946 130. Davidson 0.91 0.984 52.7 1.23 0.954 119. Randolph 0.63 0.983 52.9 0.88 0.946 130.

C. Lab & field tests (n = 66 for HL & 68 for Hh; Dav = 65 & 67)


La r2 SD (kN) h

a r2 SD (kN)

Reese 0.65 0.985 30.3 0.91 0.957 71.6 Hansen 1.00 0.988 27.1 1.39 0.959 69.9 Broms 0.85 0.983 32.8 1.19 0.966 63.9 Stevens 0.57 0.988 27.1 0.79 0.958 71.0 Davidson 0.91 0.987 28.7 1.23 0.964 64.0 Randolph 0.63 0.986 29.4 0.88 0.958 71.1 a - slope of regression through origin, giving HL = L Hu or Hh = h

Hu, with Hu = predicted lateral capacity

that can not be compared directly. This point is evident when examining part C of the table. In this combined population, the regression slopes are controlled by the field data, which may be biased as noted above.

A more correct method of comparison for these data is to use the normalized capacities, which also are plotted as histograms in Figs. 3 & 4. A complete summary of these evaluations is given in Table 2, which shows the calculated capacities normalized first by HL and second by Hh. As noted previously, HL corresponds ap-proximately to initial failure of the shaft-soil system and effec-tively represents the lower bound. In contrast, H h corresponds approximately to the ultimate limit state and effectively represents the upper bound.

Examination of these normalized results indicates that the corresponding mean, S.D., and coefficient of variation (COV) for the laboratory and field data are reasonably comparable. There-fore, both data sets can be considered equally valid for evaluating the analytical models.

Evaluation of these analytical models shows that, for Hu/HL,Broms averages 0.78, Hansen averages 0.88, Davidson et al.average 1.03, and the other three overestimate and predict 1.19 to 1.53. For Hu/Hh, Stevens and Audibert average 0.99, Randolph & Houlsby average 0.83, Reese averages 0.78, Davidson et al. av-erage 0.66, Hansen averages 0.57, and Broms averages 0.50.

252

Table 2. Summary of normalized undrained lateral capacities.

A. Normalized by HL

Test No. Calculated Hu by Authora / HLb

R H B S&A D R&H

Lab 45 mean 1.18 0.88 0.76 1.52 1.03 1.28 S.D. 0.31 0.27 0.23 0.46 0.31 0.35 COV 0.26 0.31 0.30 0.30 0.30 0.27 Field 21 mean 1.20 0.90 0.83 1.54 1.02 c 1.30 S.D. 0.38 0.26 0.36 0.46 0.35 0.39 COV 0.32 0.29 0.43 0.30 0.34 0.30 Lab 66 mean 1.19 0.88 0.78 1.53 1.03 c 1.28 & S.D. 0.34 0.26 0.28 0.46 0.32 0.37 Field COV 0.29 0.30 0.36 0.30 0.31 0.29

B. Normalized by Hh.

Test No. Calculated Hu by Authora / Hhb HL/Hh

R H B S&A D R&H

Lab 47 mean 0.76 0.56 0.49 0.97 0.66 0.82 0.65 d

S.D. 0.21 0.18 0.15 0.31 0.21 0.24 0.08 COV 0.28 0.32 0.31 0.32 0.32 0.29 0.12Field 21 mean 0.81 0.59 0.52 1.03 0.66 c 0.86 0.64 S.D. 0.23 0.17 0.21 0.29 0.21 0.24 0.09 COV 0.28 0.29 0.40 0.28 0.32 0.28 0.14Lab 68 mean 0.78 0.57 0.50 0.99 0.66 c 0.83 0.65d

& S.D. 0.22 0.18 0.17 0.31 0.21 0.24 0.08Field COV 0.28 0.32 0.34 0.31 0.32 0.29 0.12a - R = Reese, H = Hansen, B = Broms, S&A = Stevens & Audibert,

D = Davidson et al., R&H = Randolph & Houlsby b - HL = lateral/moment limit, Hh = hyperbolic capacity c - n = 20 for field, n = 65 or 67 for lab & field d - n = 44 for lab, n = 65 for lab & field

Fig. 5. HL / Hh for undrained lateral tests on rigid drilled shafts.

Clearly, the Broms, Hansen, and Davidson et al. models corre-spond to a lower bound, while the Reese, Randolph and Houlsby, and Stevens and Audibert models are upper bounds.

One final point to note about Table 2 is that the average H L/Hhis 0.65, or about 2/3, which is consistent with previous observa-tions (Mayne et al., 1992; Chen & Kulhawy, 2003). These data are shown in Fig. 5 and clearly indicate comparable laboratory and field data populations.

5 INFLUENCE OF TIP RESISTANCE

In Table 2, the calculated capacities represent the lateral soil resistance alone, except for the Davidson et al. model that includes tip resistance. However, the load test results, and subsequently HLand Hh, include some tip resistance, even though the amount is unknown and has not been measured.

To provide a qualitative assessment of this tip effect, the test results were re-analyzed, assuming that the full possible tip resis-tance (Ht) was present and was equal to A tip su (DSS). This Htvalue then was subtracted from the HL and Hh values to give HLand Hh , and the analyses were re-done for all but the Davidson et al. model. Table 3 summarizes the normalized undrained lateral capacities without the full tip effect. It should be noted that some ill-conditioning developed where H t was large compared to HL or Hh.

Comparison of the results in Tables 2 & 3 reveals some im-portant points. First, all of the normalized capacities are increased when the tip resistance is included, with a rather large increase of about 34 percent on the lower bound (H L ) comparisons and a more modest increase of about 19 percent on the upper bound (H h )comparisons. This disproportionate increase on the lower bound assumes that the full available tip resistance has been mobilized at relatively small displacements. This effect is not likely, and only a modest percentage of the available tip resistance would be mobi-lized.

Second, the field data comparisons consistently give larger ra-tios than the laboratory data comparisons, which suggest that the field tests are less likely to develop the same tip resistance effects as the laboratory tests. Intuitively, this pattern should occur be-cause there are more controls in the laboratory.

Third, the COV on the upper bound (H h ) comparisons still is about 30 percent, which suggests comparable influence of the tip resistance on the overall data interpretations. However, the COV on the lower bound (HL ) comparisons has increased to about 45 percent for the laboratory data. This significant increase further suggests a disproportionate influence on these laboratory data.

Overall evaluation of these data suggests the following guide-lines for rigid drilled shafts. The Hansen or Davidson et al.models are appropriate to evaluate HL or the lower bound, and a modest tip resistance (1/4 to 1/3 of the maximum available) could be included in the Hansen model. Tip effects already are included in the Davidson et al. model. The Reese, Randolph & Houlsby, or Stevens & Audibert models are appropriate to evaluate Hh or the upper bound, and a large component of tip resistance (1/2 to 2/3 of the maximum available) could be included. There will always be more vagaries in field data, and therefore the specific site condi-tions at the test should be assessed carefully when assessing the tip resistance.

6 MODIFICATIONS FOR NON-RIGID FIELD TESTS

The non-rigid field tests also were evaluated, even though the analytical models presented are not directly applicable. In general, these models predict capacities that are larger than the test results, largely because the full shaft depth is considered. However, for the flexible shafts, the depth below the "hinge" or "yield" point without full tip effect does not materially contribute to the capac-ity.

253

Table 3. Summary of normalized undrained lateral capacities

A. Normalized by HL'

Test No. Calculated Hu by Authora / HL'b

R H B S&A R&H

Lab 41 mean 1.55 1.14 1.00 1.99 1.67 S.D. 0.66 0.53 0.44 0.92 0.74 COV 0.43 0.46 0.44 0.46 0.44 Field 13 mean 1.78 1.23 1.30 2.16 1.87 S.D. 0.47 0.34 0.38 0.58 0.49 COV 0.26 0.28 0.29 0.27 0.26 Lab 54 mean 1.61 1.16 1.07 2.03 1.72 & S.D. 0.61 0.48 0.43 0.84 0.68 Field COV 0.38 0.41 0.40 0.41 0.40

B. Normalized by Hh'.Test No. Calculated Hu by Authora / Hh

'b

R H B S&A R&H

Lab 41 mean 0.89 0.65 0.57 1.14 0.95 S.D. 0.26 0.22 0.17 0.38 0.29 COV 0.29 0.34 0.30 0.33 0.31 Field 14 mean 0.96 0.72 0.73 1.26 1.09 S.D. 0.29 0.20 0.23 0.33 0.28 COV 0.30 0.28 0.32 0.26 0.26 Lab 55 mean 0.91 0.67 0.61 1.17 0.99 & S.D. 0.27 0.21 0.19 0.37 0.29 Field COV 0.30 0.31 0.31 0.32 0.29 a - R = Reese, H = Hansen, B = Broms, S&A = Stevens

& Audibert, R&H = Randolph & Houlsby b - HL = lateral/moment limit, Hh = hyperbolic capacity

Still, it is worthwhile to examine whether these more flexible shafts can be modeled by simple rigid shaft approaches. One criterion (Poulos & Hull, 1989) considers the shaft to be rigid when D is less than one-third of the critical depth (D c), given by 4.44 (EcIc/Es)0.25, in which Ec = concrete modulus, Ic = concrete moment of inertia, and Es = soil modulus. Using D c/3 as the shaft depth, Hu was re-calculated. No corrections are made for tip effects, because the tip influence for a flexible shaft will be neg-ligible.

The results show that, for shafts of intermediate stiffness, they are similar to those for rigid shafts and are of comparable popula-tions, using Dc/3. However, this simple correction is not appli-cable for flexible shafts, which need criteria specific to flexible shafts.

7 CONCLUSIONS

The undrained lateral capacity was evaluated for straight-sided laboratory and field-scale drilled shafts in a wide variety of cohe-sive soil profiles. The lateral capacity was evaluated using several analytical models and consistent interpretation methods for load tests. These analyses showed the following:

(a) The laboratory and field data populations give comparable results when normalized properly, and therefore both are useful for comparison with available analytical models.

(b) The Hansen or Davidson et al. models are appropriate to evaluate HL or the lower bound. A modest tip resistance (1/4 to

1/3 of the maximum available) could be included in the Hansen model. For the Hansen model (without tip resistance), the ratio of capacity to HL is 0.88 with n = 66, S.D. = 0.26, and COV = 0.30. For the Davidson et al. model, the ratio of capacity to HL is 1.03 with n = 65, S.D. = 0.32, and COV = 0.31.

(c) The Reese, Randolph and Houlsby, or Stevens and Audibert models are appropriate to evaluate Hh or the upper bound, and a larger tip resistance (1/2 to 2/3 of the maximum available) could be included. For the Reese model, the ratio of capacity to H h is 0.78, with n = 68, S.D. = 0.22, and COV = 0.28. For the Randolph and Houlsby model, the ratio of capacity to Hh is 0.83 with n = 68, S.D. = 0.29, and COV = 0.29. For the Stevens and Audibert model, the ratio of capacity to Hh is 0.99, with n = 68, S.D. = 0.31, and COV = 0.31.

(d) Shaft rigidity greatly influences the lateral capacity. For intermediate and flexible shafts, the computed capacities from the analytical models are greater than the interpreted capacities. By using Dc/3 for D in the analytical models, shafts of intermediate stiffness give results comparable to those of rigid shafts. However, this approach is not applicable for flexible shafts.

REFERENCES

Bierschwale, M.W., Coyle, H.M. & Bartoskewitz, R.E. 1981. Lateral load tests on drilled shafts. Drilled Piers & Caissons:98-113. MW O�Neill (Ed.). New York: ASCE.

Broms, B.B. 1964. Lateral resistance of piles in cohesive soils. Journal of Soil Mechanics & Foundations Division , ASCE,90(SM2): 27-63.

Carter, J.P. & Kulhawy, F.H. 1992. Analysis of laterally loaded shafts in rock. Journal of Geotechnical Engineering , ASCE118(6):839-855.

Chen, Y.J. 1993. Case history evaluation of behavior of drilled shafts under axial and lateral loading. PhD Thesis, Cornell University, Ithaca, New York, U.S.A.

Chen, Y.J. & Kulhawy, F.H. 1994. Case history evaluation of behavior of drilled shafts under axial and lateral loading. Re-port TR-104601, EPRI, Palo Alto: 356p.

Chen, YJ & Kulhawy, FH 2003. Drained lateral loading for drilled shafts. Proceedings of the 12th Asian Regional Conference on Soil Mechanics & Geotechnical Engineering : 595-598. Sin-gapore.

Davidson, H.L., Cass, P.G., Khilji, K.H. & McQuade, P.V. 1982. Laterally loaded drilled pier research. Report EL-2197, EPRI, Palo Alto: 448p.

Hansen, J.B. 1961. Ultimate resistance of rigid piles against transversal forces. Bulletin Danish Geotechnical Institute : 5-9. Copenhagen.

Hirany, A. & Kulhawy, F.H. 1989. Interpretation of load tests on drilled shafts - Part 3: Lateral and moment. Foundation Engi-neering: Current Principles & Practices (GSP 22) : 1160-1172.(Ed.) FH Kulhawy. New York: ASCE.

Kulhawy, F.H. & Chen, Y.J. 1995. A thirty year perspective on Broms' lateral loading model, as applied to drilled shafts. Proceedings Broms Symposium : 225-240. Singapore: World-wide Scientific.

Kulhawy, F.H. & Mayne, P.W. 1990. Manual on estimating soil properties for foundation design, Report EL-6800, EPRI, Palo Alto: 306.

Manoliu, I., Dimitriu, D.V., Radulescu, N. & Dobrescu, G. 1985. Load-deformation characteristics of drilled piers. Proceedings of the 11th International Conference on Soil Mechanics & Foundation Engineering (3) : 1553-1558. San Francisco.

254

Mayne, P.W., Kulhawy, F.H. & Trautmann, C.H. 1992. Experi-mental study of undrained lateral and moment behavior of drilled shafts during static and cyclic loading. Report TR-100221, EPRI, Palo Alto: 383p.

Poulos, H.G. & Davis, E.H. 1980. Pile Foundation Analysis and Design. New York: Wiley.

Poulos, H.G. & Hull, T.S. 1989. Role of analytical geomechanics in foundation engineering. Foundation Engineering: Princi-ples & Practices (GSP 22): 1578-1606. FH Kulhawy (Ed.). New York: ASCE.

Randolph, M.F. & Houlsby, G.T. 1984. Limiting pressure on circular pile loaded laterally in cohesive soil. Geotechnique34(4): 613-623.

Reese, L.C. 1958. Discussion of soil modulus for laterally loaded piles, Transactions ASCE 123: 1071-1074.

Stevens, J.B. & Audibert, J.M.E. 1979. Re-examination of p-y curve formulations. Proceedings of the 11th Offshore Tech-nology Conference (1) : 397-403. Houston.


255

A Time-Related Load Transfer Model for Bearing Behavior of Driven Piles

Y. Huang, Y.Q. Tang & W. M.Ye Department of Geotechnical Engineering, Tongji University, Shanghai 200092, China [email protected]

L. L. Zhao Shanghai Tunnel Engineering Co., Ltd., Shanghai 200032, China

Abstract: There are many researches on the phenomenon of time-dependent capacity related to pile driving. However, very few datahave been published on methods of predicting the variation of load-settlement curves with time after driving. The main objective of this paper is to quantify effects of time on the load-settlement behavior of a single pile. Based on the mechanism related to the variation of load-settlement behavior with time, a new time-related load transfer model is proposed to estimate the load-settlement behavior of a single pile after driving. Two nondimensional time factors are added into traditional load transfer functions in order to describe the in-crease in both shaft friction and base resistance with time. For verification, the proposed time-related load transfer model is used to cal-culate the load-settlement behavior after pile installation based on a case history. Good agreement is found between calculated and measured load-settlement curves of static load tests.

1 INTRODUCTION

Load-settlement curve is one of the most important behaviour of a single pile, from which geotechnical engineers may evaluate the load transfer character between piles and soil easily. It is well known that the capacity of a driven pile changes with time after installation. A lot of observed data has indicated that the bearing capacity of driven piles is not a constant value, but changes with the time, and finally tends to a certain steady value (e.g. Long et al., 1999). This phenomenon is called pile setup.

For having great economic benefits, until now, there are many researches on the phenomenon of time-dependent capacity re-lated to pile driving. However, most of them are limited in pile ultimate bearing capacity. On the contrary very few literatures have been published on methods of predicting the variation of load-settlement curves with time after driving. Therefore, the main objective of this paper is to quantify effects of time on the load-settlement behavior of a single pile.

2 MECHANISM

Previous researches have indicated that pile bearing capacity show the increasing trend with time particularly in clayey soils. From the current state of knowledge for time effects of driven piles, in general, the mechanisms related to the variation of load-settlement behavior with time mostly include the following two reasons, i.e. the migration of pore water and thixotropy.

Firstly, excess pore water pressures generates during pile in-stallation, causing a reduction in effective stress. As the excess pore water pressure dissipates with time, the effective stress in the soil will increase. Thus, shaft friction as well as the ultimate pile bearing capacity increases too. On the other hand, during pile driving, thixotropy occurs when the soil is seriously dis-turbed by vibration. Since thixotropy is a reversible process, soil

resistance recovers apparently after installation, which also leads to an increase in the bearing capacity of the pile with time. Of course, the load-settlement behavior of pile also depends on some other factors, including driven depth, pile geometrical shape, installation method, and loading conditions.

3 ANALYSIS METHOD

The current analysis methods for load-settlement behavior of a single pile, which have been commonly performed, are mainly the load transfer method, the elastic analytical method, the finite element method and so on. In this paper, we use the compara-tively simple one of them, i.e. the load transfer method (e.g. Coyle et al., 1966), from which we can get the full load-settlement curve of a single pile. This method views the single pile as a number of segments that are assumed to be connected with soil by a series of nonlinear springs. The load transfer char-acteristics between soil and pile usually referred to as load trans-fer functions. For the interface component of this model, the shaft friction per unit pile length, qs, is related to the relative dis-placement of the pile-soil interface, s. The second component re-fers to the base resistance, Qp with pile tip displacement sp. In this study, the corresponding linear elastic-perfectly plastic load transfer function is used in Fig. 1. Therefore, the load transfer function for shaft friction is:

qs=Cs · s, when s<su (1)

qs=qsu, when s=su and s>su (2)

where, qsu is the ultimate value of shaft friction at the ultimate relative displacement su. The load transfer function for base resis-tance Qp is:

256

Cs

su

qsu

qs

1

s

Fig. 1. Linear elastic-perfectly plastic load transfer function.

Qp=kp · Ap · sp (3)

where kp is the subgrade reaction coefficient, and Ap is the plan area of the pile base.

In order to describe the increase in both shaft friction and base resistance with time respectively, two nondimensional experien-tial parameters, Ist and Ipt, may be added into the above load transfer functions. Thus, the modified load transfer functions can take into account load-settlement behavior at an actual time after pile installation. The shaft friction, qst at time t can be calculated from:

qst= Ist · Cs · s, when s<su (4)

qst= Ist · qsut, when s=su and s>su (5)

where Ist is the time effect coefficient for shaft friction at time t.In addition, the base resistance is

Qpt= Ipt · kp · Ap · sp (6)

where, Ipt is the time effect coefficient for ending bearing at time t. Values of Ist and Ipt may be determined by laboratory tests, instrumented field tests, and empirical data in specific areas. In previous mechanism analysis, the time effect coefficients are determined by calculating the effects of pore water migration and thixotropy on pile bearing behavior. Above all, the influence of excess pore water may be predicted from the consolidation state. For instance, the excess pore water pressure generated during pile installation could be estimated by the elastic-plastic theory on cavity expansion. Then, from the Terzaghi or Biot consolida-tion theory, it is easy to get the dissipation ratio at any time. Thus, the increase of pile-soil strength could be obtained by the history of effective stress. In addition, the effect of pore water migration can be determined by the empirical method. Soderberg (1962) proposed a nondimensional parameter to describe this kind of in-fluence, which was related to intermission time, pile diameter and coefficient of consolidation. On the other hand, the effect of thixotropy can be obtained in terms of laboratory tests of dis-turbed sample soil. Figure 2 shows the strength increases

0 5 10 15 206

8

10

12

14

Time (day)

Fig. 2. Increase of strength with time of remolded clay (Li et al.,1992)

with time of the remolded soft saturated clay near pile side at dif-ferent time interval (Li et al., 1992).

4 APPLICATION

With the proposed method, we developed a program for simulat-ing time effects on load-settlement behavior of a single pile and then investigated the load tests of a driven precast concrete pile at the 14th day and 50th day after installation in Shanghai, China for verification. The pile is 25.6 m long and 300 mm in diameter. At the end of driving, the penetration at pile termination of about 12mm/blow (with 1.8 t hammer and 1m drop) was recorded. The soils at the site consist of deep saturated soft clayey soils. The engineering geological properties at the test site are listed in Ta-ble1.

The Incremental-Static-Load Test was carried out to deter-mine load-settlement relationships. The tests followed the stan-dard test method for piles under static load prescribed in the Shanghai Code for Design of Building Foundation (DGJ08-11-1989). The applied load was increased in equal increments, each increment being one tenth of the predicted ultimate toe resistance. The load was held at each increment for 60 min.

In the analysis, the pile is divided into 32 segments. Accord-ing to the previous research and experiences, the time effect on base resistance can be ignored in Shanghai area and the coeffi-cient Ist for shaft friction may be determined as 0.50 at the 14th day and 0.65 at the 50th day. It is necessary to point out that the values of the above two parameters were assumed based on the local experiences in this research.

The comparison between calculated results and measured re-sults are shown in Tables 2 & 3, Figs. 3 & 4. Good agreement could be found between calculated and measured load-settlement curves of static load tests.

257

Table 1. Properties of soil strata at the test site.

Layer Type Elevation/m c/kPa /

2 Silty clay 0.76~-1.88 14 14.0

3-1 Mucky silty clay -1.58~-0.14 7 10.1

3-2 Sandy silt -3.17~-1.71 4 21.9

3-3 Mucky silty clay -4.63~-3.97 7 8.6

4-1 Mucky clay -14.49~-13.61 9 8.1

4-2 Clay -17.03~-16.11 12 9.8

5-1-1 Silty clay -18.83~-17.35 12 13.1

5-1-2 Silty clay -20.95~-19.72 11 14.1

6 Silty clay -25.67~-24.01 29 15.3

0

5

10

15

20

25

30

35

40

45

0 2 4 6 8 10Q /100kN

Predicted Measured

Fig. 3. Measured load-settlement (Q-s) curve and calculated re-sults (14 days after pile driving).

Table 2. Comparison between calculated results and measured values (14 days after pile driving).

Q/kN s/mm

(Predicted)

s/mm

(Measured) Error/%

224 1.30 1.25 4.00

336 2.11 2.17 2.76

448 3.54 3.64 2.75

560 6.09 6.25 2.56 672 9.59 9.75 1.64

784 14.23 14.86 4.24

896 41.61 43.53 4.41

0

1

2

3

4

5

6

7

8

9

10

0 2 4 6 8 10 12Q /100kN

Predicted Measured

Fig. 4. Measured load-settlement (Q-s) curve and calculated re-sults (50 days after pile driving).

