MECHANICAL BEHAVIOR OF PILE FOUNDATION CONSTRUCTED …

961

i) Chief Researcher, Civil Engineering Research Institute for Cold Region, Japan (ko-tomsw＠ceri.go.jp).ii) Professor, Graduate School of Engineering, Hokkaido University, Japan.

The manuscript for this paper was received for review on May 26, 2006; approved on May 29, 2007.Written discussions on this paper should be submitted before May 1, 2008 to the Japanese Geotechnical Society, 4-38-2, Sengoku, Bunkyo-ku,Tokyo 112-0011, Japan. Upon request the closing date may be extended one month.

961

SOILS AND FOUNDATIONS Vol. 47, No. 5, 961–972, Oct. 2007Japanese Geotechnical Society

MECHANICAL BEHAVIOR OF PILE FOUNDATION CONSTRUCTEDIN COMPOSITE GROUND AND ITS EVALUATION

KOUICHI TOMISAWAi) and SEIICHI MIURAii)

ABSTRACT

This paper describes a foundation design method in which the ground is improved around the heads of pile founda-tions in soft ground or loose sandy ground and its practical eŠectiveness. The shear strength increased due to groundimprovement is re‰ected in the horizontal resistance of piles. In this design method, the in‰uence range of the horizon-tal resistance of piles and the necessary range of ground improvement are determined by taking account of three-dimensional domain formed with the gradient of the surface of passive failure. The horizontal subgrade reaction ofpiles is evaluated by converting the shear strength of improved ground to the modulus of deformation. In this study,the validity of design method for the pile foundation with ground improvement was conˆrmed through an in-situhorizontal loading test. The dynamic behavior of pile foundation constructed in improved ground was also investigat-ed through a series of centrifuge model tests and numerical analyses. The in‰uence of the diŠerence in strength be-tween the original and improved grounds on piles during an earthquake was also conˆrmed based on the numericalanalyses. The cost performance of the proposed method was discussed by comparing with the case without ground im-provement.

Key words: centrifuge shaking test, composite ground, horizontal subgrade reaction, in-situ test, pile (IGC: E4WE12WH1)

INTRODUCTION

The composite ground in this paper is deˆned as thefoundation subgrade ground with parts formed by im-proving methods such as deep mixing, preloading andsand compaction piling. When designing the foundationsof piles to be constructed in composite ground, how to setthe necessary range of ground improvement and thestrength of the ground are the key issues on evaluating thehorizontal subgrade reaction to the piles. However, thereis no rational and uniêd determination method for suchfactors. Consequently, there is also no systematic seismicdesign method that considers the in‰uence of the dynam-ic behavior of the composite ground on pile foundationsduring earthquakes.

Tomisawa and Nishikawa (2005a, b) developed areasonable design method, in which prior to constructingpiles the ground is improved at the shallow part of thepile foundations in soft ground and loose sandy ground.In this method, increased shear strength is re‰ected in thehorizontal resistance. The construction method studieduses a combination of pile foundation together with com-mon ground improvement methods, including deep mix-ing (DM), preloading and sand compaction piling (SCP),and is referred to as the composite ground pile method(Civil Engineering Research Institute of Hokkaido, 2002;

Public Works Research Center, 1999).This study presents a design procedure for the compo-

site ground pile method which evaluates the range of in-‰uence of the horizontal resistance of piles and the neces-sary range of ground improvement on the basis of en-gineering judgment and the horizontal subgrade reactionof piles from the shear strength increased due to groundimprovement. The validity of the proposed design proce-dure for the composite ground pile was conˆrmedthrough an in-situ horizontal loading test together with anumerical analysis using three-dimensional elastic ˆniteelement method under static conditions. Next, a series ofcentrifuge model tests and numerical analyses using elas-tic ˆnite element method were carried out to clarify thedynamic behavior and the horizontal bearing capacity ofthe composite ground pile during earthquake. Seismicresistance of the composite ground pile was evaluatedfrom the reductions in displacement and bending mo-ment of piles caused by earthquake motions on the basisof the results of centrifuge model tests. Since improvedground shows higher strength than the original ground inthe proposed composite ground pile method, diŠerent dy-namic response of the pile is supposed to occur in im-proved ground system. On the basis of the results of nu-merical analyses, the in‰uence of the diŠerence instrength between the original and improved grounds on

962

Fig. 1. 3-D image of the lateral resistance of pile foundation and thesetup for the range of ground improvement

962 TOMISAWA AND MIURA

the horizontal bearing capacity of piles during earth-quake was clariêd by comparing the earthquake-inducedground deformation and sectional forces of pile for thecases with and without composite ground. The proposeddesign procedure by considering the in‰uence range ofthe horizontal resistance of pile and the increased shearstrength of the composite ground enables properly settingthe necessary range of ground improvement for the bestperformance of the composite ground pile and evaluatingthe eŠects of ground improvement on the seismicresistance of the pile foundation. Finally, cost perfor-mance of the composite ground pile method was dis-cussed on these results.

