______________________________________________________________________________

The 2001 NEHRP Professional Fellowship Report

PRELIMINARY STUDY ON A
SIMPLIFIED RESPONSE SPECTRUM METHOD
FOR INCOHERENT GROUND MOTIONS OF BRIDGES

Greg Griffin
M. Saiid Saiidi

r1 x g 1 r2 x g 1

x g1
 PRELIMINARY STUDY ON A
SIMPLIFIED RESPONSE SPECTRUM METHOD
FOR INCOHERENT GROUND MOTIONS OF BRIDGES

Greg Griffin
M. Saiid Saiidi

June 9, 2003

______________________________________________________________________________

This research was supported by the 2001 Professional Fellowship,

which was funded by the Federal Emergency Management Agency and administered by the
Earthquake Engineering Research Institute.

PF2001-10
Greg Griffin, Associate Bridge Engineer, CH2M HILL in Boise, Idaho, was selected as
the Year 2001 National Earthquake Hazard Reduction Program (NEHRP) Professional
Fellow, awarded by EERI under a cooperative program funded by the Federal Emergency
Management Agency (FEMA). Griffin's research was conducted with the guidance of
Professor M. Saiid Saiidi, of the University of Nevada, Reno.
Griffin has worked as a consulting engineer since 1995 and is employed by CH2M HILL.
He earned his BSCE from the University of Idaho, and his MSCE from the University of
Nevada, Reno.

Cover graphic: Depicts the pseudo-static displacement response in the simplified two
degree-of-freedom model used to study the effects of incoherent ground motion. Unlike
uniform base motion, the different displacement history response at each support during
an earthquake will induce forces within a bridge.
 ACKNOWLEDGEMENTS

Mr. Griffin would like to thank the Earthquake Engineering Research Institute for their

generous support during the 2001 EERI/FEMA NEHRP Professional Fellowship. Mr.

Griffin would also like to thank the staff of the Civil Engineering Department at the

University of Nevada, Reno, for their assistance during the Fellowship period. The

efforts of Dr. Saiid Saiidi, his assistance and insight throughout this research project are

greatly appreciated. Dr. Raj Siddarthan, Professor of Civil Engineering, is also thanked

for his explanation of the equivalent linear ground response model. Dr. Qing-Shan Yang,

visiting professor from Northern Jiaotong University, Beijing, China, is thanked for his

insight into the coherency model.

ii
 TABLE OF CONTENTS

EXECUTIVE SUMMARY...............................................................................................vii

CHAPTER 1 - INTRODUCTION ...................................................................................... 1

1.1 Introduction .............................................................................................................. 1
1.2 Sources of Incoherency ............................................................................................ 2
1.2.1 Response History Analysis Methods..................................................................... 3
1.2.2 Response Spectrum Analysis Method................................................................... 4
1.3 Review of Previous Work ........................................................................................ 4
1.4 Object and Scope...................................................................................................... 5

CHAPTER 2 - ANALYTICAL MODELS ......................................................................... 7

2.1 Introduction .............................................................................................................. 7
2.2 Two Degree-of-Freedom System ............................................................................. 8
2.2.1 Equation of Motion Formulation........................................................................... 9
2.2.2 Influence Functions ............................................................................................. 11
2.2.3 Modal Response History Method........................................................................ 12
2.2.4 Model Verification .............................................................................................. 14
2.3 Ground Response Model........................................................................................ 15
2.3.1 Equivalent Linear Method................................................................................... 16
2.3.2 Nonlinear Analysis Methods............................................................................... 17
2.3.4 Model Verification .............................................................................................. 20
2.4 Response Spectrum Method................................................................................... 21

CHAPTER 3 - PARAMETRIC STUDIES ....................................................................... 23

3.1 Introduction ............................................................................................................ 23
3.2 Coherency Function Models .................................................................................. 24
3.3 Frequency Parameters ............................................................................................ 25
3.3.1 Coupling Ratio .................................................................................................... 26
3.3.2 Ground Motions and Soil Parameters ................................................................. 27
3.4 Results .................................................................................................................... 28
3.5 Limitations of the Parametric Study ....................................................................... 29

CHAPTER 4 - SUMMARY AND CONCLUSIONS ....................................................... 31

4.1 Summary ................................................................................................................ 31
4.2 Conclusions ............................................................................................................ 32
4.3 Future Research...................................................................................................... 33

REFERENCES.................................................................................................................. 35

APPENDIX A - MAGNIFICATION FACTORS

APPENDIX B - COMPUTER MODEL SOURCE CODE

iii
 LIST OF TABLES

Table 2.1 - Parameters used to verify the computer model .............................................. 36

Table 3.1 -: Masses used for the 2 DOF model................................................................. 36

Table 3.2 - Shear beam stiffnesses used for the 2 DOF model ......................................... 37

Table 3.3 - Soil deposit depth combinations ..................................................................... 37

iv
 LIST OF FIGURES

Figure 2.1 - 2 DOF model ................................................................................................. 38

Figure 2.2 - Pseudo-static displacement response............................................................. 38

Figure 2.3 - Dynamic displacement response comparison at DOF 1................................ 39

Figure 2.4 - Total displacement response comparison at DOF 1 ...................................... 39

Figure 2.5 - Ground motion response model..................................................................... 40

Figure 2.6 - Variation of damping ratios for sands ........................................................... 41

Figure 2.7 - Variation of shear modulus for sands ............................................................ 42

Figure 2.8 - Theoretical primary curve for a sand deposit ................................................ 43

Figure 2.9 - Primary curve used in the computer model for a sand deposit...................... 43

Figure 2.10 - Ground motion response - Northridge earthquake (Scale factor = 1) ........ 44

Figure 2.11 - Ground motion response - Northridge earthquake (Scale factor = 4) ........ 44

Figure 3.1 - Predicted absolute ground acceleration : Taft earthquake............................. 45

Figure 3.2 - Predicted absolute ground acceleration : Taft earthquake............................. 45

Figure 3.3 - Predicted absolute ground acceleration : El Centro earthquake.................... 46

Figure 3.4 - Predicted absolute ground acceleration : El Centro earthquake.................... 46

Figure 3.5 - Predicted absolute ground acceleration : Northridge earthquake.................. 47

Figure 3.6 - Predicted absolute ground acceleration : Northridge earthquake.................. 47

Figure 3.7 - Predicted relative ground displacement : Taft earthquake ............................ 48

Figure 3.8 - Predicted relative ground displacement : Taft earthquake ............................ 48

Figure 3.9 - Predicted relative ground displacement : El Centro earthquake ................... 49

Figure 3.10 - Predicted relative ground displacement : El Centro earthquake ................. 49

Figure 3.11 - Predicted relative ground displacement : Northridge earthquake ............... 50

v
Figure 3.12 - Predicted relative ground displacement : Northridge earthquake ............... 50

Figure 3.13 - Average magnification factors : Taft earthquake ........................................ 51

Figure 3.14 - Average magnification factors : El Centro earthquake ............................... 51

Figure 3.15 - Average magnification factors : Northridge earthquake ............................. 52

Figure 3.16- Average magnification factors : All earthquakes ......................................... 52

vi
 EXECUTIVE SUMMARY

Many instances arise in bridge design where non-synchronous ground motions occur at

different support locations. Although several sources of incoherent ground motion are

recognized, differing soil conditions at intermediate supports along a bridge can occur

frequently. Several methods are available to predict bridge response to incoherent

ground motions, but are typically cost prohibitive to implement in standard bridge

designs. Unlike a bridge that is excited with uniform ground motion, the behavior of

bridges undergoing non-synchronous ground motions are not well understood. For

bridges subjected to incoherent ground motions, both the ground acceleration and

displacement contribute to the overall bridge response. To better understand bridge

response due to the site-response effect, a study funded by the EERI/FEMA Professional

Fellowship was undertaken.

