See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/228873191

Treatment of seismic input and boundary conditions in nonlinear seismic

analysis of a bridge ground system

Conference Paper · July 2003

CITATIONS READS

60 2,501

7 authors, including:

Zhaohui Yang Jacobo Bielak

University of Minnesota Twin Cities Carnegie Mellon University
34 PUBLICATIONS 2,584 CITATIONS 229 PUBLICATIONS 7,440 CITATIONS

SEE PROFILE SEE PROFILE

Joel P. Conte Ahmed-Waeil Elgamal

University of California, San Diego University of California, San Diego
259 PUBLICATIONS 6,789 CITATIONS 277 PUBLICATIONS 8,436 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Joel P. Conte on 28 May 2014.

The user has requested enhancement of the downloaded file.

 16th ASCE Engineering Mechanics Conference
July 16-18, 2003, University of Washington, Seattle

TREATMENT OF SEISMIC INPUT AND BOUNDARY CONDITIONS IN

NONLINEAR SEISMIC ANALYSIS OF A BRIDGE GROUND SYSTEM
1
Yuyi Zhang
1
Zhaohui Yang , Associate Member ASCE
2
Jacobo Bielak , Member ASCE
1
Joel P. Conte , Member ASCE
1
Ahmed Elgamal , Member ASCE
ABSTRACT
The Middle Channel Humboldt Bay Bridge near Eureka, California, was selected as a testbed in the
ongoing development of a performance-based earthquake engineering design and assessment methodology
at the Pacific Earthquake Engineering Research (PEER) Center. A two-dimensional, advanced nonlinear
computational model of the structure-foundation-soil system for the Middle Channel Bridge has been
developed using OpenSees, the new PEER software framework for seismic response simulation of
structural and geotechnical systems.
This paper presents and discusses a simplified, yet improved definition of the seismic excitation along
the boundaries of the soil domain included in the finite element model of the structure-foundation-soil
system. This seismic excitation assumes vertically incident shear waves and a linear elastic, undamped, and
homogeneous semi-infinite half-space underlying the nonlinear soil region modeled. It is also consistent
with free field (rock and soil) actual earthquake records representative of the seismicity of Humboldt Bay.
The proposed treatment of the boundary conditions includes transmitting/absorbing boundaries in order to
limit the occurrence of spurious seismic wave reflections along the boundaries of the modeled soil medium.
Using simple (single wavelet) and complex (earthquake-like) incident wave motions, the dynamic
response of the bridge-foundation-soil system obtained using the above modeling assumptions is compared
to that obtained under the common assumption of a total rigid base excitation in conjunction with shear
beam constraints across the longitudinal direction of the soil domain. The computational model developed
is exercised to assess the effects of soil nonlinearity (including lateral spreading) on the seismic response of
the structure-foundation-soil system.
Keywords: Bridge, structure-foundation-soil interaction, seismic input, boundary conditions,
transmitting/absorbing boundary