EVALUATION FOR HORIZONTAL RESISTANCEOF PILES INSTALLED IN COMPOSITE GROUND

Consideration of the Range of Ground ImprovementAs indicated in the studies on the lateral resistance of

piles by Broms (1964a, b) and Reese et al. (1974), therange of in‰uence of horizontal resistance in the groundwhen horizontal force is applied to a pile spreads gradual-ly as load increases. As a result, when the failure limitstate of the ground is reached following the horizontaldisplacement of the pile, a state of equilibrium is consi-dered to be maintained between the maximum value ofthe horizontal subgrade reaction and the passive earthpressure.

In the composite ground pile method, therefore, thenecessary range of ground improvement, i.e., the neces-sary range to be considered fully improved to generatehorizontal subgrade reaction to the pile, is proposed to bea three-dimensional domain formed with the gradient ofthe surface of passive failure u＝(459＋qW2) (q: angle ofshear resistance of soil) from the depth of the characteris-tic length of piles, 1Wb, (b＝4 (kD)W4EyI), which meansthe in‰uence depth of the horizontal resistance of piles onthe basis of the limit equilibrium and the Mohr-Coulombfailure criterion. Here Ey, D and I are Young's modulus,diameter of pile and second moment of area of the pilesection, and k is coe‹cient of horizontal subgrade reac-tion. To safely set the depth of ground improvement, k oforiginal ground is proposed to be adopted for calculatingthe characteristic length of piles, 1Wb.

As a result, the necessary range of improvement is setas a three-dimensional inverted cone shape centered onthe pile. However, since it is di‹cult to conduct groundimprovement in a cone shape due to construction limita-tions, a cubic body covering the range of the invert coneshape shown in Fig. 1 was proposed for the range ofground improvement. The method for setting the rangeof ground improvement for pile group is the same as thatfor a single pile.

Method for Determining Horizontal Subgrade ReactionIn determining the horizontal resistance of piles, the

subgrade reaction p acting on a pile is considered to in-crease in proportion to the de‰ection y expressed by theequation p＝k・y, where the coe‹cient k is deˆned as the

coe‹cient of horizontal subgrade reaction (k value), onthe basis of the elastic bending theory of beam (Chang,1937). The k value depends on pile speciˆcations andground properties (Terzaghi, 1943; Vesic, 1961); the for-mula for calculating k shown in Eq. (1) is widely used forthe design of pile foundations in Japan (Japan RoadAssociation, 2002a). This means that k value for thehorizontal resistance of piles constructed in compositeground can be calculated from the modulus of deforma-tion E.

k＝1

0.3aE・Ø DWb

0.3 »－3W4

(1)

where E is the modulus of deformation of the im-proved ground (kNWm2), a is a compensation coe‹cientfor the k value which depends on the prediction methodof E, D is the diameter of pile (m), and b is the character-istic value of piles (m－1).

When DM is used for ground improvement with thecomposite ground pile method, the ground is formed byplacing improved columns with required strength aroundthe piles. The shear strength of composite ground S is cal-culated by Eqs. (2) and (3) (Civil Engineering ResearchInstitute of Hokkaido, 2002), combining the shearstrength of the improved column Sc and that of originalground S0 according to the improvement rate ap;

S＝Sc・ap＋as・S0(1－ap) (2)

Sc＝qupW2, S0＝qu0W2, ap＝ApWA (3)

where Sc is the shear strength of the improved columns(kNWm2), S0 is the shear strength of the original ground,ap is the ground improvement rate, as is the ratio ofstrength reduction of the original ground due to thefailure strain of the improved columns which is usually

963

Fig. 2. Relationship between the unconˆned compressive strength quc

of improved columns and the modulus of deformation Ec (cohesivesoil) (modiêd from Public Works Research Center, 1999)

963MECHANICAL BEHAVIOR OF PILE FOUNDATION

set as 1W2 to 1W3, qup is the unconˆned compressionstrength of improved columns (kNWm2), quo is the uncon-ˆned compression strength of the original ground, Ap isthe area of the cross-section of the improved columns(m2), and A is the distribution area per improved column.Greater strength can be expected for the compositeground when DM is adopted for ground improvement;the shear strength of improved columns Sc in the compo-site ground pile method is approximately 200 to 500 kNWm2, which is the range of commonly used design values.In this case, improved columns are installed one by onewith an improvement rate of ap＝78.5z or higher to en-sure a certain level of the horizontal subgrade reaction ofthe piles.

Consequently, to properly select the horizontal sub-grade reaction of piles in improved ground in a design forthe composite ground pile, it is necessary to evaluate theincreases in shear strength S and modulus of deformationE of the ground due to ground improvement. In DMmethod, the shear strength of improved columns Sc canbe obtained from the relationship between unconˆnedcompression strength quc and shear strength of improvedcolumns Sc (＝qucW2), as shown in Eq. (3). It is also wellestablished that the unconˆned compression strength quc

is proportional to the modulus of deformation Ec of im-proved columns. For example, the design manual for thedeep mixing method for onshore construction (PublicWorks Research Center, 1999) gives the relationship Ec＝100 quc for improved cohesive ground (Fig. 2). Therefore,it is assumed that the modulus of deformation of im-proved ground E can be treated equivalently to shearstrength S, i.e., E for calculating k value of the improvedground can be expressed as shown in the following equa-tion in the same way as the calculation of S.