Table 3. Comparison between calculated results and measured values (50 days after pile driving).

Q/kN s/mm

(Predicted)

s/mm

(Measured) Error/%

224 1.12 1.17 4.27

336 1.80 1.83 1.64

448 2.59 2.58 0.39

560 3.50 3.43 2.04

672 4.49 4.37 2.75

784 5.52 5.39 2.41

896 6.55 6.49 0.92

100.8 7.99 7.70 3.77

1120 9.45 9.37 0.85

5 CONCLUSIONS

In the paper, a method of analysis was proposed to calculate the load-settlement behavior of driven piles. Based on the mecha-nism of time effects, two nondimensional time factors are added into traditional load transfer functions in order to describe the in-crease in both shaft friction and base resistance with time respec-tively. Then, the modified load transfer functions can take into account load-settlement behavior at an actual time after pile in-stallation.

It is necessary to point out that the method belongs to the semi-empirical method. The time factors should be determined on the basis of laboratory tests and instrumented field tests as well as local experiences.

258

ACKNOWLEDGMENTS

The research reported herein was supported by the Shanghai Education Commission, through the Shanghai Municipal Key Discipline Development Project (Geotechnical Engineering). This support is gratefully acknowledged.

REFERENCES

Coyle, H.M. & Reese, L.C. 1966. Load transfer for axially loaded piles in clay. Journal of the Soil Mechanics and Foun-dations Division, ASCE 92(2): 1-26.

Li, X. & Liu, J.L. 1992. Time effect of bearing capacity of driven pile in saturated soft soil. Chinese Journal of Geotechnical Engineering 14(6): 9-16.

Long, J.H., Kerrigan, J. A. & Wysockey, M. H. 1999. Measured time effects for axial capacity of driven piling. Transportation Research Record 1663: 8-15.

Shanghai Construction Commission 1989. Shanghai Code for Design of Building Foundation (DGJ08-11-1989) .

Soderberg, L.O. 1962. Consolidation theory applied to founda-tion pile time effects. Geotechnique 12(3): 217-225.


259

Lateral Bearing Capacity Prediction of Caisson Type of Seawalls in Difficult Subsoil Conditions

N. D. Kumar Ph.D Scholar, Department of Ocean Engineering, IIT Madras, Chennai-36, India. [email protected]

S. N. Rao Professor, Department of Ocean Engineering, IIT Madras, Chennai-36, [email protected]

V. Sundar Professor, Department of Ocean Engineering, IIT Madras, Chennai-36, [email protected]

Abstract: In this paper, the lateral capacity of caisson seawall in marine clay has been brought out. The load-ground level deflection curves and the variations in lateral capacities with embedment depth, load eccentricities and consistency of soil are presented and discussed. The capacities estimated from the methods reported in literature, are over conservative at lower embedment depth ratios. The estimated lateral capacities are close to the observed values at higher embedment depth ratios and lower load eccentricity ratios. Using curve-fitting method for the observed data, an equation is suggested for the estimation of lateral bearing capacity of caissons in clay. This equation is validated for an independent data. This procedure can be extended for the estimation of capacities in caissons.

1 INTRODUCTION

In marine situations, one could find large stretches of soft clay deposits, which in general, pose lot of problems to geotechnical engineers. One of the important aspects in the design of foundation is the estimation of load carrying capacity. Caissons can be used as water front structures in marine environment. In the coastal protection works, caissons can be used as seawalls. These caissons embedded in clay, might experience large amount of lateral forces due to waves and currents. The exact prediction of lateral load carrying capacity of caissons and their depth of embedment into these deposits lead to safe performance as seawalls. The lateral capacity of this type of embedded caisson used in seawalls depends on the passive resistance of surrounding soil and the size of the structure. There are a few approaches reported in literature for the lateral capacity of caissons in clay. Some of the theories related to rigid piles are extended to study the lateral behaviour of caissons. There can be few limitations in extending the rigid pile concept to predict its lateral behaviour of caissons.

Hansen (1961) suggested a method for the estimation of ultimate lateral resistance of rigid piles founded in cohesive and cohesionless soils. Broms (1964) developed a set of charts for estimating the ultimate lateral capacity of rigid piles in cohesive soils and also suggested a simplified distribution of ultimate soil resistance with depth in terms of passive earth pressure. Girija Vallabhan (1981) used nonlinear soil spring concept for the lateral soil resistance to estimate the lateral capacity. Sastry et al.(1986) extended the lateral earth pressure theories of Terzaghi (1943) and Hansen (1961) for estimating the ultimate loads in both sands and clays for rigid piles. Brettmann & Duncan (1996) suggested a method called characteristic load method (CLM) for the analysis of laterally loaded piles and drilled shafts to estimate

the ground level deflection, accurately in clay. Prasad & Rao (1996) studied the lateral capacity of helical piles in clays. A simplified approach for lateral load carrying capacity of rigid piles in clays was developed (Rao et al., 1996). Leung et al.(1997) brought out the influence of width of caisson on lateral displacement. Zen et al. (1998) reported the field application of caissons and suggested the use of suction force in the installation of caissons. Kanatani et al. (2001) developed the deformation analysis for the caisson type of seawalls with an armoured embankment-using centrifuge shaking table tests.

Ohmaki et al. (2001) performed the studies on skirted caisson foundation and brought out the influence of embedment on lateral load carrying capacity. With increase in skirt embedment, the lateral capacity improvement was observed. Yuxia & Randolph (2002) considering the soil as normally consolidated clay in which undrained shear strength increases with depth, the soil flow mechanisms was developed for caissons with embedment ratios up to 5. It was found that with increasing embedment ratio, the soil flow mechanism changed from surface failure to a deep cavity expansion mode and this transition occurs at a higher embedment ratio.

From the aforementioned review, it is felt that there is a need to make an in depth study in predicting the lateral behaviour of caisson in clayey soil. An attempt has been made to arrive at predictions for lateral capacity of caisson in marine clay using the test results obtained from a controlled testing in a laboratory.

2 EXPERIMENTAL WORK

2.1 Model Caisson Used

Model caissons were made out of mild steel pipes of 105 mm outer diameter and 10 mm wall thickness. The embedment length

260

(L) to diameter (D) ratio (L/D) of model caissons investigated was varied from 2 to 4. The load was applied with a load of eccentricity (e), equal to 0.5, 1.0, 1.5 and 2D of the model caisson. Consistency index of soil considered in this investigation was varied from 0.22 to 0.62. Model caissons were prepared , keeping in view the fixing and removal of instrumentation and its rigidity.

The rigidity of the model caissons has been estimated as per the Poulos (1971). The pile flexibility factor is expressed as:

4

E Ip pkR

E Ls e (1)

where kR= Pile flexibility factor, Ep= Pile material Young�s modulus, Ip= Moment of inertia of pile material, Es= Soil modulus and Le= Embedment length of pile in soil. As per this criterion, all the model caissons used in this investigation are found to be rigid.

2.2 Soil Used

Marine clay from the coastal deposit of Chennai, India was used in this investigation. The liquid limit and plastic limit of soil are 48% and 18% respectively. This soil is composed of 38% clay, 32% silt and 30% fine sand. The soil is classified as medium compressible clay (CI).

2.3 Test Set Up

The tests were conducted in a mild steel rectangular tank of size, 1.2 m x 0.8 m x 0.90 m (Fig.1). The test tank is chosen such that the size effects of model caissons adopted in this investigation were minimum. A special stress controlled pneumatic loading device was designed and fabricated to apply the loading. The pneumatic loading system consisted of (i) an air compressor of adequate capacity with pressure chamber, (ii) 5 port two-way electrical double solenoid valve, (iii) double acting pneumatic power cylinder, and (iv) the pressure regulator valve with filter. Using this system, the lateral static loading was transferred to the model through an appropriate piston arrangement. A load cell was connected to the piston rod to measure the lateral load. The inductive type LVDT was placed at the ground level to monitor the deflections. The outputs through cables connected to amplifiers were recorded in the system through data acquisition unit.

2.4 Testing Procedure

Clayey soil was mixed thoroughly with the required amount of water to maintain a consistency index of 0.22, 0.42 and 0.62 (Consistency Index, Ic is the ratio of difference of Liquid limit and natural water content to plasticity Index). In the test tank, after placing the model caisson vertically, the soil was placed in layers of 50mm thickness with hand packing and pressed by jacking a template to remove entrapped air and to ensure homogeneous packing. The full saturation of soil was confirmed by the measurements of pore water pressure parameter of B=0.98 to 0.99 in a triaxial test set up (Bishop & Henkel, 1962). Static lateral load tests were conducted to arrive at ultimate lateral load at Ic = 0.22, 0.42 and 0.62 for caissons with L/D ratio of 2,3 & 4 and for e/D values of 0.5, 1.0 1.5 and 2. Static lateral loads were applied in increments through pneumatic system. At every load increment, it was waited until the stabilization in deformation had reached. The ultimate load was obtained from the lateral load-

ground level deflection plots, corresponding to a ground level deflection of 0.2 times the diameter of the model caisson (Broms, 1964). The ground level deflections and actual lateral load acting were recorded for every load increment. In this paper, the test results corresponding to L/D= 2, 3 & 4, e/D= 0.5, 1.0, 1.5 & 2.0 and for Ic of 0.22, 0.42 and 0.62 are presented and discussed.

3 RESULTS AND DISCUSSIONS

From the experimental results, some of the interesting findings in relation to lateral capacity of caissons in marine clay are brought out. Figure 2 shows a typical variation in ground level deflection with lateral load for a caisson model of L/D=4 and I c= 0.22. From these results, it is observed that with increase in e/D, there is a significant decrease in the capacities.

The ultimate lateral loads observed and estimated from the available theories (Hansen, 1961; Broms,1964; Rao et al., 1996) for various conditions are presented in Table.1.

The percentage difference in the observed values corresponding to ground level deflections of 0.1D and 0.2D are brought out and these values are presented in Table 2. From these results, it is clear that there is not much difference in lateral capacities observed between deflections of 0.1D and 0.2D. This difference is found to be 10 to 17%. This indicates that there is an enormous increase in deflection even for small load increment after a ground deflection of 0.1D.

Figures 3, 4 & 5, show the variation in ultimate lateral capacity with embedment depth, load eccentricity and consistency of soil respectively. From Fig.3, it is observed that there is a good increase in lateral load carrying capacity with embedment depth ratio (L/D) and the increase capacity is about 60% as L/D changes from 2 to 4. From Fig.4, it is observed that there is a decrease in ultimate lateral capacity with increase in load eccentricity at all the embedment depths of caisson tested.

12

1

217

13

16

15

11

10

9

7

5

3

68

14

4

1. Model caisson, 2. Soil, 3. Test tank, 4. Test frame, 5. LVDT, 6. Extension rod, 7. Load cell, 8.Piston rod, 9.Pneumatic power cylinder, 10. Pressure gauge, 11. Solenoid valve, 12. Electronic timer, 13. Pressure regulator, 14. Compressor air chamber, 15. Carrier frequency amplifier, 16. BNC box, 17. Data acquisition system.

Fig. 1. Schematic diagram of test set up.

261

Table 1. Observed and estimated ultimate lateral capacities at Ic = 0.42.

Ultimate Lateral Load (N) L/D Method

e/D

2 3 4 0.5 285.00 530.00 710.00 1.0 240.00 440.00 620.00 1.5 200.00 380.00 575.00

Observed

2.0 160.00 290.00 495.00 0.5 262.00 470.00 708.30 1.0 207.00 393.20 616.00 1.5 170.00 337.50 544.00

Hansen�s (1961)

2.0 144.00 232.00 433.20 0.5 134.00 363.00 700.00 1.0 114.00 323.00 605.00 1.5 80.00 262.00 444.00

Broms (1964)

2.0 67.00 202.00 336.00 0.5 118.00 369.00 680.00 1.0 97.00 312.00 590.00 1.5 79.00 268.00 521.00

Rao et al.(1996)

2.0 68.00 236.00 467.00

Figures 5 & 6 show the increase in ultimate lateral capacity with consistency. There is a good improvement in the lateral capacity with consistency at all the embedment depths tested, where as changes in the capacity with load eccentricity, e at different values of Ic are not significant. With increase in consistency of soil, there is a good improvement in strength and for the higher embedment depths of caisson the projected area for mobilization of passive resistance is more. Hence, the lateral capacity observed is more.

Table 2. Lateral capacities observed at ground level deflections of 0.1D and 0.2D at Ic=0.42.

Ultimate Lateral Load (N) L/D

e/D

0.2D 0.1D % Difference

0.5 285 255 11.761.0 240 212 13.211.5 200 180 11.11

2

2.0 160 140 14.290.5 530 460 15.221.0 440 375 17.331.5 380 325 16.92

3

2.0 290 250 16.000.5 710 605 17.361.0 620 535 15.891.5 575 495 16.16

4

2.0 495 425 16.47

Figure 7 explains the variation in observed and estimated ultimate lateral capacities with embedment depth. From this plot, it is seen that the observed capacities are higher than the values estimated from the reported theories. Capacities from Hansen�s (1961) theory are closely matching with observed values from the tests. The capacities estimated form Broms (1964) and Rao et al.(1996) are conservative in predicting the lateral capacities of caisson.

Fig. 2. Variation in ground level deflection with lateral load.

Ic=0.22, L/D=4

0.00

0.10

0.20

0.30

0.40

0.50

0 5 10 15 20 25 30 35 40

Ground level deflection (mm)

e/D=0.5e/D=1.0e/D=1.5e/D=2.0

Ic=0.22

0.00

0.10

0.20

0.30

0.40

0.50

1.50 2.50 3.50 4.50

Embedment depth ratio (L/D)

e/D=0.5e/D=1.0e/D=1.5e/D=2.0

Fig. 3. Variation in ultimate lateral capacity with embedment depth ratio (L/D).

262

e/D=1

0.00

0.20

0.40

0.60

0.80

1.00

1.20

0.20 0.28 0.36 0.44 0.52 0.60 0.68

Consistency Index (Ic)

L/D=4L/D=3L/D=2

Fig. 5. Variation in ultimate lateral capacity with consistency of clay (Ic).

L/D=4

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

0.20 0.28 0.36 0.44 0.52 0.60 0.68

Consistency Index (Ic)

e/D=0.5e/D=1.0e/D=1.5e/D=2.0


e/D=2, Ic=0.22

0.00

0.05

0.10

0.15

0.20

0.25

0.30

1.5 2.0 2.5 3.0 3.5 4.0 4.5

Embedment depth ratio (L/D)

ObservedHansens (1961)Rao et al (1996)Broms (1964)

Fig. 7. Variation in observed and estimated ultimate lateral capacity with embedment depth ratio of caisson (L/D).

Ic=0.22

0.00

0.10

0.20

0.30

0.40

0.50

0.00 0.50 1.00 1.50 2.00 2.50

Load eccentricity ratio (e/D)

L/D=4L/D=3L/D=2

Fig. 4. Variation in ultimate lateral capacity with embedment depth ratio (L/D).

263

From the lateral load-deflection data obtained from the model tests conducted and the data reported by earlier investigators (Mayne, 1994; Vallabhan, 1983), the normalized plots are drawn in the form of (P/Pu) and (y/D) shown in Fig 8. From this, it is seen that the observed and reported normalized data are closely matching. The independent data reported by Vallabhan et al.(1983) and Mayne et al. (1994) along with the author�s data have been used in these plots. The lateral load-deflection data obtained at different load eccentricity ratios is presented in Fig. 9, in the form of normalized plots between P/Pu and y/D. From this normalized data presented in Fig 9, a unique relation is fitted between P/Pu and y/D as shown in Eq. (2).

2

9.7805 5.8392 0.28P y y

P D Du(2)

From the above equation (2), for the known values of ultimate lateral capacity (Pu) and size (D) of the caisson, the lateral capacity (P) of caissons for a specified ground level deflection (y) can be estimated in clay.

The observed test data is put in non-dimensional form, and curves are drawn between (Pu/cuD2) and (L/D) for various values of e/D. These curves are shown in Fig 10. From the data presented in Fig 10, a linear fit has been made for the non-dimensional parameters of (Pu/cuD2), (L/D) and (e/D). An equation has been fitted from the data and is given in equation (3).

21.36 0.967 9.667 102

P L eu

c D D Du(3)

Ultimate lateral capacity of caissons can be estimated in clay from the Eq. (3) for the known values of (L/D), (e/D), D and c u.

4 CONCLUSIONS

(1) There is not much variation in observed ultimate lateral capacities corresponding to a ground level deflection between 0.1D and 0.2D.

L/D=4, e/D=1

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

0.00 0.10 0.20 0.30 0.40 0.50

(y/D)

Mayne et al (1994)Vallabhan et al (1983)Observed

Fig. 8. Comparison of test data with reported data in the normalised form (P/Pu) and (y/D).

Ic=0.42

0.00

1.00

2.00

3.00

4.00

5.00

6.00

1.50 2.00 2.50 3.00 3.50 4.00 4.50

L/D

e/D=0.5e/D=1.0e/D=1.5e/D=2.0

Fig. 10. Non-dimensional plots for ultimate lateral capacity.


Ic=0.42, L/D=4

(P/Pu) = -9.7805(y/D)2 + 5.8392(y/D) + 0.28R2 = 0.9417

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

0.00 0.10 0.20 0.30 0.40

(y/D)

e/D=0.5e/D=1.0e/D=1.5e/D=2.0Poly. (fit)

264

(2) Up to about a ground level deflection of 0.1D, the lateral load taken by caisson is more and thereafter for even small load increment, the ground level deflection of a caisson observed is quite enormous.

(3) There is significant increase in the ultimate lateral capacity with embedment depth and consistency of soil. The influence of load eccentricity on ultimate lateral capacity is not significant.

(4) At higher embedment depths of caissons, the observed and estimated lateral capacities are closely matching.

(5) The estimated lateral capacities are very much on the conservative side at lower embedment depths and whereas at higher embedment depths, the estimated and observed capacities are nearly equal.

(6) Based on the fitting of observed data, an equation has been proposed to estimate the ultimate lateral capacity of caissons in clay. Independent published data also have been used in these formulations.

(7) Keeping in view of all these, it is possible to design the embedded caisson in marine clay to serve as seawall for better performance and to resist the lateral loads.

REFERENCES

Bishop, A.W. & Henkel, D.J. 1969. The Measurement of Soil Properties in the Triaxial Test. 2nd Ed. London: Edward Arnold.

Brettmann, T. & Duncan, J. M. 1996. Computer application of CLM laterally load analysis to piles and drilled shafts. Journal of Geotechnical Engineering Division, ASCE 122(6): 496 � 498.

Broms, B. B. 1964. The lateral resistance of piles in cohesive soils. Journal of the Soil Mechanics & Foundations Division, ASCE 90(SM2): 27-63.

Girija Vallabhan. 1981. Short rigid piers in clays. Journal of Goetechnical Engineering Division, ASCE 108 (GT10): 255-272.

Hansen, B. 1961. The ultimate resistance of rigid piles against transversal forces. Bulletin 12, Geoteknisk Institut. Copenhagen.

Kanatani, M. et al. 2001. Prediction method on deformation behaviour of caisson type seawall covered with armored embankment on man-made islands during earthquakes. Soils and Foundations 41(6): 79-96.

Leung, C.F. et al. 1997. Behaviour of gravity caisson on sand. Journal of Geotechnical and Geoenvironmental Engineering, ASCE 123(3): 187-196.

Mayne, P.W., Kulhawy, F.H and Trautmann, C.H. 2001. Laboratory modeling of laterally loaded drilled shafts in clay. Journal of Geotechnical and Geoenvironmental Engineering, ASCE 121(12) 827-835.

Ohmaki, S. & Nishizaki, T. 2001. Bearing capacity and deformation characteristics of a skirted foundation on soft cohesive ground. Soft Soil Engineering 8: 231-236. Lee et al.(Eds), Swets and Zeitlinger.

Poulos, H.G. 1971. Behaviour of laterally loaded piles: I- Single piles. Journal of Soil Mechanics & Foundations Division,ASCE 97(SM5): 711-751.

Prasad, Y.V.S.N. & Rao, S.N. 1996. Lateral capacity of helical piles in clays. Journal of Geotechnical and Geoenvironmental Engineering, ASCE 122(11): 938-941.

Rao, S.N. et al. 1996. A simplified method of calculating the lateral load carrying capacity of rigid piles in clay. Ground Engineering 29(9): 38 - 40.

Sastry, V. V. R. N. et al. 1986. Behaviour of rigid piles in layered soil under eccentric and inclined loads. Canadian Geotechnical Journal (23): 451- 457.

Yuxia Hu & Mark, F. Randolph. 2002. Bearing capacity of caisson foundations on normally consolidated clay. Soils and Foundations 42(5): 71-77.

Zen, K. et al. 1998. Case history on the penetration of caisson-type foundations into seabed by the use of suction force. Journal of Soil Mechanics and Foundation Engineering603(III-44): 21-34. (In Japanese).


265

Influence of Pile Head Connection Condition on Behavior of Model Piled Raft Foundations in Sand: Shaking Table Tests at 1-G Gravitational Field

K. Fukumura, T. Matsumoto & A. Oki Kanazawa University, 2-40-20 Kodatsuno, Kanazawa 920-8667, Japan [email protected]

K. HorikoshiTechnology Center, Taisei Corporation, 344-1 Nase Cho, Totsuka-ku, Yokohama 245-0051, Japan [email protected]

Abstract: A series of seismic load tests were carried out on model piled rafts in dry sand by using a shaking table at 1-g gravitational filed. An emphasis is placed on the influence of the connection condition between the pile head and the raft on the behaviour of the piled raft foundations, such as horizontal acceleration, horizontal displacement and inclination (rocking motion) of the foundation. Fur-thermore, a series of static horizontal load tests on the same model piled rafts were conducted to compare the behaviour of the model piled rafts under static and dynamic horizontal loading. Some parts of the results from both test series are presented and discussed in this paper.

1 INTRODUCTION

Piled raft foundations have been widely recognized as an eco-nomical and rational type of pile foundations when they are sub-jected to vertical loading, because the vertical load is supported by the raft as well as the piles, resulting in smaller settlements with a reduced number of piles compared to free-standing pile groups (e.g., Poulos & Davis, 1980; Randolph, 1994; Horikoshi & Randolph, 1999; Katzenbach & Moorman, 2001). In highly seismic areas such as Japan, estimation of the be-haviour of pile groups and piled rafts subjected to horizontal loading or seismic loading becomes a vital issue in seismic de-sign of pile foundations. Behaviour of model piled rafts and model pile groups subjected to static or dynamic horizontal loads have been intensively investigated in 1-g field model tests (Pastsakorn, et al., 2002; Fukumura et al., 2003) and in centri-fuge modelling (Horikoshi et al., 2003a; Horikoshi et al.,2003b). These test results show that piled rafts are also eco-nomical and rational foundations even for horizontal loading. In the centrifuge modelling by Horikoshi et al. (2003a, b), a focus was placed on the influence of the pile head connection condi-tions, rigid or hinged, on the behaviour of piled raft models with limited test conditions.

In this paper, a series of dynamic horizontal load tests by us-ing a shaking table at 1-g gravitational field as well as static horizontal load tests were carried out on model pile rafts in dry sand, in order to compare the test results with those from Horikoshi et al. (2003a, b).

