This action might not be possible to undo. Are you sure you want to continue?
APPLICATIONS FOR TYPICAL MULTISPAN HIGHWAY
BRIDGES
A Dissertation
Presented to
The Academic Faculty
by
Murat Eröz
In Partial Fulfillment
of the Requirements for the Degree
Doctor of Philosophy in the
School of Civil and Environmental Engineering
Georgia Institute of Technology
December 2007
ADVANCED MODELS FOR SLIDING SEISMIC ISOLATION AND
APPLICATIONS FOR TYPICAL MULTISPAN HIGHWAY
BRIDGES
Approved by:
Dr. Reginald DesRoches, Advisor
School of Civil and Environmental
Engineering
Georgia Institute of Technology
Dr. Laurence Jacobs
School of Civil and Environmental
Engineering
Georgia Institute of Technology
Dr. Barry Goodno
School of Civil and Environmental
Engineering
Georgia Institute of Technology
Dr. Jeffrey Streator
School of Mechanical Engineering
Georgia Institute of Technology
Dr. Donald White
School of Civil and Environmental
Engineering
Georgia Institute of Technology
Date Approved: November 5, 2007
In loving memory of Russell and Wylene Sohn
iv
ACKNOWLEDGEMENTS
I would like to express my deepest gratitude to my advisor Dr. Reginald
DesRoches for the support, guidance and friendship he has extended. I would like to
thank Dr. Barry Goodno, Dr. Donald White, Dr. Laurence Jacobs, and Dr. Jeffrey
Streator for their time and effort in serving on my Ph.D. committee. I am grateful for the
unconditional love and support of my parents. Their inspiration and guidance has made
this dream a reality.
v
TABLE OF CONTENTS
Page
ACKNOWLEDGEMENTS iv
LIST OF TABLES ix
LIST OF FIGURES x
SUMMARY xvii
CHAPTER
1 INTRODUCTION 1
1.1 Background 1
1.2 Statement of the Problem 4
1.3 Objectives and Scope 5
1.4 Thesis Outline 6
2 CRITICAL REVIEW OF THE CURRENT STATEOFTHEART 8
2.1 Introduction 8
2.2 Basics: Seismic Isolation for Bridges 8
2.3 The Friction Pendulum System (FPS) 10
2.3.1 Normal Force, N 15
2.3.2 Coefficient of Friction, µ 21
2.3.3 Bidirectional Coupling 24
2.3.4 Large Deformation Effects 25
2.3.5 Other Parameters 26
2.4 Comparative Studies 26
2.5 Parametric Studies 29
2.6 Critical Appraisal 31
vi
3 SEISMIC ISOLATION FOR BRIDGES 33
3.1 Introduction 33
3.2. General Features of a Seismically Isolated Bridge (SIB) 33
3.3 Bridge Seismic Isolation in Design Codes 38
3.3.1 Procedure 1: Uniform Load Method 39
3.3.2 Procedure 2: Single Mode Spectral Method 40
3.3.1 Procedure 3: Multimode Spectral Method 41
3.3.1 Procedure 4: TimeHistory Method 41
3.4 Descriptions of common Isolators 42
3.4.1 The Eradiquake System (EDS) 44
3.4.2 The High Damping Rubber System (HDRS) 45
3.4.3 The Lead Rubber Bearing (LRB) 46
3.5 Conclusion 48
4 FRICTION PENDULUM SYSTEM (FPS) MODELING 49
4.1 Introduction 49
4.2. Simplified Isolator Response Modeling 49
4.3 Normal Force 52
4.4 Coefficient of Friction 53
4.5 Bidirectional Coupling 54
4.6 Large Deformation Moments 57
4.7 Mathematical Model 58
4.8 Evaluation Platform 62
4.9 Verification 64
4.9.1 Structural Properties and Loads 64
4.9.2 Modeling and Analysis 68
vii
4.9.3 Results 69
4.10 Conclusion 75
5 BRIDGE RESPONSE AS A FUNCTION OF ISOLATOR MODELING
ASSUMPTIONS 76
5.1 Introduction 76
5.2. Selection of the Class of Highway Bridges for Seismic Isolation and
Analyses 76
5.3 Seismically Isolated Bridge (SIB) Modeling 77
5.4 FPS Models 80
5.5 Dynamic Analyses 82
5.6 Results 85
5.7 Conclusions 93
6 COMPARATIVE ASSESSMENT OF SLIDING VERSUS ELASTOMERIC
SEISMIC ISOLATION FOR TYPICAL MULTISPAN BRIDGES 95
6.1 Introduction 95
6.2. The Lead Rubber Bearing (LRB) 95
6.3 ForceDeformation Characteristics of the LRB 99
6.4 Modeling of the Isolator Response 101
6.5 Bridge Model 102
6.6 Bridge Seismic Isolation 103
6.7 Dynamic Analysis 107
6.8 Results 110
6.9 Conclusion 124
7 ASSESSMENT OF THE INFLUENCE OF THE DESIGN PARAMETERS ON
THE RESPONSE OF BRIDGES ISOLATED WITH THE FRICTION
PENDULUM SYSTEM (FPS) 136
7.1 Introduction 136
viii
7.2 Influence of Bridge Design Parameters 126
7.2.1 Analyses 127
7.2.2 Results 129
7.3 Influence of a Modified Seismic Isolation Strategy 140
7.3.1 Proposed Design 142
7.3.2 Analyses 147
7.3.3 Results 149
7.4 Conclusion 152
8 SUMMARY, CONCLUSIONS AND FUTURE RESEARCH 155
8.1 Summary and Conclusions 155
8.2 Future Research 159
APPENDIX A: METHOD OVERRIDING IN C++ 161
APPENDIX B: REGRESSION LINES FOR THE EFFECTS OF BRIDGE DESIGN
PARAMETERS ON SYSTEM RESPONSE 163
REFERENCES 174
VITA 182
ix
LIST OF TABLES
Page
Table 3.1: Calculation of base shear 38
Table 5.1: Summary of model properties 81
Table 5.2: Ground motion suite 83
Table 6.1: Ground motion suite 109
Table 7.1: Parameter variation ranges 128
Table 7.2: Vibration periods of the bridges as a function of design parameters 132
Table 7.3: Modified isolator design properties to achieve a fundamental period of T
1
=2.38
s 148
x
LIST OF FIGURES
Page
Figure 1.1: Typical seismic isolation example in the American River Bridge at Lake
Natoma in Folsom, California (courtesy of Earthquake Protection Systems,
Inc.) 2
Figure 1.2: Design spectrum and the shift of spectral ordinates for an isolated structure 3
Figure 1.3: The Friction Pendulum System (FPS) (a) exterior view (b) internal
components (c) interior elevation 4
Figure 2.1: Applications of FPS seismic isolation at (a) Bolu Viaduct; (a) Rio Hondo
Busway Bridge; (b) KodiakNear Island Bridge; (c) Ataturk International
Airport Terminal; and (d) liquefied natural gas (LNG) storage tanks in Greece
(courtesy of Earthquake Protection Systems) 11
Figure 2.2: Components of the FPS 14
Figure 2.3: Sources of normal force, N, fluctuations in bridge isolators (a) vertical inertial
forces (b) lateral inertial forces (c) vibration forces due to traffic 16
Figure 2.4: Response spectra for Gazli (1976) earthquake record 17
Figure 2.5: Compression failure and bulging of piers along bent 3. Bull Creek Canyon
Channel Bridge. Photograph courtesy of Earthquake Engineering Research
Institute (Papazoglou and Elnashai 1996) 18
Figure 2.6: Resilient sliding isolation (RSI) (Iemura et al. 2005) 21
Figure 2.7: Test setup for Teflonsteel sliding surfaces (Mokha et al. 1990) 22
Figure 2.8: Dependency of the coefficient of friction on sliding velocity (Mokha et al.
1990) 23
Figure 3.1: Idealized seismicisolated bridge substructure 34
Figure 3.2: Idealized seismicisolated bridge substructure 35
Figure 3.3: Structural shapes for (a) Mode 1 (b) Mode 2 36
Figure 3.4: Modal expansions of effective earthquake forces and modal static responses
for the base 37
Figure 3.5: Pseudoacceleration design spectrum for AASHTO Type II soil profile with
acceleration coefficient A=0.2 38
xi
Figure 3.6: Single substructure and isolator idealization 40
Figure 3.7: Characteristics of bilinear isolation bearings per AASHTO Guide
Specifications 42
Figure 3.8: Typical elastomeric isolator (Taylor and Igusa 2004) 43
Figure 3.9: Clantarient`s base isolation system using a layer of talc as the isolating
medium (Naeim and Kelly 1996) 44
Figure 3.10: The Eradiquake seismic isolation bearing 45
Figure 3.11: (a) High damping rubber bearing used in the earthquake simulator tests with
dimensions in mm (b) corresponding forcedeformation hysteresis (Kikuchi
and Aiken 1997) 46
Figure 3.12: Typical lead rubber bearing (LRB) (Taylor and Igusa 2004) 47
Figure 3.13: (a) Lead rubber bearing (LRB) used in the earthquake simulator tests with
dimensions in mm (b) corresponding forcedeformation hysteresis (Kikuchi
and Aiken 1997) 48
Figure 4.1: The signum function 50
Figure 4.2: The intrinsic response components (a) friction,
µ
f , and (b) pendulum, f
R
50
Figure 4.3: Forcedeformation characteristics of the unidirectional rigidplastic response
of the FPS 51
Figure 4.4: Forcedeformation characteristics of bilinear isolators 52
Figure 4.5: Gap element forcedeformation model 53
Figure 4.6: Variation of the coefficient of friction with (a) velocity of sliding; and (b)
isolator contact pressure (Roussis and Constantinou 2006) 54
Figure 4.7: Frictional interaction surface (a) uncoupled (b) coupled response 56
Figure 4.8: Deformed shape of the seismic isolator between the superstructure and the
substructure with concave dish at the (a) bottom (b) top 58
Figure 4.9: Schematic view of the model 59
Figure 4.10: Deflections and forces acting on the slider 61
Figure 4.11: FPSelement and FPSmaterial interaction 64
Figure 4.12: Test setup studied by Mosqueda et al. (2004) 65
xii
Figure 4.13: Scaling of f
max
with respect to the relationship presented in Constantinou et
al. (1993) 66
Figure 4.14: Unidirectional load histories with amplitude, ± 12.7 cm (L1a), ± 17.8 cm
(L1b), and ± 45.0 cm (L1c) 67
Figure 4.15: Combined bidirectional loading path for L2 68
Figure 4.16: Finite element model of the test setup studied by Mosqueda et al. (2004) 69
Figure 4.17: Comparison of the normalized forcedeformation histories between the
model and the experimental results for (a) individual Isolator 3, and (b) total
isolator forces 70
Figure 4.18: Friction coefficient, µ , time history under loading L1b 71
Figure 4.19: Comparison of the forcedeformation histories of the FPS with theoretically
exact value of the friction coefficient, µ , µ =f
max
,and µ =f
min
for loading L1b
72
Figure 4.20: Comparison of the response of the small deformations model (SDM) and the
large deformations model (LDM) 73
Figure 4.21: Comparison of the bidirectional resisting forces for the experimental data,
and coupled model 74
Figure 4.22: Comparison of the bidirectional resisting forces for the coupled model and
the uncoupled models 74
Figure 5.1: Photo of example MSC Steel Girder Highway Bridge (Nielson 2005) 77
Figure 5.2: Multispan continuous (MSC) steel girder bridge general elevation and
modeling details 79
Figure 5.3: Pier configuration and bent and column discretization 79
Figure 5.4: Constitutive relationships for the modeling of (a) steel material; and (b)
concrete material 80
Figure 5.5: Mode shapes of the deck 82
Figure 5.6: Response spectrums for the suite of ground motions 83
Figure 5.7: Orientation of the 3D bridge model 84
Figure 5.8: Time history of the N/N
o
for the FPS during the Nahanni earthquake NLTH
analysis 86
xiii
Figure 5.9: Forcedeformation history of the FPS in the longitudinal directions on top of
the pier for the N. Palm Springs earthquake record with (a) Model 1; (b) Model
2; (c) Model 3; and Model 4 87
Figure 5.10: The influence of modeling assumptions on (a) MNF; (b) MND; and (c) d
max
89
Figure 5.11: The influence of constant value of µ assumptions on (a) MNF; (b) MND;
and (c) d
max
91
Figure 5.12: Comparison of the (a) total MNF transferred to the pier, ΣMNF
pier
, and the
total MNF transferred to the abutment, ΣMNF
abutment
; (b) MND on top of the
pier, MND
pier
, and abutments, MND
abutments
92
Figure 6.1: Examples of LRB applications (a) Rio Vista Bridge, Califonia; (b) Patria
Acueducto, Mexico (courtesy of Dynamic Isolation Systems) 97
Figure 6.2: The Lead Rubber Bearing (LRB) interior elevation 98
Figure 6.3: Effects of geometrical variations of the LRB on the forcedeformation
response (Priestley et al. 1996) 98
Figure 6.4: Isolator model 102
Figure 6.5: Variation of the buckling load, P
cr
, as a function of isolator inplane
deformation, δ, for the LRB 104
Figure 6.6: Variation of isolator yield force, F
y
, as a function of applied compressive
axial load, N 105
Figure 6.7: Variation of isolator postyield stiffness, k
p
, as a function of applied
compressive axial load, N 116
Figure 6.8: Bilinear idealizations of the FPS forcedeformation characteristics after
gravity loading in the bridge 106
Figure 6.9: Bilinear idealizations of the LRB forcedeformation characteristics after
gravity loading in the bridge 107
Figure 6.10: Mode shapes of the bridge deck from plan 108
Figure 6.11: Response spectrums for the suite of ground motions 109
Figure 6.12: Forcedeformation history of the FPS in the longitudinal direction on top of
the pier for the Northridge earthquake record where the vertical component,
v
) (t u& & , effect is (a) included (b) not included 111
xiv
Figure 6.13: Time history of the N/N
o
for the FPS during the Northridge earthquake
record 111
Figure 6.14: Forcedeformation history of the LRB in the longitudinal direction on top of
the pier for the Whittier Narrows earthquake record where the vertical
component,
v
) (t u& & , effect is (a) included (b) not included 112
Figure 6.15: Time history of the LRB buckling load, P
cr
, and normal force, N 113
Figure 6.16: Forcedeformation history of the LRB on top of the pier for the Gazli
earthquake in the (a), (c) longitudinal, and (b), (d) transverse directions 114
Figure 6.17: Forcedeformation history of the FPS in the longitudinal direction on top of
the pier for the Whittier Narrows earthquake record where the vertical
component,
v
) (t u& & , effect is (a) included (b) not included 125
Figure 6.18: Forcedeformation history of the FPS and LRB for the Helena earthquake
(a),(b) on top of the pier and (c),(d) on top of the abutment 127
Figure 6.19: Comparison of the (a) total MNF transferred to the pier, ΣMNF
pier
, and the
total MNF transferred to the abutment, ΣMNF
abutment
; (b) MID on top of the
pier, MID
pier
, and abutments, MID
abutments
118
Figure 6.20: Total energy dissipated for by the isolators (a) on top of the pier (b) on top
of the abutment 119
Figure 6.21: Maximum isolator forces (MIF) for the suite of ground motions on top of the
pier with vertical component,
v
) (t u& & , effect is (a) included (b) not included 121
Figure 6.22: Maximum isolator deformations (MID) for the suite of ground motions on
top of the pier with vertical component,
v
) (t u& & , effect is (a) included (b) not
included 121
Figure 6.23: Maximum column drifts, d
max
, for the suite of ground motions on top of the
pier with vertical component,
v
) (t u& & , effect is (a) included (b) not included 132
Figure 6.24: Deformation histories for the Morgan Hill earthquake of (a), (c), (e) FPS and
(b), (d), (f) LRB located on top of the pier 123
Figure 7.1: Bridge design parameters 128
Figure 7.2: Base conditions for the substructure modeled as (v) pinned (b) partiallyfixed
(c) fixed 128
Figure 7.3: Slopes, a
’response quantity
,
’
for the regression lines of the median design
parameters 131
xv
Figure 7.4: Forcedeformation history of the FPS in the longitudinal direction on top of
the pier for the Whittier Narrows earthquake record with (a) L
c
/L
co
=0.8 (b)
L
c
/L
co
=1.6; (c) timehistory of the longitudinal column tip deformations 133
Figure 7.5: Forcedeformation history of the FPS in the longitudinal direction on top of
the pier for the Helena earthquake record with (a) L
d
*
/L
do
=0.75 (b) L
d
*
/L
do
=1.75 134
Figure 7.6: Forcedeformation history of the FPS in the longitudinal direction on top of
the pier for the Northridge earthquake record with (a) L
d
/L
do
=0.75 (b) L
d
/L
do
=1.75 135
Figure 7.7: Median values of the response quantities of the bridge as a function of base
modeling assumptions 138
Figure 7.8: Structural bridge responses as a function of base modeling assumptions 139
Figure 7.9: Forcedeformation history of the FPS in the longitudinal and transverse
directions on top of the pier for the Loma Prieta earthquake record with (a), (b)
fixed base conditions; (c), (d) pinned base conditions 140
Figure 7.10: Prestressing the FPS (Kasalanati and Constantinou 2005) 142
Figure 7.11: Proposed modified design for the FPS above the abutments in (a)
undeformed (b) deformed state 144
Figure 7.12: Sample spring element to be used in the proposed modified design of the
FPS at the abutments 145
Figure 7.13: Displacementcontrolled load history 146
Figure 7.14: Forcedeformation response of the isolator with modified design 146
Figure 7.15: Modified isolator design modeling at the abutments 147
Figure 7.16: Variation of bridge response quantities as a function of FPS design
parameters 150
Figure 7.17: Forcedeformation history of the FPS in the longitudinal direction for the
Helena earthquake (a), (b) conventional design with R = 99 cm; (c), (d) new
design with R = 621 cm; (a), (c) on top of the pier; (b), (d)on top of the
abutment 152
Figure B.1: The influence of the pier concrete compressive strength, f
c
, on (a) MNF (b)
MND (c) d
max
(d) ΣMNF
pier
/ ΣMNF
abutment
(e) MND
pier
/ MND
abutment
164
Figure B.2: The influence of the column length, L
c
, on (a) MNF (b) MND (c) d
max
(d)
ΣMNF
pier
/ ΣMNF
abutment
(e) MND
pier
/ MND
abutment
166
xvi
Figure B.3: The influence of longitudinal deck length, L
d
, with constant mass on (a) MNF
(b) MND (c) d
max
(d) ΣMNF
pier
/ ΣMNF
abutment
(e) MND
pier
/ MND
abutment
168
Figure B.4: The influence of longitudinal deck length, L
d
, with adjusted mass on (a) MNF
(b) MND (c) d
max
(d) ΣMNF
pier
/ ΣMNF
abutment
(e) MND
pier
/ MND
abutment
170
Figure B.5: The influence of skew angle, α, on (a) MNF (b) MND (c) d
max
(d) ΣMNF
pier
/
ΣMNF
abutment
(e) MND
pier
/ MND
abutment
172
xvii
SUMMARY
The large number of bridge collapses that have occurred in recent earthquakes has
exposed the vulnerabilities in existing bridges. One of the emerging tools for protecting
bridges from the damaging effects of earthquakes is the use of isolation systems. Seismic
isolation is achieved via inserting flexible isolator elements into the bridge that shift the
vibration period and increase energy dissipation. To date, the structural performance of
bridges incorporating sliding seismic isolation is not wellunderstood, in part due to the
lack of adequate models that can account for the complex behavior of the isolators. This
study investigates and makes recommendations on the structural performance of bridges
utilizing sliding type seismic isolators, based on the development of stateoftheart
analytical models. Unlike previous models, these models can account simultaneously for
the variation in the normal force and friction coefficient, large deformation effects, and
the coupling of the vertical and horizontal response during motion. The intention is to
provide support for seismic risk mitigation and insight for the analysis and design of
seismically isolated bridges by quantifying response characteristics. The level of
accuracy required for isolator analytical models used in typical highway bridges are
assessed. The comparative viability of the two main isolator types (i.e. sliding and
elastomeric) for bridges is investigated. The influence of bridge and sliding isolator
design parameters on the system’s seismic response is illustrated.
1
CHAPTER I
INTRODUCTION
1.1 Background
Bridges are a crucial part of the overall transportation system and their
performance during earthquakes is important for continued functioning of a community.
The large number of bridge collapses during the 1971 San Fernando earthquake exposed
the deficiencies of the 1965 AASHTO and previous bridge design codes. There are
approximately 575,000 bridges in the United States with approximately 60% having been
constructed prior to 1970 with little or no consideration given to seismic resistance
(Cooper et al.1994). A series of revisions to the design guidelines were accompanied by
the launching of substantial retrofitting programs following the 1971 San Fernando
earthquake. Until the mid1980s, bridge retrofitting techniques in the U.S. only
incorporated individualized strengthening schemes such as steel jacketing of columns,
casting of infill walls between columns, strengthening of footings and bearing elements,
widening of the pier caps and abutments, and the use of restraining cables (Yashinsky
1998). Seismic bridge design focused on increasing the lateral strength to resist inertial
forces that occurred from ground shaking (Yashinsky and Karshenas 2003). However,
this increase was found to be selfdefeating, since strengthened members attracted larger
forces that severely damaged other elements along the bridge’s lateral load path.
The root cause of the damaging effects of earthquakes is the unfortunate
correlation between the fundamental periods of vibration of major structures and the
frequency content of the seismic input (Priestley et al. 1996). Seismic isolation
decouples the structure from the horizontal components of the ground motion with
elements that have a low horizontal stiffness (Naeim and Kelly 1996) (Figure 1.1).
2
Figure 1.1 Typical seismic isolation example in the American River Bridge at Lake
Natoma in Folsom, California (courtesy of Earthquake Protection Systems, Inc.).
Isolation shifts the response of the structure to a higher fundamental period and
increases the damping, thus reducing the corresponding pseudoacceleration in the design
spectrum and attracting smaller earthquakeinduced forces, as illustrated in Figure 1.2.
The philosophy of seismic isolation for improving earthquake resistance of a structure
departs from conventional retrofit measures because the latter attempts to strengthen
individual elements of bridges while the former improves structural performance by
reducing the overall earthquake forces that the structure acquires.
3
Figure 1.2 Design spectrum and the shift of spectral ordinates for an isolated structure.
Various isolators have been manufactured with the similar objective of providing
a period shift and additional energy dissipation to structures. Isolators can be classified
as sliding and elastomeric (Taylor and Igusa 2004; AASHTO 1999). Among others, two
isolator types that are representative of sliding and elastomeric systems are the Friction
Pendulum System (FPS) and the LeadRubber Bearings (LRB), respectively. This study
focuses on the seismic response of bridges with emphasis on the FPS. The mechanism of
the FPS is primarily based on its concave geometry and the surface properties. The
supported structure is administered into a pendulum motion as the isolators
simultaneously glide on its concave surfaces and dissipate hysteretic energy via these
frictional surfaces (Dicleli and Mansour 2003).
4
(a) (b) (c)
Figure 1.3 The Friction Pendulum System (FPS) (a) exterior view (b) internal
components (c) interior elevation.
1.2 Statement of the Problem
Bridge seismic isolation in the U.S. is a relatively new phenomenon that was
addressed by the AASHTO with the Seismic Isolation Guide Specification for the first
time in 1991. By this time elastomeric bearings were primarily used in bridge seismic
isolation (Stanton 1998). As new isolator types became available by 1995, the first
Seismic Isolation Guide Specification was essentially rewritten in 1997 to address the
advances in the industry. However, the Specification still does not provide guidance
about selecting the optimal isolator type for different bridge applications. Despite recent
advances in base isolation research, the widespread application of this technology is still
impeded by overconservative attitudes (Mayes 2002; Naeim and Kelly 1996; Clark et al.
1999). The responses of state bridge engineers on a survey asking why base isolation is
not more widely used revealed that engineers are not comfortable with seismic isolation
because they view it as a black box and that there is a lack of certainty on choosing the
optimum type of seismic isolation (Mayes 2002). Furthermore, sliding seismic isolators
make up less than 25% of the total number of isolated bridges in North America (Buckle
et al. 2006). A better understanding of the impact of isolators on the seismic behavior of
bridge response is necessary.
5
The individual response of isolators is well understood but there are limited
studies investigating the response of seismically isolated bridges (SIBs) via detailed
threedimensional (3D) models. In particular, questions such as the relative benefits of
different isolator systems for different bridge types have yet to be adequately addressed.
Also, it is not clear what parameters in a typical bridge govern the effectiveness of
various isolation systems. Finally, since isolators have a highly nonlinear response
involving the simultaneous action of multiple components, detailed models are needed to
capture the intricate behavior of these highly nonlinear elements. The influence of the
level of accuracy in the modeling assumptions of isolators for bridges has received
narrow attention. This thesis is aimed at addressing some of these issues.
1.3 Objectives and Scope
The objectives of this study are to assess the performance of bridges seismically
isolated with the FPS, with a particular emphasis on the modeling parameters of the
isolators which govern the seismic response of typical bridges. This is accomplished by
developing rigorous analytical models of isolators with particular emphasis on the FPS
and using these models to investigate the response of SIBs. The intention is to provide
support for seismic risk mitigation and insight for the analysis and design of SIBs by
quantifying response characteristics. The research tasks to accomplish these objectives
are the following:
• Identify the characteristic aspects of the FPS that contribute to the force
deformation response. Develop the nonlinear kinematics formulation of the
isolator model.
• Implement the model into a nonlinear dynamic evaluation platform and validate
response using experimental data.
6
• Develop detailed 3D bridge models isolated with the FPS and identify the
influence of the modeling assumptions of the isolator on the response of the
bridge.
• Modify the FPS model to represent the LeadRubber Bearings (LRB) force
deformation response.
• Compare and quantify the response of bridges as a function of isolator type with
emphasis on FPS and LRB.
• Investigate parametrically the influence of bridge geometric and material
properties, and isolator design parameters on the system’s response. If applicable,
propose modifications for design of the isolator to improve bridge seismic
performance.
1.4 Thesis Outline
The content of the dissertation is organized into the following chapters:
Chapter 2 provides a literature review on bridge seismic isolation. Particular
emphasis is given to the modeling aspects of the FPS, structural sensitivity to different
modeling parameters, comparative studies among different isolators, and parametric
analyses. A critical assessment of the currentstateoftheart is presented.
Chapter 3 presents the application of seismic isolation into bridges. The
dynamics of bridge seismic isolation is explained on a simplified model. Analysis
methods in current Guide Specifications are outlined. An overview of common isolators
is presented.
Chapter 4 explains the development of the FPS isolator model. Modeling aspects
of the FPS are highlighted and the nonlinear kinematics of the response is generated. The
model is incorporated into an open source finite element platform. The model is
validated using experimental data.
7
Chapter 5 examines the response of a three dimensional (3D) SIB as a function
of FPS modeling assumptions. Friction Pendulum System (FPS) models developed with
different assumptions are incorporated into bridges and subjected to timehistory
analyses.
Chapter 6 compares the performance of a typical bridge with elastomeric isolation
versus sliding isolation. An LRB model modified from the previous FPS model is
developed. The differences of the FPS and LRB response are highlighted. The responses
of two bridge models isolated with each isolator are examined under timehistory
analyses.
Chapter 7 provides a parametric investigation on the seismic response of an FPS
isolated bridge. The response of the bridge is monitored as a function of varying design
parameters.
Chapter 8 presents a summary of the research, major conclusions drawn from this
study, and recommendation for future research.
8
CHAPTER 2
CRITICAL REVIEW OF THE CURRENT STATEOFTHEART
2.1 Introduction
The introduction of seismic isolation as a practical tool has provided a rich source
of experimental and theoretical work both in the dynamics of the isolated structural
systems and in the mechanics of the isolators themselves. This chapter presents a
summary of the previous studies that address the modeling and analysis aspects of bridge
seismic isolation with particular emphasis on the Friction Pendulum System (FPS)
isolator. The basics and historical development of seismic isolation is outlined.
Experimental and analytical research conducted on the response characteristics of the
FPS is elaborated. Analytical research aimed at comparing the two main classes of
isolators, the sliding and the elastomeric, are reviewed. Parametric studies conducted on
isolator and bridge design properties are summarized. A critical appraisal of the current
literature is presented.
2.2 Seismic Isolation for Bridges: Basics
Recent earthquakes have illustrated the vulnerability of bridges to damage and
collapse (Cooper and Friedland 1994; Yashinsky 1998). One of the emerging tools for
protecting bridges from the damaging effects of earthquakes is the use of seismic
isolation systems. An insightful definition of ‘seismic isolation’ given by Skinner et al.
(1993) is as follows:
‘Seismic isolation consists essentially of the installation of mechanisms which
decouple the structure, and/or its contents, from potentially damaging earthquake
9
induced, ground or support, motions. This decoupling is achieved by increasing the
flexibility of the system, together with providing appropriate damping.’
The two fundamental structural improvements provided by seismic isolation is the
reduction of lateral forces and the concentration of lateral displacements at the isolation
interface (Taylor and Igusa 2004). Seismic isolators are typically installed between
piers, abutments, and deck (Priestley et al. 1996). Although patents for seismic isolation
schemes were obtained as early as 130 years ago, only in the last four decades has
industrial capabilities enabled the manufacturing of practical isolation devices, and only
in the last decade has seismically isolated structural design been widely adopted (Taylor
and Igusa 2004). Currently, isolation systems are most commonly classified as
elastomeric and sliding. The fundamental concept of base isolation was first studied on
an example building on balls by Professor John Milne who was a faculty member in the
Mining Engineering Department of Tokyo University between 1876 and 1895 (Naeim
and Kelly 1996). The first building that employed a rubber isolation system was a school
at Skopje, Yugoslavia in 1969 (Naeim and Kelly 1996). The first seismically isolated
building in the U.S.A. was the Foothill Communities Law and Justice Center in 1984
1985 in California, which was located only 19.3 km west of the San Andreas Fault
(Taylor and Igusa 2004). The California Department of Transportation (Caltrans) was
the first U.S. transportation agency to use seismic isolation on a bridge at the Sierra Point
Overlook in 1985 (Taylor and Igusa 2004).
Currently seismic isolation is wellintegrated into the code provisions in the U.S.
for both buildings and bridges (International 2003; NEHRP 2003; AASHTO 1999;
FEMA 356). However, provisions developed for seismic isolation of bridges are unique
due to fundamental differences in the structural response of bridges compared to
buildings. Taylor and Igusa (2004) identified the distinct properties of the structural
response of bridges from that of buildings as the following:
10
1. Spatial variations in the ground motion may become important because bridges
are long in one direction
2. Bridges are supported by flexible piers while buildings often have relatively rigid
foundations
3. The flexibility of the substructure may change significantly in bridges due to the
need of accommodating variations in terrain
4. Bridges are typically more flexible in the vertical direction compared to buildings
because of long spans. This makes vertical ground motions important for bridges.
5. The philosophy in building isolation is to limit the forces in the superstructure
while the primary concern in bridges is typically the substructure.
In addition to these factors, bridge isolators are subjected to more severe routine live load
conditions than those observed in buildings and may be more exposed to environmental
conditions such as, freezing, rain, sun light, salt water, and debris compared building
isolators located at the foundation levels.
2.3 The Friction Pendulum System (FPS)
The Friction Pendulum System (FPS) is a sliding type seismic isolator that was
developed in 1986 by Earthquake Protection Systems, Inc. (Zayas et al. 1987). The FPS
was first used to retrofit an apartment building in California in 1989 (Naeim and Kelly
1996). Since than, the FPS have been used to isolate buildings (Washington State
Emergency Operations Center at Camp Murray, the U.S. Court of Appeals Building in
San Francisco), bridges (BeniciaMartinez Bridge in the San Francisco Bay Area,
American River Bridge at Lake Natoma in Folsom), and storage tanks (LNG storage
tanks on Revithoussa Island near Athens) (Jangid 2005) (Figure 2.1). The FPS has been
incorporated into the design codes (AASHTO 1999; International 2000).
11
(a)
Figure 2.1 Applications of FPS seismic isolation at (a) Bolu Viaduct; (a) Rio Hondo
Busway Bridge; (b) KodiakNear Island Bridge; (c) Ataturk International Airport
Terminal; and (d) liquefied natural gas (LNG) storage tanks in Greece (courtesy of
Earthquake Protection Systems).
12
(b)
(c)
Figure 2.1 continued
13
(d)
(e)
Figure 2.1 continued
14
The FPS consists of a spherical stainless steel surface, an articulated slider and a
housing plate (Zayas et al. 1987) (Figure 2.2). The sliding surface of the FPS consists of
stainless steel and a Teflonbased custom material. The radius of the FPS isolator
controls the concavity related stiffness and the isolation period of the structure (Naeim
and Kelly 1996). As the slider displaces over the concave surface, a continuous
recentering force is provided by the gravity load of the supported mass. Simultaneously,
the friction force opposes the motion of the slider and dissipates hysteretic energy.
Figure 2.2 Components of the FPS.
Findings of previous research provide ample evidence that the dynamic response
of seismically isolated structures is governed by the characteristics of the mechanisms of
the isolators (Dicleli 2002). This is an indication that the modeling assumptions adopted
for the response of the FPS will affect the estimated response quantities of SIBs. The
forcedeformation response of the FPS is typically modeled using a unidirectional
idealization (Zayas et al. 1987):
15
}
R
µ
f
f
δ
R
N
δ N f + =
48 47 6
&
) sgn( µ (2.1)
where N is the normal force acting on the sliding surface, µ is the friction coefficient
between the sliding surfaces, R is the radius of the concave surface, δ is the sliding
deformation, δ
&
is the sliding velocity, and ) sgn(δ
&
is the signum function. The signum
function is equal to +1 or 1 depending on whether δ
&
is negative or positive,
respectively. The forcedeformation response of the FPS is further elaborated in Chapter
4. However, it is important to underline the fundamental assumptions inherent in this
equation: (1) N is constant; (2) µ is constant; (3) the horizontal response is uncoupled in
the orthogonal directions; and (4) isolator deformations are small and planar. The
following sections describe theoretical and experimental research performed to quantify
the influence of these simplifications.
2.3.1. Normal Force, N
The restoring mechanism of the FPS is dependent on the normal force, N (see
Equation 1). Takashi et al. (2000) identified three sources for the fluctuation of N in the
isolators of bridges: (1) inertial forces due to vertical ground motions; (2) rocking
behavior of the girder due to horizontal ground motion; and (3) deflection vibration of the
girder due to vertical ground motion service loads and curvature (Figure 2.3).
16
(a)
(b)
(c)
Figure 2.3 Sources of normal force, N, fluctuations in bridge isolators (a) vertical inertial
forces (b) lateral inertial forces (c) vibration forces due to traffic.
There is a general tendency among bridge engineers to neglect the effects of
vertical ground motions on the structural response (Button et al. 2002). Although the