E＝Ec・ap＋as・E0(1－ap) (4)

where Ec and E0 are the moduli of deformation of the im-

proved columns and the original ground respectively.Similarly, when using preloading, vacuum consolida-

tion, or other surcharge methods for ground improve-ment around piles, the shear strength S of the ground af-ter consolidation is determined by

S＝Su＋DS＝Su＋m・DP・U (5)

where Su is the undrained shear strength of the originalground, DS is the increment of shear strength due to con-solidation, m is the rate of increase in shear strength, DPis the stress increment in the ground and U is the averagedegree of consolidation of the soft ground layer (PublicWorks Research Center, 1999). The rate of increase inshear strength, m, needs to be set in detail through soiltests for diŠerent types of ground materials.

By regarding the increment in shear strength DS gener-ated by consolidation as equivalent to the increment inthe modulus of deformation DE achieved by ground im-provement, the modulus of deformation E for the fullyimproved ground can be calculated in the same way, evenwhen using the surcharge method.

When SCP is applied to the composite ground pilemethod for ground improvement around piles, k value inthe composite ground is calculated by Eq. (6) (e.g.Japanese Geotechnical Society, 1988) using the improve-ment ratio of sand piles as;

k＝ks・as＋kc・(1－as) (6)

where ks is k value of sand piles (kNWm3), and kc is k valueof the original ground among sand piles. This means thatks and kc can be calculated directly from the respectivemodulus of deformation which is evaluated from SPT N-value or by the horizontal loading test in a borehole. N-value of sand piles formed by SCP is approximately 10 to15, and therefore a greater improvement rate as cannot beset. However, since the installation of sand piles increasesthe density of ground among sand piles, an increase in thehorizontal resistance of piles can be expected.

By re‰ecting the increased ground strength from anyparticular ground improvement in the horizontal sub-grade reaction, as for the composite ground pile men-tioned above, it becomes possible to adopt a greaterhorizontal resistance for piles in improved ground, whichre‰ects closely the actual ground conditions. Althoughground improved by the composite ground pile methodhas higher strength than the original ground, the rigidityof piles is much greater than that of the improvedground, and the strength of the improved ground is alsopresumed to be non-uniform in most cases due to theconstruction methods of ground improvement. There-fore, the improved part around piles for the compositeground pile method should not be considered as a struc-ture ˆxed to the pile heads, but as a ground with relativelyhigh rigidity in the same way as the case when pile foun-dation is constructed in common grounds.

964

Fig. 3. A bridge for which the composite ground pile method was adopted

Fig. 4. Horizontal loading test device and reinforcing bar strain gaugeinstallation plan


VERIFICATION OF THE PROPOSED DESIGNMETHOD THROUGH IN-SITU HORIZONTAL PILELOADING TESTS

To conˆrm the validity of the proposed design method,as a case study, a series of in-situ horizontal loading testswas carried out on foundation piles of an abutment fordirectly evaluating k value for the composite ground pilemethod. The site selected for the loading test was a cast-in-place pile foundation of a box-type abutment con-structed in soft ground. In this site, the horizontal sub-grade reaction of piles was too small to satisfy the designcriterion without ground improvement, and it was ap-propriate to use DM for the composite ground pilemethod. The in-situ loading tests focused mainly to rev-eal the behavior of horizontally loaded piles installed inthe improved ground.

The cast-in-place piles were 1.2 m in diameter and 17 min length. In the area surrounding the cast-in-place pilesof the abutment, DM was adopted with a design uncon-ˆned compression strength of qup＝200 kNWm2 and an im-provement rate of ap＝78.5z for the improved columns.DM was applied in the same way as when it was employedfor ensuring slip stability at the back of the abutment andas a countermeasure to prevent lateral spreading. Therange of ground improvement was set as a three-dimen-sional cubic body around the piles to cover a space withthe gradient of surface of passive failure u＝(459＋qW2)from the depth 1Wb＝3.65 m, i.e., the whole layer of softground ( see the method shown in Fig. 1). Figure 3 showsa general view of the bridge abutments and the founda-

tions together with the N-value of the original ground.The horizontal loading test was carried out using a

multi-cycle load control system and was performed in ac-cordance with the guidelines for pile loading tests of theJapanese Geotechnical Society (1983). A load cell, ahydraulic jack, and a loading tower were installed be-tween the test and reaction piles, and a static load was ap-plied in one direction until pile head displacementreached approximately 1z (0.012 m) of the pile di-ameter. To measure the bending stress of the test piles,reinforcing bar strain gauges were installed on two linesof the main reinforcing bars of the test piles at regular in-tervals along the depth. Figure 4 shows a diagram of thehorizontal loading test system and the detailed installa-tion of the reinforcing bar strain gauges. The strain

965

Table 1. Coe‹cients of horizontal subgrade reaction of piles

Coe‹cient ofhorizontalsubgradereaction(MNWm

3)