1
Department of Structural Engineering, University of California at San Diego, La Jolla, CA 92093. E-mails:
[email protected], [email protected], [email protected], [email protected].
2
Department of Civil & Environmental Engineering, Carnegie Mellon University, Pittsburgh, PA 15217. E-mail:
[email protected].
INTRODUCTION
The Humboldt Bay Middle Channel Bridge (Fig. 1) is one of three bridges that cross
Humboldt Bay on California State Route 255. Owned and maintained by the California
Department of Transportation (Caltrans), the bridge (no 04-0229) is a 330 meter, nine-span
composite structure with precast and prestressed concrete I-girders and cast-in-place concrete
slabs to provide continuity. It is supported on eight pile groups, each of which consists of 5 to 16
prestressed concrete piles, in soils potentially vulnerable to liquefaction (under extreme
earthquake shaking conditions). The river channel has an average slope from the banks to the
center of about 7% (4 degrees). The foundation soil is composed mainly of dense fine-to-medium
sand (SP/SM), organic silt (OL), and stiff clay layers (Fig. 2). In addition, thin layers of loose and
soft clay (OL/SM) are located near the ground surface (see Fig. 2). The bridge was designed in
1968 and built in 1971 and has been the object of two Caltrans seismic retrofit efforts, the first
one designed in 1985 and completed in 1987, and the second designed in 2001 to be completed in
2005. The objectives of the first retrofit effort were to mitigate the potential for unseating and
diaphragm damage, while those of the second retrofit are to strengthen the piers, pilecaps, and
pile groups. It is an interesting bridge because it is fairly representative of older
AASHTO-Caltrans girder bridges with moderate traffic loads (average daily traffic of 4,600 to
15,800 in 1992), designed before ductile detailing was common. Furthermore, since the bridge is
founded on poor soils, permanent ground deformation is a serious concern.
Borehole 1 at Caltran’s Samoa Bridge Geotechnical downhole array (~0.25 mile north-west of
the west abutment of the Middle Channel Bridge) provides a shear wave velocity profile down to
a depth of 220 meters, where the borehole encountered bedrock (shear wave velocity > 850
m/sec). The shear wave velocities are about 180 m/sec in the upper 20 meters, lie in the range of
200 to 400 m/sec in the depth range of 20 to 60 meters, and fall in the range of 400 to 600 m/sec
in the depth range of 60 to 220 meters (Somerville and Collins, 2002).

FIG. 1. Humboldt Bay Middle Channel Bridge (courtesy of Caltrans)

FINITE ELEMENT MODELING

A two-dimensional nonlinear model of the Middle Channel Bridge, including the
superstructure, piers, and supporting piles was developed using OpenSees (McKenna and Fenves,
2001), as shown in Fig. 2 (Conte et al., 2002). The bridge piers are modeled using 2-D nonlinear
material, fiber beam-column elements (with five Gauss-Lobato points along their length)
formulated using the flexibility approach based on the exact interpolation of the internal forces
(Spacone et al., 1996). The column cross-section is discretized into concrete and steel fibers. The
uniaxial Kent-Scott-Park constitutive model with degraded linear unloading/reloading stiffness is
used to model the concrete fibers, while the uniaxial bilinear model (or uniaxial J2 plasticity

2
model with linear kinematic hardening) is used to model the reinforcing steel. Confined and
unconfined concrete materials are modeled using different sets of material constitutive parameters.
The pile groups supporting the bridge piers are first lumped in the bridge transversal direction and
then modeled using 2-D nonlinear material, fiber beam-column elements (with two Gauss-Lobato
points in the length direction). The superstructure is idealized using equivalent linear elastic
beam-column elements. The interior expansion joints and the abutment joints are modeled using
zero-length elasto-plastic gap-hook elements. The soil domain is analyzed under the assumption
of plane strain condition. It is spatially discretized using four-noded, bilinear, isoparametric finite
elements with four integration points each. The materials in the various layers of the soil domain
are modeled using a multi-surface plasticity model incorporating liquefaction effects and
formulated in effective stresses (Elgamal et al., 2003). The lateral and base boundary conditions
for the computational soil domain are modeled using a modified Lysmer transmitting/absorbing
boundary. A set of viscous dampers normal and tangential to the soil boundaries is used to
implement these transmitting boundaries.

Eureka Channel Samoa Channel

(South-East) 1 2 3 4 5 6 7 8 (North-West)

A B

FIG. 2. OpenSees finite element model of structure-foundation-soil system

(based on blue prints courtesy of Caltrans, mesh generated using GID software)

LYSMER-KUHLEMEYER TRANSMITTING/ABSORBING BOUNDARY

For numerical accuracy, the maximum dimension of any soil element must be limited to
one-eighth thru one-fifth of the shortest wavelength considered in the analysis. For computational
efficiency, on the other hand, it is desirable to decrease the size of the soil domain included in the
finite element model. As the size of the soil domain decreases, the treatment of its boundary
conditions becomes increasingly more important. At the site of the Humboldt Bay Middle
Channel Bridge, bedrock lies about 220 meters below the ground surface. In reality, the wave
energy traveling away from the bridge site is dissipated by the soil medium and is never reflected
back into the computational domain. In the finite element model developed, the soil domain is
1100 meters in length and 75 meters in depth. Since the soil domain is of finite size, the
Lysmer-Kuhlemeyer boundary (Lysmer and Kuhlemeyer, 1969) is used to limit spurious wave
reflections at the soil mesh boundary. This boundary absorbs the propagating waves in such a way
that the incident wave is transmitted entirely into the soil domain of the finite element model
without distortion and no waves are transmitted back to the exterior domain.
The one-dimensional vertical shear wave equation can be written in the form