vertical components of ground motions typically have lower energy content compared to
the horizontal components, the energy content is concentrated in a narrow, high
frequency band which may inflict considerable structural damage (Collier and Elnashai
2001). For nearfield earthquakes (<1015 km), the vertical spectra may significantly
exceed the horizontal spectra for short periods (Silva 1997). This is illustrated in Figure
2.4 for the Gazli record taken 3 km from the source. Button et al. (2002) reported that the
vertical component of the earthquake may be consequential for bridges located within 60
km of the source.
17
Period, T (s)
0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50 2.75 3.00
S
p
e
c
t
r
a
l
a
c
c
e
l
e
r
a
t
i
o
n
,
S
a
(
g
)
f
o
r
ξ
=
5
°
0
1
2
3
4
5
Component, UP
Component, 00
Component, 90
Figure 2.4 Response spectra for Gazli (1976) earthquake record.
Serious substructure damage was observed in bridges during the Northridge and
Kobe earthquakes (Papazoglou and Elnashai 1996) (Figure 2.5). It was shown that the
design values for dynamic response increases for bridge members with the inclusion of
the vertical components of the ground motions (Gloyd 1997; Yu 1996). However,
structural damage was noted in bridge members only at relatively high magnitudes of
accelerations in the vertical direction. For example, Saadeghvaziri (1991) concluded that
considerable damage would occur in columns if the vertical component of the ground
motions exceeded 0.7g.
18
Figure 2.5. Compression failure and bulging of piers along bent 3. Bull Creek Canyon
Channel Bridge. Photograph courtesy of Earthquake Engineering Research Institute
(Papazoglou and Elnashai 1996).
The aforementioned studies focus on nonisolated bridges. Since structures are
inherently stronger and stiffer in the vertical direction compared to the horizontal
directions, they are normally not isolated from vertical earthquake motions (Taylor and
Igusa 2004). However, theoretical and experimental evidence suggests that the
horizontal response of the FPS is coupled with the vertical response. Takashi et al.
(2000) performed shaking table tests on a girder model supported by a set of four
frictional isolators and two rubber buffers to assess the dependency of the structural
response to vertical and horizontal excitations. The authors reported that although the
rocking of the structure altered the normal force and behavior of individual isolators, it
did not have considerable influence on the overall response. Takashi et al. mentioned
that the sum of the areas of the hysteretic loops for cases with and without vertical ground
motion was acutely similar, and concluded that the effect of vertical ground motion on
the overall response was trivial. The authors have imposed vertical ground motion five
times higher than that of initial test input to assess the effects of extremely large vertical
19
ground motion on the response. Takashi et al. reported that a loss of normal force on the
isolators emanated, but the overall horizontal response did not languish even under
extremely large vertical ground motions.
Nakajima et al. (2004) performed pseudodynamic tests to verify the effect of
vertical motions on the response of sliding isolators. The experimental setup consisted of
a sliding bearing that generated friction damping and of a rubber bearing that provided
restoring force. The authors concluded that although the friction coefficient of the
bearings alternated as a function of the bearing pressure, the effect of vertical motion was
inconsequential in the overall response. It was observed that the friction coefficient
increased proportionally with the sliding velocity and virtually flattened after exceeding a
certain velocity. The authors stated that the friction coefficient decreased linearly with
increasing bearing pressure.
Mosqueda et al. (2004) performed unidirectional and tridirectional tests on a
rigidblock frame supported by four FPS. The authors concluded from the results of the
tridirectional tests that the vertical component of the ground motion had negligible effect
on the forcedeformation response of the FPS. However, the authors noted that rotation
of the superstructure in bridges caused by lateral ground motions could significantly
influence the behavior of FPS.
Almazan et al. (1998) generated the exact analytical equations of motions for the
FPS, which considered the large deformation effects. The authors analyzed fourstory
building models under horizontal and vertical ground motion. It was concluded that the
global response of the structure might be estimated within the vicinity of 20% error if
vertical motions are disregarded. The authors further mentioned that uplift occurred at
several instants of the response and column base shears were 3 times larger when vertical
motion is considered.
Dicleli (2002) performed modal and nonlinear timehistory analyses on a six span
slabongirder deck isolated with the FPS. Analysis results showed that the first modes
20
of vibrations were those involving the isolation system, with Modes 1, 2 and 4 being the
transverse modes of vibration and mode 3 being the longitudinal. In the longitudinal
direction, all isolator were found to have a uniform displacement due to the large axial
rigidity of the deck in this direction. However, in the transverse direction, bearing
displacements varied along the bridge caused by the unusual flexibility of the bridge in
his direction. Furthermore, the dead load reactions at the abutments were only 20% of
those at the piers due to smaller tributary weight of shorter end spans and uplift reactions
created by much longer adjacent spans. Since the lateral resistance of the FPS is directly
proportional to the dead load reactions acting on the bearings, a very small equivalent
stiffness was obtained at the abutments. This was found to produce even larger bearing
displacements as the seismically induced forces acquired the shape of the deflected
structure. Although the abutments were structurally stronger than the piers due to their
massive size, only 7% of the total seismic force was transferred to the abutments. It was
concluded that hybrid isolation of the bridge using FPS and laminated elastomeric
bearings produced results that are more favorable than the bridge isolated with FPS alone.
Iemura et al. (2005) performed shake table tests on scaled models of two highway
bridges seismically isolated with a combined rubber and sliding bearings, a system
referred as ‘resilient sliding isolation (RSI)’ (Figure 2.6). The objective was to quantify
the effects of the vertical accelerations and rocking of the deck on sliding isolators. The
authors concluded that the rocking motion had considerable effect on the individual
response of the RSI however this effect was dampened in the response of the total
system.
21
Figure 2.6 Resilient sliding isolation (RSI) (Iemura et al. 2005).
2.3.2. Coefficient of Friction, µ
Sliding isolators typically employ interfaces of steel and Polytetrafluoroethylene
(PTFE or Teflon) (Mokha et al. 1991). Teflon is extremely resistant to attack by corrosive
reagents or solvents (Billmeyer 1984). Furthermore, this polymer is not hard, but is
slippery and waxy to touch, and has very low coefficient of friction on most substances.
For all practical purposes the polymer is completely unaffected by water. Its thermal
stability is such that its mechanical properties do not change for long intervals (months)
at temperatures as high as C 250
o
. Resistance to wear and to deformation under load,
stiffness, and compressive strength of Teflon can be enhanced by the use of different
fillers such as glass fibers, graphite, carbon and bronze. The sliding of the two surfaces
of the FPS is an integral part of the forcedeformation response.
Mokha et al. (1990) underlined the absence of experimental data on the sliding
response of Teflon surfaces for velocities that are of interest to seismic isolation bearings.
The authors performed experiments to investigate the characteristics of the steelTeflon
sliding surfaces (Figure 2.7). The following conclusions were made from the test results:
1) The coefficient of friction increases and reaches a flat plateau beyond a certain
point with increasing velocity (Figure 2.8).
22
2) Sliding initiates at initial motion of the isolator and motion reversals (stick
slip). The magnitude of the static friction coefficient attained at these
instances is substantially larger than the magnitude during sliding.
3) The magnitude of the friction coefficient reduced with increasing normal force
acting on the plane of the sliding motion.
4) The effect of the dwell of the load, acceleration of the ground motion, and the
specimen size is negligible to the magnitude of the coefficient of friction
Figure 2.7 Test setup for Teflonsteel sliding surfaces (Mokha et al. 1990).
23
Sliding velocity (cm/s)
0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00
C
o
e
f
f
i
c
i
e
n
t
o
f
f
r
i
c
t
i
o
n
(
%
)
0
3
6
9
12
15
6.9 MPa
17.3 MPa
44.9 MPa
Figure 2.8 Dependency of the coefficient of friction on sliding velocity (Mokha et al.
1990).
Mosqueda et al. (2004) examined the effect of sliding velocity on the coefficient
of friction by tests performed on four FPS isolators under a rigid frame that was subjected
to bidirectional load histories. The velocity dependence of the coefficient of friction was
similar to the one depicted by Mokha et al. (1990) with the exception of a slight decrease
after 25.4 cm/s. The authors associated the reduction of the friction coefficient at high
velocities with the escalation of the temperature, and the deposition of the composite
material on the stainless steel surface. Test results showed that the steady state response
of the friction coefficient started to develop after 2.54 cm/s, which was earlier than 12.7
cm/s reported by Mokha et al. (1990). The authors concluded this steadiness favorable
for describing the coefficient of friction as a constant value postulated by the Coulomb’s
law of friction.
24
Jangid (2004) performed a parametric study to ascertain the effect of the friction
coefficient of FPS on the seismic response of buildings and bridges to nearfault ground
motions. The author analyzed a multistorey building model and a three span continuous
deck bridge model under near fault ground motions. Jangid (2004) concluded from the
analyses that there exists an optimum value of the coefficient of friction for the FPS that
reduces isolator displacements, which simultaneously minimizes the superstructure
acceleration. The bridge model revealed similar results to those obtained for buildings
which implied the presence an optimum value for the friction coefficient of the FPS that
minimizes pier base shear and deck accelerations. The author suggested the use of
coefficient of friction values between 0.07 and 0.19 for bridge isolators, and 0.05 to 0.15
for building isolators where the nearfault ground motions are expected.
Nakajima et al. (2000) analyzed a one degreeoffreedom model with a natural
period equivalent to their test setup. The authors stated that omitting the variation of the
friction coefficient as a function of the velocity in the modeling process caused
inconsistency with experimental results.
2.3.3 Bidirectional Coupling
Available earthquake records indicate that the horizontal ground motions are two
dimensional (PEER 2000). Theoretical studies have confirmed that there exists
significant coupling between the orthogonal components of the response in structures that
extend into the nonlinear range. Mokha et al. (1993) showed that neglecting the
orthogonal coupling of the steelTeflon interfaces in models results in underestimation of
displacements and overestimation of forces. Similar theoretical and experimental
research performed with the FPS confirmed these findings.
The AASHTO Guide Specifications (1999) provide a method for estimating
effects of bidirectional input by combining 100% plus 30% of orthogonal maxima. Warn
and Whittaker (2004) performed nonlinear timehistory analyses on an FPSisolated
25
bridge model to quantify the effect of bidirectional excitation and reviewed the accuracy
of the AASHTO Guide Specifications (1999) estimating maximum isolator
displacements. The authors concluded that bidirectional excitation produces significantly
larger isolator displacements than unidirectional excitation. The authors associated these
results with two factors, namely, the orthogonal component of excitation and the coupled
behavior of seismic isolators. It was concluded that the AASHTO Guide Specifications
(1999) procedures underestimate the isolator deformations. Anderson and Mahin (2004)
analyzed generalized bridge models to asses the accuracy of the AASHTO Guide
Specifications (1999) method to account for bidirectional response effects and reached
similar conclusions with Warn and Whittaker (2004).
Mosqueda et al. (2004) performed unidirectional and tridirectional tests on a
rigidblock frame supported by four FPS and examined how different mathematical
models conjectured the response of these bearings under different excitations. The
authors underlined that unidirectional tests overestimate the reduction of the friction
coefficient because the friction is computed from the motion over an invariable path that
accumulates the temperature. The authors stated that the bearings of fullscale bridges
abide a chaotic bidirectional path that endures lower temperatures and consequently
higher friction coefficients.
2.3.4 Large Deformation Effects
The large deformation aspects pertaining to FPS seismic isolation have been
theoretically addressed by Almazan and Llera (2002). The authors later cast the
nonlinear kinematic equations of the FPS response into an element format (Almazan and
LLera 2003). The authors reported that large deformation effects in the FPS was
influential on individual isolator response rather than the substructure response. It was
also noted that the orientation of the FPS was a controlling parameter in transferring large
deformation moments to the substructure or superstructure.
26
2.3.5 Other Parameters
The dynamic performance characteristics of FPS, specifically, stiffness, damping
and energy dissipation was found to have relatively low sensitivity to temperature
extremes (Zayas and Stanley 1999; HITEC 1998). The performance of the isolators did
not change at C 49
o
and C 40
o
− . Fatigue tests performed by 10,000 cycles of service
movements showed that deterioration from fatigue and wear was not evident (HITEC
1998). Test results showed that the FPS was mildly frequency dependent. The stiffness
and energy dissipation characteristics of the FPS generally increased with increasing
periods of the excitation.
2.4 Comparative Studies
Seismic isolators serve the common objective of decoupling the structure from the
horizontal components of the ground motion to minimize the seismic loads on the load
carrying components. However, there exist considerable differences in the vertical
response characteristics of elastomeric and sliding isolators. The LeadRubber Bearing is
a widely used elastomeric isolator (Buckle and Mayes 1990). The details of the LRB
characteristics are elaborated in Chapters 3 and 6. The conventional FPS is essentially
rigid under compression and has no tensile load capacity while the LRB has relatively
less compression stiffness and able to resist a limited amount of tensile loading (Kelly
2003; Almazan et al. 1998; Naeim and Kelly 1996). Both the postyield stiffness and the
yield force of the two types of isolators are known to be affected by the normal force
being imposed, but at a different rate and form (Almazan et al. 1998; Ryan and Chopra
2003). Normal forcedependent FPS models have been developed previously to show
that this effect may result in considerable variation on the estimated isolator response
(Almazan et al. 1998). However, LRB models that account for bidirectional coupling
27
has not yet been extended to account for normal forcedependency and implemented in
bridge analyses to the authors` knowledge. High variation of the normal loads may result
in fracture and a considerable change in the horizontal response of the isolators (Priestley
et al. 1996). It has been shown that excluding the inplane coupling of the orthogonal
response for the isolators may result in significant underestimation of the displacements
and forces for both type of isolators (Mosqueda et al. 2004; Jangid 2004). The bilinear
forcedeformation idealization of isolators allowed by the Specifications is based on the
assumptions that the response is unidirectional and the normal force acting on the
isolators is constant. Consequently, the unique response of the isolators may not be
adequately captured with this simplified modeling approach.
Dicleli and Buddaram (2005) compared the response of an idealized bridge
substructure utilizing seismic isolation devices that have the characteristic stiffness values
of leadrubber, high damping rubber and frictionbased bearings. The authors
highlighted that the peak isolator displacements decrease and forces increase as the post
yield stiffness of isolators increase. The effects of postyield stiffness on isolator
response were more pronounced for ground motions with low frequency content. Initial
stiffness of the isolators was noted to have negligible affects on the seismic response.
Matsagar and Jangid (2004) analyzed the effects of different yield displacement
and yield force properties in bilinear forcedeformation models of elastomeric and
friction isolators for a multistory building model. A higher yield displacement resulted
in substantial decrease in peak superstructure accelerations and marginal increase in
bearing displacement. An increase of the yield strength resulted in higher superstructure
accelerations and lower bearing displacements. This was attributed to the structure
remaining within the elastic range for a longer period of time thus decreasing the amount
of energy dissipation.
28
Ordonez et al. (2003) investigated the earthquake response of a twodegree of
freedom model as a function of different isolators. The four types of isolators were the
pure friction system, the friction pendulum system, the laminated rubber isolator, and the
New Zealand (leadrubber) isolator. The authors reported base displacements in
descending order as the leadrubber isolator, the laminated rubber isolator, the pure
friction system and the friction pendulum system. Ordonez et al. (2003) concluded that
base displacements are smaller but interstory drifts are larger for frictional systems
compared to neoprene systems.
Barrosso et al. (2002) compared the seismic performance of three different
structural control methods employed on steel frame buildings. The three schemes of
structural control were the: (1) friction pendulum system; (2) linear viscous dampers; and
(3) active tendon brace system. The authors identified normalized hysteretic energy and
interstory drifts as a definitive measure for describing the response of the frame. It was
concluded from the analysis that no control system was consistently better than another,
and that they all reduced the amount of energy dissipated by the structural system and
interstory drifts to negligible values.
Jangid and Kelly (2001) analyzed a two degreeoffreedom isolated building
model. The separate isolation schemes included elastic rubber isolators, the high
damping rubber bearings, the leadrubber bearing, and ElectricdeFrance system under
nearfault ground motions. Linear rubber and lead rubber bearings resulted in almost the
same response that corresponded to larger deformations but smaller superstructure
accelerations compared to other isolators. The authors concluded from the study that
highdamping rubber bearings was influential in reducing bearing displacement but
transmitted higher accelerations into the structure compared to other isolators. It was
concluded that there existed an optimal value of isolator damping when caused higher
accelerations to the superstructure exceeded. The authors showed that the Electricde
29
France acquires lower deformations compared to the rubber isolators but transmits
similar accelerations to the superstructure.
Sugiyama (2000) compared the seismic response of a continuous steel box girder
bridge isolated either with sliding or rubber isolation. The author reported higher
superstructure accelerations with sliding isolators compared to rubber isolators for strong
earthquakes, which is conversely different from the case in buildings. Additionally, the
author underlined that the difference of the bridge isolated was negligible for weak
earthquakes.
2.5. Parametric Studies
Studies aimed at investigating the effects of different design parameters in SIBs
are limited. Ghobarah (1988) investigated the parametric effects of LRB isolator
stiffness, pier stiffness and pier eccentricity on the response of SIB models. The author
showed that the increase of the flexibility of piers results in an increase in the force and
displacement demands at the abutments and reduced shears at the pier. The forces and
displacements at the abutments became larger as the pier offset was increased. Larger
elastic stiffness of the isolators reduced the deck displacements and the behavior
approached to a case of hinged supports.
Ghobarah and Ali (1988) compared the response of a seismically LRB isolated
and nonisolated bridge model and investigated the effects of different design parameters
in the response of the SIB response. The authors parametrically quantified the error for
assuming a rigid superstructure in the models in the fundamental period as a function of
deck to pier stiffness ratio. It was observed that the rigid deck assumption
underestimated the fundamental period and the error is more pronounced for stiffer
isolators. However, the error was within the order of 6% for typical highway bridges.
The authors parameterized the location of the effect of energy dissipation of the isolators
throughout the bridge. Isolation schemes involving higher energy dissipation at the
30
abutments compared to the piers resulted in considerable reduction in the seismic forces
acting on the piers. A parametric investigation of the magnitude of the yield force of
isolators showed that higher yield forces result in reduced shear forces in the pier and the
displacements of the deck.
Turkington et al. (1989) performed parametric analyses on a bridge seismically
isolated with combined LRB and elastomeric bearings to develop a design method. This
study investigated the effects of a range of isolator characteristics and the strengths of
various components of the bridge on the overall response. It was concluded that as long
as the isolators’ yield strength remained within 410% of the superstructure weight, the
seismic response is not significantly affected. The effectiveness of the isolators reduced
considerably as the superstructure flexibility increased. Increasing the LRB height,
which is equivalent to increasing the postyield stiffness, was found to result in greater
period shifts of the bridge.
Bridges may be constructed with a skew angle to accommodate traffic and site
conditions. Nielson (2005) defined bridge skew, θ , as “the angle measured between the
center line of the bridge supports and the line perpendicular to the bridge center line. It
was also noted that a 15.3% to 53% of all the bridge types considered in the Central and
Southeastern United states are skewed. An important structural aspect of skewed bridges
is that their vibrational modes do not uncouple in orthogonal directions as in the case of
nonskewed bridges (Maleki 2001b). Meng and Lui (2000) analyzed the response of a
skew concrete box girder bridge by accounting for deck flexibility and column boundary
conditions. The models of the same bridge were generated either as elastic shell
elements, as rigid deck or elastic beam elements. The authors noted that large skewness
may lead to torsional and lateral vibrations. Maleki (2001a) derived closed form
solutions for the translational and torsional periods of skewed bridges supported on
elastomeric bearings and have crossframe diaphragms. The author showed that the
31
fundamental period of the bridges with elastomeric bearings increases as the skew angle
increases and the second translational period is independent from the skew angle.
Dicleli et al. (2005) investigated the effects of soilstructure interaction (SSI) in
SIBs using iterative multimode response spectrum analyses. This study considered two
bridge types that had different superstructure and substructure weights. It was concluded
that SSI effects were negligible for SIBs with heavy superstructure and light
superstructure located on stiff soil. However, SSI effects were influential on isolator
forces and displacements for SIBs with light superstructure and heavy substructure
regardless of the stiffness of the soil.
Vlassis and Spyrakos (2001) performed parametric analyses to asses the influence
of SSI on the dynamic response of a SIB pier located over shallow soil stratum overlying
a rigid bedrock. The authors proposed closed form solutions to account for SSI effects in
design equations used by the AASHTO. It was shown that including the SSI effects
reduced the estimated base obtained from the AASHTO design procedures. Additionally,
it was concluded that the fundamental period of the structure may increase substantially
by accounting for SSI effects.
Thakkar and Maheshwari (1995) compared the response of a SIB model located
on soft, medium and hard soil. The authors concluded that stiffer soil resulted in higher
isolator forces but lower bending moments at the base of the piers.
2.6 Critical Appraisal
A review of the current stateoftheart illustrates that the mechanism of the FPS
has been thoroughly studied. The individual response of the conventional FPS has been
established with experimental and analytical research. However, there are still issues
pertaining to bridge seismic isolation, in particular with the FPS that need further
clarification. The three main gaps in the literature were identified as the following:
32
1) The FPS has a highly nonlinear response that involves the variation and
coupling of different parameters. Previous research considering the effects of different
aspects of nonlinearities in the response of the FPS showed that there may be a
significant divergence from a bilinear idealization. There is a need to develop a better
understanding of the modeling assumptions and the required level of accuracy for the
FPS in threedimensional (3D) bridge models.
2) The number of studies that compared the response of SIBs with different
isolator types is limited. Available studies in this area did not consider the vertical
components of ground motions, used twodimensional structural models and idealized the
forcedeformation response of the isolators as bilinear which overlooked some of the
distinguishing aspects of the response of the two isolator systems. There is a need for
further assessment of the comparative response of SIBs via detailed isolator models that
can capture the distinctions in the mechanism of sliding and elastomeric isolators.
3) Previous research on the parametric affects of design parameters in SIBs have
focused primarily on bridges utilizing elastomeric systems and was generally confined to
twodimensional models that excluded the vertical components of ground motions.
Further insight on the influence of design parameters in bridges isolated with the FPS is
needed.
33
CHAPTER 3
SEISMIC ISOLATION FOR BRIDGES
3.1 Introduction
The general topic of bridge seismic isolation requires an understanding of
structural analysis and dynamics, corresponding code aspects and the characteristics of
the isolators. Decoupling of the structure from the horizontal components of the ground
motion via an isolator results in a redistribution of seismic forces. The governing design
codes present different methods of analysis for SIBs. Various isolators are available for
bridges. The objective of this chapter is to elaborate on these aspects of bridge seismic
isolation. The effects of seismic isolation in bridges are illustrated on a simplified bridge
model via modal analysis. An outline of the analysis methods in the governing design
code of bridge seismic isolation is also presented. The characteristics of most commonly
used isolators in bridges are described.
3.2 General Features of a Seismically Isolated Bridge (SIB)
Typically, the primary objective of bridge seismic isolation is the protection of the
piers and the foundations and in some cases the abutments (Skinner et al. 1993). This is
accomplished by installing seismic isolators between the superstructure and these
components. The superstructure is seldom of concern due to inherent strength in design
for vehicle loads. In superstructure isolation, the substructure is not isolated from the
ground motions but decoupled from the relatively larger mass of the superstructure. The
effects of seismic isolation is examined through an idealized bridge structure with a
lumped superstructure and substructure mass, m
1
and m
2
, respectively, substructure
lateral linearelastic stiffness k
1
and isolator linearelastic stiffness k
2
. Each lumped mass
is designated a lateral degreeoffreedom, u
1
and u
2
(Figure 3.1).
34
k
1
k
2
m
1
m
2
u
1
u
2
Figure 3.1 Idealized seismicisolated bridge substructure.
The mass and the stiffness matrices of this system are:
m=
(
¸
(
¸
2
0
0
m
m
1
(3.1)
k=
(
¸
(
¸
−
− +
2 2
2 2 1
k k
k k k
(3.2)
The characteristic equation (frequency equation) is (Chopra 2000):
  0 det
2
= − m k
n
ω (3.3)
where
n
ω is the natural frequency of the n
th
mode. Given the positive definite property
of k and m, the expanded form of Equation 1 has real and positive roots for
2
n
ω :
0 ) (
2 4
= + − − − +
2 1 n 1 2 2 2 2 1 n 2 1
k k m k m k m k m m ω ω (3.4)
Consider the structure to have m
1
=0.70 kNs
2
/cm, m
2
=0.18 kNs
2
/cm, and k
1
=280.2
kN/cm. From the solution of Equation 3.4, the natural period of the n
th
mode, T
n
=
n
/ ω π 2 , as a function of k
2
/ k
1
is as given in Figure 3.2. Large values of k
2
/ k
1
imply a
rigid layer between m
1
and m
2
. In this case the structure essentially reduces to a
cantilever with a lumped mass at the tip and has a T
1
=0.35 s. It is observed that for
values of k
2
/ k
1
< 1, there is a notable increase in the fundamental period of the structure.
The effect of introducing a flexible layer to the second period is negligible. The
35
installation of a flexible layer, where k
2
/k
1
<1, is an effective approach for increasing the
fundamental period of the structure with little influence on the second period.
k
2
/ k
1
0.0 0.2 0.4 0.6 0.8 1.0 1.2
P
e
r
i
o
d
,
T
n
(
s
)
0
1
2
3
4
T
1
T
2
k
1
k
2
m
1
m
2
u
1
u
2
Figure 3.2 First, T
1
, and second, T
2
, mode structural periods as a function of k
2
/ k
1
(m
1
=0.70 kNs
2
/cm, , m
2
=0.18 kNs
2
/cm, and k
1
=280.2 kN/cm ).
To highlight the dynamics of superstructure isolation, consider the system to have
seismic isolation with the objective of achieving three times increase from the non
isolated period of 0.35 s. From Figure 3.2, this corresponds to approximately k
2
=26.3
kN/cm in which the first two periods become T
1
=1.05 s and T
2
=0.15 s. The n
th
mode
shape,
n
φ , is determined from:
  0
2
= −
n n
φ ω m k (3.5)
The solution of Equation 3.5 by assuming unit displacements at the first degrees
offreedom in each of the two modes is given in Figure 3.3. It is observed that in the first
36
mode shape of the structure, the isolator undergoes significant deformation compared to
the substructure. In the second mode the structural deformation is larger, however, this
mode is subsequently shown to make insignificant contribution to the earthquakeinduced
forces of the structure, based on its low participation factor.
(a) (b)
Figure 3.3 Structural shapes for (a) Mode 1 (b) Mode 2.
The spatial distribution of the effective earthquake forces are defined by :
∑
=
=
N
n
n n
1
φ Γ m s (3.6)
where
n n n
/M L = Γ , mi
T
n n
L φ = , and
n
T
n n
M φ φ m = , and i is the influence vector. The
modal expansion of these forces and the modal static responses for the base shear, V
bn
st
,
and base moment, M
bn
st
, for the n
th
mode are given in Figure 3.4. An important
implication of this result is that the first mode forces are essentially the same as the total
forces. The second mode components of the static forces are negligible compared to
those in the first mode. Consequently, the second mode response which involves the
structural deformations is expected to make little contribution to the earthquake response
of the structure.
37
Figure 3.4 Modal expansions of effective earthquake forces and modal static responses
for the base.
The earthquake response of the structure at any given time, t, is obtained by
combining the contributions of all modes:
∑
=
=
N
n
n
st
n
t A r t r
1
) ( ) ( (3.7)
where
st
n
r is the modal static response, ) ( ) ( t D ω t A
n
2
n n
= is the pseudoacceleration
timehistory and ) (t D
n
is the displacement of the n
th
mode. Particular interest is the peak
response of the structure over the duration of the earthquakeinduced response. This is
obtained from:
n
st
n n
A r r =
O
(3.8)
where A
n
is the peak value of ) (t A
n
for a particular T
n
provided by a pseudoacceleration
spectrum. Consider the pseudoacceleration design spectrum of AASHTO in the
Memphis, TN region with Type II soil profile and acceleration coefficient A=0.2 (Figure
3.5).
The base shear calculation of each mode and their combination via the square
rootofsumofsquares (SRSS) is given in Table 3.1. Despite a larger pseudo
acceleration that corresponds to the second mode of the response, this effect is dampened
38
due to considerably smaller static response. The design pseudoacceleration for the non
isolated structure corresponds to 0.5g and a peak base shear of 0.438 kNs
2
/cm. The
seismic isolation scheme reduced the base shear by 72%.
Period, T (s)
0.0 0.5 1.0 1.5 2.0
S
p
e
c
t
r
a
l
a
c
c
e
l
e
r
a
t
i
o
n
,
S
a
(
g
)
f
o
r
ξ
=
5
%
0.1
0.2
0.3
0.4
0.5
0.6
0.278g
0.50g
T
=
0
.
3
5
s
T
=
0
.
1
5
s
T
=
1
.
0
5
s
Figure 3.5 Pseudoacceleration design spectrum for AASHTO Type II soil profile with
acceleration coefficient A=0.2.
Table 3.1 Calculation of base shear
Mode A / g Static Base shear, V
b
st
(kNs
2
/cm) Peak Base shear, V
bo
(kNs
2
/cm)
1 0.278 0.875 0.244
2 0.500 0.145 0.073
SRSS 0.255
3.3 Bridge Seismic Isolation in Design Codes
The American Association of Highway and Transportation Officials (AASHTO)
Guide Specifications for Seismic Isolation Design provides the “guide specifications for
39
the seismic isolation design of highway bridges”. This is a supplemental document to the
AASHTO Standard Specifications for Highway Bridges Division IA: Seismic Design.
The first Seismic Isolation Guide Specification for bridges by the AASHTO was
available in 1991. By this time elastomeric bearings were primarily used in bridge
seismic isolation (Stanton 1998). As new designs became available by 1995 the first
Seismic Isolation Guide Specification was essentially rewritten in 1997 to address the
advances in the industry. Four procedures are available in the Guide Specification for the
analysis of SIBs. The following is a brief overview of these procedures.
3.3.1 Procedure 1: Uniform Load Method
The Uniform Load Method is essentially the method of approximating earthquake
loads with an equivalent static force. This statically equivalent seismic force is:
W C F
s
= (3.9)
where W is the total vertical load for design of the isolation system and C
s
is the elastic
seismic response coefficient computed from:
W
d K
C
eff
s
= (3.10)
where d is the total deck displacement relative to ground and K
eff
is the sum of the
effective linear stiffnesses of all bearings and substructures. In calculating the K
eff
, the
configuration, flexibility and individual stiffnesses of the isolator units and substructure
shall be taken into account:
∑