Ratio of coe‹cientof horizontal

subgradereaction to thedesign value

Design value for improved ground k 47.8 1.00

Original ground k0 9.3 0.19

Value calculated from the unconˆnedcompression strength kU

107.0 2.24

Value estimated from the in-situhorizontal loading test kH

100.0 2.09

Fig. 5. Relationship between horizontal load H and displacement atpile head y

Fig. 6. Distribution of pile bending stress obtained from the in situhorizontal loading test together with the calculated values


gauges were carefully waterproofed.Table 1 shows a comparison of the coe‹cients of

horizontal subgrade reaction of piles, the design value forthe improved ground k, the value for original ground k0,the value calculated from the measured unconˆned com-pression strength of improved columns kU and the valueestimated from the results of the horizontal loading testkH together with their ratios to the design value of im-proved columns k. The value of the original ground k0

was found to be 9.3 MNWm3 from the modulus of defor-mation of the original ground E0 where N§5 or lower.The design value for the improved ground was calculatedto be k＝47.8 MNWm3 by Eqs. (1), (2) and (3) based onthe design unconˆned compression strength quc＝200 kNWm2 and improvement rate ap＝78.5z, in which therelationship between the modulus of deformation of theimproved columns Ec and quc shown in Fig. 2 was adopt-ed. However, the average unconˆned compressionstrength of the improved columns at the material age of28 days after construction was quc＝408 kNWm2 at thedepth of 1Wb, and the calculated value of kU was 107.0MNWm3, which was almost twice as high as the designvalue. The value kH was found to be around 100 MNWm3,from the secant gradient between the horizontal load Hand the displacement of pile-head y, at which standarddisplacement of 1z of the pile diameter ( y＝0.012 m)was adopted. The displacement of pile-head y was deter-mined according to Hayashi and Chang's method(Hayashi, 1921; Chang, 1937) as:

y＝HW2EyIb3 (7)

Figure 5 shows (1) the relationship between the meas-ured horizontal load H and pile head displacement y ob-tained from the in-situ horizontal loading tests of pilesand (2) the relationship between H and y calculated byEq. (7). As a result, the value kH§100 MNWm3 estimatedfrom the horizontal loading test at the standard displace-ment of the pile head was coincident with the calculatedvalue kU＝107.0 MNWm3 from the proposed designmethod. Therefore, it could be considered that the re-quired horizontal subgrade reaction of the pile horizon-tally loaded is governed by the shear strength of theground in the range of passive failure at the shallow partof the pile.

Figure 6 shows the pile bending stress calculated by thelinear elastic subgrade reaction method compared withthe pile bending stress measured from the in-situ horizon-tal loading test at the horizontal load level of Hmax＝1800kN. The bending stress calculated using kU was close tothe measured value, but was diŠerent from the value cal-culated from k0.

The bending stress distribution of piles obtained fromthe analysis using the three-dimensional elastic ˆnite ele-ment method is also shown in Fig. 6. For the analysismodel, a cubic body of improvement was reproducedaround the piles with a depth of 1Wb in accordance withthe proposed setting method, and a horizontal load whichis equivalent to that of the in-situ horizontal loading testwas applied to the pile head. In the three-dimensionalelastic ˆnite element analysis, the modulus of deforma-tion of the original ground was set on the basis of test

966

Fig. 7. Model setup for the centrifuge shaking test (composite groundpile model)

Table 2. Physical properties of kaolin clay and Toyoura sand

Kaolin clay Toyoura sand

Unit weight kNWm3 10.101 15.574

Particlepercentge

Sand z — 97.3

Silt z 50.3 0.8

Clay z 49.7 1.9

Cone index qc MNWm2 1.0 3.3

Shear modulus G MNWm2 4.3 23.8


results, and the modulus of deformation for improvedcolumns was set as E＝40000 kNWm2, which was calculat-ed from qup＝400 kNWm2. Since it was an elastic analysis,cohesion C and internal friction angle q were not takeninto account. However, the eŠect of nonlinearity on thepile rigidity was taken into account depending on thestrain level. As a result, the bending stress distribution ofpiles obtained from the analysis using the three-dimen-sional elastic ˆnite element method was closely coincidentwith the measured bending stress in the same way as thevalue obtained from the analysis using the elastic sub-grade reaction method.

The validity of the proposed design method for thecomposite ground pile, i.e., setting the range of groundimprovement and calculating the horizontal subgradereaction from the increased shear strength due to groundimprovement, was conˆrmed from the in-situ loading testunder static condition described above.

EVALUATION OF EARTHQUAKE RESISTANCEFOR THE COMPOSITE GROUND PILE METHOD

Centrifuge Model Tests for Veriˆcation of the DynamicBehavior of Composite Ground Pile with DiŠerent Rigid-ity between Improved and Original Grounds

The validity of the proposed design method for compo-site ground pile was conˆrmed through an in-situhorizontal loading test under the static conditions. Toverify the dynamic behavior of composite ground pilewith diŠerent rigidity between improved and originalgrounds, a series of centrifuge model tests was conductedto investigate the dynamic responses of both the groundand the piles. To simplify the model for testing, auniform sand ground with relatively high rigidity wasadopted as the improved ground around the pile head,and soft cohesive soil was assumed to be the originalground. As the materials of model grounds, air-driedToyoura sand and kaolin clay were used for the improvedand original ground respectively.