2 SIMILARITY RULE FOR 1-G FIELD MODEL TEST

It is important to consider the similarity rule to deduce the behav-iour of a prototype structure from the behaviour of the corre-sponding model. Dry Toyoura sand was used for the model grounds in both of the 1-g model tests in this study and the cen-trifuge modelling by Horikoshi et al. (2003a, b). The physical properties of Toyoura sand are summarised in Table 1. The rela-

tive density, Dr, of the model ground used in the centrifuge mod-elling was about 60 %, while that of the model ground used in the 1-g model tests was 95 %.

A series of triaxial consolidated drained shear tests (CD test) were carried out on soil specimens of Dr = 95 %, 100 mm high and 50 mm radius, with different confining pressures, p0, of 50, 100, 200 and 300 kPa. Another series of CD tests were carried out on soil specimens of Dr = 65 % which was nearly equal to Drof the model ground used in the centrifuge modelling. The deviator stress, q, versus, axial strain, a, obtained from the CD tests on the soil specimens of Dr = 95 % are shown in Fig. 1. The internal friction angle, ', was obtained as 45 degrees. From initial linear part of each curve, the shear modulus, G, was estimated as G = q/( a- r) where r is radial strain, and is plotted against the confining pressure, p0, in Fig. 2. In Fig. 2, the G ver-sus p0 of Toyoura sand with Dr = 65 % is also indicated. The measured values of G are fitted by the lines given by Eq. (1):

0.5ref ref ( / )G G p p (1)

where pref is a reference value of confining pressure ( =100 kPa) and Gref is the value of G at p = pref. The values of Gref are 29163 kPa and 21086 kPa for soil specimens of Dr = 95 % and 65 %, respectively.

Table 1. Physical properties of Toyoura sand. Property Value Density at 1-g test Density at centrifuge modelling

t

t

1.64 t/m3

1.52 t/m3

Relative density in 1-g test Dr 95 % Relative density in centrifuge modelling Dr 60 % Maximum density dmax 1.65 t/m3

Minimum density Density of soil particle

dmin

s

1.35 t/m3

2.66 t/m3

Mean grain size D50 0.162 mm Internal friction angle ' 45 deg.

266

0 1 2 3 4 5 6 7 8 9 100

200

400

600

800

1000

Axial strain, a (%)Fig. 1. Deviator stress versus axial strain obtained from CD tests of Toyoura sand of Dr = 95 %.

0 100 200 300 4000

10000

20000

30000

40000

50000

60000

Dr=95 %Dr=65 %

Confining pressure, p0 (kN/m2)

Fig. 2. Shear modulus versus confining pressure, together with the fitting line.

It can be seen from Fig. 2 that the shear modulus of Toyoura sand is proportional to the square root of the confining pressure, p0, regardless of the relative density. Hence, the similarity rule at 1-g field proposed by Iai (1989) can be applied to the model tests in this study with the similitude for strain = where is the scale factor (prototype scale/model scale).

3 TEST APPARATUS AND TEST PROCEDURE

3.1 Model piled raft

Figure 3 shows the model rafts having different pile head connection conditions: (a) rigid connection and (b) hinged connection. The square model rafts, with a breadth of 80 mm, was made of aluminum plates with thicknesses of 25 mm for the rigid connection and 40 mm for the hinged connection. The mass of the model raft for the rigid connection was 0.4 kg and that for the hinged connection was 0.94 kg. In order to increase the friction at the raft base, the base was roughened. The interface frictional angle between the raft base and the model ground was 30.5 degrees in the rigid connection model and 22.9 degrees in the hinged connection model, respectively, i.e., the coefficient of frictional angle was 0.59 and 0.42.

Aluminum pipes with an outer diameter of 10 mm, an inner diameter of 8 mm and a length of 170 mm were used for the model piles. Each pile toe was capped with a thin aluminum plate. Young�s modulus, Ep, and Poisson�s ratio, p, were determined from bending tests of the model piles. Each pile was instrumented with foil strain gauges along the pile shaft as shown in Fig. 3 in order to obtain the distributions of the axial forces, the shear forces and the bending moments of the pile. The geometrical and mechanical properties of the model pile are listed in Table 2.

Four model piles were connected to the model rafts with a pile spacing of 40 mm. A universal joint was attached to the head of

each pile for the rigid pile connection piled raft model so that the pile could rotate freely in any direction.

Note that configurations of the model piled rafts used in this study are almost the same as those used in the centrifuge model-ling by Horikoshi et al. (2003a, b). In Table 3, the geometrical and mechanical properties of the model pile used in the centri-fuge modelling by Horikoshi et al. (2003a, b) are also shown.

80

Pi l e 1Pi l e 2

Pi l e 3Pi l e 4

Al umi num pi peOD: 10 mmI D: 8 mm

Al umi num cap

4020 20

Axi al st r ai nShear st r ai n

80

Pi l e 1Pi l e 2

Pi l e 3Pi l e 4

4020 20

Uni t : mm

Al umi num cap

(a) rigid pile head connection (b) hinged pile head connection Fig. 3. Model piled rafts.

3.2 Test set-up and test procedure

Figure 4 shows an illustration of the final stage of the test set-up just before starting a dynamic (seismic) load test. The model foundation was set near the centre location of a laminar box with a special rig before making the model ground. The laminar box with dimensions of 210 mm in width, 560 mm in length and 310 mm in depth was consisted of 16 layers of aluminum frames with a thickness of 20 mm. Dry Toyoura sand was poured in the laminar box and compacted to Dr = 95 % by applying small vibrations using the shaking table. After the completion of the preparation of the model ground, a loading mass (model superstructure) of 22 kg was bolted on the top of the raft.

Accelerometers were embedded in the model ground (Acc. 2, to 4) and attached to the side of the model raft (Acc. 5), and the side and the top of the model superstructure (Acc. 6 to 9). An accelerometer (Acc. 1) was placed on the shaking table to measure the input acceleration.

267

A series of dynamic horizontal load tests were carried out with target amplitude of 100 gal. Sinusoidal input waves of the frequencies from 5 to 95 Hz at an interval of 5 Hz were applied.

Static horizontal load tests were also carried out using the model ground prepared in a rigid acrylic box with dimensions of 500 mm in width, 840 mm in length and 300 mm in depth (Fig. 5).

Acc. 7

Acc. 6Acc. 5

Acc. 4

Acc. 3

Acc. 2Toyour a sand Dr = 95 %

80

Pi l e 1

Pi l e 3Pi l e 4

Pi l e 2

Acc. 1

I nput mot i on: Accel er omet er

560

Acc. 8 Acc. 9

260

: LDT

LDT

Uni t : mm

Pi l e 1Pi l e 2

Fig. 4. Test set-up for dynamic load test using shaking table.

LDT

Pi l e 1Toyour a sand Dr = 95 %

Pi l e 2

DG 1 DG 2Wi nch Load cel l

Wi r e840

Fig. 5. Test set-up for static horizontal load test.

The horizontal load was applied at the level of the gravity cen-tre of the loading mass by pulling the loading mass by means of a winch and a wire at a slow displacement rate less than 1 mm/min. The horizontal displacement of the raft was measured by a laser displacement transducer (LDT), and the vertical displacements of the loading mass were measured at two points by dial gauges (DG) to obtain the inclination of the loading mass.

Note that the test procedure and the model pile rafts used in the 1-g tests in this study are similar to those used in the centri-fuge modelling by Horikoshi et al. (2003a, b). Minor difference is the pile lengths in both tests. The pile length used in the centri-fuge modelling was 180 mm, while that used in the 1-g tests is 170 mm. However, the soil conditions in both tests are largely different as shown in Table 2. In Table 2, the effective vertical stresses, v', the effective mean stresses, p, and the values of the shear modulus at a depth of 170 mm in the model ground (pile tip depth in the 1-g tests) are shown. The corresponding values in the prototype (scale factor = 50) for the 1-g test and the centri-fuge modelling are also indicated. In estimation of the shear modulus, the relation of Eq. (1) was used with p = (1+2K0) v'/3where K0 is the coefficient of earth pressure at rest and was esti-mated from the empirical equation by Jâky:

0 1 sin 'K (2)

If we have an interest in the prototype scale, the stress level and the shear modulus are comparable between the prototypes of the 1-g test and the centrifuge modelling. However, the longitu-dinal rigidity, EpA, and the bending rigidity, EpI, of the prototype pile in the 1-g test are very large, compared with those in the cen-trifuge modelling. Equivalent Young's modulus of the prototype pile as a solid pile is 41.7 GPa in the centrifuge modelling when the compatibility of EpI is considered, indicating that the proto-type pile simulates a concrete pile. On the other hand, equivalent Young's modulus of the prototype pile as a solid pile is 280 GPa in the 1-g test when the compatibility of EpI is considered, indi-cating that the prototype pile in the 1-g test is substantially 'rigid' in bending and deforms as 'short' pile.

4 TEST RESULTS

The results of the dynamic horizontal load tests and the static horizontal load tests listed in Table 3 are presented and compared.

Table 2. Geometrical and mechanical properties of the model piles and the corresponding prototype piles. Model Prototype ( = 50)

1-g field test Centrifuge model-ling at 50-g

1-g field test Centrifuge modelling

Outer diameter, ro (mm) 10 10 500 500 Wall thickness, tw (mm) 1 1 50 50 Length, L (mm) 170 180 8500 9000 Cross section area, A (mm2) 28.3 28.3 70685.8 70685.8 Young's modulus, Ep (GPa) 67.1 71.0 474.5 71.0 Poisson's ratio, p 0.345 0.345 0.345 0.345 Longitudinal rigidity, EpA (GN) 1.90 10-3 2.0 10-3 33.53 5.0 Bending rigidity, EpI (GNm2) 1.94 10-8 1.94 10-8 0.859 0.128 Effective vertical stress (kPa)* Mean effective stress (kPa)* Shear modulus of soil at the pile base, Gb (kPa)*

2.7251.4423502

126.466.8

17232

136.372.1

24763

126.466.8

17232Equivalent Young's modulus as a solid pile (GPa) (compatible with EpI)

279.8 41.7

* at depth of 170 mm in model (at depth of 8.5 m in prototype)

268

Table 3. Test name and test conditions. Test name

Gravity centre

from G.L. (mm)

Pile head condition

Type of loading

Loading mass (kg)

Vertical load

ratio* (%)

DRL 49.3 Rigid Dynamic 22.4 73.4 DHL 63.2 Hinged Dynamic 22.9 72.5 SRL 49.3 Rigid Static 22.4 79.6 SHL 63.2 Hinged Static 22.9 78.5 *Proportion of vertical load carried by piles before load test.

The proportions of the vertical load carried by the piles just after the loading mass was placed on the raft were about 73 % in the dynamic load tests and about 79 % in the static load tests. This difference of the vertical load proportions is thought to be caused by the different soil containers used in the dynamic and the static load tests, laminar and rigid soil containers.

Figure 6 shows the transfer functions of the horizontal accel-erations at the ground surface (Acc. 4), side of the raft (Acc. 5), the gravity center of loading mass (Acc. 6) and the top of loading mass (Acc. 7). The response factor is the ratio of the response ac-celeration to the input acceleration measured at the shaking table (Acc. 1). It can be seen that the natural frequencies of the both model piled rafts were 15 Hz. It can be also seen that the re-sponse factor of the top of the loading mass was larger than that of the raft at input frequencies between 5 Hz to about 30 Hz in both models. On the other hand, the response factor of the raft became larger than that of the top of the loading mass at inputfrequencies larger than about 30 Hz where the response factor of the ground surface became large.

Hereafter, the test results of the both models at the input fre-quency of 15 Hz (0.8 Hz in prototype scale) are compared.

Figure 7 shows the input acceleration waves. Sinusoidal waves with amplitude of about 0.70 m/s2 and a frequency of 15 Hz were applied to the both models.

Figure 8 shows the input acceleration and the response accel-eration at the gravity center of the loading mass for a loading cy-cle indicated by the shaded area in Fig. 7. It can be clearly seen that there was time lag of 0.02 second between the input and the response accelerations in both models. At this input frequency, the response factors were 2.8 and 2.5, respectively, in the rigid and the hinged pile head connection models.

0 20 40 60 80 1000

2

4

6 Top of

superstructure Gravity center

of superstructure Raft Ground surface

Input frequency (Hz) (a) Rigid connection model (DRL)

0 20 40 60 80 1000

2

4

6

Input frequency (Hz)(b) Hinged connection model (DHL) Fig. 6. Transfer function of horizontal acceleration

1 2 3-1.0

-0.5

0.0

0.5

1.0

Time (s)1 2 3

-1.0

-0.5

0.0

0.5

1.0

Time (s)(a) Rigid connection model (DRL) (b) Hinged connection model (DHL) Fig. 7. Input acceleration waves.

2.20 2.22 2.24 2.26

-2

-1

0

1

2

765

4

32

1

Input Response

Time (s)2.50 2.52 2.54 2.56

-2

-1

0

1

2

76

5

432

1

Input Response

Time (s)(a) Rigid connection model (DRL) (b) Hinged connection model (DHL) Fig. 8. Input and response acceleration waves.

Figures 9 and 10 show the time histories of the horizontal load and the pile resistance, respectively. The horizontal load was calculated as the product of the acceleration measured at the gravity center of superstructure and the total mass of superstruc-ture (raft and loading mass). The pile resistance was the total of shear forces at the pile heads of 4 piles. In both models, large amplitude of the horizontal load was observed at the initial stage of dynamic loading. After that, the amplitude decayed, then it be-came almost constant. It can be seen that the horizontal load and the pile resistance of the rigid connection model were slightly larger than those of the hinged connection model. In both mod-els, the pile resistance was smaller than the horizontal load, and this reduction of load was due to the contribution of the raft resis-tance.

Figure 11 shows the time history of the horizontal load pro-portion carried by the piles. A rapid decrease in the pile load was observed during the initial stage of dynamic loading in both models. The horizontal load proportion carried by the piles in DRL concentrated in a range from 50 % to 70 %, and that in DHL concentrated in a range from 50 % to 60 %.

1 2 3-60

-30

0

30

60

Time (s)1 2 3

-60

-30

0

30

60

Time (s)(a) Rigid connection model (DRL) (b) Hinged connection model (DHL) Fig. 9. Time history of horizontal load.

1 2 3-60

-30

0

30

60

Time (s)1 2 3

-60

-30

0

30

60

Time (s)(a) Rigid connection model (DRL) (b) Hinged connection model (DHL) Fig. 10. Time history of pile resistance.

269

1 2 30

20

40

60

80

100

Time (s)1 2 3

0

20

40

60

80

100

Time (s)(a) Rigid connection model (DRL) (b) Hinged connection model (DHL) Fig. 11. Horizontal load proportion carried by piles.

In the centrifuge modelling by Horikoshi et al. (2003b), the raft initially carries more load than the piles, with larger displace-ments the piles more than the raft in the piled raft with rigid pile head connection. In the piled raft with hinged pile head connec-tion, the contribution of the piles is much smaller. Higher propor-tions of the horizontal load carried by the piles in both the rigid and hinged connections in the 1-g tests (Fig. 11) could be attrib-uted to substantially 'rigid' piles used in the 1-g model tests.

Figure 12 shows the time history of the vertical load propor-tion carried by the piles. It can be seen that the vertical load pro-portion carried by the piles was reduced during dynamic loading. However, the proportion of the vertical load carried by the piles almost recovers to the initial value before dynamic loading.

1 2 30

20

40

60

80

100

Time (s)1 2 3

0

20

40

60

80

100

Time (s)

(a) Rigid connection model (DRL) (b) Hinged connection model (DHL) Fig. 12. Time history of vertical load proportion carried by piles.

Figure 13 shows the relationship between the total horizontal resistance including the pile resistance and the raft resistance, and the horizontal displacement of the raft in a loading cycle. The horizontal displacement of the raft measured by the laser dis-placement transducer is the relative displacement between the shaking table and the center height of the raft. It can be seen from both figures that the relationships of the horizontal load and the horizontal displacement of the rigid and the hinged connection models exhibited similar hysteresis loops. In both models, the raft resistance was effectively mobilized even during dynamic load-ing.

On the other hand, in the static load tests, the horizontal stiff-ness of the hinged connection model was larger than that of the rigid connection model.

-0.2 -0.1 0.0 0.1 0.2-60

-30

0

30

60

7

54

2 1

Total (DRL) Pile (DRL) Total (SRL) Pile (SRL)

Horizontal displacement (mm)-0.2 -0.1 0.0 0.1 0.

-60

-30

0

30

60

64

5

21

Total (DHL) Pile (DHL) Total (SHL) Pile (SHL)


(a) Rigid connection model (DRL) (b) Hinged connection model (DHL) Fig. 13. Horizontal resistance-horizontal displacement

relationship.

Figure 14 shows the relationship between the inclination and the horizontal displacement of the raft. Inclination in clockwise direction was taken as positive. The inclination of the raft tended to increase with increasing the horizontal displacement in both models. Comparing both models, the inclination of the raft of DHL was almost equal to that of DRL. This result was different from the results from the static horizontal loading tests, in which the inclination of the rigid connection model (SRL) was larger than that of the hinged connection model (SHL).

-0.2 -0.1 0.0 0.1 0.2-0.0010

-0.0005

0.0000

0.0005

0.0010

3

4

73

2

6

2

6

1

1

DRL DHL SRL SHL

Horizontal displacement (mm)Fig. 14. Inclination of the raft vs Horizontal displacement.

Figure 15 shows the distributions of the axial forces, the shear forces and the bending moments along the pile shaft observed in the cases of DRL and DHL. Positions of pile 1 and pile 2 can be referred to Fig. 4. The distributions at the horizontal load of 40 N are shown for the cases of DRL and DHL for comparison. At this time moment, pile 1 was front pile, while pile 2 was rear pile.

160140120100806040200

-20 0 20 40 60 80 100

HL= 40 N

DRL Pile 1 DRL Pile 2 DHL Pile 1 DHL Pile 2

Axial force (N)(a) Distribution of axial force

160140120100806040200

-5 0 5 10 15 20

HL= 40 N


Shear force (N) (b) Distribution of shear force

160140120100806040200

0.0 0.2 0.4 0.6

HL= 40 N


Bending moment (Nm)(c) Distribution of bending moment Fig. 15. Distributions of axial force, shear force and

bending moment.

270

The raft inclined in clockwise direction at this time moment, so pile 1 was compressed resulting in the increase in the axial forces as shown in Fig. 15(a).

In both models, the axial force, the shear force and the bend-ing moment of the front pile (pile 1) were larger than those of the rear pile (pile 2). It can be seen from Fig. 15(a), (b), (c) that the difference between pile 1 and pile 2 in the case of DHL was smaller than that in the case of DRL.

As for the bending moment, the maximum bending moment was generated at a depth of 80 mm from the pile head in the case of DRL, while that was generated at a depth of 40 mm in the case of DHL. In the case of DRL, it can be inferred from the measured distribution of the bending moments that a large bending moment is also generated at the pile head.

The distributions of the shear forces and the bending moments of the pile shaft obtained from the dynamic tests and the static load tests are compared in Fig. 16 and Fig. 17 , respectively. The distributions at a horizontal load of 40 N are shown. Even in the static load tests, larger shear forces and bending moments are generated in the front piles compared with the rear piles. How-ever, it can be seen that difference of pile behaviour between the front and the rear piles is smaller in the static load tests.

The results in Figs. 16 and 17 suggest that the hinged pile connection model has an advantage for reducing failure of piles especially in dynamic loading.

160140120100806040200

-10 0 10 20

HL= 40 N

DRL Pile 1 DRL Pile 2 SRL Pile 1 SRL Pile 2

Shear force (N)

160140120100806040200

-10 0 10 20

HL= 40 N

DHL Pile 1 DHL Pile 2 SHL Pile 1 SHL Pile 2

Shear force (N)

(a) DRL & SRL (b) DHL & SHL

Fig. 16. Distribution of shear forces.

160140120100806040200

-0.2 0.0 0.2 0.4 0.6

HL= 40 N

DRL Pile 1 DRL Pile 2 SRL Pile 1 SRL Pile 2

Bending moment (Nm)

160140120100806040200

-0.2 0.0 0.2 0.4 0.6

HL= 40 N

DHL Pile 1 DHL Pile 2 SHL Pile 1 SHL Pile 2

Bending moment (Nm)

(a) DRL & SRL (b) DHL & SHL

Fig. 17. Distribution of bending moments.

5 CONCLUSIONS

Principal findings from this study are as follows:

1) The responses of horizontal accelerations of the raft in the rigid and the hinged pile head connection models were al-most equal, although the response in the rigid connection model was a little bit of large.

2) The proportion of the vertical load carried by the piles is reduced during dynamic loading in both piled raft models. However, the proportion of the vertical load carried by the piles almost recovers to the initial value before dynamic loading.

3) As for the proportion of the horizontal load carried by each component, the piles carry more than 60 % of the total horizontal load in both the dynamic and the static load tests.

4) In the dynamic load tests, the relationships of the horizon-tal load and the horizontal displacement of the rigid and the hinged connection models exhibited similar hysteresis loops. On the other hand, in static load tests, the horizontal stiffness of the hinged connection model was larger than that of the rigid connection model.

5) In the dynamic load tests, the relationships of the inclina-tion and the horizontal displacement of the rigid and the hinged connection models exhibited similar loop where the inclination tends to increase with the horizontal displace-ment. On the other hand, in static load tests, the rigid con-nection model exhibited larger inclination compared with the hinged connection model.

6) In the dynamic and the static load tests, much more shear forces and bending moments are generated in the front piles compared with the rear piles. This behaviour is pro-nounced in the rigid pile head connection model and in the dynamic load tests.

The findings 3) to 5) are somewhat different from the find-ings from the centrifuge modelling by Horikoshi et al. (2003a, b). This maybe attributed to the difference of the shear modulus and the stress level of the model grounds used in this study and the centrifuge modelling, i.e. substantially 'rigid' pile in the 1-g tests and flexible pile in the centrifuge modelling. For more un-derstanding of the results from the 1-g test and the centrifuge modelling, analytical study will be required.

REFERENCES

Iai, S. 1989. Similitude for shaking table tests on soil-structure-fluid model in 1g gravitational field, Soils and Foundations29(1): 105-118.

Fukumura, K., Matsumoto, T., Ohno, A. & Hashizume, Y. 2003. Experimental study on behavior of piled raft foundations in sand using shaking table at 1-g gravitational filed. Proceed-ings of the BGA International Conference on Foundations: Innovations, observations, design and practice , Dundee: 307-320.

Horikoshi, K., Matsumoto, T., Hashizume, Y., Watanabe, T. & Fukuyama, H. 2003a. Performance of piled raft foundations subjected to static horizontal loads. International Journal of Physical Modelling in Geomechanics 3(2): 37-50.

Horikoshi, K., Matsumoto, T., Hashizume, Y., & Watanabe, T. 2003b. Performance of piled raft foundations subjected to dy-namic loading. International Journal of Physical Modelling in Geomechanics 3(2): 51-62.

Horikoshi, K. & Randolph, M.F. 1999. Estimation of overall settlement of piled rafts, Soils and Foundations 39(2), 59-68.

