Download as pdf or txt
Download as pdf or txt
You are on page 1of 20

SOILS AND FOUNDATIONS Vol. 40, No. 5, 113-132, Oct.

2000
Japanese Geotechnical Society

SEISMIC EVALUATIONS OF PILE FOUNDATIONS WITH THREE DIFFERENT


METHODS BASED ON THREE-DIMENSIONAL ELASTO-PLASTIC FINITE ELEMENT
ANALYSIS

MAKOTO KIMURAi) and FENG ZHANoii)

ABSTRACT
In this paper, three methods for seismic evaluation, namely, a separated method using a dynamic analysis on a sim-
plified sway-rocking model (S-R model), a responding displacement method, and a dynamic analysis on a full system,
are investigated by conducting a series of static and dynamic three-dimensional elasto-plastic finite element analyses.
At first, the validity of a newly developed finite element analytical code DGPILE-3D, which is used throughout this
paper, is checked through a dynamic model test on a pile foundation. Then, an elevated railway bridge with a pile
foundation is investigated with the separated method and the responding displacement analysis, which are often used
in seismic design, and a three-dimensional elasto-plastic dynamic analysis on a full system. Two types of ground are
considered in order to understand the influence of the soil on the mechanical behaviors of upper structures and pile
foundations. From these analyses, it is possible to understand the difference of the mechanical behaviors of pile foun-
dation predicted by different methods. Based on the results obtained in this paper, fundamental knowledge needed for
developing an applicable and easily conducted seismic design method with three-dimensional finite element analysis, is
achieved.

Key words: dynamic, elasto-plasticity, pile group, responding displacement method, seismic evaluation, separated
method, three-dimensional finite element analysis (IGC: E4)

three-dimensional dynamic effective-stress analysis of a


INTRODUCTION light structure buried in saturated sandy ground. Taji et
After the Hyogoken-Nambu earthquake, the design al. (1997) conducted a large-scale shaking-table test and a
codes for highway and railway bridges were revised to centrifuge model test to investigate the soil-pile-structure
meet the needs of much higher levels of strength for struc- interaction in potentially liquefying sand. Fukutake
tures (Design Codes of Japan Highway Bridge, 1996; De- (1997) conducted a seismic evaluation of a group-pile
sign Codes for Foundations and Earth-Retaining Struc- foundation surrounded by a frame wall using a 3-D non-
tures of Japan Railway, 1997). In the revised railway linear dynamic analysis. Wakai et al. (1997) conducted a
code, a nonlinear dynamic analysis is strongly recom- seismic analysis of a bridge-ground system using a 3-D
mended for the design of bridges. It is commonly known, nonlinear dynamic analysis. Taguchi et al. (1997) con-
however, that the seismic behavior of a bridge is not only ducted a numerical simulation of the soil-foundation in-
related to the upper structure of the bridge, but also to teraction with a 3-D nonlinear dynamic analysis consider-
the foundation of the bridge and to the. ground upon ing subsoil liquefaction.
which the structure is built. It is recommended, there- It is known that in order to simulate the mechanical be-
fore, that a full system, which consists of upper struc- havior of soils under generalized stress condition, a
ture, foundation and ground, be considered in the sophisticated constitutive model is necessary, which
dynamic analysis because of the merit that relatively few often means that the determination of a few parameters
assumptions are adopted. based on laboratory tests and in situ measurements is
Dealing with the full system in a dynamic analysis is needed. For this reason, a three-dimensional nonlinear
usually thought to be difficult when the nonlinearity of dynamic analysis based on a sophisticated model cannot
both the structure and the ground must be considered. A be commonly used in the seismic design of pile founda-
few studies have been done in this field through both ex- tions.
periments and numerical analyses. Matsuda et al. (1994) It is obviously that to describe the general behavior of
performed a shaking table model test and conducted a soils with a four-parameter model is too ambitious. For
i) Associate Professor, Dept. of Civil Engineering, Kyoto University, Kyoto 606-8317.
ii) Associate Professor, Dept. of Civil Engineering, Gifu University, Gifu 501-1193.
Manuscript was received for review on September 16, 1999.
Written discussions on this paper should be submitted before November 1, 2001 to the Japanese Geotechnical Society, Sugayama Bldg. 4F,
Kanda Awaji-cho 2-23, Chiyoda-ku, Tokyo 101-0063, Japan. Upon request the closing date may be extended one month.

113

This is an Open Access article under the CC-BY-NC-ND license.


NII-Electronic Library Service
114 KIMURA AND ZHANG

Seismic evaluation of group-pile foundation

method Responding displacement method Dynamic analysis on


1. Determining the interaction between pile 1. Calculating the maximum displacement at a full
foundation and ground with equivalent springs ground surface from an earthquake 3-D dynamic FEM is conducted
by static 3-D FEM vibration analysis of ground only directly on a structure-
2. Simplifying a full system to an S-R model 2. Assuming a distribution of the deformation foundation-ground system with
3. Conducting a dynamic analysis on ·the S-R along the vertical direction in ground relatively few simplifications.
model and calculating the maximum forces in 3. Applying the above assumed deformation
the equivalent springs on the ground to evaluate the sectional
4. Using the maximum forces of the springs as forces of piles
external forces, a static 3-D FEM is again
conducted on the pile foundation to calculate
the sectional forces ofthe piles
Comparing the mechanical behavior of upper
structure (pier) and piles
• Dynamic analysis Understanding the difference of the mechanical
on S-Rmodel behaviors predicted by different methods
• Static analysis of
pile group on finite Developing an applicable and easily conducted
element model seismic design method based on 3-D FEM

Verification of the code


through a model test under
....oPiiE~3fi ... ·:
elastic condition
(Murono et al., 1997) •C •si;ti~ ~r
load
··· 3:n· FEM ·;o~·siJ;ri~g ·c;,~ii~ Tri-linear model for :
• piles in monotonous :
2. Elasto-plastic isotropic hardening model for soils : load, considering the :
3. Tri-linear model for piles and pier considering : influence of axial :
the hysteresis ofM-<1> relation :force on M-<P relation :
4. Simulation of pile with hybrid element
~ ••.•...•.....•.... ~

considering pile-volume influence


5. Rayleigh type of attenuation
••6••• J?!r;.e~! !~t~~~~t!~~ ~~f!t~~ .~f.~~~.P:~.k;~••••••••

--••~ Flow chart of research


.................. Type of andalysis
Jill""'"" use
~ ........
......... Features ofthe FEM codes

Fig. 1. Flow chart of research in this paper and brief introduction to DGPILE-3D

this reason, the numerical analyses in this paper are con- foundation-ground system.
ducted under a total stress condition, which means that a The first purpose of the paper, is to provide a relatively
potentially liquefying ground is beyond the scope of this simple numerical method based on three-dimensional
paper. For a potentially liquefying ground, a soil-water finite element analysis, in which the nonlinearity of a
coupling analysis is necessary. ground is described with a simple elasto-plastic model
Kimura et al. (1991) developed a three-dimensional and its four parameters can be determined based on the
finite element analysis program, GPILE-3D (Ground Pile N value of standard penetration test (SPT}, without los-
Interaction of Laterally Loading Effect), in which the ing too much accuracy of the description for soils.
stress-strain relation of the ground is described by an The second purpose of the paper is to clarify the fea-
elasto-plastic model with an associated flow rule of tures, the merits and demerits of the three methods
Drucker-Prager's yielding criteria. Various improve- among which the concepts of the separated method and
ments were later added to the program (Adachi et al., the responding displacement method are commonly used
1994; Kimura and Zhang, 1997). GPILE-3D is developed in seismic design.
for static analysis and is limited to a monotonous load- Figure 1 shows a flow chart of the research in this
ing. Its main feature is that the nonlinearity of pile is de- paper and a brief introduction to the code DGPILE-3D.
scribed by a tri-linear model, considering the influence of Detailed explanation of the figure will be given in the fol-
axial force on the moment-curvature relation. By develop- lowing sections.
ing the three-dimensional static and dynamic finite ele-
ment analytical code DGPILE-3D (Kimura and Zhang,
1999), the authors try to discover whether or not it is pos- NONLINEAR BEHAVIOR OF GROUND AND PILES
sible and/ or practical to conduct a three-dimensional Nonlinearity of Ground
nonlinear finite element dynamic analysis on a structure- It is known that in the case of Drucker-Prager's associ-

NII-Electronic Library Service


SEISMIC EVALUATION OF PILE FOUNDATION 115

ated flow rule, the plastic potential is the same as the


yielding function. Therefore, the plastic volumetric
strain will always be negative, which means that swelling
will always occur during shearing. The flow rule of the
Drucker-Prager model is