For the dynamic centrifuge model test, a 1:50 model ofground and piles was prepared in a laminar model con-tainer with inner dimensions of 0.7 m×0.2 m×0.35 m. A50 g (g: gravitational acceleration) centrifugal accelera-tion êld was adopted for the test to satisfy the similaritylaw on the stress level. A model pile with an outer di-ameter d＝0.01 m, thickness t＝0.002 m, and pile lengthL＝0.4 m was made from steel pipe and specially ˆn-ished. A prototype scale steel pile with outer diameter d＝0.5 m and thickness t＝0.01 m was simulated in the cen-trifugal acceleration êld for this model pile according tothe similarity law. A steel block with a weight of W＝3.92N (equivalent to 490 kN at the prototype scale) was ˆxedto the pile head to simulate the substructure and super-structure, and the lower end of the model pile was ˆxed inthe model base ground with gypsum. Strain gauges wereattached to the model pile to measure both axial andbending strain generated along the axis of the pile, andacceleration sensors were installed on the pile head and inthe ground to measure the responses of both pile and

ground during shaking. Figure 7 shows the setup of thetest model. To form the model ground, a single layer ofair dried kaolin clay was prepared to simulate theuniform soft ground, and the improved ground was pre-pared by replacing kaolin clay with Toyoura sand in thearea around the pile head to simulate the improvedground with relatively high rigidity. The improvedground using Toyoura sand was a rectangular body withlength and width of 0.2 m and height of 0.1 m centeredaround the pile to simulate the range of in‰uence of thehorizontal resistance of the pile, which was assumed tocover the inverted cone formed by the gradient of the sur-face of passive failure u＝(459＋qW2) from the depth ofthe characteristic length of piles, 1Wb, based on theproposed setting method. The model grounds were pre-pared by pouring both kaolin clay and Toyoura sandfrom a certain height. Table 2 shows the physical proper-ties of kaolin clay and Toyoura sand. The relative densityof improved ground with Toyoura sand was set at 85z.As the mechanical properties of the model grounds, theratio of strength expressed by the cone index qc at 1 g con-dition between kaolin clay and Toyoura sand was ap-proximately 1:3, and the ratio of rigidity expressed by theshear modulus G drawn from bender element tests be-tween them was approximately 1:5. The details of the

967

Fig. 8. Pile frequency for uniform ground of kaolin clay and im-proved ground with Toyoura sand

Fig. 9. Distribution of horizontal displacement of pile during shaking


bender element test have been described by Timpong etal. (2005).

A sine-wave was used as input motion in the shakingtests. The acceleration level of the input motion in themodel was set as 10 mWs2 (equivalent to 20 gal at the pro-totype scale) to simulate the behavior of the pile andground over a small range of deformation (the horizontaldisplacement of pile head was nearly equivalent to 1¿2zof the pile diameter).

Figure 8 shows a comparison of the frequency curvesfor piles between uniform ground of kaolin clay and im-proved ground using Toyoura sand. These data were ob-tained from shaking tests using sine-waves with variousfrequencies. The vertical axis in the ˆgure represents theratio of Fourier amplitude of the acceleration response atthe pile head to that of the input acceleration at the baseground, i.e., the ratio of transfer functions and curve ˆt-ting for the frequency curves was performed to removenoise by using a three-order function. As a result, thenatural frequency of the pile was found to be 0.95 Hz inthe uniform ground of kaolin clay and 1.4 Hz in the im-proved ground using Toyoura sand. Because the im-proved ground has greater shear strength, the period ofthe pile in composite ground becomes shorter. Since thedamping eŠect of the whole bridge system (superstructureand substructure) changes with the shortening of theperiod (heightening of the frequency) of the pile founda-tion, careful attention should be paid to the radiationdamping phenomenon from the pile foundation to theground when considering the earthquake resistance ofpiles in improved ground (Railway Technical ResearchInstitute, 1999).

Figures 9 and 10 show the distributions of horizontaldisplacement and bending moment respectively along theaxes of piles installed in the uniform ground of kaolinclay and improved ground using Toyoura sand at a timeinterval of Dt＝0.1 sec. The data for these ˆgures wereobtained during shaking using the sine-waves with natur-al frequencies of piles as the input motions. The maxi-mum displacement of the piles in the improved groundusing Toyoura sand was only 1W6 of that of piles inuniform ground of kaolin clay (dmax for the improved

groundWdmax for the uniform ground §0.0016 mW0.0107m). The maximum bending moment occurred within therange of depth 1Wb was 1W2.5 of that in uniform groundof kaolin clay (Mmax for the improved groundWMmax forthe uniform ground §25.2 kN-mW64.3 kN-m) and con-verged within the range of improved ground with Toy-oura sand. This was considered to be due to the restric-tion on the pile head in improved ground and the decreasein displacement amplitude of the pile during shaking. Asa result, by using ground improvement around the pilehead, a certain amount of earthquake resistance that res-tricts pile displacement and bending moment againstearthquake motion was achieved, though careful atten-tion should be paid to the shortening of the natural fre-quency of pile foundations in improved ground.

