4th International Conference on Earthquake Engineering

Taipei, Taiwan
October 12-13, 2006

Paper No. 034

RECENT STUDIES ON SEISMIC SOIL-PILE-STRUCTURE-

INTERACTION IN SOFT CLAY

Juan M. Mayoral1 and Miguel P. Romo2

ABSTRACT

The evaluation of performance of deep foundation systems under seismic loading is one of the most
complex problems in earthquake engineering, due to the varying degrees of nonlinear soil response
observed during severe earthquakes. Strong shaking may lead to large excess pore pressures and
strain-induced softening, increasing the amount of nonlinear behavior in the near, intermediate and
free field domains. These effects overall become more significant for multi-directional shaking,
which may reduce the soil stiffness and increase permanent deformations of the structure and/or soil
even further. In particular, in the near field, lateral multi-directional movements of the pile can cause
radial degradation of the soil stiffness, and can create radial gapping, which in turn, reduces the
confinement of the pile, leading to an increase in bending moments on the structural members.
Similarly, two-directional shaking may generate larger shear strains in the intermediate and free field.
In practice, however, the seismic soil-pile-structure interaction problem is often oversimplified and
the nonlinear effects are ignored. This paper revisits recent studies and developments directed to
improve our understanding of these phenomena in soft clay, including analytical and experimental
results. The predicting potential of non-conventional analysis techniques, including soft computing
and nonlinear dynamics is discussed.

Keywords: Soil-pile-structure, seismic, multi-directional, interfaces, soft clay, soft computing,

nonlinear dynamics

INTRODUCTION

With few exceptions, the analysis of seismic soil-pile-structure interaction problems is often
oversimplified and the effects of two-directional shaking are ignored. Nevertheless, two-directional
shaking may significantly increase the amount of non-linear behavior in the near, intermediate and
free field domains. In addition, larger excess pore pressures and strain-induced softening may be
generated due to two-directional loading, which can reduce the soil stiffness and increase permanent
deformations of the structure and/or soil. Cyclic movements/displacements in one direction can soften
the near field soil, and can significantly affect soil-pile interaction for subsequent movements in other
directions. Usually, the response of the near field soil is represented with uni-directional p-y curves.
However, two-directional lateral movements of the pile can cause radial degradation of the soil
stiffness and create radial gapping, which is not captured by uni-directional p-y representations. The
correct characterization of the radial stiffness degradation and gaping is crucial because this effect
reduces the confinement of the pile, leading to an increase in the bending moment on the structural
foundation member as well as a typically larger demand in the superstructure. The purpose of this
paper is to review recent research efforts aimed at addressing these issues. In particular, in the near

1
Associated Professor, Institute of Engineering at UNAM, Mexico City, Mexico, jmayoralv@iingen.unam.mx
2
Professor, Institute of Engineering at UNAM, Mexico City, Mexico, mpr@pumas.iingen.unam.mx
field, this includes a two-directional pot testing device and an experimental methodology developed to
characterize the multi-directional response of soil-pile interfaces, several series of cyclic multi-
directional displacement control pot tests were performed, and a numerical model to described the
observed behavior was proposed. The multi-directional pot testing device allows the study of soil-
pile units to better characterize the multi-directional response of the near field during cyclic loading.
The results obtained from these tests fill an important gap in experimental information regarding
multi-directional p-y curves. In the far field a model proposed for multi-directional non-linear analysis
of the free field soil response to seismic loading is commented. This model includes explicitly
hysteretic behavior, plastic deformations, and pore pressure generation during cyclic loading.