Katzenbach, R. & Moormann, C. 2001. Recommendations for the design and construction of piled rafts, Proceedings of the 15th ICSMGE 2: 927-930.

Pastsakorn, K., Hashizume, Y. & Matsumoto, T. 2002. Lateral load tests on model pile groups and piled raft foundations in sand. Proceedings of the International Conference on Physi-cal Modelling in Geotechnics , St. John's, Canada: 709-714.

Poulos, H.G. & Davis, E.H. 1980. Pile Foundation Analysis and Design. New York: John Wiley and Sons.

Randolph, M. F. 1994. Design methods for pile groups and piled rafts, Proceedings of the 13th ICSMFE 1994, New Delhi 2: 61-546.

Yamashita, K., Kakurai, M. & Yamada, T. 1994. Investigation of a piled raft foundation on stiff clay. Proceedings of the 3rd In-ternational Geotechnical Seminar Deep Foundation on Bored and Auger Piles, Belgium 1: 457-464.


271

Modeling of Pile Foundations Retrofitting Strategies AgainstSeismically Induced Lateral Spreading

T. H. AbdounRensselaer Polytechnic Institute, JEC 4049, 110 8th street, Troy, NY 12180, USA. [email protected]

Abstract: Experiences from earthquakes and centrifuge models have shown the great importance of the shallow nonliquefiable soil in increasing the forces and moments imposed on the pile cap and pile foundation subjected to liquefaction-induced lateral spreading. This paper focuses on evaluating retrofitting strategies, with emphasis on the placement of a soft or frangible material near the founda-tion in the shallow nonliquefiable layer. While this shallow soft material reduces the stiffness and the strength of the pile foundation with respect to the superstructure inertia forces, it constitutes an extremely effective way to mitigate the effect of lateral spreading cases in which the resistance to inertia is provided by other foundation elements are one area of application of the proposed retrofitting strategies. Results of three centrifuge tests, Models 2, 2r1 and 2r2, are presented to illustrate the effectiveness of the implemented ret-rofitting method. The experimental results for each of the centrifuge pile models are reviewed and compared. After implementing the proposed retrofitting strategies, a dramatic reduction in the maximum bending moment is observed at the upper boundary of the lique-fied layer (2 m depth). A reduction of up to 35% in the measured maximum bending moment is also observed at the lower boundary of the liquefied layer (8 m depth). These significant reductions in measured pile maximum bending moments in Model 2r, together withthe associated reduction of up to 50% in the measured pile head displacement, demonstrate the effectiveness of the implemented retro-fitting method.

1 INTRODUCTION

Earthquakes are among the major natural disasters. Case history indicates that in most large earthquakes soil liquefaction-related failures have occurred. For instance, the 1971 San Fernando, California earthquake caused more than five hundred million in damage (NRC 1982). In 1976, the Tangshan, China earthquake resulted in collapse and severe damage of many buildings and in the death of several hundred thousand people (NRC 1982). The 1995 HyogoKen Nanbu earthquake in Kobe, Japan, caused more than one hundred billion in total damage. The recent Turkey, Taiwan and Greece 1999 earthquakes also caused tremendous de-struction. In all these earthquakes, much of the damage was re-lated to liquefaction and associated induced lateral spreading.

Evaluation of case history and physical models reveals the significance of several factors influencing the deformation of deep foundations as well as bending moments and cracking of damaged piles. These factors include: free-field permanent lat-eral ground displacement; thicknesses and properties of soil strata penetrated by the piles; and the geometry and properties of the pile foundation. The observed damage and cracking to the piles is often concentrated at the upper and lower boundaries of the liquefied sand layer where there is a sudden change in soil prop-erties (Hamada 1992; Yokoyama et al. 1997; Tokimatsu 1999; Abdoun & Dobry 2002; Abdoun et al. 2002 and Dobry et al.2002) or at the connections between pile and pile cap (e.g., (Hamada, 1992 & Hamada, 2000).

A series of centrifuge tests were conducted earlier to study the great importance of the shallow nonliquefiable soil in increasing the bending response of the pile foundation (Adboun & Dobry, 2002), (Abdoun et al., 2002) and (Dobry et al., 2002). This paper presents a promising rehabilitation approach of existing founda-tions to replace the shallow soil in a trench around piles and pile cap by a soft material that will yield under constant lateral soil

forces (Figs. 1b &1c). This would decrease both bending mo-ments and foundation deformations while allowing the ground lateral spreading to take place without interference from the foundation. As this retrofitting scheme also decreases the lateral resistance of the foundation to inertial loading, a desired material should remain resilient under the transient inertial loading while yielding to static force. In this study, the trench surrounding the foundation will be filled with soft clay. Trenches may be located directly around the foundation (Fig. 1b) or may be located at some distance from the foundation so as to increase the resistance to inertial loading (Fig. 1c).

This paper presents the results of three centrifuge experiments performed to study lateral spreading and its effect on single pile foundations for retrofitted and non-retrofitted piles (Fig. 1). The objectives of this study are accomplished mainly using the centri-fuge experiments and corresponding interpretations and compari-sons. These objectives can be summaries as follows: i) study the three-layer lateral spreading soil response in the free field during earthquake shaking including the magnitude and profile with depth of the maximum induced lateral displacement; and ii) study the soil-pile interaction in three-layer soil system during liquefac-tion and lateral spreading, with and without the implementation of retrofitting strategies.

2 DESCRIPTION OF CENTRIFUGE MODELS

Sketches of RPI�s laminar box (0.25 m (W) 0.46 m (L) 0.25 m (H), in model units) including the instrumentation used for all the centrifuge models are presented in Fig.. 1. The laminar box consists of a stack of up to 39 rectangular rings separated by lin-ear roller bearings, arranged to permit relative movement be-tween rings in the long direction with minimal friction. A relative displacement of up to 0.006 m between adjacent rings is possible, and the design permits an overall shear strain up to

272

Fig. 1. Setups and instrument locations of six centrifuge models; Model 2 (1a), Model 2r1 (1b) and Model 2r2 (1c).

(1a)

(1b)

(1c)

Strain Gage LVDT Accelerometer TransducerPore Pressure

273

20%. The laminar box is optimized to accommodate and accu-rately measure a wide range of cyclic and permanent lateral strains occurring in the soil model. More detailed information on RPI laminar box used in this study is presented by (Van Laak et al. 1994) and www.cee.rpi.edu/centrifuge.

The prototype being simulated in Model 2 (Fig. 1a) involves a single solid pile of diameter 0.60 m, length 10 m and EI = 8000 kN-m2, with a pile cap embedded in the top cemented sand layer. An Aluminum pile cap was rigidly clamped to the to top of the pile model. The pile cap has dimensions of 2 m (W) 2.5 m (L)

0.5 m (H), in prototype units. In model units, the model height is approximately 0.20 m, simulating a 10 m prototype soil deposit and pile length at 50g. The prototype profile includes a bottom layer of 2.0 m slightly cemented sand, topped by a 6.0 m layer of uniform Nevada sand placed at a relative density of about 40%, topped by a 2.0 m layer of the same slightly cemented sand. At 50g, the Dr = 40% fine Nevada sand layer in the model simulates a liquefiable, Dr = 40% coarse Nevada sand layer in the proto-type. The soil profile is fully saturated with water, inclined 2 0 to the horizontal corresponding to 4.8 0 inclination in the field after the instrumental correction (Taboada 1995), and spun at a centri-fuge acceleration of 50g.

The model was excited by 40 cycles of a 100 Hz sinusoidal input parallel to the base of the laminar box, with uniform accel-eration amplitude of about 15g. For the 50g centrifuge accelera-tion of the test, this corresponds respectively, to a frequency of 2 Hz and peak acceleration of 0.3g in prototype units. The horizon-tal accelerations outside the laminar box and in the soil, excess pore water pressures in the liquefiable sand layer, lateral dis-placements of the rings, and bending moments along the pile were measured.

The proposed rehabilitation approach of existing foundations by replacing the shallow soil in a trench around piles and pile cap by a soft material that will yield under constant lateral soil forces (Figs. 1b and 1c). Models 2r1 and 2r2 shown in Figs. 1b and 1c, respectively, represent the two strategies used in retrofitting exist-ing single piles with pile cap embedded in a three layer soil sys-tem. In Model 2r1 (Strategy 1), the slightly cemented sand di-rectly surrounding the pile cap is replaced by 1m width of a slurry wall made by bentonite and water mixture. The depth of the slurry wall is about 1 m. A 1 m width ring of clay wall is di-rectly surrounding the pile from 1m below the slightly cemented sand surface to the boundary of the cemented sand layer and li-quefiable sand layer. In Model 2r2 (Strategy 2), a slurry wall made by bentonite and water mixture was placed around the foundation about 1 m away from the pile cap. The slurry wall has dimensions of 1 m width and 2m depth.

3 SUMMARY OF CENTRIFUGE TESTS FREE FIELD MEASUREMENTS

It is important to analyze the free field experimental results measured away from the piles to verify tests repeatability. A comparison between excess pore pressure profiles, and soil lat-eral displacement profiles for all models presented in Fig. 1, at different times during shaking are presented respectively in Figs. 2 and 3. These free field records are very similar, confirming that the presence of the pile did not affect the free field soil response, and also validating the good repeatability of these centrifuge tests. In all models the lateral displacement occurred essentially within the 6.0 m thickness of the liquefiable sand layer (Fig. 3). The permanent ground surface lateral displacements at the end of

shaking for all models are listed in Table 1. A similar good com-parison, not presented in this paper, was also obtained between the recorded acceleration time histories for the seven centrifuge models (Wang, 2001).

after 5 cycles of shaking

at end of shaking

Excess pore pressure (kPa)

0 20 40 60 80

0

2

4

6

8

10

Nevada sand (Dr=40%)

Slightlycemented sand

ru= 1.0


0 20 40 60 80

0

2

4

6

8

10

Nevada sand (Dr=40%)


ru= 1.0


Model 2

Model 2r2Model 2r1

Fig. 2. Profiles of free field excess pore pressures measured dur-ing and at the end of shaking.

4 COMPARISON OF RESULTS MEASURED IN MODELS 2, 2R1 AND 2R2; PILE RETROFITTING WITH NO INER-TIAL EFFECTS

4.1 Lateral Displacement

Figure 3 shows the recorded lateral displacements of the soil sys-tem at various elevations. The permanent prototype lateral dis-placements of the ground surface after shaking were about 0.70, 0.74 and 0.78 m for Models 2, 2r1 and 2r2, respectively (Table 1).

A comparison of soil and pile lateral displacement profiles of Models 2, 2r1 and 2r2 is presented in Fig. 4. Smaller pile head displacements were measured in Models 2r1 and 2r2 than those measured in Model 2. This indicates that the implemented reme-diation did decrease the pile lateral displacement. Figure 5 pre-sents photos of soil condition around the pile cap after the test of Model 2r1. The pictures indicate that the implemented remedia-tion did decrease the pile lateral displacement as the clay in the upslope of the pile cap was crushed while a gap was left on the downslope side of the cap. The permanent prototype lateral dis-placements of the pile head at the end of shaking were about0.85, 0.42 and 0.60 m for Models 2. 2r1 and 2r2, respectively (Table 1).

4.2 Bending Moments

The effect of the remediation can be best seen in the comparison of the bending moments. Figure 6 shows the bending moment measured along the pile in Models 2, 2r1 and 2r2 plotted at dif-ferent times during shaking as the ground surfaces lateral dis-placement (DH) was increasing. Before remediation, the meas-ured moments at the interface of the top cemented layer and the liquefiable layer is about 200 kN-m in prototype units. After remediation, it drops to almost zero in Model 2r1 and only 20 kN-m in Model 2r2. Before remediation, the maximum bending moment between the bottom layer and the liquefied layer is about 305 kN-m. It drops to about 200 kN-m in Model 2r1 and drops to about 250 kN-m in Model 2r2.

274

Table 1. Summary of single pile foundation characteristics and measurements during centrifuge tests (D r = 40 % in all the tests for the liquefiable layer).

Measurements on piles

Bending moment, Mmax

(kN-m) Model No. No. of soil layers

Pile Cap Mass Retrofitting

strategy Upper boundary

Lower bound-ary

Pile deflec-tion, Dpmax

(m)

Free field ground surface lat-eral displacement

(m)

2

2r1

2r2

3

3

3

Yes

Yes

Yes

No

No

No

No

Strategy 1

Strategy 2

270

0

30

305

220

290

0.85

0.42

0.60

0.70

0.74

0.78

Model 2 Model 2r2Model 2r1

cemented sand

(Dr=40%)Nevada sand

Slightly

T = 5 sec

Displacement (cm)

0 20 40 60 80

0

2

4

6

8

10

cemented sandSlightly

Displacement (cm)


(Dr=40%)Nevada sand


T = 10 sec

0 20 40 60 80

0

2

4

6

8

10

Displacement (cm)


(Dr=40%)Nevada sand


T = 25 sec

0 20 40 60 80

0

2

4

6

8

10

Fig. 3. Lateral displacement profiles at different times during shaking, Models 2, 2r1 and 2r2.

Fig. 4. Pile and soil lateral displacement profiles at the end of shaking, Models 2, 2r1 and 2r2, estimated pile displacements were cal-culated using Beam-on-Winkler springs (BWS).

cemented sand

(Dr=40%)Nevada sand

Slightly

Model 2

Displacement (cm)

0 20 40 60 80 100

0

2

4

6

8

10


Displacement (cm)


(Dr=40%)Nevada sand


Model 2r2

0 20 40 60 80 100

0

2

4

6

8

10

Displacement (cm)


(Dr=40%)Nevada sand


Model 2r1

0 20 40 60 80 100

0

2

4

6

8

10

Measured Soil Disp. Measured Pile Disp. Estimated Pile Disp.

275

DIRECTION OF FREE FIELD LATERAL DISPLACEMENT

Fig. 5. Photos of soil condition around pile cap after the test, Model 2r1.

Fig. 6. Bending Moments Along the Pile in Model 2, Model 2r1 and Model 2r2.

5 CONCLUSIONS

The results from three centrifuge model tests of sand liquefaction with and without retrofitting strategies for the single pile founda-tion are reported in this paper, with detailed data interpretations and discussions. Some major conclusions may be drawn from these results as follows:

1. This work demonstrates that centrifuge modeling of soil-pile interaction during liquefaction is both realistic and useful.

Consistent results were obtained from all centrifuge model tests, which provided detailed information on bending mo-ment response of single pile foundations subjected to lateral spreading with or without retrofitting.

2. As expected, the maximum bending moment was still meas-ured at the lower boundary between liquefied and nonlique-fied soil, while the bending moment at the upper boundaries between liquefied and nonliquefied soil was greatly reduced after retrofitting.

Model 2 Model 2r1 Model 2r2

DH=20cm DH=40cm

-200 0 200

0

2

4

6

8

10-200 0 200

0

2

4

6

8

10

DH=70cmDH=60cm

-200 0 200

0

2

4

6

8

10-200 0 200

0

2

4

6

8

10

Moment (kN-m)

276

3. Test results indicate that the lateral pressure, exerted by the shallow nonliquefied soil layer controls the bending moments developed along the pile.

4. In the absence of inertia force, a significant reduction of the measured maximum bending moment is achieved by the implementation of the two proposed retrofitting strategies aimed at reducing the lateral pressures exerted by the shallow nonliquefied layer.

ACKNOWLEDGEMENTS

The work reported herein was partially supported by the Multid-isciplinary Center for Earthquake Engineering Research (MCEER) and the National Science Foundation (NSF). This sup-port is gratefully acknowledged.

REFERENCES

Abdoun, T. & Dobry, R. 2002. Evaluation of pile foundation re-sponse to lateral spreading. Soil Dynamics and Earthquake Engineering Journal 22 (9-12) 1069-1076.

Abdoun, T., Dobry, R., T. D. O�Rourke & Goh, S.H. 2002. Pile Response to lateral spreads: Centrifuge modeling. Journal of Geotechnical and Geoenvironmental Engineering , ASCE (in print).

Dobry, R., T. Abdoun, T. D. O�Rourke & Goh, S.H. 2002. Piles in lateral spreading: Field bending moment evaluation. Jour-nal of Geotechnical and Geoenvironmental Engineering ,ASCE (in print).

Hamada, M. 1992. Large ground deformations and their effects on lifelines: 1964 Niigata Earthquake, Chapter 3 of Hamada and O�Rourke: 3-1 to 3-123.

Hamada, M. 2000. Performances of foundations against liquefac-tion-induced permanent ground displacement. Proceedings of the 12th World Conf. On Earthquake Engineering , Paper 1754.

NRC 1982. Earthquake Engineering Research - 1982, Overview and recommendations, Report by the Committee on Earth-quake Engineering, National Research Council, National Academy Press. Washington, DC.

Taboada, V. 1995. Centrifuge modeling of earthquake-induced lateral spreading in sand using a laminar box. Ph.D. Thesis, Rensselaer Polytechnic Institute, Troy, NY.

Tokimatsu, K. 1999. Performance of pile foundations in laterally spreading soils. Proceedings of the 2nd International Confer-ence on Earthquake Geotechnical Engineering 3: 957-964. Lisbon, Portugal.

Van Laak, P., Taboada, V., Dobry, R. & Elgamal, A. W., 1994. Earthquake centrifuge modeling using a laminar box. Dynamic Geotechnical Testing Journal , American Society for Testing and Materials, Philadelphia.

Wang, Y., 2001. Evaluation of pile foundation retrofitting against lateral spreading and inertial effects during liquefaction using centrifuge models. MS Thesis, Department of Civil Engineer-ing, Rensselaer Polytechnic Institute, Troy, NY.

Yokoyama, K., Tamura, K. & Matsuo, O. 1997. Design methods of bridge foundations against soil liquefaction and liquefac-tion-induced ground flow. Proceedings of the 2nd Italy-Japan Workshop on Seismic Design and Retrofit of Bridges Rome ,Italy: 109-131.


277

Single Pile Settlement: A Practical Assessment

J. T. Chin JT Geodesign, 36A Jalan BRP ½, Bukit Rahman Putra, 47000 Sg. Buloh, Selangor Darul Ehsan, Malaysia [email protected]

Abstract: Various methods are available for computing vertical settlement of single pile. These available methods vary in degree of complexity. The rigorous methods generally have greater flexibility but are not economical to use for routine initial design assessments. Therefore, the need for a practical and quick method of assessment of the single pile settlement is desired. This paper presents elastic settlement design charts for axially-loaded single vertical piles embedded in a two-layer soil continuum with the homogeneous profile as a special case. It can be shown that these design charts can be readily programmed into an efficient spreadsheet formulation. This spreadsheet can then be easily and routinely used for computing elastic pile settlements at design working loads without recourse to the computer program, resulting in savings in time and cost. The solutions obtained using these design charts approach are shown to com-pare favorably to some field measured static axial pile load test results.

1 INTRODUCTION

Pile foundation is one of the most common foundation systems used for the support of many buildings, bridges, towers and other structures. The piles may be used as individual single piles or as individual pile groups. From a practical viewpoint, the ability to estimate quickly the single pile performance for given pile sizes and pile lengths during the initial design stage is an advantage. Commercial and proprietary computer programs are available for the analysis of the pile deformation response (e.g. PIGLET, DEFPIG, PGROUP). These methods of analysis vary in degree of complexity. Elastic design charts for specific class of problems have also been reported by Butterfield & Douglas (1981) and Poulos & Davis (1980). The structural-based methods, which ig-nore or simplify the soil continuum, generally have more restric-tions and limitations in its application to practical problems. On the other hand, the more rigorous boundary and finite element methods, which considers the soil continuum, have greater flexi-bility and modeling capabilities (e.g. Butterfield & Banerjee 1971; Ottaviani 1975). The rigorous methods are however gener-ally not economical to use for routine initial design assessments. Therefore, the need for a practical and quick method of assess-ment of the single pile deformation response is desired.

In this paper, elastic design charts for axial pile settlement re-sponse derived from the elastic continuum simplified boundary element method for piles embedded in a two-layer soil continuum are presented. The homogeneous soil profile represents a special case of the design charts. The presented elastic design charts cover a wide practical range of normalized pile and soil parame-ters. For practical purposes, elastic analysis is generally applica-ble for assessments at service working load conditions where as-sumption of elastic pile-soil response is acceptable.

It can be shown that the present elastic design charts can be readily programmed into an efficient single page spreadsheet formulation. This spreadsheet can then be easily and routinely used for computing elastic single pile settlements at design work-ing loads without recourse to the computer program, resulting in time and cost savings. Finally, the solutions obtained using the present elastic design charts spreadsheet formulation are com-pared to two reported field measured static single pile axial load test results.


The simplified boundary element method analysis procedure of Chow et al. (1990) for piles embedded in a two-layer soil contin-uum is used for assessment of the elastic pile-soil interaction problem. Full details of the analysis procedures can be obtained from that reference. However, for clarity brief details are de-scribed herein. Figure 1 shows a schematic of the single pile problem installed through an upper soil layer and socketed into a lower stiffer bearing layer.

Fig. 1. Single pile problem embedded in two-layer soil profile.

The governing relationship for the pile-soil interaction prob-lem can be shown to be given by Eq. (1).

([Kp] + [Ks]) {wp} = {P} (1)

where [Kp] and [Ks] are the assembled pile and soil stiffness ma-trices respectively, {wp} and {P} are the vectors of pile nodal vertical deformations and applied axial compression load respec-tively. The soil stiffness matrix [Ks] is obtained using the elastic solutions of Chan et al. (1974) for a two-layer soil continuum. Equation (1) can be solved to obtain the pile head settlement for

h

e

d

P

Upper soil layer

Lower soil layer

(E1 , 1)

(E2 , 2)

278

a given axial compression load at the pile head. The analysis pro-cedure has been coded into a fortran computer program �SPILE�.

3 ELASTIC DESIGN CHARTS

A series of elastic single pile settlement design charts has been generated using the computer program SPILE for different values of the pile-soil stiffness ratio (Ep/E1), soil stiffness ratio (E2/E1),normalized upper soil layer thickness (h/d) and the normalized pile socket length in the lower bearing layer (e/d). These charts cover the following practical ranges of pile and soil parameters:

h/d = 5 to 200 e/d = 0 to 40 Ep/E1 = 100 to 10000 E2/E1 = 1 to 200

For clarity of presentation, the sample elastic single pile design charts are collectively shown in Figs. 2 to 4 for pile-soil stiffness ratios of Ep/E1 = 100, 1000 and 10000 respectively. The present sets of design charts are more detailed than those presented ear-lier by Chow et al. (1990). It can be shown that these sets of de-sign charts can be readily programmed into an efficient spread-sheet formulation. For intermediate pile and soil parameter values than those shown in Figs. 2 to 4, linear and logarithmic (to base 10) interpolations for relevant �h/d� and �e/d� values, and �Ep/E1� and �E2/E1� values can be used respectively. It is obvious that increased accuracy would be obtained for more intermediate values of Ep/E1 design charts. However, for initial design assess-ment purposes the presented sets of design charts are sufficient for most practical situations. The single page spreadsheet can then be used to (a) obtain a quick assessment of the elastic pile settlements at service working loads for different pile-soil prob-lems without the need to run the computer program SPILE; (b) back-figure relevant soil deformation parameters by �matching� the field measured pile load-settlement response.