(1)

where jy is the yielding function, fp is the plastic potential


and A is a positive parameter.
The yielding function is expressed as follows:
(2)
- - - ~=0.018
where ]z is the second invariant of the deviatoric stress ········- ~=0.0018 (a=0.018)
tensor and am is the mean stress. In addition, a and Ks are ,.-..
~
0 ~=0.0
material constants which can be determined from cJ>, the '-'

internal frictional angle of the soil, and c, the cohesion of ·§


.l:l 0.8
the soil, as rl)
C,.)
0.6
2 sin cJ> 6c cos cJ>
(3)
"€
-13(3 +sine/>)' Ks .f3(3 +sin cJ>) ' ] 0.4
0
The plastic potential in the theory is expressed as > 0.2
.~
......
fp=(Jz) 112 -3fiam-Ks=O (4) ~ 0_3 0 3
0:: Shear strain (%)
where p is the dilatancy parameter which can be com-
(b) Dilatancy
puted with Eq. (3) by substituting the frictional angle
with dilatant angle p. If P=O, it means no dilatancy. If Fig. 2. Stress-strain and stress-dilatancy relations in hollow cylinder
P=a, it means an associated flow rule. test with Drucker-Prager model (Confining and axial pressures are
In Eq. (3), a and Ks are evaluated based on the condi- kept constant.)
tion that the Drucker-Prager yielding surface coincides
with the Mohr-Coulomb yielding surface in triaxial exten-
sion. In triaxial compression, this may result in an un- and a half cycles of torsional shearing strain while the
derestimation of the strength of soils. Considering the confining and axial stresses are kept constant. It is found
fact that most of the elements around a pile group are that the stress-strain relation remains the same during
not in purely triaxial compression, the way of fitting the each cycle for different P values (fi=O.O, P=0.0018,
Drucker-Prager yielding surface with the Mohr- P=0.018, a=0.018). The plastic volumetric strain in Fig.
Coulomb yielding surface at extension is preferred. 2(b) is different, however, if p takes a different value. The
The loading and unloading conditions are expressed as larger the P value is, the larger the plastic volumetric
strain will be.
jy=O, dfy>O: loading If the simulation is conducted under the same condi-
jy=O, d/y=O: neutral (5) tions but with the volumetric strain restricted to zero, it is
jy=O, d/y<O: unloading. found from Fig. 3 that in the case of the associated flow
rule, the shear strength increases during each loading cy-
The incremental stress-strain relation can then be given cle because of the increase in mean stress. If P=O (non-as-
as sociated flow rule), the shear strength is kept constant.
dau=Dijkt" dekz, In soft clay a positive dilatancy usually does not occur.
In the case of sandy ground, the dilatancy is dependent
afp aJy
e -
D ijqr - -aD emnkl
aUqr on the density of the sand. For loose sand, negative
Umn dilatancy usually occurs. In the seismic evaluation of pile
(6)
a/y aJP foundation, most concern is on soft ground. It is, there-
--D'Jnnqr-
aamn aaqr fore, reasonable to suggest that when a boundary value
problem related to a pile foundation in soft ground is con-
D~jkt=A.~u~kt+ G(~ik~jt+~u~jk),
sidered, the condition that P is equal to or less than zero
Ev E is preferred. For simplicity, the condition of P=O is as-
A. , G (7)
(1 + v)(1- 2v) 2(1 + v) sumed throughout this paper.
Figure 2 shows a theoretical simulation of stress-strain The advantage of the model is that it is very simple and
relation in a hollow cylindrical torsional test on sand. In only four parameters need to be determined. In the de-
the simulation, the cyclic shearing load is applied by two sign stage, 1apanese engineers prefer to determine these
parameters with N, the value of SPT, instead of with

NII-Electronic Library Service


116 KIMURA AND ZHANG

(a) (~=0.018, a=0.018)


0. 6 I""T""ri""T'"T""T""T"'"I..,...,...T""rT"i""T'"T""T""T"'"I""T""T""T""rT"i""T'"T"'T""T""1

0.4
0.2
0
-0.2
-0.4
-0.6 L..J...I...I...I....L....I...I...J....J...J....J...J...J....L...J;;;L..I.J...I....J...J....J...J...J..J...J....J...J....I..J

-3 -2 -1 0 2 3

_....
t'\S

~
...._,
0.4
U)
0.2 Fig. 4. Possible stress paths on failure line of Drucker-Prager model
U)
(1.)
tl 0
U)

~ -0.2
(1.)
,..d -0.4
r:/)

(c) (~=0.0, a=0.018)


0 .6 r-r-T"T""T""I""T""T'""T""T'""''...,...,....,...,....,........,...I""T""T'""T""T'""''...,...,...,...,....,...,....,

0.4
0.2
0
··········
-0.2
-0.4 Fig. 5. Tri-linear model for piles and pier
-0.6- 3~..U...:.::-L-L-L...L...I:-J...J-j,.~.J..J....y..J-.I...J...J..~....L...1.:!

modified model is called a modified Drucker-Prager


Fig. 3. Stress-strain relation in hollow cylinder test with Drucker- model. It should be kept in mind that the Drucker-Prager
Prager model (Volumetric strain is restricted to zero.) model is an elasto-perfect plastic model and its Young's
modulus is independent of confining stress am.

laboratory tests. In this case, the following empirical for- Nonlinearity of Piles
mulae are recommended (Design Codes for Foundations In GPILE-3D, only monotonous loading is considered
and Earth-Retaining Structures of Japan Railway, 1997): (Adachi et al., 1994). Therefore, the nonlinearity of the
relation between the bending moment and the curvature
c=qu/2=(1/160)N(MPa) for clay (Terzaghi-Peck) (8)
of the pile (M- rf> relation) is simulated by a trilinear
~ )
06
<t>= 1.85 ( , I . +26 ( o) for sand, model in which the influence of axial force on M- rp rela-
av ao+ 0 .7 tion is considered. In DGPILE-3D, however, cyclic load-
ac)=0.1 MPa (9) ings should be considered. In cyclic loading condition, it
is difficult to give a proper description of the influence of
E= 1.4 to 2.8N (MPa) (10)
axial force on the M- rf> relation. For this reason, the
The disadvantage of the model is that when a stress M- rf> relation of pile is simulated here by a kinematic
state reaches the failure line, it will move along the line if hardening trilinear model, taking into consideration the
the stress increment is judged as a loading, as shown in hysteresis of loading and unloading, as shown in Fig. 5.
Fig. 4. In the case of normally consolidated clay under The influence of axial force on M- rp relation, is not con-
drained and undrained conditions, or in the case of sand sidered, as explained in Figure 1. In the M- rp relation,
under a drained condition, however, it is known that the Me, My, and Mu represent the cracking, yielding and ulti-
shear strength of the soil is kept almost constant on the mate moments, while rf>c, rf>y, and rf>u represent the corre-
failure line. For this reason, a simple modification is sponding curvatures. In this paper, these values are deter-
added to the model in such a way that, when a stress state mined under the condition of static axial force.
is at the failure line and its stress increment is judged as a In simulating a pile, the normal way in the finite ele-
loading according to Eq. (5), then an adjustment is made ment analysis is to use a beam element that is known to
to keep the stress state stationary along the failure line by be a non-volume element. In the finite element analysis
setting the Young's modulus in Eq. (7) at a very small of a pile group, the pile-volume influence cannot be ig-
value while Eq. (6) remains valid. In this paper, the nored. If the area and volume of piles are not considered