Two-dimensional Elastic Finite Element Analyses forClariˆcation of the Dynamic Response of the Pile in theBoundary between Improved and Original Grounds

The greatest concern about the horizontal dynamic be-havior of piles in composite ground seems to be theresponse of piles at the boundary between improved andoriginal grounds, where diŠerent ground rigidity appears.This problem was examined for the composite ground

968

Fig. 10. Distribution of bending moment of pile during shaking

Fig. 11. Model used for the dynamic two-dimensional elastic ˆnite ele-ment method analysis


pile when DM was adopted through numerical analyses.To qualitatively clarify the dynamic response of the

pile in the boundary between improved and originalgrounds, a series of dynamic two-dimensional ˆnite ele-ment analyses was conducted. The analyses were mainlyfocused to reveal the earthquake resistance of compositeground piles where original ground with diŠerent thick-ness exists between composite ground and base. Themodel used for analysis was a one-span abutment foun-dation shown in Fig. 11. This site has typical conditionsfor the composite ground pile method using DM as theground improvement method, and the purpose of numer-ical analyses was to acquire basic information on theearthquake resistance of the composite ground pile suchas the in‰uence of the diŠerence in strength between theoriginal and improved ground, and the earthquake in-duced ground deformation and sectional forces of pileunder diŠerent ground conditions. The composite groundpile method was applied to the upper layer composed ofloose silt and ˆne sand. The pile model consisted of cast-in-place piles (pile diameter D＝1.2 m, pile length L＝11m, pile arrangement n＝5×4＝20) of an abutment foun-dation supported by mudstone bedrock. The rigidity offour piles installed in the direction perpendicular to the

axis of the bridge was taken into consideration for analy-sis. The range of ground improvement was set as a cubicbody with a depth of 5.2 m, equivalent to the depth of1Wb, and the same width as the outside piles. b was esti-mated by calculating the coe‹cient of horizontal sub-grade reaction k0 from the modulus of deformation E0＝2100 kNWm2 at the upper layer of the original ground inaccordance with the proposed setting method. The im-provement rate was set at ap＝78.5z of DM for the oneby one installation pattern. In the analysis model, theweight of superstructures such as the abutment andbridge beam was treated as a vertical load, and thehorizontal load applied to the pile foundation was set asthe sum of the inertial force of the substructure and thedynamic earth pressure acting on the abutment during anearthquake at the seismic intensity method level. The ver-tical and horizontal loads were applied to the center ofthe bottom of the footing respectively. The inputparameters for the improved columns and originalground are shown in Table 3. The modulus of deforma-tion of the composite ground E was calculated by usingEq. (4), in which the modulus of deformation of theoriginal ground E0 was obtained from soil tests and thatof the improved columns Ec was derived from therelationship between Ec and quc given in Fig. 2. For Pois-son's ratio n, a commonly used value was adopted for theoriginal ground, and for the improved columns, n＝1W6was adopted by referring to past test results published bythe Public Works Research Center (1999).

The dynamic shear modulus of the original ground GD,was calculated from the elastic shear wave velocity Vs

through Eq. (8) (Japan Road Association, 2002b).

969

Table 3. Input parameters for the ground materials in dynamic two-dimensional ˆnite element analysis

Unit weightgt (kNWm

3)

Modulus ofdeformationE (kNWm

2)

Dynamic shearmodulus

GD (kNWm2)

Poisson'sratio

n

Improvedground

17.0 63000 1250 0.17

Sandy silt 18.0 8004 950 0.35

Sandy soil 17.0 39002 2670 0.35

Bedrock ofmudstone

20.0 140000 10000 0.35

Fig. 12. Shear rigidity ratio GWG0 and damping h-shear strain g curve


GD＝gt・V 2s Wg (8)

where, gt is the unit weight (kNWm3), Vs is the elastic shearwave velocity (mWs) and g is the gravitational accelera-tion. The dynamic shear modulus for the improvedcolumns GD was converted from static shear modulus ofelasticity GS calculated by using the static modulus ofdeformation E as shown in Eq. (9):

GD＝51.0・(GSW100)0.781 (9)

where, GS＝EW2(1＋n). In the analysis model, cast-in-place piles were treated as beam elements. Improvedcolumns in the composite ground were treated as linearmaterial because they have a certain amount of shearstrength. The nonlinearity of sandy silt and sandy soilwas taken into account.

To clarify the pile stress at the boundary between theimproved and original grounds during an earthquake, thefollowing three cases with diŠerent analysis models werecarried out: (1) the range of improvement was to a depthof 1Wb and reached the stiŠ layer (NÀ10), i.e., the casewith the analysis model as shown in Fig. 11 (Case 1); (2)the improved ground did not reach the stiŠ layer andthere was a thin weak layer with a thickness of 1 m be-tween the improved ground and the stiŠ layer (Case 2);and (3) the improved ground did not reach the stiŠ layerand there was a thick weak layer of 5 m between the im-proved ground and the stiŠ layer (Case 3). The weak lay-ers in Case 2 and 3 were considered to be the surface siltlayer, and its shear strength was set as 1W6 that of the stiŠlayer.

Since the dynamic analysis mainly focused on thediŠerence in response characteristics of diŠerent groundconditions, two-dimensional ˆnite element method usingcomplex response analysis of the equivalent linearity wasadopted. In the analysis, the equivalent width of a singlepile perpendicular to the axis of the bridge was 3.25 m,the boundary conditions were set as the energy transmis-sion boundaries on the right and left sides, and the cohe-sion boundary was the bottom of the bedrock. The Kai-hoku Bridge motion recorded during the Miyagiken-Okiearthquake in 1978 was induced as the input seismicwave. The peak acceleration amplitude of the inputearthquake motion was adjusted to be 105 gal based onthe seismic intensity method.