∂ 2 u ( x, t ) 2 ∂ u ( x, t )
2
= vs (1)
∂t 2 ∂x 2
where u denotes the soil particle displacement (perpendicular to the direction of wave propagation)

3
and v s = G ρ is the shear wave velocity. The solution of the above wave equation has the
form
x x
u ( x, t ) = u r (t − ) + u i (t + ) (2)
vs vs

where ur(…) and ui(…) can be any arbitrary functions of (t − x v s ) and (t + x v s ) , respectively.
The term u r (t − x v s ) represents the wave traveling at velocity vs in the positive x-direction,
while u i (t + x v s ) represents the wave traveling at the same speed in the negative x-direction.
Thus, ui is the incident wave if it points upwards into the computational domain and ur is the
reflected wave. Taking the partial derivative with respect to time of both sides of the above
equation and multiplying by ρv s gives

∂u ( x , t )
ρ vs = ρvs ur' (t − x vs ) + ρvs ui' (t + x vs ) (3)
∂t
where the prime superscript denotes the derivative of the associated function with respect to its
argument. Now, the linear elastic uniaxial shear stress - shear strain relation is given by
∂u ( x, t ) G G
τ ( x, t ) = G = − u r' (t − x v s ) + u i' (t + x v s ) (4)
∂x vs vs
where τ ( x, t ) is the shear stress. Combining equations (3) and (4), we get

∂u ( x, t )
τ ( x , t ) = − ρv s + 2 ρv s u i' (t + x / v s ) (5)
∂t
Note that ∂u ( x, t ) / ∂t represents the velocity of the total soil particle motion, while
u i' (t + x / v s ) = ∂u i (t + x / v s ) / ∂t is the velocity of the incident motion. Therefore, the first
term on the right hand side of equation (5) is equivalent to the force (per unit area) generated by a
dashpot of coefficient ρvs , while the second term is equivalent to a force (per unit area), which is
proportional to the velocity of the incident wave. As a result, the soil below the soil domain of
interest can be replaced with a dashpot and an equivalent force, which defines the seismic input.
In our two-dimensional finite element model, we assume that the response of the soil deposit is
predominantly caused by vertically propagating shear waves. We model the soil below the base
(at a depth of about 75 m) of the computational domain as a linear elastic, undamped, and
homogeneous semi-infinite half-space. It is expected that any nonlinearities below the base of the
computational domain would remain small, since this soil is significantly stiffer than the soil
within the computational domain. On each node at the base and on the lateral boundaries of the
soil domain, we add two dashpots, normal and tangential to the boundary, respectively. The
normal dashpots are set to absorb the reflected compressive waves while the tangential ones are
set to absorb the reflected shear waves. And on each node at the base, an equivalent force, which
is proportional to the velocity of the incident wave and the tributary surface area of the node, is
applied in the horizontal direction to represent the vertically incident SV-wave.

4
DEFINITION OF SEISMIC INPUT
As we stated above, the seismic input is defined in terms of equivalent nodal forces (or
effective earthquake forces), which are proportional to the velocity of the incident seismic wave,
applied in the horizontal direction along the base of the soil domain included in the finite element
model. Three sets of scaled actual earthquake ground motions corresponding to three different
hazard levels (50%, 10%, and 2% in 50 years, respectively) including both soil and rock site
conditions are used in the project (Somerville and Collins, 2002), yielding a total of 51 free
surface ground motions. These free field ground motions are consistent with the local seismology
at Humboldt Bay and with the magnitude-distance deaggregation of the seismic hazard, defined
as the probability of the spectral acceleration at the fundamental period of the bridge-ground
system exceeding a threshold level. For rock site free field ground motions, we assume that the
free surface motion represents the rock outcropping motion. In the case of vertically propagating
shear waves in a homogeneous, undamped, elastic half-space, the amplitude of the free surface
total motion is twice the amplitude of the incident motion at any point in the half-space (Kramer,
1996). Thus, for each of the rock site ground motions, we take as incident seismic wave half the
free surface ground motion. For the soil site free field ground motions, the software SHAKE
(Schnabel et al, 1972) is used to deconvolve these free surface motions in order to obtain the
corresponding incident wave motion at the base of the computational domain. For the
deconvolution operation, it is assumed that the soil region underlying the soil domain included in
the finite element model is homogeneous, viscoelastic, with material properties representative of
the exterior soil domain.