NII-Electronic Library Service


SEISMIC EVALUATION OF PILE FOUNDATION 117

Hybrid element Separated Method Using a Dynamic Analysis on S-R


Pile
Solid element Model
(El)solid (EA)solid There are a few methods for evaluating the equivalent
springs that can represent the interaction between a pile
Beam element foundation and a ground. Kimura and Zhang (1997)
(El)beam (EA)beam proposed a separated method, in which a static elasto-
plastic three-dimensional finite element analysis (GPILE-
3D) is conducted to evaluate the equivalent springs, con-
sidering the nonlinearity of both the ground and the
piles. The stiffness matrix of the 3-dimensional equiva-
lent springs is given in Eq. (11) in which kx, ky, kz, k 8x,
key, kez, represent the springs in six directions. Here kyex
EI=(EI)solid+(EI)beam
and kxey are associated springs which can be evaluated
EA=(EA)solid+(EA)beam
along with springs kx and ky respectively.
Fig. 6. Hybrid element and its mechanism
Hx kx 0 0 0 kxey 0 <5x
Hy 0 ky 0 kyex 0 0 <5y
properly, it is almost impossible to simulate well the inter- Hz 0 0 kz 0 0 0 <>z
action between the piles and the ground that is heavily de- (11)
Mex 0 kyex 0 kex 0 0 Ox
pendent on the geometry of the piles. For this reason, a
hybrid element that consists of a beam element and a few Mey kxey 0 0 0 kez 0 Oy
solid elements is introduced to simulate the pile, as Mez 0 0 0 0 0 kez (}z
shown in Fig. 6.
The stiffness of the pile is shared by the beam element The feature of the method, which is different from the
and several solid elements in such a way that the bending usual ways used in the seismic evaluation of pile founda-
stiffness of pile EI is equal to the sum of the bending stiff- tions, is that in calculating these equivalent springs, the
ness of beam element (EI)beam and solid elements (EI)solid· efficiency of a pile group can be properly evaluated in an
The sharing ratio between the stiffness of beam element arbitrary arrangement because of the capability of
and the solid should be determined in such a way that the GPILE-3D and DGPILE-3D to take into consideration
mechanical behavior of a cantilever beam simulated by a such factors as the pile-volume influence, and the non-
single beam element and the hybrid element will be the linearity of soils and piles. Three-dimensional calculation
same. It is found that the higher the sharing ratio is, the needs to be done six times in order to evaluate these
closer the mechanical behavior between a single beam ele- springs. In the calculation, the load can be applied by
ment and the hybrid element will be. On the other hand either prescribed loads or prescribed displacements. In
considering the main purpose of introducing the hybrid the present study, prescribed load is used in the calcula-
element, the solid elements within the hybrid element tion.
should be 'hard' enough to represent the pile-size In the separated method, a full system is firstly sim-
influence. Therefore, the stiffness of the solid elements plified to an S-R model (a frame-spring model) by
should be kept to a certain level. Based on the above dis- representing the interaction between a ground and a foun-
cussion, the sharing ratio between the stiffness of the dation with the equivalent springs. A dynamic analysis is
beam element and the solid elements is selected as 9 to 1. then conducted only on the S-R model, which is much
If we keep in mind that the stiffness of concrete is in the easier if compared with the dynamic analysis on a full sys-
order of 104 MPa and the stiffness of soft soil is in the or- tem. From the dynamic analysis on the S-R model, it is
der of 10 MPa, the stiffness of the solid elements within possible to evaluate the maximum forces acting on the
the hybrid element is stiff enough to represent the pile- equivalent springs. Finally, using the maximum forces of
size influence. In the case of a nonlinear M -l/> relation, the springs obtained from the dynamic analysis, a static
the Young's modulus of the beam element is determined analysis is again conducted on the pile foundation to cal-
by the kinematic hardening trilinear model shown in Fig. culate the sectional forces of the piles. In this method,
5 to which the Young's modulus of the solid elements is the forces of the piles caused by the deformation of the
linked. ground are not considered.

Responding Displacement Method


THREE METHODS OF SEISMIC EVALUATIONS The responding displacement method is a method in
FOR PILE FOUNDATIONS which the main force of an earthquake is regarded as the
In the following, the three methods of seismic evalua- deformation of the ground. In this case, the sectional
tion for pile foundations shown in Fig. 1 are explained in forces of piles can be evaluated with a static finite ele-
detail. ment analysis in which a prescribed displacement, usu-
ally evaluated from an earthquake vibration analysis of
the ground only, is applied to the ground. The inertial

NII-Electronic Library Service


118 KIMURA AND ZHANG

force from the upper structure is disregarded. In this of a full system is badly needed to understand the dynam-
case, the position where a prescribed displacement is ap- ic interaction between a ground and a pile foundation.
plied may affect the behavior of the pile foundation. The advantage of the dynamic analysis of a full system
There are two ways in regard to the application of the based on three-dimensional FEM over the first two
prescribed displacement in the responding displacement methods is out of question. The problems are, (1) how to
method based on FEM. One is that the prescribed dis- properly evaluate the nonlinear behavior of the ground
placements in all nodes are firstly transformed to nodal and the structure, which may greatly affect the mechani-
forces based on an elastic finite element analysis, and cal behavior of the pile foundation subjected to cyclic
then these nodal forces are applied at all nodes. The lateral loading up to the ultimate state; (2) how to make
other is that the prescribed displacements are applied di- it easy and applicable to conduct the dynamic analysis on
rectly in some places, but not at all nodes. Each method a full system in the seismic evaluation of a pile founda-
has its advantages and disadvantages. The former has the tion for highway and railway bridges.
advantage of being independent of loading position but
is considered under elastic condition. The latter has the
disadvantage of being dependent on loading position but NUMERICAL SIMULATION OF A DYNAMIC
is considered under plastic condition. In this paper, the MODEL TEST ON A FULL SYSTEM
latter is preferred because everything should be consid- Murano et al. (1997) conducted a dynamic model test
ered under plastic condition in a major earthquake. on a structure-foundation-ground system to investigate
The advantage of the first two methods is that finite ele- the seismic behavior of group-pile foundation. The
ment analyses related to a ground can be conducted un- material of the model ground is silicon gum, and the
der static condition, which is relatively easy and can save material of both the piles and the pier is teflon. Figure 7
a lot of calculating time. The disadvantages of the shows the overall view of the model ground, the pier, the
methods are that, (1) the dynamic interaction between a pile foundation and the layout of the measurement. Ta-
pile foundation and a ground is evaluated statically and ble 1 shows the parameters of the model ground and the
(2) it is very difficult to evaluate the phase difference be- structures. The pile foundation consists of four piles,
tween the inertial force from an upper structure and the 51.5 em in length and 2 em in diameter. The pier also con-
deformation of a ground during an earthquake. Murano sists of four individual thin rectangular bars that are
et al. (1997) discussed the phase difference experimentally fixed together at the top and the bottom. The eigen
under elastic condition. As a matter of fact, if the stress- periods of the model ground and upper structure, ob-
strain relation of soils exceeds elastic region, it is almost tained from the model test, are 0.192 and 0.101 seconds,
impossible to evaluate the phase difference without con- respectively. In the test, acceleration meters were buried
ducting a dynamic analysis. Therefore, dynamic analysis in different layers of the model ground, as shown in Fig.

Plane view

(a) Overall view of ground, pile foundation, upper (b) Sectional views of upper structure
structure and the layout of measurement and pile foundation

Fig. 7. Dynamic model test on structure-foundation-ground system (Murono et al., 1997)

NII-Electronic Library Service


SEISMIC EVALUATION OF PILE FOUNDATION 119

Table 1. Parameters of model ground and structures Figure 10 shows a comparison of the responding ac-
celerations, obtained from the numerical analysis and the
Moment of
Material
E
inertial
Area Height Density Poisson's model test, at the points where the acceleration meters
(MPa) 4 (cm2) (em) (g/cm3) ratio are installed, as shown in Fig. 7. It is clear that the maxi-
(cm )
mum accelerations as well as the shapes of the wave are
ground silicon 0.3525 - - 51.5 1.0 0.49
almost the same for the numerical analysis and the test.
pile teflon 578.43 0.7854 3.142 51.5 2.26 0.20 This phenomenon can also be verified from Fig. 11, in
pier teflon 578.43 1.333 4.000 13.6 2.26 0.20 which a comparison of the responding acceleration spec-
trum obtained from the numerical analysis and the test is
given. Except in the area in which the period is less than
0.3 seconds and the numerical results are a little larger
than the test results, the two spectrums are the same. The
accuracy of the numerical analysis is proven to be rather
high.