¹

\

+
=
j eff sub
eff sub
eff
k k
k k
K (3.11)
40
where Σ extends over all substructures, and k
sub
is the substructure lateral stiffness and
k
eff
is the isolator unit lateral stiffness calculated at maximum displacement capacity
(Figure 3.6). The displacement d (mm) is given by:
B
T AS
d
eff i
250
= (3.12)
where A is the acceleration coefficient defined in Table C31 of the Guide Specification,
S
i
is the numerical coefficient per sitesoil profile as defined in Table 51 of the Guide
Specification, T
eff
is the period of the seismically isolated structure in seconds in the
direction under consideration and B is the damping coefficient.
Figure 3.6 Single substructure and isolator idealization.
3.3.2 Procedure 2: Single Mode Spectral Method
The Single Mode Spectral Method is the same approach in article 4.4 of
AASHTO Standard Specifications (Division IA: Seismic Design). However, the method
is simplified for SIBs due to the rigid body deformation of the superstructure. The first
three steps of the procedure is devoted to finding the deflection, slope and moments in the
41
bridge superstructure via the double integration method. These results are used to
establish the intensity of the equivalent static seismic loading applied to represent the
primary mode of vibration, p
e
(x). For SIBs the Guide Specification Commentary C7.2
gives:
s e
C x x p ) ( w ) ( = (3.13)
where C
s
is calculated from Equation 10 and w(x) is the deadloadperunit length of the
bridge superstructure. The loading, p
e
(x), is used for calculating resulting member forces
and displacements.
3.3.3 Procedure 3: Multimode Spectral Method
Different from the previous two procedures, the Multimode Spectral Method
requires a detailed model of the bridge in a computer program. The Guide Specifications
refers to article 4.5 of AASHTO Standard Specifications (Division IA: Seismic Design)
for the specifics on modeling. Isolators are modeled by their effective stiffness based on
design displacements. The procedure is essentially performing an equivalent linear
response spectrum analysis.
3.3.4 Procedure 4: TimeHistory Method
The Time History Method of analysis is the most sophisticated among the four
permitted in the Guide Specification (Stanton 1998). This Procedure involves the time
history analysis of the bridge models with isolation bearings that have nonlinear
deformation characteristics. The AASHTO Guide Specifications (1999) Section 7 on
Analysis Procedures states that: “To simplify the nonlinear behavior of the isolator unit,
a bilinear simplification may be used”. This bilinear hysteretic model is characterized
by the elastic stiffness, k
u
, postelastic stiffness, k
d
, characteristic strength, Q
d
, yield force,
F
y
, and maximum bearing displacement,
max
∆ (Figure 3.7). The bridge model is to be
42
subjected to the orthogonal components of no less than three ground motion records. The
ground motion records may be frequencyscaled to match the characteristics of the
corresponding site. The 5%damped response spectrum is established by taking the
square root of the sum of the squares of the orthogonal components of the ground
motions. The maximum response parameter among the three ground motion records
determines the design. If seven or more ground motions are used, than the average value
of response parameter may be used for design. This method is required if the structure
acquires effective periods greater than 3 seconds.
Figure 3.7 Characteristics of bilinear isolation bearings per AASHTO Guide
Specifications.
3.4 Descriptions of Common Isolators
Isolators can be classified as sliding and elastomeric (Taylor and Igusa 2004;
AASHTO 1999). The Guide Specifications specifically mention two isolator types from
each category: LeadRubber and High Damping Rubber for elastomeric, and Friction
Pendulum System and Eradiquake for sliding. The development of elastomeric isolators,
(shown in Figure 3.8) is considered an extension of elastomeric bridge bearings and
43
bearings for vibration control of buildings (Naeim and Kelly 1996). These systems have
become a practical tool for seismic isolation in the late 70s. Elastomeric isolators are
fabricated through a process called vulcanization, which is the bonding of steel plates
with rubber (HITEC).
Figure 3.8 Typical elastomeric isolator (Taylor and Igusa 2004).
Sliding isolators have been modified from Teflon sliding bearings that are
commonly used in bridge applications to accommodate movements from factors such as
thermal expansions, creep, shrinkage or prestressing (HITEC 1998). These isolators
typically possess two surfaces with different finish to slide over one another (AASHTO
1999). Naeim and Kelly (1996) noted that a purely sliding system with talc proposed by
Johannes Avetican Clanterients in 1909 was the earliest seismic isolation system (Figure
3.9). However, widespread use of the sliding isolators corresponds to early 1990s.
Despite the emphasis of this study on the sliding isolator FPS, a general overview of the
four systems mentioned in the Guide Specifications are presented for completeness.
44
Figure 3.9 Clantarient`s base isolation system using a layer of talc as the isolating
medium (Naeim and Kelly 1996).
3.4.1. The Eradiquake System (EDS)
The Eradiquake System (EDS) isolator is made up of essentially two components:
(1) a sliding multirotational bearing assembly (2) a maintenancefree restoring device
called the Mass Energy Regulator (MER) (Figure 3.10). The system restores through the
MER and simultaneously dissipates energy via the steel and composite sliding bearing at
the center. It is possible to design the bearing to have different energy dissipation and
stiffness characteristics in the inplane orthogonal directions. The value of Q
d
is a
function of the dynamic friction coefficient of the central sliding bearing. The value of
K
d
is governed by the properties of the MER (AASHTO 1999).
45
Figure 3.10 The Eradiquake seismic isolation bearing.
3.4.2. The High Damping Rubber System (HDRS)
The High Damping Rubber System (HDRS) is essentially the aforementioned
elastomeric bearings with the only difference of having a modified rubber compound that
acquires increased damping characteristics (Figure 3.11) (Naeim and Kelly 1996). High
damping rubber layers deform under shear to reduce earthquake loads and dissipate
energy (HITEC 1998). The isolator stiffens and acquires a higher level of energy
dissipation at large deformations due to the strain crystallization process in the rubber
(Naeim and Kelly 1996). The values of Q
d
and
K
d
are a function of the additives to the
rubber (AASHTO 1999).
46
(a) (b)
Figure 3.11 (a) High damping rubber bearing used in the earthquake simulator tests with
dimensions in mm (b) corresponding forcedeformation hysteresis (Kikuchi and Aiken
1997).
3.4.3. The Lead Rubber Bearing (LRB)
The Lead Rubber Bearing (LRB) consists of steel plates, rubber and a lead core
(Figure 3.12). A lead core is inserted down the center of the bearing for energy
dissipation and stiffness (Priestley et al. 1996). Lead is a feasible option because it yields
in shear at relatively low stresses, 10 MPa, and has good fatigue properties (Skinner et al.
1993). The steel layers placed between the rubber serves to limit the edgebulging of the
rubber (Tyler 1991). Additionally, the steel plates force the lead plug bearing to deform
in shear (Naeim and Kelly 1996). As a multilayered elastomeric type bearing, the LRB is
susceptible to a buckling type of instability. The forcedeformation response is typically
modeled as bilinear (Naeim and Kelly 1996). The value of Q
d
is a function of the lead
core and the value of
K
d
is a function of the rubber (AASHTO 1999). An important
characteristic of the LRB is that the in cold temperatures, natural rubber causes a
significant increase in the Q
d
(AASHTO 1999).
47
The LRB is a type of isolator widely used in bridge applications (Buckle and
Mayes 1990). It has been also reported based on information provided by the
manufacturers that the cost, size and energy dissipation of the FPS and LRB may be
comparable in bridge applications (Dicleli 2002). Consequently, the LRB is used in
Chapter 6 to compare the applications of sliding versus elastomeric seismic isolation of
bridges. The mechanical properties of the LRB is elaborated further in Chapter 6.
Figure 3.12 Typical lead rubber bearing (LRB) (Taylor and Igusa 2004).
48
(a) (b)
Figure 3.13 (a) Lead rubber bearing (LRB) used in the earthquake simulator tests with
dimensions in mm (b) corresponding forcedeformation hysteresis (Kikuchi and Aiken
1997).
3.5 Conclusion
Seismic isolation of bridges is an effective approach for reducing the forces
imparted by earthquakes. The favorable effects of seismic isolation is achieved
essentially by decoupling the response of the structure from the ground motion and
shifting the period for lower pseudo acceleration in the design spectrum. It is concluded
from the modal analysis of a simplified bridge model that the stiffness characteristics of
the isolators significantly control the dynamic response of the system. There are four
analysis procedures available in the governing code of SIBs, the Guide Specifications.
The most sophisticated one of these procedures, the timehistory method, will be the
basis of the analyses throughout this study. An overview of common isolators mentioned
in the Guide Specifications has been presented.
49
CHAPTER 4
FRICTION PENDULUM SYSTEM (FPS) MODELING
4.1 Introduction
This chapter is devoted to the development of a new finite element (FE) model
that can represent the inherent nonlinear and coupled response of the Friction Pendulum
System (FPS). The general characteristics of the simplified bilinear model of the isolator
are explained. The analogy between the simplified bilinear response of the FPS and the
bilinear modeling presented in the AASHTO Guide Specifications (1999) is described.
The equations and modeling techniques used to represent the nonlinear and coupled
response of the isolator is reviewed, and the nonlinear kinematics of the isolator response
is developed. The implementation of the response of the isolator into OpenSees via
developing new C++ classes is explained. The FE model is verified with experimental
data. Preliminary results on the influence of different modeling assumptions for the FPS
are highlighted.
4.2 Simplified Isolator Response Modeling
The mobilized response of the FPS is representative of a mass sliding on a
perfectly spherical surface with a coefficient of friction µ , and a radius of curvature R.
The two components of the intrinsic forces of the FPS consists of the pendulum motion
of the mass, f
R
, and the friction between the mass and the sliding surface,
µ
f . Assuming
small deformations, the unidirectional forcedeformation response of the FPS is (Zayas et
al. 1987):
50
}
R
µ
f
f
δ
R
N
δ N f + =
48 47 6
&
) sgn( µ (4.1)
where N is the normal force acting on the sliding surface, R is the radius of the concave
surface, δ is the sliding deformation, δ
&
is the sliding velocity, and ) sgn(δ
&
is the signum
function. The signum function is equal to +1 or 1 depending on whether δ
&
is negative
or positive, respectively (Figure 4.1). The friction response,
µ
f , and the pendulum
motion response, f
R
, in Equation 4.1 are representative of plastic and elastic models,
respectively, illustrated in Figure 4.2.
Figure 4.1 The signum function.
(a) (b)
Figure 4.2 The intrinsic response components (a) friction,
µ
f , and (b) pendulum, f
R
.
51
The combination of f
R
and f
µ
corresponds to a unidirectional rigidplastic
hysteretic model given in Figure 4.3. The AASHTO Guide Specifications (1999) Section
7 on Analysis Procedures states that: “To simplify the nonlinear behavior of the isolator
unit, a bilinear simplification may be used” (Figure 4.4). This bilinear hysteretic model
is characterized by the elastic stiffness, K
u
, postelastic stiffness, K
d
, characteristic
strength, Q
d
, yield force, F
y
, and maximum isolator displacement,
max
∆ . If the yield
displacements of steelTeflon sliding surfaces reported in the order of 0.050.02 cm by
Constantinou et al. (1990) for conditions relevant to the FPS is considered, Figure 4.3
takes up the characteristics of the bilinear model in Figure 4.4. This bilinear model for
the FPS given in Figure 4.3 is based on the assumptions that: (1) N is constant; (2) µ is
constant; (3) the horizontal response is uncoupled in the orthogonal directions; and (4)
isolator deformations are small and planar. The following sections elaborate on these
aspects and how they may be incorporated into the response.
Figure 4.3 Forcedeformation characteristics of the unidirectional rigidplastic response
of the FPS.
52
Figure 4.4 Forcedeformation characteristics of bilinear isolators.
4.3 Normal Force
The normal force, N, acting on the FPS is inherent in both resisting force
components,
µ
f and f
R
, of the response. An increase in the magnitude of N is indicative
of a higher yield force which may delay the mobilization of the FPS under dynamic loads
and a higher postyield stiffness which may reduce the flexibility of the isolator.
Additionally, N changes the magnitude of µ , however this relationship is discussed
subsequently. The conventional FPS does not have resistance in tension and it is
approximately rigid in compression (Zayas et al. 1987). This behavior closely matches
the response of a zerolength gap element defined with a force:
¹
´
¦
>
≤
=
0 0 if 0 0
0 0 if
g
g
. .
. δ k
f
g g
g
δ
δ
(4.2)
where k
g
is a high compression stiffness and
g
δ is the deformation (Figure 4.5).
Modeling the vertical response of the FPS with a gap element allows simultaneously the
monitoring of the variations in N and capturing the effects of uplift and impact in the FPS
(Almazan and Llera 2003).
53
Figure 4.5 Gap element forcedeformation model.
4.4 Coefficient of Friction
The coefficient of friction, µ , in addition to the material properties of the surface,
was found to be primarily a function of δ
&
and N (Mokha et al. 1990). Accurate
mathematical models have been developed by Constantinou et al. (1990) to capture the
value of µ for a range of δ
&
and N that is of interest to the response of the FPS. The
influence of δ
&
on the µ was approximated via the aid of experimental results as:
) δ a (
f max
e D f µ
&
−
− = (4.3)
where, f
max
and f
min
are the values of coefficient of friction at large and small sliding
velocities, respectively, D
f
is the difference between f
max
and f
min
, and a is a constant,
having units of time per unit length, that controls the variation of the coefficient of
friction with velocity. Only the dependency of f
max
to pressure, P, is considered in this
study as the influence of pressure on f
min
and
a were shown to be negligible by Tsopelas et
al.(1994). The term f
max
as a function of P was given as:
) (εP tanh D f f
fmax max,0 max
− = (4.4)
where f
max,0
and f
max,p
are the values of f
max
at very low and high pressures respectively,
D
fmax
is the difference between f
max,0
and f
max,p
, and ε is a constant that controls the
variation of f
max
between very low and very high pressures (Figure 4.6). Equations 4.3
54
and 4.4 may be updated for the values of δ
&
and N in an iterative solution scheme to
account for the changes in µ .
(a)
(b)
Figure 4.6 Variation of the coefficient of friction with (a) velocity of sliding; and (b)
isolator contact pressure (Roussis and Constantinou 2006).
4.5 Bidirectional Coupling
Bidirectional motion may commence in the FPS isolator subject to
multidirectional excitations. The two important characteristics of the bidirectional sliding
motion are that: (1) the inplane forcedeformation response of the sliding is isotropic;
and (2) the behavior shifts from stick and slip conditions (Constantinou et al. 1990). A
simplified approach for modeling the planar frictional response is to consider two
independent unidirectional elements according to the Coulomb’s model in the orthogonal
55
x and y directions of the horizontal plane based on Equation 4.1. In this case the planar
frictional force is:
(
(
¸
(
¸
=
(
¸
(
¸
=
) sgn(
) sgn(
y
x
δ
δ
µ
&
&
N
f
f
µy
µx
µ
f (4.5)
where,
x
f
µ
and
x
f
µ
are the components of the friction force, and
x
δ
&
and
y
δ
&
, are the
components of the velocity in the x and y directions, respectively. However, the
calculation of
µ
f with two independent Coulomb’s models for plane motion,
overestimates the resistance, produces inaccuracies in capturing stickslip conditions and
raises complications in numerical computation (Constantinou et al. 1990).
The planar sliding forcedeformation response of the FPS is isotropic. This
implies that the interaction surface of
x
f
µ
and
x
f
µ
is circular and the resultant sliding
friction force magnitude, 
µ
f 
y
2
x
f f
µ µ
+ = , is equal to µ N regardless of the sliding
direction. The planar frictional force that satisfies these conditions during sliding is:
(
¸
(
¸
=
(
¸
(
¸
=
) sin(
) cos(
θ
θ
N
f
f
µy
µx
µ
µ
f (4.6)
where, ) ( tan
1
x y
δ / δ
& &
−
= θ defines the direction of the sliding motion. Unlike the case
presented by Equation 4.5, the components of
µ
f in Equation 4.6 are insensitive to the
variations in the magnitudes of respective sliding velocities. Consequently, if the
response of
µ
f is modeled by Equation 4.5, the magnitude of 
µ
f  during sliding ranges
between µ N , if sliding along the x or y axes, and µ N 2 , if sliding in a path along the 45
degree direction. The shape of the interaction surface between
x
f
µ
and
x
f
µ
is circular if
Equation 4.5 is used and square if Equation 4.6 is used (Figure 4.7).
56
Figure 4.7 Frictional interaction surface (a) uncoupled (b) coupled response.
The response of the sliding system may shift between two phases: (1) the sliding
phase, where motion commences in both directions; and (2) the sticking phase, where one
or both of the components of the velocity are zero or very low (Constantinou et al. 1990).
In the Coulomb’s model, the transition between the two phases is independent from the
magnitude of the sliding velocity and discontinuous (Figure 4.1). Constantinou et al.
(1990) reported that the response of Teflonsteel sliding surfaces predicted via the
Coulomb’s model contained highfrequency components that did not prevail in the
experiments. This was attributed to the significantly more sticking phases that developed
in the Coulomb’s model as a result of independence of resistance to the magnitude of the
sliding velocity. There are complications in using Coulomb’s model in numerical
solutions because of the low rate of convergence caused by this discontinuity (Feldstein
and Goodman 1973) and the difficulty of extending Equation 4.5 to plane motion (Younis
et al. 1983).
Constantinou et al. (1990) extended the work of Park et al. (1986) and presented
bidimensional hysteretic parameters, = η [
y x
η η ]
T
,
to evaluate
µ
f in planar steel
Teflon sliding interfaces. The frictional force vector in this case is defined as:
η f
µ
Nµ = (4.7)
57
where η is dimensionless hysteretic parameters that evolve according to the following
coupled set of differential equations:
(
(
¸
(
¸
(
(
¸
(
¸
− −
− −
=
(
¸
(
¸
=
y
x
2
x x y x x
y x y
2
x x
y
x
δ
δ
η a A η η a
η η a η a A
∆s
1
η
η
&
&
&
&
η (4.8)
where
x
δ
&
and
y
δ
&
, are the components of the sliding velocity in the x and y directions,
respectively, ) sgn(
x
x
.
x
η δ γ β a + = , ) sgn(
y
y
.
y
η δ γ β a + = ,
Y
∆ is the yield displacements,
A, β , γ are dimensionless constants that control the shape of the hysteretic loops.
Constantinou et al. (1990) showed that for 1 A = + ) /( γ β , the solution of Equation 4.7:
(1) describes a circular interaction curve; (2) for sliding conditions the hysteretic
parameters are θ cos =
x
η and θ sin =
y
η , where ) arctan( x
.
y
.
/ δ δ θ = ; and (3) for sticking
conditions 0 1. < η . The solution of the of the coupled differential Equation (4.8) can be
solved via numerical algorithms presented for the common computer languages like C++
(Lee and Schiesser 2004; Press et al. 2003). Additionally, common numerical calculation
software such as Mathcad (Pritchard 1998) and Matlab (Shampine et al. 2003) have a
variety of builtin functions to solve coupled differential equation systems. In the model
being developed for this study, the semiimplicit method presented by Rosenbrock (1963)
have been utilized as the solution algorithm due to its: (1) relative simplicity; (2) ability
to handle stiff problems; (3) common stability; and (4) acceptable accuracy (<10
4
– 10
5
)
(Press et al. 2003).
4.6 Large Deformation Moments
In bridge applications, the FPS is installed above piers and abutments as either the
concave dish facing up or down (EPS 2002). The orientation of the FPS controls whether
the P ∆ moments occur at the structural members below or above the isolator (EPS
58
2002). This unique feature of the FPS does not have implications on the inplane force
deformation response and allows for diverting P ∆ moments from weak elements of the
structure (Almazan and Llera 2003). Figure 4.8 is a schematic of the displaced shape of
an FPS between a simplified bridge superstructure and the column. The normal force, N,
is transmitted through the slider to the concave dish. Assuming that the rotations at the
superstructure and the top of the column are negligible, the displaced configuration of the
FPS results in an internal moment M=Nδ . This internal moment, M, is balanced at the
tip of the column if the concave dish is at the bottom and by the superstructure if the
concave dish is at the top. Almazan and Llera (2003) presented a nonlinear
transformation matrix for their FPS model to account for this aspect, which is elaborated
in the subsequent section.
M
N
Column
Superstructure
Figure 4.8 Deformed shape of the seismic isolator between the superstructure and the
substructure with concave dish at the (a) bottom (b) top.
4.7 Mathematical Model
The exact three dimensional (3D) kinematics equations considering large
deformation effects of the FPS were developed by Almazan and Llera (2003). The zero
59
length element developed in OpenSees to model the response of the FPS is constructed
based on these principles by assuming no nodal rotations. The author does not claim any
innovation for implementing this simplification and refer the reader to Almazan and Llera
(2003) for a more detailed presentation of these principles. Here, only a brief summary of
the mathematical formulation is presented with similar nomenclature as the original
equations. The zerolength element is 3D with 6 degrees of freedom (DOF) per node
(Figure 4.9). This model accounts for the variations of the N via an inherent gap element
described by Equation 4.2 and the variations of the µ via Equations 4.3 and 4.4 at each
integration time step. The coupling of the sliding forces are included via the hysteretic
parameters, = η [
y x
η η ]
T
, as described in the previous sections. The P ∆ moments are
transferred to the nodes of the model via a nonlinear transformation matrix.
Figure 4.9 Schematic view of the model.
60
Assuming that the nodal rotations are negligible, the local slider and global
coordinates coincide. The nodal deformations of the element are defined as u=[u
(J)
u
(I)
]
T
, where u
(J)
=[u
x
(J)
u
y
(J)
u
z
(J)
r
x
(J)
r
y
(J)
r
z
(J)
]
T
and u
(I)
=[u
x
(I)
u
y
(I)
u
z
(I)
r
x
(I)
r
y
(I)
r
z
(I)
]
T
define
the motions of nodes J and I, respectively. The instantaneous position and velocity of the
slider is defined by the vectors = δ [
z y x
δ δ δ ] and = δ
&
[
z y x
δ δ δ
& & &
], respectively
(Figure 4.10). The slider’s motion is bounded by the spherical surface of the concave
dish defined as:
0
2 2 2
= − + + = ) ( R δ δ δ G
z y x
(4.9)
The unitary vectors in the outward normal direction and the tangential to the trajectory of
the slider are:
 
T
1
) ( n R δ δ δ
R G
G
z y x
− =
∇
∇
= (4.10)
and
T
cos cos
(
(
¸
(
¸
= = α α α sin
η
η
y
x
η η
δ
δ
s
&
&
(4.11)
respectively (Figure 4.10), where,





¹

\

−
+
=
z
y
y
x
x
R δ
η
δ
η
δ
α
η η
arctan (4.12)
denotes the angle between the frictional force component and the xy plane. The local
slider restoring forces for all phases is:
f = 
T
z y x
f f f =Nr (4.13)
where, r, is the restoring force orientation vector that is constituted from the normal and
tangential components of the slider as:
61
}
8 7 6
component tangential
component normal
s η n r µ + = (4.14)
The concavity and frictionbased components of the isolator are N n f
R
= and
s η f
µ
µ N = , respectively. The normal force in the isolator is:
z
g
r
f
N = (4.15)
where r
z
is the axial component of the vector r.
Figure 4.10 Deflections and forces acting on the slider.
The nodal force vector F=[F
(J)
F
(I)
]
T
, where F
(J)
=[F
x
(J)
F
y
(J)
F
z
(J)
M
x
(J)
M
y
(J)
M
z
(J)
]
T
and F
(I)
=[F
x
(I)
F
y
(I)
F
z
(I)
M
x
(I)
M
y
(I)
M
z
(I)
]
T
define the forces at nodes J and I, respectively, is:
T
L
ˆ
= F f (4.16)
where
T
L
ˆ
is the transform of the nonlinear transformation matrix defined as:
(
(
(
¸
(
¸
± − ±
− −
± ± ±
=
0 1 0 0 0 0 0 1 0 0
0 0 1 0 0 0 0 0 1 0
0 0 0 1 0 0 0 0 0 1
x y
x z
y z
L
ˆ
δ δ
δ δ
δ δ
m
m
(4.17)
The top and bottom signs in Equation 4.17 are used to differentiate between the
downward and upward positions of the FPS, respectively. The transformation matrix, L
ˆ
,
62
depends exclusively on geometry and is nonlinear to account for the variation of the P ∆
moments. The exclusion of the vertical rise in the concave dish corresponds to
z
δ = 0, α
= 0, and N = f
g
. In this case Equation 4.13 becomes:
f = 
T
z y x
f f f =
T
1
(
¸
(
¸
− + + µ η
R
δ
µ η
R
δ
N
y
y
x
x
(4.18)
4.8 Evaluation Platform
The Open System for Earthquake Engineering Simulation (OpenSees) is an open
source software framework for simulating the earthquake response of structural and
geotechnical systems (Mazzoni et al. 2006). OpenSees has an open source object
oriented architecture in the C++ programming language that maximizes its modularity,
thus making it a viable choice for research purposes. OpenSees is chosen as the
simulation platform for this study mainly because of its ability to access its source code
to incorporate new material and element models without the need to perform changes in
the existing solution algorithms. Material and element models describing the hysteretic
response of new structural members can be developed as C++ classes and inserted into
the existing library of OpenSees for analyses. This powerful attribute of OpenSees
allows researchers to analyze a wide range of aspects of innovative materials and
elements in larger models. Although OpenSees provides a variety of hysteretic uniaxial
forcedeformation response models, none of them was found to be capable of adequately
representing the mathematical model described previously. Consequently, a 3D zero
length element
1
class and a complimentary material class that can optionally
include/exclude the modeling aspects of the FPS is implemented into OpenSees. The
implementation of this procedure requires a combined knowledge of C++ programming,
1
Italicized word is used for C++ class
63
objectorienteddesign and the definitions and architecture of OpenSees framework.
There is no source, to the author’s knowledge, that provides a stepbystep explanation of
adding new elements or materials into the existing library of OpenSees. The annually
held OpenSees Developer Workshop at Berkeley, California gives limited yet valuable
insight about these procedures (McKenna 2005a; McKenna 2005b).
A typical structural member in OpenSees is constructed via the material and
element classes. The material class receives nodal displacements and velocities as input
from the element class and returns trial force and tangent stiffness values according to a
predefined forcedeformation law. The element class is responsible for the generation of
time dependent stiffness and transformation matrices and equilibrium iterations. The
element class has access to all the forces from the materials in different directions. The
new material class developed for the FPS model, FPSmaterial, is responsible for
evaluating the corresponding components of f via Equation 4.13. In addition to the
strain and velocity, Equation 4.13 requires that, N, the corresponding component of η,
and µ , at each integration time step be delivered from the element class. This is
accomplished by adding these parameters to the corresponding abstract method via
overriding (see Appendix A). The new element class developed for the FPS model,
FPSelement, is designed to interact with FPSmaterial. In addition to the typical parent
element class in OpenSees, the FPSelement is designed to evaluate large deformation
effects in N via Equation 4.2, solve the parameters of η via the semiimplicit Rosenbrock
(1963) method in Equation 4.8, update the values of µ via Equations 4.3 and 4.4 and α
via Equation 4.12, and to construct the L
ˆ
via Equation 4.17 at each time integration step
(Figure 4.11). The C++ script for differential equation solution with the Rosenbrock
Methods presented by Press et al. (2003) has been incorporated into the FPSelement class
to solve the parameters of η.
64
Figure 4.11 FPSelement and FPSmaterial interaction.
4.9 Verification
The FE model developed for the FPS in OpenSees is verified with respect to data
obtained from experimental studies performed on a rigid isolated frame by Mosqueda et
al. (2004) and the influence of modeling assumptions are monitored. A 3D model of the
experimental setup is developed in OpenSees. Two loading schemes comprised of
unidirectional and bidirectional paths are applied to the model to illustrate the
characteristics of the FPS forcedeformation response.
4.9.1. Structural Properties and Loads
The structural model considered herein is the test setup studied at the Earthquake
Engineering Research Center (EERC) headquartered at the University of California,
Berkeley by Mosqueda et al. (2004). The test setup consisted of four FPS isolators
installed under a rigid concrete block (Figure 4.12). The objective of the test was to
examine the bidirectional response of the FPS isolators. The frame was loaded with a
1.78 m (x direction) by 2.85 m (y direction) rectangular rigid mass totaling approximately
290 kN, to produce a target vertical load on each isolator of 72.5 kN. The isolators had a
concave surface with a 76.2 cm radius and the µ at low and high velocities, f
min
and f
max
,
65
were given as 0.05 and 0.11, respectively. The maximum displacement capacity of the
isolators was ± 17.8 cm. However, the maximum displacement capacity of the simulator
was ± 12.7 cm. The yield displacement of the isolators,
y
∆ , was approximately 0.025
cm. The constant describing the rate of transition from low to high velocities, a, was
found to be 1.5 by Mosqueda et al. (2004). However, the constant, ε , that controls the
variation of f
max
between very low and very high pressures is not provided. Constantinuo
et al. (1993) gave this relationship as f
max ) tanh( 07 0 12 0 εp . . − =
. This relationship is
scaled by 1.2 to match the value of f
max
=0.11 in the Mosqueda et al.(2004) study at a
normal load level of 72.5 kN (Figure 4.13).
Figure 4.12 Test setup studied by Mosqueda et al. (2004).
66
N (kN)
0 100 200 300 400
f
m
a
x
0.00
0.05
0.10
0.15
0.20
Constantinou et al. (1993)
Scaled
N
=
7
2
.
3
k
N
f
max
=0.09
f
max
=0.11
Figure 4.13 Scaling of f
max
with respect to the relationship presented in Constantinou et
al. (1993).
The frame is analyzed under displacement controlled unidirectional (L1) and
bidirectional (L2) loadings. The unidirectional loading (L1) is categorized as L1a, L1b
and L1c sinusoidal motions with peak displacement values of 12.7, 17.8 and 45.0 cm,
respectively (Figure 4.14). The loading L1a has a maximum displacement that is limited
by the simulator capacity of the test setup and is used to verify only forcedeformation
response per tests data. The second loading, L1b, has the maximum displacement
capacity of the isolators and it is used to observe bounds of the µ on the force
deformation response. The third load path, L1c, exceeds the maximum capacity of the
isolators and it is purely theoretical. This path is used to monitor the range which large
deformation effects become significant in the response of the isolators.
67
t
(s)
0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0
δ
x
(
c
m
)
60
40
20
0
20
40
60
L1a
L1b
L1c
12.7 cm
17.8 cm
45.0 cm
Figure 4.14 Unidirectional load histories with amplitude, ± 12.7 cm (L1a), ± 17.8 cm
(L1b), and ± 45.0 cm (L1c).
The bidirectional loading, L2, is used to verify the planar response of the isolators
and monitor the influence of the effects of bidirectional sliding interaction in the
response. The history of the displacements in each orthogonal direction is defined by
Bsin(wt), where B=5 is the amplitude of the motion, t is the time, and w is the frequency
equal to 1 in the x direction and 2 in the y direction (Figure 4.15).
68
δ
x
(cm)
20 15 10 5 0 5 10 15 20
δ
y
(
c
m
)
30
20
10
0
10
20
30
1.78 m
2.85 m
x
y
z
Figure 4.15 Combined bidirectional loading path for L2.
4.9.2 Modeling and Analysis
A three dimensional (3D) model of the setup is developed in OpenSees via rigid
beam elements and the zerolength element developed for the FPS (Figure 4.16). The
base of the model is fixed at the four corners in all directions. The mass of the rigid
block is lumped at the four corners and these corners are connected via rigid beam
elements. After applying only the gravity portion of the corresponding load case, the
gravity loads are held constant and the lateral load is applied to the model. This allows
for the development of the frictional forces in zerolength elements. All analyses are
geometrically nonlinear and a NewtonRaphson solution algorithm is used for the
solutions.
69
Figure 4.16 Finite element model of the test setup studied by Mosqueda et al. (2004).
4.9.3 Results
Mosqueda et al. (2004) reported that as a result of the rigidity of the supported
block, minor changes in the vertical alignment of the setup caused significant
redistribution of the normal forces acting on the isolators, and in some cases complete
unloading. The authors have also underlined the difficulty of adequately modeling the
friction response of the FPS. The normalized forcedeformation history of Isolator 3 and
the overall superstructure computed from the model and the experimental results for L1 is
given in Figure 4.17. Given the complexity of the response, results predicted by the
model are in agreement with that reported from the experiment.
70
δ
x
/ R
0.20 0.15 0.10 0.05 0.00 0.05 0.10 0.15 0.20
f
x
/
N
0.4
0.3
0.2
0.1
0.0
0.1
0.2
0.3
0.4
Model
Experiment
FPS
fx
(a)
δ
x
/ R
0.20 0.15 0.10 0.05 0.00 0.05 0.10 0.15 0.20
Σ
f
x
/
Σ
N
0.4
0.3
0.2
0.1
0.0
0.1
0.2
0.3
0.4
Model
Experiment
FPS
fx
(b)
Figure 4.17 Comparison of the normalized forcedeformation histories between the
model and the experimental results for (a) individual Isolator 3, and (b) total isolator
forces.
Figure 4.18 shows the variation of µ for loading path L1b. Despite the
assumption of constant N on the FPS, the value µ ranged between 0.05 to 0.11 due to
71
changes in the sliding velocity. This is equivalent to a 110% variation in the frictional
force component,
µ
f , of the FPS during response. The upper and lower bounds of the
isolator response as a function of f
min
=0.05 and f
max
=0.11 is given in Figure 4.19. The
maximum isolator force (MIF) calculated by the model by accounting for velocity and
pressure effects in the friction coefficient was 24.02 kN. This value was overestimated
by 9.6% and underestimated by 10.5% when the µ was assumed to be f
min
and f
max
,
respectively. The shape of the forcedeformation histories given in Figure 4.19 also
reveal that the energy dissipated per cycle under different assumptions for the value of
the µ is proportionally variant as in the case of MIF.
t
(s)
0 5 10 15 20
µ
0.025
0.050
0.075
0.100
0.125
0.150
µ = f
max
µ = f
min
Figure 4.18 Friction coefficient, µ , time history under loading L1b.
72
δ
x
/ R
0.20 0.15 0.10 0.05 0.00 0.05 0.10 0.15 0.20
f
x
/
N
0.4
0.3
0.2
0.1
0.0
0.1
0.2
0.3
0.4
µ = exact
µ=f
max
µ=f
min
FPS
fx
Figure 4.19 Comparison of the forcedeformation histories of the FPS with theoretically
exact value of the friction coefficient, µ , µ =f
max
,and µ =f
min
for loading L1b.
Isolator response under L1c with the small deformations model (SDM) and the
large deformations model (LDM) are given in Figure 4.20. It is observed that the
difference between the SDM and the LDM is negligible under even the maximum
isolator displacement capacities which correspond to δ
x
/R=0.18. Inclusion of the large
deformation effect reduced the MIF at theoretical displacements that exceed the isolator
deformation capacity. This softening is attributed mainly to the inclusion of the geometric
angle α in the calculation of the N at large displacements. The difference in the MIF
between the LDM and the SDM for this particular case was less than 4% at the maximum
displacements of L1c loading.
73
δ
x
/ R
0.60.50.40.30.20.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6
f
x
/
N
0.8
0.6
0.4
0.2
0.0
0.2
0.4
0.6
0.8
LDM model
SDM model
δ
x
/
R
=