To simplify the analysis and understand the pseudo-static displacement and

acceleration behavior, a response history computer model for a two degree-of-freedom

(DOF) system was developed. A shear beam connected both DOF to represent the

superstructure of a bridge. Superposition of the pseudo-static and acceleration response

was used to determine the total DOF response. A SDOF, nonlinear ground response

model was used to predict the site-response effect. An iteration scheme using the

Newton-Raphson method was implemented and verified with an equivalent linear

stiffness method. Parametric studies were conducted considering different structure

masses and shear beam stiffnesses. Three different earthquake ground motions were used

and the soil depths were varied at the supports. A stiffness parameter, defined as the

Coupling Ratio (CR), was defined to quantify the effect of the shear beam stiffness on the

vii
overall structure behavior. CR varies from 0 (each DOF responds independently) to 1

(rigid body response). Magnification factors were calculated as the ratio of the relative

displacement at each DOF due to the incoherent ground motion to the uniform base

motion response. The magnification factors can be used to account for ground motion

incoherency by modifying the response of a two DOF system subjected to uniform base

motion.

viii
 CHAPTER 1

INTRODUCTION

1.1 Introduction

Many instances arise in bridge design that cause differing earthquake ground motions at

supports of a structure. Accurate estimates of seismic induced forces are critical to

provide reliable designs for superstructure and substructure elements. To simplify the

seismic force calculations, design codes assume that all supports have the same intensity

of ground shaking, although many bridges have conditions that modify the intensity and

frequency content of seismic shaking (Ref. 1). Bridge responses to incoherent earthquake

ground motions are due to a dynamic portion and differing support movements (denoted

as “pseudo-static” displacements). Due to the nature of non-synchronous effects on

structures, it is not apparent whether the dynamic or pseudo-static response will increase

bridge forces as compared to uniform base motions. Some studies suggest incoherency

cancels peak amplitudes and reduces force levels. Another study indicates that the

pseudo-static response increases force levels in stiff structures.

One method to determine the effects of incoherent ground motion has been

developed using response history analysis. Using this method, the analytical model is

subjected to differing earthquake ground motions and the response histories of the bridge

elements are calculated. A drawback of the response history method is determining

representative earthquake ground motions for the site and different supports. Depending

upon the site and bridge length, considerable costs can be involved in developing ground

motions that encompass the design earthquake. Typical bridge designs for short and

medium span lengths do not use response history analysis due to the increase in cost to

1
develop and implement site-specific ground motions. The most common method used to

determine dynamic response in bridges is the response spectrum method (RSM).

Widespread acceptability of the RSM is due to the relative simplicity in definition of the

design earthquake and the use of the linear elastic analysis to simulate actual behavior

during earthquakes. The RSM currently used does not account for non-synchronous

ground motions.

1.2 Sources of Incoherency

Four different sources of incoherency (Ref. 5) are typically recognized in seismic ground

motions: 1) the loss of coherency due to the scattering of seismic waves and the

superposition of waveforms from differing sources (denoted as the “incoherency effect”),

2) difference in arrival times of the waveform as they propagate through the underlying

strata (denoted as the “wave passage effect”), 3) decay of wave amplitude with distance

due to spreading and energy dissipation (denoted as the “attenuation effect”) and 4)

variation of waveforms emanating from a similar bedrock layer through differing soil

strata, which modify the amplitude and frequency of the waveforms at the base of the

structure (denoted as the “site-response effect”). The attenuation effect has been shown

to have limited influence on the coherency function and can be ignored (Ref. 6). In this

study, only the site-response effect will be considered as the source of non-synchronous

ground motion. It is believed that other effects can be critical only for relatively long

bridges.

2
1.2.1 Response History Analysis Methods

Response history analysis methods have been developed and are available to estimate the

effects of incoherent ground motion (Ref. 3). The equations of motion can be written to

include the contribution of individual supports to the degrees-of-freedom within the

structure. A complete formulation of the equations of motion to include non-

synchronous ground motions will be presented in Chapter 2. Fenves (Ref. 4) used the

computer program SAP90 (Ref. 5) with response history capabilities to study the

Dumbarten Bridge and compare analytical to measured response due to the Loma Prieta

earthquake. Because SAP90 does not have the capability to include the effects of non-

synchronous ground motions, a fictious spring at each support was introduced to model

the incoherency using time varying loads. Currently, state-of-the-art software packages,

such as SAP2000 (Ref. 6), are able to account for differing support ground motions using

the free-field ground displacement histories as input to the analytical model. Results of

the study indicate that the Dumbarten Bridge was not sensitive to incoherent ground

motion, but indicate that the pier displacements increased somewhat as compared to

uniform base motion. This study was limited to incoherency due to the site-response

effect. It was concluded that the non-synchronous ground motions did not substantially

increase forces levels for that particular bridge. Due to similar ground displacement and

acceleration histories along the length of the bridge at different supports, the overall

bridge response should not differ greatly from the uniform base motion response.

3
1.2.2 Response Spectrum Analysis Method

The response spectrum method (RSM) is used in practice to conduct bridge dynamic

analysis. An inherent assumption of the RSM is that all supports of the structure are

excited with the same earthquake ground motion. Der Kiuregian (Ref. 5) has developed

one method to estimate the response to incoherent motion called the Multiple Support

Response Spectrum (MSRS) rule. This method was developed based on random

vibration theory and solution of the governing equations of motion in the frequency

domain instead of the time domain. All sources of differing ground motions are depicted

using frequency-based coherency functions, and structure response quantities are

determined using correlation coefficients to determine support contribution of response

due to each degree of freedom. The MSRS rule has significant computation requirements

to determine the cross correlation coefficients. Results from two separate parametric

studies conducted on four span bridges indicate that 1) an envelope of the site response

spectrums can be unconservative and should not be used to estimate the seismic response,

and 2) a single-degree-of-freedom ground response model can reasonably predict site

response.

1.3 Review of Previous Work

Past studies of incoherent ground motions have mainly been project specific using

response history analysis methods. Some researchers have been developing simplified

models to better understand the components of bridge behavior under non-synchronous

ground motions. Mylonakis has studied a simplified model with sinusoidal support

movements and presented observations (Ref. 2). Studies on multi-span bridges have been

4
limited to the longitudinal axis assuming that the deck is rigid. Very few in-depth studies

have been conducted to determine the effect of incoherent ground motions on bridges and

to develop methods or design guidelines that can be readily implemented in design. The

remainder of this section will discuss response history analysis methods, and response

spectrum methods.

1.4 Object and Scope

The main objective of this study was to provide the basis for a practical RSM that can

account for non-synchronous ground motions to estimate bridge response due to the site-

response effect. Another goal of this study was to provide insight into bridge response

due to incoherent ground motions using a simplified computer model. To study the non-

synchronous effects, a two-degree-of-freedom computer model using response history

analysis was developed. A nonlinear, SDOF ground response model was developed to

estimate the site-response effect. Using the results of parametric studies, factors were

calculated that compare the incoherent to the uniform base motion response. The factors

can be applied to a typical response history analysis results for uniform base motion to

estimate the bridge response under incoherent ground motion. Parameters varied in the

study were the mass and shear beam stiffness, while three different earthquakes and

multiple soil depths were used.

Chapter 2 describes the two-degree-of-freedom computer model used in the study.

Development of the non-linear and equivalent linear SDOF ground response model is

also presented in Chapter 2. Chapter 3 describes a brief overview of a coherency

function model to provide insight into the complexities of incoherent ground motions.