J_::J::: :
-2000 5 10 15 20
Figure 12 shows a comparison of the transfer function
in the surface of the ground and at the top of the pier ob-
tained from the numerical analysis and the test. In the sur-
Time (sec) face of the ground, the calculated resonance frequency
agrees well with the tested response, while the calculated
Fig. 8. Input wave
transfer function is a little larger than the tested one. At
the top of the pier, there are two resonance frequencies.
One is inherent from the ground and the other is from
the upper structure. The calculated transfer functions are
larger than the tested ones in a lower frequency area but
smaller than the tested ones in a higher frequency area.
The calculated transfer function decreases to zero in a
very high frequency area. From the above discussion,
however, it can be concluded that DGPILE-3D can simu-
late the mechanical behavior of the structure-founda-
tion-ground system well in a dynamic test.
Node: 1540
Element: 1152

SEISMIC EVALUATIONS OF A PILE


Fig. 9. Finite element mesh used for simulation of dynamic model test FOUNDATION
An elevated railway bridge with a group-pile founda-
tion is considered in a seismic evaluation. The bridge,
7. The input wave, with a maximum acceleration of which is designed for a common light-rail transportation
amax= 114 gal, is shown in Fig. 8. As the model ground system in urban area is supported by a group-pile founda-
and the structure are made of typical elastic materials, tion made of 2 x 2 cast-in-place reinforced concrete piles,
the mechanical behavior of the model ground and the 1.2 meters in diameter (D) and 20 meters in length, as
structure were observed to be elastic in the dynamic shown in Fig. 13. The distance between the centers of the
model test. two piles is 2.5D, that is, 3 meters. For simplicity, the
A three-dimensional dynamic analysis is conducted in round-shaped piles are simulated with square-shaped ele-
this paper to simulate the model test, by which the validi- ments in the finite element analyses, making sure that the
ty of the code DGPILE-3D can be confirmed. Figure 9 bending stiffness EI of the representative hybrid element
shows the finite element mesh used for the simulation of is equal to the original one. Figure 13 also shows the
the dynamic model test. Because of the symmetric condi- finite element mesh that consists of 2680 nodes and 2200
tion of the geometry and the load, only half of the 8-node isoparametric solid elements, with slide bounda-
domain is taken under consideration. The boundary con- ries on the side surfaces and fixed boundaries at the bot-
ditions are the same as the model test, that is, free at the tom.
two sides whose normal direction is parallel to x-axis, slid- Two types of ground, shown in Fig. 14, are considered
ing at the two sides whose normal direction is parallel to in order to investigate the influence of the ground materi-
y-axis and fixed at the bottom. The mesh consists of 1540 als. The two grounds are each composed of three layers.
nodes and 1152 8-node isoparametric solid elements. In The first two layers are alluvial and the third layer is
the dynamic analysis, a Rayleigh type of attenuation is diluvial. The pile group is laid on this third layer. The
adopted and the attenuation constants of the ground, the thicknesses of the first two layers are 4 and 16m, respec-
piles and the pier are assumed as 5%. The direct integra- tively. In Ground 1, the 1st layer is silt clay and the 2nd
tion method of Newmark-P is used in the dynamic analy- layer is sand. In Ground 2, the 1st layer is sand and the
sis and the time interval is 0.01 seconds. 2nd layer is a very soft clay. The bottom layers of the two

NII-Electronic Library Service


120 KIMURA AND ZHANG

40
20

-20
-40
20
10

-10

-20
20
10

,.-...
~-10
OJ) 1-100
'-' '-'

§ -20 § -200
·-g 20 ·~ 20
t Q)
a)
] 10 C) 10
~
C)
<r;

-10 -10

-20
20
10

-10
-20
20

Time (sec)
(a) Calculation (b) Experiment

Fig. 10. Comparison of responded acceleration obtained from calculation and experiment

grounds are the same. They are supposed to be an elastic tributes from -4 m to -20m, however, the value of c
material in the numerical analyses. For simplicity, the determined from Eq. (8) would be so small that the soil
material properties of the grounds are determined in such would fail at the k 0 condition. Considering the empirical
a way that an average value is assumed for each soil lay- relation CuI p' for soft clay, the average value of the layer
er, as shown in Table 2. In determining the parameters c is assumed as 0.044 MPa for the whole layer. Similarly,
and¢, Equations 8 and 9 are usually used. In the case of an average value of ¢ in the second layer of Ground 1 is
Ground 2 where a soft clay layer with theN value of 2 dis- assumed. The material properties of the piles and the pier

NII-Electronic Library Service


SEISMIC EVALUATION OF PILE FOUNDATION 121

experiment
- - - calculation

..-15
Q

~10
10
~ 5

Period (sec)
5 10 15 20 25 30
Frequency (Hz)
(a) Ground surface

1
Period (sec)

5 10 15 20 25 30
Frequency (Hz)
Period (sec)
(b) Top of pier

Fig. 11. Comparison of responded acceleration spectrum obtained Fig. 12. Comparison of transfer function obtained from experiment
from experiment and calculation and calculation

Upper structure
(m=300 ton) Pier (H=7 m)

1st layer

2nd layer

3rd layer

(a) A full system of structure-foundation-ground Node: 2680


Element: 2200

(b) Footing

Fig. 13. Geometry of toll bridge with group-pile foundation investigated in seismic evaluation

NII-Electronic Library Service


122 KIMURA AND ZHANG

-~- load-displacement curve


s Silty clay
N=8
s
"¢ Isand
N=8
- - o - - equivalent spring

Clay
s s
~""g 10
Sand N=2
\0
....... N=20 ~

0
N=50 .....:l
s N=50
.......
elastic s
......., el~tic
"'
......
t
"'' -- t
'
20 40
(a) Ground 1 (b) Ground 2 Displacement (em)

Fig. 14. Geologic profiles of model grounds

(b) Ground 2
Table 2. Material properties of grounds p

Layer
Thickness
Soil
Density
N
E
v
c cf>
~ 10
""g
y3

(m) (g/ cm3 ) (MPa) (MPa) (0) 0


.....:l
Ground 1 4 silty 1.60 8 20.0 0.40 0.044 4
clay
1 2 16 sand 1.80 20 50.0 0.33 0.0 30 20 40 60
Displacement (em)
3 1 elastic 2.00 50 100.0 0.30 - -

Ground 1 4 sand 1.80 8 20.0 0.33 0.0 30 Fig. 15. Load-displacement curve and skeleton curve of horizontal
equivalent springs (kx)
2 2 16 clay 1.60 2 5.0 0.40 0.044 4
3 1 elastic 2.00 50 100.0 0.30 - -
Table 5. Nonlinear properties of equivalent horizontal springs
(kx=ky)
Table 3. Material properties of concrete and reinforcement
Py 1 (MN) Py2 (MN) Py3 (MN) r5y 1 (em) r5y2 (em) r5y3 (em)
Strength of Strength of Ground 1 10.4 12.8 16.4 3.93 9.54 58.1
Reinforcement
concrete reinforcement
Ground 2 10.0 16.0 20.0 2.49 5.02 53.8
Pile 24.0 (MPa) 300.0 (MPa) D29-24

Pier 24.0 (MPa) 400.0 (MPa) D35-44


Table 6. Values of elastic equivalent springs
Young's modulus of concrete=2.5 x 104 (MPa)
kz kox key
(MN*m/m) (MN*m/rad) (MN*m/rad)
Table 4. Nonlinear parameters of pier and piles
Ground 1 1724.1 9857.6 9857.6
Me My Mu
C/Jc (1/m) C/JY (1/m) cpu (1/m) Ground 2 1052.6 5521.5 5521.5
(kN*m) (kN*m) (kN*m)

Pile 702.2 1591.0 2230.2 0.000276 0.00233 0.0110

Pier 2591.0 8799.0 11867.0 0.000201 0.00189 0.00629 lation is that the head of the pile is fixed with the footing
and the toe of the pile is free. It is clear from this figure
that the interaction between the group-pile foundation
are shown in Tables 3 and 4. The initial stress condition and the ground shows a nonlinear behavior. It is also
of the grounds is assumed to be a gravitational stress field found that there is a strong hysteresis in the load-displace-
without the existence of piles. ment relation. Therefore, a trilinear model that considers
the hysteresis is assumed for the horizontal equivalent
Evaluation with Separated Method spring in which the skeleton curve is taken as the load-dis-
Figure 15 shows the horizontal load-displacement rela- placement relation in the loading process, as shown in
tions of the group-pile foundation under conditions Fig. 15. Table 5 shows kx, the stiffness of the horizontal
where the bending moment acting on the footing is zero equivalent springs in the case of Ground 1 and Ground 2.
and only the load along the x-direction is applied. Be- In Fig. 15 and Table 5, Pyt, Pyz, Py3, ~yt, ~y2 and ~y3,
cause the influence of the associated springs kxoy and kyox represent the values of the force and the displacements at
is not important in this case, their values are assumed to three points of the tri-linear skeleton curve, that is, the
be zero. The boundary condition of the piles in the calcu- first yielding point, the second yielding point and the ulti-