0
.
1
8
δ
x
/
R
=
0
.
1
8
Softening
Figure 4.20 Comparison of the response of the small deformations model (SDM) and the
large deformations model (LDM).
The forcedeformation histories and the combined resisting force paths in the two
orthogonal directions of the combined isolators under L2 are given in Figures 4.21. This
figure shows that the force paths in the orthogonal directions predicted by the coupled
model are in agreement with test results. A comparison of this response with an
uncoupled model is presented in Figure 4.22. It is observed that, the uncoupled model
deviates significantly from the path of the coupled model as the resisting forces change
directions. The variation of the components of the instantaneous velocity in the
orthogonal directions had significant influence on the force history in the x direction.
This is attributed mainly to the variation of the coupled parameters of the η varying
continuously as a function of time. The MIF of 24.5 kN with the coupled modeled was
overestimated as 29.4 kN (20% difference) with the uncoupled model.
74
Σ f
x
/ Σ N
0.5 0.4 0.3 0.2 0.1 0.0 0.1 0.2 0.3 0.4 0.5
Σ
f
y
/
Σ
N
0.5
0.4
0.3
0.2
0.1
0.0
0.1
0.2
0.3
0.4
0.5
Model
Experiment
Figure 4.21 Comparison of the bidirectional resisting forces for the experimental data,
and coupled model.
Σ f
x
/ Σ N
0.5 0.4 0.3 0.2 0.1 0.0 0.1 0.2 0.3 0.4 0.5
Σ
f
y
/
Σ
N
0.5
0.4
0.3
0.2
0.1
0.0
0.1
0.2
0.3
0.4
0.5
Coupled model
Uncoupled model
Figure 4.22 Comparison of the bidirectional resisting forces for the coupled model and
the uncoupled models.
75
4.10 Conclusion
In this chapter, the simplified bilinear modeling of the Friction Pendulum System
(FPS) has been explained. The simplifying assumptions in this approach and how aspects
pertaining to these simplifications can be represented in the finite element (FE) model of
the FPS has been highlighted. A new 3D zerolength FE of the FPS has been developed
in OpenSees and verified using experimental data. The influence of neglecting certain
modeling aspects of the FPS response has been presented. The following conclusions can
be drawn from this chapter:
(1) There exists sufficient theoretical and experimental findings on different modeling
aspects of the FPS; however a comprehensive FE model that combines these effects
is absent. An FE model to fill this gap has been developed and compiled from
existing research and implemented as a new element into OpenSees.
(2) The FE model developed in OpenSees for the response of the FPS provides good
agreement with experimental findings of Mosqueda et al. (2004) under both
unidirectional and bidirectional loadings.
(3) It was observed that the different assumptions in modeling the FPS caused
significant variations in the paths of the forcedeformation histories of the isolators.
These affects need to be quantified for bridge applications. Different assumptions
pertaining to the modeling of structures seismically isolated with the FPS may lead
to over or under design of isolators.
76
CHAPTER 5
BRIDGE RESPONSE AS A FUNCTION OF ISOLATOR
MODELING ASSUMPTIONS
5.1 Introduction
This chapter investigates the response of typical highway bridges isolated with the
Friction Pendulum System (FPS) as a function of isolator modeling assumptions. The
selection and detailed modeling of the bridges considered for seismic isolation with the
FPS are presented. Seven models of a threedimensional (3D) MultiSpan Continuous
(MSC) Steel Girder bridge with different assumptions of the FPS are generated.
Nonlinear time history analyses (NLTH) are performed for the bridge to examine the
effect of modeling parameters of the FPS on the response. The influence of the variations
in isolator normal force, N, and coefficient of friction, µ , inplane bidirectional sliding
interaction, large deformation, P ∆ effects, and the orientation of the FPS isolators are
highlighted. Maximum normalized force (MNF) and deformation (MND) of the isolators
and column drifts are used as the parameters to characterize the response of the models.
5.2 Selection of the Class of Highway Bridges for Seismic Isolation and Analyses
A detailed survey of 163,433 bridges in the Central and Southeastern US (CSUS)
was performed by Nielson (2005). The results of this study showed that MultiSpan
Continuous (MSC) Steel Girder and MultiSpan Simply Supported (MSSS) Steel Girder
bridges and MSC and MSSS Concrete Girder bridges are among the most common
classes of bridges found in the CSUS inventory (Figure 5.1). Nielson (2005) further
performed the fragility assessment of the classes of bridges and concluded that the MSC
and MSSS Steel Girder bridges were among the most vulnerable to damage, followed by
77
the MSC and MSSS Concrete Girder bridges. Previous research identified significant
vulnerabilities of the steel fixed and rocker bearings employed in these bridges to seismic
loads (Mander et al. 1996). Seismic isolation of these bridges via replacing the existing
steel bearings with the Friction Pendulum System (FPS) may be an effective tool for
improving the earthquake performance. Assuming that the superstructure of SIBs
remains within the elastic range, the modeling of the steel and concrete bridges are
similar. Particular emphasis is given to these highway bridges in subsequent sections of
this dissertation.
Figure 5.1 Picture of example MSC Steel Girder Highway Bridge (Nielson 2005).
5.3 Seismically Isolated Bridge (SIB) Modeling
The bridge type selected for the NLTH analysis in this chapter is an MSC Steel
Girder Bridge seismically isolated with FPS isolators (Nielson 2005). The 3D SIB
model was developed in OpenSees. This model includes material and geometric
nonlinearities. The geometry and modeling approach for the bridge is illustrated in
Figures 5.2 and 5.3. The bridge has three spans and a continuous slabongirder deck
with a total of eight steel girders. The seismic isolation of the bridge is achieved via
placing FPS isolators under each of the eight girders above the piers and abutments. The
78
width of the expansion joints at the abutments is 7.7 cm. The FPS isolators are selected
to achieve approximately a 2.02.5 second fundamental period which corresponds to R =
99 cm with an inplane displacement capacity of 23 cm. The isolators are assumed to be
positioned as the concave dish at the top. The slider diameter has 7.7 cm to ensure
pressures below 275 MPa under gravity and earthquake loads in accordance with the
recommendations of the manufacturer. The characteristic properties of the µ are
selected as: a = 59.1 s/m,
s
∆ = 0.025 cm, ε = 0.012 MPa
1
, f
max
= 0.12 and f
min
= 0.05
(Mosqueda et al. 2004; Constantinou et al. 1993).
The superstructure is expected to remain within the linear elastic range, thus, the
deck elements are modeled using elastic beam column elements, using the composite
section properties. The section properties for the columns and the bent beams are created
using fiber elements with appropriate constitutive models for both the concrete and the
steel reinforcement. The reinforcing steel is modeled as a bilinear material with a yield
strength, f
ys
= 414 MPa, and an elastic modulus, E
s
= 200 GPa. A strain hardening ratio
of 0.018 is used for this material (Figure 5.4). The unconfined and confined concrete
behavior is modeled via the KentScottPark model which utilizes a degraded linear
uploading/reloading stiffness and a residual stress. The concrete compressive strength, f
c
,
and associated strain,
c
ε , are 27.6 MPa and 2.10
3
for the unconfined case and 28.5 MPa
and (2.062)10
3
for the confined case, respectively (Figure 5.4). The bridge has footings
which are 2.44 m square and use eight piles. The horizontal, k
t
, and rotational, k
r
,
stiffnesses of the foundation are 130.5 kN/mm and (6.06)10
5
kNm/rad, respectively.
Structural damping is assumed to be 5%.
79
Figure 5.2 Multispan continuous (MSC) steel girder bridge general elevation and
modeling details.
Figure 5.3 Pier configuration and bent and column discretization.
80
(a)
(b)
Figure 5.4 Constitutive relationships for the modeling of (a) steel material; and (b)
concrete material.
5.4 FPS Models
Seven SIB models are generated with the above properties where the only
difference is in the FPS modeling assumptions. The first model is theoretically exact,
i.e., accounts for the variations of the N and µ , has bidirectional coupling of the sliding
forces and incorporates P ∆ effects. The second model is a simplified bilinear model
that is insensitive to the variations in N and µ , with uncoupled bidirectional sliding
forces and small deformation assumptions. In Model 2, the constant value of N is taken
as the corresponding value after gravity load analysis and µ as 0.07. The third model is
developed to monitor the influence of not accounting for the variations of N on the
response of the FPS. It is the same model as Model 1 with the only difference of
assuming a constant N of the corresponding value after gravity load analysis. The fourth
model is developed to identify the influence of the bidirectional coupling in estimating
81
the response of the FPS. It is the same model as Model 1 with the only difference of
assuming the orthogonal sliding forces of the FPS isolators to be uncoupled. This is
achieved by assuming η = [ ) sgn( ) sgn(
y x
δ δ
& &
]
T
.
The fifth model is developed to monitor
the influence of not accounting for the inclination due to the concavity in the FPS. This
is achieved by computing f with Equation 4.18 . The sixth model is generated to identify
the influence of the FPS orientation. Model 6 is same as Model 1 with the only
difference being that the FPS isolators are positioned with the concave dish at the bottom
which is accommodated as the corresponding sign shift in the L
ˆ
in Equation 4.17. The
seventh model, is developed to monitor the influence of the assumptions on the value
of µ . Model 7 is established with the same principles as Model 1 with the only
difference of having a µ that is constant, i.e. insensitive to variations in pressure and
sliding velocity. Model 7 is discussed separately from the other models and analyzed for
a constant value of µ ranging from 0.05 to 0.12 with increments of 0.01. The properties
of the models are summarized in Table 5.1.
Table 5.1 Summary of model properties
Model
1
Modeling aspect 1 2 3 4 5 6
2
7
3
Normal force x x x x x
Bidirectional coupling x x x x x
Large deformations x x x x x
Fiction coefficient x x x x x
1
‘x’ denotes exact modeling
2
Concave dish of the FPS at the bottom
3
Seven models with value of µ ranging from 0.05 to 0.12 with increments of 0.01
82
5.5 Dynamic Analyses
The modal properties of the SIB in Model 1 are established by assigning linear
effective stiffness to the FPS isolators. The first three modes of vibration are those
involving the isolation system which shows that the characteristics and the design of the
FPS isolators govern the dynamic response of the bridge (Figure 5.5). The first three
modal periods of the SIB are 2.22 s, 2.15 s, and 1.93 s, respectively. The first mode is
longitudinal, the second mode is transverse and the third mode is torsional.
Figure 5.5 Mode shapes of the deck.
Seismically isolated bridge (SIB) models were subjected to NLTH analyses.
OpenSees allows the user to select the integration technique and solution algorithm for
the analysis. Newmark’s average acceleration timestepping scheme, which is an
unconditionally stable numerical integration algorithm, was used in integrating the
nonlinear dynamic equilibrium equations. The equations of motion were solved
numerically using the NewtonRaphson method. The time interval for solving the
equations of motion was taken as 0.005 s.
An important recommendation by the bridge engineering community is the use of
design earthquakes that have a 2% probability of exceedance in 50 years (an earthquake
with a mean recurrence interval of 2475 years) (FEMA 1997). A suite of ten earthquake
83
records from rock sites is used in the NLTH analysis of the bridges (Table 5.2). The
geometric mean of the longitudinal and transverse component of each record is scaled to
match the spectral value of 0.118g at a period of 2.22 s corresponding to a 2% probability
of exceedance in 50 years hazard level earthquake in Memphis, TN. The response
spectra of the scaled ground motion records for 5% damping, ξ, and their median are
given in Figure 5.6.
Table 5.2 Ground motion suite
Component PGA (g)
No. Earthquake record Longitudinal Transverse Vertical Scale
1 Morgan Hill 1984/04/24 0.098 0.069 0.092 7.932
2 Northridge 1994/01/17 1.285 1.585 1.229 0.767
3 Whittier Narrows 1987/10/01 0.304 0.199 0.227 3.279
4 Loma Prieta 1989/10/18 0.473 0.411 0.209 1.263
5 Gazli, USSR 1976/05/17 0.718 0.608 1.264 0.421
6 N. Palm Springs 1986/07/08 0.492 0.612 0.471 1.434
7 Helena, Montana 1935/10/31 0.173 0.15 0.102 3.654
8 Loma Prieta 1989/10/18 0.453 0.501 0.507 1.248
9 Nahanni, Canada 1985/12/23 0.978 1.096 2.086 0.787
10 Landers 1992/06/28 0.721 0.785 0.818 0.552
Period, T (s)
0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50 2.75 3.00
S
p
e
c
t
r
a
l
a
c
c
e
l
e
r
a
t
i
o
n
,
S
a
(
g
)
f
o
r
ξ
=
5
°
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
2% in 50 years
Median
T=2.22 s, S
a
=0.118g
Figure 5.6 Response spectrums for the suite of ground motions.
84
The three components of the acceleration histories of each scaled ground motion
are applied to the SIB models (Figure 5.7). The inplane orthogonal components of the
earthquakes are oriented to result in the maximum demands on the columns for all cases.
The SIB models were first analyzed for gravity loads and sequentially subjected to NLTH
analyses using simultaneously the longitudinal, transverse and vertical acceleration
records of the given earthquake. It is found from the gravity load analysis that each
isolator above the pier and the abutments carry a gravity load , N
o
, of approximately 125
kN and 258 kN respectively (neglecting the normal load variation between the isolators at
the exterior and the interior ends at the same pier and abutment).
Figure 5.7 Orientation of the 3D bridge model.
The structural response of the isolators and columns along the same transverse
axis were essentially the same. Therefore, the results are presented for one of the
isolators on top of the piers and the abutments and one of the columns. The main
response quantities monitored for the FPS isolators are the maximum normalized force,
MNF = max(
o T L
N / f f
2 2
+ ), where f
L
is the longitudinal and f
T
is the transverse
isolator force, respectively, and the maximum normalized displacement, MND =
Vertical
Transverse
Longitudinal
ü
T
(t)
ü
L
(t)
ü
V
(t)
85
max( /R
T L
2 2
δ δ + ), where
L
δ is the longitudinal and
T
δ is the transverse isolator
displacement, respectively, for the FPS above the pier. Maximum column drifts, d
max
, are
selected as the response quantity to monitor the structural demands on the SIB.
5.6 Results
It was observed from the NLTH analyses of Model 1 that the maximum allowable
displacements at the expansion joints were exceeded in an all records except Morgan
Hill, Gazli and Nahanni. This indicates that pounding would occur between the abutment
and the deck in the longitudinal direction. The impact forces in the deck are difficult to
correlate to damage levels and may impede the utilization of the full capacity of the
isolators. Additionally, uplift took place between the sliding surfaces of the FPS isolators
in the vertical direction for all of the records except for the Loma Prieta, Helena and
Landers. The timehistory of the N/N
o
of the Model 1 FPS isolator for the Nahanni
earthquake is given in Figure 5.8. The maximum allowable N is limited by the allowable
pressure of 310 MPa on the slider, which corresponds to N/N
o
=5.4. This ratio was not
exceeded during any of the NLTH analyses, however, during the Nahanni earthquake a
peak value of N/N
o
=3.51 was reached. This substantial increase is indicative of a
proportional increase in the postyield stiffness and yield force of the isolator. It is
observed from Figure 5.8 that the contact between the two sliding surfaces was lost at
least once which resulted in N/N
o
=0. This uplift caused instantaneous yet complete loss
of stiffness of the isolators during the earthquakes. However, due to the indeterminacy of
the model there was no instability.
86
Time (s)
0 2 4 6 8 10 12 14 16 18 20 22 24
N
/
N
o
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
max(N/N
o
)=3.51
FPS
N
min(N/N
o
)=0.0
Figure 5.8 Time history of the N/N
o
for the FPS during the Nahanni earthquake NLTH
analysis.
Figure 5.9 shows the normalized forcedeformation (NFND) histories of the FPS
isolators on top of the piers among Models 1 to 4 in the longitudinal direction of the
bridge during the N. Palm Springs record. Model 1 can capture the abrupt changes in
isolator force and instances of uplift in the vertical direction. These two aspects of the
isolator response could not be observed in Model 2. Additionally, Model 2
underestimated both the MNF and the MND in comparison to Model 1. These
differences between Model 1 and Model 2 NFND histories can be explained by the
response observed in Models 3 and 4. Model 3 was unable to capture peak isolator forces
indicating that the normal components of the ground motion were influential in this
response quantity. Although Model 4 was able to account for the significant variations
in isolator forces, the peak isolator force was overestimated and the peak isolator
deformation was underestimated. This implies a stiffer isolator response when the
bidirectional effects are neglected.
87
δ
L
/ R
0.20 0.15 0.10 0.05 0.00 0.05 0.10 0.15 0.20
f
L
/
N
o
0.6
0.4
0.2
0.0
0.2
0.4
0.6
max( δ
L
/ R 
) = 0.118
max(  f
L
/ N
o

) = 0.365
FPS
fL
δ
L
/ R
0.20 0.15 0.10 0.05 0.00 0.05 0.10 0.15 0.20
f
L
/
N
o
0.6
0.4
0.2
0.0
0.2
0.4
0.6
max( δ
L
/ R 
) = 0.096
max(  f
L
/ N
o

) = 0.195
FPS
fL
(a)
(b)
δ
L
/ R
0.20 0.15 0.10 0.05 0.00 0.05 0.10 0.15 0.20
f
L
/
N
o
0.6
0.4
0.2
0.0
0.2
0.4
0.6
max( δ
L
/ R 
) = 0.102
max(  f
L
/ N
o

)= 0.180
FPS
fL
δ
L
/ R
0.20 0.15 0.10 0.05 0.00 0.05 0.10 0.15 0.20
f
L
/
N
o
0.6
0.4
0.2
0.0
0.2
0.4
0.6
max( δ
L
/ R 
) = 0.112
max(  f
L
/ N
o

) = 0.410
FPS
fL
(c) (d)
Figure 5.9 Forcedeformation history of the FPS in the longitudinal directions on top of the
pier for the N. Palm Springs earthquake record with (a) Model 1 (b) Model 2 (c) Model 3 and
Model 4.
The influence of the isolator modeling parameters on MNF, MND and d
max
for the
suite of ground motions is illustrated via box plots given in Figure 5.10. Box plots are a
useful way of presenting the graphical description of variability of data (Montgomery
2005). This information provides an overview of the expected demands on the isolators
88
and the structural system as well as the scatter in the results. The statistical interpretation
of the results are presented with numerical values of the median and plots of the 10
th
,
25
th
, 10
th
, 75
th
, and 90
th
percentile cumulative probabilities.
It is observed that Model 2 underestimated the median of the MNF by 20% and the
peak MNF as 44% of Model 1. Similar results were observed for Model 3 which
indicated that the normal components of the force are influential in design level isolator
forces. It was observed that not including the influence of the variations in the normal
forces acting on the isolators causes a loss in the variability of the MNF results. Peak
MND values for Models 2 to 7 had negligible difference with Model 1. However, there
was a notable variation in the median values of the MND as a function of the magnitude
of the constant value of µ. This effect is elaborated in the subsequent section. Model 2
overestimated the median of the d
max
by 12% and underestimated the peak d
max
as 69% of
Model 1.
The absence of the variability in the d
max
with Model 2 stems from the inability
to account for normal force variations on the isolators. On the other hand, Model 2
attained a general increase in the median of the d
max
, which is attributed to the
overestimation of the stiffness caused by uncoupled response in the orthogonal directions
of the isolator lateral motion. Another factor that contributed to the increase in the
median of the d
max
in Model 2 was not accounting for the variation of the µ. Models 5
and 6 predicted the MNF, MND and d
max
approximately the same as Model 1 for the
whole suite of ground motion records. The exclusion of the exact kinematics pertaining
to the concavity of the FPS in Model 6 was insignificant since the MND was limited to
0.19 for the suite of ground motions. It is concluded that large deformation effects
associated with the orientation and exact kinematics were not significant in the response
where average MND was smaller than 0.20.
89
M
N
F
0.0
0.2
0.4
0.6
0.8
1.0
0.260
1 2 3 4 5 6
Model
0.208
0.182
0.263
0.260
0.260 0.263
0.362
7
µ=0.05 µ=0.12
(a)
M
N
D
0.000
0.025
0.050
0.075
0.100
0.125
0.150
0.175
0.200
0.095
1 2 3 4 5 6
Model
0.098
0.094
0.099
0.095 0.095
0.131
0.086
7
µ=0.05 µ=0.12
(b)
d
m
a
x
(
%
)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.469
1 2 3 4 5 6
Model
0.526
0.505 0.513
0.469 0.469
0.495
0.555
7
µ=0.05 µ=0.12
(c)
Figure 5.10 The influence of modeling assumptions on (a) MNF; (b) MND; and (c) d
max
.
90
The influence of different magnitudes of µ in Model 7 on MNF, MND and d
max
for the suite of ground motions is illustrated via box plots given in Figure 5.11. It is
observed that both the median and the peak MNF for the suite of ground motions increase
consistently with increasing values of µ. Although the median MND generally decreases
with the decreasing values of µ, the peak MND remains essentially the same. However,
there is a notable increase in the variability of MND with increasing values of µ. The
peak and median of the d
max
are for all values of µ overestimated by Model 7 in
comparison to Model 1. The peak and median of the d
max
attain optimal values at µ =
0.08 and µ = 0.09, respectively.
91
M
N
F
0.0
0.2
0.4
0.6
0.8
1.0
0.275
0.05 0.06 0.07 0.08 0.09 0.10
µ
0.289 0.292
0.298
0.334
0.317
0.337
0.362
0.11 0.12
(a)
M
N
D
0.000
0.025
0.050
0.075
0.100
0.125
0.150
0.175
0.200
0.131
0.123
0.111
0.100
0.096 0.094 0.095
0.086
0.05 0.06 0.07 0.08 0.09 0.10
µ
0.11 0.12
(b)
d
m
a
x
(
%
)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.504 0.495 0.488
0.479
0.490
0.514
0.556 0.555
0.05 0.06 0.07 0.08 0.09 0.10
µ
0.11 0.12
(c)
Figure 5.11 The influence of constant value of µ assumptions on (a) MNF; (b) MND;
and (c) d
max
.
92
Smaller N
o
developed in the FPS isolators at the abutments than at the piers due to
the difference in the corresponding tributary mass of the superstructure. Figure 5.12
shows the comparison of the total MNF transferred to the pier, ΣMNF
pier
, and the total
MNF transferred to the abutment, ΣMNF
abutment
; and MND on top of the pier, MND
pier
,
and abutments, MND
abutment
for the suite of ground motions. It is observed that the
isolators transferred almost twice as large as the force to the piers in comparison to the
abutments on the median. On the other hand MND
pier
were approximately 16% less than
MND
abutment
on the median. This indicates that the deck engaged into torsional vibration.
Abutments may be further engaged into resisting earthquake induced loads in SIB by
designing the FPS isolators at the abutments with higher stiffness properties.
Σ
M
N
F
p
i
e
r
/
Σ
M
N
F
a
b
u
t
m
e
n
t
0.00
0.25
0.50
0.75
1.00
1.25
1.50
1.75
2.00
2.25
2.50
2.75
3.00
1.97
M
N
D
p
i
e
r
/
M
N
D
a
b
u
t
m
e
n
t
0.00
0.25
0.50
0.75
1.00
1.25
1.50
1.75
2.00
2.25
2.50
0.84
(a) (b)
Figure 5.12 Comparison of the (a) total MNF transferred to the pier, ΣMNF
pier
,
and the total MNF transferred to the abutment, ΣMNF
abutment
; (b) MND on top of the pier,
MND
pier
, and abutments, MND
abutments
.
93
5.7 Conclusion
In this chapter, the modeling of a typical highway bridge seismically isolated with
the FPS has been presented. The influence of FPS modeling assumptions on normal
force, N, and friction coefficient, µ , orthogonal coupling and large deformation, P ∆,
effects in a seismically isolated multispan continuous (MSC) steel girder bridge has been
highlighted via nonlinear timehistory (NLTH) analyses. The following conclusions are
made:
(1) The simplified bilinear idealization of the FPS response was unable to capture the
variability in the results. This model underestimated the maximum column drifts
(d
max
) by up to 31%. This was mainly a result of not accounting for the effects of
vertical components of ground motions, bidirectional coupling and the variable
magnitude of the friction coefficient.
(2) The uplift and pounding of the deck in the vertical direction had notable affects in
the response of the FPS that in one case caused an increase of up to 3.51 times in
the initial gravity load acting on the isolators (N
o
).
(3) Excluding the bidirectional coupling of the FPS isolators generally resulted in
overestimating the isolator maximum normalized forces (MNF) and
underestimating the isolator maximum normalized displacements (MND). This
indicates an overestimation of the stiffness of the isolators.
(4) The incorporation of the effects of orientation and the exact concave geometry of
the FPS in to the response had negligible affects. This is mainly a result of the
MND remaining under 0.20 for the suite of ground motions.
(5) The peak MND of the isolators among the suite of ground motions acquired
negligible variations among all the modeling assumptions. However, the median
MND was influenced by the assumptions in the magnitude of µ .
(6) The structural demands transferred by the isolators to the abutments and the piers
were significantly different. Abutment forces were twice of those at the piers in the
94
median and the isolators acquired 18% larger deformations in the median at the
abutments in comparison to those at the piers. This is a results of the uneven
distribution of isolator stiffness properties along the bridge as a function of deck
tributary mass.
95
CHAPTER 6
COMPARATIVE ASSESSMENT OF SLIDING VERSUS
ELASTOMERIC SEISMIC ISOLATION FOR TYPICAL MULTI
SPAN BRIDGES
6.1 Introduction
This chapter compares sliding versus elastomeric seismic isolation of a typical
MultiSpan Continuous Concrete Girder (MSCCG) bridge with advanced isolator
models. The Friction Pendulum System (FPS) and the Lead Rubber Bearing (LRB) are
selected as representative examples of sliding and elastomeric isolators, respectively.
Isolators serve the common objective of lengthening the period of the structure and
providing additional energy dissipation, however there exists considerable differences in
their mechanisms. In spite of existing research findings on the dependency of LRB in
plane response to the magnitude of inplane deformation and normal load, existing
nonlinear models do not account for these effects. A detailed isolator model for the LRB
that can account for the inplane and vertical coupling of the response is developed in
OpenSees. Particular emphasis is given to the distinct vertical load dependency
modeling of the isolators. A seismic evaluation of the bridge, isolated in one case with
the LRB and in another case with the FPS, is performed for a hazard level of 7% in 75
years using a nonlinear threedimensional (3D) analytical model. Maximum isolator
forces and displacements, and column drifts are selected as response quantities.
6.2 The Lead Rubber Bearing (LRB)
The Lead Rubber Bearing (LRB) was invented in April 1975 by W H Robinson
(Skinner et al. 1993). The LRB is one of the most commonly used elastomeric isolator
96
types in bridges (Buckle and Mayes 1990) (Figure 6.1). Analytical and experimental
research and observed performances of bridges during earthquakes isolated with the LRB
showed that the LRB may have a substantial impact on improving the structural
performance of bridges prone to seismic loads (Jangid 2004; Kelly and Buckle 1986; DIS
1996; Lee et al. 2002; Ghobarah 1988). It has been also reported based on information
provided by the manufacturers that the cost, size and energy dissipation of the FPS and
LRB may be comparable in bridge applications (Dicleli, M 2002). Additionally, both
isolators have been incorporated into the design codes (AASHTO 1999; International
2000).
97
(a)
(b)
Figure 6.1 Examples of LRB applications (a) Rio Vista Bridge, Califonia (b) Patria
Acueducto, Mexico (courtesy of Dynamic Isolation Systems).
98
The LRB consists of steel plates, rubber and a lead core (Figure 6.2). A lead core
is inserted in the center of the bearing for energy dissipation and stiffness (Priestley et al.
1996). Lead is a feasible option because it yields in shear at relatively low stresses, 10
MPa, and has good fatigue properties (Skinner et al. 1993). The steel layers placed
between the rubber serves to limit the edgebulging of the rubber (Tyler 1991).
Additionally, the steel plates force the lead plug bearing to deform in shear (Naeim and
Kelly 1996). The forcedeformation response of the LRB is typically idealized as
bilinear (Ghobarah 1987; AASHTO 1999; Naeim and Kelly 1996). The LRB provides
the advantage of attaining versatile forcedeformation characteristics via the geometrical
variations of both the lead core and the rubber (see Figure 6.3) (Priestley et al. 1996).
Figure 6.2 The Lead Rubber Bearing (LRB) interior elevation.
Deformation
Increasing lead
core diameter
S
h
e
a
r
F
o
r
c
e
Figure 6.3 Effects of geometrical variations of the LRB on the forcedeformation
response (Priestley et al. 1996).
99
6.3 ForceDeformation Characteristics of the LRB
As a multilayered elastomeric type bearing, the LRB is susceptible to a buckling
type of instability (Kelly 1997). The critical buckling load established from the elastic
theory in undisplaced position is (Naeim and Kelly 1996):
e s cr
P P P =
o
(6.1)
where
GA P
s
= (6.2)
and
2
r
e
t
EI
P
eff
2
) ( π
= (6.3)
where G is the shear modulus of the rubber, t
r
is the total thickness of the rubber, A is the
crosssectional area of the isolator and (EI)
eff
is the effective rigidity. The critical
buckling load capacity of the LRB reduces as the isolator deforms (Buckle and Liu
1993):
(
¸
(
¸
− =
D
P P
cro cr
δ
1 (6.4)
where D is the diameter of the isolator and δ is the lateral bearing displacement.
The LRB response softens and yield force increases with increasing level of N
(Ryan and Chopra 2005). The postyield stiffness of the LRB as a function of the N is
(Buckle and Kelly 1986):
(
(
¸
(
¸


¹

\

− =
2
1
cr
po LRB p,
P
N
k k (6.5)
where stiffness k
po
is the nominal (meaning absent of N effects) postyield stiffness. If
the large deformation (P∆) effects described in Chapter 4 are neglected, the post yield
stiffness of the FPS is:
100
R
N
k
FPS p,
=
The yield force of the LRB was observed to not achieve the theoretical strength under
low N (Hwang and Hsu 2000). The yield force of the LRB is approximated as (Ryan
and Chopra 2005):
( )
 
o
N/P
Y LRB Y,
F F
−
− = e 1
o
(6.6)
where
o Y
F is the nominal yield strength of the isolator which can be computed from the
yield stress of the lead core, and P
o
is the normal load corresponding to approximately
63% of nominal strength. The initial elastic stiffness of the LRB, k
i,LRB
, is typically
estimated as 10k
p,LRB
(Naeim and Kelly 1996). The yield displacement of elastomeric
isolators is typically larger compared to sliding isolators (Matsagar and Jangid 2004).
The yield strength of the FPS is (Earthquake 2003):
µ N F
FPS Y,
= (6.7)
The LRB compression stiffness is (Naeim and Kelly 1996):
h
A E
k
c
v
= (6.8)
where E
c
is the compression modulus and h is the total isolator height. The stiffness of
the LRB remains elastic under tension with the same magnitude as in compression until
reaching cavitation at strains ) 2 ( 1
2
S /
c
= ε , where S is the shape factor of the isolator
(Kelly 2003; Mori et al. 1996). This behavior closely matches the response of a zero
length symmetrically elastic element bound by P
cr
under compression and
c
ε under
tension:
¹
´
¦
< < −
≤ <
=
cavitation 0 for
0 buckling for
e e e
e e e
e
δ k
δ k
f
δ
δ
(6.9)
where k
e
is the vertical stiffness of the elastic element equal to that of the isolator, δ
e
is
the elastic deformation with a () and (+) sign for tension and compression, respectively.
101
This is a similar approach adopted for modeling of the FPS vertical response in Chapter
4.
6.4 Modeling of the Isolator Response
The bilinear idealization of the LRB forcedeformation response is based on
similar simplifications adopted for the bilinear idealization of the FPS response: (1) N is
constant; (2) F
Y,LRB
is constant; (3) the horizontal response is uncoupled in the orthogonal
directions; and (4) isolator deformations are small. It is possible to extend the model
developed in Chapter 4 for the FPS to represent the nonlinear and coupled response of the
LRB using the relationships described in the previous section.
The hysteretic forcedeformation response of the LRB is modeled by
implementing a 3D zerolength element class and a complimentary material class in
OpenSees. The procedure for this approach is essentially the same as the FPS modeling
described in Chapter 4. The only difference is in evaluating the postyield stiffness, yield
force, and vertical response from Equations (6.5), (6.5) and (6.9), respectively. The large
deformation affects are assumed to be negligible (as described in Chapter 5). The FPS
model developed in Chapter 4 is used in the subsequent analyses with the only additional
assumption of neglecting largedeformation effects. It has been shown that excluding the
inplane coupling of the orthogonal response for the LRB may result in significant
underestimation of the displacements and forces (Jangid 2004). Consequently, this
interaction is included in the response of the LRB using the methodology described in
Chapter 4. In retrospect, the general forcedeformation relationship in the x, y, and z
directions for the FPS and the LRB are:
f = 
T
z y x
f f f = 
T
N F k F k
y Y y p x Y x p
η δ η δ + + (6.10)
102
where f and δ are the isolator forces and displacements in the direction denoted with the
subscript, k
p
is the postyield stiffness, F
y
is the yield force described by corresponding
Equations in the previous section for the two isolators (Figure 6.4).
Figure 6.4 Isolator model.
6.5 Bridge Model
The MultiSpan Continuous Concrete Girder (MSCCG) bridge considered in this
Chapter is essentially the same with the one used in Chapter 5 with the only difference in
the geometric and material properties of the superstructure. The superstructure is
comprised of a continuous slabongirder concrete deck with a total of eight concrete
girders. The superstructure is expected to remain within the linear elastic range and
modeled as a beam element. The total weight of the superstructure is W=12272 kN with
a moment of inertia about the strong and weak axes as I
y
=75.03 m
4
and I
x
=0.12 m
4
and
modulus of elasticity of E
c
=25.6 GPa. Additionally, the pounding between the deck and
103
the abutments are neglected to focus on the differences in the response of the two
isolators. Details of the substructure modeling of the bridge are given in Chapter 5.
6.6 Bridge Seismic Isolation
The seismic isolation of the bridge is achieved via placing isolators under each of
the eight girders above the piers and abutments. It is assumed that a single isolator size is
used throughout the bridge to save on the cost of an extra mold. According to AASHTO
(1999) the isolation period is to be determined from the isolation system effective
stiffness based on maximum design displacement. Due to the absence of maximum
design displacement the period is determined with the k
p
of the isolation system. The
effective stiffness and k
p
are correlative and typically close (Jangid 2004). Additionally,
the k
p
of the isolators are more influential for the seismic response of bridges compared to
the initial stiffness (Dicleli and Buddaram 2005). To achieve a target isolation period of
approximately T=2 seconds for a onedegreeoffreedom mass of W=12272 kN the
stiffness is 123.7 kN/cm. Assuming an equivalent distribution of superstructure weight on
32 isolators, the stiffness of a single isolator is 1 4. k
i
= kN/cm with a static weight of
387 =
i
W kN. An approximate FPS design to match this k
i
has R=99 cm and 05 0. = µ .
The slider diameter is D
s
=12.7 cm. The LRB design properties are chosen by
considering: (1) a vertical load capacity of at least three times the initial gravity load, N
o
,
(2) a shape factor S>8 (3) a postyield stiffness of approximately 4.1 kN/cm (4) a yield
force of approximately the same value to the average FPS to acquire a comparative
seismic isolation scheme. The following are the properties selected to achieve these
design objectives: bearing diameter, D=35.6 cm, lead core diameter, D
c
=5.8 cm, single
104
layer thickness for rubber, t=0.953 cm, number of layers, n=17, shear modulus, G=0.76
MPa of rubber. This design has an S=8.7 and a P
cro
=1600 kN. The yield displacements
for the FPS and the LRB are assumed to be at 0.026 cm and 0.585 cm respectively. The
two systems possess a yield force of approximately F
y
=19 kN for N=W
i
. The variation of
the P
cr
as a function of δ is given in Figure 6.5.
δ (cm)
0 8 16 24 32 40
P
c
r
(
k
N
)
0
250
500
750
1000
1250
1500
1750
2000
D=35.6 cm
P
cro
=1600 kN
Figure 6.5 Variation of the buckling load, P
cr
, as a function of isolator inplane
deformation, δ, for the LRB.
Two models of the bridge are generated and one is isolated with the LRB and the
other is isolated with the FPS, using the design parameters described above. The yield
force, F
y
, and postyield stiffness, k
p
, of the two isolators as a function of the N are given
in Figures 6.6 and 6.7 respectively. The tributary mass supported by each isolator varies
once they are installed into the bridge and subjected to gravity loading. Each isolator
above the pier and the abutments carry a gravity load, N
o
, of approximately 258 kN and
105
512 kN respectively (neglecting the load variation between the bearings at the exterior
and the interior ends at the same pier and abutment). The corresponding bilinear force
deformation idealization of the two isolation systems after gravity loading is given in
Figures 6.8 and 6.9 respectively. It is observed that the FPS sustains a considerable
variation of the F
y
and k
p
throughout the longitudinal axis of the bridge due to
corresponding tributary gravity load from the superstructure. The F
y
and k
p
of the FPS
isolators above the piers become twice in magnitude of those above the abutments.
Consequently, the isolators above the piers become stiffer and the ones atop the
abutments become more flexible compared to the initial design with W
i
. Since the rate of
dependency of the LRB response to N is weaker compared to FPS, the variation of the F
y
and k
p
located at different parts of the bridge are smaller. Another notable distinction is
that the LRB isolators become stiffer at the abutments and more flexible at the piers
compared to design with W
i
because the k
p
is inversely proportional to N in Equation 6.5.
Normal force, N (kN)
0 300 600 900 1200 1500 1800
Y
i
e
l
d
f
o
r
c
e
,
F
y
(
k
N
)
0
15
30
45
60
75
LRB
FPS
Force
Deformation
F
y
F
y
(N=387 kN) = 11.2 kN
Figure 6.6 Variation of isolator yield force, F
y
, as a function of applied compressive axial
load, N.
106
Normal force, N (kN)
0 300 600 900 1200 1500 1800
P
o
s
t

y
i
e
l
d
s
t
i
f
f
n
e
s
s
,
k
p
(
k
N
/
c
m
)
0
2
4
6
8
10
12
14
LRB
FPS
Force
Deformation
k
p
k
p
(N=387 kN) = 4.1 kN/cm
Figure 6.7 Variation of isolator postyield stiffness, k
p
, as a function of applied
compressive axial load, N.
Deformation, d (cm)
0 2 4 6 8 10 12
F
o
r
c
e
,
F
(
k
N
)
0
15
30
45
60
75
90
Design (N=387 kN)
Piers (N=512 kN)
Abutments (N=258 kN)
kp
=
5
.1
7
k
N
/
c
m
Fy
=
2
5
.8
k
i
p
k
p
=
4
.0
3
k
N
/c
m
Fy
=
1
9
.6
k
N
k
p
=
2
.6
1
k
N
/cm
F
y
=
1
2
.9
k
ip
Isolators at abutments
Isolators at piers
∆
y
=0.26 cm
Figure 6.8 Bilinear idealizations of the FPS forcedeformation characteristics after
gravity loading in the bridge.
107
Deformation, d (cm)
0 2 4 6 8 10 12
F
o
r
c
e
,
F
(
k
N
)
0
15
30
45
60
75
90
Design (N=387 kN)
Piers (N=512 kN)
Abutments (N=258 kN)
k
p
= 3.68 kN/cm F
y
= 20.24 kN
k
p
= 4.03 kN/cm
F
y
= 19.58 kN
k
p
= 4.30 kN/cm
F
y
= 17.80 kN
Isolators at abutments
Isolators at piers
∆
y
=054 cm
Figure 6.9 Bilinear idealizations of the LRB forcedeformation characteristics after
gravity loading in the bridge.
6.7 Dynamic Analysis
The structural periods of the two bridges utilizing either FPS or LRB were
established per k
p
of the isolators. The first three structural periods for the FPSisolated
and LRBisolated bridge were T
1
=2.38, T
2
=2.26, T
3
=2.00 and T
1
=2.30, T
2
=2.16, T
3
=1.64,
respectively. The first three mode shapes were those involving the seismic isolation and
were essentially the same for the two bridges (Figure 6.10). The modal characteristics of
the two bridges, isolated with the FPS in one case and with the LRB in the other, are
considered to be sufficiently close for comparative assessment. A notable distinction
among the vibration characteristics of the two bridges arises for the torsional mode, T
3
.
The FPSisolated bridge acquired a higher period due to the more flexible isolators at the
abutments compared to the LRBisolated bridge.
108
Figure 6.10 Mode shapes of the bridge deck from plan.
A suite of ten earthquake records from rock sites is used in the NLTH analysis of
the bridges (Table 6.1). The geometric mean of the longitudinal and transverse
component of each record is scaled to match the spectral value of 0.056g at a period of
2.44 s (arithmetic average of the T
1
of the two bridges) corresponding to a 7% probability
of exceedance in 75 years hazard level earthquake in Memphis, TN. The response
spectra of the scaled ground motion records for 5% damping, ξ, and their median are
given in Figure 6.11. The three components of the acceleration histories of each scaled
ground motion are applied to the models using NLTH analyses. The inplane orthogonal
components of the earthquakes are oriented to result in the maximum demands on the
columns for all cases. The bridge models were also analyzed without the vertical
acceleration history component of the ground motions,
v
) (t u& & , to examine the effects of N
variations on the response of the bridge and the isolators.
109
Table 6.1 Ground motion suite
Component PGA (g)
No. Earthquake record Longitudinal Transverse Vertical Scale
1 Morgan Hill 1984/04/24 0.098 0.069 0.092 5.302
2 Northridge 1994/01/17 1.285 1.585 1.229 0.595
3 Whittier Narrows 1987/10/01 0.304 0.199 0.227 2.617
4 Loma Prieta 1989/10/18 0.473 0.411 0.209 0.845
5 Gazli, USSR 1976/05/17 0.718 0.608 1.264 0.290
6 N. Palm Springs 1986/07/08 0.492 0.612 0.471 1.135
7 Helena, Montana 1935/10/31 0.173 0.15 0.102 2.820
8 Loma Prieta 1989/10/18 0.453 0.501 0.507 0.814
9 Nahanni, Canada 1985/12/23 0.978 1.096 2.086 0.500
10 Landers 1992/06/28 0.721 0.785 0.818 0.320
Period, T (s)
0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50 2.75 3.00
S
p
e
c
t
r
a
l
a
c
c
e
l
e
r
a
t
i
o
n
,
S
a
(
g
)
f
o
r
ξ
=
5
°
0.00
0.25
0.50
0.75
1.00
1.25
1.50
1.75
7% in 75 years
Median
T=2.34 s, S
a
=0.056g
Figure 6.11 Response spectrums for the suite of ground motions.
110
6.8 Results
The structural response of the isolators and columns along the same transverse
axis were essentially the same. Therefore, the results are presented for one of the
isolators on top of the piers and the abutments and one of the columns. The main
response quantities monitored for the isolators are the: (1) maximum isolator force, MIF
= max(
2 2
T L
f f + ), where f
L
is the longitudinal and f
T
is the transverse isolator force,
respectively; (2) the maximum isolator displacement, MID = max(
2 2
T L
δ δ + ), where
L
δ is the longitudinal and
T
δ is the transverse isolator displacement, respectively; and (3)
maximum column drift, d
max
, in a given earthquake timehistory analysis. Maximum and
minimum values of other response quantities are denoted by ‘max(response quantity)’
and ‘min(response quantity)’, respectively.
The LRB on top of the piers were observed to buckle (N>P
cr
) under the
Northridge earthquake record when the N increase due to
v
) (t u& & and P
cr
reduction due to
isolator deformation was considered in the models. When the bridge was analyzed
without the
v
) (t u& & effect on the isolator, the buckling condition (N>P
cr
) of the LRB did
not prevail. It is concluded that neglecting the effects of
v
) (t u& & in the isolator model may
result in overlooking a fundamental failure mode of the LRB. The f
L

