Title

Author(s)

THE EFFECT OF SOIL-STRUCTURE INTERACTION ON

SEISMIC ASSESSMENTS FOR BRIDGES
C. W. HUANG; H. H. HUNG; C. C. CHEN; K. K. JENG

Citation

Issue Date

2013-09-13

DOI

Doc URL

http://hdl.handle.net/2115/54444

Right

Type

proceedings

Additional
Information
File
Information

easec13-H-1-5.pdf

Instructions for use

Hokkaido University Collection of Scholarly and Academic Papers : HUSCAP

THE EFFECT OF SOIL-STRUCTURE INTERACTION ON SEISMIC

ASSESSMENTS FOR BRIDGES
C. W. HUANG1*, H. H. HUNG2, C. C. CHEN1, and K. K. JENG1
1

Department of Civil Engineering, Chung Yuan Christian University, Taiwan

2
National Center for Research on Earthquake Engineering, Taiwan

ABSTRACT
The effect of soilstructure interaction on the seismic assessment of bridges is studied in this paper.
Nonlinear static pushover analyses and dynamic time history analyses of single column bent bridges
are performed using two separate types of finite element models. The first type model assumes that
bridge columns are rigidly connected to the foundation without considering soil-structure
interactions while the second type model incorporate soil-structure interactions through the use of
equivalent springs. In addition, parametric study is carried out to investigate the effects of soil types
and buried depth of piles on the seismic assessments for bridges. Numerical results show a
markedly different seismic behaviour when the soil-structure interaction is included in such
analyses, rather than simply considering a fixed support as usually done in previous studies.
Furthermore, for stronger excitations, it is seen that as inelastic mechanisms are introduced and
boundary conditions changed, the considerations of the foundation and soil compliance play an
increasingly important role that can potentially modify the anticipated failure hierarchy, as well as
the ensuing pushover curves in the transverse direction of the bridge.
Keywords: Soilstructure interaction, seismic assessment, pushover analysis, bridge.
1.

INTRODUCTION

The seismic design and assessment of structures has become an essential issue for the sustainable
development in recent years. A rational seismic design and assessment for a structure has to be
based on the accurate prediction of its inelastic displacement demand and capacity. The
displacement-based seismic evaluation and design has increasingly become the main stream for the
design code of the next generation. The most accurate analytical procedure to estimate the seismic
displacement demands of a structure responding in the nonlinear range is to carry out the nonlinear
dynamic time-history analyses. However, since this procedure requires the consideration of a large
number of earthquakes and is relatively complicated and time-consuming, the use of nonlinear static

Corresponding author: Email: cwhuang@cycu.edu.tw

Presenter: Email: cwhuang@cycu.edu.tw

pushover analysis is generally considered to be a more suitable standard practice for the seismic
design and evaluation of structures.
The static nonlinear pushover analysis is an analyzing procedure whereby an incremental-iterative
process has been carried out to obtain the response of a structure subjected to a specified
monotonically increasing lateral load pattern. However, most pushover analyses of structures have
been derived based on the assumption that structures are supported on rigid foundations. This
assumption could minimize the computation cost, but would result in an overestimation on the
foundation stiffness. Today, most building codes adopt idealized envelope response spectra which
attain constant acceleration values up to certain period (of order of 0.41.0 s at most) and then
decrease monotonically with period (for example as T-2/3). Therefore, the consideration of the
soil-structure interaction (SSI) leads to smaller accelerations and stresses in the structure and
thereby smaller forces onto the foundation. It is widely believed that the SSI is beneficial to the
behavior of the structural system under earthquake loading (Jeremic et al. 2004).
A lot of studies have investigated the influence of the SSI on behaviors of bridges in recent years
(Kappos and Sextos 2001, Silva and Manzari 2008, Kwon and Elnashai 2010). The beneficial role
of SSI has been essentially turned to dogma for many structural engineers. It is worth to note that
the SSI also leads to larger displacements under earthquake loading, which could violate the
performance-based design requirements. The object of this paper is to identify the effect of SSI on
the seismic assessment of bridges while the SSI is simulated by lump springs in finite element
models. In addition, parametric studies are also carried out to discuss the influences of different soil
and foundation buried depth on the magnification factor of lateral displacements. Preliminary
analytical studies comparing the response of fixed-based models with simplified SSI models are
expected to provide important information on the need for considering SSI effects in the design
process.
2.