NEAR FIELD RESPONSE

Recent observations during the Turkey (1999) and Taiwan (1999) earthquakes have showcased the
multi-directional nature of strong shaking and the importance of directivity effects on ground motions.
Following the research program that has been conducted at UC Berkeley for over a decade, to develop
and calibrate numerical tools for assessing the seismic performance of structures supported on deep
foundations in soft clay, Mayoral et al. (2002) designed, constructed and calibrated a new testing
device to study the multi-directional response of clay-pile interfaces using the p-y spring concept. As a
part of this research effort several multi-directional pot tests were conducted in a model soil. The
model soil consisted of a mixture of 72% kaolinite, 24% bentonite, and 4% type C fly ash (by weight).
This soil mixture was engineered to have similar mechanical properties to San Francisco Young Bay
Mud with a scaling factor, , of 8 for 1-g shaking table experiments (Wartman, 1996). This soil was
developed originally to be used in large-scale Seismic Soil-Pile-Superstructure Interaction (SSPSI)
model tests, which were performed on the 6.1 m x 6.1 m U.C. Berkeley/PEER Center multi-directional
shaking table to generate a suite of case histories to calibrate an advanced numerical code for SSPSI
analysis (Meymand, 1998). Based on the experimental findings, a simple analytical discrete element
model to simulate clay-pile interfaces subjected to bi-directional loading was developed and validated.

Multi-directional Pot Testing Sep Up

As described by Mayoral et al. (2005), the multi-directional pot testing device is an extension of the
original pot test proposed by Matlock (1970). The main components are the instrumented model pile
and a testing device capable of imposing a suite of prescribed soil/pile lateral relative displacements in
two orthogonal directions.

Bi-directional pot testing device

Figs. 1a and 1b show schematically a cross section and a plan view of the new device illustrating some
of the main components. The device is comprised of the following elements: loading unit, external and
internal transducers, roller table, a loading frame and a data acquisition system. The loading unit has
two pneumatic servo-valves and two orthogonally aligned loading pistons. The external transducers
consist of two load cells and two DC-wire potentiometers. Wire potentiometers were preferred over
LVDTs because of the large displacements in each direction required during the test. The wire
potentiometers are used both for data collection and for control of the closed-loop loading system. The
wire pots are mounted on the loading frame with one of them connected to the roller table and the
other one to the pile cap. The internal transducers are composed of orthogonal pairs of temperature
compensated electrical resistance foil strain gauges placed at three elevations along the model pile.
The equipment is completed with the model pile cap, model pile and soil container. A container with
the model soil and the instrumented model pile is placed on the rolling table. The model pile can be
loaded and/or displaced laterally independently in two orthogonal directions.

Model pile
The model-pile is a 51 mm diameter aluminum tube with 0.51 mm wall thickness and its mechanical
properties are summarized in Table 1. The pipe pile has a rigid tip composed of a conical tip with a
cylindrical termination that fits tightly into one end. The model pile was instrumented with orthogonal
pairs of temperature compensated electrical resistance foil strain gauges at three elevations 50, 126
and 202 mm.

Data
Acquisition
System
Servo- A.P.
Valve 1 Regulator 1
Piston 1

Instrumented A.P.
Model Pile Regulator 2
Roller Table Load Air Pressure
Cell 2 Supply
Servo-
Bucket with Model Soil Valve 2
Wire Potentiometer 2

Piston 2

Roller Blocks and Table Guiding System

a) Cross Section
Wire Potentiometer 1
Data
Acquisition
System
Soil
Roller Block &
Container Load
Pile Cap Guiding
System Cell 2
Table/Container
Model Movement
Pile

Model Piston 2
Pile Cap

Servo-
A.P.
Valve 2
Regulator 2

Roler
Wire Potentiometer 2 Table Guiding
Blocks
System

Load
Servo-Valve 1
Cell 2 A.P.
Regulator 1
Air Pressure
Pile Movement Piston 1 Supply

b) Plan View

Figure 1. Multi-directional Pot Testing Device (not to scale, After Mayoral et al., 2005)