L
δ history of the
FPS with and without the
v
) (t u& & effects on the isolator located on top of the pier for the
Northridge earthquake record is given in Figure 6.12. The timehistory of the N/N
o
for
the same earthquake is given in Figure 6.13. It is observed that the N/N
o
makes a notable
drop to 0.14 which is proportional to the decrease of the k
p,FPS
. This effect can be
observed in Figure 6.12a as deviation from the idealized bilinear forcedeformation path
of the isolator. The analysis without the effects of
v
) (t u& & on the isolator models can not
account for the increased flexibility of the isolator and consequently underestimated the
111
max(
L
δ ). The maximum allowable pressure stress of 275 MPa at the slider of the FPS
was not exceeded in any of the analyses.
δ
L
(cm)
21 14 7 0 7 14 21
f
L
(
k
N
)
180
135
90
45
0
45
90
135
180
max( δ
L

) = 11.5
max(  f
L

) = 74.3
FPS
fL
δ
L
(cm)
21 14 7 0 7 14 21
f
L
(
k
N
)
180
135
90
45
0
45
90
135
180
max( δ
L

) = 9.4
max(  f
L

) = 72.1
FPS
fL
(a) (b)
Figure 6.12 Forcedeformation history of the FPS in the longitudinal direction on top of the
pier for the Northridge earthquake record where the vertical component,
v
) (t u& & , effect is (a)
included (b) not included.
Time (s)
0 2 4 6 8 10 12 14 16
N
/
N
o
0.0
0.5
1.0
1.5
2.0
2.5
max(N/N
o
)=2.14
FPS
N
min(N/N
o
)=0.13
Figure 6.13 Time history of the N/N
o
for the FPS during the Northridge earthquake
record.
112
The f
L

L
δ history of the LRB with and without the
v
) (t u& & effect on the isolators
located on top of the pier for the Whittier Narrows earthquake is given in Figure 6.14. It
is observed that excluding the
v
) (t u& & resulted in an overestimation of the max( f
L
) and
underestimation of the max(
L
δ ). This result can be explained via the timehistory of
the P
cr
and the N of the isolator given in Figure 6.15. The reduction factor for the post
yield stiffness, k
pf
= ( )
2
1
cr
N/P − , vary during the earthquake response of the isolator. An
increase of the N by the random pulses of the
v
) (t u& & at large deformations compounds the
reduction of the P
cr
and results in smaller k
pf
. The value of k
pf
reaches a minimum of
0.27, which increases the flexibility and the MID of the isolator.
δ
L
(cm)
21 14 7 0 7 14 21
f
L
(
k
N
)
180
135
90
45
0
45
90
135
180
max( δ
L

) = 15.5
fL
LRB
max(  f
L

) = 32.3
δ
L
(cm)
21 14 7 0 7 14 21
f
L
(
k
N
)
180
135
90
45
0
45
90
135
180
max(  f
L

) = 59.7
max( δ
L

) = 13.5
fL
LRB
(a) (b)
Figure 6.14 Forcedeformation history of the LRB in the longitudinal direction on top of
the pier for the Whittier Narrows earthquake record where the vertical component,
v
) (t u& & ,
effect is (a) included (b) not included.
113
Time (s)
0 5 10 15 20 25
F
o
r
c
e
(
k
N
)
0
450
900
1350
1800
N
P
cr
N
LRB
min[1(N/P
cr
)
2
]=0.27
Figure 6.15 Time history of the LRB buckling load, P
cr
, and normal force, N.
The ΜΙD
of the LRB
may also be overestimated by not including the
v
) (t u& & effect
on the isolator models. Figure 6.16 gives the f
L

L
δ and f
T

T
δ of the LRB with and
without the
v
) (t u& & effect on the isolator located on top of the pier for the Gazli earthquake.
It is observed that although the longitudinal response of the LRB isolator was essentially
the same with and without the
v
) (t u& & effect, there was a considerable difference in the
transverse direction. The max(
T
δ ) with and without the
v
) (t u& & effect on the isolator was
4.4 cm and 8.7 cm, respectively. The reduction in the k
p
from higher N corresponded to a
more flexible response in the transverse direction thus attracting smaller earthquake
induced displacements on the isolators in this direction.
.
114
v
) (t u& & included
δ
L
(in)
8 6 4 2 0 2 4 6 8
f
L
(
k
i
p
s
)
40
30
20
10
0
10
20
30
40
max(  f
L

) = 10.6
max( δ
L

) = 3.8
fL
LRB
δ
T
(cm)
21 14 7 0 7 14 21
f
T
(
k
N
)
180
135
90
45
0
45
90
135
180
max(  f
T

) = 29.0
max( δ
T

) = 4.4
fT
LRB
(a) (b)
0 ) (
v
= t u& &
δ
L
(in)
8 6 4 2 0 2 4 6 8
f
L
(
k
i
p
s
)
40
30
20
10
0
10
20
30
40
max(  f
L

) = 10.8
max( δ
L

) = 3.6
fL
LRB
δ
T
(in)
21 14 7 0 7 14 21
f
T
(
k
N
)
180
135
90
45
0
45
90
135
180
max(  f
T

) = 45.9
max( δ
T

) = 8.7
fT
LRB
(c) (d)
Figure 6.16 Forcedeformation history of the LRB on top of the pier for the Gazli
earthquake in the (a), (c) longitudinal, and (b), (d) transverse directions.
The f
L

L
δ history of the FPS with and without the
v
) (t u& & effect on isolator the
isolator located on top of the pier for the Whittier Narrows earthquake is given in Figure
6.17. It is observed that the exclusion of the
v
) (t u& & resulted in a slight overestimation of
max(
L
δ ) , however, the max( f
L
) was underestimated by approximately 47%. The
115
increase of the N by inertial forces from the
v
) (t u& & resulted in instantaneous but
considerable stiffening of the FPS response. Although at no instance was there separation
between the two surfaces of the FPS or was the maximum allowable pressure exceeded,
the N reached a maximum of 837 kN and a minimum of 290 kN, which corresponds to a
61% increase and a 56 % decrease from N
o
respectively.
δ
L
(cm)
21 14 7 0 7 14 21
f
L
(
k
N
)
180
135
90
45
0
45
90
135
180
max( δ
L

) = 10.7
max(  f
L

) = 129.0
FPS
fL
δ
L
(cm)
21 14 7 0 7 14 21
f
L
(
k
N
)
180
135
90
45
0
45
90
135
180
max( δ
L

) = 11.0
max(  f
L

) = 88.1
FPS
fL
(a) (b)
Figure 6.17 Forcedeformation history of the FPS in the longitudinal direction on top of
the pier for the Whittier Narrows earthquake record where the vertical component,
v
) (t u& & ,
effect is (a) included (b) not included.
It is concluded from the aforementioned results that excluding the normal load
dependency of the isolators in modeling the forcedeformation response may produce
considerable errors in MIF and MID and mislead to similarities between the FPS and the
LRB. The forcedeformation histories of the LRB and the FPS given in Figures 11(a)
and 14(a) show that there are notable differences in the response of the two isolators. In
addition to acquiring a higher initial and postyield stiffness, the FPS hysteresis has short
duration wriggles compared to the LRB. This is attributed to the stronger dependency of
116
the FPS response to
v
) (t u& & . It is observed from Figures 6.14(b) and 6.17(b) that the force
deformation response of the two isolators is smoother and similar to the bilinear
idealization when the
v
) (t u& & effects in the isolator models are excluded.
Figure 6.18 illustrates the differences in the forcedeformation responses of the
FPS and LRB located on the same beam line but on top of the pier and the abutment for
the Helena earthquake. It is observed that the longitudinal deformation of both types of
isolators is larger at the abutments compared to those at the piers. This is attributed
mainly to the torsional vibration of the bridges deck. However, the difference between
the abutment and pier isolator deformations is greater for the FPS compared to the LRB.
This indicates a larger torsional effect in the response of the FPSisolated bridge. It is
observed that the forcedeformation hysteresis of the FPS gets shorter in width and
deeper in height on top of the pier compared to the response on top of the abutment. This
is an attribute of the strong fluctuation of the k
p,FPS
and F
Y,FPS
to N.
117
δ
L
(cm)
21 14 7 0 7 14 21
f
L
(
k
N
)
180
135
90
45
0
45
90
135
180
max( δ
L

) = 7.0
max(  f
L

) = 77.9
FPS
fL
δ
L
(cm)
21 14 7 0 7 14 21
f
L
(
k
N
)
180
135
90
45
0
45
90
135
180
max( δ
L

) = 10.2
fL
LRB
max(  f
L

) = 54.7
(a) (b)
δ
L
(cm)
21 14 7 0 7 14 21
f
L
(
k
N
)
180
135
90
45
0
45
90
135
180
max( δ
L

) = 10.5
max(  f
L

) = 40.7
FPS
fL
δ
L
(cm)
21 14 7 0 7 14 21
f
L
(
k
N
)
180
135
90
45
0
45
90
135
180
max( δ
L

) = 11.9
fL
LRB
max(  f
L

) = 63.4
(c) (d)
Figure 6.18 Forcedeformation history of the FPS and LRB for the Helena earthquake (a),(b)
on top of the pier and (c),(d) on top of the abutment.
Figure 6.19 shows the box plots for the ratio of the total MIF transferred to the
pier, ΣMIF
pier
, and the total MIF transferred to the abutment, ΣMIF
abutment
; and MND on
top of the pier, MND
pier
, and abutments, MND
abutment
for the 9 earthquake NLTH analyses
(the Northridge earthquake excluded from the initial 10 due to LRB buckling condition).
The FPS inflicted more than twice the amount of force to the piers than the abutments on
the median. A notable difference with the LRB is that the difference in the isolator forces
transmitted to the abutments compared to the piers were closer in magnitude compared to
118
the FPS. Contrary to the FPS, the LRB transmitted larger forces to the abutments
compared to the piers. The MID for both isolator types were larger on top of the
abutments compared to the piers. This is attributed to the torsional motion of the bridge
deck. However, this effect was observed to a lesser extent with the LRBisolated bridge
because the stiffness distribution of the isolators along the longitudinal axis of the bridge
was more uniform compared to the FPSisolated bridge.
FPS
LRB
Median
90
th
percentile
10
th
percentile
75
th
percentile
25
th
percentile
Σ
M
N
F
p
i
e
r
/
Σ
M
N
F
a
b
u
t
m
e
n
t
0.00
0.25
0.50
0.75
1.00
1.25
1.50
1.75
2.00
2.25
2.50
2.18
0.87
M
I
D
p
i
e
r
/
M
I
D
a
b
u
t
m
e
n
t
0.00
0.25
0.50
0.75
1.00
1.25
1.50
1.75
2.00
2.25
2.50
0.80
0.93
(a) (b)
Figure 6.19 Comparison of the (a) total MNF transferred to the pier, ΣMNF
pier
, and the
total MNF transferred to the abutment, ΣMNF
abutment
; (b) MID on top of the pier, MID
pier
,
and abutments, MID
abutments
.
Figure 6.20 gives the total energy dissipated by FPS and LRB located on the same
beam line but on top of the pier and the abutment for the 9 earthquake timehistory
analyses. It is observed that the median energy dissipation of the FPS at the abutment is
approximately 55% of that at the abutment. The energy dissipation capacities of the LRB
on top of the pier and the abutment are closer compared to the case in the FPS.
119
Furthermore, unlike the case in the FPS, the energy dissipation in the LRB is higher at the
abutment compared to the pier.
FPS
LRB
Median
90
th
percentile
10
th
percentile
75
th
percentile
25
th
percentile
T
o
t
a
l
e
n
e
r
g
y
d
i
s
s
i
p
a
t
e
d
(
k
N
.
c
m
)
0
600
1200
1800
2400
3000
3600
4200
4800
5400
6000
1785
Above pier
1735
997
2084
Above abutment
(a) (b)
Figure 6.20 Total energy dissipated for by the isolators (a) on top of the pier (b) on top of
the abutment.
The MIF and MID acquired on top of the pier for the 9 earthquake timehistory
analyses are given in Figures 6.21 and 6.22, respectively. The maximum column drifts,
d
max
, are given in Figure 6.23. The statistical interpretation of the results are presented
with numerical values of the median and plots of the 10
th
, 25
th
, 10
th
, 75
th
, and 90
th
percentile cumulative probabilities. A general comparative assessment from the median
values in Figures 6.21 to 6.22 shows that the FPS acquired smaller MID and larger MIF,
and placed larger demands on the columns compared to the LRB. It is observed from
Figure 19 that the median MIF in the FPS are approximately 46% larger than the LRB.
This difference reaches a peak of 139% for the Whittier Narrows earthquake where the
MIF in the FPS and LRB were 116 kN and 50 kN, respectively. The MIF of the FPS for
120
the suite of ground motions is more scattered then the LRB. Exclusion of the
v
) (t u& & in the
timehistory analyses resulted in the underestimation of the median of the MIF in the two
systems by approximately 15%. The median of the MID of the LRB was approximately
32% larger than the FPS. In particular, the difference between the FPS and LRB was
maximum for the Nahanni earthquake with 8.2 cm and 4.6 cm. It is observed that the
absence of
v
) (t u& & in MID prediction resulted in negligible difference in the FPS and
slight overestimation in the LRB on the median. The median of the d
max
in the FPS
isolated bridge is approximately 17% larger than the LRBisolated bridge. However,
similar to the results obtained for the MIF, the scatter for the FPSisolated bridge d
max
is
higher. For example the d
max
=0.78% in the FPSisolated bridge for the Whittier
Narrows earthquake is approximately 82% higher than the LRBisolated bridge. The
median of the d
max
for the suite of ground motions did not change significantly by
excluding the
v
) (t u& & for either isolator.
FPS
LRB
Median
90
th
percentile
10
th
percentile
75
th
percentile
25
th
percentile
M
I
F
(
k
N
)
0
25
50
75
100
125
150
64.1
0 ) (
v
= t u& &
v
) (t u& &
included
45.4
34.8
48.1
(a) (b)
Figure 6.21 Maximum isolator forces (MIF) for the suite of ground motions on top of the
pier with vertical component,
v
) (t u& & , effect is (a) included (b) not included.
121
FPS
LRB
Median
90
th
percentile
10
th
percentile
75
th
percentile
25
th
percentile
M
I
D
(
c
m
)
0
3
6
9
12
15
18
6.1
0 ) (
v
= t u& &
v
) (t u& &
included
3.2
6.4
9.2
(a) (b)
Figure 6.22 Maximum isolator deformations (MID) for the suite of ground motions on
top of the pier with vertical component,
v
) (t u& & , effect is (a) included (b) not included.
FPS
LRB
Median
90
th
percentile
10
th
percentile
75
th
percentile
25
th
percentile
d
m
a
x
(
%
)
0.0
0.2
0.4
0.6
0.8
1.0
0.42
0 ) (
v
= t u& &
v
) (t u& &
included
0.36
0.42
0.33
(a) (b)
Figure 6.23 Maximum column drifts, d
max
, for the suite of ground motions on top of the
pier with vertical component,
v
) (t u& & , effect is (a) included (b) not included.
122
Figure 6.24 gives the deformation histories of the FPS and LRB on top of the pier
for the Morgan Hill earthquake. The displacement trajectories in Figure 6.24(a) and (b)
reveal the fundamentally different responses of the two types of isolators. The LRB
acquired a larger MID compared to the FPS. The total sum of the deformation of the
LRB and the FPS are 29.1 cm and 47.9 cm, respectively. This can be explained further
by Figure 6.24(c)(d). The LRB deformation history has a higher frequency content
compared to the FPS. Consequently, the LRB engages into motion at smaller dynamic
vibrations compared to the FPS. This is attributed to lower elastic stiffness of the LRB
compared to the FPS. It is concluded that the LRB is more sensitive to the frequency
content of the ground motion compared to the FPS.
123
δ
L
4 3 2 1 0 1 2 3 4
δ
T
4
3
2
1
0
1
2
3
4
MID=3.56
δ
L
4 3 2 1 0 1 2 3 4
δ
T
4
3
2
1
0
1
2
3
4
MID=4.32
(a)
(b)
t
(s)
0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0 22.5 25.0 27.5 30.0
δ
L
(
c
m
)
4
3
2
1
0
1
2
3
4
max( δ
L
 ) =2.21
t
(s)
0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0 22.5 25.0 27.5 30.0
δ
L
(
c
m
)
4
3
2
1
0
1
2
3
4
max( δ
L
 ) =3.03
(c)
(d)
t
(s)
0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0 22.5 25.0 27.5 30.0
δ
T
(
c
m
)
4
3
2
1
0
1
2
3
4
max( δ
T
 ) =3.30
t
(s)
0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0 22.5 25.0 27.5 30.0
δ
T
(
c
m
)
4
3
2
1
0
1
2
3
4
max( δ
T
 ) =3.72
(e)
(f)
Figure 6.24 Deformation histories for the Morgan Hill earthquake of (a), (c), (e) FPS and
(b), (d), (f) LRB located on top of the pier.
124
6.9 Conclusion
In this paper, the distinctions of the response characteristics of sample elastomeric
and sliding isolator types, the Lead Rubber Bearing (LRB) and the Friction Pendulum
System (FPS) respectively, have been highlighted. An LRB model to account for both
the coupling in the inplane and vertical directions has been developed. Two models of a
threedimensional (3D) MultiSpan Continuous Concrete Girder (MSCCG) bridge, one
isolated with the FPS and the other with the LRB, was generated. Nonlinear timehistory
analyses using a suite of 10 ground motions that corresponded to a hazard level of 7% in
75 years was performed for the two bridges. Following conclusions are made:
(1) Despite attaining similar seismic isolation periods, the choice between the LRB
or the FPS resulted in considerable differences in the response of the bridge and
the isolator forces and displacements.
(2) Excluding the vertical component of the ground motion in the analyses dampened
the distinctions that exist between the forcedeformation response of the FPS and
the LRB. The median maximum isolator forces and displacement values were
underestimated for both types of isolators in the absence of vertical components
of the ground motions in the analyses. However, the effects of the vertical
components of the ground motion on the isolator response had negligible effect
on maximum column drifts.
(3) The LRB acquired higher displacements, however, placed smaller demands on
the columns compared to the FPS.
(4) The LRB had a more uniform distribution of lateral stiffness throughout the
bridge compared to the FPS. The FPS stiffened on top of the piers while the LRB
was stiffer on top of the abutments. These distinctions arise from the variant
dependency of the two isolators to normal loads.
125
(5) The FPS is capable of accommodating vertical components of the ground motion
records that might result in a buckling failure of the LRB. Excluding the vertical
component of the ground motion in the modeling and analysis phases of the LRB
may result in overlooking a fundamental failure mode.
126
CHAPTER 7
ASSESSMENT OF THE INFLUENCE OF DESIGN PARAMETERS
ON THE RESPONSE OF BRIDGES SEISMICALLY ISOLATED
WITH FRICTION PENDULUM SYSTEM (FPS)
7.1 Introduction
Seismic isolation is applicable to a wide range of bridges that differ in geometric
and material properties. Additionally, the design characteristics of the FPS may be
modified to alter the seismic response of the bridge. A further understanding of the
effects of design parameters will provide insight on both the level of complexity required
in modeling and the seismic response for a wider range of bridge designs. This chapter
investigates the effects of bridge design parameters and an innovative isolation design
strategy on bridge seismic response. In the first section, the “singlefactor“ method and
regression analyses are performed with nonlineartime history (NLTH) analyses to
quantify the significance of bridge design parameters on the system’s response quantities.
In the second section, a new seismic isolation strategy that involves a practical design
modification to the FPS at the abutments is presented. This approach aims to improve the
seismic response of the bridge by creating a more uniform response between the isolators
located at the piers and the abutments. The implications of possible design variations in
this new seismic isolation strategy are illustrated.
7.2 Influence of Bridge Design Parameters
To ascertain the effects of the design parameters of the typical highway bridges
on the system’s seismic response, an experimental design is constructed. The term
“experimental design” can be defined as: “the experimental structure used to generate
127
practical data for interpretative purposes” (Gardiner and Gettinby 1998). The “single
factor design”, which is a form of experimental design to analyze the influence of
different levels of a controllable parameter on a measured response, is selected for this
study (Gardiner and Gettinby 1998).
7.2.1 Analyses
The threedimensional bridge model with FPS isolation developed in Chapter 6 is
used as the base model to which the variation of the design parameters is compared. The
details of this model are not repeated in this chapter. The design parameters of the base
model are denoted by a subscript ‘o’. The NLTH analyses are performed using the suite
of ground motions described in Chapter 6 for four equally spaced values of selected
bridge design parameters. These parameters are detailed in Table 7.1 and illustrated in
Figure 7.1. The upper and lower bounds of the selected design parameters in Table 7.1
are selected from the bridge inventory analysis performed by Nielson (2005) for the
Central and Southeastern United States. Reinforcement ratio and distribution is assumed
to remain the same for the range of column height, L
c
, considered. The span length is
varied in two different ways: (1) mass adjusted to increasing length, L
d
*, (2) mass kept
constant at all lengths, L
d
. Increasing the span length by keeping the mass and cross
section properties constant provides the opportunity to monitor the effects of
superstructure flexibility. Given the wide range of material properties available,
superstructure sections with the same structural characteristics may be constructed for
constant mass. Adjusting the mass for the same crosssection in the superstructure
provides insight on the influence of additional dead weight. The influence of pinned and
fixed modeling assumptions for the base is illustrated to provide the bounds of the
response (Figure 7.2). However, the results from this analysis are handled separately
since this variation is associated with modeling assumptions as opposed to a design
parameter in a bridge.
128
Table 7.1 Parameter variation ranges
Parameter Abbreviation Range
Degree of skew α 040 (degrees)
Column height L
c
3.67.5 (m)
Span length L
d
2060 (m)
Span length* L
d
* 2060 (m)
Concrete nominal strength f
c
20.748.3 (MPa)
Base conditions  Pinned, elastic, fixed
*Mass adjusted in accordance with increased length
Figure 7.1 Bridge design parameters.
(a) (b) (c)
Figure 7.2 Base conditions for the substructure modeled as (a) pinned (b) partiallyfixed
(c) fixed.
129
Performance of the bridge is monitored via the maximum normalized force and
displacement of the FPS isolators located on top of the pier, MNF and MND, maximum
column drifts, d
max
, the ratio of the total MNF transferred to the pier and the total MNF
transferred to the abutment, ΣMNF
pier
/ ΣMNF
abutment
, the ratio of the MND on top of the
abutment and MND on top of the pier, MND
pier
/ MND
abutment
. The variation of these
response quantities is plotted against the selected design parameters for the ten ground
motion records used in Chapter 6. Linear regression curves in the form: y(x) = ax + b, for
the median values of the response quantities are developed. The slope of these regression
curves, a
‘response quantity’
, is used to quantify the significance of the design parameters.
7.2.2 Results
Linear regression curves for the bridge response quantities as a function of the
variation of the design parameters are given in Appendix B. Figure 7.3 summarizes the
absolute values of the slope, a
’response quantity’
of the regression curves in Appendix B. The
exact values of the slopes are denoted on top of the bars with the corresponding signs.
Positive and negative signs indicate an increase and decrease of the linear regression
curves as the design parameter magnitude increases, respectively. It is observed from
Figure 7.3 that the two design parameters that have made the highest impact on the
response quantities were L
c
and L
d
*
. The variation of these two parameters caused the
largest shift in the fundamental vibration period of the structure (Table 7.7). The first
three vibration periods for the base model were T
1
=2.38 s, T
2
=2.26 s, and T
3
=2.00 s (see
Figure 6.10). Although the shape of the first three modes of vibration remained the same
with the base model, a notable change in the first period, T
1
, was observed in the bridge
as a function of L
c
and L
d
*
.
130

a
M
N
F

0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.002
0.044
0.093
0.024
0.001
f
c
/ f
co
L
d
/ L
do
L
d
*/ L
do
L
c
/ L
co
α
(a)

a
M
N
D

0.00
0.01
0.02
0.03
0.04
0.05
0.06
0.001
0.042
0.016
0.0 0.0
f
c
/ f
co
L
d
/ L
do
L
d
*/ L
do
L
c
/ L
co
α
(b)

a
d
m
a
x

0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.080
0.351
0.647
0.001
0.001
f
c
/ f
co
L
d
/ L
do
L
d
*/ L
do
L
c
/ L
co
α
(c)
131

a
Σ
M
N
F
p
i
e
r
/
Σ
M
N
F
a
b
u
t
m
e
n
t

0.0
0.2
0.4
0.6
0.8
1.0
0.106
0.792
0.643
0.222
0.002
f
c
/ f
co
L
d
/ L
do
L
d
*/ L
do
L
c
/ L
co
α
(d)

a
M
N
D
p
i
e
r
/
M
N
D
a
b
u
t
m
e
n
t

0.0
0.2
0.4
0.6
0.8
0.056
0.497
0.211
0.051
0.0
f
c
/ f
co
L
d
/ L
do
L
d
*/ L
do
L
c
/ L
co
α
(e)
Figure 7.3 Slopes, a
’response quantity
,
’
for the regression lines of the median design
parameters.
132
Table 7.2 Vibration periods of the bridges as a function of design parameters
L
c
/L
co
L
d
*/L
do
Period 0.8 1.2 1.4 1.6 0.75 1.25 1.5 1.75
T
1
2.328 2.461 2.551 2.654 2.343 2.427 2.454 2.592
T
2
2.274 2.303 2.32 2.354 2.268 2.296 2.307 2.519
T
3
1.999 1.999 2.008 2.018 1.98 2.008 2.019 2.327
The increase of L
c
created a more flexible structure with increased fundamental
vibration. Consequently, the bridge and attracted lower seismic forces. Figure 7.4 shows
the longitudinal normalized forcedeformation response of the FPS on top of the pier and
the history of the column tip displacements, δ
column
, for the Whittier Narrows earthquake
record for the cases of stiff (L
c
/L
co
=0.8) and flexible (L
c
/L
co
=1.6) substructure. It is
observed that the increased flexibility of the columns resulted in higher structural
displacements of the substructure at the isolator level and dampened the effect of seismic
isolation. Larger L
c
/L
co
resulted in flexible supports for the isolators located on top of the
piers. However, the fixity of the supports for the FPS located on top of the abutments
remained the same. Consequently, the terms indicating the difference in the responses
between isolators at the piers and abutment, MND
pier
/ MND
abutment
and ΣMNF
pier
/
ΣMNF
abutment
increased considerably as the substructure flexibility increased. The
magnitude of d
max
increased with larger substructure flexibility.
133
δ
L
/ R
0.20 0.15 0.10 0.05 0.00 0.05 0.10 0.15 0.20
f
L
/
N
o
0.3
0.2
0.1
0.0
0.1
0.2
0.3
max( f
L
/ N ) = 0.29
max( δ
L
/ R ) = 0.14
δ
L
/ R
0.20 0.15 0.10 0.05 0.00 0.05 0.10 0.15 0.20
f
L
/
N
o
0.3
0.2
0.1
0.0
0.1
0.2
0.3
max( f
L
/ N ) = 0.13
max( δ
L
/ R ) = 0.05
(a) (b)
Time (s)
0 2 4 6 8 10 12 14 16 18 20 22 24
δ
c
o
l
u
m
n
(
c
m
)
10
5
0
5
10
L
c
/ L
co
=1.6
L
c
/ L
co
=0.8
max( δ
column