5
Chapter 3 also discusses the parametric study variables used for the bridge model and site

conditions, and presents the results and observations from the study. A summary of the

study, important conclusions and recommendations for future work are presented in

Chapter 4. An appendix has been included with graphs that present the magnification

factors developed in the parametric study. Another appendix is included with portions of

the computer code developed for the two-degree-of-freedom model and the non-linear

SDOF ground response model.

6
 CHAPTER 2

ANALYTICAL MODELS

2.1 Introduction

To estimate the response of bridges during earthquakes, several different methods are

available. As a state-of-the-practice, a linear elastic analysis is conducted using the

Response Spectrum Method (RSM). An inherent assumption in the RSM is that all the

supports of the structure are subjected to the same ground motion. For more complicated

or important bridges, response history analysis may be used to determine the seismic

response. Because force levels in bridges may be affected by non-synchronous ground

motion, an accurate response analysis should be conducted to properly size bridge

members and foundations.

Structural analysis software has not been readily available to estimate incoherent

ground motion response. One researcher used fictious springs at the supports to simulate

the pseudo-static support displacements (Ref. 4). Recently, finite element analysis

packages have been developed to calculate the dynamic response due to incoherent

ground motions using the response history method (Ref. 7). A response history analysis

is typically not feasible to produce cost-effective bridge designs for ordinary span length

bridges. A method that has been proposed to conduct dynamic analysis using incoherent

ground motions is the Multiple Support Response Spectrum (MSRS) rule (Ref. 7). Due

to the complexity in determination of different factors and input ground motion response,

this method has not received widespread use in bridge design. As a simplified method,

several researchers have suggested that the site response coefficients may be averaged to

account for incoherency (Ref. 8).

7
 To study the effects of incoherency, a simplified two degree-of-freedom (DOF)

system consisting of two masses and springs connected by a shear beam was developed.

To gain a better basic understanding of bridge response, a detailed finite element model

of different bridges was not considered. In-depth parametric studies that vary the bridge

mass and stiffness and the soil depths were conducted and will be presented in Chapter 3.

In this chapter, the two DOF system connected with a shear beam will be discussed.

Development of the equations of motion will be presented using the superposition

method and solution using the displacement histories. Influence functions will be derived

for a generalized structure and applied to the simplified model. Development of the

modal response history method using modified mode participation factors will be

presented. A single DOF soil column response model was developed using a non-linear

and equivalent linear method and will be presented. An overview of the response

spectrum method will be discussed.

2.2 Two Degree-of-Freedom System

To understand and study the effects of incoherent ground motion, a two DOF model

connected with a shear beam was developed and shown in Figure 2.1. A 2 DOF model

was selected for study to reduce the complexities of the analysis and allow the pseudo-

static and dynamic response components to be studied. Each DOF in the model consists

of a mass connected by a support spring and a shear spring. The support springs

represent the substructure of the bridge including the foundation springs. Superstructure

stiffness is represented with the shear beam spring. As stated previously, this study will

be limited to only the site-response effect.

8
2.2.1 Equation of Motion Formulation

To determine the response of a structure due to incoherent earthquake ground motions,

the total response can be determined from a superposition of the dynamic response and

the pseudo-static ground displacement response. An alternative formulation of the

governing equations of motion can be solved using the support ground displacement

history. Using the superposition method, the governing equations for the dynamic

response of the structure due to ground acceleration at support n can be described as

[M ]{x }+ [C ]{x }+ [K ]{x }= −[M ][r ]x

dyn dyn dyn gn
(2.1)

in which M is the mass matrix, C is the damping matrix, K is the stiffness matrix, xdyn ,

x dyn , x dyn are the structure acceleration, velocity and displacement, respectively, relative

to the undeflected position, r is the influence factor matrix and xgn is the ground

acceleration at support n. Equation 2.1 can be solved using a direct integration approach

such as Newmark’s Beta Method or superposition of a modal analysis. For this study, the

equations of motion was solved using modal analysis. The pseudo–static displacement

response at DOF i due to support n can be determined by partitioning of the stiffness

matrix into nodes at DOF and support nodes as

 K ff K fs  U f   Pf 
K  = 
K ss  U s   Ps 
(2.2)
 sf

in which the subscripts f denote a DOF and s denotes a support node. Assuming that

external loads are not applied to the structure, the displacement at DOF i can be

determined due to a unit displacement at support n described as

9
K ff U ff + K fsU ss = 0 (2.3)

Equation 2.3 was developed using the first matrix equation in equation 2.2. Solving

equation 2.2.3 for the displacement at the free node

U st = U ff = − K −ff1 K fsU ss (2.4)

in which Ust are the pseudo-static displacements. Equation 2.4 describes the

displacement at the structural DOF due to support displacements (denoted as “pseudo-

static displacement”) and shown in Figure 2.2. For the case in which Uss is equal to a

unit displacement, Uff is defined as the influence functions and will be developed for the

2 DOF system in the following section. The total response at the structural DOF can be

determined using superposition

U T = U dyn + U st (2.5)

in which UT are the total displacements. An alternate formulation of the governing

equations of motion can be developed using

[M ]U T
+ [K ]U dyn = 0 (2.6)

in which damping effects are considered negligible. Substituting equation 2.5 into

equation 2.6

[M ]U T
+ [K ]U T = [K ]rU ss (2.7)

Solution of equation 2.7 will provide the total displacement at the structural degrees-of-

freedom due to dynamic and pseudo-static response. In equation 2.7, Uss must be defined

as the total bedrock and free-field ground displacement history. Otherwise, the dynamic

contribution will not include the response due to the bedrock ground motion.

10
2.2.2 Influence Functions

The influence functions are defined as the displacement at DOF i due to a unit

displacement at support n and can be determined using equation 2.4. For the two DOF

system, the stiffness matrix can be developed

1  K1 − K1 
2  − K 1 K 1 + K SB − K SB 

K= (2.8)
3 − K SB K 2 + K SB − K2 
 
4 − K2 K2 

in which K1 and K2 are the support spring stiffnesses and KSB is the shear beam stiffness.

Rearranging the matrix to form the partitions defined in equation 2.2

2  K 1 + K SB − K SB − K1 
3  − K SB K 2 + K SB − K2 

K= (2.9)
1  − K1 K1 
 
4 − K2 K2 

Note that nodes 2 and 3 are the DOF and nodes 1 and 4 are the restraints. Using the

partitioned sub-matrices from equation 2.9 and substituting into equation 2.2 the

influence function matrix is defined as

2  K ( K + K SB ) K 2 K SB  1 
r=  1 2    (2.10)
3 K 1 K SB K 2 ( K 1 + K SB )  K 1 K 2 + K 2 K SB + K 1 K SB 

A property of the influence matrix is

ntot

∑r
n =1
in = 1.0 (2.11)

in which the subscript i denotes the DOF, n denotes the support and ntot are the total

number of supports. It should also be noted that the ground acceleration distributes from

11
an individual support to the structural DOF using the influence function matrix due to the

fact that the acceleration is a function of the displacement.