RESPONSE ANALYSIS
To illustrate and validate the above procedure for defining the seismic input, a rectangular,
homogeneous, undamped, elastic soil domain (1,100 by 40 meters in size) under plane stress
condition is discretized into the finite element model shown in Fig. 3. Horizontal and vertical
dashpots are defined at every node along the lateral and base boundaries of the soil domain. The
dashpot coefficients (per unit area) are defined as ρvs and ρv p (where v p denotes the
compressive or P-wave velocity) for the dashpots tangential and normal to the boundaries,
respectively.
Since we apply the results derived for one-dimensional wave propagation to a finite size
two-dimensional soil domain, the soil domain is made sufficiently long so that the response in the
center part of the mesh is close to its counterpart for the one-dimensional case. A sinusoidal
incident wave is applied at the base through a set of nodal equivalent forces. The total
acceleration response histories along the height of the two control soil columns shown in Fig. 3
are plotted in Fig. 4. We can clearly see the shear wave propagating vertically from the base to the
top of the soil domain and being reflected at the free surface. The amplitude of the total
acceleration on the ground surface is twice the amplitude of the incident wave. The vertical
acceleration at points along the height of each control soil column remains very small compared
to the horizontal acceleration at the same level, which verifies that the response of the soil domain
is predominantly caused by vertically propagating shear waves. Thus, the results of
one-dimensional wave propagation apply very well, even near the lateral boundaries (e.g., control
soil column # 1), to the propagation of vertically incident shear waves through the finite size
rectangular soil domain considered here and endowed with the Lysmer-Kuhlemeyer boundaries.

5
 Soil Column #1 Soil Column #2

FIG. 3. Rectangular homogenous linear elastic soil domain

It is more complicated to implement the Lysmer-Kuhlemeyer (L-K) transmitting/absorbing

boundaries in the nonlinear soil domain of the bridge-ground system. In fact, since confining
pressure dependent material plasticity models are used to model the behavior of the various soil
layers, lateral confinement is needed for these soil layers to develop some strength. However, the
dashpots cannot provide any lateral static constraint. Therefore, we need to proceed in several
steps to apply the L-K boundaries for a static gravity analysis followed by a dynamic (earthquake
response) analysis of the nonlinear bridge-ground system. First, we fix the base and lateral
boundaries of the soil domain, set the various soil constitutive models to be linear elastic, and
apply gravity. In the second step, the soil constitutive models are switched from linear elastic to
elastoplastic accounting for liquefaction effects and the new static equilibrium state under gravity
of the soil domain is obtained iteratively. Then the nonlinear material model of the bridge
structure and foundations is added to the model of the soil domain and the gravity of the bridge is
applied statically to the nonlinear bridge-ground system. In a third step, all the displacement
constraints along the boundaries of the soil domain are removed and replaced with the
corresponding support reactions recorded at the end of the second step. After balancing the
internal and external forces, dashpots in both the horizontal and vertical directions are added to
the base and lateral boundaries of the soil domain to model the L-K transmitting boundaries.
Finally, in a fourth step, the seismic excitation is applied, in the form of equivalent nodal forces
defined earlier, from the static equilibrium configuration under gravity loads.