According to the technical manual of liquefactioncountermeasures (Coastal Development Institute ofTechnology, 1997), nonlinearity of the ground was ad-dressed by changing the ratio of the shear modulus ofelasticity GWGD and the damping constant h depending onthe strain level for diŠerent types of ground (sandy andcohesive) as shown in Fig. 12.

Figure 13 shows the distribution of the maximum shearstrain occurring around the piles A, C, and E in thehorizontal direction of the ground for Cases 1, 2 and 3.While no shear strain of the ground was observed at thetop and bottom ends of the piles, relatively large shearstrain occurred at the boundary between the improvedand original grounds. In Case 2, there was signiˆcantshear strain of the ground (gmax＝4.5z) at the boundarybetween the bottom of the improved grounds and the un-derlying thin weak layer. This shear strain value was con-sidered to be caused by the yielding of the weak ground,and there was a possibility that a high level of stresswould be concentrated at such point of the piles follow-ing the deformation of the ground. However, the maxi-mum shear strain at the boundaries of the grounds inCases 1 and 3 was equal to or less than gmax＝0.8z, ap-proximately 1W6 of the value in Case 2.

Figure 14 shows the bending moment M, axial force N,and shear force S for pile A (front side) in Cases 1, 2, and3. The bending moment distribution of pile M varied ineach case. While a relatively large bending moment wasobserved at the pile heads and bottom of the improvedground, especially in Case 2, the maximum bending mo-ment was Mmax＝2554 kN-m and the bending stress of thepiles was calculated as smax＝241 MNWm2, which is withinthe range of allowable stress for reinforcing bars duringan earthquake ssa＝300 MNWm2 and design standardstrength of concrete sck＝24 MNWm2. The axial force ofpiles N showed a general decreasing tendency with depth.A relatively small axial force of the pile in Case 3 wasconsidered to be in‰uenced by the phase diŠerence. While

970

Fig. 13. Distribution of the maximum shear strain occurring aroundthe piles A, C, and E in the horizontal direction of the ground forCases 1, 2 and 3

Fig. 14. Sectional force diagrams of bending moment, axial force, andshear force of pile A for Cases 1, 2 and 3


the shear forces S were relatively small for Cases 1 and 3,a large shear force (Smax＝1787.0 kN) equivalent to theshear resistance of piles S＝1,800 kN developed at theboundary between the improved and original ground inCase 2, which was similar to the behavior of the maxi-mum shear strain of the ground.

As with the results, in the composite ground pilemethod, relatively large shear strain of the ground and

sectional forces of the pile in the boundary between im-proved and original grounds were induced during earth-quake when thin layer of unimproved weak groundremained between the improved ground and stiŠ baseground as the condition of Case 2. Therefore, it is clari-êd that when deciding the depth of ground improve-ment, basically it is set as the characteristic depth of piles1Wb, but need to check carefully if there is thin week layer

971

Fig. 15. Comparison of pile foundation shapes in the conventionalmethod and composite ground pile method


between improved ground and stiŠ base ground will beremained or not to ensure earthquake resistance. Sinceground conditions are usually complicated for most casesin the êld, it will be necessary to verify the earthquakeperformance of piles at the boundary using dynamicresponse analysis (Japan Road Association, 2002b) whenthe dynamic behavior of piers and other foundationstructures are expected to be under particular conditions.

USEFULNESS OF THE COMPOSITE GROUND PILEMETHOD

The proposed design procedure that accounts for thein‰uence range of the horizontal resistance of pile and theincreased shear strength of the composite ground enablesproperly setting the necessary range of ground improve-ment for the reasonable design of the composite groundpile and evaluating the eŠects of ground improvement onthe seismic resistance of the pile foundation. High costperformance by using composite ground pile method wasachieved comparing with the conventional constructionmethods.

Figure 15 shows a comparison of the foundation shapefor composite ground piles and that for the conventionalmethod without ground improvement under the same de-sign conditions of the site described in the previous para-graphs, in which a horizontal loading test of full-scalepiles was carried out as a typical case study of the compo-site ground pile method.

When using the conventional method, unless embank-

ment reductions or other special measures are taken at theback of the abutment, an unrealistic number of piles, n＝14×5＝70, is necessary and the scale of constructionmust be increased to ensure adequate horizontalresistance against soft ground. By using the compositeground pile method, the horizontal resistance of piles canbe ensured by ground improvement around cast-in-placepiles. It is therefore possible to make the abutment sizesmaller by reducing the number of piles to n＝3×4＝12,although the pile diameter needs to be changed from 1.0m to 1.2 m. As a result, in this case the constructioncosts, including ground improvement around the pilefoundation, would be reduced by approximately 45z byadopting the composite ground pile method comparedwith the conventional method.

To date, the composite ground pile method has beenadopted as the countermeasure for soft ground or loosesandy ground at ten sites. The reduction in constructioncost is estimated to have been 20z to 50z, demonstrat-ing the usefulness of this method. The reduction in con-struction cost mentioned here is the ratio of the cost forone bridge abutment, including the cost of the abutmentbody, pile, and ground improvement, when the compo-site ground pile method is adopted to when it is notadopted. In this case, the construction cost is strongly de-pendent on the number of piles.