NII-Electronic Library Service


SEISMIC EVALUATION OF PILE FOUNDATION 123

mate point. the same as that shown in Fig. 13. Here, an equal-dis-
Because of the symmetric condition, the horizontal placement-boundary condition is used between the two
equivalent springs kx and ky are the same. The rotational side boundaries whose normal direction is parallel to the
equivalent springs kox and koy are also the same. The tor- x-axis to simulate the infinite boundary in real situation.
sional spring koz is not considered here because of the In this case, the group-pile foundation and the upper
symmetric condition. The other springs except the structure are not included in the mesh. Figure 16 shows
horizontal springs behave elastically and their values are the input earthquake wave used in the analysis, which is
listed in Table 6. an artificial earthquake wave specified for the seismic de-
In seismic engineering design, when a dynamic analysis sign of railway bridges in Japan. It has a maximum ac-
is conducted on an S-R model, the input earthquake celeration of 722 gal in a horizontal direction. In the anal-
wave should be a surface wave. However, in the case of a ysis, the attenuation constant of the ground is assumed
full system, the input earthquake wave is a base wave, as 10%. Figure 17 shows the surface waves obtained
which is applied at the boundary between the surface-lay- from the analysis of the two grounds. It is clear that the
er ground and the base ground. In this paper, the same maximum surface acceleration of Ground 1 is much
way is used. That is, the surface wave is calculated from larger than that of Ground 2, while the maximum dis-
the earthquake vibration analysis of the ground in which placement of Ground 1 is much smaller than that of
the input base wave is the same as the wave used in the Ground 2.
analysis on a full system. Therefore it is possible to com- A dynamic analysis on an S-R model can then be done
pare the results obtained from different methods in the under the conditions that the attenuation constants of
same condition. the ground and the equivalent springs are 5% and 10%,
Based on the above discussion and the values of the respectively. The time interval of the integration in the dy-
equivalent springs, a dynamic analysis on the simplified namic analysis is 0.01 sec. Figure 18 shows the time
S-R model is conducted. At first, an earthquake vibra- responses and the maximum forces acting on the horizon-
tion analysis of the ground is conducted to calculate the tal spring (/max) and the rotating spring (Mmax). Using the
surface wave. A direct integration method of Newmark-P maximum forces obtained from the above analyses as the
is adopted and the finite element mesh for the ground is external forces, it is possible to calculate the sectional
forces of the piles with a three-dimensional elasto-plastic
soo~~~~~~~~~~~~~~~
finite element analysis on the pile foundation. The calcu-
~ lated results will be shown later in order to compare them
!:9 400 with the results obtained from other methods.
1::
0
·~ 0 6-Jv.WW~/MIPIUAIIT '\1\A,i/U
~ Evaluation with Responding Displacement Method
8 -400
(..)
In the separated method, the earthquake force is not
<_goo~~~~~~~~~~~~~~~~ sufficient because it does not consider the influence of the
0 5 10 15 deformation of the ground in which a pile foundation is
Time (sec)
buried during an earthquake. Here, a responding dis-
Fig. 16. Input wave placement analysis is conducted with DGPILE-3D. The

,....._ a = -669 gal (t=9.21sec) ,....._ d =-27.7cm (time=8.81sec)


1 800 .,.,...,..,....,."'T"T..,....,.."'""T"'1r-T""T-Tm~ax~"'""T"'1r-T""T"'T"T..,....,..'T""'I"""T""'""'T""''""'T'..,...,..'T""'I"""~'""'~""~ ~ 50 .,.,...,r-T""T"'T"T-r-r-"'""T"'1~-rm""T"ax-r-r-,...,...,,...,..-r"TT-r-r-T"T"r-r-r"TTi"''T"T"r-r-:1
~ 400 "S 25
0 ~
·.g
~
0 s
~
0
C) -400 Jg -25
<u
-800
0 5 10 15
Ground 1
~
&;o.l..,j,.J....I..I....l...J...J....I..............I....I...i............l.-j,j.....I....J.....J..,j,.,.....I....J.....J..,j,.,.....I....J.....I.-j,j............

20
o..
.~ -50
° ....~.-~...I,....L...oi............
I;;;,J,j,.................

0 5
Ground 1
""-"-J...........I....I...I.......................l...l...lo.....................l.-l.................

10 15 20
Time (sec) Time (sec)
252 1 ( 5 86 ) d = 42.2 em (time=8.36sec)
i 800~~~~~~am~ax~=~-~~g~a~t~=~·~se~c~~ ~ 50~~~~~~m~ax~~~~~~~~~~
§ 400 "S 25
·~ 0 ~ 0
C) -400 ~ -25
u Ground 2 o.. Ground 2
....._._.~..~....~...~............~....~...~...........~ .~ -5o~~~~~.....l....l...~~~~~wu~~~~
<-80oi;;;,J,J,..........
0
l...l...lo.J....I..I....l...J...J....I................._._...........

5 10
Time (sec)
15 20 ° 0 5 10
Time (sec)
15 20

(a) Acceleration (b) Displacement

Fig. 17. Responses of acceleration and displacement at ground surface

NII-Electronic Library Service


124 KIMURA AND ZHANG

S 15 r="T""T"~""T'""r-T"'T'"M,....,-lfl'¥-=I"'T1"'T'"4"r"T.9-rr2..,-MNrT'-r-T""T*m.. ,--r""lt(r-1'"=..,_.9.I"'T37"'T'"s"r"Tec_,. ).,. .)r-T"'::

~~
0
Q)

s
~-15 ~~~~~~~~~~~~~~~
5 10 15 20 0 5 10 15 20
Time (sec) Time (sec)
f =1.56 MN (t=5.92sec))
3 ~~~~~m~~~~~~~~~~~~~~~ s 15 M =7.801 MN*m (t=5.95sec))
r:-r-r-r-r-r-r-T""T""T-r-r'-'+"1'-T""T""T-r-T"'"T'""T""T""T"""""'1"""T"'"T'""'r.,...,....,""'1"""T"'"T'""'r"T""T"""1....,....;'::1

~0
Q) ~ 0
~ ~
0
~ s
~ -15 o~.J...J....J....J....J.....L...J.....J-I....I...I...J...J....J....J....J.....L1~o~..J......I...JL-I-I...~~~-I.....I...:l
5 10 20
Time (sec) Time (sec)
(a) Forces of horizontal springs (b) Moments of rotating springs
Fig. 18. Responses of spring forces