(a) Horizontal acceleration (b) Horizontal acceleration

Ground surface motion Vertical acceleration Ground surface motion Vertical acceleration
Total acceleration

Total acceleration

Incident wave
Incident wave

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Time (sec) Time (sec)

FIG. 4. Wave propagation through the rectangular homogeneous linear elastic soil
domain, (a) control soil column # 1, (b) control soil column # 2

We investigated the wave propagation in the soil domain, shown in Fig. 5, of the bridge
ground system, modeled as both a multi-layered linear elastic undamped and elastoplastic soil
medium. A single Ricker wavelet with a dominant frequency of 3.9 Hz is used as the incident

6
wave, see Fig. 6. The total horizontal acceleration response histories at points along the height of
the two control soil columns shown in Fig. 5 are plotted in Fig. 6. Observe that the amplitude of
the linear response is greater than that of the nonlinear response and that the difference between
these two response amplitudes decreases with depth. In both cases, the total acceleration response
at the base of the soil domain is very close to that of the incident wave. For the linear case, we can
clearly see the reflected incident wave reaching the base level, while for the nonlinear case the
reflected wave dies out before it reaches the base. This is due to the fact that there is no source of
energy dissipation in the linear soil, while energy is dissipated through inelastic action in the
nonlinear soil.
Soil Column #1 Soil Column #2

FIG. 5. Soil mesh of the Humboldt Bay Middle Channel Bridge

(a) Linear soil (b) Linear soil

Nonlinear soil Ground surface motion Nonlinear soil
Ground surface motion
Total horizontal acceleration
Total horizontal acceleration

Incident wave Incident wave

0 0.5 1 1.5 2 0 0.5 1 1.5 2
Time (sec) Time (sec)

FIG. 6. Linear soil versus nonlinear soil total horizontal acceleration response (single Ricker
wavelet as incident wave, no bridge present), (a) control soil column # 1, (b) control soil
column # 2
Before implementing the transmitting/absorbing L-K boundaries in our current (second
generation) model of the Humboldt Bay bridge-foundation-soil system, in our first generation
model of this bridge system, we implemented a shear beam constraint across the horizontal
direction of the soil domain. According to this shear beam constraint, the two lateral boundaries
(vertical sides) of the soil domain are constrained to deform similarly in the lateral direction by
imposing equal horizontal degree of freedom constraints to pairs of lateral nodes at the same
elevation. Then, the seismic input is applied to all nodes at the base of the soil domain assuming
that the base moves rigidly according to the total ground acceleration input, thus neglecting
spatial variability of the seismic input. It is of interest to compare the response simulation results
obtained using the two types of soil boundary conditions and definition of the seismic input.
Comparative results are shown in Figs. 7, 8, and 9. Fig. 7 corresponds to simulation cases without
the bridge, while Figs. 8 and 9 show results of simulation cases with the bridge present. The total
horizontal acceleration response histories at points along the height of control soil columns # 1

7
and 2 due to a single Ricker wavelet incident motion applied to the nonlinear soil without the
bridge are shown in Fig. 7 for the two models of the soil boundary conditions and seismic input.
Fig. 7 indicates that the soil acceleration responses for the two treatments of boundary conditions
and seismic input are very close when the shear wave velocity of the soil below the computational
soil domain is high (i.e., vs = 600 m/sec). Then, an actual earthquake ground motion record (with
a 2% in 50 years hazard level), was applied as incident wave motion to the bridge-ground system.
Fig. 8 compares the soil total horizontal acceleration response at points along the height of control
soil columns # 1 and 2 for the two models of the soil boundary conditions and seismic input. The
transmitting/absorbing L-K boundaries were based on a shear wave velocity of vs = 600 m/sec. A
more significant difference between the soil responses for the two modeling assumptions is
observed than in the case of the single wavelet incident motion (see Fig. 7). This could be due to
the presence of the bridge structure and its foundations and/or to the fact that earthquake
excitation has a greater frequency bandwidth than the Ricker wavelet considered and that high
frequency waves are more sensitive to soil heterogeneities than relatively low frequency waves.
Figs. 9 (a) and (b) show the shear stress – shear strain soil response during the earthquake at
control points A and B, respectively, of the computational domain (see Fig. 2) for the two models
of the soil boundary conditions and seismic input. Comparison of the responses of the bridge
structure itself for both modeling assumptions is of the highest interest in this study. As an
illustration, Fig. 9(c) shows the top-to-bottom relative horizontal displacement of the bridge
column # 4 (see Fig. 2), while Fig. 9(d) displays the moment-curvature history of the base section
of the bridge column # 1 (see Fig. 2). From Figs. 9, it is clear that the treatment of the soil
boundary conditions and seismic input has a significant effect on the simulated local response of
the soil material and response of the bridge structure itself.