)=6.7
min( δ
column

)=1.8
(c)
Figure 7.4 Forcedeformation history of the FPS in the longitudinal direction on top of the
pier for the Whittier Narrows earthquake record with (a) L
c
/L
co
=0.8 (b) L
c
/L
co
=1.6; (c)
timehistory of the longitudinal column tip deformations.
The influence of increased mass simulated via adjusted span length, L
d
*, is
unique. Although the larger gravity load has stiffened the response of the isolators and
resulted in larger MNF, it has also increased the fundamental period of the structure.
Figure 7.5 shows the longitudinal normalized forcedeformation of the FPS on top of the
pier for the Whittier Narrows earthquake record for the cases of L
d
*
/L
do
=0.75 and L
d
*
/L
do
=1.75). It is observed that larger N increased the stiffness and postyield values of the
134
FPS and caused the isolators to deform less. Larger MNF was acquired with increased
superstructure mass and this resulted in higher structural demands in the columns
observed from increased values of d
max
.
δ
L
/ R
0.20 0.15 0.10 0.05 0.00 0.05 0.10 0.15 0.20
f
L
/
N
o
0.3
0.2
0.1
0.0
0.1
0.2
0.3
max( f
L
/ N ) = 0.24
max( δ
L
/ R ) = 0.13
δ
L
/ R
0.20 0.15 0.10 0.05 0.00 0.05 0.10 0.15 0.20
f
L
/
N
o
0.3
0.2
0.1
0.0
0.1
0.2
0.3
max( f
L
/ N ) = 0.30
max( δ
L
/ R ) = 0.08
(a) (b)
Figure 7.5 Forcedeformation history of the FPS in the longitudinal direction on top of
the pier for the Helena earthquake record with (a) L
d
*
/L
do
=0.75 (b) L
d
*
/L
do
=1.75.
Increased flexibility of the superstructure had negligible effect on d
max
and MND.
However, the MNF, MND
pier
/ MND
abutment
, and ΣMNF
pier
/ ΣMNF
abutment
were
moderately affected. Figure 7.6 shows the longitudinal normalized forcedeformation
and the N history normalized with the initial gravity load, N
o
, of the FPS on top of the
pier for the Northridge earthquake record for the cases of stiff (L
d
/L
do
=0.75) and flexible
(L
d
/L
do
=1.75) superstructure. It is observed that the stiff superstructure caused
considerably larger MND and MNF and resulted in the uplift of the isolator when
compared to the flexible superstructure. This is attributed to the reduced vertical
flexibility gained by larger L
d
. As L
d
increases the vertical flexibility of the structure
increases and dampens the vertical effects of the ground motions. Consequently, it is
135
observed that the max(N/N
o
) for the flexible superstructure is approximately 60% of the
stiff superstructure. This stiffening affect increases MNF and d
max
. This indicates that
the vertical components of the ground motion are less influential for bridges with large
superstructure flexibility.
δ
L
/ R
0.20 0.15 0.10 0.05 0.00 0.05 0.10 0.15 0.20
f
L
/
N
o
0.3
0.2
0.1
0.0
0.1
0.2
0.3
max( f
L
/ N ) = 0.21
max( δ
L
/ R ) = 0.12
δ
L
/ R
0.20 0.15 0.10 0.05 0.00 0.05 0.10 0.15 0.20
f
L
/
N
o
0.3
0.2
0.1
0.0
0.1
0.2
0.3
max( f
L
/ N ) = 0.15
max( δ
L
/ R ) = 0.10
(a)
(b)
Time (s)
0 2 4 6 8 10 12 14 16
N
/
N
o
0.0
0.5
1.0
1.5
2.0
2.5
max(N/N
o
)=2.19
min(N/N
o
)=0.0
Time (s)
0 2 4 6 8 10 12 14 16
N
/
N
o
0.0
0.5
1.0
1.5
2.0
2.5
max(N/N
o
)=1.35
min(N/N
o
)=0.77
(c) (d)
Figure 7.6 Forcedeformation history of the FPS in the longitudinal direction on top of the
pier for the Northridge earthquake record with (a) L
d
/L
do
=0.75 (b) L
d
/L
do
=1.75.
136
Bridge skew, α, had negligible effect on the response quantities of the bridge.
This is attributed to the concentration of the seismic response at the isolators and the
uncoupling of the superstructure from the substructure. Similarly, the influence of f
c
on
the structural response quantities was negligible.
The influence of pinned, partiallyfixed, and fixed base modeling assumptions on
the bridge response quantities is given in Figure 7.7. The first three vibration periods of
the fixedbase and pinnedbase were T
1
=2.31 s, T
2
=2.26 s, T
3
=2.00 s and T
1
=4.80 s,
T
2
=2.31 s, T
3
=2.01 s, respectively. It is observed that the pinnedbase modeling
assumption results in substantial overestimation of the T
1
and a general underestimation
of the response quantities compared to partiallyfixed and fixed base modeling
assumptions. The base bridge model response quantities are in general closer to the
fixedbase modeling assumption.
In the pinnedbase model, the effect of the isolators at the piers was reduced
because the substructure flexibility allowed for accommodation of more of the seismic
deflections compared to the fixedbase condition. Figure 7.8 illustrates this aspect using
approximate deflected shapes of the bridge where the bases are modeled as either pinned
or fixed. The isolators located on top of the piers in the pinnedbase model attained
negligible deformations due to the deformability of the substructure. Seismic
deformations of the substructure were essentially resisted by the isolators on top of the
abutments. This is a similar phenomenon to that observed with the increased values of
L
c
. Figure 7.9 shows the longitudinal and transverse normalized forcedeformation
history of the FPS on top of the pier for the Loma Prieta earthquake record for the pinned
and fixed base conditions of the bridge model. It is observed that in the pinned base
model, the longitudinal response of the isolators located on top of the piers become
negligible in the longitudinal direction. The transverse action of the isolators is due to the
frame rigidity of the piers in these directions. It is concluded that the pinned modeling
assumption may result in considerable underestimation of isolator design parameters and
137
reduce the effectiveness of seismic isolation. Increased substructure flexibility
contributes to the difference in the response of the isolators located at the piers and the
abutments.
138
M
N
F
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.087
0.140
0.135
M
N
D
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0.032
0.068 0.067
(a)
(b)
M
N
D
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0.032
0.068 0.067
Σ
M
N
F
p
i
e
r
/
Σ
M
N
F
a
b
u
t
m
e
n
t
0.5
1.0
1.5
2.0
2.5
3.0
1.318
2.146 2.133
(c)
(d)
M
N
D
p
i
e
r
/
M
N
D
a
b
u
t
m
e
n
t
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
0.318
0.849
0.801
(e)
Figure 7.7 Median values of the response quantities of the bridge as a function of base
modeling assumptions.
139
Figure 7.8 Structural bridge responses as a function of base modeling assumptions.
140
δ
L
/ R
0.20 0.15 0.10 0.05 0.00 0.05 0.10 0.15 0.20
f
L
/
N
o
0.3
0.2
0.1
0.0
0.1
0.2
0.3
max( f
L
/ N ) = 0.11
max( δ
L
/ R ) = 0.06
δ
T
/ R
0.20 0.15 0.10 0.05 0.00 0.05 0.10 0.15 0.20
f
T
/
N
o
0.3
0.2
0.1
0.0
0.1
0.2
0.3
max( f
T
/ N ) = 0.09
max( δ
T
/ R ) = 0.04
(a)
(b)
δ
L
/ R
0.20 0.15 0.10 0.05 0.00 0.05 0.10 0.15 0.20
f
L
/
N
o
0.3
0.2
0.1
0.0
0.1
0.2
0.3
δ
T
/ R
0.20 0.15 0.10 0.05 0.00 0.05 0.10 0.15 0.20
f
T
/
N
o
0.3
0.2
0.1
0.0
0.1
0.2
0.3
max( f
T
/ N ) = 0.09
max( δ
T
/ R ) = 0.02
(c) (d)
Figure 7.9 Forcedeformation history of the FPS in the longitudinal and transverse
directions on top of the pier for the Loma Prieta earthquake record with (a), (b) fixed base
conditions; (c), (d) pinned base conditions.
7.3 Influence of a Modified Seismic Isolation Strategy
The FPS design parameters that can be investigated for a range of values are limited.
A notable disadvantage of the FPS is that it is essentially a oneparameter system based
on the radius of its concave dish, R (Naeim and Kelly 1999). It has been shown in
Chapters 4 and 5 that achieving constant prescribed friction forces is difficult due
substantial changes in the magnitude of the friction coefficient as a function of sliding
141
velocity and pressure. The smallest R of the FPS in standard manufacture by the
Earthquake Protection Systems is 99 cm (Earthquake 2003). Isolators with smaller R
have been manufactured, however, for experimental research purposes. Examples
include R = 48 cm (Erdik and Uckan 2004), R = 56 cm (Constantinou et al. 1993), and
R = 76 cm (Mosqueda et al. 2004). An important implication of smaller R is the
reduction of maximum lateral deformation capacity, ∆
max
, of the isolator. Standard radii
in manufacture for the FPS are 99, 155, 224, 305, 396, and 620 cm. It has been shown in
Chapters 5 and 6 that the vibration characteristics of bridges are essentially a function of
the isolator properties. Hence, using larger R for the FPS without additional measures
may shift the fundamental period of the structure to a region out of interest in the design
spectra.
It has been shown in Chapters 5 and 6 that there exists a notable variation in the
stiffness and yield force properties of the FPS isolators located above the piers and
abutments. Specifically, the isolators at the abutments acquired smaller yield force and
postyield stiffness compared to those located at the piers. This lead to torsional modes
of vibration of the superstructure during earthquake induced loads. Additionally, larger
displacement demands and lower transfer of seismic forces were observed at the isolators
located at the abutments compared to those located at the piers. A more favorable
response may be obtained by acquiring stiffer FPS forcedeformation response at the
structurally stronger abutments compared to the piers.
One option to increase the stiffness and yield force of the FPS located above the
abutments is to increase the magnitude of the normal force, N. Kasalanati and
Constantinou (2005) showed that the FPS isolators can be effectively prestressed to
achieve higher N. However, the primary objective of this approach is the prevention of
tensile force and uplift of the isolators. Additionally, tendons must be configured to
sustain a limited amount of tensile strain as shown in Figure 7.10. This setup may not be
practical for the configuration of the isolators located at the abutments
142
(a)
(b)
Figure 7.10 Prestressing the FPS (Kasalanati and Constantinou 2005).
7.3.1 Proposed Design
Considering the limitations on the design parameters of the FPS explained above,
a modified approach for the seismic isolation of the bridge is proposed. The objective of
the design is to:
1) Contribute more to reducing the structural demands on the columns by
achieving smaller d
max
2) Create a more balanced ΣMNF
pier
/ ΣMNF
abutment
and MND
pier
/ MND
abutment
The primary constraint is to maintain the same R for all the isolators in the
bridge to avoid additional mold costs.
The proposed bridge seismic isolation approach consists of increasing the R
throughout the isolators in the bridge and introducing a modified design to the isolators at
the abutments. The conceptual drawing of the proposed design modification for the FPS
located at the abutments is given in Figure 7.11. This simple modification consists of the
introduction of extension spring elements around the perimeter of the isolator (Figure
7.12). The objective of this supplemental design approach is to generate a stiffer force
143
deformation response for the isolators at the abutments. The springs are pinconnected to
the upper and lower steel plates of the isolator and provide supplemental source of
stiffness and energy dissipation as the isolator displace (Figure 7.11). Additionally, the
vertical position of the springs also provides tensile restraint in the event of uplift. The
number of springs may be varied as a function of the required amount of supplemental
stiffness and energy dissipation. Extension springs are proposed for the task because of
their existing widespread use in a vast variety of industries, capability to undergo
deformations that are of interest to isolators, and wide range of design parameters
(Carlson 1978; Chironis 1961). Other alternatives, such as rubber springs, may further
be explored; however, this is outside the scope of this study (see Göbel and Brichta
1974).
144
(a)
(b)
Figure 7.11 Proposed modified design for the FPS above the abutments in (a)
undeformed (b) deformed state.
145
Figure 7.12 Sample spring element to be used in the proposed modified design of the
FPS at the abutments.
The force deformationresponse of the springs is assumed to be elastic. Although
it is possible to design extension springs for pretensioned loads, this property is
neglected. The influence of the supplemental spring in the forcedeformation response of
the FPS is analyzed for the displacementcontrolled load history given in Figure 7.13. It
is assumed that a single FPS isolator with the design properties used in the analyzed
bridge and N=258 kN, is introduced in the supplemental springs that have a stiffness,
k
s
=1.286 kN/cm. Figure 7.14 gives the forcedeformation response of the FPS, spring
and the combined FPS and spring response. It is observed that the addition of the elastic
spring increases the postyield stiffness and the energy dissipation capacity of the
isolator.
146
t
(s)
0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0
δ
(
c
m
)
25
20
15
10
5
0
5
10
15
20
25
Figure 7.13 Displacementcontrolled load history
δ
/ R
0.20 0.15 0.10 0.05 0.00 0.05 0.10 0.15 0.20
f
/
N
o
0.5
0.4
0.3
0.2
0.1
0.0
0.1
0.2
0.3
0.4
0.5
FPS
Spring + FPS
Spring
Figure 7.14 Forcedeformation response of the isolator with modified design.
147
Figure 7.15 shows the configuration and the modeling of the modified FPS design
in the 3D bridge model. The supplemental springs are modeled in OpenSees as zero
length elements with elastic uniaxial material object in the x, y and z directions. The
response of the springs in the three main directions is uncoupled. This is an additional
advantage of the proposed design because it allows for the assignment of separate
stiffness characteristics in different directions to give more control over the seismic
response over the isolator. The spring element essentially acts in parallel in all three
principal directions to the response of the FPS isolator.
Figure 7.15 Modified isolator design modeling at the abutments.
7.3.2 Analyses
The bridge seismically isolated with the proposed modified design is analyzed
with the same methodology used in the previous sections for a range of R and k
s
. The
148
objective is to quantify the effects of the modified bridge seismic isolation approach. The
fundamental period of the base bridge model was T
1
=2.38 s. Five new modified designs
are introduced. Each design is attained by one of the standard R’s of the isolators
provided by the manufacturing company. The stiffness of the supplemental springs in the
isolators located at the abutments is selected to result in the same fundamental seismic
isolation period of the bridges (Table 7.3).
Table 7.3 Modified isolator design properties to achieve a fundamental period of T
1
=2.38
s
R k
p
/ k
s
99 
155 0.640
224 0.287
305 0.176
396 0.120
620 0.070
Although the shape of the vibration modes in the modified isolation designs
remained the same, the period of the third mode shapes of the bridges had notable
differences compared to the base model. The value of the T
3
of the bridges for the five
values of the R ranging from 155 cm to 620 cm in Table 7.3 were, 1.62, 1.48, 1.42, 1.38
and 1.34 s, respectively. This indicates that the torsional vibration characteristics of the
modified designs are stiffer than the base model. It is also observed that although the
postyield stiffness of the isolators is reduced with increasing R, the torionsal stiffness
increases. This is an attribute of the supplemental stiffness provided at the abutment
149
isolators to acquire the same fundamental period in the bridges isolated with the modified
design approach.
7.3.3 Results
The median of the response quantities as a function of the R of the isolators are
given in Figure 7.16. It is observed that the new seismic isolation strategy may result in
considerable improvements in the response of the bridge. The median values of the MNF
and d
max
reduces and reaches a flatter plateau as the R increases. Additionally, the
highest value of the d
max
among the suite of ground motions reduces with the new seismic
isolation strategy. The increased R implies more flexible isolators at the piers and
consequently the MND slightly increases. However, the MND
pier
/ MND
abutment
is
increased with the new isolation strategy. The ratio ΣMNF
pier
/ ΣMNF
abutment
, reduces,
indicating more energy and force transfer to the abutments. This ratio becomes
approximately equal to 1, implying a completely uniform MND distribution throughout
the isolators, for approximately R = 620 cm. Similar uniformity in the MNF is captured
at approximately R = 244 cm.
Figure 7.17 shows the longitudinal forcedeformation, of the FPS on top of the
pier and abutment for the Helena earthquake record for the cases of conventional design
with R = 99 cm and the new design with R = 620 cm. It is observed that the new design
allows for considerable reduction in the MNF and stiffness on top of the isolator located
on top of the pier. This is the fundamental principle underlying the objective of the new
design. However, the stiffness of the isolator on top of the abutment has increased. This
allowed higher isolator forces to be transferred to the abutments, which are typically
structurally stronger components compared to the piers. Additionally, the isolator
deformations of the abutment slightly reduced.
150
R (cm)
0 100 200 300 400 500 600 700
M
N
F
0.00
0.05
0.10
0.15
0.20
0.25
0.30
(a)
R (cm)
0 100 200 300 400 500 600 700
M
N
D
(
R
=
9
9
c
m
)
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
(b)
R (cm)
0 100 200 300 400 500 600 700
d
m
a
x
(
%
)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
(c)
151
R (cm)
0 100 200 300 400 500 600 700
Σ
M
N
F
p
i
e
r
/
Σ
M
N
F
a
b
u
t
m
e
n
t
0.0
0.5
1.0
1.5
2.0
2.5
3.0
(d)
R (cm)
0 100 200 300 400 500 600 700
M
N
D
p
i
e
r
/
M
N
D
a
b
u
t
m
e
n
t
(
R
=
9
9
c
m
)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
(e)
Figure 7.16 Variation of bridge response quantities as a function of FPS design
parameters.
152
δ
L
(cm)
16 12 8 4 0 4 8 12 16
f
L
(
k
N
)
150
100
50
0
50
100
150
max( f
L
) = 63.5
max( δ
L
) = 8.5
δ
L
(cm)
16 12 8 4 0 4 8 12 16
f
L
(
k
N
)
150
100
50
0
50
100
150
max( f
L
) = 38.9
max( δ
L
) = 11.1
(a)
(b)
δ
L
(cm)
16 12 8 4 0 4 8 12 16
f
L
(
k
N
)
150
100
50
0
50
100
150
max( f
L
) = 38.5
max( δ
L
) = 10.0
δ
L
(cm)
16 12 8 4 0 4 8 12 16
f
L
(
k
N
)
150
100
50
0
50
100
150
max( f
L
) = 74.1
max( δ
L
) = 10.2
(c) (d)
Figure 7.17 Forcedeformation history of the FPS in the longitudinal direction for the
Helena earthquake (a), (b) conventional design with R = 99 cm; (c), (d) new design with R
= 621 cm; (a), (c) on top of the pier; (b), (d)on top of the abutment.
7.4 Conclusion
In this Chapter the influence of the common bridge design parameters on the
system’s response quantities have been investigated. A new seismic isolation strategy
using a modified FPS design at the abutments has been presented. The influence of the
design parameters in this new seismic isolation approach has been parametrically
investigated. The following conclusions have been made:
153
1) The two design parameters that had the highest effect on the response quantities
of the SIB were column height, L
c
, and superstructure dead weight which was
imposed via additional superstructure length, L
d
*
. Both design parameters also had
a notable effect on the fundamental period of the bridge
2) Larger substructure flexibility may result form base modeling assumptions or
longer piers. SIBs with flexible substructures acquire higher vibration periods
thus lower earthquake forces. However, flexible substructure reduces the
effectiveness of seismic isolation. As the flexibility of the substructure of SIBs
increase, the capacity to utilize the isolators at the piers tend to reduce and the
deviation between the responses of the isolators located on top of the piers and
abutments increase.
3) SIBs with larger superstructure mass acquire higher vibration periods. Increased
superstructure mass, reduces the isolator deformations and increases the demands
on the substructure components.
4) Increased superstructure flexibility acquired via longer span length dampens the
inertial loads caused by the vertical components of the superstructure. This
results in lower isolator forces and structural demands on the substructure.
However, increased superstructure flexibility contributed to the variation between
the responses of the isolators located on top the piers and the abutments.
5) Bridge skew has negligible effect on SIBs. This is mainly due to the uncoupling
of the superstructure from the superstructure. The substructure nominal concrete
strength had negligible effect on the response of the bridge.
6) The modified isolator design effectively increases the postyield stiffness and
energy dissipation capacity of the FPS utilizing springs widely found in the
industry. This new design can be tailored to achieve different isolator force
deformation properties by changing the design, material and number of springs.
154
7) The new seismic isolation strategy allowed for a more balanced distribution of
isolator forces and displacements throughout the bridge. This was achieved using
the standard isolator dimensions manufactured by the parent company.
8) When compared to the conventional seismic isolation approach, the new strategy
reduced the isolator deformation demands by utilizing the isolators more evenly at
the abutments and piers.
155
CHAPTER 8
SUMMARY, CONCLUSIONS, AND FUTURE RESEARCH
8.1 Summary and Conclusions
Seismic isolation is an effective tool for improving the structural performance of
bridges susceptible to earthquake induced loads. The dynamic response of bridges may
be shifted towards higher periods via isolators to attract lower seismic forces. Isolators
typically govern the dynamic characteristics of bridges and may be a powerful tool to
calibrate the system’s seismic response to a desired level by the designer. Currently,
seismic isolators are classified as sliding and elastomeric. Despite being considered a
relatively mature technology, sliding seismic isolation has been incorporated in bridge
design codes only in 1997 and has not found widespread use particularly in the highway
bridge community. One reason for this is the absence adequate analytical models that
can capture the highly nonlinear behavior of the isolators. In addition, current seismic
isolation design codes do not provide any guidance about the selection of isolator types.
Examination of the affects of bridge and isolator design parameters on the system’s
seismic response has been limited. The objective of this study is to enhance the
understanding of the structural response of bridges utilizing sliding seismic isolation via
models that can capture the complex behavior of the isolators. This study was intended
to provide support for seismic risk mitigation and insight for the analysis and design of
SIBs by quantifying response characteristics. The Fiction Pendulum System (FPS) was
selected as the representative isolator and given particular emphasis in the analyses.
The review of the currentstateoftheart on bridge seismic isolation studies
showed that there are three issues that need further clarification:
1) The level of accuracy required for modeling the forcedeformation behavior of
isolators;
156
2) The comparative response of the two main types of seismic isolators (i.e.
sliding and elastomeric);
3) The influence of bridge and isolator design parameters on the system’s
response.
The influence of seismic isolation in a simplified bridge model was examined via
modal analysis. It was shown that: (1) dynamic loads are substantially reduced via the
insertion of a flexible element between the substructure and the superstructure; (2)
vibration modes are governed by those involving the isolators; (3) the characteristics of
the isolators are an important determinant of the dynamic response of the bridge. The
nonlinear timehistory (NLTH) analysis was selected as the method for further
examination of SIB response.
A new finite element (FE) model of the FPS was implemented into the existing
library of OpenSees. New C++ material and element classes were developed and
compiled into the open source framework of OpenSees to model the response of the FPS.
This model makes use of the existing research findings on the response components of
the FPS. Unlike previous models in the literature, this model can account simultaneously
for the variation in the normal force and friction coefficient, large deformation effects,
and the coupling of the vertical and horizontal response during motion. The FE model
was validated using experimental data from shake table tests of structures seismically
isolated with the FPS. The influence of the modeling assumptions on the force
deformation response of the FPS was monitored on the developed for the verification
study.
The response of typical highway bridges isolated with the FPS was investigated as
a function of isolator modeling assumptions. A MultiSpan Continuous Steel Girder
(MSCSG) bridge was used for this purpose. Threedimensional (3D) model of the bridge
isolated with the FPS was developed. The bridge was modeled with a high degree of
detail of the substructure components, and base and other boundary conditions. The new
157
FE model developed for the FPS was incorporated into the bridge model. Seven other
models of the SIB were generated with the only difference in the FPS modeling
assumptions. Each model Isolator and bridge response characteristics were monitored
under NLTH analyses that utilized 2% in 50 years hazard level earthquakes. It was
concluded that, the most important modeling aspects of the FPS in bridge applications are
the normal force and friction coefficient variations, and bidirectional coupling. These
parameters may have considerable effects on isolator design. However, the effects of the
accuracy of these modeling parameters on bridge structural response are weak to
moderate. Large deformation effects of the isolators were found to be negligible for the
systems considered in this study. Pounding between the deck and the abutments may
occur due to increased deformations at the isolator level. This occurrence limits the
effectiveness of the isolators and may lead to unanticipated structural damage.
The seismic response characteristics of a MultiSpan Continuous Concrete Girder
(MSCCG) bridge with: (a) sliding (b) elastomeric seismic isolation was compared using
advanced isolator models. The FPS and the LeadRubber Bearings (LRB) were selected
as representative isolators for sliding and elastomeric seismic isolators types,
respectively. A new FE model of the LRB has been implemented into the existing library
of OpenSees using a similar approach adopted for the FPS. In addition to accounting for
bidirectional coupling effects available in models found in the literature, this model can
account simultaneously for the variation in the normal force and large deformation
effects. Two models of a MultiSpan Continuous Concrete Girder (MSCCG) bridge were
generated and one was isolated with the LRB and the other was isolated with the FPS
with approximately the same seismic isolation periods. The two bridge models were
subjected to NLTH with a suite of 7% probability of exceedance in 75 years hazard level
earthquake records. The influence of vertical components of the ground motions on the
two isolators was assessed. Isolator and bridge response characteristics were monitored
for the two seismic isolation schemes. It was concluded that unlike the FPS, the LRB
158
forcedeformation response is significantly affected from large deformations. There is a
notable loss of lateral stiffness in LRB at large deformations that may lead also lead to
buckling. The influence of normal load variations on the forcedeformation
characteristics of the LRB was weak. The choice of elastomeric or sliding seismic
isolation of bridges may have considerable effects despite attaining similar vibration
periods. The LRB acquired higher displacements, however, placed smaller demands on
the columns compared to the FPS. The FPS is capable of accommodating vertical
components of the ground motion records that might result in a buckling failure of the
LRB. The LRB possesses a more uniform response among isolators located at the
abutments and piers in comparison to the FPS.
The influence of the design characteristics of the bridge and the isolators on the
system’s seismic response was examined. The previously established MSCCG bridge
models were used for this task. Five bridge design parameters were selected. The range
of these design parameters were established based on previous inventory analyses
performed in the Central and Southeastern United States. Nonlinear time history
analyses were performed by changing the design parameters within the four equally
spaced values of their established bounds. Previously selected suite of 7% probability of
exceedance in 75 years hazard level earthquake records were used in the analyses. Linear
regression curves for the median values of the response quantities were developed from
the results of the analyses. The slope of these regression curves were used to quantify the
significance of the design parameters on the system’s response. A new bridge seismic
isolation strategy that involves a practical modification in the design of the FPS was
proposed. The objective was to increase the effectiveness of seismic isolation with the
FPS by achieving a more uniform response among isolators located at the piers and the
abutments. The implications of the proposed design on the FPS forcedeformation
response were examined. The seismic response of the system was investigated as a
function of different parameters of this new seismic isolation strategy. It was concluded
159
that seismic isolation with the FPS is less effective for bridges that have flexible
substructure characteristics, compared with those that are less flexible. Base conditions
and column height were identified as the two parameters to impact substructure flexibility
the most. Bridges with large superstructure flexibility tend to be effected less from the
effects of the vertical components of the ground motions. Bridges with large
superstructure mass acquires less isolator displacements but place larger demands on the
load carrying elements in the substructure. Unlike nonisolated bridges, the seismic
response characteristics of bridges seismically isolated with the FPS were not affected by
skew. The two disadvantages of the FPS are: (1) the limitation of the design parameters
to attain desired forcedeformation response characteristics and; (2) the significant
variation in the forcedeformation responses characteristics of the isolators located at the
piers and the abutments. A new design approach for the FPS is proposed to overcome
these limitations. This modification involves installing supplemental springs around the
isolator to provide latitude for alternate forcedeformation characteristics. It was
possible to reduce structural demands on the load carrying elements of the bridge by
creating a more uniform response among isolators located at the piers and abutments via
the new design.
8.2 Future Research
The following are possible areas which this research can be extended to:
• Ground motions have a wide range of variability in their characteristics stemming
from proximity, frequency content and directivity. The influence of ground
motion characteristics on the seismic response of SIBs should be examined.
• Seismic isolation is applicable to virtually any type of bridge. The structural and
dynamic characteristics of these bridges may have significant variations. The
effectiveness of isolators on different bridge types should be quantified.
160
• A costbenefit analysis should be performed for a range of isolator and bridge
types. This will contribute to the wider application of the seismic isolation
technology by making these elements a catalog commodity for design.
• Buildings typically possess larger overturning moments compared to bridges.
Additionally, the design objectives and the placement of isolators in buildings
have significant differences from bridges. The influence of isolator modeling
assumptions and the selection of the type of seismic isolator on building response
should be investigated.
• Possible improvements with hybrid seismic isolation schemes should be explored.
In addition to considering different isolator types, these scenarios should include
retrofit techniques based on strengthening such as jacketing, and other protective
devices such as dampers.
161
APPENDIX A
METHOD OVERRIDING IN C++
Open System for Earthquake Engineering Simulation (OpenSees) is an open
source software framework for simulating the earthquake response of structural and
geotechnical systems (Mazzoni et al. 2006). OpenSees has an open source object
oriented architecture in the C++ programming language that maximizes its modularity,
thus making it a viable choice for research purposes. This framework has been used to
simulate the complex forcedeformation response of the FPS and the LRB isolators.
Despite the availability of versatile material and element models inside the library of
OpenSees, none has the capability of effectively modeling the complex force
deformation response of the isolators described in chapters 4 and 6. A material class,
FPSmaterial, and a zerolength element class, FPSelement, has been added to the
existing library of OpenSees to model the forcedeformation response of the FPS
(McKenna 2005a; McKenna 2005b). These classes were derived as child classes of the
existing generic UniaxialMaterial and ZeroLength parent classes. A critical approach
used to achieve this task was ‘overriding’.
The parent UniaxialMaterial class has a method called: setTrialStrain which
is designed to take in two parameters: (1) strain (strain) and; (2) strain rate
(strainRate) at step i of the analysis and update the values of trial stress (Tstress) and
trial tangent stiffness (Ttangent) values based on the predefined constitutive relationship
for step i+1. The setTrialStrain method defined in the parent UniaxialMaterial
class is pure virtual base class thus no objects of it's type can be instantiated. Classes
derived from the UniaxialMaterial class must implement the setTrialStrain method
to avoid a compiler error. However, the constitutive relationship defined for the FPS in
162
Chapter 4 requires additional parameters to the default ones. To achieve this, the
setTrialStrain is kept inside the code but designed to not function by:
int FPSMaterial::setTrialStrain(double strain, double strainRate)
{return 0;}
A new setTrialStrain method to override the existing default one in FPSmaterial is
generated from:
int FPSMaterial::setTrialStrain(double frictionC, double Alpha, double
zz, double n_modified, double strain, double strainRate)
where frictionC is µ (Equation 4.3), Alpha isα (Equation 4.12), zz is the
corresponding parameter of η (Equation 4.8), n_modified is N (Equation 4.2). This
overriding method is also added to the header file of the parent UniaxialMaterial class:
virtual int setTrialStrain (double frictionC, double Alpha, double zz,
double n_modified, double strain, double strainRate) {return 0;}
The utilization of the new setTrialStrain method inside the FPSelement is as
following:
for (int mat=0; mat<3; mat++) {
if(mat==0){
strain = this>computeCurrentStrain1d(0,diff);
strainRate = this>computeCurrentStrain1d(0,diffv);
ret += theMaterial1d[mat]
>setTrialStrain(frictionC_x,Alpha,y1,n_modified,strain,strainRate);}
else if(mat==1) {
strain = this>computeCurrentStrain1d(1,diff)
strainRate = this>computeCurrentStrain1d(1,diffv);
ret += theMaterial1d[mat]
>setTrialStrain(frictionC_y,Alpha,y2,n_modified,strain,strainRate);}
else {
strain = this>computeCurrentStrain1d(mat,diff );
strainRate = this>computeCurrentStrain1d(mat,diffv);
ret += theMaterial1d[mat]>setTrialStrain(strain,strainRate);}
}
163
APPENDIX B
REGRESSION LINES FOR THE EFFECTS OF BRIDGE DESIGN
PARAMETERS ON SYSTEM RESPONSE
The following figures present the variation of maximum normalized force and
displacement of the FPS isolators located on top of the pier, MNF and MND, maximum
column drifts, d
max
, the ratio of the total MNF transferred to the pier and the total MNF
transferred to the abutment, ΣMNF
pier
/ ΣMNF
abutment
, the ratio of the MND on top of the
abutment and MND on top of the pier, MND
pier
/ MND
abutment
, for the earthquakes used in
chapter 7. The bridges design parameters considered are column height, L
c
, length of the
superstructure with constant mass, L
d
, length of the superstructure with adjusted mass,
L
d
*, degree of skew, a, and the nominal concrete strength of the substructure, f
c
. Figures
a re given as a ratio of the design parameters to the base model design values denoted by
a subscript ‘o’ of the design parameter. The term a denotes the slope of the linear
regression curve established from the median values of the response quantities as a
function of the design parameters. The term R
2
is the coefficient of multiple
determination used to measure the amount of reduction in the variability of the response
quantity median by the design parameter variables.
164
a = 0.002
R
2
= 0.495
f
co
= 34.5 MPa
f
c
/ f
co
0.4 0.6 0.8 1.0 1.2 1.4 1.6
M
N
F
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.137
0.140 0.140 0.140
0.139
(a)
a = 0.001
R
2
= 0.013
f
co
= 34.5 MPa
f
c
/ f
co
0.4 0.6 0.8 1.0 1.2 1.4 1.6
M
N
D
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0.065
0.071
0.066
0.067
0.068
(b)
a = 0.080
R
2
= 0.988
f
co
= 34.5 MPa
f
c
/ f
co
0.4 0.6 0.8 1.0 1.2 1.4 1.6
d
m
a
x
(
%
)
0.0
0.2
0.4
0.6
0.8
1.0
0.473
0.463
0.434
0.445
0.409
(c)
Figure B.1 The influence of the pier concrete compressive strength, f
c
, on (a) MNF (b)
MND (c) d
max
(d) ΣMNF
pier
/ ΣMNF
abutment
(e) MND
pier
/ MND
abutment
165
a = 0.106
R
2
= 0.912
f
co
= 34.5 MPa
f
c
/ f
co
0.4 0.6 0.8 1.0 1.2 1.4 1.6
Σ
M
N
F
p
i
e
r
/
Σ
M
N
F
a
b
u
t
m
e
n
t
0.5
1.0
1.5
2.0
2.5
3.0
2.053
2.087
2.125
2.133
2.139
(d)
a = 0.056
R
2
= 0.896
f
co
= 34.5 MPa
f
c
/ f
co
0.4 0.6 0.8 1.0 1.2 1.4 1.6
M
N
D
p
i
e
r
/
M
N
D
a
b
u
t
m
e
n
t
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
0.760
0.779
0.789 0.801
0.810
(e)
Figure B.1 continued
166
a = 0.044
R
2
= 0.906
L
co
= 4.6 m
L
c
/ L
co
0.6 0.8 1.0 1.2 1.4 1.6 1.8
M
N
F
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.137
0.140
0.120
0.127
0.105
(a)
a = 0.042
R
2
= 0.972
L
co
= 4.6 m
L
c
/ L
co
0.6 0.8 1.0 1.2 1.4 1.6 1.8
M
N
D
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0.071
0.067
0.053
0.063
0.040
(b)
a = 0.351
R
2
= 0.965
L
co
= 4.6 m
L
c
/ L
co
0.6 0.8 1.0 1.2 1.4 1.6 1.8
d
m
a
x
(
%
)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.386
0.445
0.572
0.563
0.671
(c)
Figure B.2 The influence of the column length, L
c
, on (a) MNF (b) MND (c) d
max
(d)
ΣMNF
pier
/ ΣMNF
abutment
(e) MND
pier
/ MND
abutment
167
a = 0.792
R
2
= 0.954
L
co
= 4.6 m
L
c
/ L
co
0.6 0.8 1.0 1.2 1.4 1.6 1.8
Σ
M
N
F
p
i
e
r
/
Σ
M
N
F
a
b
u
t
m
e
n
t
0.0
0.5
1.0
1.5
2.0
2.5
3.0
2.117
2.133
1.726
1.861
1.540
(d)
a = 0.497
R
2
= 0.994
L
co
= 4.6 m
L
c
/ L
co
0.6 0.8 1.0 1.2 1.4 1.6 1.8
M
N
D
p
i
e
r
/
M
N
D
a
b
u
t
m
e
n
t
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
0.898
0.801
0.607
0.675
0.495
(e)
Figure B.2 continued
168
a = 0.024
R
2
= 0.918
L
do
= 30.3 m
L
d
/ L
do
0.50 0.75 1.00 1.25 1.50 1.75 2.00
M
N
F
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.153
0.140
0.130 0.135
0.127
(a)
a = 0.0
R
2
= 0.0
L
do
= 30.3 m
L
d
/ L
do
0.50 0.75 1.00 1.25 1.50 1.75 2.00
M
N
D
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0.065
0.060
0.060
0.064
0.065
(b)
a = 0.001
R
2
= 0.001
L
do
= 30.3 m
L
d
/ L
do
0.50 0.75 1.00 1.25 1.50 1.75 2.00
d
m
a
x
(
%
)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.455
0.445
0.459
0.473
0.447
(c)
Figure B.3 The influence of longitudinal deck length, L
d
, with constant mass on (a) MNF
(b) MND (c) d
max
(d) ΣMNF
pier
/ ΣMNF
abutment
(e) MND
pier
/ MND
abutment
169
a = 0.222
R
2
= 0.708
L
do
= 30.3 m
L
d
/ L
do
0.50 0.75 1.00 1.25 1.50 1.75 2.00
Σ
M
N
F
p
i
e
r
/
Σ
M
N
F
a
b
u
t
m
e
n
t
0.5
1.0
1.5
2.0
2.5
3.0
2.148 2.133
1.956
1.939
1.952
(d)
a = 0.051
R
2
= 0.466
L
do
= 30.3 m
L
d
/ L
do
0.50 0.75 1.00 1.25 1.50 1.75 2.00
M
N
D
p
i
e
r
/
M
N
D
a
b
u
t
m
e
n
t
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
0.774
0.801
0.727
0.738
0.746
(e)
Figure B.3 continued
170
a = 0.093
R
2
= 0.975
L
do
= 30.3 m
L
d
/ L
do
0.50 0.75 1.00 1.25 1.50 1.75 2.00
M
N
F
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.116
0.140
0.178
0.156
0.219
(a)
a = 0.016
R
2
= 0.796
L
do
= 30.3 m
L
d
/ L
do
0.50 0.75 1.00 1.25 1.50 1.75 2.00
M
N
D
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0.072
0.061
0.061
0.062
0.052
(b)
a = 0.674
R
2
= 0.987
L
do
= 30.3 m
L
d
/ L
do
0.50 0.75 1.00 1.25 1.50 1.75 2.00
d
m
a
x
(
%
)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.351
0.445
0.623
0.573
0.772
(c)
Figure B.4 The influence of longitudinal deck length, L
d
, with adjusted mass on (a) MNF
(b) MND (c) d
max
(d) ΣMNF
pier
/ ΣMNF
abutment
(e) MND
pier
/ MND
abutment
171
L
do
= 30.3 m
a = 0.643
R
2
= 0.978
L
d
/ L
do
0.50 0.75 1.00 1.25 1.50 1.75 2.00
Σ
M
N
F
p
i
e
r
/
Σ
M
N
F