Moment (MN*m)
o~.n"n~~~~~~~~~~~~

5 5

--o.6m
-cr-I.9m
gw --o.6m
~~~~~~----t -cr-I.9m
5 ..9m
--¢----
0 15 5.9m
--¢----

20 20

-2 0 2 -2 0 2 4
(a) Ground 1 (b) Ground2

Fig. 19. Distribution of moments of pile obtained from responding displacement method

deformation of the ground along a vertical direction is as- the moments by the responding displacement method at
sumed to be a cosine distribution, while the maximum dis- different loading distances. In the case of Ground 1, the
placement at the ground surface is 27.7 em in the case of distribution of the moments does not change too much
Ground 1 and 42.2 em in the case of Ground 2, which is due to the difference in loading distance. Because
evaluated from the dynamic analysis on a ground as Ground 1 is relatively stiff compared to Ground 2, it is
shown in Fig. 17. A prescribed plane-distributed displace- easy to transfer the load to the pile foundation even if the
ment is applied in the x-direction, at two vertical planes load is applied at a great distance. In the case of Ground
away from each side of the footing. The distance from 2, however, the distribution of the moments shows a big
the edge of the footing to the center of a pile is 1.2 m. difference with various loading distances. The larger the
The inertial force that comes from the upper structure loading distance is, the smaller the bending moment will
during an earthquake is not considered in the analysis. be. In soft ground, therefore, it is somewhat difficult to
The heads of the piles are fixed in the calculation. choose a suitable loading distance for the responding dis-
The influence of the distance from the point at which placement method. Much attention should be paid to the
the prescribed displacement is applied to the footing in proper choice of a loading distance.
the responding displacement method is investigated, giv- Figure 20 shows the distributions of displacements in x
ing consideration to three cases with different distances direction. It is clear from the figure that in the case of
(S) of 0.6 m, 1.9 m, and 5.9 m. Ground 1, a sliding surface exists in the ground due to
Figure 19 shows a comparison of the distribution of the failure of the surface-layer ground. In the case of

NII-Electronic Library Service


SEISMIC EVALUATION OF PILE FOUNDATION 125

(b)-1 S=0.6 m

(a)-2 S=1.9 m

(b )-3 S=5 .9 m
(a) Ground 1 (b) Ground 2
Fig. 20. Horizontal displacement (em) of grounds obtained from responding displacement method (Given displacements are applied from left to
right.)

Ground 2, the displacement is evenly distributed. gen periods in the case of Ground 1 are 0.856 and 0.807
sec., respectively, and 2.524 and 1.965 sec. in the case of
Dynamic Analysis on a Full System Ground 2. The eigenvalue analysis is conducted with a
In this paper, a Rayleigh type of attenuation is adopt- hybrid of Jacobian and subspace methods. In the earth-
ed and the attenuation constants of the structures and the quake vibration analysis, a direct integration method of
ground are assumed as 5% and 10%, respectively, in the Newmark-P, the same method as is used in the S-R
dynamic analysis of a full system. It is known that the model, is adopted and the time interval of the integration
value of attenuation that is very difficult to be evaluated is 0.01 sec.
for a full system may greatly affect an elastic dynamic Figures 21 and 22 show the hysteresis of the stress-
response. In the plastic region, however, the hysteretic at- strain relation (a zx- ezx) at different positions for Ground
tenuation from the elasto-plastic behaviors of soils and 1 and Ground 2. In Ground 1, the stress-strain relations
piles may become a dominant factor. In engineering de- of elements e1 - e9 behave elasto-plastically and a big hys-
sign, the hysteretic attenuation is usually equalized to an teresis loop exists in the stress-strain relation. The stress-
equivalent viscous attenuation, by taking a large value of strain relations in the surface-layer ground behave elasti-
attenuation for strong elasto-plastic materials. For exam- cally. In Ground 2, however, the stress-strain relation is
ple, the value of attenuation for soft clay under a strong quite different from the relations in Ground 1. Except for
earthquake movement may take a value of 20%. There- element e1, in which a big hysteresis loop exists in the
fore, the value of attenuation assumed here is not deter- stress-strain relation, the stress-strain relations of the
mined by a strict analysis and is only an empirical one. ground behave elastically. This phenomenon implies that
Although the stiffness of the ground, the piles, and the if the bottom-layer ground fails during an earthquake,
pier may change because of the nonlinearity of these the vibration wave will not transfer to the upper layers,
materials, the viscous matrix calculated from the Ray- on the condition that the shear strength of the ground
leigh type of attenuation is assumed to be constant in along the vertical direction is the same. The failed soil lay-
spite of the changes in the stiffness matrix. In order to cal- er looks like a shelter for preventing the strong waves
culate the viscous matrix, an eigenvalue analysis for the from reaching the surface. The displacement on the
full system is conducted to evaluate the first two eigen- ground surface, however, may be very large.
values. The eigenvalue analysis shows that the first two ei- Figures 23 and 24 show the hysteresis of the moment-

NII-Electronic Library Service


126 KIMURA AND ZHANG

~ ~ ~
~100 ~100 ~100

II!
b~ b~ b~
~ 0 ~ 0 ~ 0
~ ~ ~
~100 ~-100 ~-100
] ] ]
C/) C/) C/)

-200 ~............................................................
J;;I..I.I...I.J,..J..I..J...I..I..I....I..I..I.,j............ -200 C..U...L.L.U..I..I..J...L.U..U..I...I-I..J............U..U..J....I..U.J..U..I..J...I..I..I....J;;J -200 ~.I.J,..J..I.~~ ~.L.L.U.~~~~
............

-0.04 -0.02 0 0.02 0.04 -0.04 -0.02 0 0.02 0.04 -0.04


Shear strain E Shear strain Ezx Shear strain E
200 zx 200 ..,...,..,..,...,.....,...,...,....,...,.....,...,...,....,...,.....,...,...,....,...,.....r-rT"r.,...,.....r-rT"r'"1"'1 200..,...,..,..,...,.....~.,...,.....~.,...,.....~~zx~.,...,.....~~
~ ~ ~
~100 ~100 ~100
b~ b~ b~
~ 0 / ~ 0 00 0
~ ~ ~
~-100 ~100 ~-100
~ ] ]
v• C/) C/)
-200 t:U.J...u..l..u.J..L..L.J....U..J..J...I..I..l..L.L..u..l..u.J..L..L.J....U...u..J..l..J..I.,.I;;I -200 tl...U..W..U...J....L.L.L..L.J..J.~~~~.L..L.J..J...u..I..L..U..J:I -200 t:U.J...u..l..u.J..L..L.J....U..I..U.I..l..L.L..u..l..u.J..L..L.J....U...u..J..~
-0~04 -0.02 0 0.02 0.04 -0.04 -0.02 0 0.02 0.04 -0.04 -0.02 0 0.02 0.04
Shear strain E Shear strain E Shear strain E
200..,...,..,..,...,.....~.,...,.....r-rT"r.,...,.....~.,...,.....zx~.,...,.....~~ 200..,...,..,..,...,.....r-rT"r.,...,.....~.,...,.....~.,...,.....zxr-rT"r.,...,.....~~ 200..,...,..,..,...,.....r-rT"r.,...,.....~.,...,.....~~~.,...,.....~~

Shear strain E Shear strain E


zx
200~.,...,.....r-rT"r"T"TTr-rT"r"T"TTr-rT"r"T"TTzxr-rT"r.,...,.....r-rT"r~
~ ~
~100 ~100
b~ b~
~ 0 / ~ 0
~ ~
~100 ~100
63 ~
-200 tl.J.J...L...U.Ju..J...L...L.J,...U..J..J...I..I..l..L.L..L...U.Ju..J...L...L.J,...U...u..J..L..J..J..J;;J -200 t:U.J.~L.,U...!...L.J....U...u..J..I..l..L.L~.........................u..J..J....I..I....I;
-0.04 -0.02 0 0.02 0.04 -0.04 -0.02 0 0.02 0.04
Shear strain E Shear strain E
200~~~~~rrrr~~~~~ 200 p;;-rrrrT"TjTrn-r;ry::r~:t:t::C:~TTTT"1'"'TTtl

-0.02 0 0.02 0.04


Shear strain E
zx
Fig. 21. Hysteresis of stress-strain relation at different position in ground (Ground 1)

curvature relations at different positions of the piles in important role in the moment-curvature relation of piles.
Ground 1 and Ground 2. The difference of M- cp rela- Figure 25 shows the deformation patterns of the
tions at various positions in the ground is not as large as grounds calculated from the dynamic analyses on the full