(a) Shear Beam

Transmitting/absorbing,Vs=600m/sec
(b) Shear Beam
Transmitting/absorbing,Vs=600m/sec
Transmitting/absorbing,Vs=300m/sec Transmitting/absorbing,Vs=300m/sec
Total horizontal Acceleration

Total horizontal Acceleration

Incident wave Incident wave

0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 0.2 0.4 0.6 0.8 1 1.2 1.4
Time (sec) Time (sec)

FIG. 7. Nonlinear soil total horizontal acceleration response at points along the two control
soil columns (single Ricker wavelet as incident wave, no bridge present), (a) control soil
column #1, (b) control soil column # 2

8
 Shear Beam
Total horizontal acceleration (a) Transmitting/absorbing
(b) Shear Beam
Transmitting/absorbing

Total horizontal acceleration

Incident wave Incident wave
0 5 10 15 20 0 5 10 15 20
Time (sec) Time (sec)

(c) Shear Beam

Transmitting/absorbing (d) Shear Beam
Transmitting/absorbing
Total horizontal acceleration

Total horizontal acceleration

Incident wave Incident wave

8 8.5 9 9.5 10 10.5 11 11.5 12 8 8.5 9 9.5 10 10.5 11 11.5 12
Time (sec) Time (sec)

FIG. 8. Nonlinear soil total horizontal acceleration response at points along the two control
soil columns (earthquake ground motion as incident wave, bridge present),
(a) control soil column # 1, (b) control soil column # 2 , (c) zoom of (a), (d) zoom of (b)

10 80
(a) 60
(b)

5 40
Shear stress (KPa)
Shear stress (KPa)

20

0 0

−20

−5 −40

Shear Beam Boundary −60 Shear Beam Boundary

Transmitting/Absorbing Boundary Transmitting/Absorbing Boundary
−10 −80
−0.002 0 0.002 0.004 −0.004 −0.002 0 0.002 0.004 0.006
Shear strain Shear strain

9
 Column #4 Base section of column #1
0.2 10000
Shear Beam Boundary
(c) Transmitting/Absorbing Boundary 8000 (d)
0.15
6000

Moment (KN−M)
4000
Moment (KN−M)

0.1

2000
0.05
0

0 −2000

−4000
−0.05
−6000 Shear Beam Boundary
Transmitting/Absorbing Boundary
−0.1 −8000
0 10 20 30 40 50 60 70 −0.004 −0.002 0 0.002 0.004 0.006 0.008
Time (sec) Curvature

FIG. 9. Nonlinear soil and bridge structure responses (earthquake motion as incident wave),
(a) shear stress –shear strain soil response at control point A (see Fig. 2), (b) shear stress –
shear strain soil response at control point B (see Fig. 2), (c) top-bottom horizontal relative
displacement of bridge column # 4, (d) moment-curvature at base section of bridge column
#1