CONCLUSIONS

Through a series of experimental and numerical stu-dies, a procedure on the composite ground pile methodwas proposed. Design for the composite ground pile onthe basis of the conventional design of pile foundations ispossible by considering the following issues:(1) The range of in‰uence of the horizontal resistance of

composite ground piles, or the necessary range ofground improvement, is a cubic body that covers theshape of the gradient of the surface of passivefailure u＝(459＋qW2) from the depth of the charac-teristic length of piles, 1Wb, based on existing designconsideration and engineering judgment.

(2) The validity and safety of the composite ground piledesign method was veriêd with the existing linearelastic subgrade reaction method and three-dimen-sional elastic ˆnite element analysis, which wereconducted based on the results of in-situ horizontalloading tests of piles for an actual bridge founda-tion.

(3) The dynamic behavior of composite ground pilewith diŠerent rigidity between improved and origi-nal grounds was conˆrmed through a series of dy-namic centrifuge model tests. Su‹cient earthquakeresistance of piles in composite ground, such as therestriction of horizontal pile displacement and bend-ing moment against earthquake motions was veriêdon the basis of the test results.

(4) The in‰uence of the boundary between improvedand original grounds on the dynamic response of thepiles was clariêd through dynamic elastic ˆnite ele-


ment analyses. It was conˆrmed that thin weak layerremained between improved ground and stiŠ baseground will induce relatively large shear strain of theground and sectional forces of the pile in the bound-ary between improved and original grounds.

(5) Based on the case studies of several practical con-struction sites, the cost reduction of the compositeground pile method was conˆrmed to be 20z to 50z compared with the conventional method if it isapplied to sites on soft ground or loose sandyground where the speciˆcations of piles are con-trolled by horizontal resistance.

ACKNOWLEDGEMENTS

The authors would like to thank Dr. Jun'ichi Nishika-wa from the Civil Engineering Research Institute forCold Region, Prof. Makoto Kimura from the Interna-tional Innovation Center of Kyoto University and Dr.Liming Li from R&D Center of Nippon Koei Corpora-tion for their valuable advice on the analysis of the behav-ior of pile foundations.

REFERENCES

1) Broms, B. B. (1964a): Lateral resistance of piles in cohesive soils,Proc. ASCE, 90(SM3), 27–63.

2) Broms, B. B. (1964b): Lateral resistance of piles in cohesionlesssoils, Proc. ASCE, 90(SM3), 123–157.

3) Chang, Y. L. (1937): Discussion on ``lateral pile-loading tests'' byL. B. Feagin, Trans ASCE, 102, 272–278.

4) Civil Engineering Research Institute of Hokkaido (CERI) (2002):Manual for Countermeasure against Peat Soft Ground, 95–129 (inJapanese).

5) Coastal Development Institute of Technology (1997): Handbook ofLiquefaction Countermeasures for Reclaimed Land (in Japanese).

6) Hayashi, K. (1921): Theorie des tr äagers auf elastischer unterlageund ihre anwendung auf den tiefbau, Springer Verlag.

7) Japanese Geotechnical Society (1983): Horizontal loading testmethod for pile foundations, JGS Standard (in Japanese).

8) Japanese Geotechnical Society (1988): Countermeasures againstSoft Ground, 132–133 (in Japanese).

9) Japan Road Association (2002a): Speciˆcations for HighwayBridges Vol. IV: Substructures (in Japanese).

10) Japan Road Association (2002b): Speciˆcations for HighwayBridges Vol. V: Seismic Design (in Japanese).

11) Public Works Research Center (1999): Deep Mixing Method forOnshore Construction—Design and Construction Manual—,48–148 (in Japanese).

12) Public Works Research Institute (1999): Manual for design andconstruction of countermeasure against liquefaction, ResearchReport, (186) (in Japanese).

13) Railway Technical Research Institute (1999): Design Criteria forRailway Structures and Instruction Manual—Seismic Design— (inJapanese).

14) Reese, L. C., Cox, W. R. and Koop, F. D. (1974): Analysis of later-ally loaded pile in sand, Proc. OŠshore Technology Conference,Houston, TX, OTC2080.

15) Terzaghi, K. (1943): Evaluation of coe‹cient of subgrade reaction,Geotechnique, 5(4), 118–143.

16) Timpong, S., Miura, S. and Yara, K. (2005): EŠect of consolida-tion time on shear modulus of crushable volcanic soils, Soils andFoundations, 45(5), 115–119.

17) Tomisawa, K. and Nishikawa, J. (2005a): Pile design method incomposite ground formed by deep mixing method, Journal of Ge-otechnical Engineering, Japan Society of Civil Engineers, (799WIII–72), 183–193 (in Japanese).

18) Tomisawa, K. and Nishikawa, J. (2005b): A design method con-cerning horizontal resistance of piles constructed in improvedground, 16th ICSMGE, September 12–16, 2, 2187–2192.

19) Vesic, A. (1961): Bending of beams resting on isotropic elasticsolid, Journal of Engineering Mechanics, ASCE, 87(2), 35–53.

MECHANICAL BEHAVIOR OF PILE FOUNDATION CONSTRUCTED …

Documents