2.2.3 Modal Response History Method

Unlike the direct integration response history analysis method in which the governing

equations are solved to determine the total response of the DOF, in modal analysis the

response of individual modes are superimposed to determine the total response. By

assuming that damping effects are negligible, the ground acceleration is zero, and

defining x = −ω x , equation 2.1 can be reduced to

2

[K − ω 2
n M ]{φ in }= 0 (2.12a)

Det (K − ω n2 M ) = 0 (2.12b)

in which ω is the natural circular frequency for mode n, and φ are the modal

displacements or mode shape at DOF i for mode n. Equation 2.12b will yield a non-

trivial solution to equation 2.12a in which modal frequencies and modal displacements

can be calculated. To determine the response of individual modes, the displacement at

DOF i, xi, can be described as

Tot
xi = ∑ φ ni z n (2.13)
n =1

in which z is the displacement response of mode n. To determine the dynamic response

for mode n, equation 2.13 is substituted into equation 2.1 and multiplied by φ mT

φ mT m ∑ φ n zn  + φ mT c ∑ φ n z n  + φ mT k  ∑ φ n z n  = −φ mT rx g
 Tot   Tot   Tot 
(2.14)
 n =1   n =1   n =1 

12
in which equation 2.14 and equation 2.1 are termed the coupled equations of motion,

which refers to the fact that the solution must be obtained by direct integration of all

equations of motion. To uncouple the equations of motion such that the dynamic

response of the structure can be determined using the superposition of results from

individual modes, orthogonality relationships can be developed

∑φ T
m mφ n = 0 (2.15)

∑φ T
m cφ n = 0 (2.16)

∑φ T
m kφ n = 0 (2.17)

Equations 2.15 through 2.17 are valid when m ≠ n . Applying the orthogonality

relationships in equations 2.15 through 2.17 to equation 2.14 and dividing the resulting

equation by the modal mass, φ T mφ ,

z + 2ε iω i z + ω i2 x = −γ pm xg (2.18)

in which  is the damping coefficient for mode i, and γ pm are the modified mode

participation factors defined as

 φ T mr 
γ pm =  T  (2.19)
 φ m φ 
For uniform base motion, r is defined as a matrix vector of 1’s and corresponds to the

uniform distribution of acceleration to all DOF. For the case in which non-synchronous

ground motions are considered, r is defined using equation 2.10. Total response for the

structure can be determined using equation 2.5 when a response history analysis is

13
conducted. Equation 2.18 can be recognized as the response of a single degree-of-

freedom (SDOF) oscillator. When non-synchronous ground motion response is

considered, equation 2.18 will further reduce to the SDOF response due to individual

support ground motions. The total number of SDOF oscillators to be considered are

dependant on the number of modes to be included in the analysis and the number of

supports with similar input ground motions.

2.2.4 Model Verification

TM
A Windows based computer program, IGMR (Incoherent Ground Motion Response),

was developed to calculate the total response due to incoherent ground motions using the

response history analysis method. A superposition of the dynamic and pseudo-static

displacement response using modal analysis was implemented. To verify the accuracy of

the computer implementation, results from IGMR were compared with SAP2000 (Ref.

7). To conduct an incoherent ground motion response analysis using SAP2000, the

following steps are done:

1) Apply unit displacements at each support with similar time history functions.

SAP2000 will numerically evaluate the influence function matrix, and

2) Assign displacement time histories to the supports. SAP2000 uses the displacement

response history method in equation 2.7 to determine the total DOF response.

Parameters used in the structural model are listed in Table 2.1. Bedrock ground motion

used was the 1952 Taft earthquake. Figure 2.3 depicts the displacement response history

at DOF 1 for the uniform ground motion case using a 50-foot deep soil deposit and

indicates excellent agreement. Free-field ground motion displacement histories were

14
applied to the different supports in SAP2000 using the above procedure. Figure 2.4

depicts the displacement response history using non-synchronous ground motion due to a

soil column of 50-feet at support 1 and 150-feet at support 2. Displacement response

histories calculated show good agreement. Minor differences in peak displacement can

be noted and are attributed to inaccuracies that arise in the numerical integration.

2.3 Ground Response Model

Four different sources of incoherency were briefly presented in Chapter 1, although for

this study, only the site-response effect will be considered. To determine the free-field

ground response, the following assumptions were made:

1) Bedrock ground motion at each support are the same,

2) Underlying soil strata can be defined as a shear beam, and

3) Soil deposit response is due to vertically propagating shear waves.

To estimate the free-field ground motion response, a SDOF model was developed and

implemented into IGMR. One study recommends that a MDOF model should be used

and divided into numerous layers depending upon the fundamental period of the deposit

(Ref. 9). These findings were developed based upon comparison of results from a

lumped mass model to a distributed parameter model. Past studies (Ref. 7) have shown

that the SDOF soil model reasonably approximates a more detailed analysis. Figure 2.5

depicts the ground motion response model. The equation of motion of the soil mass can

be described as

x + 2ξωx + ω 2 x = − xg (2.20)

15
in which x , x , x are the relative ground acceleration, velocity and displacement,

respectively, as compared to the bedrock, ξ is the damping ratio, ω is the natural

circular frequency of the deposit, and xg is the bedrock acceleration. Using a lumped

mass assumption for the SDOF model, ω can be defined as

Vs
ω= 2 (2.21)
H

in which Vs is the shear wave velocity of the deposit and H is the depth of the soil

deposit. Vs can be defined from G = ρVs2, in which G is the shear modulus and ρ is the

soil density. Theoretical studies (Ref. 9) have shown that the frequency of the soil

deposit with distributed parameters is defined as

π V
ω =  s (2.22)
2H

Within a soil deposit, the damping ratio and shear modulus are nonlinear functions that

vary with shear strain. The experimental data used in this study are shown in Figures 2.6

and 2.7. The damping ratio represents the hysteretic damping due the shear strain in the

deposit. To estimate the free-field response of the soil deposit, equation 2.20 can be

solved using an equivalent linear or nonlinear analysis method.

2.3.1 Equivalent Linear Method

Due to complexities in nonlinear analysis, Seed and Idriss (Ref. 9) proposed an

equivalent linear method to predict soil deposit response. Both the soil deposit stiffness

and damping in this model can be represented using an equivalent spring and damper

16
based on the maximum shear strain within the soil deposit. The equivalent linear method

can be conducted using the following steps:

1) Assume a shear strain, γi, in the soil deposit,

2) Determine Gi and εi based on γi,

3) Calculate ωi using equation 2.22,

4) Calculate the response history using equation 2.20 and determine γmax,

5) Determine the average cyclic shear stress, γavg = γi+1 = 0.65γmax,

6) Check for convergence of the solution

γi
1− γ i +1 < Tol the solution has converged. G = Gi and ε = εi.

γi γ = γ i +1 .
1−
γ i +1 > Tol the solution has not converged. Go to step 2) and use i

in which Tol is defined as the tolerance for the solution.

In step 5 of the above procedure, the constant 0.65 represents the ratio of the average

uniform shear stress to the maximum shear stress during the earthquake (Ref. 12).

2.3.2 Nonlinear Analysis Methods

Two different nonlinear methods are available to predict the free-field ground motion

response. One method to account for the nonlinear behavior of the soil deposit can be to

model the soil stiffness with an appropriate model such as a bilinear model (Ref. 9). The

second method to account for nonlinear soil behavior is to use a secant stiffness method

in which the shear modulus and damping factors are updated at the end of each

integration time step during the response analysis. The benefit to using a nonlinear

model is that energy dissipation and stiffness variations are intrinsically included in the

17
analysis. Any model that represents the actual stress-strain curves for the soil could be

implemented in the equation of motion.

In this study, nonlinear behavior of the soil deposit was modeled with the secant

stiffness method using the modulus reduction and damping coefficient curves. An

iteration scheme at the end of each time step was introduced to solve for the soil response

in equation 2.20. This method results in the determination of the effective shear modulus

and damping ratio at the end of each time step. The primary curve for a given soil

deposit can be constructed using the modulus reduction curve and plotting values of shear

force and displacement defined as

τ
G= (2.23)
ε

∆
ε= (2.24)
H

in which τ is the shear stress in the deposit, ε is the shear strain in the deposit, and ∆ is

the displacement at the top of the soil deposit. Substituting equation 2.24 into equation

2.23 and solving for the shear stress results in the force per unit area of soil column

Geff
F= ∆ (2.25)
H

in which Geff / H represents the effective stiffness and

Geff = Gmax (G RED ) (2.26)

in which Gmax is the low strain shear modulus and GRED is a modulus reduction factor.