SOIL-STRUCTURE INTERACTION

The properties of soil include friction, cohesion, dilation/contraction and buildup/dissipation of pore
water pressure. In addition, the randomness of material properties of soil is much higher than that of
common structural materials. To avoid the huge computational cost of a full-scale
three-dimensional finite element model which consists of solid elements with nonlinear material
properties to consider the effect of soil-structure interaction, we adopted a lumped nonlinear soil
spring approach to simplify the computation in this paper. Furthermore, plastic hinges, instead of
fiber-element, are used to simulate nonlinear deformation of the piers and piles.
2.1. Nonlinear soil spring
Elastic-perfect plastic Winkler-type soil springs which were attached at the pile nodes were adopted
for modeling the soil stiffness. The lateral subgrade coefficient for the piles, kH, provided by the soil
was taken as:
2

k H 0.34( E0 )1.10 D 0.31 ( EI ) 0.103

(1)

the and E0 (kgf/cm2) denote the coefficient of subgrade reaction under earthquake and soil
modulus of elasticity, respectively, and are equal to 2 and 25N where N is the value of the standard
penetration test. In addition, D and EI are the diameter (cm) and flexural rigidity (kgf-cm2) of the
pile. The ultimate lateral subgrade reaction force per unit area is given by:
pu p

(2)

where is the modification factor and equals 0.9; p is the passive earth pressure in front of piles
with unit of tf/m2. The relationship between horizontal subgrade reaction force and horizontal
displacement is shown in Fig. 1(a).

(a) Horizontal soil spring

(b) Vertical soil spring

Figure 1: The nonlinear behaviors of the soil springs

Similarly, the vertical soil model surrounding pile caps is also simulated as springs with stiffness
which was computed according to the expression
kVP aAp E p / L

(3)

where Ap and Ep are the cross section area (cm2) and modulus of elasticity (kgf/cm2) of the pile cap,
respectively. Moreover, L is the pile length and the constant a is equal to 0.031(L/D)-0.15 for
case-in-place piles. The ultimate compressive bearing force puP and tensile bearing force plP for pile
are obtained, respectively, from Eqs. (4a) and (4b):
puP 0.8( f s As qb Ab )

(4a)

plP 0.5( f s As qb Ab )

(4b)

where fs is friction resistance on the pile surface, As is the surface area of the pile, qb is the ultimate
bearing pressure of the pile, and Ab is the cross section area of the pile. The relationship between
vertical resistance force and vertical displacement of the pile is shown in Fig. 1(a).

2.2. Plastic hinges

Apart from soil flexibility, nonlinear deformation of piers and piles is also considered in terms of
the potential plastic hinge development. For piers, the plastic hinges are assumed to be located near
the bottom of piers, as shown in Fig. 2(a). The moment-rotational angle curve of the plastic hinge is
determined by the envelope of the two different moment-rotational angle curves obtained by
considering flexural and shears failure criteria. The equivalent plastic hinge length, Lp, of piers is
defined by (Priestley et al. 1996):
L p 0.08L 0.022 f yl dbl 0.044 f yl dbl

(5)

where L is the pile length, fyl is the yield strength of the longitudinal bars, and dbl represents the
diameter of the longitudinal reinforcement.
On the other hand, when the plastic hinge method is used in the nonlinear analyses of a pile-soil
system, difficulties arise from the impossibility of predetermining the location of the plastic zone. In
this paper, the distributed plastic hinge model was adopted along an expected plastic zone of a pile
(Chiou et al. 2009). The tributary length ldp of a plastic hinges is regarded as the plastic hinge length
as shown in Fig. 2(b). The yielding plastic hinges define an actual plastic zone. While the definition
of plastic curvature in ATC-40 is modified to

pm

M
EI e

(6)

where and M/EIe represent the total curvature and the elastic curvature, respectively.

0.5 Lp

Plastic hinge
on piers

(a) Pier

(b) Pile

Figure 2: The plastic hinges on RC members

3.