a
b
u
t
m
e
n
t
0.5
1.0
1.5
2.0
2.5
3.0
2.331
2.133
1.809
1.877
1.673
(d)
L
do
= 30.3 m
a = 0.211
R
2
= 0.979
L
d
/ L
do
0.50 0.75 1.00 1.25 1.50 1.75 2.00
M
N
D
p
i
e
r
/
M
N
D
a
b
u
t
m
e
n
t
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
0.840
0.801
0.701
0.724
0.628
(e)
Figure B.4 continued
172
a = 0.001
R
2
= 0.200
α
o
= 0
o
α
10 0 10 20 30 40 50
M
N
F
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.140
0.147
0.127
0.139
0.140
(a)
a = 0.0
R
2
= 0.0
α
o
= 0
o
α
10 0 10 20 30 40 50
M
N
D
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0.067
0.075 0.074
0.075
0.067
(b)
a = 0.001
R
2
= 0.530
α
o
= 0
o
α
10 0 10 20 30 40 50
d
m
a
x
(
%
)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.445 0.455
0.431
0.401
0.397
(c)
Figure B.5 The influence of skew angle, α, on (a) MNF (b) MND (c) d
max
(d) ΣMNF
pier
/
ΣMNF
abutment
(e) MND
pier
/ MND
abutment
173
a = 0.002
R
2
= 0.157
α
o
= 0
o
α
10 0 10 20 30 40 50
Σ
M
N
F
p
i
e
r
/
Σ
M
N
F
a
b
u
t
m
e
n
t
0.5
1.0
1.5
2.0
2.5
3.0
2.133
2.121
1.944 1.978
2.112
(d)
a = 0.0
R
2
= 0.003
α
o
= 0
o
α
10 0 10 20 30 40 50
M
N
D
p
i
e
r
/
M
N
D
a
b
u
t
m
e
n
t
0.4
0.6
0.8
1.0
1.2
1.4
0.801
0.852
0.761 0.762
0.842
(e)
Figure B.5 continued
174
REFERENCES
AASHO (1965). Standard specifications for highway bridges, Washington D.C.
AASHTO (1999). Guide specifications for seismic isolation design, Washington, D.C.
Almazan, JL, Llera, JCDL, and Inaudi, JA (1998). “Modeling aspects of structures
isolated with the frictional pendulum system,” Earthquake Engineering and
Structural Dynamics, 27, 845867
Almazan, JC and Llera, JCDL (2002). “Analytical model of structures with frictional
pendulum isolators,” Earthquake Engineering and Structural Dynamics, 31,
305332
Almazan, JC and Llera, JCDL (2003). “Physical model for dynamic analysis of structures
with FPS isolators,” Earthquake Engineering and Structural Dynamics, 32,
11571184
Anderson, EL, and Mahin, A (2004). “An evaluation of bidirectional earthquake shaking
on the provisions of the AASHTO Guide Specifications for Seismic Isolation
Design,” 13th World Conference on Earthquake Engineering, Vancouver, B.C.,
Canada
Buckle, IG, Constantinou, MC, Dicleli, M, and Ghasemi, H (2006). Seismic isolation of
highway bridges, Technical Report MCEER06SP07, Multidisciplinary Center
for Earthquake Engineering Research, State University of New York at Buffalo
Buckle, IG and Liu, H (1993). “Stability of elastomeric seismic isolation systems,” ATC
Seminar on Seismic Isolation, Passive Energy Dissipation and Control,
Redwood City, California
Buckle, IG and Mayes, RL (1990). “Seismic isolation: history, application and
performance – a world review,” Earthquake Spectra, 6(2)
Buckle, IG and Kelly, JM (1986). “Properties of slender elastomeric isolation bearings
during shake table studies of a largescale model bridge deck,” Joint Sealing
and Bearing Systems for Concrete Structures (ACI), Vol. 1, 247269
Button, MR, Cronin, CJ, Mayes, RL (2002). “Effect of vertical ground motions on
seismic response of highway bridges,” Journal of Structural Engineering,
128(12), 15511564
Carlson, H (1978). Spring designer’s handbook, Marcel Dekker, New York
175
Chironis, NP (1961). Spring designer and application, McGraw Hill, New York
Chopra, AK (2001). Dynamics of structures theory and applications to earthquake
engineering, 2
nd
Ed., Prentice Hall, New Jersey
Collier, CJ and Elnashai, AS (2001). “A procedure for combining vertical and horizontal
seismic action effects,” Journal of Earthquake Engineering, 5(4), 521539
Constantinou, MC, Mokha, A, Reinhorn, A (1990). “Teflon bearings in base isolation. II:
modeling,” Journal of Structural Engineering, 116(2), 455474
Constantinou, MC, Tsopelas, P, Kim, YS, and Okamoto, S (1993). NCEERTaisei
corporation research program on sliding seismic isolation systems for bridges:
experimental and analytical study of a friction pendulum system (fps), Technical
Report NCEER930020, National Center for Earthquake Engineering
Research, State University of New York at Buffalo
Cook, RD, Malkus, DS, Plesha, ME, and Witt, RJ (2002). Concepts and applications of
finite element analysis, 4
th
Ed., Kohn Wiley & Sons, New York
Cooper, J, Friedland, IM, Buckle, IG, Nimis, RB, and Bobb, NM (1994). “The
Northridge: progress made, lessons learned in seismic design of bridges,”
Public Roads, 58(1)
DIS (1996). Earthquake performance of structures on leadrubber isolations system,
Lafayette, California
Dicleli, M and Buddaram, S (2006). “Effect of isolator and ground motion characteristics
on the performance of seismicisolated bridges,” Earthquake Engineering and
Structural Dynamics, 35, 233250
Dicleli, M, Albhaisi, S and Mansour, MY (2005). “Static soilstructure interaction effects
in seismicisolated bridges” Practice Periodical on Structural Design and
Construction, 10(1), 2233
Dicleli, M and Mansour, MY (2003). “Seismic retrofitting of bridges in Illinois using
friction pendulum seismic isolation bearings and modeling procedures,”
Engineering Structures, 25, 11391156
Dicleli, M (2002). “Seismic design of lifeline bridge using hybrid seismic isolation,”
Journal of Bridge Engineering, 7(2), 94103
Earthquake Protection Systems (2003). Technical Characteristics of Friction Pendulum
Bearings, Vallejo, California
176
Erdik, M, and Uckan (2004). “Earthquake response of a 1:4 scaled three story steel
structure seismically isolated by the FPS type sliding isolation system,”
Proceedings of the Third European Conference on Structural Control, Vienna,
Austria
FEMA (1997). “NEHRP Recommended Provisions for Seismic Regulations for New
Buildings and Other Structures, Part 1: Provisions.” Report No. FEMA 302,
Federal Emergency Management Agency
Fenves, GL (2005). “Brief notes on objectoriented software design and programming
with C++,” OpenSees Developer`s Workshop, Berkeley, California
Gardiner, WP, and Gettinby, G (1998). Experimental design techniques in statistical
practice, Horwood Publishing Limited, West Sussex, England
Ghobarah, A and Ali, HM (1988). “Seismic performance of highway bridges,”
Engineering Structures, 10, 157166
Ghobarah, A (1988). “Seismic behavior of highway bridges with base isolation,”
Canadian Journal of Civil Engineering, 15, 7278
Göbel, EF and Brichta, AM (1978). Rubber spring design, NewnesButterworths,
London
Hwang, JS and Hsu, TY (2000). “Experimental study of isolated building under triaxial
ground excitations,” ASCE Journal of Structural Engineering, 126(8), 879886
Iemura, H, Taghikhany, T, Takahashi, Y and Jain, SK (2005). “Effect of variation of
normal force on seismic performance of resilient sliding isolation systems in
highway bridges,” Earthquake Engineering and Structural Dynamics, 34, 1777
1797
International Code Council (2000). International building code, Falls Church, Virginia
Jangid, RS (2004). “Seismic response of isolated bridges,” Journal of Bridge
Engineering, Vol. 9, No. 2, 156166
Jangid, RS (2004). “Optimum friction pendulum system for nearfault motions,”
Engineering Structures, 27(3), 349359
Jangid, RS and Kelly, JM (2001). “Base isolation for nearfault motions,” Earthquake
Engineering and Structural Dynamics, 30, 691707
Kasalanati, A and Constantinou, MC ( 2005). “Testing and modeling of prestressed
isolators,” ASCE Journal of Structural Engineering, 131(6), 857866
177
Kelly, J (2003). “Tension buckling in multilayer elastomeric bearings,” Journal of
Engineering Mechanics, December, 13631368
Kelly, J (1997). Earthquakeresistant design with rubber, SpringerVerlag, London
Kelly, JM, Buckle, IG, Tsai, HC (1986). Earthquake simulator testing of a baseisolated
bridge deck, Report No. EERC 198509, University of California, Berkeley
Kreyszeig, E (1999). Advanced engineering mathematics, 8
th
Ed., John Wiley & Sons,
New York
Lee, HJ, and Schiesser, WE (2004). Ordinary and partial differential equation routines in
C, C++, Fortran, Java, Maple, and Matlab, Chapman & Hall/Rc, Boca Raton
Lee, GC, Kitane, Y and Buckl, IG (2002). “Literature review of the observed
performance of SIBs,” Research Progress and Accomplishments: 20012003,
MCEER, University at Buffalo State University of New York
Liao, W, Loh, CH, and Lee, BH (2004). “Comparison of dynamic response of isolated
and nonisolated continuous girder bridges subjected to nearfault ground
motions,” Engineering Structures, 26(2004), 21732183
Maleki, S (2001a). “Free vibration of skewed bridges,” Journal of Vibration and Control,
7, 935952
Maleki, S (2001b). “Effects of diaphragms on seismic response of skewed bridges,”
1First MIT Conference on Computational Fluid and Solid Mechanics, June 12
15 2001
Mander, JB, Kim, DK, Chem, SS, and Premus, GJ (1996). “Response of steel bridge
bearings to the reversed cyclic loading,” Report No. NCEER 960014, NCEER,
Buffalo, NY
Matsagar, VA and Jangid, RS (2004). “Influence of isolator characteristics on the
response of baseisolated structures,” Engineering Structures, 26, 17351749
Mayes, R (2002). “Impediments to the implementation of seismic isolation,” ATC
Seminar on Response Modification Technologies for PerformanceBased
Seismic Design, Los Angeles, California
Mazzoni, S, McKenna, FT, Scott, MH and Fenves, GL (2007). “Open System for
Earthquake Engineering Simulation User Manual,” <http://peer.berkeley.edu/
~silvia/OpenSees/manual/html/>
178
McKenna, F (2005a). “Introducing a new uniaxial material into OpenSees,” OpenSees
Developer`s Workshop, Berkeley, California
McKenna, F (2005b). “Adding an element into OpenSees,” OpenSees Developer`s
Workshop, Berkeley, California
Mokha, AS, Constantinou, MC, Reinhorn, AM (1991). “Verification of friction model of
Teflon bearings under triaxial load,” ASCE Journal of Structural Engineering,
119(1), 240261
Mokha, AS, Constantinou, MC, Reinhorn, AM (1991). “Further results on frictional
properties of teflon bearings,” Journal of Structural Engineering, 117(2), 622
626
Mokha, AS, Constantinou, MC, Reinhorn, AM (1990). “Teflon bearings in base isolation.
I: testing.” Journal of Structural Engineering, 116(2), 455475
Mokha, AS, Constantinou, MC, Reinhorn, AM (1990). “Teflon bearings in base isolation.
II: modeling.” Journal of Structural Engineering, 116(2), 439454
Montgomery, DC (2005). Design and analysis of experiments, 6
th
Edition, John Wiley &
Sons, U.S.A.
Mosqueda, G, Whittaker, AS, Fenves, GL (2004). “Characterization and modeling of
friction pendulum bearings subjected to multiple components of excitation,”
Journal of Structural Engineering, 130(3), 433442
Muthukumar, S. (2003). “A Contact Element Approach with Hysteresis Damping for the
Analysis and Design of Pounding in Bridges,” PhD thesis, Georgia Institute of
Technology.
Naeim, F and Kelly, J (1996). Design of seismic isolated structures, 1
st
Ed., Wiley, New
York
Nakajima, K, Iemura, H, Takahashi, Y, Ogawa, K. (2000). “Pseudo dynamic tests and
implementation of sliding bridge isolators with vertical motion,” 12
th
World
Conference on Earthquake Engineering, Vancouver, Canada
Nielson, B (2005). “Analytical Fragility Curves for Highway Bridges in Moderate
Seismic Zones,” PhD thesis, Georgia Institute of Technology.
Ordonez, D, Foti, D and Bozzo, L (2003). “Comparative study of the inelastic response
of base isolated buodings,” Earthquake Engineering and Structural Dynamics,
32, 151164
179
Palfalvi, B, Amin, A, Mokha, A, Fatehi, H, and Lee, P (1993). “Implementation issues in
seismic isolation retrofit of government buildings.” ATC171 Seminar on
Seismic Isolation, Passive Energy Dissipation and Active Control, San
Francisco, CA, March.
Papazoglou, AJ and Elnashai, AS (1996). “Analytical and field evidence of the damaging
effect of vertical earthquake ground motion,” Earthquake Engineering and
Structural Dynamics, 25, 11091137
PEER Strong Motion Database. (2000). Retrieved October 10, 2007 from
http://peer.berkeley.edu/smcat/
Press, WH, Vetterling, WT, Teukolsky, SA, and Flannery, BP (2003). Numerical recipes
in C++: the art of scientific computing, Cambridge University Press,
Cambridge
Priestley, MJN, Seible, F and Calvi, GM (1996). Seismic design and retrofit of bridges,
John Wiley and Sons, New York
Pritchard, PJ (1998). MathCad: a tool for engineering problem solving, McGrawHill
Science
Rosenbrock, HH (1963). “Some general implicit processes for the numerical solution of
differential equations” The Computer Journal, 5(4), 329330
Ryan, KL and Chopra, AK (2005). Estimating the seismic response of baseisolated
buildings including torsion, rocking and axialload effects, Report No. EERC
200501, University of California, Berkeley
Saadeghvaziri, MA, and Futch, DA (1991). “Dynamic behavior of R/C highway bridges
under the combined effect of vertical and horizontal earthquake motions,”
Earthquake Engineering and Structural Dynamics, 20, 535549
Shenton, HW (1996). Guidelines for prequalification, prototype and quality control
testing of seismic isolation systems, National Institute of Standards and
Technology, Gaithersburg, Maryland
Shampine, LF, Gladwell, I, and Thompson, S (2003). Solving ODEs with Matlab,
Cambridge University Press, New York
Silva, WJ (1997). “Characteristics of vertical strong ground motions for applications to
engineering design,” Proceedings of the FHWA/NCEER Workshop on the
National Representation of Seismic Ground Motion for New and Existing
Highway Facilities, Technical Report NCEER970010
180
Skinner, RI, Robinson, WH, and McVerry, GH (1993). An introduction to seismic
isolation, John Wiley & Sons, New York
Stanton, JF (1998). “The 1997 AASHTO seismic isolation guide specification,”
Proceedings of the U.S.Italy Workshop on Seismic Protective Systems for
Bridges, Technical Report MCEER980015, New York
Stanton, J and Roeder, C (1991). “Advantages and limitations of seismic isolation,”
Earthquake Spectr, 7(2), 301323
Stevens, A (1997). Teach yourself C++, 5
th
Edition, Henry Holt & Co., New York
Takahashi, Y, Iemura, H, Yanagawa, S, Hibi, M (2004). “Shaking table test for frictional
isolated bridges and tribological numerical model of frictional isolator,” 13
th
World Conference on Earthquake Engineering, Vancouver, Canada
Taylor, AW and Igusa, T (2004). Primer on seismic isolation, 1
st
Ed., American Society
of Civil Engineers, Virginia
Thakkar, SK and Maheshwari R (1995). “Study of seismic base isolation of bridge
considering soil structure interaction”, Third Int. Conf. on Recent Advances in
Geotechnical Earthquake Engineering and Soil Dynamics, Univ. of Missouri
Rolla, Rolla, Missouri, Vol. 1, 397400
Tsopelas, P and Constantinou, MC (1996). “Experimental study of FPS system in bridge
seismic isolation,” Earthquake Engineering and Structural Dynamics, 25, 6578
Turkington, DH, Cooke, N, Moss, PJ and Carr, AJ (1989). “Development of design
procedure for bridges on leadrubber bearings”, Engineering Structures, Vol. 11, 38
Tyler, RG (1991). “Rubber bearings in baseisolated structures – a summary paper,”
Bulletin of the New Zealand National society for Earthquake Engineering, 24,
251274
Tyler, RG and Robinson, WH (1984). “Highstrain tests on leadrubber bearings for
earthquake loadings,” Bulletin of the New Zealand National society for
Earthquake Engineering, 17(2), 90105
Vlassis, AG and Spyrakos, CC (2001). “SIB piers on shallow soil stratum with soil
structure interaction,” Computers and Structures, 79, 28472861
Warn, GP and Whittaker, AS (2004). “Performance estimates in SIB structures.”
Engineering Structures, 26, 12611278
Yashinsky, M and Karshenas, MJ (2003). Fundamentals of seismic protection for
bridges, 1
st
Ed., Earthquake Engineering Research Institute, California
181
Yashinsky, M (1998). “Performance of bridge seismic retrofits during Northridge
earthquake,” Journal of Bridge Engineering, 3(1), 114
Zayas, VA and Low, S (1999). “Seismic isolation for extreme cold temperatures,” 8
th
Canadian Conference on Earthquake Engineering, Canadian Association for
Earthquake Engineering, Vancouver, 391395
Zayas, VA and Low, SS, and Mahin, SA (1987). The fps earthquake resisting system
experimental report, Report No. EERC 198701, University of California,
Berkeley
182
VITA
Murat Eröz was born on November 3, 1980 in Susurluk, Turkey. He earned a
B.S. degree in Civil Engineering from Istanbul Technical University, Istanbul, Turkey in
2002. He earned an M.S. degree in 2003 from the Structural Engineering, Mechanics,
and Materials group of Civil Engineering, Georgia Institute of Technology. He continued
to pursue a Ph.D. degree in the same department. His major concentration was in the area
of earthquake engineering and seismic isolation of structures, with a minor in Structural
Mechanics.
ADVANCED MODELS FOR SLIDING SEISMIC ISOLATION AND APPLICATIONS FOR TYPICAL MULTISPAN HIGHWAY BRIDGES
Approved by: Dr. Reginald DesRoches, Advisor School of Civil and Environmental Engineering Georgia Institute of Technology Dr. Barry Goodno School of Civil and Environmental Engineering Georgia Institute of Technology Dr. Donald White School of Civil and Environmental Engineering Georgia Institute of Technology Date Approved: November 5, 2007 Dr. Laurence Jacobs School of Civil and Environmental Engineering Georgia Institute of Technology Dr. Jeffrey Streator School of Mechanical Engineering Georgia Institute of Technology
In loving memory of Russell and Wylene Sohn
Laurence Jacobs. Reginald DesRoches for the support.D. Dr. Jeffrey Streator for their time and effort in serving on my Ph. I would like to thank Dr. and Dr.ACKNOWLEDGEMENTS I would like to express my deepest gratitude to my advisor Dr. Barry Goodno. guidance and friendship he has extended. Donald White. I am grateful for the unconditional love and support of my parents. Their inspiration and guidance has made this dream a reality. Dr. committee. iv .
3.3 Bidirectional Coupling 2.1 Normal Force.4 Thesis Outline 2 CRITICAL REVIEW OF THE CURRENT STATEOFTHEART 2.TABLE OF CONTENTS Page ACKNOWLEDGEMENTS LIST OF TABLES LIST OF FIGURES SUMMARY CHAPTER 1 INTRODUCTION 1.4 Large Deformation Effects 2.3 Objectives and Scope 1.3. µ 2.2 Basics: Seismic Isolation for Bridges 2.2 Coefficient of Friction. N 2.2 Statement of the Problem 1.6 Critical Appraisal 1 1 4 5 6 8 8 8 10 15 21 24 25 26 26 29 31 iv ix x xvii v .4 Comparative Studies 2.3.3 The Friction Pendulum System (FPS) 2.5 Parametric Studies 2.3.1 Introduction 2.1 Background 1.3.5 Other Parameters 2.
8 Evaluation Platform 4.2.3 Normal Force 4.4.1 Introduction 4.2 Modeling and Analysis vi .9 Verification 4.4 Coefficient of Friction 4.3.4 Descriptions of common Isolators 3.2.3 The Lead Rubber Bearing (LRB) 3.7 Mathematical Model 4.5 Conclusion 33 33 33 38 39 40 41 41 42 44 45 46 48 49 49 49 52 53 54 57 58 62 64 64 68 4 FRICTION PENDULUM SYSTEM (FPS) MODELING 4.4.2 Procedure 2: Single Mode Spectral Method 3.1 The Eradiquake System (EDS) 3.3 Bridge Seismic Isolation in Design Codes 3.6 Large Deformation Moments 4.3.9. General Features of a Seismically Isolated Bridge (SIB) 3.1 Procedure 3: Multimode Spectral Method 3.1 Procedure 1: Uniform Load Method 3.2 The High Damping Rubber System (HDRS) 3.1 Procedure 4: TimeHistory Method 3.5 Bidirectional Coupling 4.3 SEISMIC ISOLATION FOR BRIDGES 3.9.3.3.4.1 Introduction 3. Simplified Isolator Response Modeling 4.1 Structural Properties and Loads 4.
The Lead Rubber Bearing (LRB) 6.1 Introduction 6.5 Bridge Model 6.3 Results 4.2.1 Introduction 136 vii .9 Conclusion 95 95 99 101 102 103 107 110 124 7 ASSESSMENT OF THE INFLUENCE OF THE DESIGN PARAMETERS ON THE RESPONSE OF BRIDGES ISOLATED WITH THE FRICTION PENDULUM SYSTEM (FPS) 136 7.3 Seismically Isolated Bridge (SIB) Modeling 5.9.7 Dynamic Analysis 6.6 Bridge Seismic Isolation 6.10 Conclusion 5 BRIDGE RESPONSE AS A FUNCTION OF ISOLATOR MODELING ASSUMPTIONS 5.4 Modeling of the Isolator Response 6.4.8 Results 6.3 ForceDeformation Characteristics of the LRB 6. Selection of the Class of Highway Bridges for Seismic Isolation and Analyses 76 5.4 FPS Models 5.1 Introduction 69 75 76 76 5.5 Dynamic Analyses 5.6 Results 5.2.7 Conclusions 6 77 80 82 85 93 COMPARATIVE ASSESSMENT OF SLIDING VERSUS ELASTOMERIC SEISMIC ISOLATION FOR TYPICAL MULTISPAN BRIDGES 95 6.
3 Results 7.2.4 Conclusion 8 SUMMARY.1 Analyses 7. CONCLUSIONS AND FUTURE RESEARCH 8.3.7.2 Results 7.1 Proposed Design 7.1 Summary and Conclusions 8.3.3.2.2 Analyses 7.2 Future Research APPENDIX A: METHOD OVERRIDING IN C++ APPENDIX B: REFERENCES VITA 126 127 129 140 142 147 149 152 155 155 159 161 REGRESSION LINES FOR THE EFFECTS OF BRIDGE DESIGN PARAMETERS ON SYSTEM RESPONSE 163 174 182 viii .2 Influence of Bridge Design Parameters 7.3 Influence of a Modified Seismic Isolation Strategy 7.
1: Ground motion suite Table 7.1: Summary of model properties Table 5.1: Parameter variation ranges Table 7.LIST OF TABLES Page Table 3.2: Vibration periods of the bridges as a function of design parameters 38 81 83 109 128 132 Table 7.2: Ground motion suite Table 6.1: Calculation of base shear Table 5.38 s 148 ix .3: Modified isolator design properties to achieve a fundamental period of T1=2.
3: Sources of normal force.1: Typical seismic isolation example in the American River Bridge at Lake Natoma in Folsom.4: Modal expansions of effective earthquake forces and modal static responses for the base 37 Figure 3.4: Response spectra for Gazli (1976) earthquake record 17 Figure 2.3: The Friction Pendulum System (FPS) (a) exterior view (b) internal components (c) interior elevation 4 Figure 2. and (d) liquefied natural gas (LNG) storage tanks in Greece (courtesy of Earthquake Protection Systems) 11 Figure 2. (b) KodiakNear Island Bridge. (c) Ataturk International Airport Terminal.5: Compression failure and bulging of piers along bent 3. (a) Rio Hondo Busway Bridge. Inc.3: Structural shapes for (a) Mode 1 (b) Mode 2 34 35 36 Figure 3.2 38 x . 1990) 21 22 Figure 2.1: Idealized seismicisolated bridge substructure Figure 3. Bull Creek Canyon Channel Bridge.1: Applications of FPS seismic isolation at (a) Bolu Viaduct. N. 2005) Figure 2. 1990) 23 Figure 3.2: Components of the FPS 14 Figure 2. California (courtesy of Earthquake Protection Systems.8: Dependency of the coefficient of friction on sliding velocity (Mokha et al.2: Idealized seismicisolated bridge substructure Figure 3. Photograph courtesy of Earthquake Engineering Research Institute (Papazoglou and Elnashai 1996) 18 Figure 2.6: Resilient sliding isolation (RSI) (Iemura et al.) 2 Figure 1.5: Pseudoacceleration design spectrum for AASHTO Type II soil profile with acceleration coefficient A=0.7: Test setup for Teflonsteel sliding surfaces (Mokha et al.2: Design spectrum and the shift of spectral ordinates for an isolated structure 3 Figure 1.LIST OF FIGURES Page Figure 1. fluctuations in bridge isolators (a) vertical inertial forces (b) lateral inertial forces (c) vibration forces due to traffic 16 Figure 2.
f µ .3: Forcedeformation characteristics of the unidirectional rigidplastic response of the FPS 51 Figure 4.12: Typical lead rubber bearing (LRB) (Taylor and Igusa 2004) 47 Figure 3.11: (a) High damping rubber bearing used in the earthquake simulator tests with dimensions in mm (b) corresponding forcedeformation hysteresis (Kikuchi and Aiken 1997) 46 Figure 3.8: Typical elastomeric isolator (Taylor and Igusa 2004) Figure 3.4: Forcedeformation characteristics of bilinear isolators Figure 4.1: The signum function 50 Figure 4. and (b) isolator contact pressure (Roussis and Constantinou 2006) 54 Figure 4. fR 50 Figure 4.Figure 3. (2004) 59 61 64 65 xi . and (b) pendulum.9: Schematic view of the model Figure 4.12: Test setup studied by Mosqueda et al.13: (a) Lead rubber bearing (LRB) used in the earthquake simulator tests with dimensions in mm (b) corresponding forcedeformation hysteresis (Kikuchi and Aiken 1997) 48 Figure 4.8: Deformed shape of the seismic isolator between the superstructure and the substructure with concave dish at the (a) bottom (b) top 58 Figure 4.2: The intrinsic response components (a) friction.11: FPSelement and FPSmaterial interaction Figure 4.10: The Eradiquake seismic isolation bearing 40 42 43 44 45 Figure 3.9: Clantarient`s base isolation system using a layer of talc as the isolating medium (Naeim and Kelly 1996) Figure 3.7: Frictional interaction surface (a) uncoupled (b) coupled response 56 Figure 4.6: Variation of the coefficient of friction with (a) velocity of sliding.10: Deflections and forces acting on the slider Figure 4.7: Characteristics of bilinear isolation bearings per AASHTO Guide Specifications Figure 3.6: Single substructure and isolator idealization Figure 3.5: Gap element forcedeformation model 52 53 Figure 4.
20: Comparison of the response of the small deformations model (SDM) and the large deformations model (LDM) 73 Figure 4.5: Mode shapes of the deck Figure 5.and µ =fmin for loading L1b 72 Figure 4. ± 17. and coupled model 74 Figure 4.0 cm (L1c) 67 Figure 4. and (b) concrete material Figure 5.22: Comparison of the bidirectional resisting forces for the coupled model and the uncoupled models 74 Figure 5.15: Combined bidirectional loading path for L2 68 Figure 4.19: Comparison of the forcedeformation histories of the FPS with theoretically exact value of the friction coefficient.4: Constitutive relationships for the modeling of (a) steel material.2: Multispan continuous (MSC) steel girder bridge general elevation and modeling details Figure 5.13: Scaling of fmax with respect to the relationship presented in Constantinou et al.3: Pier configuration and bent and column discretization Figure 5.16: Finite element model of the test setup studied by Mosqueda et al. µ .21: Comparison of the bidirectional resisting forces for the experimental data. µ =fmax.7: Orientation of the 3D bridge model 77 79 79 80 82 83 84 Figure 5. µ .6: Response spectrums for the suite of ground motions Figure 5. ± 12. and ± 45.8 cm (L1b).14: Unidirectional load histories with amplitude.7 cm (L1a). (2004) 69 Figure 4.17: Comparison of the normalized forcedeformation histories between the model and the experimental results for (a) individual Isolator 3. and (b) total isolator forces 70 Figure 4.8: Time history of the N/No for the FPS during the Nahanni earthquake NLTH analysis 86 xii .18: Friction coefficient. (1993) 66 Figure 4. time history under loading L1b 71 Figure 4.1: Photo of example MSC Steel Girder Highway Bridge (Nielson 2005) Figure 5.Figure 4.
and Model 4 87 Figure 5. (b) MND. MNDpier.2: The Lead Rubber Bearing (LRB) interior elevation Figure 6.6: Variation of isolator yield force. as a function of applied compressive axial load. as a function of applied compressive axial load. δ.9: Bilinear idealizations of the LRB forcedeformation characteristics after gravity loading in the bridge 107 Figure 6.12: Forcedeformation history of the FPS in the longitudinal direction on top of the pier for the Northridge earthquake record where the vertical component. (b) Model 2. as a function of isolator inplane deformation. for the LRB 98 98 102 104 Figure 6. (b) MND. (b) MND on top of the pier. effect is (a) included (b) not included 111 xiii . Fy. (b) Patria Acueducto.8: Bilinear idealizations of the FPS forcedeformation characteristics after gravity loading in the bridge 106 Figure 6.12: Comparison of the (a) total MNF transferred to the pier.Figure 5.3: Effects of geometrical variations of the LRB on the forcedeformation response (Priestley et al. and (c) dmax 89 Figure 5. and abutments. ΣMNFabutment. && u (t ) v . 1996) Figure 6.7: Variation of isolator postyield stiffness. Califonia.11: The influence of constant value of µ assumptions on (a) MNF. Mexico (courtesy of Dynamic Isolation Systems) 97 Figure 6. and the total MNF transferred to the abutment. Palm Springs earthquake record with (a) Model 1. and (c) dmax 91 Figure 5. N 105 Figure 6.10: Mode shapes of the bridge deck from plan Figure 6. ΣMNFpier.5: Variation of the buckling load.4: Isolator model Figure 6. (c) Model 3.10: The influence of modeling assumptions on (a) MNF. MNDabutments 92 Figure 6. N 116 Figure 6.9: Forcedeformation history of the FPS in the longitudinal directions on top of the pier for the N. kp.11: Response spectrums for the suite of ground motions 108 109 Figure 6.1: Examples of LRB applications (a) Rio Vista Bridge. Pcr.
u (t ) v . Pcr. effect is (a) included (b) not included 121 Figure 6. (b) MID on top of the pier.18: Forcedeformation history of the FPS and LRB for the Helena earthquake (a).20: Total energy dissipated for by the isolators (a) on top of the pier (b) on top of the abutment 119 Figure 6. u (t ) v .19: Comparison of the (a) total MNF transferred to the pier. effect is (a) included (b) not included 121 Figure 6.24: Deformation histories for the Morgan Hill earthquake of (a). u (t ) v .(b) on top of the pier and (c).15: Time history of the LRB buckling load. a’response quantity. u (t ) v .13: Time history of the N/No for the FPS during the Northridge earthquake record 111 Figure 6.(d) on top of the abutment 127 Figure 6.1: Bridge design parameters 128 Figure 7. (d). (c). (e) FPS and (b).22: Maximum isolator deformations (MID) for the suite of ground motions on && top of the pier with vertical component. and normal force.21: Maximum isolator forces (MIF) for the suite of ground motions on top of the && pier with vertical component.16: Forcedeformation history of the LRB on top of the pier for the Gazli earthquake in the (a). ΣMNFabutment.Figure 6. effect is (a) included (b) not included 112 Figure 6. u (t ) v .’ for the regression lines of the median design parameters 131 xiv .14: Forcedeformation history of the LRB in the longitudinal direction on top of the pier for the Whittier Narrows earthquake record where the vertical && component.3: Slopes. for the suite of ground motions on top of the && pier with vertical component. MIDpier. (c) longitudinal. ΣMNFpier.23: Maximum column drifts. and (b). and abutments. (d) transverse directions 113 114 Figure 6.2: Base conditions for the substructure modeled as (v) pinned (b) partiallyfixed (c) fixed 128 Figure 7. (f) LRB located on top of the pier 123 Figure 7. dmax. MIDabutments 118 Figure 6.17: Forcedeformation history of the FPS in the longitudinal direction on top of the pier for the Whittier Narrows earthquake record where the vertical && component. and the total MNF transferred to the abutment. N Figure 6. effect is (a) included (b) not included 125 Figure 6. effect is (a) included (b) not included 132 Figure 6.
8: Structural bridge responses as a function of base modeling assumptions 139 Figure 7. on (a) MNF (b) MND (c) dmax (d) ΣMNFpier / ΣMNFabutment (e) MNDpier / MNDabutment 166 xv . (c) timehistory of the longitudinal column tip deformations 133 Figure 7.75 (b) Ld*/Ldo =1.2: The influence of the column length.15: Modified isolator design modeling at the abutments Figure 7. (c). (c). fc.17: Forcedeformation history of the FPS in the longitudinal direction for the Helena earthquake (a).5: Forcedeformation history of the FPS in the longitudinal direction on top of the pier for the Helena earthquake record with (a) Ld*/Ldo=0.75 135 Figure 7. (c) on top of the pier.6. (d) pinned base conditions 140 Figure 7.16: Variation of bridge response quantities as a function of FPS design parameters 146 146 147 150 Figure 7. on (a) MNF (b) MND (c) dmax (d) ΣMNFpier / ΣMNFabutment (e) MNDpier / MNDabutment 164 Figure B.75 134 Figure 7.4: Forcedeformation history of the FPS in the longitudinal direction on top of the pier for the Whittier Narrows earthquake record with (a) Lc/Lco=0.12: Sample spring element to be used in the proposed modified design of the FPS at the abutments 145 Figure 7.1: The influence of the pier concrete compressive strength. (d) new design with R = 621 cm.10: Prestressing the FPS (Kasalanati and Constantinou 2005) Figure 7. Lc.Figure 7.13: Displacementcontrolled load history Figure 7.6: Forcedeformation history of the FPS in the longitudinal direction on top of the pier for the Northridge earthquake record with (a) Ld/Ldo=0.11: Proposed modified design for the FPS above the abutments in (a) undeformed (b) deformed state 142 144 Figure 7.75 (b) Ld/Ldo =1. (a).8 (b) Lc/Lco=1. (b) fixed base conditions. (b) conventional design with R = 99 cm. (b).14: Forcedeformation response of the isolator with modified design Figure 7.7: Median values of the response quantities of the bridge as a function of base modeling assumptions 138 Figure 7. (d)on top of the abutment 152 Figure B.9: Forcedeformation history of the FPS in the longitudinal and transverse directions on top of the pier for the Loma Prieta earthquake record with (a).
α. Ld. on (a) MNF (b) MND (c) dmax (d) ΣMNFpier / ΣMNFabutment (e) MNDpier / MNDabutment 172 xvi .5: The influence of skew angle.3: The influence of longitudinal deck length. with constant mass on (a) MNF (b) MND (c) dmax (d) ΣMNFpier / ΣMNFabutment (e) MNDpier / MNDabutment 168 Figure B.Figure B.4: The influence of longitudinal deck length. Ld. with adjusted mass on (a) MNF (b) MND (c) dmax (d) ΣMNFpier / ΣMNFabutment (e) MNDpier / MNDabutment 170 Figure B.
large deformation effects. One of the emerging tools for protecting bridges from the damaging effects of earthquakes is the use of isolation systems. in part due to the lack of adequate models that can account for the complex behavior of the isolators. Seismic isolation is achieved via inserting flexible isolator elements into the bridge that shift the vibration period and increase energy dissipation. Unlike previous models.e. The comparative viability of the two main isolator types (i. This study investigates and makes recommendations on the structural performance of bridges utilizing sliding type seismic isolators. sliding and elastomeric) for bridges is investigated. To date. these models can account simultaneously for the variation in the normal force and friction coefficient.SUMMARY The large number of bridge collapses that have occurred in recent earthquakes has exposed the vulnerabilities in existing bridges. the structural performance of bridges incorporating sliding seismic isolation is not wellunderstood. xvii . The influence of bridge and sliding isolator design parameters on the system’s seismic response is illustrated. The intention is to provide support for seismic risk mitigation and insight for the analysis and design of seismically isolated bridges by quantifying response characteristics. and the coupling of the vertical and horizontal response during motion. The level of accuracy required for isolator analytical models used in typical highway bridges are assessed. based on the development of stateoftheart analytical models.
Seismic bridge design focused on increasing the lateral strength to resist inertial forces that occurred from ground shaking (Yashinsky and Karshenas 2003).CHAPTER I INTRODUCTION 1.1).1 Background Bridges are a crucial part of the overall transportation system and their performance during earthquakes is important for continued functioning of a community.S. A series of revisions to the design guidelines were accompanied by the launching of substantial retrofitting programs following the 1971 San Fernando earthquake. 1996). The large number of bridge collapses during the 1971 San Fernando earthquake exposed the deficiencies of the 1965 AASHTO and previous bridge design codes.000 bridges in the United States with approximately 60% having been constructed prior to 1970 with little or no consideration given to seismic resistance (Cooper et al. casting of infill walls between columns. only incorporated individualized strengthening schemes such as steel jacketing of columns. Until the mid1980s. The root cause of the damaging effects of earthquakes is the unfortunate correlation between the fundamental periods of vibration of major structures and the frequency content of the seismic input (Priestley et al. bridge retrofitting techniques in the U. and the use of restraining cables (Yashinsky 1998). However.1994). There are approximately 575. since strengthened members attracted larger forces that severely damaged other elements along the bridge’s lateral load path. widening of the pier caps and abutments. strengthening of footings and bearing elements. 1 . Seismic isolation decouples the structure from the horizontal components of the ground motion with elements that have a low horizontal stiffness (Naeim and Kelly 1996) (Figure 1. this increase was found to be selfdefeating.
Isolation shifts the response of the structure to a higher fundamental period and increases the damping.Figure 1.).1 Typical seismic isolation example in the American River Bridge at Lake Natoma in Folsom. Inc.2. California (courtesy of Earthquake Protection Systems. 2 . The philosophy of seismic isolation for improving earthquake resistance of a structure departs from conventional retrofit measures because the latter attempts to strengthen individual elements of bridges while the former improves structural performance by reducing the overall earthquake forces that the structure acquires. thus reducing the corresponding pseudoacceleration in the design spectrum and attracting smaller earthquakeinduced forces. as illustrated in Figure 1.
3 . AASHTO 1999). two isolator types that are representative of sliding and elastomeric systems are the Friction Pendulum System (FPS) and the LeadRubber Bearings (LRB). Among others. Various isolators have been manufactured with the similar objective of providing a period shift and additional energy dissipation to structures. respectively. The supported structure is administered into a pendulum motion as the isolators simultaneously glide on its concave surfaces and dissipate hysteretic energy via these frictional surfaces (Dicleli and Mansour 2003). The mechanism of the FPS is primarily based on its concave geometry and the surface properties. Isolators can be classified as sliding and elastomeric (Taylor and Igusa 2004.Figure 1. This study focuses on the seismic response of bridges with emphasis on the FPS.2 Design spectrum and the shift of spectral ordinates for an isolated structure.
However. Clark et al. 1999). Furthermore. By this time elastomeric bearings were primarily used in bridge seismic isolation (Stanton 1998). As new isolator types became available by 1995. 2006). 4 .S. 1.3 The Friction Pendulum System (FPS) (a) exterior view (b) internal components (c) interior elevation.(a) (b) (c) Figure 1. is a relatively new phenomenon that was addressed by the AASHTO with the Seismic Isolation Guide Specification for the first time in 1991. Naeim and Kelly 1996. the widespread application of this technology is still impeded by overconservative attitudes (Mayes 2002. A better understanding of the impact of isolators on the seismic behavior of bridge response is necessary. the first Seismic Isolation Guide Specification was essentially rewritten in 1997 to address the advances in the industry. sliding seismic isolators make up less than 25% of the total number of isolated bridges in North America (Buckle et al. Despite recent advances in base isolation research. the Specification still does not provide guidance about selecting the optimal isolator type for different bridge applications. The responses of state bridge engineers on a survey asking why base isolation is not more widely used revealed that engineers are not comfortable with seismic isolation because they view it as a black box and that there is a lack of certainty on choosing the optimum type of seismic isolation (Mayes 2002).2 Statement of the Problem Bridge seismic isolation in the U.
questions such as the relative benefits of different isolator systems for different bridge types have yet to be adequately addressed. isolator model. since isolators have a highly nonlinear response involving the simultaneous action of multiple components. This thesis is aimed at addressing some of these issues. with a particular emphasis on the modeling parameters of the isolators which govern the seismic response of typical bridges. This is accomplished by developing rigorous analytical models of isolators with particular emphasis on the FPS and using these models to investigate the response of SIBs.The individual response of isolators is well understood but there are limited studies investigating the response of seismically isolated bridges (SIBs) via detailed threedimensional (3D) models. Develop the nonlinear kinematics formulation of the 5 . The intention is to provide support for seismic risk mitigation and insight for the analysis and design of SIBs by quantifying response characteristics. The influence of the level of accuracy in the modeling assumptions of isolators for bridges has received narrow attention. it is not clear what parameters in a typical bridge govern the effectiveness of various isolation systems.3 Objectives and Scope The objectives of this study are to assess the performance of bridges seismically isolated with the FPS. The research tasks to accomplish these objectives are the following: • Identify the characteristic aspects of the FPS that contribute to the forcedeformation response. 1. In particular. detailed models are needed to capture the intricate behavior of these highly nonlinear elements. Finally. • Implement the model into a nonlinear dynamic evaluation platform and validate response using experimental data. Also.