To solve equation 2.20 using the nonlinear method, the following procedure was used:

1) Solve equation 2.20 for the incremental displacement response, dxi, using an initial

value for G and ε,

18
2) Determine an improved estimate of the displacement using the Newton-Raphson

Second Method,

−1
 f ′′( xi ) f ′( xi ) 
dxi +1 = dxi +  −  (2.27)
 2 f ′( xi ) f ( xi ) 

in which dxi+1 is the incremental displacement for iteration i+1.

3) Define the functionals in equation 2.27 as

f (dxi ) = ω 2* dxi − a * (2.28)

6 6ε i ω i
ω 2* = 2
+ + ω i2 (2.29)
h h

6 x
a* = − ∆xg + (1 + ε iωh ) + x(3 + ε iω i h) (2.30)
h

in which equation 2.28 was developed using equation 2.20 and assuming a linear

acceleration distribution.

1
f ′(dxi ) = [− f (dxi − 3h) + 9 f ( x − 2h) − 45 f ( x − h)
60h (2.31)
+ 45 f ( x + h) − 9 f ( x + 2h) + f ( x + 3h)]

1
f ′′(dxi ) = [2 f ( x − 3h) − 27 f ( x − 2h) + 270 f ( x − h) − 490 f ( x)
180h 2 (2.32)
+ 270 f ( x + h) − 27 f ( x + 2h) + 2 f ( x + 3h)]

in which the functional expressions in equation 2.31 and equation 2.32 are derived using

a central difference evaluation for the first and second derivative, respectively (Ref. 10).

4) Check convergence of the solution.

dxi +1 − dxi < Tol then solution has converged, continue using step 1)

dxi +1 − dxi > Tol then use dxi+1 in step 2) and repeat.

19
2.3.4 Model Verification

To verify the ground response model, two different approaches were taken:

1) Develop the primary curve using the equations 23 through 26 and compare with the

force-displacement response from the computer model using the nonlinear method.

2) Compare the ground displacement response history using the equivalent linear

method and nonlinear method.

The soil deposit was assumed to be 100 feet deep and use the modulus reduction

and damping factors shown in Figures 2.6 and 2.7. For all cases, the Northridge

earthquake was used as the bedrock ground motion, and the soil deposit was assumed to

be sand using G = 2,500 ksf. Excellent correlation of the theoretical primary curve in

Figure 2.8 and the force-displacement response of the soil deposit shown in Figure 2.9

can be noted. Agreement of the primary curve calculated using equations 2.23 through

2.26 as compared to the force-displacement response of the computer model indicate that

the iteration scheme employed for the Nonlinear Method converges correctly. Figure

2.10 shows the free-field displacement time history response using the equivalent linear

and nonlinear methods. Although the frequency content differs, both methods calculate

the maximum ground displacement of approximately 0.12 feet. This result is expected

because the equivalent linear method was developed to capture the maximum ground

response. Applying a scale factor of 4.0 to the Northridge Sylmar earthquake, Figure

2.11 indicates that the equivalent linear method predicts a maximum ground displacement

of approximately 50% greater than the nonlinear method. Although this is an unexpected

20
result, the nonlinear method should estimate lower ground displacements due to

significant nonlinear behavior associated with the large bedrock accelerations.

2.4 Response Spectrum Method

Unlike the response history method in which the governing equations are solved to

determine the DOF response, the Response Spectrum Method (RSM) combines the

maximum response of individual modes to estimate the response because the maximum

values do not occur simultaneously. To determine the dynamic displacement, xi, the

following mode combination methods can be used

∑ (φ γ pn z n )
Tot
xi =
2
ni
(2.33)
n =1

∑∑α (φ γ pn )(φ miγ pm )ρ nm z m z n

Tot Tot
xi = nm ni
(2.34)
n =1 m =1

in which φ is the mode shape for mode n at DOF i, γp is the mode participation factor for

mode n, z is the maximum modal response for mode n, and α and ρ are factors to

correlate the response of mode m and mode n. Equation 2.33 is defined as the Square

Root of the Sum of the Squares (SRSS) method and equation 2.34 is defined as the

Complete Quadratic Combination (CQC) method. A complete derivation of the SRSS

and CQC methods are available in Ref. 3.

For non-synchronous ground motion response, the displacement at each DOF can

be estimated using

xic = γ MF xuni (2.35)

21
in which xic is the maximum displacement response due to incoherent ground motion, γMF

are proposed magnification factors to account for the incoherent ground motion, and xuni

is the maximum displacement response calculated using equation 2.33 or 2.34. γMF are

empirically derived factors that are a function of the soil deposit, ground motion and

structural frequency and will be discussed in Chapter 3.

22
 CHAPTER 3

PARAMETRIC STUDIES

3.1 Introduction

Prediction of earthquake force levels in bridges is critical to properly size members to

resist seismic forces. The effects of non-synchronous ground motions have not been well

understood due to the nature of the response. When a bridge is assumed to be excited

with uniform base motion, the total response of the structure can be determined using

well-established dynamic theories. During non-synchronous ground motion, the bridge

response consists of dynamic and pseudo-static support components. Bridge response

due to incoherent ground motions can be predicted using a response history analysis as

described in Chapter 2. However, in the state-of-the-practice, the Response Spectrum

Method (RSM) is the analysis procedure that is most commonly used to evaluate bridge

response due to seismic loading. Difficulty arises in the RSM to correlate not only the

dynamic response of the structure due to different support ground motions, but also the

pseudo-static displacement response. To explain the behavior of the non-synchronous

ground motion, past researchers have used coherency function models to correlate the

out-of-phase waves from different input earthquake motions. Although these models

have been shown to accurately predict the correlation between different ground motions,

definition of the coherency function models require an analysis using response history

functions of the earthquake ground motions. If actual or synthetic time histories are

available, response history analysis methods will provide the best results.

The purpose of the parametric studies is to identify different aspects of incoherent

ground motions that affect bridge response. A simplified bridge model with two degrees-

23
of-freedom (DOF) was studied to better gain an understanding of non-synchronous

ground motion. This chapter will discuss a brief overview of coherency function models,

frequency parameters used in the study, the derivation of the coupling ratio as a measure

of structure stiffness, parameters used to model the soil columns, results from the

parametric studies, and limitations of the parametric study.

3.2 Coherency Function Models

To understand and accurately solve a structure subjected to asynchronous ground motion

using a RSM, coherency of the input ground motions must be defined. A coherency

function model that has been proposed (Ref. 7) can be expressed as

Gu k ul (ω )
γ kl (ω ) =
 
(3.1)
Gu k uk (ω )Gul ul (ω )
   

in which γ kl (ω ) is a complex valued function and defines the statistical dependency of

ground motions k and l, and G is the auto-spectral densities at k and l. Interpretation of

the coherency function is that when γ kl = 1 the ground motions are identical and in-phase

at frequency ω . Equation 3.1 can be re-written as

γ kl (ω ) = γ kl (ω ) exp[iθ kl (ω )] (3.2)

in which γ kl is the bounded modulus and varies 0 ≤ γ kl ≤ 1 , and θ kl (ω ) is termed as

the phase angle between the motions at k and l and is defined as

Im γ kl (ω )
θ kl (ω ) = (3.3)
Re γ kl (ω )

in which Im is the imaginary and Re is the real parts of γ kl (ω ) , respectively. It has been

shown that θ kl (ω ) characterizes the site-response effect (Ref. 7). Significant

24
computational effort is required to develop the coherency function using equations 3.1

through 3.3 from a site response analysis.