The accuracy of the spreadsheet approach using the presented design charts is compared to the results obtained using the com-puter program SPILE, as shown in Figs. 5 and 6. Good agree-ment is obtained between the estimated values using the spread-sheet and the computed values using the program SPILE.

4 COMPARISON WITH SINGLE PILE TEST RESULTS

4.1 Bored Pile in Singapore Old Alluvium Formation

Chin (1996) reported the static load test results of an instru-mented 800mm diameter bored pile installed in Singapore Old Alluvium formation. The design working load of the pile is 4000 kN. The bored pile was installed through about 10m of loose to medium dense clayey silty sand, 4m of dense silty sand and a fi-nal 10m of very dense silty sand. Details of the pile instrumenta-tion, test loading procedures and site soil conditions may be ob-tained from that reference.

For the present elastic design charts approach using the pro-grammed spreadsheet, the following relevant pile and representa-tive two-layer soil parameters adopted by Chin (1996) are used.

Pile Parameters: Pile length, L (=h+e) = 24.0 m Pile diameter, d = 800 mm Modulus of pile, Ep = 28000 MPa

Soil Parameters: Thickness of upper soil layer, h = 14.0 m Upper soil modulus, E 1 = 50 MPa Lower soil modulus, E2 = 450 MPa

Using Spreadsheet (Design Charts): with Ep/E1 = 560, E2/E1 = 9, h/d = 17.5, e/d = 12.5, the estimated pile stiffness, K [= P/(w*E1*d)] = 24.8, and computed pile settlement = 4.04mm at service working load of 4000 kN.

Figure 7 shows the comparison between the spreadsheet com-puted pile settlement value at the service working load of 4000 kN, the elastic results obtained from program SPILE and the field measured load-settlement data. As shown, the computed spread-sheet results compare favorably to the SPILE results and the field measurements. For practical purposes, the programmed spread-sheet using the presented elastic design charts is therefore capa-ble of providing a quick assessment of the elastic bored pile set-tlement at service working load condition.

4.2 Driven Steel Pipe Pile in Stiff Over-Consolidated Clays

O�Neill et al. (1982) reported the axial load test results of a 274mm external diameter (wall thickness = 9.3mm) steel pipe pile driven 13.1m into stiff over-consolidated clays. The adopted service working load of the pile is 200 kN for the present case.

Chow (1986) has analyzed this reported field pile test results, and the relevant reported parameters in that reference are adopted for the present assessment. The adopted parameters for the pre-sent two-layer soil continuum model are tabulated below.

Pile Parameters: Pile length, L (=h+e) = 13.1 m Pile diameter, d = 274 mm (external diameter) Wall thickness, t = 9.3 mm Modulus of pile, Ep = 205000 MPa

Soil Parameters: Thickness of upper soil layer, h = 13.1 m Upper soil modulus, E 1 = 298.5 MPa Lower soil modulus, E2 = 465.9 MPa

Using Spreadsheet (Design Charts): with Ep/E1 = 90.1, E2/E1 = 1.56, h/d = 47.8, e/d = 0.0, the estimated pile stiffness, K [= P/(w*E1*d)] = 6.61, and computed pile settlement = 0.37mm at service working load of 200 kN.

Figure 8 shows the comparison between the spreadsheet com-puted pile settlement value at the service working load of 200 kN, the elastic and non-elastic results obtained from program SPILE and the field measured load-settlement data. As shown, the computed spreadsheet results compare favorably to the SPILE elastic results. The present field test results show a larger magni-tude of the measured pile head settlement indicating some pile-soil interface yielding has occurred at the applied axial load of 200 kN. A non-elastic analysis using computer program �SPILE� shows good agreement between the computed and measured re-sults up to the applied axial load of 200 kN. For practical pur-poses, the programmed spreadsheet using the presented elastic design charts is shown to be capable of providing a quick and ac-ceptable assessment of the elastic steel pipe pile settlement at service working load condition.

279

(a) (c)

(b) (d)

Fig. 2. Variations of pile stiffness for Ep/E1=100 for (a) E2/E1=1, (b) E2/E1=10, (c) E2/E1=20 and (d) E2/E1=50 & 100.

Pile Stiffness for Ep/E1 = 100 (E2/E1 = 1)

4.0

4.5

5.0

5.5

6.0

6.5

7.0

0 20 40 60 80 100h/d

e/d = 0

20,40

10 5

1

2


6

8

10

12

14

16

0 20 40 60 80 100h/d

e/d = 0

1,2,5,10,20,40


6

7

8

9

10

11

12

13

0 20 40 60 80 100h/d

e/d = 0

1

2,5,10,20,40

Pile Stiffness for Ep/E1 = 100 (E2/E1 = 50 & 100)

6

8

10

12

14

16

0 20 40 60 80 100h/d

e/d = 0,1,2,5,10,20,40

280

(a) (d)

(b) (e)

(c) (f)

Fig. 3. Variations of pile stiffness for E p/E1=1000 for (a) E2/E1=1, (b) E2/E1=10, (c) E2/E1=20, (d) E2/E1=50, (e) E2/E1=100 and E2/E1=200.


0

4

8

12

16

20

0 40 80 120 160 200h/d

e/d = 40

2010

5

2,1,0


0

20

40

60

80

100

0 20 40 60 80 100h/d

e/d = 0

1

5,10,20,40 2


0

10

20

30

40

50

60

0 40 80 120 160 200h/d

e/d = 0 1

25

10,20,40


0

20

40

60

80

100

120

0 20 40 60 80 100h/d

e/d = 0

1 5,10,20,40


0

10

20

30

40

50

60

70

0 20 40 60 80 100h/d

e/d = 0

12

5 10 10,20,40


0

20

40

60

80

100

120

0 20 40 60 80 100h/d

e/d = 0

1,2,5,10,20,40

281

(a) (d)

(b) (e)

(c) (f)

Fig. 4. Variations of pile stiffness for E p/E1=10000 for (a) E2/E1=1, (b) E2/E1=10, (c) E2/E1=20, (d) E2/E1=50, (e) E2/E1=100 and E2/E1=200.


0

10

20

30

40

50

60

0 40 80 120 160 200h/d

e/d = 40

2010

5,2,1,0


0

80

160

240

320

400

0 40 80 120 160 200h/d

e/d = 01

5

20,40

2

10


020406080

100120140160

0 40 80 120 160 200h/d

e/d = 01

25

10

20

40


0

80

160

240

320

400

480

0 40 80 120 160 200h/d

e/d = 0

1

2

10,20,40

5


0

40

80

120

160

200

240

0 40 80 120 160 200h/d

e/d = 0

125

10

20

40


0

100

200

300

400

500

600

700

0 40 80 120 160 200h/d

e/d = 0

1

5,10,20,40

2

282

Fig. 5. Comparison between results from computer program SPILE and spreadsheet using design charts (for h/d=35, e/d=4).

Fig. 6. Comparison between results from computer program

SPILE and spreadsheet using design charts (for E 2/E1=6, e/d=6).

Fig. 8. Computed and measured steel pipe pile static axial re-sponse.

5 CONCLUSIONS

The present paper has provided single pile elastic settlement design charts for an axially loaded vertical pile embedded in a two-layer soil profile, with the homogeneous profile as a special case. It has been shown that these design charts can be readily programmed into an efficient spreadsheet formulation which can be used to obtain a quick estimate of the single pile settlement with-out recourse to the computer program. Comparisons with two re-ported field static axial pile test results show favorable agreement between the computed and measured responses at service working load condition. The proposed spreadsheet approach presents a practical, economical and quick method for routine initial de-sign assessments of the single pile settlements at service working loads.

REFERENCES

Butterfield, R. & Banerjee, P.K. 1971. Geotechnique 21: 43-60. Butterfield, R. & Douglas, R.A. 1981. Technical Note 108.

CIRIA, London. Chan, K.S. et al. 1974. International Journal of Solids and

Structures 10: 1179-1199. Chin, J.T. 1996. Proceedings of the 12th Southeast Asian Geo-

technical Conference 1: 441-446.Chow, Y.K. 1986. International Journal for Numerical and Ana-

lytical Methods in Geomechanics 10: 59-72. Chow, Y.K. et al. 1990. Journal of Geotechnical Engineering

Division, ASCE 116: 1171-1184. O�Neill et al. 1982. Journal of Geotechnical Engineering Divi-

sion, ASCE. 108: 1605-1623. Ottaviani, M. 1975. Geotechnique 25: 159-174. Poulos, H.G. & Davis, E.H. 1980. Pile Foundation Analysis and

Design. New York, U.S.A.: John Wiley & Sons Publisher. .

Fig. 7. Computed and measured bored-pile static axial response.

0

40

80

120

160

200

100 1000 10000Ep/E1

E2/E1 = 15 (program) E2/E1 = 15 (charts)E2/E1 = 60 (program) E2/E1 = 60 (charts)

0

20

40

60

80

100

0 40 80 120 160 200h/d

Ep/E1 = 2500 (program)Ep/E1 = 2500 (charts)Ep/E1 = 8500 (program)Ep/E1 = 8500 (charts)

0

200

400

600

800

1000

0 2 4 6 8 10 12 14Settlement (mm)

computed (using program)

measured

computed (design chart)

0

2

4

6

8

100 200 400 600 800

Load (kN)

MeasuredComputed (non-elastic)Computed (Elastic)Design Charts (Elastic)

single pile test


283

A Comparative Study on Results of Static Pile Load Test at Rock Socketed Drilled Shaft and Bearing Capacity Equations

B. S. Chun & S. S. Whang Department of Civil Engineering, Hanyang University #17, Haeng-Dang Dong, Seong-Dong Ku, Seoul,Korea, [email protected]

Abstract: The driven pile has environmental problems such as vibration and noise. Especially, if the site consists of gravel, cobble and weathered rock, the driven pile cannot be applied. Therefore, the application of the drilled shafts is increasing in Korea. However, the bearing capacity values which were estimated by the suggested theoretical formulas are generally considered too conservative. In this paper, static load tests for the rock socketed drilled shaft were done. To study the load transfer mechanism, strain gauges were used. The bearing capacity values by the theoretical formula are compared. Even the static load tests did not reached the ultimate bearing capacity nor by end bearings but rather by side friction resistances. Based on the above results, several suggestions are proposed for the drilled shaft design.

1 GUIDELINES

The pile which has been used in this country is divided into two types: driven pile and drilled shaft. The study of drilled shaft has been widely made abroad since 1960 because of non-vibration, anti-noise and good workability for the ground composed of gravel, boulder layer or weathered rock. Though the use of drilled shaft has increased in the country, there is much to be desired in the study of drilled shaft.

Therefore, design criteria and bearing capacity formulas of foreign country such as America and Canada are used to design drilled shaft in Korea.

However, it is known that because the design criteria and bearing capacity formulas are for specified foreign ground and solid rock, there have been a problem with applying for domestic ground and rock and are underestimated in comparison with ultimate bearing capacity of actual ground (Bae et al, 1996; Kyungsung University, 2000). Therefore, to use foreign design criteria in domestic fields, the applicability has to be examined in comparison with static load test data of drilled shaft. Moreover, it is necessary to establish the design criteria for the design of domestic drilled shaft. In this paper, static load test data for the rock socketed drilled shaft at Gwangan Highway and Suyoung 3rd Bridge are analyzed. The bearing capacity from field test data and theoretical formulas are compared and analyzed. Also, this study seeks about to suggest the ultimate bearing capacity formulas which are suitable to the domestic field and make it possible to design economically.

2 TEST RESULTS AND ANALYSIS

2.1 Field Data

In this paper, the static pile load test data, which were carried out in Gwangan Highway 5th construction field of Suyoung Bay filled ground and Suyoung 3rd Bridge construction site of Pusan Centum penetration road, are used to compare with the existing proposed bearing capacity formula. Tables 1, 2, and 3 show the specifications of drilled shaft and the geotechnical physical property value of ground (Kyungsung Univ. 2000; Baek, 2002). In case of Suyoung 3rd Bridge, because the data analysis results were still incomplete, except for N value, geotechnical physical property values (unit weight, unconfined compressive strength, undrained strength, internal friction angle) of other soil part are presumed to be the upper and the lower limit value using correlation with N value that is proposed by Bowles (1977). Unconfined compressive strength of weathered rock layer used the proposed value about the weakest soft rock among classification categories that was provided from ISRM (International Society for Rock Mechanics).

Rock, differently from soil part, has much difference in the value of physical properties such as internal friction angle according to the type of rock and the weathering level. Therefore, without physical property test about the field rock, the use of data in practical design must be avoided.

Table 1. Pile specifications.

Gwangan Highway 5th construction field Suyoung 3rd Bridge

The embedded length of pile 32m 30m The type of pile The drilled shaft of weathered rock The drilled shaft of weathered rock

The diameter of pile 1m 1.5m maximum static pile load 1500 ton 2050 ton

284

Table 2. The value of geotechnical properties applied for bearing capacity formula (Gwangan Highway 5th construction field).

Depth(m) Layer Saturated unit weight t (t/m2)

Unconfined compressive

strength qu(t/m2)

Internal frictionangle

Undrained strength Cu(t/m2)

0.0 9.1 alluvium 1.85 3.974 32 1.997 9.1 11.5 weathered soil 1.89 7.948 35 3.994 11.5 32 weathered rock 2.26 99.862 36 49.9311

Table 3. The value of geotechnical physical properties applied for bearing capacity formulas (Suyoung 3 rd Bridge).

Depth(m) Layer NavgSaturated unit weight t(t/m2)

Unconfined compressive strength

qu(t/m2)

Undrained strength Cu(t/m2)

Internal friction angle (°)

0.0-5.2 clay, gravel 7/30 1.76- 4.98-9.78 2.45-4.89 - 5.2-10.7 sand 7/30 1.44-1.85 - - 27-32

10.7-15.5 clay 4/30 1.61-1.92 2.45-4.89 1.23-2.45 - 15.5-16.8 sand gravel 25/30 1.76-2.08 - - 30-35

16.8-24.5 argillaceous weathered soil 35/30 1.92-2.24 39.13- 19.57 -

24.5-30.0 weathered rock 50/4 2.00-2.50 100-500 - 35-40

2.2. The Evaluation of Ultimate Bearing Capacity Using Existing Formulas

Ultimate bearing capacity was evaluated by the existing design standard and the suggested bearing capacity formulas to compare with the surveying value of static pile load test to use a ground property value of Tables 2 to 5.

The used bearing capacity formulas were proposed by Reese & O�Neil (1988), Das (1984), Bowles (1988) etc., and comply with Canadian Geotechnical Society (1992) and FHWA (1999), an existing design standard of Highway Corporation and a reformed design standard of Highway Corporation in this year.

In case of Suyoung 3rd Bridge, when there were no exact data as calculating ultimate bearing capacity according to the upper and the lower limit value of properties which are calculated from N value of soil department and a general statement of rock, it had

been compared with the ultimate bearing capacity which varies as the designer�s selection of property value.

As the result, there was at most the treble difference of total ultimate bearing capacity as design methods and Uniaxial Compressive Strength of rock masses was potent influence

Therefore, in case of drilled shaft socketed to rock masses, in order to reduce the difference of bearing capacity by designer �s selection, not only the properties of sand but also accurate properties of rock masses (unit weight, uniaxial compressive strength, internal friction angle) are very important. Because of few data about rock joints, end bearing capacity using design criteria of FHWA (1999) was calculated on the most critical assumption that joint spacing is 305mm and joint aperture was 5mm long.

Table 4. Ultimate bearing capacities through the theory formulas of bearing capacity (Gwangan Highway 5 th construction site).

Gwangan Highway 5th

construction field

Reese&

O�Neil(1988)

Das(1984)

Bowles (1988)

Canadian Geotechni-cal Society

(1992)

Highway bridge design standard of established

road construc-tion

(2000)

Highway bridge design

standard of reformed

load construction

(2002)

FHWA (1999)

The measured value of

static pile load test

0.0~9.1m 3.78 1.14 1.21 1.55 3.78 3.78 3.78 24.5 9.1~11.5m 6.29 2.63 3.08 3.52 6.29 6.29 6.29 29.2

The static pile load test

which is followed by depth (t/m2)

11.5~32m (rock mass) 8.13 4.16 8.28 19.9 1.01 19.9 15 8.7

Skin friction (ton) 678 320 591 1353 220 1009 1121 1480 End bearing capacity unit

(t/m2) 430 913.9 200 90 130 430 40.8 25.5

End bearing capacity (ton) 338 718 157 71 338 338 32 20 Gross ultimate bearing ca-

pacity (ton) 1017 1038 748 1424 558 1347 1153 1500

* The surveying value of static pile load test is not ultimate bearing capacity but the distribution load of skin friction and end bear-ing capacity during the maximum loading condition

** : the ratio of settlement to a pile-top diameter

285

Table 5. Ultimate bearing capacities calculated by theoretical equation.

Reese & O'Neil (1988) Das (1984) Bowels

(1988)

Canadian Geo-technical Soci-

ety (1992)

Highway bridge design standard of es-tablished road construction

(2000)

Highway bridge design standard of re-formed load construction

(2002)

FHWA (1999)

Suyoung 3rd Bidge

Upper limit

Lowerlimit

Upperlimit

Lower limit

Upperlimit

Lowerlimit

Upperlimit

Lowerlimit

Upperlimit

Lowerlimit

Upper limit

Lower limit

Upper limit

Lower limit

The real

value ofloading

test

0.0~5.2m 2.69 1.35 1.96 0.98 2.45 1.21 2.45 1.23 2.69 1.35 2.69 1.35 2.69 1.35 - 5.2~10.7m 6.43 4.18 2.34 1.44 2.48 1.31 3.18 2.06 6.43 4.18 6.43 4.18 3.01 1.95 - 10.7~15.5m 1.35 0.68 0.98 0.49 1.23 0.62 1.23 0.62 1.35 0.68 1.35 0.68 1.35 0.68 - 15.5~16.8m 7.94 5.04 4.60 2.83 5.40 2.83 6.16 3.92 7.94 5.04 7.94 5.04 12.20 7.76 - 16.8~24.5m 10.76 10.76 7.83 7.83 9.79 9.79 5.22 3.46 10.76 10.76 10.76 10.76 10.76 10.76 -

Mean unit skin fric-

tionper depth

(t/ ) 24.5~30.0m (base rock) 6.58 4.45 6.95 4.64 12.49 7.04 44.55 19.90 2.64 1.01 44.60 19.90 46.87 15.00 -

Skin friction(ton) 854 684 609 487 847 624 1535 754 751 595 1453 913 1835 916 1904 Unit end bearing capacity

of pile(t/ ) 430 430 2526.6 869.7 133.3 133.3 370.5 74.1 364.2 364.2 364.2 364.2 197.5 39.5 82.6

End bearing capacity of pile (ton) 760 760 4465 1537 236 236 655 131 644 644 644 644 349 70 146

Total ultimate bearing capacity(ton) 1614 1444 5074 2024 1083 860 2190 885 1395 1239 2097 1557 2184 986 2050

*The surveying value of static pile load test is not ultimate bearing capacity but the distribution load of skin friction and end bearing capacity during the maximum loading condition.

**Mean unit skin friction is not obtained because of lack of data. *** : the ratio of settlement to a pile-top diameter.

This method is suggested in the case of such rock as more than 305mm joint spacing, less than 5mm joint aperture, and more than 305mm foundation width. Therefore, end bearing capacity which is calculated with the formula of FHWA (1999) was too small to be less than those which are calculated with other formulas. This means that special investigation for joints of rock masses should be carried out to use the bearing capacity formula of FHWA in design.

2.3 The Comparison Between Static Load Test and Theoretical Formulas

Though general load test is applied only to measure the strain quantity on the pile tip, the static load test, carried out in Busan Gwangan Highway 5th construction field and Suyoung 3 rd Bridge, made it possible to measure not only load-strain on the pile tip but also skin friction and end bearing capacity independently by axial transfer analysis according to depth using auto-monitoring system.

During the loading test, the test pile in Gwangan Highway 5 th

construction field and Suyoung 3 rd Bridge was given maximum load (1500ton, 2050ton), but failure of pile did not occur and load-displacement curve showed that the behavior of test piles was elastic. According to regular Load Transfer Curve between end bearing capacity and settlement suggested by FHWA (1999), the skin friction and end bearing capacity of sand socketed drilled shaft have maximum value when the settlement is over 1% and over 5% of end diameter.

As the results of the Load Test, the settlements of pile tip were each 11mm and 35mm at Gwangan Highway 5th construction field and Suyoung 3 rd Bridge and is actually 0mm and 28mm

without elastic strain. These values are each 0% and 1.87% of end diameter.

The rock layers that test pile penetrated is completely weathered to sand (silty sand, gravel). In comparison with FHWA, it can be shown that the static load is much smaller than ultimate bearing capacity. Thus, the skin friction and the end bearing capacity, calculated by the axial load transfer of static loading test, are distributed load measured around pile and on the tip under maximum condition.

Nevertheless, the ultimate bearing capacity calculated by design equation is almost equal to distribution load of static pile load test. Also, in the case of actual design, the safety factor of ultimate bearing capacity is generally 3 for general bearing capacity formula and 2 for static load test. Therefore, the difference of ultimate bearing capacity is more remarkable. Accordingly, if existing bearing capacity formula is used in the actual design of piles, the construction design will be conservative and uneconomical.

3 CONCLUSIONS

From the result of comparative study on bearing capacity equation and static pile load test at Gwangan Highway 5 th

Construction Field and Suyoung 3 rd Bridge of Test Pile, the following conclusion can be obtained:

1) First, the actual measured value obtained by static pile load test is not ultimate bearing capacity but skin friction of maximum load and distribution load of toe resistance. Therefore, the ultimate bearing capacity of actual ground should be even larger than maximum load. Nevertheless, the ultimate bearing capacity calculated by capacity equation is small or nearly equal as

286

compared with the load by static loading test. Moreover, when the factor of safety of 3 and 2 is applied in the design, this difference becomes remarkable. This means that the method by current capacity equations have a problem with calculating ultimate bearing capacity. In comparison with the surveyed value of static loading test used in this study, it has a problem with the safety of piles that the value of safety of 2 can be applied to bearing capacity obtained by capacity equations

2) In case of frictional resistance, there is no value of Bearing Capacity Formulas that exceeded the measured value by the Static Load Test. It means that Bearing Capacity Formulas underestimated the frictional resistance of Drilled Shaft very much. Among theoretical bearing values, the calculated values of frictional resistance by Canadian Geotechnical Society (1992) and FHWA (1999) were the most approximate to the measured value of frictional resistance by the Static Load Test. Therefore, it is desired to put a calculation equation of frictional resistance by Canadian Geotechnical Society (1992) and FHWA (1999) among proposed equations compared in this paper in order to calculate frictional resistance of drilled shaft penetrating domestic weathered rock without field load test.