NUMERICAL MODLES

Nonlinear dynamic time history analyses and nonlinear static analyses (i.e., pushover) were
performed on a continuous bridge with two different boundary conditions of the foundation. The
first model assumes the bridge columns to be rigidly connected to the foundation without
consideration of the SSI. The second model incorporates SSI using nonlinear soil springs attached
4

on the pile foundation. The soil conditions considered in this paper is assumed to be consisted of
uniform deposit sand for simplicity.
3.1. Description of structure
The selected system is a four-span continuous bridge with equal pier height of 10 m. The length of
each span is 40 m and the total length is 160 m. The superstructure of the reference bridge, shown
in Fig. 3(a), consists of a concrete deck on top of four continuous steel girders and the
superstructure mass was 15 ton/m all throughout. The substructure consists of five piers supported
by RC pile foundations. Each pier has six piles which length is 40 m. The bearing systems are pin
supports at the intermediate piers P2, P3 and P4, and roller supports at expansion joints P1 and P5.
All piers are circular RC columns with a diameter of 2.5 m. Also, all piers have the same
reinforcement details of 74-D32 longitudinal reinforcing bars as shown in Fig. 3(b), and are
transversely reinforced with 19 mm bar hoop spaced at 8 cm. In the first model, the bridge columns
are rigidly connected to the foundation as shown in Fig 4(a). On the other hand, each column of the
bridge is supported by six piles which are circular RC columns with a diameter of 0.9 m as shown
in Fig. 4(b). All piles have the same reinforcement details of 16-D25 longitudinal reinforcing bars
and 13 mm bar hoop spaced at 30 cm for transversely reinforcement.
2.5 m

3
2

74-D32
D19@8cm

A = 5.27 m2I22 = 3.1394m4I33 = 39.0233m4

E = 2.49107 kN/m2m = 15 ton/m
(a) Superstructure

(b) Substructure

Figure 3: The cross section of the reference bridge.

(a) Model 1 (without piles)

(b) Model 2 (with piles)

Figure 4: Two finite element models of the considered bridge.

The stiffness of equivalent horizontal and vertical soil springs can be determined using empirical
formula with values of the standard penetration tests. In this paper, three different combinations of
N values and pile depth are selected to investigate the effects of soil types and buried depth of piles
on the seismic responses for the considered bridges. The corresponding fundamental periods of four
models are listed in Table 1. From Table 1, one can observe that the fundamental period of the
bridge with soil-structure interaction is almost twice of that of the bridge with fixed support. In
addition, the fundamental period of the considered bridge decreases with increasing the pile length
or the value of the standard penetration test. However, the effects of soil types and pile length on the
fundamental period are not obvious.
Table 1: Fundamental period for different finite element modes
Model no.

SPT- N

Pile length (m)

Fundamental period (sec)

Model 1

0.520

Model 2A

40

1.056

Model 2B

15

20

1.061

Model 2C

15

40

1.035

3.2. Nonlinear analyses

The nonlinear static pushover analyses and nonlinear dynamic time history analyses were carried
out using the commercial finite element software, SAP 2000N (CSI 2012). Superstructures were
simulated by elastic beam elements, and the nonlinear behaviors of piers and piles were modeled by
M3 hinge. The fundamental mode shapes in the lateral direction of these finite element models were
adopted as the lateral load patterns in static pushover analyses. Since the results of pushover
analyses depend on the choice of the monitored point, the concept of system displacement and
system mass (Kowalsky 2002) was recommended in the current study to calculate the equivalent
capacity curve. In addition, the design based earthquake (DBE with SDS = 0.7 and SD1= 0.52) and
maximum possible earthquake (MPE with SDS = 0.9 and SD1= 0.55) are considered in this study to
take the effects of earthquake magnitude into account. The results of nonlinear dynamic time
history analyses are the average displacements at deck nodes obtained by applying five
code-compatible artificial earthquakes on these bridges. A typical time history and the
corresponding elastic spectrum of a code-compatible artificial DBE are shown in Fig. 5.

(a) Time history

(b) Elastic spectrum

Figure 5: Typical time history and the corresponding elastic spectrum of DBE.
6

4.