3.3 Frequency Parameters

To determine the non-synchronous ground motion effects on bridge response, a wide

range of dynamic parameters were used. For a certain range of mass and stiffness

properties with fundamental periods near the predominant period of the earthquake

ground motion, it would be expected that the dynamic portion of the response would be

dominant. In flexible structures having periods that are much larger than the predominant

earthquake motions, it is expected that the pseudo-static displacements would dominate

the bridge response. To quantify the range for typical bridge types, bridges with

fundamental periods, Tp, ranging from 0.5s ≤ Tp ≤ 3.0s were considered. To simplify the

parameters, the period ranges studied were based on the uncoupled periods of each DOF.

In each period range considered, the ratio of the periods (TR) for the two DOF were

varied from

TR = T2 T1 = 01
. ,0.4,0.7,10
. (3.4)

in which T1 and T2 are the uncoupled periods for DOF 1 and DOF 2, respectively. In all

cases, the damping ratio, ε , was assumed to be 5% and the spring support stiffness was

100 k/ft. Changing the masses at the DOF varied the period range and ratios. To account

for the shear beam stiffness effect to the overall DOF response, a new parameter denoted

as the Coupling Ratio (CR) was developed and will be discussed in the following section.

For each period range and period ratio, CR was varied

CR = 0.1,0.4,0.7,1.0 (3.5)

25
A CR = 0 indicates that each DOF responds as a single-degree-of-freedom oscillator,

while a CR = 1.0 indicates that the bridge responds as a rigid body. Mass and shear beam

stiffness parameters used in the study are listed in Table 3.1 and Table 3.2, respectively.

3.3.1 Coupling Ratio

As mentioned in the previous section, a parameter that was developed to quantify the

effect of the shear beam stiffness was the Coupling Ratio and is defined as

 T2 
 T1  c
CR = 1 − (3.6)
 T2 
 T1  u

in which CR is defined as the Coupling Ratio and varies from 0 ≤ CR ≤ 1 , T1 and T2 are

the periods of the 2 DOF systems, and the subscripts c and u denote the coupled and

uncoupled period ratios, respectively. Due to the shear beam, the DOF 1 response must

be solved considering the response of DOF 2. Equation 3.6 was developed based on the

following observations of a two DOF system. For a system with an extremely flexible

shear beam,

 T2  ≈  T2  (3.7)
 T1  c  T1  u

For a system with an extremely stiff shear beam,

 T2  ≈ 0 (3.8)
 T1  c

Equation 3.8 approaches 0 because the two DOF’s do not respond independently and

approach rigid body response. Several examples representative of CR approaches 1 are :

1) In the longitudinal direction due to the high axial stiffness of the superstructure, and 2)

26
In the transverse direction with a high aspect ratio (width of bridge deck / length of

bridge).

3.3.2 Ground Motions and Soil Parameters

Three different bedrock ground motions were used in the parametric study: the 1952 Taft,

the 1940 El Centro, and the 1994 Northridge earthquakes. As mentioned earlier,

differing soil conditions at the supports were considered as the sole source of

incoherency. For each bedrock ground motion, ground motion response was predicted

for soil depths ranging from 50-feet to 300-feet deep. The parametric studies limited the

soil depth difference at the supports to 100-feet based on practical limits encountered in

practice. Table 3.3 summarizes the soil depths considered. The nonlinear ground

response model described in Chapter 2 was used to provide the most accurate prediction

of the free-field ground motion response. Absolute acceleration predicted from the

ground response model is compared to each the bedrock motion in Figure 3.1 through

Figure 3.6. Amplification of the ground acceleration was noted for the Taft and

Northridge earthquakes, but the El Centro earthquake was attenuated. Ground

displacement relative to the bedrock was predicted for each earthquake and compared in

Figure 3.7 through Figure 3.12. Calculation of the absolute ground displacement time

history was not required to predict the pseudo-static support response of the bridge.

For all soil deposits considered, the low-strain shear modulus, G, was equal to

2,500 ksf and the unit weight of soil was equal to 120 pcf. To apply the results of the

parametric study for deposits with different soil properties, an equivalent soil column

height, HEQ, can be determined using

27
 819 H
H EQ = (3.9)
Vs

in which H is the actual soil depth in feet, and Vs is the shear wave velocity of the actual

deposit in ft/sec. Equation 3.9 was the result of equating the natural frequency of the

deposit using the actual and the parametric study parameters (equation 2.22).

3.4 Results

Parametric studies were conducted using the frequency parameters discussed in this

chapter. Using the computer program IGMR, discussed in Chapter 2, the response

histories were calculated with the different earthquake ground motions and soil depth

variations. To quantify the effects of non-synchronous ground motion, magnification

factors were determined using

γ MF = δ IC δ UNI (3.10)

in which γMF is the magnification factor, δIC is the maximum relative displacement at

DOF i due to the incoherent ground motion, and δUNI is the maximum relative

displacement at DOF i due to the uniform base motion. γMF is an empirically derived

response coherency function that includes site amplification effects and pseudo-static

support displacements. Values for γMF are presented in Appendix A. Non-synchronous

relative displacement response can be determined with equation 2.35 for a given uniform

base response. To estimate the non-synchronous response with a different earthquake

ground motion, an averaged γMF should be used. Although the response history analyses

were conducted using three specific earthquakes, certain trends can be established upon

observation of γMF:

28
1) Based on the average magnification factors of earthquakes, increased force levels can

be expected (Figure 3.13 through Figure 3.15) due to the incoherent ground motion.

For all earthquakes with characteristic periods less than 1.5 seconds, the

magnification factors change rapidly and appear to be associated with the large

contribution of the dynamic response component. As the dynamic response of the

structure becomes smaller for larger values of T1, the magnification factors tend to

become constant. For the Northridge earthquake (Figure 3.15), significant frequency

contents are present at higher periods that increase the dynamic displacement

contribution as compared to the other earthquakes. To explain the significance of the

dynamic displacement, see observation 2.

2) As the fundamental period (T1) of the structure approaches the characteristic period

(Tpe) of the ground motion, a uniform seismic base input will yield the greatest bridge

response (Figure 3.16). This figure was developed based on the average

magnification factors of all earthquakes and frequency parameters. This result can be

expected based on the Dynamic Response Factor, R, from elementary structural

dynamic theory (Ref. 3). For frequency ratios (ωpe / ω) significantly less than 1, R

approaches 1. R approaches infinity (ε = 0) for ω = ωpe, while R becomes 0 for large

values of ωpe / ω. In the previous discussion, ω is the natural circular frequency of

the structure and ωpe is the predominant frequency of the forcing function. Stiff

structures would typically have increased force levels for large values of ωpe / ω.

3.5 Limitations of the Parametric Study

As discussed in this chapter, an in-depth parametric study was conducted using the two

DOF model. The limitations to this study are:

29
1) The two DOF model precludes the contribution of higher modes. In some instances,

including higher modes may not affect the overall response greatly. Higher mode

response will affect the dynamic contribution only, while the pseudo-static

displacement response will be independent of the number of modes included in the

analysis.

2) Mass variations were used to achieve the different frequency parameters while the

support spring stiffness remained constant. Differing support springs would cause the

influence function matrix to be unsymmetrical and the pseudo-static displacement

response would be different for a given T1, TR and CR.

30
 CHAPTER 4

SUMMARY AND CONCLUSIONS

4.1 Summary

Bridge response to uniform ground motion is generally well understood and can be

estimated using a response history or response spectrum method. For bridges with

different input ground motions at the supports, the expected behavior is not as apparent.