3) In case of FHWA (1999), the frictional resistance is the most approximate measured value by Static Load Test but end bearing capacity of pile is calculated were very small. The data for rock joint assumed most critical condition because there were no data of rock joint for calculation of end bearing capacity of a pile, and, henceforth, rock joint survey must be performed at ground survey in other to use the theory equation of FWHA for design

4) In case of performing loading test not for verifying design load but for design, to verify ultimate bearing capacity, it is essential to control the load till failure.

5)In case of rock socketed drilled shaft, ground survey for pre-design must measure the properties for rock as well as soil because rock properties have much influence on ultimate bearing capacity.

6) Therefore, until suitable design method for domestic ground is found in the future, Load Transfer Analysis through axial load measuring system has to be used with loading test.

REFERENCES

Baek K.G.C. 2002. Suyoung 3rd Bridge for the bearing capacity comparison of drilled shaft Report No. PTR-02-115.

Bae, S.W., Hwang, S.I. & Cho, N.J. 1996. Prediction of load settlement curves for drilled shafts, Korean Geotechnical Society: 327-336.

Bowles, J.E. 1977. Foundation Analysis and Design . New York:Graw-Hill Book Company.

Bowles, J.E. 1996. Foundation Analysis and Design : New York:Graw-Hill Book Company.

Canadian Geotechnical Society 1992. Canadian Foundation Engineering Manual, Canadian Geotechnical Society Technical Committee on Foundations, Ottawa.

Choi, Y.K. 1989. An experimental study on the sealing ability of base-grouted pile, Dissertation, Seoul National.

Das, B.M. 1999. Principles of Foundation Engineering, Book/Cole.

Korean Geotechnical Society 1997. Ground Exploration. Geotechnical Engineering Series, Gu Mi Book Publisher.

Korean Society of Civil Engineers 2001. Design and Performance of Road Pavements Explanation.

Kyungsung University 2000. The review on the Gwangan bridge construction work is during static pile load test for a large diameter socketed pipe pile and numerical analysis. Report No. KSU/GT-00-1.

Peck, R.B., Hanson, W.E., & Thornburn, T.H. 1974. FoundationEngineering, John Wiley and Sons: 361-374.

Reese, L.C. & O'Neill, M.W. 1988. Drilled Shafts Student Workbook. NHI Course No. 13214, Federal Highway Administration, August.

U.S. Department of Transportation, Federal Highway Administration 1999. Drilled Shafts: Construction Publication No. FHWA-IF-99_025

Vesic, A. S. 1977. Design of pile foundation, NCHRP synthesis of highway practice No. 42 , Transportation Research Board, National Research Council, Washington, D.C.


287

The Increase of Passive Earth Pressure due to the Dowel Effect of Foundation Piles

R. Katzenbach, G. Bachmann, C. Gutberlet & J. Turek Institute and Laboratory of Geotechnics, Technische Universität Darmstadt, Petersenstrasse 13, D-64287 Darmstadt, Germany [email protected]

Abstract: The construction of a CPRF is usually combined with a deep excavation. Geotechnical measurements on retaining structureswithin different high-rise building projects in Frankfurt am Main (Katzenbach et al. 1999) have shown that the observed wall dis-placements are much less than predicted which is partially caused by the dowel effect of the foundation piles nailing the earth wedge in front of the wall. For quantifying the increase of the passive earth pressure due to this effect a �numerical test series� � built up on a small scale model test series � was carried out with more than 500 single finite element simulations in which geometrical dimen sions like the distance between the first pile row and the retaining wall and material parameters were varied within the investigation to gain knowledge about their influence on the earth pressure. The results of the numerical investigations show an increase of earth pressure leading to an effective reduction of the embedment depth of the walls and thus to a cost-optimization for excavation pits.

1 INTRODUCTION

During the construction process of high-rise buildings founded on a CPRF or a pile foundation the piles used for the foundation are constructed from a pre-excavation level above the groundwa-ter table before the remaining part of the pit is excavated. The in-teraction of the foundation piles, the soil continuum and the re-taining wall during the excavation process � which leads to a strengthening of the earth wedge in front of the wall � is worth-while to be investigated, for example the impact of the piles on the displacement dependent earth pressure in front of the wall. The impact of lateral ground movements on piles has been de-scribed by Heyman (1965) who reported about the influence of lateral earth pressure on pile foundations and Chen & Poulos (1997) who developed linear elastic solutions for simple soil movement profiles to enable approximate assessment of the pile head deflection and the maximum bending moment. Furthermore, Nalcakan & Ergun (2001) report about model tests on laterally loaded passive piles.

Í³¿´´ ½¿´» ³±¼»´ ¬»¬ »®·»

Ò«³»®·½¿´ ¬»¬ »®·»

Ê¿´·¼¿¬·±² ±º ¬¸» ÚÛó³±¼»´

Ð¿®¿³»¬»® ¬«¼§

Ï«¿²¬·º·½¿¬·±² ±º ¬¸» ¼±©»´ »ºº»½¬

Í¬»° ï

Í¬»° î

Í¬»° í

Í¬»° ì

Í¬»° ë

Fig. 1. Sequence of the investigation.

This paper deals with small scale model tests combined with numerical parameter studies with the aim to identify the parame-ters which have influence on the magnitude of the earth pressure and to develop a method for the consideration of the dowel effect of foundation piles for the design of retaining walls. The se-quence of the investigation is shown in Fig. 1.

2 SMALL SCALE MODEL TEST SERIES

The 1g small scale model test series (Moormann 2003) was set up to gain both knowledge about the quantity of the dowel effect and a data basis for the subsequent numerical simulation series. The dimensions of the small scale model test series were deter-mined based on typical dimensions of CPRFs and excavation pits in Frankfurt am Main and a model scale factor for the length of 1/50. The earth pressure activation is achieved by moving the up-per part of the test setup wall into the soil. A schematic sketch of the investigated problem is shown in Fig. 2.

Fig. 2. Section of the doweled earth wedge.

É¿´´ ¼·°´¿½»³»²¬ «

Û³¾»¼³»²¬¼»°¬¸ ¬

Ð·´»

ß«³»¼¸»¿® ¾¿²¼

Î»¬¿·²·²¹ ©¿´´

288

Within the testing series a variation of several parameters influ-encing the increase of passive earth pressure due to the dowel ef-fect was performed. The following parameters were varied:

- Distance between the first row of piles and the retaining wall e0 (Fig. 3)

- Distance between the piles within the pile grid e p (Fig. 3) - Embedment depth of the retaining wall t (Fig. 2) - Wall displacement type (parallel translation, rotation about

wall top, rotation about wall base, rotation about a deep fixed point in a distance of 1.25 t from the wall base)

The test setup consists of a rigid box of 720 mm x 900 mm x 800 mm (Fig. 3). The diameter of the installed model piles (tubes made of polycarbonate) has been set to 30 mm which corre-sponds to a diameter of 1.50 m in real dimensions. The embedded length of the piles was 600 mm (30 m in real dimensions).

Fig. 3. Exemplary layout and photo of the model test setup.

Some material parameters of the sand used for the tests are dis-played in Table 1.

Table 1. Material parameters of the dry quartz sand.

Material Parameter Symbol Dimension Value Angle of friction � [ ° ] 38.0 Cohesion c� [kN/m²] - Unit weight [kN/m³] 16.5 Void ratio e [ - ] 0.59 Water content w [%] 0.11

The density of the sand used in model tests has a strong influ-ence on the test results. Therefore a method of forming artificial beds of sand which are homogeneously reproducible is required (Walker & Whitaker 1967; Heineke et al. 2001). This require-ment was met by pouring the sand into the test box with a rainfall method which was successfully applied for several model tests in Darmstadt (e.g. Turek & Katzenbach 2003).

The most important results of the test series are:

- The increase of earth pressure due to the dowel effect is non-linearly dependent on the wall displacement.

- The increase of earth pressure depends severely on the type of the wall displacement.

- The earth pressure increases with decreasing geometrical pa-rameters e0 and ep.

The test results give a first glance on the dependencies and impact of the dowel effect on the earth pressure.

3 NUMERICAL SIMULATION OF THE SMALL SCALE MODEL TEST SERIES

3.1 Aim of the Numerical Investigation

The numerical investigation was carried out to derive a numerical basic model calibrated on the results of the small scale test series. This model was transferred to Frankfurt am Main soil conditions for an investigation of the quantities of the earth pressure in-crease factor which is described in chapter 4.

3.2 Description of the Constitutive Law for the Quartz Sand

The constitutive law used for the sand is an ideally-plastic Mohr-Coulomb model combined with linear elasticity.

The angle of friction, cohesion and density were identified by several laboratory tests. The remaining parameters were deter-mined within the back-analysis (Table 2) supported by laboratory tests. The low value of the Young�s modulus has to be ascribed to its stress dependency. The low stress level which exists all over the height of the test box causes the low stiffness of the im-plemented quartz sand.

Table 2. Material parameters used for the Mohr-Coulomb model.

Material Parameter Symbol Dimension Value Angle of friction � [ ° ] 38.0 Cohesion c� [kN/m²] - Angle of dilatancy [ ° ] 10.0 Young�s modulus E [kN/m²] 2000 Poisson�s ratio [ - ] 0.3 Unit weight [kN/m³] 16.5

3.3 Pre-Analysis

Before the actual numerical simulation series have been carried out a pre-analysis was performed to examine the acceptability of the basic numerical model. Because of the multiple symmetry of the test setup the numerical basic model was supposed to be re-duced to a 3-dimensional slice (Figs. 4 & 6). Therefore an inves-tigation of the influence of the friction between the lateral walls and the soil and herewith the feasibility of the basic model with respect to this friction problem was performed.

This investigation consisted of two single simulations. In the first simulation half of the whole test setup was modelled � under consideration of the onefold symmetry of the setup. As it is a model test without any piles, the results of the described simula-tion can be directly compared with the second simulation, a plain-strain simulation of the same model test (Fig. 5).

The difference between both curves is caused by the shear stresses at the lateral walls. The increase of the measured earth pressure due to the shear stresses is about 20 % which means the friction influence of the lateral walls has necessarily to be consid-ered in the 3-dimensional slice model.

çð ½³

éî ½³

»°

»ð

»°

çð ½³

èð ½³

îð ½³

êð ½³

ßß

Ù®±«²¼ °´±¬

Í»½¬·±² ßóß

Ó±ª¿¾´» ©¿´´

Ú·¨»¼ ©¿´´

Í½®»© ¬¸®»¿¼

Ð·´»

289

çð ½³

éî ½³

»°

»ð

»°

íÜóÍ´·½»

Fig. 4. 3-D-Slice (Top view).

Fig. 5. Influence of the lateral wall friction.

Further investigations concerning the influence of the wall friction of the retaining wall and of the material parameters like angle of dilatancy or Young�s modulus were carried out. The achievements of the pre-analysis were implemented in the basic model.

Ü±©»´ Ð·´»

Ú±®½»¼Ü·°´¿½»³»²¬

Fig. 6. 3-D-Slice (Displaced finite element mesh).

3.4 Assessment of the Results of the Numerical Investigation

In total 12 model tests which are listed in Table 3 were simulated. Although all simulations were set up with the same basic model and identical material parameters a good agreement between the results of the simulations and the small scale test series was achieved.

The basic form of the earth pressure displacement curves of all model tests performed without piles is equal. After reaching a maximum the earth pressure decreases to a steady level with in-

creasing displacements. This earth pressure decrease is not ob-served within the numerical simulations because of the constitu-tive law used (Fig. 7).

Table 3. Notations of the numerical simulated small scale model tests.

Notation Type of wall movement No. of piles Depth of the

wall P0P 10 Parallel 0 10 cm P0P 20 Parallel 0 20 cm P0P 30 Parallel 0 30 cm P8P 10 Parallel 8 10 cm P8P 20 Parallel 8 20 cm P8P 30 Parallel 8 30 cm K0P 20 Combined 0 20 cm K0P 30 Combined 0 30 cm K8P 20 Combined 8 20 cm K8P 30 Combined 8 30 cm K30P 20 Combined 30 20 cm K30P 30 Combined 30 30 cm

Fig. 7. Earth pressure-displacement relation of P0P 10, model test and simulation.

Fig. 8. Earth pressure-displacement relation of P8P 10, model test and simulation.

The earth pressure decrease does not appear in systems rein-forced by piles. The behaviour of the composite material �soil and piles� is distinctly different from conventional soil behaviour

Eart

h Pr

essu

re

E [N

]

Eart

h Pr

essu

re

E [N

] Ea

rth

Pres

sure

E

[N]

290

so the best accordance between test and simulation was achieved with the pile reinforced systems (Fig. 8).

The test series and the numerical simulations provide the fol-lowing important results:

- The influence of the geometrical parameter e0, ep and t that has been investigated by the small scale model test series was also observed within the numerical analysis.

- The main increase of earth pressure has to be ascribed to the first row of piles. If there are more than one pile row the fur-ther increase of earth pressure is only of minor importance.

- The behaviour of reinforced earth wedges clearly differs from the conventional soil behaviour.

The numerical basic model is validated by a good agreement between the results of the small scale model test series and the numerical simulation.

3.5 Quantification of the Earth Pressure Increase

The earth pressure increase is quantified by the earth pressure in-crease factor EEF which is defined by the ratio of the earth pres-sure measured in the test with piles and the earth pressure of the tests without piles:

)()()(

/

uuEuE

EEFpilesow

pileswithEEF

(1)

Thus, a displacement dependent increase factor is introduced. As it is shown in Fig. 9 the earth pressure increase factor rises with increasing displacement.

Fig. 9. Earth pressure increase factor EEF related to the wall dis-placement u, P8P 20, model test and numerical simulation.

3.6 Dependencies of the Earth Pressure Increase Factor EEF

In the following section the earth pressure increase factor�s de-pendencies on the angle of friction and the Young �s modulus is described. For this examination a parameter variation of these pa-rameters was carried out while the remaining parameters main-tained their primary values. The simulation series derived the subsequent results.

The earth pressure increase is influenced by the variation of the angle of friction at displacements of 2 mm matching 10 cm in real dimensions (Fig. 10). At small wall displacements (1 mm) the earth pressure increase is nearly independent of the angle of friction.

Fig. 10. Earth pressure increase factor EEF related to the wall displacement and friction angle, P8P 10, numerical simulation.

Fig. 11. Earth pressure increase factor EEF related to the wall displacement and Young�s modulus, P8P 10, numerical simula-tion.

The influence of the Young�s modulus on the earth pressure increase is much larger than the influence of the friction angle. Higher soil stiffness is connected to an increasing earth pressure (Fig. 11).

4 APPLICATION OF THE NUMERICAL BASIC MODEL ON FRANKFURT CLAY

The basic numerical model was transferred to the soil conditions of Frankfurt am Main to describe the earth pressure increase due to the dowel effect of foundation piles for excavation pits in the Frankfurt clay.

î

í

ï

Ú½

Ø§¼®±¬¿¬·½¸» ß½¸»

Ú

Ú¬

Ø§¼®±¬¿¬·½ ß¨·

Fig. 12. Yield surfaces of the Modified Drucker-Prager/ Cap model in the principal stress space.

Eart

h Pr

essu

re In

crea

se F

acto

r [-

]

Eart

h Pr

essu

re In

crea

se F

acto

r [-

] Ea

rth

Pres

sure

Incr

ease

Fac

tor

[-]

291

For the simulation of the material behaviour of the Frankfurt clay the modified Drucker-Prager/Cap model was used. This con-stitutive law uses two main yield surface segments: the pressure dependent, perfectly shear failure surface and the compression cap yield surface (Katzenbach et al., 1998). Changes of stress within the yield surfaces cause only elastic deformations while stress changes on the yield surface lead to plastic deformations. The shear failure surface is perfectly plastic whereas volumetric plastic strains can lead to hardening or softening by changing the cap position.

The material parameters for the Frankfurt clay are summarized in Table 4.

Table 4. Material parameters of the Frankfurt clay.

Material Parameter Symbol Dimension Value Angle of friction � [ ° ] 20.0 Cohesion c� [kN/m²] 20.0 Young�s modulus E [kN/m²] 50000 Poisson�s ratio [ - ] 0.25 Buoyant unit weight � [kN/m³] 9.0

To achieve knowledge about the influence of several parame-ters a variation of four geometrical parameters based on the re-sults of the model test series has been started. The following pa-rameters were varied:

- Embedment depth of the retaining wall t - Distance between the first row of piles and the

retaining wall e0- Distance between the piles within the pile grid e p- Diameter of the foundation piles D

The subsequent figures show the earth pressure increase factor EEF in dependence of the above listed parameters exemplarily

for the configuration e0 = 1.5 m, ep = 3 m, t = 3 m and D = 1.5 m. During the analysis only one of the parameters was varied while the others were held constant. Displayed are the earth pressure increase factor EEF for a wall movement of u = 5 cm and u = 10 cm.

Fig. 13. Earth pressure increase factor EEF in dependence of the pile-wall-distance e0 (ep = 3 m; t = 3 m; D = 1.5 m).

With decreasing pile-wall-distance e0 the earth pressure in-crease factor EEF rises. This effect is apparently caused by the smaller buffer of soil between the first pile row and the wall (Figs. 13 & 14).

Û¿®¬¸ É»¼¹»Û¿®¬¸ É»¼¹»

Î»¬¿·²·²¹ É¿´´Î»¬¿·²·²¹ É¿´´

¬¬

»ð»ð

Ú±«²¼¿¬·±² Ð·´»

Þ«ºº»® ¾»¬©»»² ¬¸» º·®¬°·´» ®±© ¿²¼ ¬¸» ®»¬¿·²·²¹ ©¿´´

Ú±«²¼¿¬·±² Ð·´»

Fig. 14. Section view through the earth wedge at different pile-wall-distances e0.

Fig. 15. Earth pressure increase factor EEF in dependence of the pile-pile-distance ep (e0 = 1.5 m; t = 3 m; D = 1.5 m).

Fig. 16. Earth pressure increase factor EEF in dependence of the diameter of the pile D (e0 = 1,5 m; ep = 3 m; t = 3 m).

A decreasing pile-pile-distance ep and a larger pile diameter D lead to a higher stiffness of the whole earth wedge and therefore the earth pressure increase factor EEF rises (Figs. 15 & 16).

Increasing the embedment depth of the retaining wall leads to superproportionally larger earth wedges. Thus, the ratio of the cross sectional area of the pile and the area of the whole shear band decreases.

Eart

h Pr

essu

re In

crea

se F

acto

r [-

] Ea

rth

Pres

sure

Incr

ease

Fac

tor

[-]

Eart

h Pr

essu

re In

crea

se F

acto

r [-

]

292

Fig. 17. Earth pressure increase factor EEF in dependence of the embedment depth of the wall t (e0 = 1.5 m; ep = 3 m; D = 1.5 m).

Û¿®¬¸ É»¹¼» Û¿®¬¸ É»¼¹»

Î»¬¿·²·²¹ É¿´´

¬

¬

»ð

»ð

Ú±«²¼¿¬·±² Ð·´»

Î»¬¿·²·²¹ É¿´´

Ú±«²¼¿¬·±² Ð·´»

Fig. 18. Section view through the earth wedge at different em-bedment depths t.

The maximum values for the earth pressure increase factor EEF for the exemplarily listed configuration e0 = 1.5 m, ep = 3 m,

t = 3 m and D = 1.5 m are (Figs. 17 and 18): - EEF = 1,97 for a wall movement of 5 cm and - EEF = 2,54 for a wall movement of 10 cm.

6 CONCLUSIONS

With the small model test series the increase of the earth pressure related to the dowel effect of piles can be simulated. The devel-oped and validated numerical model is appropriate to simulate the dowel effect of piles in front of retaining walls. The results of the numerical simulation form a basis for the quantification of the passive earth pressure increase due to the dowel effect of founda-tion piles.

The following conclusions can be drawn: - With increasing wall movement the earth pressure in-

crease factor EEF rises. - With decreasing pile-wall-distance e0 the earth pressure

increase factor EEF rises superproportionally. - With decreasing pile-pile-distance ep the earth pressure

increase factor EEF increases. - With increasing diameter of the piles the earth pressure

increase factor EEF increases. - With decreasing embedment depth of the wall the earth

pressure increase factor EEF rises. Due to the increased passive earth pressure the embedment depth of retaining walls can be distinctly reduced. So the consideration of the dowel effect of foundation piles in front of retaining walls consequently offers a technical and economical optimization for retaining structures.

REFERENCES

Chen, L. & Poulos, H. 1997. Piles subjected to lateral soil movements. Journal of Geotechnical & Geoenvironmental Engineering, ASCE 123 (9): 802-811 .

Heineke, S.T., Katzenbach, R., & Arslan, U. 2001. Model scale investigations on the deformation of the subsoil under railway traffic. Proceedings 15th International Conference on Soil Mechanics and Geotechnical Engineering (3): 2077-2080, Istanbul, Turkey.

Heyman, L. 1965. Measurement of the influence of lateral earth pressure on pile foundation. Proceedings 6th International Conference on Soil Mechanics and Foundation Engineering (2): 257 � 260, Montreal, Canada.

Katzenbach, R., Arslan, U., Moormann, C. & Reul, O 1998. Piled raft foundation � Interaction between piles and raft.Proceedings International Conference on Soil-Structure In-teraction in Urban Civil Engineering 4(2): 279 � 296.

Katzenbach, R., Arslan, U. & Moormann, C. 1999. Piled raft foundation projects in Germany. In: Design Applications of Raft Foundations, (J.A. Hemsley, ed.), Thomas Telford, Lon-don: 323-391.

Moormann, C. 2003. Zur Tragwirkung von Gründungspfählen beim Baugrubenaushub. Pfahl-Symposium 2003, Mitteilungen des Instituts für Grundbau und Bodenmechanik, Technische Universität Braunschweig, Heft (71): 351 � 378.

Nalcakan, M.S. & Ergun, M.U. 2001. Lateral loading of a row of model passive piles in a cohesive soil. Proceedings 15th Inter-national Conference on Soil Mechanics and Geotechnical En-gineering 2: 1219-1222, Istanbul, Turkey.

Turek, J. & Katzenbach, R. 2003. Small-scale model tests with combined pile-raft foundations. Proceedings 4th International Geotechnical Seminar on Deep Foundations on Bored and Auger Piles, Ghent, Belgium, Millpress, Rotterdam: 409-413.

Walker, B.P. & Whitaker, T. 1967. An apparatus for forming uniform beds of sand for model foundation tests. Géotech-nique 17: 161-167.

Eart

h Pr

essu

re In

crea

se F

acto

r [-

]


293

Laboratory Study on the Behaviour of Piled Raft on Granular Soils

V. Balakumar Research Scholar, Division of Soil Mechanics and Foundation Engineering, Anna University, Chennai [email protected]

K. Ilamparuthi Professor & Head, Division of Soil Mechanics and Foundation Engineering, Anna University, Chennai

Abstract : In order to understand the settlement behaviour of piled raft on granular soil, a series of small scale 1g tests were conducted in the geotechnical engineering laboratory of Anna University. The constituent elements of the model were determined based on the typical dimensions of tank foundations in around this place. The pile length was so chosen that the length is not more than the raft dimension. The tests were carried out on plain raft, combined piled raft and free standing pile group embedded in sand beds of three different densities. The performance of square piled raft and circular piled raft were compared. The factors influencing the load-settlement behaviour of piled raft were identified. Further this paper focuses on the results of the parametric study and establishes the governing factor on the settlement reduction of the raft due to the presence of piles.