RESULTS AND DISCUSSIONS

Figures 6 demonstrated that the pushover curves in the lateral direction form these four models. One
can observe that the fixed model (Model 1) has more stiffness than spring models (Model 2A, 2B,
and 2C) in the linear range. For bridge with fixed base, the pier response becomes nonlinear early in
the response while the yielding taking place with the lateral displacement approaching 0.09 m. On
the other hand, for bridges with equivalent soil-structure springs, the pier response is essentially
elastic till 0.18 m of lateral displacement. Moreover, the pushover curves of three different
combinations of soil types and pile length almost coincide with each other, which implies that the
soil types and pile length have little influence on the pushover curves.
16000

Load (kN)

12000

8000
Model 1 (without soil-structure interaction)
Model 2A (N=8, pile length = 40 m)
4000

Model 2B (N=15, pile length = 20 m)

Model 2C (N=15, pile length = 40 m)

0
0

0.1

0.2

0.3

0.4

0.5

Displacement (m)

Figure 6: Pushover curves for different finite element models.

Following the procedure of seismic assessment, the capacity spectra developed for different finite
element models and the performance points corresponding to the design-based earthquake and
maximum possible earthquake can be obtained. Based on the spectral displacement and spectral
acceleration at the performance point, the lateral displacements at decks level under the
design-based earthquake and maximum possible earthquake can be obtained for the multiple
monitored points, respectively. The calculated transverse displacement profiles from different
models are given in Fig. 7. In order to realize the performance of nonlinear static pushover analyses,
inelastic dynamic analyses were also performed and the corresponding results were shown in Fig. 7.
Considering the average of results of inelastic dynamic analyses as reference values, the transverse
displacements from static pushover analyses are smaller than those from dynamic analyses for the
fixed base model. However, the transverse displacements from static pushover analyses are larger
than those from dynamic analyses for the soil spring models. These trends can be observed both for
the design-base earthquake and maximum possible earthquake. These results imply that the static
pushover analyses for fixed base models may underestimate the maximum transverse displacements
and result in non-conservative designs. On the other hand, the static pushover analyses for models

with soil-structure interaction may overestimate the maximum transverse displacements and result
in conservative designs.
0.18

0.2

0.16

0.18
0.16

Displacement (m)

Displacement (m)

0.14
0.12
0.1
0.08

0.14
0.12
0.1
0.08

Model 1(Dynamic)

Model 2A (Dynamic)

Model 1(Dynamic)

Model 2A (Dynamic)

Model 2B (Dynamic)

Model 2C (Dynamic)

0.06

Model 2B (Dynamic)

Model 2B (Dynamic)

0.04

"Model 1 (Static)

Model 2A (Static)

0.04

Model 1 (Static)

Model 2A (Static)

0.02

Model 2B (Static)

Model 2C (Static)

0.02

Model 2B (Static)

Model 2C (Static)

0.06

0
0

40

80

120

160

40

80

120

160

Position (m)

Position (m)

(a) Design-based earthquake

(b) Maximum possible earthquake

Figure 7: Transverse displacement distribution on the desk of the considered bridge.

5.

CONCLUSIONS

In this paper, the influence of SSI on a typical regular continuous bridge is evaluated. It is shown
that the SSI leads to smaller accelerations and stresses in the structure. For bridge with fixed base,
the pier response becomes nonlinear early in the response. In addition, the soil types and pile length
have little influence on the pushover curves. However, the transverse displacement would be
magnified which could violate the performance-based design requirements. The magnification
factors are almost the same for design-based earthquake and maximum possible earthquake. Finally,
the static pushover analyses for models with soil-structure interaction may overestimate the
maximum transverse displacements and result in conservative designs.
REFERENCES
Jeremic B, Kunnath S, and Xiong F (2004). Influence of soil-foundation-structure interaction on seismic response of the
I-880 viaduct. Engineering Structures. 26(3), pp. 391-402.
Kappos AJ and Sextos AG (2001). Effect of foundation type and compliance on seismic response of RC bridges.
Journal of Bridge Engineering. 6(2), pp. 120-130.
Silva PF and Manzari MT (2008). Nonlinear pushover analysis of bridge columns supported on full-moment connection
CISS piles on clays. Earthquake Spectra. 24(3), pp. 751-774.
Kwon OS and Elnashai AS (2010). Fragility analysis of a highway over-crossing bridge with consideration of
soil-structure interactions. Structure and Infrastructure Engineering. 6(1-2), pp. 159-178.
Priestley MJN, Seible F, and Calvi GM (1996). Seismic Design and Retrofit of Bridges. Wiley Interscience Publications.
Fourth Edition. New York.
Kowalsky MJ (2002). A displacement-based approach for the seismic design of continuous concrete bridges.
Earthquake Engineering and Structural Dynamics, 31(3), pp. 719-747.