Table 1. Summary of the Scale Pile Properties

Wall
Young Modulus, Diameter, Flexural Rigidity,
Material Thickness, t
E, (kPa) D (mm) EI (N.m)
(mm)
Aluminum 140 51 0.3 2945
 Multi-directional p-y Response
Multi-directional displacement controlled tests were conducted with the new device to model typical
free-field ground displacement paths (also referred to as orbits) recorded in recent earthquakes, such
as those depicted in Fig. 2. The x and y horizontal components of these paths are normalized by the
Peak Ground Displacement, PGD, defined as the maximum displacement in one of the two
components. These two examples cover a wide range of PGDs from approximately 3 to 90 cms for
earthquakes with moment magnitudes ranging from 6.5 to 7.0. Although the orbits are very complex,
the largest loop can be approximated by one (or a combination of several) simplified displacement
paths, such as: a) linear, b) elliptical/circular, c) figure-8 and d) parabolic. In particular, a multi-
directional p-y curve corresponding to a figure-8 pile displacement path is shown in Fig. 3. The typical
flat portion in the central portion of the p-y curve is noticeable. The maximum normalized
displacement amplitudes in the x and y directions for a given pile transit path are given by Ax/D and
Ay/D, respectively. Fx and Fy are the loading frequencies for a cyclic loading in the x and y directions,
respectively. The lateral forces, px and py, were normalized with respect the shear strength corrected
for rate effects for the direction of maximum displacement, su*, multiplied by the model pile diameter,
D (i.e., p/(su*.D)). The displacements in the directions x and y are normalized by the pile diameter (i.e.,
x/D and y/D). The resulting p-y response is dimensionless and it is only a function of the embedment
ratio (L/D). The corresponding undrained shear strength and the residual shear strength were
determined using a vane shear apparatus in three to four undisturbed portions of the model clay.
1.0 1.0

El Centro Array 6 Sta

Imperial Valley, 1979
PGD=0.028m, M =6.5
w
0.5 0.5
Normal Component/PGD

0.0 0.0

-0.5 -0.5
Gilroy Historic Bldg Sta
Loma Prieta, 1989
PGD=0.855m, M =7.0
w

-1.0 -1.0
-1.0 -0.5 0.0 0.5 1.0 -1.0 -0.5 0.0 0.5 1.0
Parallel Component/PGD Parallel Component/PGD
Figure 2. Examples of free-field ground displacement paths observed during recent earthquakes.

The data gathered from this experimental work fill the current lack of experimental data to formulate
and calibrate p-y curves in multi-directional loading conditions. From this data, Mayoral et al. (2002 &
2005) concluded that the shapes of the p-y curves were strongly affected by the loading path and were
a function of the evolution of the gap during the cyclic loading. This leads to a coupling between the
passive and frictional components. This effect is more significant for circular and elliptical
displacement transit paths. The number of cycles of loading required to achieve fully stabilized
gapping behavior for multi-directional pot test was path dependent as well.

Modeling Soil-Pile Interfaces Subjected to Multi-directional Cyclic Loading

A discrete p-y spring element was introduced by Mayoral et al. (2006) to describe the nonlinear
response of the near field for soft clays, including the progressive formation of gapping for multi-
directional loading conditions. The nonlinear stiffness of the spring is controlled by a plastic and a
loading surface that maps radially onto a boundary surface. The loading surface has two limiting
stages, which define the first loading and loading conditions after the gap has been fully stabilized. In
the first stage, the gap is not open and in the second stage, the gap is completely stabilized. The plastic
surface keeps track of the displacement history of the pile. An interpolation function, which depends
on the accumulation of plastic deformations, allows for a better representation of the gap evolution on
the transition zone of the p-y curve, as can be seen in Fig. 4, which presents a comparison between
model predictions and experimental data for a figure 8 pile transit path. Although radiation damping
is not incorporated explicitly in the formulation, this nonlinear p-y spring can be used in series with a
dashpot to model radiation damping (i.e., geometric damping) (e.g., Lok, 1999, Lok et al., 2000).

6
Force-Displacement- x component
Model Pile D= 51 mm, L/D = 5
4 & Pile Cap
Normalized Force, p /(s *.D)

Movement of
u

2 Container
x

x-direction
0

-2
Pile-cap
Model Movement
-4 y-direction
Soil

-6
1.00 1.00
2-D Displacement Path Force-Displacement- y component
0.75 0.75
Normalized Displacement, y/D

0.50 0.50

0.25 0.25

0.00 0.00

-0.25 -0.25

-0.50 -0.50

-0.75 -0.75

-1.00 -1.00
-1.00 -0.75 -0.50 -0.25 0.00 0.25 0.50 0.75 1.00 -6 -4 -2 0 2 4 6
Normalized Displacement, x/D Normalized Force, p /(s *.D)
y u