1 INTRODUCTION

Piled raft is essentially raft foundation enhanced with piles with the aim of reducing the settlement. This concept of raft enhanced with piles as settlement reducer has been in the minds of geotechnical engineers as early as 1957. Even though the concept exists for a long time, this has not been an automatic choice due to the conservative approach of our building codes and the end users. Various researchers have worked on the behaviour of the piled raft. They can be grouped broadly under there heads, namely (a) analytical modeling (b) filed observations on proto type piled raft and (c) laboratory modeling. Analytical modeling has been on the process of development since 1971. Butterfield & Banerjee (1971) have used boundary element theory to study the effect of the pile cap in load sharing. Plate and spring approach, (winkler approach) has also been used in studying the load sharing mechanism (Yamashita et al., 1990; Clancy & Randolph,1993; Poulos, 1994). Combined models using boundary elements for piles and finite elements for raft have also been studied.(Hain & Lee, 1973; Frank et al.,1994; Ta & Small,1996;). Limited studies have been reported on the performance of real size structure on piled raft (Hooper, 1973; Yamashita et al., 1993; Katzenbach et al., 1999; Reul, 2000; Balakumar & Ilamparuthi, 2003;). Poulos (2001) has summarized many of the studies conducted on real size structures by various researchers and scholars. He has concluded that increase in the number of piles will not always produce best foundation performance.

Laboratory model studies on piled raft have also been reported to a limited extent. Weisner & Brown (1980) studied the behaviour of model piled raft of sizes 118mm x 118mm and 152mm x 89mm with piles of 9.6mm diameter and 250mm long founded on kaolin clay bed and validated their findings through finite element analysis by modeling the soil as elastic continuam. Cooke (1986) concluded through the model studies that the spacing and the length of the pile have influence on the behaviour of piled raft. He further reported that longer the length of the pile lesser is the settlement of the piled raft system. The influence of various parameters such as diameter of pile, length

of pile and number of piles on the load settlement behaviour of a combined piled raft foundation was investigated by Thaher & Jeeseberger (1991). Horikoshi & Randolph (1996) brought out the effect of closely spaced(small centered) pile group on the settlement of the piled raft through centrifuge model tests on an over consolidated clay bed. Recently few studies carried out on granular beds have been reported. Kim et al. (2002) conducted model tests on piled raft embedded in sand bed and developed a genetic algorithms based analysis for optimizing pile locations. Turek & Katzenbach (2003) reported the results of model tests conducted on loose and dense sand bed. Tests were conducted on instrumented model piles of 30mm diameter in a five pile group, piled raft model. It was reported that the reduction in settlement is 30% and 50% in loose and dense sand respectively.

The research on the behaviour of piled raft on sand has gained momentum recently, though it is an accepted practice to support large storage tanks on the piled raft system in loose granular deposits. Further the allowable settlement in sand is lesser than clay. Therefore, control of total and differential settlements are very important. Conventionally in the design, the presence of raft and its contribution in sharing the load is ignored, assuming that the entire load is taken by piles. If the presence of raft and its contribution in sharing the load are taken, it is quite possible, to use a relatively smaller number and shorter length of pile.

Most of the laboratory studies reported covers only the effect of spacing of piles on load sharing between raft and pile. Further studies in the laboratory covering many other important parameters such as length of the piles, spacing of piles, shape of the raft etc. are very limited. Hence it was decided to conduct an elaborate study on circular and square piled raft in granular strata. Accordingly tests were conducted on model piled raft by varying the length, diameter and spacing of the piles. Spacing were so arranged as to represent more of piled raft behaviour rather than pile group behaviour. The influence of pile length, diameter of piles, thickness of raft and spacing between the piles on settlement reduction as well as piled raft coefficient are brought out in this study inclusive of effect of density of sand bed.

294

2 SMALL SCALE MODEL TEST

The small scale model tests described in this paper were conducted at 1 g level in the geotechnical engineering laboratory of Anna University, Chennai. The shape and dimensions of the model was determined based on typical dimensions of the tanks commonly used with a scale of 1/100. Commonly used scaling law relationship have been given due consideration (Frank & Mirth 1985).

Table 1. Properties of sand used for test.

Silt 2% Fine Sand 35% Medium Sand 61% Coarse Sand 2% Max. Void Ratio 0.46 Min. Void Ratio 0.84 Coefficient of Curvature 1.22 Uniformity Coefficient 2.63 D10 0.38

In the present study vertical loading tests have been performed on plain raft and piled raft with and without contact between raft and soil. raft

2.1 Test Bed

All the tests were carried out in a steel tank of 1000mm x 600mm x 600mm. Sufficient thickness of plates and stiffeners were used to keep the tank rigid. Thick Perspex sheet was used on one face. A rigid door was kept on one side to clear the box, after finishing the test. Fig 1 indicate the test tank with loading arrangements.

2.2 Sand

Dry sand obtained from Palar river bed has been used. The grain size and the properties are given in Table 1. The tests were conducted on three different densities of sand bed namely loose (14.5kN/m3),medium dense ( 15.5kN/m3), and dense(16.5kN/m3)conditions. The corresponding values are 340, 370 and 410

respectively .

2.3 Raft

Details of the model raft and raft with piles are given in Fig. 2. Perspex sheets of 6mm, 8mm, 10mm, 12mm thickness(t) have been chosen for raft, keeping in mind commonly used raft thickness. The raft dimension has been kept as 200mm dia in the case of circular raft, and 200mm X 200mm in the case of the square raft. The top of the raft has countersunk holes to fasten the pile using stainless steel screw. Corresponding to each model of piled raft, separate plain raft was also made.

2.4 Piles

Perspex solid rods have been used as piles. 6mm, 8mm and 10mm diameters(D) have been chosen. The length of the piles (L)chosen were 200mm, 160mm, 120mm, 100m. Sufficient care was taken to see that the specimen is geometrically accurate. Threads were provided on the top end to facilitate accurate fastening, and to generate monolithic action between the two.

Fig. 1. Test setup.

Fig. 2. Plan detail of piled raft.

2.5 Head Condition

The piles were fixed to the raft at required locations using stainless steel screws. This ensures the true fixed head condition.

2.6 Loading Arrangement

The foundation was vertically loaded using a hydraulic jack fitted to a loading frame. The loading was measured by a proving ring of 20 kN capacity having the required accuracy. A rigid plate was fixed to the proving ring bottom. A loading platten with closely spaced ball shaped buttons was used to transfer the load from the jack to the foundation system uniformly. The displacements were measured using highly sensitive mechanical dial gauges placed on the diagonally opposite corners of the model.

3 PREPARATION OF THE TEST BED AND TESTING PROCEDURE

One important aspect that has considerable influence on the test behaviour is the bed preparation. It is imperative that the bed must have uniform density. In order to achieve this, sand was poured from a constant height in layers. Each layer was compacted with specially made rammer. The height of pouring and ramming were calibrated to achieve the required density. The bed was prepared for each test independently. In the case of loose sand, the height of pouring was so adjusted that the density is achieved constantly at each test.

Sand bed was prepared under three densities viz. 14.5kN/m 3

15.5kN/m3 and 16.5kN/m3 representing loose, medium dense, dense, state of compactness. A specially made template was placed on the bed and piles were driven through an outer sleeve

295

0

4

8

12

16

20

24

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0

LOAD (kN)

FREE STANDING GROUP PILED RAFT GROUP

Fig. 3. Load settlement � pile group.

0

4

8

12

16

20

24

0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0

LOAD (kN)

RAFT PILE PILED RAFT

N = 21L = 200mmD = 10mmS = 4DB = 200mmt = 8mmf = 37 °

Fig. 4. Load settlement curve � (circular piled raft).

fitted to the template. 30mm long piles were left out and the raft was fixed to it as stated earlier.

The entire pile group was pushed down as a single unit till the raft touched down the sand. While doing so, the reading on the proving ring was noted to evaluate the unit friction. The whole installation procedure represents the driving of piles. This procedure ensures the pile to soil and pile to pile interaction and ensures field condition to certain extent. A monolithic single sheet of perspex with required size and thickness was used to represent the raft in order to achieve a reasonable representation of proto type. For each density corresponding plain raft of same size and thickness was used to get the load settlement curve. After checking the levels and accuracy, a small seating load was applied. The settlement gauges were reset. The load was applied in very small increments and the corresponding settlements were recorded.

4 TEST RESULTS

The tests performed comprise of tests on plain raft and raft supported on piles with raft in contact with soil and the raft without contact. Initially a series of tests were performed and they were repeated twice to get the consistency over the preparation of bed and testing procedure. The results presented in the paper are from 24 tests on circular piled raft and 33 tests on square piled raft. Figure 3 gives the load settlement curve for a free standing pile group and the pile group with raft in contact with the soil. The load settlement curves for typical circular piled raft and square piled raft on medium dense sand are given in Figs. 4 & 5.

The corresponding load settlement curves for plain raft and pile group have also been incorporated. In all the cases the

0

4

8

12

16

20

24

0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0

LOAD (kN)

RAFT PILE PILED RAFT

N = 25L = 200mmD = 10mmS = 4DB = 200mmt = 8mmf = 37 °

Fig. 5. Load settlement curve � (square piled raft).

0.000.100.200.300.400.50

0.600.700.800.901.00

40 80 120 160 200 240 280

LENGTH OF PILE (mm)

S

MEDIUM DENSE LOOSE

N = 21D = 10m mS = 4DB = 200m mt = 8m m

f = 37 °

f = 34 °

Fig. 6. Length Vs SR (circular piled raft).

settlement reduction for the corresponding plain raft capacity was of the order of 50 to 60%.

5 LOAD SETTLEMENT BEHAVIOUR

The study of Fig. 3 indicate that the load carrying capacity of pile group, in the case of raft in contact with the soil is higher than free standing pile group except for the settlements lesser than 2mm to 3mm. At this settlement free standing piles reached their limiting resistance, there after piles exhibited uncontrollable settlement under limiting load. In the case of raft in contact with the soil the resistance of piles increased steadily with settlement and reached nearly with the load of free standing pile at 20mm settlement. Similar observations were made in tests on piled raft of various length and spacing irrespective of density of sand bed. This behaviour is due to the increase in the normal stress, as the result of the transfer of the load from the raft to the soil. The general load settlement curve as given in Figs. 4 & 5 indicate that the settlement reduction corresponding to 20mm settlement of plain raft is of the order of 47.5% for circular raft with radial pile configuration and 62% in the case of square raft. The number of piles for circular raft and square raft were so provided that area of the piles to the area of the raft is the same in both the cases. A small variation of 5% could not be avoided due to the layout requirement. The load settlement curve indicates that in the initial stages more load is transferred to the pile group, and the raft takes lesser load. As the settlement increases beyond a particular magnitude, the raft starts taking more load than the pile group. This trend has been found practically in all the tests. Also this behaviour is in agreement with the views published by Horikoshi & Randolph (1996). It is evident from the curves that as the settlement crosses 2.5mm to 3mm, the friction is over come by the piles and the raft starts taking the load. Such behaviour made the piles to function as a settlement reducer,

N = 21 L = 200mm D = 10mm S = 4D = 37°

=

=

= 37°

296

0.000.100.200.300.400.500.600.700.800.901.00

80 120 160 200 240

LENGTH OF PILE (mm)DENSE MEDIUM DENSE LOOSE

N = 9D = 10mmS = 6DB = 200mmt = 8mm

f = 41 ° f = 37 ° f = 34 °

Fig. 7. Length Vs SR (Square piled raft).

0.00

0.20

0.40

0.60

0.80

1.00

1.4 1.5 1.6 1.7

DENSITY (kN/m 3)

S

SQ. - 25% SQ. - 100%CIR. - 25% CIR. - 100%

SQ. RAFTN = 9L = 160mmD = 10mmS = 6Dt = 8mm

CIR. RAFT N = 21L = 160mmD = 10mmS = 4D

Fig. 8. Density Vs SR.

which is in conformity with the expected performance of the piledraft.

At a settlement of 20mm which is 10% of the least lateral dimension of the raft, the piles in the raft resisted 28% of the total load in the case of circular raft where as piles in square raft resisted 33% . On the other hand, for the load corresponding to 20mm settlement of plain raft, the settlement of piled raft is 10mm and 8mm for circular and square shape piled raft respectively. From the results of model rafts the piles in raft are very effective in reducing the settlement rather than contributing in resisting the load.

6 PARAMETRIC STUDY ON SETTLEMENT VARIATION

In order to understand the influence of various parameters of piled raft system on the reduction of settlement a parameter called settlement ratio, SR was used and was defined as follows

( pr - r) ) SR = ------------ (1) r where:

pr settlement of piled raft for a given load r settlement of raft for the same load.

6.1 Effect of Pile Length on Settlement Reduction

Figures 6 & 7 represent the effect of pile length on the settlement reduction for circular and square piled raft. It can be observed that the settlement reduction ratio increases as the length increases. In the case of medium dense sand the settlement reduction increases more rapidly with length for circular piled

Fig. 9. Piled diameter Vs SR (Circular raft).

raft. For square raft the increase is gradual, probably because the piles were spaced at 6D. The raft contact area becomes more in this case and hence the effect of increase in pile length was not very much prominent. However when the spacing was 4D the settlement reduction was of the order of 63% as against 50% of 6D spacing. So it can be concluded when the pile spacing is smaller, the length plays a vital role.

6.2 Effect of Bed Density

The variation in settlement reduction of circular and square piled raft are compared in Fig.8. The comparison has been done at two stages of loading namely 25% of the load at which the raft settlement becomes 20mm and at final load corresponding to a settlement of 20mm. It can be seen that the reduction in settlement in higher for lesser load irrespective of the density and shape of the raft.

As the loading becomes higher the performance is very consistent for both square and circular rafts in the reducing settlement. In general the square piled raft perform better than the circular piled raft in reducing the settlement. This may be due to the loading of square raft loads higher confinement than circular raft which leads to the development of higher frictional resistance on piles.

6.3 Effect of Pile Diameter

Three different diameter of piles have been used to study the effect on settlement reduction. The reduction in settlements of circular piled raft are presented for the settlement of 2mm/10mm and 20mmin Fig. 9. It can be seen from the Fig. 9, as the diameter increases the settlement reduction ratio also increases. In the initial stages, the diameter has got more pronounced effect due to the fact that the load transfer to the pile is higher and by shaft friction. As the settlement increases the raft starts taking more load and hence the curve becomes more linear. In other words the relation between the diameter of the pile and the settlement reduction is more linear at higher level of settlement.

7 LOAD SHARING

The distribution of the load between the two main components namely raft and the piles can well be designated by a factor called load distribution factor, pr which represents the ratio between the total load taken by the piles to the total load on the piled raft corresponding to a given settlement. The value of pr =o when the system is raft alone and = 1, when the system is totally piles. The value of pr(Figs. 10 & 11) in the initial stages

0 .00

0 .10

0 .20

0 .30

0 .40

0 .50

0 .60

0 .70

0 .80

4 5 6 7 8 9 1 0 1 1 1 2

PILE DIA M ETER (mm)

2 M M SETTLEM EN T 1 0 M M SETTLEM ENT2 0 M M SETTLEM EN T

N = 2 1L = 2 00 mmD = 1 0 mmS = 4 DB = 2 0 0mmt = 8mm

= 3 7 °

297

increase and then falls down with the settlement. This indicates that the proportion of the load taken by the pile reduces as settlement increases. This has also been indicated by Horikoshi & Randolph (1996), Turek & Katzenbach (2003).The value of prappear to decrease rapidly at initial stages of loading and until the settlement ( about 3mm to 4mm) required to over come the limiting friction value of piles. There after, though the piles take more load, its contribution is not significant which is evident from the almost constant pr values at higher settlement of piled raft. This trend has been found in all the tests. This phenomenon is happening at settlements around 3 to 5mm. In other words at this settlement the friction reaches a limiting value. The pile raft starts behaving as a block with raft sharing more load. However there is a small increase is seen in the friction. This is perhaps due to the increase in normal stress on account of the loading on the raft. The tests conducted on a free standing pile group and the pile group with raft in contact with the soil confirms this effect as the load taken by the pile group of the piled raft is higher than the load taken by the free standing pile group.

7.1 Comparison of Square and Circular Piled Raft

In Figs. 10 & 11 the variation of pr Vs settlement for both circular and square piled raft for a particular length and thickness of the raft are compared. The figures indicate that the square and circular piled raft perform well in the case of loose sand compared to medium dense and dense. In the case of circular raft the pr does not vary much between loose and dense. In the case of square raft while there is not much of difference in the prvalue between dense and medium dense, the value of pr in the case of loose sand is much higher than the medium dense and dense sand. But in either case pr value reduces as the settlement increases. The prvalue is higher in the initial stages of settlement of around 3mm to 5mm beyond this the pr value reached more or less constant value with increase in settlement. This trend confirms that the pile group takes more load initially then the raft

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0 5 10 15 20 25SETTLEMENT (mm)

loose medium dense

Fig. 10. Pr Vs settlement (Circular Raft).

Fig. 11. Pr Vs settlement (Square Piled Raft).

shares most part of the applied load despite that there is some increase in the pile resistance. This behaviour clearly shows that settlement of higher order is required for active involvement of raft in load sharing mechanism. At initial stages of loading most part of the load goes to the pile, however there is a increase in pile resistance due to increase in normal stress on piles due to load transfer mechanism between raft and soil. The tests conducted on the freestanding pile groups confirmed this.

At higher settlement the raft starts taking more load generating block action leading to a fall in the pr value. This confirms the fact that the increase in the soil strength below the raft plays a vital role in making the piles function efficiently as settlement reducer. The bearing effect of the raft becomes more important at higher settlement. It can be seen very clearly that in the case of circular raft piles take more load in the initial stages of loading than the square raft but as the settlement increases the square pile raft transfers more load to the pile. Hence it is quite evident that the bearing of the raft also plays a very important role in the behaviour of piled raft system. The reduction in settlement corresponding to a plain raft capacity at 20mm settlement is 50% for circular raft and 60% for square raft.

7.2 Effect of Pile Length

In Fig. 12, the variation of pr with length of pile is presented. As the length increases the pr value increases irrespective of the order of settlement of the piled raft. The pr value is maximum for the pile length of 200mm which is equal to the width of the raft. The increase in pr value with the length is almost linear irrespective of the settlement and density of sand bed. The variation in pr for settlements higher than 10mm is almost negligible irrespective of lengths. Similar observation was recorded in other tests on circular raft tested in loose and dense

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0 50 100 150 200 250Length of pile (mm)

2 mm sett 10 mm sett 20 mm sett

Fig. 12. Pr Vs pile Length (Circular piled raft).

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

4 5 6 7 8 9 10 11 12

PILE D IAME TER (m m )

2MM SETTLEM ENT 10MM S ETTLEME NT

20MM SETTLE ME NT

N = 21L = 200m mD = 10m mS = 4DB = 200m mt = 8m m

= 37 °

Fig. 13. Pr Vs pile diameter (Circular Piled Raft).

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0 5 10 15 20 25

SETTLEMENT (mm)

loose medium dense

n = 21 = 37°

t = 8mm

298

sand conditions. Since the study concentrates on relatively shorter piles, effect of pile length larger than the raft dimension is not discussed here. However, the effect of length is not very much pronounced in the case of tests on the square piled raft with larger spacings (6D). This is perhaps due to the fact that the square raft with larger spacing has more contact area.

7.3 Effect of Pile Diameter

Figure13 indicates the effect of pile diameter on PR for a given density. The diameter has a predominant role in the load sharing in the initial stages of settlement. This is perhaps the load taken by the pile gets transferred in the form of shaft friction. At higher level of settlement, the variation of pr is gradual with increase in diameter and is lesser in magnitude than .pr of initial settlement.

8 CONCLUSIONS

1. The load on piles of piled raft system is higher than the free standing pile groups irrespective of length, diameter, spacing of piles and density of sand bed. The contribution by the piles against load beyond the settlements corresponding to limiting friction is stable and function of intensity of load on the piled raft.

2. The settlement of square piled raft is lesser than the circular piled raft for a given set of condition of the piled raft.

3. A term called settlement reduction ratio (SR) is used to understand the influence of various parameters of piled raft on settlement variation. SR increases with increase in length of pile and pile diameter. However SR showed decreasing trend with increasing in density of sand bed.

4. The load sharing between the piles and raft of a piled raft system is mainly depends on the settlement. At settlements lesser than the deformation required for limiting friction, Most of the applied load is taken by the piles. Beyond this settlement, raft shares most part of the load irrespective of the thickness and shape of the raft, length and the spacing of the piles.

5. The load distribution factor pr decreased rapidly at the initial stages of loading, and almost reached a constant value with increase in settlement. The pr value is higher in loose sand than denser states of sand bed. However this effect is more pronounced in square raft than the circular raft.

6. The pr value is higher for longer length of pile and higher diameter irrespective of order of settlement of piled raft.

7. The thickness of the raft investigated in this study does not show significant effect on load sharing irrespective of sand bed and various parameters of pile.

ACKNOWLEDGEMENT

The authors wish to express their deep sense of gratitude for the inspiring encouragement of Dr. H. G. Poulos, Senior Principal, Coffey Geosciences Limited, Sydney. Also the authors expresses his gratitude to M/S Simplex Concrete Piles (India) Limited, and Indian Roof Crafts, Chennai for their financial support in fabrication of experimental set up and models.

REFERENCES

Butterfield, R. & Banerjee, P.K. 1971. The Problem of Pile Group and Pile Cap Interaction. Geotechnique 21(2):135-142.

Balakumar, V. & Ilamparuthi, K. 2003. Field study on piled raft foundations of twelve storied building at Chennai. Proceedings of the 6th International Symposium on Field Measurements in Geo Mechanics � Oslo, Norway, 17-22.

Clancy, P. & Randolph, M.F.1993. Simple design tools for piled raft foundations. Geotechnique 46(2): 313-328.

Cooke, R.W. 1986. Piled raft foundations on stiff clays- a contribution to design philosophy, Geotechnique 36 (2): 169-203.

Frank, E. & Mirth, G. 1985. Scale effects in 1 g model tests on horizontally loaded piles. Proceedings of the 11th

International Conference on Soil Mechanics and Foundation Engineering, San Francisco, U.S.A: 1011-1014.

Frank, E .et al., 1994. Measurements and numerical modeling of high rise building foundations on Frankfurt clay. Vert. And Horzl. Deformations of Foundations and Embankments. , ASCE Geotech. spec. pub. no. 40(2):1325-1336.

Hooper, J.A. 1973. Observations on the behaviour a piled raft foundations on London clay. Journal of Institution of Civil Engineers 55(2) 77-90.

Horikoshi, K. & Randolph, M.F. 1996. Centrifuge modeling of piled raft foundations on clay. Geotechnique 46(4) :741-752

Katzenbach, R. et al., 1999. Piled raft foundation projects in Germany. Design applications of raft foundations and ground slabs. Ed. By J.A.Helmsley, Thomas Telford Ltd.

Kim, H.T., Yoo, H.K. & Kang, I.K. 2002. Genetic algorithm optimum design of piled raft foundations with model tests Journal of South East Asian Geotechnical Society : 1 �9.