CONCLUSIONS
This paper briefly describes a 2D nonlinear finite element model of the
structure-foundation-soil system for the Middle Channel Bridge at Humboldt Bay near Eureka in
Northern California. This model was developed using OpenSees, the new software framework for
seismic response simulation of structural and geotechnical systems developed by the Pacific
Earthquake Engineering Research (PEER) Center. This paper especially focuses on the
implementation of the Lysmer-Kuhlemeyer transmitting/absorbing boundary for this bridge
model. Several response simulations were conducted to verify this implementation. The effect of
linear elastic versus elastoplastic soil constitutive behavior on the propagation of vertically
incident shear waves in the computational domain is examined. It is found that soil nonlinearity
has a significant effect on the wave propagation through the foundation soil. The difference
between two approximate modeling approaches for the soil boundary conditions and definition of
the seismic input was investigated in terms of local and global, soil and bridge structure responses
to vertically incident shear waves. These two modeling approaches are: (1) shear beam constraints
across the longitudinal direction of the computational soil domain used in conjunction with “rigid
soil” excitation under the base of the computational soil domain; and (2) use of
Lysmer-Kuhlemeyer transmitting/ absorbing boundaries along the lateral sides and base of the
computational soil domain in conjunction with a definition of the seismic input in terms of
vertically incident shear waves. It is found that these two modeling approaches can produce quite
different responses of the bridge structure itself.
This study is part of the PEER Center ongoing development of a probabilistic seismic
performance assessment methodology for civil structures such as buildings and highway bridges.
This methodology integrates probabilistic seismic hazard analysis, advanced computational
modeling of the bridge ground system, probabilistic seismic demand analysis, and probabilistic
damage analysis or fragility analysis (Cornell and Krawinkler, 2000).

10
 ACKNOWLEDGMENTS
Support of this work was provided primarily by the Earthquake Engineering Research Centers
Program of the National Science Foundation, under Award Number EEC-9701568 through the
Pacific Earthquake Engineering Research Center (PEER). This support is gratefully
acknowledged. The authors wish to thank Mr. Patrick Hipley, Mr. Cliff Roblee, Mr. Charles
Sikorsky, and Mr. Mark Yashinsky of Caltrans for providing all the requested information
regarding the initial design and retrofits of the Middle Channel Bridge. Prof. Greg Fenves, Dr.
Frank McKenna, Mr. Michael Scott (U.C. Berkeley) and Prof. Boris Jeremic (U.C. Davis) helped
with the OpenSees modeling and analysis framework. Prof. Pedro Arduino (Univ. of Washington,
Seattle) suggested and helped with the use of GID (http://gid.cimne.upc.es/), a commercial pre-
and post-processing system for geometric modeling and visualization. Their help is gratefully
acknowledged.

REFERENCES
Conte, J.P., A. Elgamal, Z. Yang, Y. Zhang, G. Acero, and F. Seible (2002), "Nonlinear Seismic
th
Analysis of a Bridge Ground System," (CD-ROM), Proceedings of the15 ASCE Engineering
Mechanics Conference, Columbia University, New York, June 2-5, 2002.
Cornell, C.A. and H. Krawinkler (2000), "Progress and Challenges in Seismic Performance
Assessment," PEER Center News, 3(2), Spring 2000.
Elgamal, A., Z. Yang, E. Parra, and A. Ragheb (2003), "Modeling of Cyclic Mobility in Saturated
Cohesionless Soils," International Journal of Plasticity, 19(6), 883-905.
Kramer, S. L. (1996), Geotechnical Earthquake Engineering, Prentice Hall, Upper Saddle River,
New Jersey, USA.
Lysmer, J., and R.L. Kuhlemeyer (1969), “Finite Dynamic Model for Infinite Media,” Journal of
the Engineering Mechanics Division, ASCE, 95(EM4), 859-877.
McKenna, F. and G.L. Fenves (2001), "The OpenSees Command Language Manual, Version 1.2,"
Pacific Earthquake Engineering Research Center, University of California at Berkeley,
Berkeley, CA (http://opensees.berkeley.edu/ ).
Schnabel, P.B., J. Lysmer, and H.B. Seed (1972), “SHAKE: A Computer Program for Earthquake
Response Analysis of Horizontally Layered Sites”, Report No. EERC 72-12, Earthquake
Engineering Research Center, University of California, Berkeley, CA.
Somerville, P., and N. Collins (2002), “Ground Motion Time Histories for the Humboldt Bay
Bridge,” Report of the PEER Performance Based Earthquake Engineering Methodology
Testbed Program, Pacific Earthquake Engineering Research Center, University of California
at Berkeley, Berkeley, CA (http://peer.berkeley.edu/ ).
Spacone, E., F.C. Filippou, and F.F. Taucer (1996), " Fibre Beam-Column Model for Non-Linear
Analysis of R/C Frames: Part I. Formulation," Earthquake Engineering and Structural
Dynamics, 25 (7), 711-725.

11

View publication stats