TEST INFORMATION SOIL PROPERTIES

Test # Path A /D A /A F (Hz) F (Hz) (%) s * (kPa)
x x y x y u

f81312 Figure 8 0.60 3:1 0.05 0.10 108 3.88

Figure 3. Multi-Directional p-y Test: Figure 8 Displacement Path

 8

6 Model Pile D= 51 mm, L/D= 5

& Pile Cap
4
Movement of
Normalized Force, px/(su*.D)

Container
2 x-direction
0

-2
Model Pile-cap
-4 Soil Movement
y-direction
Model
-6 Data

-8
1
Normalized Displacement, y/D

0.5

-0.5

-1
-1 -0.5 0 0.5 1
-8 -6 -4 -2 0 2 4 6 8
Normalized Displacement, x/D Normalized Force, py/(su*.D)
Figure 4. Model Prediction and Measure Response for a Figure-8 Displacement Path

FAR FIELD RESPONSE

In an attempt to model soil nonlinearities during multi-directional loading in the free field, Mayoral et
al. (2006b), proposed a simplified plastic hysteretic model, which is based on an existing effective
stress constitutive law developed to describe the cyclic behavior of lightly over consolidated clays in
simple shear (Pestana et al., 2000), but modified to account for hysteretic behavior. The model was
implemented in an 8-noded solid finite element to study the effects of two directional dynamic loading
on the response of soil deposits, and validated throughout comparisons with other implementations
and experimental results. Then, the predictions capabilities of the model are explored. Among the
cases studied is included the analysis of a deposit of generic stiff soil that was previously studied by
Rodriguez-Marek, (2000), using a total stress-based bounding surface plasticity model (Borja et al.,
1999) for clays (c.f. Fig. 5).
 0.08 0.08

0.06 Surface 0.06

(Normal)
0.04 0.04
Displacement, m

0.02 0.02

0 0

-0.02 -0.02

-0.04 -0.04

-0.06 -0.06

-0.08 -0.08
15
0 5 10 15
Normal

Parallel
10
Surface
Soil
(Parallel)
Column of
Generic 45m
Stiff Soil 5

0
-0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08
Displacement, m
-1 1

Surface
-0.5 0.5
Accelerations, g

0 0

0.5 -0.5
BPM
EHM
1 -1
1 1

Pacoima Dam: Normal Comp. Pacoima Dam: Parallel Comp.

0.5
PGArock=0.2g PGArock=0.5g 0.5
Accelerations, g

0 0

-0.5 -0.5

-1 -1
0 5 10 15 0 5 10 15
Time (s) Time (s)

Figure 5. Comparisons of the Predicted Response using a Total Stress Based

Bounding Plasticity Model (BPM) and the Enhnaced Hysteretic Model (EHM)
 ALTERNATIVE METHODOLOGIES

Nonlinear effects in soft soils can also be studied through the application of non conventional
advanced mathematical tools, such as those based on soft computing and nonlinear dynamics. In
particular Romo (1999) showed that properly designed network architectures are capable of
forecasting both earthquake ground movements and the seismic response of buildings in soft soils. He
analyzed a seventeen story-concrete structure supported by a friction pile-box foundation. A scheme of
the lateral faade is depicted in Fig. 6, where the strong motion instruments layout is indicated.
0.1

Acceleration, g's
0.05
Floor 17 0
-0.05 0 50 100
-0.1

93/10/24
0.1 M=6.6
Floor 12 0.05
0
-0.05 0 50 100
-0.1

0.1
Floor 6 0.05
0
-0.05 0 50 100
GL -0.1

0.1
0.05
Basement 0
-0.05 0 50 100
C -0.1
Time, sec
Accelerometer

Figure 6. Lateral faade of Plaza Crdoba building, Mexico City (After Romo, 1999)