Response history methods are currently available to account for the incoherency, but can

be difficult and cost prohibitive to use for routine bridge design. During non-

synchronous ground motions, structure response consists of both a dynamic response

portion and pseudo-static support displacements. Four different sources of incoherency

typically considered are the incoherence effect, the wave passage effect, the attenuation

effect and the site-response effect. This study was limited to incoherency due to the site-

response effect. To understand the effects of incoherent ground motion on bridge

response, a simplified two DOF response history computer model was developed.

Ground response was estimated from a nonlinear, SDOF response model. Parametric

studies were conducted using three different earthquake ground motions and soil deposit

depths. Objectives of the study were 1) to provide insight into bridge response due to the

incoherency from the site-response effect, and 2) to provide the ground-work for a

practical response spectrum method that can account for non-synchronous ground

motions.

Chapter 2 described the development and verification of the computer model used

to study the site-response effects. Total response of each DOF was determined from

superposition of the pseudo-static and dynamic response. Pseudo-static support

31
displacements can be calculated using the influence function matrix. Influence functions

define the displacement at each DOF due to a unit support displacement and can be

developed using stiffness matrix partitioning. Development of the equations of motion

shows that only the mode participation factors need to be modified using the influence

functions. A nonlinear, SDOF ground response model using shear modulus and damping

factors was developed. An equivalent linear method was used to verify the nonlinear

model.

Chapter 3 described the parametric studies conducted using the two DOF model.

To provide insight into the complexities of non-synchronous ground motions, a

coherency function model was discussed that correlates the waveforms of different input

ground motions in the frequency domain. This model can be used to correlate the input

ground motions of two separate functions at different frequencies. Response history

analysis does not require the use of coherency functions. Parametric studies were

conducted that varied the masses at the DOF, shear beam stiffness, and soil deposit depth.

Three different earthquakes were selected and free-field ground motions were determined

using the nonlinear ground response model described in Chapter 2. Magnification factors

were calculated that compared the maximum relative displacement responses at each

DOF due to the incoherent and uniform base motions.

4.2 Conclusions

Practicing bridge engineers typically use the response spectrum method to estimate

earthquake-induced forces. Response history analysis is usually impractical and cost

prohibitive for most bridge designs. Although this study did not result in a response

32
spectrum method to account for incoherent ground motions, insight into the site-response

effect on bridge response was observed from the parametric studies:

1) Increased force levels were predicted in period ranges less than 1.5 seconds based on

average magnification factors for three different earthquakes. Dynamic effects

appeared to contribute to rapid changes in the magnification factors. When the

dynamic response was not significant, the magnification factors approached a

constant level. This result indicates that bridge columns would have increased

ductility requirements and additional reinforcement may be required to increase the

ductility capacity. Due to a potential increase in column over-strength forces,

increased foundation costs could result.

2) As the fundamental period of the bridge approached the characteristic period of the

ground motion, uniform base motion developed greater force levels. This result can

be expected because the Dynamic Response Modification factor increases

dramatically when the frequency of the bridge and ground motion are close. As a

result, uniform base motion would predict maximum column ductility demands and

incoherent ground motion response could be ignored.

3) As the fundamental period of the bridge exceeded the characteristic period of the

ground motion, combination of different soil depths at the supports did result in

amplification and reduction of forces. No apparent trends were noted.

4.3 Future Research

Additional studies that should be addressed to implement this study in bridge design

practice are:

33
1) A response history analysis of an actual bridge should be compared with a response

history analysis of an equivalent two DOF system to show the applicability of the

method.

2) Additional studies need to be conducted to quantify the effect of different spring

support stiffnesses. Magnification factors developed in Chapter 3 were based on

varying the DOF mass instead of the support spring stiffness. In an actual bridge, an

equivalent two DOF system would most likely have different masses and support

springs. Different support spring stiffnesses will affect the influence function matrix

and distribution of the ground acceleration and pseudo-static support displacements.

3) Higher modes of an actual bridge that could be excited during the incoherent

earthquake motions should be studied to determine the possible limitations of the two

DOF model.

4) Based on the parametric study results, determine the contribution of the dynamic and

pseudo-static responses to the overall response. Separation of the effects will confirm

the observations made in Chapter 3, and may simplify implementation into a response

spectrum method.

5) Use the parametric study results to develop response spectrum modification factors.

Site response factors for each soil deposit and earthquake can be determined and the

contribution to the overall response can be calculated.

6) Use the methodology developed in item 5 to apply the site response factors to a

design response spectrum. To verify the approach, develop ground motions that

simulate the design response spectrum and compare response history results to the

simplified method.

34
 REFERENCES

1) AASHTO. “Standard Specifications for Highway Bridges,” 16th Ed., American

Association of State Highway And Transportation Officials, Washington D.C.,

2) G. Mylonakis, “Simplified Modeling of Bridge Response on Soft Soil to Nonuniform

Seismic Excitation,” Journal of Bridge Engineering,Vol. 6, No. 6, Nov./Dec. 2001.

3) R.W. Clough and J. Penzien, Dynamics of Structures, 2nd Ed., McGraw-Hill Inc.,
C1993.

4) G. Fenves, “Earthquake Analysis and Response of Multi-span Bridges and Viaduct

Structures,” Proceedings, Seismic Design and Retrofit of Bridges, June 1992.

5) Computers and Structures, Inc., “SAP90 – Structural Analysis Program,” version

5.10, copyright 1978 – 1989.

6) Computers and Structures, Inc., “SAP2000 – Structural Analysis Program,” version

7.4, copyright 1984 – 2000.

7) Der Kiureghian, P. Keshishian, A. Hakobian, “Multiple Support Response Spectrum

Analysis of Bridges Including the Site-Response Effect & the MSRS Code,” Report No.
UCB/EERC-97/02, College of Engineering, University of California, Berkeley, Feb.
1997.

8) M.J.N. Priestly, F. Seible, G.M. Calvi, Seismic Design of Retrofit of Bridges, John
Wiley & Sons Inc., C1996.

9) H.B. Seed and I.M. Idriss, “Soil Moduli and Damping Factors for Dynamic Response
Analysis,” Report EERC 70-10, Earthquake Engineering Research Center, University of
California, Berkeley, 1970.

10) I..M. Idriss and H. B. Seed, “Seismic Response of Horizontal Soil Layers,” Journal of
the Soil Mechanics and Foundations Division, Vol. 94, No. SM4, July 1968.

11) A.W. Al-Khafaji and J.R. Tooley, Numerical Methods in Engineering Practice, HRW
Publishing, C1986

12) H.B. Seed and I.M. Idriss, Ground Motions and Soil Liquefaction During
Earthquakes, EERI Monograph Series, C1982.