the stress-strain relations shown in Figures 21 and 22. system. In the case of Ground 1, there is a clear rocking
This implies that the deformation of the ground and not phenomenon in the ground around the pile foundation.
the vibration (usually referred to as acceleration) plays an The vertical displacement of the ground on two sides of

NII-Electronic Library Service


SEISMIC EVALUATION OF PILE FOUNDATION 127

"2' "2' "2'


~ 50 ~ 50 ~ 50
b~ b~ b~
~ 0 ~ 0 ~ 0
~ ~ ~
ta -50 ta -50 ta -50
~ ~ ~
-1 00 t:I..U.J..U..I..U...U,J.J..I...L.J..U..I..I...I..J.J..I..U.J..U..I...I...L.I..JL.J..U..I..,I;;l -1 00 t:I..U.J..U..I..U...U,J.J..I...L.J..U..I..I...I..J.J..I..U.J..U..I..U...UL.J..U..I..,I;;l -1 00 t:I..U.J..U..I...I...L.I..JL.J..U..I..I...I..J.J..I..U...U...U...I...L.I..JL.J..U..I..I...I..J.J..,J;J
-0.04 -0.02 0 0.02 0.04 -0.04 -0.02 0 0.02 0.04 -0.04 -0.02 0 0.02 0.04
Shear strain 8 Shear strain 8 Shear strain 8
100,..,.,..,rT"''"T''T"'!"M"'!"T"T"'T"''"MrT"''"T''!"'!"T"'~zx~T'T'n'!"'!"T"'~ 100 zx 100 ,..,.,..,"'!"T"T"T"n"l"' "T"T"'T"'!"MrT'T"T''!"'!"T"'T'T'ZXT"n"l"'!"T"T"'T"'!"M"TT't::
"2' "2' "2'
~ 50 ~ 50 ~50
b~ b~ bt:l
~ o 1 ~ o ~ 0
~ ~ ~
ta -50 ta -50 ta -50
~ ~ ]
00
-100 t:I..U...J...I...WL..U.J.J.J.J...I..J..J.J..J..J.J..L..I..I.J..U..U.J..U.J~.l..J..W;J -100 t;U,.J,..j,J,J.,U...U..L.U.J...u..J...U..U..j,J,J.,U...U..L.U.J..I..U.~ -1 00 !;l..I,J...I..J..l.JU...U..L.U.J..I..U..U...U..I..U..U..U..I..J..l.JU..U..............
-0.04 -0.02 0 0.02 0.04 -0.04 -0.02 0 0.02 0.04 -0.12 -0.06 0 0.06 0.12
Shear strain 8 Shear strain 8 Shear strain 8
100 zx 100 ...,..,...,."''T' '",..,..,..,...,...,..,..,..,..,..,."''T' '",..,..,..,...,...,..,..zx,..,..,..,...,...,..,..,..,..,.,."t:'J zx
"2' "2'
~ 50 ~50
b~ b~
~ o 1 ~ o
~ ~
ta -50 ta -50
] ~
00 00
-1 00 C..U.J..U..I..U...UL.J..U..L.J..U...I..J..I..I...I...U..L..U...UL.J..U..I..I...I..J.,u;j -1 00 t;U,.J,..I.J..L.I..J..j,J,JL.J..U..I..I...I..J.J..I..U.J..I..U...U..U..Io..U..L.J..U.......
-0.04 -0.02 0 0.02 0.04 -0.04 -0.02 0 0.02 0.04
Shear strain 8 Shear strain 8
100 100 IT'T"l''T"T'T'T..,..,.,..,"''"T'T"M"M",.,..,...,.T"'"T'1'~..,..,.,..,"'1"T'T"~
"2' "2'
~50 ~ 50
b~ bt:l
~ 0 ~ 0
~ ~
ta -50 ta -50
] ]
00 00
-100 tl..I....L..J...i...WU..U.J.J.J...I..J..J.J..J..J.J..L..j,J,J.,U..U...I..I.J..~.l..J..W;J -100 t;U..J...j,J,J.,U...U..U...U.,J..J,J,..U..U..I..J..l.JU..U..U...U.,J..J,J,.~
-0.04 -0.02 0 0.02 0.04 -0.04 -0.02 0 0.02 0.04
Shear strain 8 Shear strain 8
zx zx

Fig. 22. Hysteresis of stress-strain relation at different position in ground (Ground 2)

the footing is in up and down directions, while the failure in the ground near the pile foundation. The
horizontal displacement is not disturbed very much. In horizontal displacement, however, is not disturbed very
the case of Ground 2, a rather complicated distribution much either, being similar to the case of Ground 1. It is
of vertical displacement is observed in the ground around also found that the disturbed area in which the displace-
the pile foundation, showing that a vigorous change in ment changes vigorously is restricted only to the vicinity
vertical displacement occurs, which may be caused by a of the pile foundation.

NII-Electronic Library Service


128 KIMURA AND ZHANG

s*
~
'--'
EQ)
2

0
[!
-..
*
~
s

E
Q)
2

0
s*
~
'--'
E
4

2
0
b30
s
0
-1 s
0
-1
Q)

s
0
-2
::;s ::;s ::;s
-2 -2 -4
-3 -2 -1 0 1 2 -3 -2 -1 0 1 -6
Curvature (10-3/m) Curvature (10-3/m)

2 2 2
-..
s* s s*
[l1J
*
~ 0 ~
'--' 0 ~
'--' 0
i:l
Q)
EQ) E
s
0
-1 s
0
-1 Q)

s
0
-1
::;s ::;s ::;s
-2
-2 -1 0 1
3
2 -3 -2 -2 -1 0 1 2
Curvature (10- /m) Curvature (10-3/m)

2 2 2
-.. -..
s s* s

tt?
* *
~ ~ ~
'--'
EQ)

s
0
-1
0 '--'
i:l
Q)

s
0
0

-1
'--'
EQ)

s
0
0

-1
I
::;s ::;s ::;s
-2 -2
-3 -2 -1 0 1
3
2 -3 -2 -1 0 J 2 3
Curvature (10- /m) Curvature (10- /m)
2 2
-.. -.. b5
s s

[ll!l
* *
~
'--' 0 ~
'--' 0
E E
Q)

s
0
-1 s
0
-1
::;s ::;s
-2 -1 0 1 2
3
Curvature (10- /m)
2 2
-..
s
* *
s
~
'--' 0 ~
'--' 0
EQ)
i:l
Q)

s
0
-1 s
0
-1
::;s ::;s
-2
-3 -2 -1 0 1 2 3
3
Curvature (10- /m)

Fig. 23. Hysteresis of moment-curvature relation at different position of pile (Ground 1)

Comparison of the Mechanical Behaviors of the Upper from the dynamic analyses on the S-R model and the full
Structure and the Piles with Different Methods system. It is found that the maximum moment and the
Comparisons of the results obtained from three differ- hysteresis of the M -l/J relation are quite similar to each
ent methods are made to understand the characteristics other in both grounds.
of these methods in the seismic evaluation of a pile foun- Figure 27 shows a comparison between the time history
dation. Figure 26 shows a comparison of the hysteresis of of the responding acceleration at the top of the pier ob-
the M- l/J relation at the bottom of the pier obtained tained from dynamic analyses with different models. It is

NII-Electronic Library Service


SEISMIC EVALUATION OF PILE FOUNDATION 129

s
~
~
0
s-1
0
~

,...... 2 ,...... 2 ,...... 2


s s s

(!111
* * *
~ ~ ~
'-"

~
~
Q)
s
0
0

-1
'-"
~
Q)

~
§
0

-1
'-"
~
Q)

~
s
0
0

-1
Jl
-2
-3 -2 -1 0 1 3
3
Curvature (1 o- /m)

2 2 2
s* s* s*
~
~
Q)

~
s
0
0

-1 IJt7 ~
'-"
-a
s
~
Q)

0
0

-1
~
'-"
1::!
Q)

§
~
0

-1

-2
-3
I
2 2
s* s*
~
'-"
0
~ 0
-a
Q)
1::!
Q)

§ -1 § -1
~ ~
-2
-3 -2 -1 0 1 2
Curvature (1 o-3/m)
,...... 2 2

*
s s
*
~
'-"
~
'-"
0 0
1::!
Q)
-a
Q)
s
0 -1 § -1
~ ~

Fig. 24. Hysteresis of moment-curvature relation at different position of pile (Ground 2)

found that the maximum values of the acceleration as placements are very large, and the displacements are in a
well as the wave shapes obtained from different analyses different order. It is clear that the displacement calcu-
are almost the same for both grounds. lated from the S-R model does not take into considera-
Figure 28 shows a comparison between the time history tion the deformation of the ground, which in the dynam-
of the responding displacement at the top of the pier ob- ic analyses on the full system, may be as large as 30 em. If
tained from dynamic analyses with different models. The the displacement is added to the one obtained from the S-
differences in amplitude between the two responding dis- R model, it is reasonable to say that the responding dis-

NII-Electronic Library Service


130 KIMURA AND ZHANG

Horizontal
displacement

Vertical Vertical
displacement
displacement
(a) Ground 1 at t=8 sec. (b) Ground 2 at t=7 sec.

Fig. 25. Displacements (em) of two grounds obtained from dynamic analyses on full system

12~~~~~~~~~~~~ 12~~~~~~~~~~~~

s 8 s 8

~ ~
4 4
0 '-" 0
~ ~
s0
-4 s -4
0
::;E -8 ::;E -8
-12 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .j,U,j,J,.j,U,j,J,~
-4 -2 0 2 4 6 -6 -4 -2 0 2 4 6
3 3
Curvature (1 0- /m) Curvature (10- /m)
12~~~~~~~~~~~~ 12~~~~~~~~~~~~

s 8 s 8

~~
~
~
'-" 0
4

s0
-4
~
s
0
-4
::;E -8 ::;E -8
-12 ~u.u.l.i...l,,,...j,.J..I..,j,j,j.l..J..I..J..L.I.I.u...iU.U.U...I,,,...j,.J.l..J..I..J..L.~~..I.,I;;I -12 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...u.u.u=~
-6 -4 -2 0 2 4 6 -6 -4 -2 0 2 4 6
3 3
Curvature (1 0- /m) Curvature (1 o- /m)
(a) Full system (b) S-R model
Fig. 26. Comparison of moment-curvature relations at bottom of pier

placements at the top of the pier obtained from the two The distribution of the moment obtained from the dy-
dynamic analyses are of the same order. namic analysis on the full system is the one at the time
Based on the above discussion, it can be concluded when a maximum bending moment occurs in the piles
that if concern is restricted to the upper structure, the (t=8.42 sec. for Ground 1 and t= 11.84 sec. for Ground
separated method with a dynamic analysis on a simplified 2). It is found from the figure that the results from the
S-R model developed in this paper can obtain reasonable separated method are much smaller than the others and
results. its maximum value is only about one third of the result