The forecasting of unseen recorded acceleration time histories depicted in Fig. 7 is also highly
accurate, although some peaks of the time series at floors 6 and 12 are slightly missed. In the
prediction mode, the correlation was 0.989. The results point out toward the great potential of
Artificial Neural Networks to increase the capabilities of computational tools to solve in practice the
seismic-soil-structure interaction problem in soft clay. Similar techniques have been applied by Garcia
et al. (2001) to study Mexico City ground motion response using neuronal networks (NNs). This
research showed that knowledge-based procedures are capable of modeling accurately the seismic
amplifications observed not only at predominantly clayed soils deposits existing in Mexico City (lake
zone) but also at silty and sandy soils (transition zone). NNs were able to predict with a good degree of
approximation ground surface responses to seismic events that come from different zones along the
subduction source and interslab source, as is the case of the Tehuacan seismic event (99/06/15). This
result is particularly encouraging because it shows the flexibility and predicting capabilities of NNs.
On the other hand, procedures based on empirical relative transfer functions failed to produce reliable
results (Singh et al, 1999). Similarly, analytical tools do not seem to predict accurately ground
responses for intermediate-depth seismic events. Other approaches, based on recurrence quantification
analysis (RQA), have also shown potential for practical applications (Garcia et al. 2002). Proper
interpretation of this powerful discriminatory tool in the NNs model of a dynamical system can yield
definitive clues for the qualitative assessment of time series. With RQA one can graphically detect
hidden patterns and structural changes in data or see similarities in patterns across the time series
under study. In this sense, the addition of RQA parameters improves drastically the model forecasting
capabilities.
 1.5 1.5
Floor 6 Floor 6
1 1
0.5 0.5
0 0
50 55 60 65 70
-0.5 -0.5 50 55 60 65 70

-1 -1
-1.5 -1.5

1.5 1.5
Acceleration, in g's(x10 )

Floor 12 Floor 12
-1

Measured Measured
1

Acceleration, in g's(x10 )
1

-1
ANN ANN
0.5 0.5
0 0
-0.5 50 55 60 65 70
-0.5 50 55 60 65 70

-1 -1
-1.5 -1.5

1.5 1.5
Floor 17 Floor 17
1 1
0.5 0.5

0 0
-0.5 50 55 60 65 70 -0.5 50 55 60 65 70

-1 -1
-1.5 -1.5

a) October 24, 1993 seismic event b) September 14, 1995 seismic event

Figure 7. Measured and ANN predicted motions at Plaza Crdoba building (After Romo 1999)

CONCLUSIONS
Seismic performance evaluations of soil-pile-structure systems usually are oversimplified. Thus,
multi-directional loading effects are seldom included in the dynamic response of the near field or the
free field. These effects, however, may increase the amount of nonlinear behavior existing in the
different parts of the system.
A new testing device to evaluate the multi-directional p-y response of soil-pile interfaces was
developed. The results obtained for multi-directional pile transit paths show that the maximum
normalized lateral forces and the initial tangent stiffnesses are independent of the loading path.
However, the shape of the p-y curve is affected by the loading path, and is a function of the evolution
of the gap and soil degradation during the cyclic loading. The number of cycles of loading required to
achieve full stabilization of degradation (and gapping) for multi-directional pot tests was path-
dependent as well. The data gathered showed that the multi-directional movement of the pile with
respect to the near field has the following effects on the clay-pile interface response:
(1) Loading induced anisotropy, which can significantly affect the behavior of the near field soil.
(2) Generation of preferential deformation directions due to previous loading history that is
controlled by the associated gap and softened soil zone.
The residual force after stabilization seems to be independent of the loading path. The reduction of the
number of cycles needed to reach stabilization appears to be controlled by the accumulation of plastic
deformations with progressive multi-directional loading.
A simplified discrete model formulated in the time domain was proposed to simulate the response
observed during uni-directional and multi-directional loading of soil-pile segments. Overall, the
formulation captures non-linear response associated with gapping and slippage of the pile during two-
directional loading. The model was able to describe the evolution of the stiffness in the transition
regime, accounting for the progressive formation of the gap around the pile.
Regarding the far field, a fully nonlinear constitutive model was implemented in an 8-noded solid
finite element. The constitutive law is a derivative of a simple model to describe the response of clays
in the simple shear mode of deformation which has been used to describe the response of lightly
overconsolidated clays in submarine environments (Pestana et al., 2000). Model capabilities to
simulate the free field response to two-directional ground motion were evaluated in several case
studies. Overall, bi-directional site response analysis using the enhanced hysteretic model predicted
better the measured response than that obtained with an uncoupled uni-directional site response
analysis. The plastic hysteretic model seems to provide a better representation of the cyclic soil
response during transient and steady state conditions.