35
 Table 2.1 : Parameters used to verify the computer model
2 2
M1 (k-s /ft) K1 (k/ft) KSB (k/ft) M2 (k-s /ft) K2 (k/ft)
0.6333 100 305 0.1013 100

Table 3.1 : Masses used for the 2 DOF model

T1 (s) TR M1 (k-s2/ft) M2 (k-s2/ft)
0.50 0.10 0.633 0.006
0.40 0.633 0.101
0.70 0.633 0.310
1.00 0.633 0.633
1.00 0.10 2.533 0.025
0.40 2.533 0.405
0.70 2.533 1.241
1.00 2.533 2.533
1.50 0.10 5.699 0.057
0.40 5.699 0.912
0.70 5.699 2.793
1.00 5.699 5.699
2.00 0.10 10.132 0.101
0.40 10.132 1.621
0.70 10.132 4.965
1.00 10.132 10.132
2.50 0.10 15.831 0.158
0.40 15.831 2.533
0.70 15.831 7.757
1.00 15.831 15.831
3.00 0.10 22.797 0.228
0.40 22.797 3.648
0.70 22.797 11.171
1.00 22.797 22.797

36
Table 3.2 : Shear beam stiffnesses used in the 2 DOF model
TR (s) CR KSB (k/ft)
0.10 0.10 76
0.40 392
0.70 2,030
1.00 100,000
0.40 0.10 62
0.40 305
0.70 1,550
1.00 100,000
0.70 0.10 39
0.40 186
0.70 940
1.00 100,000
1.00 0.10 12
0.40 89
0.70 505
1.00 100,000

Table 3.3 : Soil deposit depth combinations

Support i Support j
Rock Rock
50'
100'
50' 50'
100'
150'
100' 100'
150'
200'
150' 150'
200'
250'
200' 200'
250'
300'

37
 KSB
M1 M2

K1 K2

Figure 2.1 : 2 DOF Model

r1 x g 1 r2 x g 1

x g1

Figure 2.2 : Pseudo-static Displacement Response

38
 0.040

SAP2000
0.030
IGMR

0.020
Displacement (feet)

0.010

0.000

-0.010

-0.020

-0.030

-0.040
15

20
Time (s)

Figure 2.3 : Dynamic displacement response comparison at DOF 1

0.025
SAP2000
0.020 IGMR

0.015

0.010
Displacement (feet)

0.005

0.000

-0.005

-0.010

-0.015

-0.020

-0.025
15

20

Time (s)

Figure 2.4 : Total displacement response comparison at DOF 1

39
 x g , x g , xg

xrock

Figure 2.5 : Ground Motion Response Model

40
41
42
 0.80

0.60

0.40

0.20
Force (k)

0.00

-0.20

-0.40

-0.60

-0.80
-0.10 -0.05 0.00 0.05 0.10
Displacement (ft)

Figure 2.8 : Theoretical primary curve for a sand deposit

0.80

0.60

0.40

0.20
Force (k)

0.00

-0.20

-0.40

-0.60

-0.80
-0.10 -0.05 0.00 0.05 0.10
Displacement (ft)

Figure 2.9 : Primary curve used in the computer model for a sand deposit

43
 0.10
Equivalent Linear
Nonlinear

0.05
Displacement (ft)

0.00

-0.05

-0.10

-0.15
10

15

20
Time (s)

Figure 2.10 : Ground motion response - Northridge earthquake (Scale = 1)

0.60

Equivalent Linear
0.40 Nonlinear

0.20
Displacement (ft)

0.00

-0.20

-0.40

-0.60
10

15

20

Time (s)

Figure 2.11 : Ground motion response - Northridge earthquake (Scale = 4)

44
 0.20

0.15

0.10

0.05
Acceleration (g's)

0.00

-0.05

-0.10

-0.15

-0.20

Rock 50 Foot Depth 100 Foot Depth 150 Foot Depth

-0.25
10

11

12

13
8

Time (s)

Figure 3.1 : Predicted Absolute Ground Acceleration : Taft Earthquake

0.15

0.10

0.05
Acceleration (g's)

0.00

-0.05

-0.10

-0.15

-0.20 Rock 200 Foot Depth 250 Foot Depth 300 Foot Deposit
10

11

12

13
8

Time (s)

Figure 3.2 : Predicted Absolute Ground Acceleration : Taft Earthquake

45
 0.15

0.10

0.05
Acceleration (g's)

0.00

-0.05

-0.10

Rock 50 Foot Deposit 100 Foot Deposit 150 Foot Deposit

-0.15
3

6
Time (s)

Figure 3.3 : Predicted Absolute Ground Acceleration : El Centro Earthquake

0.15

0.10

0.05
Acceleration (g's)

0.00

-0.05

-0.10

Rock 200 Foot Deposit 250 Foot Deposit 300 Foot Deposit
-0.15
3

Time (s)

Figure 3.4 : Predicted Absolute Ground Acceleration : El Centro Earthquake

46
 0.60

0.40

0.20
Acceleration (g's)

0.00

-0.20

-0.40

-0.60 Rock 50 Foot Depth 100 Foot Depth 150 Foot Depth
15

16

17

18

19

20
Time (s)

Figure 3.5 : Predicted Absolute Ground Acceleration : Northridge Earthquake

0.50

0.40

0.30

0.20
Acceleration (g's)

0.10

0.00

-0.10

-0.20

-0.30

Rock 200 Foot Depth 250 Foot Depth 300 Foot Depth
-0.40
15

16

17

18

19

20

Time (s)

Figure 3.6 : Predicted Absolute Ground Acceleration : Northridge Earthquake

47
 0.80

0.60

0.40
Relative Displacement (inches)

0.20

0.00

-0.20

-0.40

-0.60
50 Foot Depth 100 Foot Depth 150 Foot Depth

-0.80
10

11

12

13
8

Time (s)

Figure 3.7 : Predicted Relative Ground Displacement : Taft Earthquake

1.00

0.50
Relative Displacement (inches)

0.00

-0.50

-1.00

200 Foot Depth 250 Foot Depth 300 Foot Depth

-1.50
10

11

12

13
8

Time (s)

Figure 3.8 : Predicted Relative Ground Displacement : Taft Earthquake

48
 0.60

0.40
Relative Displacement (inches)

0.20

0.00

-0.20

-0.40

-0.60 50 Foot Depth 100 Foot Depth 150 Foot Depth

-0.80
3

6
Time (s)

Figure 3.9 : Predicted Relative Ground Displacement : El Centro Earthquake

0.60

0.40

0.20
Relative Displacement (inches)

0.00

-0.20

-0.40

-0.60

200 Foot Depth 250 Foot Depth 300 Foot Depth

-0.80
3

Time (s)

Figure 3.10 : Predicted Relative Ground Displacement : El Centro Earthquake

49
 1.50

1.00
Relative Displacement (inches)

0.50

0.00

-0.50

-1.00

-1.50

-2.00 50 Foot Depth 100 Foot Depth 150 Foot Depth

15

16

17

18

19

20
Time (s)

Figure 3.11 : Predicted Relative Ground Displacement : Northridge Earthquake

2.50

2.00

1.50
Relative Displacement (inches)

1.00

0.50

0.00

-0.50

-1.00

-1.50

200 Foot Depth 250 Foot Depth 300 Foot Depth

-2.00
15

16

17

18

19

20

Time (s)

Figure 3.12 : Predicted Relative Ground Displacement : Northridge Earthquake

50
 4.00

3.50

3.00

2.50
γ MF

2.00

1.50

1.00

0.50
Rock 50' 100' 150' 200'

0.00
0.5 1.0 1.5 2.0 2.5 3.0
T1 (sec)

Figure 3.13 : Average Magnification Factors : Taft Earthquake

2.50

2.00

1.50
γ MF

1.00

0.50

Rock 50' 100' 150' 200'

0.00
0.5 1.0 1.5 2.0 2.5 3.0
T1 (sec)

Figure 3.14 : Average Magnification Factors : El Centro Earthquake

51
 3.00

2.50

2.00
γ MF

1.50

1.00

0.50

Rock 50' 100' 150' 200'

0.00
0.5 1.0 1.5 2.0 2.5 3.0
T 1 (sec)

Figure 3.15 : Average Magnification Factors : Northridge Earthquake

2.00

Average MF Values Maximum MF Values Uniform Base Motion

1.80

1.60

1.40

1.20
γ MF

1.00

0.80

0.60

0.40

0.20

0.00
0.70 0.80 0.90 1.00 1.10 1.20 1.30 1.40 1.50
T 1 / T pe

Figure 3.16 : Average Magnification Factors : All Earthquakes

52