Poulos, H. G. 1994. An approximate numerical analysis of pile-raft interaction. International Journal of Numerical & Analytical methods in Geomechanics 18: 73-92.

Poulos, H. G. 2001. Piled Raft Foundation: Design and Application, Geotechnique 51(2): 95-113.

Prakoso, W.A. & Kulhawy, F.H. 2001. Contribution of piled raft foundation, Journal of Geotechinical & Geo-environmental Engineering, ASCE: 17-24.

Reul, O. 2000. In-Situ Messungen Und Numerische Studien Zum Tragverhalten Der Kombinierten Pfahl-Plattngrundung, Mitteillungen Desinstituts Und Der Versuchanstilt Fur Geotecnique Der Technischen Universitadt.Darmstadt.Heft.53

Ta, L.D. & Small, J.C. 1996. Analysis of piled raft systems in layered soil .International Journal of Numerical & Analytical methods in Geomechanics 20:57-72.

Thaher & Jesseberger, H.L. 1991. Investigation of the behaviour of pile raft foundations by centrifuge modeling. Proceeding of the 10th ECSMFE, Florence, Italy: 597-603.

Turek. J. & Katzenbach. R. 2003. Small scale model tests with combined pile raft foundations. Proceedings of the 4th

International Geotechnical Seminar on Deep Foundations on Bored and Angered piles, Ghent, Belgium: 409-413.

Wiesner T. & Brown P. T. 1980. Laboratory tests on model piled raft, Journal of Geotechnical Engineering , ASCE: 767-783.

Yamashita, M. & Kakurai, M. 1991. Settlement Behaviour of the Raft Foundation with Friction Piles, Proceedings of the 4th Int. Conf. on Piling and Deep Foundation: 461-466.

Yamashita, K. et al., 1993. Settlement behavior of a five story building on a piled raft foundation. Proceedings of the 2nd Int. Geot. Sem. on Deep Foundation on Bored & Augerpiles, ghent.A.A. Balkema Rotterdam: 3512-3516.


299

Analysis of Load Tests on Large Diameter Bored Piles in Very Dense Cemented Sands

N. F. Ismael Civil Engineering Department, Kuwait University, POB. 5969, Safat 13060, [email protected]

Abstract: The ultimate bearing capacity of short, large diameter bored piles was examined in connection with the construction of a multistory building on the Arabian Gulf in Kuwait. Testing included 0.45 m, 1.0 m, and 1.2 m piles installed through loose sand into a bearing stratum consisting of very dense cemented sands. All tests were carried out to failure. The base resistance and shaft friction were calculated using the Meyerhof method for a layered soil profile. The method employs the standard penetration test N-values. The results indicate that the piles are point bearing piles with a great portion of the pile capacity due to base resistance. The mobilized skin friction is small. The calculated pile capacities were very close to the measured values with the maximum difference not exceeding 10%.

1 INTRODUCTION

Short, large diameter bored piles are used with increasing frequency for heavy structures in Kuwait, and the Gulf region where ground conditions indicate loose deposits of calcareous sands. The piles range in length from 8 m to 15 m with their base located in the dense-to-very dense, cemented sand bearing deposit. The design of these piles requires knowledge of the skin friction and point resistance in these soils. While field tests on driven piles are available (Ismael 1989 & Ismael 1999), field test data on large diameter bored piles installed through loose calcareous sands into dense-to-very dense cemented or partially cemented sands are very limited at present. The results of three load tests carried out recently on bored piles at one site in Kuwait have become available. In these tests the failure load was either reached or can be extrapolated from the test results. It is important to examine and analyze the test results for the benefit of the geotechnical research and design engineers.

This paper presents the soil conditions, pile installation details, and the pile load test data at the test site located in Shuwaikh, Kuwait. The site is facing the Arabian Gulf. Test results were directly analyzed and the point resistance, and skin friction were calculated. The ultimate capacities of the piles were calculated and compared with the measured values. The influence of the pile diameter on the settlement at ultimate and at working loads is examined.

2 SOIL CONDITIONS AND PILE INSTALLATION

Figure 1 shows a brief summary of soil conditions and the dimensions of the test piles. At the site of each test pile a surface layer of loose fine sand or silty sand is underlain by very dense cemented silty sand. The cemented sand extends well below the pile tip elevations and has a Standard Penetration Test SPT N value 50. The penetration of the piles into this layer varies from 3.3 m to 6 m. Pile 1 is 0.45 m diameter and 10.8 m long. Pile 2 is 1 m diameter and 8.7 m long. Pile 3 is 1.2 m diameter and 8.5 m long.

All piles were installed by drilling using a steel casing for the full length of the piles. The steel case was then installed followed by concreting. Upon completion of concreting, the casing was removed.

3 PILE TESTING

All piles were tested in accordance with ASTM standard 1143-81 (ASTM 1994). At least three load cycles were carried out in each test corresponding to 100%, 200%, and 300% of the proposed working load. For pile 1, the test was carried out to large displacement and until the ultimate bearing capacity failure occurred. This was evident by the continuous settlement at constant load. However, for piles 2 and 3 failure was progressive in nature, and the failure load was taken by the slope tangent method at the point of intersection of the initial and final tangents to the load settlement curve. Figures 2 to 4 show the load settlement curves for the test piles. The vertical arrows in the figures indicate the failure load.

4 ANALYSIS OF TEST RESULTS

A summary of all pile data and design parameters is given in Table 1. The base resistance and shaft friction were calculated using the Meyerhof method for a layered soil profile (Meyerhof 1976). Accordingly, the skin friction fs for low or no displacement bored piles is first calculated as

2s kN/mN0.96tsf

100Nf (1)

where N is the average SPT-N value along the pile shaft in the bearing stratum.

Fs = fs * Db * p (2)

where Fs is the frictional load, Db is the depth of the pile in the

bearing stratum, and p is the pile perimeter.

300

Fig. 1. Soil conditions and bored pile details.

Fig. 2. Load-settlement curve for 0.45 m diameter pile.


301

Table 2. Settlement data of the test piles.

Parameter Pile 1 Pile 2 Pile 3

Pile width, B 0.45 m 1.0 m 1.2 m Total penetration 10.8 m 8.7 m 8.5 m Working load 500 kN 3600 kN 5400 kN Settlement at working load 1.6 mm 9.3 mm 7.3 mm Settlement ratio at working load Se/B 0.36% 0.93% 0.61% Failure load 2700 kN 9500 kN* 11000 kN* Settlement at failure 43.7 mm 32.5 mm 18 mm Settlement ratio at failure Sf/B 9.71% 3.25% 1.5% *These values are the initial failure loads.

Table 1. Design parameters for calculated pipe capacities.

Parameter Pile 1 Pile 2 Pile 3

Pile width, B 0.45 m 1.0 m 1.2 m Total penetration 10.8 m 8.7 m 8.5 m Depth to top of bearing stratum 7.5 m 3.5 m 3.5 m Penetration into bearing stratum Db 3.3 m 5.2 m 5.0 m N = average N-value within length Db 50 blows/0.3 m 50 blows/0.3 m 50 blows/0.3 m N-value at final penetration 50 blows/0.3 m 50 blows/0.3 m 50 blows/0.3 m

fs = 100N tsf (0.96 N kN/m2) 48 kPa 48 kPa 48 kPa

qb = 0.4 N BbD 4 N tsf, (0.383 N MPa) 14.04 MPa 9.96 MPa 7.98 MPa

Friction above bearing stratum 0 0 0 Fs = fs Db B 224 kN 784 kN 905 kN

End bearing, Q = qb 4

2B 2233 kN 7821 kN 9024 kN

Total calculated capacity 2457 kN 8605 kN 9929 kN Load tested capacity 2700 kN 9500 kN 11000 kN Ratio of calculated to measured capacity 91% 91% 90.3%


302

Friction within the upper layer of loose sand was judged as insignificant and was ignored in the analysis.

The base resistance qb is given by:

qb = 0.4 N Db/B 4 N tsf, (0.383 N MPa) (3)

Therefore, for BbD 10 qb = 0.383 N MPa (4)

for BbD < 10 qb = 0.383 N

B10bD MPa (5)

where B is the pile diameter.

The SPT-N values averaged 50 along the depth of the piles in the bearing stratum and was 50 in the vicinity of the pile tips. It was taken as 50 in the calculations shown in Table 1. A close examination of Table 1 reveals that despite the empirical nature of the method of analysis the calculated pile capacities are very close to the measured values. The ratio of the calculated to measured capacity is ~91% for all three piles. Moreover, the frictional resistance is about 9% of the total calculated capacity indicating that the piles are predominantly point bearing piles.

5 SETTLEMENT

Table 2 shows a summary of the test pile settlement at both the working and the ultimate loading. For the proposed working load, the corresponding settlement is very small and does not exceed one percent of the pile diameter. However, at failure, the settlement ratio is 9.71% for the 0.45 m diameter pile decreasing to 3.25% and 1.5% for the 1 m and 1.2 m diameter piles respectively. The reason for this discrepancy is related to the method of taking the failure loads for piles 2 and 3. Ultimate failure was not reached for these piles and the failure load taken by the slope tangent method represents the initial failure of a progressive local shear failure mode. Thus piles 2 and 3 could resist a larger load at the ultimate failure condition than the loads determined herein.

6 CONCLUSIONS

Short large diameter bored piles installed through calcareous sand with their bases in very dense cemented sands were load tested to failure. The results were analyzed and the calculated ultimate capacities were compared with the measured values. The following conclusions are made:

1. The piles usually penetrate a loose-to-compact calcareous surface deposit underlain by a competent dense-to-very dense cemented sand layer. The pile bases and the lower part of the pile shafts are located in the lower layer, which is siliceous with low carbonate content.

2. A major portion of the calculated capacity is derived from tip or base resistance. The small skin friction component of ~10% of the total capacity is derived from the bearing layer along the pile shaft.

3. The calculated pile capacities based on the Standard Penetration Test N values, using the classical Meyerhof method for a layered soil profile yielded accurate predictions. They were within 9% of the measured values.

4. For large diameter bored piles ultimate failure would occur at very large load which is well beyond the loading capacity that can be reached in the field. The failure load taken by the slope tangent method is considered representation of the failure condition.

5. Large diameter bored piles installed in cemented sands has the advantage of supporting large axial and lateral loads compared with driven piles.

REFERENCES

ASTM Standards. 1994. Annual Book of ASTM Standards, Vol. 4.08, ASTM-D1143-81, American Society for Testing and Materials, Philadelphia, Pa, U.S.A.

Ismael, N.F. 1989. Skin friction of driven piles in calcareous sands. Journal of Geotechnical Engineering Division, ASCE115(1):135-139.

Ismael, N.F. 1999. Analysis of load tests on piles driven through calcareous desert sands. Journal of Geotechnical & Geoenvironmental Engineering Division, ASCE 125(10): 905-908.

Meyerhof, G.G. 1976. Bearing capacity and settlement of pile foundation. Journal of Geotechnical Engineering. Division,ASCE 102(3): 195-228.


303

�Swiss-Cheese� � A Method of Degrading Soil Crust and MinimisingRisk to Punch Through Problem on the Installation of Mobile OffshoreDrilling Unit (MODU)

P. Handidjaja, P. Somehsa, & M. Manoj Somehsa Geosciences Pte. Ltd, 10, Jalan Tembusu Singapore, 438225 www.somehsa.com

Abstract: A number of potential problems can affect the performance of jack-up rig footings. The most significant problem withrespect to jack-up rig foundation performance is rapid leg penetration termed �punch-through�. Typically punch-through problems are associated with stratified soil deposits, in which, a strong soil layer overlies a weak soil layer. Most serious case of punch � through is when bearing capacity of the layered system is more than the pre-load, but bearing capacity to pre-load ratio is not sufficient enough to preclude punch-through. One of the methods to reduce punch-through potential in such a situation, is to reduce the bearing capacity of the soil by making perforations in the soil. The reduced bearing capacity facilitates the penetration of spudcan through the strong layer to the underlying weak soil layer. This method of reducing the bearing capacity by making perforations in a strong soil layer overlying a weak layer is known as �Swiss Cheese�. Swiss Cheese operation requires careful planning. Some of the geotechnical aspects anddrilling constraints to be considered during planning of Swiss cheese operation are discussed in this paper.

1 INTRODUCTION

Most of the offshore drilling is performed from self- elevating mobile platforms, commonly known as jack-ups. The foundation of jack-ups is a approximate large inverted cones, known as �spudcan�. Spudcans are generally hexagonal or roughly circular in plan and typically have a shallow conical underside with a sharp protruding spigot or cone tip.

In its operation a jack up mobile unit floats to location with its legs lifted. At the drilling site, the legs are lowered to mudline and jacked down in to the soil until the bearing capacity of soil becomes equal to the load carried by each leg. At this point the hull begins to lift out of water. Jacking continues until the air gap is about 1.5 m and the rig is levelled to within 0.1 degrees. Pre-loading by ballasting with water is then performed until all the leg penetration has stopped. Static test is then performed by holding preload water for about one to four hours. Then, pre-load water is dumped and finally hull is jacked to operation-level.

Certain geotechnical and geological conditions pose potential hazards to safe foundation performance of mobile jack ups. These hazards should be investigated and remedial measures taken to reduce the risk of failure.

Stratified soil deposits at shallow depths often pose serious risk to jack � up rig installation. Soil stratification, in which, a

strong soil (�crust�) overlies a relatively weak soil strata, gives rise to three general situations:

1. Bearing capacity is sufficiently less than pre-load so that spudcan penetrate through the strong soil layer to the underlying weak soil.

2. Bearing capacity is significantly more than that of pre-load so that no punch through takes place.

3. Bearing capacity is more than pre-load but the bearing capacity/pre-load ratio is not sufficient enough to preclude punch � through.

The third condition is the most serious as jack-up legs will penetrate until the strong soil layer, which can support it, and stops temporarily. During pre-loading or operation, the legs may suddenly punch � through the strong layer leading to serious consequences. If soil investigation suggests this possibility, then remedial measures should be taken to alleviate the potential

Fig. 1. Typical spudcan cross section.

Fig. 2. Mobile jack-up rig with individual footings foundation.

304

punch through condition. One of the methods, which has been found to be fairly successful, is making perforations in the strong layer in order to sufficiently reduce the bearing capacity and thereby facilitate the penetration of legs through the strong soil layer. This process of reducing the bearing capacity by making perforations in soil is known in the industry as �Swiss Cheese�.

Another situation requiring Swiss Cheese operation is, when a mobile offshore drilling unit is to re-visit a platform location where a different MODU type of rig was placed before. If configurations of spudcan imprints for both rigs do not match, it will lead to serious consequences such as spudcan following or sliding into old hole. As a result, the trusses may bend and jack and pinion arrangement get over-stressed, resulting in foundation tilt and even failure. In such situations, the existing imprint shall be widened or deepened to match the configuration and required bearing capacity of a new rig by using Swiss Cheese operation.

Swiss Cheese operations were fairly successful carried out for a number of platforms in South China Sea.

In this paper, goetechnical aspects and installation considerations of Swiss Cheese operation are discussed.

2 OCCURRENCE AND DISTRIBUTION OF SOIL �CRUST�

Soil �crust� exists when the soil resistance at the crust is higher than the leg load of the jack-up rig and soil resistance below crust is lower than the leg load to support the rig. Also, thickness of �crust� is less than half of the footing diameter of the spudcan.

This crustal zone is commonly present in South China Sea, Gulf of Thailand, Sunda Shelf, including Java Sea. Figure 3 shows the general distribution map of crustal layer in the region. The first crustal zone appears, generally, between 80 to 90m below present sea level. Thickness of this upper crustal zone varies from 1 to 3m. It is characterised by high shear strength and may have been formed by weathering during world-wide lowering of sea level related to Pleistocene glaciation. Figure 4 shows a typical boring log with a �crust�.

The soil layer overlying the crust, formed after this glaciation period may have inconsistent deposition and distribution over the area. Hence, the level and thickness of crust may vary laterally.

3 FACTOR OF SAFETY AGAINST PUNCH-THROUGH

The Factor of Safety against punch through is defined as the peak bearing capacity (peak soil resistance) divided by the footing load (spudcan reaction). A factor of safety greater than one indicates the bearing capacity exceeds the footing load. Conversely, a factor of safety less than one indicates footing is likely to penetrate through the strong stratum. A factor of Safety greater than 1.5 with respect to spucan reaction indicates that punch through is unlikely to occur. Figure 5 shows a typical bearing capacity vs depth plot of a potential punch-through condition.

4 GENERAL ANALYSIS PROCEDURE

General procedure is to reduce the bearing area of �crust�, and thereby reduce the bearing capacity. Bearing area reduction is achieved by creating perforations in crust. Area reduction required to lower the factor of safety to a value less than one is then estimated. Then the number of boreholes such that the total combined area is equal to the area to be reduced, is determined.

During drilling the soil around the circumference is disturbed. However, for the estimation of number of boreholes required, this disturbance is normally neglected.

5 DRILLING PLAN AND PROCEDURE

Swiss Cheese operation can be performed from MODU or Jack-up by utilising the main draw-work and drilling system after positioning the drill string over each of the proposed spudcan leg location. Swiss Cheese operation can also be carried out from a coring vessel, with 4 point mooring arrangement, by utilising soil boring drilling rig, either mounted over the side or at the centre.

Fig. 3. Map showing distribution of �crust�.

Fig. 4. Example of a boring log with �crust�.

305

The later will provide better option to prepare the site in advance prior to rig installation.

It is common practice to distribute the boreholes evenly in the spudcan foot print area. A typical pattern is shown in Fig. 6.

Successful Swiss Cheese operation require careful planning considering all the factors influencing the operation. The factors influencing the Swiss Cheese operation are discussed below.

5.1 Diameter of Holes

Diameter of boreholes depends on the availability of drilling bits and drilling rigs. During the selection of drill bits and drilling rigs, consideration should be given to the torque required for the drilling bit to pierce the crust vertically. The drilling rig should be such that, it can exert enough torque to penetrate the hard crust without damaging the unit and at efficient rate of penetration. The bottom hole assembly (BHA) for Swiss Cheese operation, normally consists of a minimum of 26� tricon drilling bit with an optional 36� hole-opener, as shown in Fig. 7.

Obviously, drilling with bigger bit will cut down the number of holes to drill and save cost and time. However, to handle bigger bit is limited to the availability of Crane and space available in coring vessel or jack-up rig.

5.2 Spacing of Boreholes

Experience suggests that borehole spacing is a very important parameter for the efficient drilling operations. For a given site condition and drill bit, there appears to be a critical minimum spacing, below which efficient drilling becomes difficult. When the spacing is less than this critical value, drill bit tends to slip to the adjacent drilled borehole. The critical minimum spacing is found to be influenced by:

Diameter of borehole Strength of hard crust Thickness of crust Depth of crust below mudline Pumping pressure and flow Weather conditions (currents)

It is difficult to quantify the effect of these factors on the minimum spacing of boreholes. Therefore for any Swiss Cheese is advisable to perform a trial operation, hence determine the minimum spacing. This trial should be carried out very near to the proposed site for Swiss Cheese operation.

5.3 Drilling Pattern

Drilling pattern, the arrangement of boreholes within the spudcan foot print area, is another important factor deciding the efficiency of Swiss cheese operation. Though the standard procedure is to distribute the boreholes evenly, experience suggest that, it is better to have more holes along the periphery of the spudcan foot print area.

When spudcan tip is trying to penetrate hard �crust�, due to the conical shape of spudcan, there will be a lateral (radial) force as well as a downward force.

The lateral thrust will push the �crust� sideways and downward force will push the crust in to the under-laying softer soil (Fig. 8).

Therefore the most efficient borehole pattern is the one which facilitates this failure mechanism. One of the drilling pattern which facilitates this failure mechanism is to have as many bore

peak

Fig. 5. Typical bearing capacity vs depth.

Fig. 6. Standard borehole distribution pattern.

Fig. 7. One of the drilling bits used for Swiss Cheese operation.

306

holes as possible along the periphery of the spudcan foot print area . Figure. 9 shows recommended pattern of boreholes.

The resistance of the crustal layer against spudcan penetration consists of: a) Shear resistance provided by the crust along the perimeter

over the thickness of crust and b) Bearing capacity of softer soil below the crust

Resistance given by the crust depends mainly on the effective perimeter of the spudcan foot print area, shear strength of crust and thickness of crust. Here the effective perimeter is defined as:

Effective perimeter = Total perimeter - diameter of holes x number of holes along perimeter

Therefore, more the number of holes along the perimeter of

spudcan foot print area, lesser the effective perimeter and hence lower the resistance against spudcan penetration. Therefore it is advisable to distribute more holes along the periphery of the spudcan foot print area in order to reduce the bearing capacity of the crust. The area to be Swiss Cheesed should be slightly more than that of the proposed spudcan.

5.4 Pump Pressure

During drilling, water is pumped at high pressure through the drill string. The jetting action of this water coming out through drill bits, remove cuttings from the bottom of hole. There appears to be an optimum water pressure and flow rate for efficient Swiss Cheese operation. This optimum water pressure can be determined during trial run

5.5 Weather Condition

Sliding of drill bit to the previously drilled adjacent borehole depends to certain extend on weather condition. Probability of sliding appears to be high during rough weather and strong bottom currents. This is because rough weather and currents make it difficult to keep the movement of vessel (roll and pitch) steady and hence drill string movement is unavoidable, especially when cutting the top soil layer which is often soft, prior advancing the crust layer.

6 CONCLUSIONS

Swiss Cheese procedure is increasingly being used to minimise the risk of punch through. The risk factor is high when Swiss Cheese is to be done very near to an existing platform. The procedure is based on the assumption that reducing bearing area of crust reduces the resistance of the crust against spudcan penetration. The standard practice is to estimate the number of boreholes required for reducing the bearing area and then distribute these boreholes evenly in the spudcan foot print area. In this paper on Swiss cheese operation, various geotechnical aspects in general and drilling constraints in particular are discussed, based on author�s experience.

7 FURTHER STUDIES

Further studies are needed to evaluate the probability of success of Swiss Cheese operation. Detailed experimental/theoretical studies are required to understand the failure mechanism during spudcan penetration, after Swiss Cheese operation. Such studies will enable us to understand the influence of various factors such as shear strength and thickness of crust, diameter and number of bore holes, distribution of boreholes etc. Also effect of Swiss Cheese operation on foundation of adjacent structures and the reduction in leg fixity and consequent reduction in lateral load capacity also require further detailed studies.

REFERENCES

Recommended Practice for Site Specific Assessment of Mobile Jack-up Units, SNAME, 1994. First Edition.

Maung, M. and Che, A. 2000. Swiss cheesing to bring in a jack-up Rig at Anding location, IADC/SPE 62755, Petronas Carigali SDN, BHD.

Castleberry.J.P & Prebaharan, N. 1985. Clay crusts of the Sunda Shelf. Proceedings of the 8th Southeast Asian Geotechnical Conference, Kuala Lampur, Malaysia.

Fig. 8. Likely failure mechanism.

Fig . 9. Recommended pattern of boreholes.

Foundation in Difficult Subsoil

Documents

Transcript of Foundation in Difficult Subsoil