REFERENCES
Biscontin, G. and Pestana, J., (2001) "Influence of peripheral velocity on measurements of undrained shear
strength for an artificial soil," ASTM, Geotechnical Testing Journal, Vol. 24, No. 4, 2001, pp. 423-429.
Borja, R.I., Chao, H. Y., Montans, F.J., and Lin, Chao-Hua (1999). Nonlinear Ground Response at Lotung LSST
Site. Journal of Geotech. Engrg., ASCE, 125(3), March, 187-197.
Garca S, Romo M P and Sarmiento N, (2001), Artificial Neural Network Modeling of Mexico City Ground
Seismic Response Proceedings of the XVth International Conference on Soil Mechanics and Geotechnical
Engineering, Istanbul, Turkey, Satellite Conference on Lessons learned from recent strong earthquakes,
pp 155-159, August
Garca S R, Romo M P, Figueroa-Nazuno J and Ramos A, (2002), A recurrence quantification approach for the
modeling of Mexico City ground Motion 12th European Conference on Earthquake Engineering,
Barbican Centre, London, UK., September, paper #231
Lok, T. (1999) Numerical Modeling of Seismic Soil-Pile-Structure Interaction in Soft Clay, Ph.D. Dissertation,
Univ. of California, Berkeley.
Lok T, Pestana J M, and Seed R., (2000) Numerical Modeling of Seismic Soil-Pile-Superstructure Interaction
12th World Conference on Earthquake Engineering, New Zealand 2000, 2000, Paper number 0914.
Meymand, P. (1998) Shaking Table Scale Model Test of Non-Linear Soil-Pile-Superstructure Interaction In
Soft Clay Ph.D. Dissertation, Univ. of California, Berkeley
Pestana, J.M., Biscontin, G., Nadim, F. and Andersen, K., (2000). Modeling cyclic behavior of lightly
overconsolidated clays in simple shear Soil Dynamics and Earthquake Engineering 19, 501-519.
Romo M. P. (1999) Earthquake Geotechnical Engineering and Artificial Neural Networks IV Arthur
Casagrande Lecture, XI Panamerican Congress on Soil Mechanics and Geoenvironmental Engineering,
Foz do Iguassu, Brasil, Vol. IV, Agosto
Mayoral, J. M. (2002) Two-directional Effects in Soil-Pile-Structure Interaction in Soft Clay, Ph.D.
Dissertation, Univ. of California, Berkeley.
Mayoral, J.M., Pestana, J.M. and Seed, R.B. (2002) Determination of Multi-directional p-y curves for soft
clays, GeoEngineering Report RCB/GE/2002-05, University of California, Berkeley, 2002
Mayoral J. M., Pestana J. M., and Seed R. B. (2005) Determination of Multi-directional p-y curves for soft
clays, Geotechnical Testing Journal, 2005, Vol. 28, No. 3
Mayoral, J. M., Pestana, J. M. and Seed R. B., (2006a) Modeling of clay-pile interface behavior during multi-
directional Cyclic Loading, International Conference of Earthquake Engineering, San Francisco, CA,
April, EERI
Mayoral, J. M., Pestana, J. M. and Seed R. B., Romo, M.P., (2006b) A simplified plastic hysteretic model for
multi-directional nonlinear site response in soft soils, 4th International Conference on Soft Soil
Engineering , Vancouver, Canada, October 2006, ISSM
Matlock, H., (1970) "Correlation for design of laterally loaded piles in soft clay, " 2nd Offshore Technology
Conference, OTC 1204, Houston, Vol. 1, 1970, 577-594 pp.
Meymand, P. (1998) Shaking Table Scale Model Test of Non-Linear Soil-Pile-Superstructure Interaction In
Soft Clay Ph.D. Dissertation, Univ. of California, Berkeley
Singh, S. K. et al. (1999), "A Preliminary Report on the Tehuacan Mexico Earthquake of June 15, 1999",
Seismological Research Letters, Vol. 70, No. 5, pp 489-504
Wartman, J. (1996) A Laboratory Study of the Effects of Fly Ash on the Geotechnical Properties of Soft Clay,
M.E. Thesis, Univ. of California, Berkeley