Figure 29 shows a comparison of the distribution of from the responding displacement method. The results
the moments obtained from the three different methods. obtained from the responding displacement analysis and

NII-Electronic Library Service


SEISMIC EVALUATION OF PILE FOUNDATION 131

Ground2 Ground2
5 10 15 5 10 15 20
Time (sec) Time (sec)
(a) Full system (b) S-R model
Fig. 27. Comparison of responses of the acceleration at top of pier

5 60~~~~~~~~~~~~~~~~
"-"
"SQ) 30
5 0
C)
..$ -30
a
0..
-60
0 5 10 15 5 10 15 20
Time (sec) Time (sec)
1f 60~~~~~~~~~~~~~~~~~ 1f 6~~~~~~rn~~~nn~~~~~rr~
"-" "-"
"S 30 "S 3
~ 0 ~ 0
~ -30 Ground 2
~
]. - 3 Ground 2
~
0 -600
5 10
15 20 o -6 ol;;;.l....I...J...I..J,....J,....J....J..J...5.1...J...J...J....I..J....J,....J....J..J...11....Lo..I-.I...J.-"-I..-L..J....I.-11....L5..1-.L...J.-"-1..-~20
Time (sec) Time (sec)
(a) Full system (b) S-R model
Fig. 28. Comparison of responses of displacement at top of pier

0
Soil1
5 5

s 10
'-"
Soi12
~
Q 15

20 20

25~~~~~~~~~~~~~~~~
-3 -2 -1 0 1 2 -3 -1.5 0 1.5
Moment (MN*m) Moment (MN*m)
(a) Ground 1 (b) Ground 2

- dynamic analysis (t=8.42 sec.) ·······-- Mc=0.702 MN*m


-o--- seperated method --------- My=1.591 MN*m
-<r-- responding displacement method (S=5.9) --------- Mu=2.230 MN*m

Fig. 29. Comparison of distributions of moments in piles obtained from different methods

NII-Electronic Library Service


132 KIMURA AND ZHANG

the dynamic analysis of the full system are of the same takes into consideration the interaction between soils and
order and in a similar distribution along a vertical direc- foundations, is possible nowadays because of the quick
tion. At a depth of 15- 17m, the distributions of the mo- development of computer and numerical technology.
ment obtained from the dynamic analysis and the The constitutive model of the ground considered in this
responding displacement method show a very sharp paper is simple and only four parameters are needed.
change in both grounds. This phenomenon is thought to These parameters can be determined with the N value of
be the result of the vigorous changes in deformation near SPT, which is quite familiar to engineers.
the area.
It is obvious that the moment in the pile caused by the
deformation of the ground in the present analyses is far ACKNOWLEDGEMENTS
more predominant than the moment caused by the iner- Sincere thanks are given to Dr. A. Nishimura and Dr.
tial force from the upper structure. The maximum bend- T. Murono of the Japan Railway Research Institute, for
ing moment occurs within a ground instead of at the top providing the precious experimental data used in this
of the piles, which has been verified in the reports about paper.
the damage investigation after Hyogoken-Nambu earth-
quake (Horikoshi, et al., 1996). Therefore, the influence
of the deformation of a ground on piles must be consi- REFERENCES
dered. 1) Adachi, T., Kimura, M. and Zhang, F. (1994): "Analyses on ulti-
mate behavior of lateral loading cast-in-place concrete piles by
three-dimensional elasto-plastic FEM," Proc. of the 8th Int. Conf.
CONCLUSION Computer Method and Advance in Geomechanics, Vol. 3, pp.
2279-2284, Morgantown: Balkema.
By conducting a series of static and dynamic three- 2) Design Codes of Foundations and Earth-Retaining Structures of
dimensional finite element analyses (DGPILE-3D), a Japan Railway (1997): Japan Ministry of Transportation, ISBN 4-
number of methods used in the seismic evaluation of a 621-04315-3 C 3051, Maruzen Print Co. Ltd. (in Japanese).
pile foundation are discussed in detail. The following con- 3) Design Codes of Japan Highway Bridge, Vol. V, Seismic Design-
ing, 1996, Japan Highway Association, ISBN 4-88950-234-3,
clusions can be made: Maruzen Print Co. Ltd. (in Japanese).
(1) A model test on a full system that consists of a 4) Fukutake, K. (1997): "Seismic evaluation of a group-pile founda-
pier, a group-pile foundation, and a ground is simulated tion surrounded by a frame wall using a 3-D nonlinear dynamic
with a three-dimensional dynamic analysis under an elas- analysis," Proc. of the 24th Symp. of Japan Earthquake Engrg.,
tic condition. The mechanical behaviors, such as the pp. 673-676 (in Japanese).
5) Horikoshi, K., Ohtsu, H., Kimura, M. and Oka, F. (1996): "Inves-
responding accelerations of the ground and the structure, tigation of piles damages by the 1995 Hyogoken-Nambu Earth-
the spectra of the accelerations, and the transfer function quake," Tsuchi-to-Kiso, Vol. 44, No. 11, pp. 27-29 (in Japanese).
of the ground and the pier can be well simulated. 6) Kimura, M., Yashima, A. and Shibata, T. (1991): "Three-dimen-
(2) If a seismic evaluation is restricted to the upper sional finite element analysis of laterally loaded piles," Proc. of the
structure of a structure-foundation-ground system, the 7th Int. Conf. Computer Method and Advance in Geomechanics,
pp. 145-150, Cairns: Balkema.
separated method proposed in this paper, using a dynam- 7) Kimura, M. and Zhang, F. (1997): "Seismic evaluation of pile foun-
ic analysis on a simplified S-R model, can obtain reasona- dation," Proc. of the 6th Int. Symp. Numerical Models in Ge-
ble results. For the piles, however, the results may omechanics, pp. 545-548, Montreal: Balkema.
underestimate the sectional forces of the piles. 8) Kimura, M. and Zhang, F. (1999): "Seismic evaluation of pile foun-
(3) The loading position of a prescribed displacement dation by static and dynamic 3-D finite element analyses,'' Proc. of
the 11th Asian Regional Conf. of Int. Society for Soil Mechanics
in the responding displacement method may affect the dis- and Geotech. Engrg., Vol. 1, pp. 507-510.
tribution of sectional forces in a pile, especially in a case 9) Matsuda, T., Ito, K. and Toriihara, M. (1994): "Shaking table test
where the ground considered is very soft. Precautions and dynamic effective stress analysis of light structure under satu-
should be taken before using the method in the seismic rated sandy ground," Proc. of 8th Int. Conf. Computer Method
evaluation of a pile foundation. and Advance in Geomechanics. Vol. 2, pp. 935-941, Morgantown:
Balkema.
(4) During a major earthquake, the influence of the 10) Murono, T., Nishimura, A. and Nagazuma, M. (1997): "Model
deformation of a ground on the piles cannot be disre- test on a group-pile foundation considering the vibration of
garded. The moment in the pile caused by the deforma- ground in seismic evaluation," Proc. of the 24th Symp. of Japan
tion of a ground may be far more predominant than the Earthquake Engrg., pp. 625-628 (in Japanese).
moment caused by the inertial force from the upper struc- 11) Taji, Y., Sato, M., Kagawa, T. and Minowa, C. (1997): "Investiga-
tion on soil-pile-structure interaction in liquefying sand by a large
ture. In the seismic evaluation of a group-pile founda- scale shaking-table test and centrifuge test," Proc. of the 24th
tion, the moment near the head of the piles is usually Symp. of Japan Earthquake Engrg., pp. 277-280 (in Japanese).
checked very carefully, while the moment of the piles at a 12) Taguchi, Y., Tateishi, A., Oka, F. and Yashima, A. (1997): "Nu-
deep place does not attract too much of the engineers' at- merical simulation of soil-foundation interaction behavior during
tention. Based on the results obtained in this paper, it is subsoil liquefaction," Proc. of the 6th Int. Symp. Numerical
Models in Geomechanics, pp. 561-564, Montreal: Balkema.
worth emphasizing that the influence of the deformation 13) Wakai, A., Ugai, K. and Matsuda, T. (1997): "Seismic analysis of
of a ground on the piles must be considered carefully. a bridge-ground system excited in the axial direction of the
(5) Dynamic analysis on a full system, which has few- bridge," Proc. of the 6th Int. Symp. Numerical Models in Ge-
er assumption compared with other methods and directly omechanics, pp. 709-712, Montreal: Balkema.

NII-Electronic Library Service

You might also like