• Modify the FPS model to represent the LeadRubber Bearings (LRB) forcedeformation response. and parametric analyses. Particular emphasis is given to the modeling aspects of the FPS. If applicable. and isolator design parameters on the system’s response. Chapter 4 explains the development of the FPS isolator model. Chapter 3 presents the application of seismic isolation into bridges. An overview of common isolators is presented. • Compare and quantify the response of bridges as a function of isolator type with emphasis on FPS and LRB. A critical assessment of the currentstateoftheart is presented. dynamics of bridge seismic isolation is explained on a simplified model. propose modifications for design of the isolator to improve bridge seismic performance. • Investigate parametrically the influence of bridge geometric and material properties.4 Thesis Outline The content of the dissertation is organized into the following chapters: Chapter 2 provides a literature review on bridge seismic isolation. The model is incorporated into an open source finite element platform. The Analysis methods in current Guide Specifications are outlined. comparative studies among different isolators. 1. Modeling aspects of the FPS are highlighted and the nonlinear kinematics of the response is generated.• Develop detailed 3D bridge models isolated with the FPS and identify the influence of the modeling assumptions of the isolator on the response of the bridge. validated using experimental data. structural sensitivity to different modeling parameters. The model is 6 .
Chapter 8 presents a summary of the research. 7 . An LRB model modified from the previous FPS model is developed. and recommendation for future research. The response of the bridge is monitored as a function of varying design parameters. The responses of two bridge models isolated with each isolator are examined under timehistory analyses. major conclusions drawn from this study. Chapter 6 compares the performance of a typical bridge with elastomeric isolation versus sliding isolation. Friction Pendulum System (FPS) models developed with different assumptions are incorporated into bridges and subjected to timehistory analyses. Chapter 7 provides a parametric investigation on the seismic response of an FPSisolated bridge.Chapter 5 examines the response of a three dimensional (3D) SIB as a function of FPS modeling assumptions. The differences of the FPS and LRB response are highlighted.
2 Seismic Isolation for Bridges: Basics Recent earthquakes have illustrated the vulnerability of bridges to damage and collapse (Cooper and Friedland 1994. One of the emerging tools for protecting bridges from the damaging effects of earthquakes is the use of seismic isolation systems. The basics and historical development of seismic isolation is outlined. An insightful definition of ‘seismic isolation’ given by Skinner et al. are reviewed. Experimental and analytical research conducted on the response characteristics of the FPS is elaborated. 2. Analytical research aimed at comparing the two main classes of isolators. from potentially damaging earthquake 8 . and/or its contents. Parametric studies conducted on isolator and bridge design properties are summarized.1 Introduction The introduction of seismic isolation as a practical tool has provided a rich source of experimental and theoretical work both in the dynamics of the isolated structural systems and in the mechanics of the isolators themselves. This chapter presents a summary of the previous studies that address the modeling and analysis aspects of bridge seismic isolation with particular emphasis on the Friction Pendulum System (FPS) isolator. Yashinsky 1998). (1993) is as follows: ‘Seismic isolation consists essentially of the installation of mechanisms which decouple the structure. the sliding and the elastomeric.CHAPTER 2 CRITICAL REVIEW OF THE CURRENT STATEOFTHEART 2. A critical appraisal of the current literature is presented.
S.S. This decoupling is achieved by increasing the flexibility of the system. and deck (Priestley et al. Seismic isolators are typically installed between piers.S. AASHTO 1999. was the Foothill Communities Law and Justice Center in 19841985 in California. motions. FEMA 356). The fundamental concept of base isolation was first studied on an example building on balls by Professor John Milne who was a faculty member in the Mining Engineering Department of Tokyo University between 1876 and 1895 (Naeim and Kelly 1996). only in the last four decades has industrial capabilities enabled the manufacturing of practical isolation devices. Currently seismic isolation is wellintegrated into the code provisions in the U. Taylor and Igusa (2004) identified the distinct properties of the structural response of bridges from that of buildings as the following: 9 . The first building that employed a rubber isolation system was a school at Skopje. The first seismically isolated building in the U. provisions developed for seismic isolation of bridges are unique due to fundamental differences in the structural response of bridges compared to buildings.induced.A. isolation systems are most commonly classified as elastomeric and sliding. which was located only 19. Although patents for seismic isolation schemes were obtained as early as 130 years ago. together with providing appropriate damping. Currently. Yugoslavia in 1969 (Naeim and Kelly 1996). The California Department of Transportation (Caltrans) was the first U. 1996). for both buildings and bridges (International 2003.3 km west of the San Andreas Fault (Taylor and Igusa 2004).’ The two fundamental structural improvements provided by seismic isolation is the reduction of lateral forces and the concentration of lateral displacements at the isolation interface (Taylor and Igusa 2004). However. NEHRP 2003. abutments. ground or support. transportation agency to use seismic isolation on a bridge at the Sierra Point Overlook in 1985 (Taylor and Igusa 2004). and only in the last decade has seismically isolated structural design been widely adopted (Taylor and Igusa 2004).
American River Bridge at Lake Natoma in Folsom). 10 . bridges (BeniciaMartinez Bridge in the San Francisco Bay Area. 1987). Inc. salt water. the U. Since than. 5.S. Bridges are supported by flexible piers while buildings often have relatively rigid foundations 3. Bridges are typically more flexible in the vertical direction compared to buildings because of long spans. sun light. bridge isolators are subjected to more severe routine live load conditions than those observed in buildings and may be more exposed to environmental conditions such as. and debris compared building isolators located at the foundation levels. Spatial variations in the ground motion may become important because bridges are long in one direction 2. 2. The flexibility of the substructure may change significantly in bridges due to the need of accommodating variations in terrain 4.3 The Friction Pendulum System (FPS) The Friction Pendulum System (FPS) is a sliding type seismic isolator that was developed in 1986 by Earthquake Protection Systems. The FPS was first used to retrofit an apartment building in California in 1989 (Naeim and Kelly 1996). the FPS have been used to isolate buildings (Washington State Emergency Operations Center at Camp Murray. rain. In addition to these factors. Court of Appeals Building in San Francisco).1). International 2000). freezing. (Zayas et al. and storage tanks (LNG storage tanks on Revithoussa Island near Athens) (Jangid 2005) (Figure 2. The FPS has been incorporated into the design codes (AASHTO 1999. This makes vertical ground motions important for bridges. The philosophy in building isolation is to limit the forces in the superstructure while the primary concern in bridges is typically the substructure.1.
(a) Figure 2. 11 .1 Applications of FPS seismic isolation at (a) Bolu Viaduct. (c) Ataturk International Airport Terminal. (a) Rio Hondo Busway Bridge. (b) KodiakNear Island Bridge. and (d) liquefied natural gas (LNG) storage tanks in Greece (courtesy of Earthquake Protection Systems).
(b) (c) Figure 2.1 continued 12 .
1 continued 13 .(d) (e) Figure 2.
Figure 2. the friction force opposes the motion of the slider and dissipates hysteretic energy. Findings of previous research provide ample evidence that the dynamic response of seismically isolated structures is governed by the characteristics of the mechanisms of the isolators (Dicleli 2002). 1987): 14 . 1987) (Figure 2.2 Components of the FPS. a continuous recentering force is provided by the gravity load of the supported mass. Simultaneously.2).The FPS consists of a spherical stainless steel surface. an articulated slider and a housing plate (Zayas et al. The radius of the FPS isolator controls the concavity related stiffness and the isolation period of the structure (Naeim and Kelly 1996). This is an indication that the modeling assumptions adopted for the response of the FPS will affect the estimated response quantities of SIBs. The sliding surface of the FPS consists of stainless steel and a Teflonbased custom material. The forcedeformation response of the FPS is typically modeled using a unidirectional idealization (Zayas et al. As the slider displaces over the concave surface.
(2) rocking behavior of the girder due to horizontal ground motion. Normal Force. (2000) identified three sources for the fluctuation of N in the isolators of bridges: (1) inertial forces due to vertical ground motions. However. and (4) isolator deformations are small and planar. (2) µ is constant. Takashi et al. δ is the sliding & & deformation.1. δ is the sliding velocity. 15 . N The restoring mechanism of the FPS is dependent on the normal force. µ is the friction coefficient between the sliding surfaces. (3) the horizontal response is uncoupled in the orthogonal directions. it is important to underline the fundamental assumptions inherent in this equation: (1) N is constant. R is the radius of the concave surface. The forcedeformation response of the FPS is further elaborated in Chapter 4. respectively.3). 2. and sgn(δ ) is the signum function.1) where N is the normal force acting on the sliding surface. N (see Equation 1). The signum & function is equal to +1 or 1 depending on whether δ is negative or positive.6 74 4µ 8 } & N f = N µsgn(δ ) + δ R f fR (2. The following sections describe theoretical and experimental research performed to quantify the influence of these simplifications.3. and (3) deflection vibration of the girder due to vertical ground motion service loads and curvature (Figure 2.
This is illustrated in Figure 2. high frequency band which may inflict considerable structural damage (Collier and Elnashai 2001). 2002). (2002) reported that the vertical component of the earthquake may be consequential for bridges located within 60 km of the source.3 Sources of normal force. the energy content is concentrated in a narrow.(a) (b) (c) Figure 2. 16 . There is a general tendency among bridge engineers to neglect the effects of vertical ground motions on the structural response (Button et al. Although the vertical components of ground motions typically have lower energy content compared to the horizontal components.4 for the Gazli record taken 3 km from the source. Button et al. For nearfield earthquakes (<1015 km). fluctuations in bridge isolators (a) vertical inertial forces (b) lateral inertial forces (c) vibration forces due to traffic. N. the vertical spectra may significantly exceed the horizontal spectra for short periods (Silva 1997).
UP Component. It was shown that the design values for dynamic response increases for bridge members with the inclusion of the vertical components of the ground motions (Gloyd 1997.50 0.7g.25 1.00 2.75 1. However.50 1. 17 .25 2.5).50 2. 00 Component.Spectral acceleration. T (s) Component. 90 Figure 2.00 1. structural damage was noted in bridge members only at relatively high magnitudes of accelerations in the vertical direction.75 2. For example. Sa (g) for ξ=5% 5 4 3 2 1 0 0. Yu 1996).25 0. Saadeghvaziri (1991) concluded that considerable damage would occur in columns if the vertical component of the ground motions exceeded 0.4 Response spectra for Gazli (1976) earthquake record.00 0.00 Period.75 3. Serious substructure damage was observed in bridges during the Northridge and Kobe earthquakes (Papazoglou and Elnashai 1996) (Figure 2.
it did not have considerable influence on the overall response. Takashi et al. However.5. and concluded that the effect of vertical ground motion on the overall response was trivial. Bull Creek Canyon Channel Bridge. (2000) performed shaking table tests on a girder model supported by a set of four frictional isolators and two rubber buffers to assess the dependency of the structural response to vertical and horizontal excitations. The aforementioned studies focus on nonisolated bridges. The authors reported that although the rocking of the structure altered the normal force and behavior of individual isolators. Takashi et al. The authors have imposed vertical ground motion five times higher than that of initial test input to assess the effects of extremely large vertical 18 . they are normally not isolated from vertical earthquake motions (Taylor and Igusa 2004). theoretical and experimental evidence suggests that the horizontal response of the FPS is coupled with the vertical response. Photograph courtesy of Earthquake Engineering Research Institute (Papazoglou and Elnashai 1996). Compression failure and bulging of piers along bent 3. Since structures are inherently stronger and stiffer in the vertical direction compared to the horizontal directions.Figure 2. mentioned that the sum of the areas of the hysteretic loops for cases with and without vertical ground motion was acutely similar.
ground motion on the response. Takashi et al. reported that a loss of normal force on the isolators emanated, but the overall horizontal response did not languish even under extremely large vertical ground motions. Nakajima et al. (2004) performed pseudodynamic tests to verify the effect of vertical motions on the response of sliding isolators. The experimental setup consisted of a sliding bearing that generated friction damping and of a rubber bearing that provided restoring force. The authors concluded that although the friction coefficient of the
bearings alternated as a function of the bearing pressure, the effect of vertical motion was inconsequential in the overall response. It was observed that the friction coefficient increased proportionally with the sliding velocity and virtually flattened after exceeding a certain velocity. The authors stated that the friction coefficient decreased linearly with increasing bearing pressure. Mosqueda et al. (2004) performed unidirectional and tridirectional tests on a rigidblock frame supported by four FPS. The authors concluded from the results of the tridirectional tests that the vertical component of the ground motion had negligible effect on the forcedeformation response of the FPS. However, the authors noted that rotation of the superstructure in bridges caused by lateral ground motions could significantly influence the behavior of FPS. Almazan et al. (1998) generated the exact analytical equations of motions for the FPS, which considered the large deformation effects. The authors analyzed fourstory building models under horizontal and vertical ground motion. It was concluded that the global response of the structure might be estimated within the vicinity of 20% error if vertical motions are disregarded. The authors further mentioned that uplift occurred at several instants of the response and column base shears were 3 times larger when vertical motion is considered. Dicleli (2002) performed modal and nonlinear timehistory analyses on a six span slabongirder deck isolated with the FPS. Analysis results showed that the first modes
19
of vibrations were those involving the isolation system, with Modes 1, 2 and 4 being the transverse modes of vibration and mode 3 being the longitudinal. In the longitudinal direction, all isolator were found to have a uniform displacement due to the large axial rigidity of the deck in this direction. However, in the transverse direction, bearing displacements varied along the bridge caused by the unusual flexibility of the bridge in his direction. Furthermore, the dead load reactions at the abutments were only 20% of those at the piers due to smaller tributary weight of shorter end spans and uplift reactions created by much longer adjacent spans. Since the lateral resistance of the FPS is directly proportional to the dead load reactions acting on the bearings, a very small equivalent stiffness was obtained at the abutments. This was found to produce even larger bearing displacements as the seismically induced forces acquired the shape of the deflected structure. Although the abutments were structurally stronger than the piers due to their massive size, only 7% of the total seismic force was transferred to the abutments. It was concluded that hybrid isolation of the bridge using FPS and laminated elastomeric bearings produced results that are more favorable than the bridge isolated with FPS alone. Iemura et al. (2005) performed shake table tests on scaled models of two highway bridges seismically isolated with a combined rubber and sliding bearings, a system referred as ‘resilient sliding isolation (RSI)’ (Figure 2.6). The objective was to quantify the effects of the vertical accelerations and rocking of the deck on sliding isolators. The authors concluded that the rocking motion had considerable effect on the individual response of the RSI however this effect was dampened in the response of the total system.
20
Figure 2.6 Resilient sliding isolation (RSI) (Iemura et al. 2005).
2.3.2. Coefficient of Friction, µ Sliding isolators typically employ interfaces of steel and Polytetrafluoroethylene (PTFE or Teflon) (Mokha et al. 1991). Teflon is extremely resistant to attack by corrosive reagents or solvents (Billmeyer 1984). Furthermore, this polymer is not hard, but is slippery and waxy to touch, and has very low coefficient of friction on most substances. For all practical purposes the polymer is completely unaffected by water. Its thermal stability is such that its mechanical properties do not change for long intervals (months) at temperatures as high as 250 o C . Resistance to wear and to deformation under load, stiffness, and compressive strength of Teflon can be enhanced by the use of different fillers such as glass fibers, graphite, carbon and bronze. The sliding of the two surfaces of the FPS is an integral part of the forcedeformation response. Mokha et al. (1990) underlined the absence of experimental data on the sliding response of Teflon surfaces for velocities that are of interest to seismic isolation bearings. The authors performed experiments to investigate the characteristics of the steelTeflon sliding surfaces (Figure 2.7). The following conclusions were made from the test results: 1) The coefficient of friction increases and reaches a flat plateau beyond a certain point with increasing velocity (Figure 2.8).
21
2) Sliding initiates at initial motion of the isolator and motion reversals (stickslip). 4) The effect of the dwell of the load. 1990). 3) The magnitude of the friction coefficient reduced with increasing normal force acting on the plane of the sliding motion. acceleration of the ground motion. The magnitude of the static friction coefficient attained at these instances is substantially larger than the magnitude during sliding. and the specimen size is negligible to the magnitude of the coefficient of friction Figure 2. 22 .7 Test setup for Teflonsteel sliding surfaces (Mokha et al.
54 cm/s. Mosqueda et al. 1990).9 MPa 17. (2004) examined the effect of sliding velocity on the coefficient of friction by tests performed on four FPS isolators under a rigid frame that was subjected to bidirectional load histories. which was earlier than 12.00 10. (1990).4 cm/s.00 40.8 Dependency of the coefficient of friction on sliding velocity (Mokha et al.00 50.00 60. and the deposition of the composite material on the stainless steel surface. The authors concluded this steadiness favorable for describing the coefficient of friction as a constant value postulated by the Coulomb’s law of friction.7 cm/s reported by Mokha et al. The authors associated the reduction of the friction coefficient at high velocities with the escalation of the temperature.00 Sliding velocity (cm/s) Figure 2.00 70.00 20. (1990) with the exception of a slight decrease after 25.3 MPa 44. The velocity dependence of the coefficient of friction was similar to the one depicted by Mokha et al.6.00 30. Test results showed that the steady state response of the friction coefficient started to develop after 2. 23 .9 MPa 15 Coefficient of friction (%) 12 9 6 3 0 0.
and 0.Jangid (2004) performed a parametric study to ascertain the effect of the friction coefficient of FPS on the seismic response of buildings and bridges to nearfault ground motions. Jangid (2004) concluded from the analyses that there exists an optimum value of the coefficient of friction for the FPS that reduces isolator displacements. which simultaneously minimizes the superstructure acceleration. Similar theoretical and experimental research performed with the FPS confirmed these findings.05 to 0. The AASHTO Guide Specifications (1999) provide a method for estimating effects of bidirectional input by combining 100% plus 30% of orthogonal maxima. Warn and Whittaker (2004) performed nonlinear timehistory analyses on an FPSisolated 24 .3 Bidirectional Coupling Available earthquake records indicate that the horizontal ground motions are two dimensional (PEER 2000). The authors stated that omitting the variation of the friction coefficient as a function of the velocity in the modeling process caused inconsistency with experimental results. Nakajima et al. Mokha et al. (1993) showed that neglecting the orthogonal coupling of the steelTeflon interfaces in models results in underestimation of displacements and overestimation of forces.3.07 and 0. Theoretical studies have confirmed that there exists significant coupling between the orthogonal components of the response in structures that extend into the nonlinear range.15 for building isolators where the nearfault ground motions are expected.19 for bridge isolators. (2000) analyzed a one degreeoffreedom model with a natural period equivalent to their test setup. The author analyzed a multistorey building model and a three span continuous deck bridge model under near fault ground motions. The bridge model revealed similar results to those obtained for buildings which implied the presence an optimum value for the friction coefficient of the FPS that minimizes pier base shear and deck accelerations. 2. The author suggested the use of coefficient of friction values between 0.
the orthogonal component of excitation and the coupled behavior of seismic isolators. The authors stated that the bearings of fullscale bridges abide a chaotic bidirectional path that endures lower temperatures and consequently higher friction coefficients. The authors reported that large deformation effects in the FPS was influential on individual isolator response rather than the substructure response. Anderson and Mahin (2004) analyzed generalized bridge models to asses the accuracy of the AASHTO Guide Specifications (1999) method to account for bidirectional response effects and reached similar conclusions with Warn and Whittaker (2004). Mosqueda et al.bridge model to quantify the effect of bidirectional excitation and reviewed the accuracy of the AASHTO Guide Specifications (1999) estimating maximum isolator displacements.3.4 Large Deformation Effects The large deformation aspects pertaining to FPS seismic isolation have been theoretically addressed by Almazan and Llera (2002). The authors concluded that bidirectional excitation produces significantly larger isolator displacements than unidirectional excitation. namely. 2. The authors underlined that unidirectional tests overestimate the reduction of the friction coefficient because the friction is computed from the motion over an invariable path that accumulates the temperature. The authors associated these results with two factors. 25 . The authors later cast the nonlinear kinematic equations of the FPS response into an element format (Almazan and LLera 2003). It was also noted that the orientation of the FPS was a controlling parameter in transferring large deformation moments to the substructure or superstructure. It was concluded that the AASHTO Guide Specifications (1999) procedures underestimate the isolator deformations. (2004) performed unidirectional and tridirectional tests on a rigidblock frame supported by four FPS and examined how different mathematical models conjectured the response of these bearings under different excitations.
but at a different rate and form (Almazan et al. Naeim and Kelly 1996). 1998. The conventional FPS is essentially rigid under compression and has no tensile load capacity while the LRB has relatively less compression stiffness and able to resist a limited amount of tensile loading (Kelly 2003. there exist considerable differences in the vertical response characteristics of elastomeric and sliding isolators.3. Fatigue tests performed by 10. However.4 Comparative Studies Seismic isolators serve the common objective of decoupling the structure from the horizontal components of the ground motion to minimize the seismic loads on the loadcarrying components. The LeadRubber Bearing is a widely used elastomeric isolator (Buckle and Mayes 1990). Normal forcedependent FPS models have been developed previously to show that this effect may result in considerable variation on the estimated isolator response (Almazan et al. The performance of the isolators did not change at 49o C and − 40o C . Ryan and Chopra 2003).5 Other Parameters The dynamic performance characteristics of FPS. The details of the LRB characteristics are elaborated in Chapters 3 and 6. Test results showed that the FPS was mildly frequency dependent. However. LRB models that account for bidirectional coupling 26 . Almazan et al. HITEC 1998). The stiffness and energy dissipation characteristics of the FPS generally increased with increasing periods of the excitation. 1998. specifically. stiffness.000 cycles of service movements showed that deterioration from fatigue and wear was not evident (HITEC 1998). Both the postyield stiffness and the yield force of the two types of isolators are known to be affected by the normal force being imposed. 2. damping and energy dissipation was found to have relatively low sensitivity to temperature extremes (Zayas and Stanley 1999. 1998).2.
The effects of postyield stiffness on isolator response were more pronounced for ground motions with low frequency content.has not yet been extended to account for normal forcedependency and implemented in bridge analyses to the authors` knowledge. the unique response of the isolators may not be adequately captured with this simplified modeling approach. This was attributed to the structure remaining within the elastic range for a longer period of time thus decreasing the amount of energy dissipation. Initial stiffness of the isolators was noted to have negligible affects on the seismic response. It has been shown that excluding the inplane coupling of the orthogonal response for the isolators may result in significant underestimation of the displacements and forces for both type of isolators (Mosqueda et al. An increase of the yield strength resulted in higher superstructure accelerations and lower bearing displacements. 27 . Dicleli and Buddaram (2005) compared the response of an idealized bridge substructure utilizing seismic isolation devices that have the characteristic stiffness values of leadrubber. The authors highlighted that the peak isolator displacements decrease and forces increase as the postyield stiffness of isolators increase. Jangid 2004). Matsagar and Jangid (2004) analyzed the effects of different yield displacement and yield force properties in bilinear forcedeformation models of elastomeric and friction isolators for a multistory building model. High variation of the normal loads may result in fracture and a considerable change in the horizontal response of the isolators (Priestley et al. 1996). Consequently. 2004. high damping rubber and frictionbased bearings. A higher yield displacement resulted in substantial decrease in peak superstructure accelerations and marginal increase in bearing displacement. The bilinear forcedeformation idealization of isolators allowed by the Specifications is based on the assumptions that the response is unidirectional and the normal force acting on the isolators is constant.
The four types of isolators were the pure friction system. The authors concluded from the study that highdamping rubber bearings was influential in reducing bearing displacement but transmitted higher accelerations into the structure compared to other isolators. The authors identified normalized hysteretic energy and interstory drifts as a definitive measure for describing the response of the frame. the friction pendulum system. the high damping rubber bearings. the pure friction system and the friction pendulum system. and the New Zealand (leadrubber) isolator. (2003) concluded that base displacements are smaller but interstory drifts are larger for frictional systems compared to neoprene systems. Ordonez et al. the laminated rubber isolator. the laminated rubber isolator. The separate isolation schemes included elastic rubber isolators. (2) linear viscous dampers. the leadrubber bearing. It was concluded from the analysis that no control system was consistently better than another. and ElectricdeFrance system under nearfault ground motions.Ordonez et al. It was concluded that there existed an optimal value of isolator damping when caused higher accelerations to the superstructure exceeded. (2002) compared the seismic performance of three different structural control methods employed on steel frame buildings. The authors showed that the Electricde 28 . Linear rubber and lead rubber bearings resulted in almost the same response that corresponded to larger deformations but smaller superstructure accelerations compared to other isolators. The three schemes of structural control were the: (1) friction pendulum system. Jangid and Kelly (2001) analyzed a two degreeoffreedom isolated building model. (2003) investigated the earthquake response of a twodegree of freedom model as a function of different isolators. and that they all reduced the amount of energy dissipated by the structural system and interstory drifts to negligible values. The authors reported base displacements in descending order as the leadrubber isolator. and (3) active tendon brace system. Barrosso et al.
Ghobarah (1988) investigated the parametric effects of LRB isolator stiffness. pier stiffness and pier eccentricity on the response of SIB models. However. The author reported higher superstructure accelerations with sliding isolators compared to rubber isolators for strong earthquakes. 2. The authors parametrically quantified the error for assuming a rigid superstructure in the models in the fundamental period as a function of deck to pier stiffness ratio. which is conversely different from the case in buildings.France acquires lower deformations compared to the rubber isolators but transmits similar accelerations to the superstructure. The author showed that the increase of the flexibility of piers results in an increase in the force and displacement demands at the abutments and reduced shears at the pier. Parametric Studies Studies aimed at investigating the effects of different design parameters in SIBs are limited. It was observed that the rigid deck assumption underestimated the fundamental period and the error is more pronounced for stiffer isolators. The authors parameterized the location of the effect of energy dissipation of the isolators throughout the bridge. the error was within the order of 6% for typical highway bridges.5. Larger elastic stiffness of the isolators reduced the deck displacements and the behavior approached to a case of hinged supports. The forces and displacements at the abutments became larger as the pier offset was increased. Ghobarah and Ali (1988) compared the response of a seismically LRB isolated and nonisolated bridge model and investigated the effects of different design parameters in the response of the SIB response. Isolation schemes involving higher energy dissipation at the 29 . Sugiyama (2000) compared the seismic response of a continuous steel box girder bridge isolated either with sliding or rubber isolation. the author underlined that the difference of the bridge isolated was negligible for weak earthquakes. Additionally.
It was concluded that as long as the isolators’ yield strength remained within 410% of the superstructure weight. the seismic response is not significantly affected. Meng and Lui (2000) analyzed the response of a skew concrete box girder bridge by accounting for deck flexibility and column boundary conditions.abutments compared to the piers resulted in considerable reduction in the seismic forces acting on the piers. as rigid deck or elastic beam elements.3% to 53% of all the bridge types considered in the Central and Southeastern United states are skewed. as “the angle measured between the center line of the bridge supports and the line perpendicular to the bridge center line. The effectiveness of the isolators reduced considerably as the superstructure flexibility increased. A parametric investigation of the magnitude of the yield force of isolators showed that higher yield forces result in reduced shear forces in the pier and the displacements of the deck. was found to result in greater period shifts of the bridge. The author showed that the 30 . The models of the same bridge were generated either as elastic shell elements. This study investigated the effects of a range of isolator characteristics and the strengths of various components of the bridge on the overall response. Turkington et al. Maleki (2001a) derived closed form solutions for the translational and torsional periods of skewed bridges supported on elastomeric bearings and have crossframe diaphragms. Bridges may be constructed with a skew angle to accommodate traffic and site conditions. The authors noted that large skewness may lead to torsional and lateral vibrations. (1989) performed parametric analyses on a bridge seismically isolated with combined LRB and elastomeric bearings to develop a design method. θ . Nielson (2005) defined bridge skew. which is equivalent to increasing the postyield stiffness. It was also noted that a 15. An important structural aspect of skewed bridges is that their vibrational modes do not uncouple in orthogonal directions as in the case of nonskewed bridges (Maleki 2001b). Increasing the LRB height.
However. there are still issues pertaining to bridge seismic isolation. (2005) investigated the effects of soilstructure interaction (SSI) in SIBs using iterative multimode response spectrum analyses.fundamental period of the bridges with elastomeric bearings increases as the skew angle increases and the second translational period is independent from the skew angle. The authors proposed closed form solutions to account for SSI effects in design equations used by the AASHTO.6 Critical Appraisal A review of the current stateoftheart illustrates that the mechanism of the FPS has been thoroughly studied. It was concluded that SSI effects were negligible for SIBs with heavy superstructure and light superstructure located on stiff soil. 2. medium and hard soil. The authors concluded that stiffer soil resulted in higher isolator forces but lower bending moments at the base of the piers. Thakkar and Maheshwari (1995) compared the response of a SIB model located on soft. it was concluded that the fundamental period of the structure may increase substantially by accounting for SSI effects. Additionally. It was shown that including the SSI effects reduced the estimated base obtained from the AASHTO design procedures. Dicleli et al. The three main gaps in the literature were identified as the following: 31 . However. SSI effects were influential on isolator forces and displacements for SIBs with light superstructure and heavy substructure regardless of the stiffness of the soil. Vlassis and Spyrakos (2001) performed parametric analyses to asses the influence of SSI on the dynamic response of a SIB pier located over shallow soil stratum overlying a rigid bedrock. This study considered two bridge types that had different superstructure and substructure weights. The individual response of the conventional FPS has been established with experimental and analytical research. in particular with the FPS that need further clarification.
used twodimensional structural models and idealized the forcedeformation response of the isolators as bilinear which overlooked some of the distinguishing aspects of the response of the two isolator systems. Previous research considering the effects of different aspects of nonlinearities in the response of the FPS showed that there may be a significant divergence from a bilinear idealization. There is a need for further assessment of the comparative response of SIBs via detailed isolator models that can capture the distinctions in the mechanism of sliding and elastomeric isolators. There is a need to develop a better understanding of the modeling assumptions and the required level of accuracy for the FPS in threedimensional (3D) bridge models. 2) The number of studies that compared the response of SIBs with different isolator types is limited.1) The FPS has a highly nonlinear response that involves the variation and coupling of different parameters. 3) Previous research on the parametric affects of design parameters in SIBs have focused primarily on bridges utilizing elastomeric systems and was generally confined to twodimensional models that excluded the vertical components of ground motions. Available studies in this area did not consider the vertical components of ground motions. Further insight on the influence of design parameters in bridges isolated with the FPS is needed. 32 .
substructure lateral linearelastic stiffness k1 and isolator linearelastic stiffness k2. the primary objective of bridge seismic isolation is the protection of the piers and the foundations and in some cases the abutments (Skinner et al. respectively. 33 . The effects of seismic isolation is examined through an idealized bridge structure with a lumped superstructure and substructure mass. This is accomplished by installing seismic isolators between the superstructure and these components. corresponding code aspects and the characteristics of the isolators. The objective of this chapter is to elaborate on these aspects of bridge seismic isolation. u1 and u2 (Figure 3.1 Introduction The general topic of bridge seismic isolation requires an understanding of structural analysis and dynamics. 3. The characteristics of most commonly used isolators in bridges are described.1). The superstructure is seldom of concern due to inherent strength in design for vehicle loads. The governing design codes present different methods of analysis for SIBs.2 General Features of a Seismically Isolated Bridge (SIB) Typically. An outline of the analysis methods in the governing design code of bridge seismic isolation is also presented. 1993). In superstructure isolation. the substructure is not isolated from the ground motions but decoupled from the relatively larger mass of the superstructure. m1 and m2.CHAPTER 3 SEISMIC ISOLATION FOR BRIDGES 3. Decoupling of the structure from the horizontal components of the ground motion via an isolator results in a redistribution of seismic forces. Various isolators are available for bridges. The effects of seismic isolation in bridges are illustrated on a simplified bridge model via modal analysis. Each lumped mass is designated a lateral degreeoffreedom.
18 kNs2/cm.2) det k − ω n m = 0 2 [ ] (3. The mass and the stiffness matrices of this system are: 0 m m= 1 0 m2 k + k − k 2 k= 1 2 k2 − k2 The characteristic equation (frequency equation) is (Chopra 2000): (3. and k1=280.2. The effect of introducing a flexible layer to the second period is negligible.35 s. the natural period of the nth mode.k2 m2 u2 u1 m1 k1 Figure 3. there is a notable increase in the fundamental period of the structure.1) (3. Tn= 2π / ω n . 34 The .3) where ωn is the natural frequency of the nth mode. Given the positive definite property of k and m. m2=0. It is observed that for values of k2/ k1 < 1. In this case the structure essentially reduces to a cantilever with a lumped mass at the tip and has a T1 =0.4. From the solution of Equation 3.1 Idealized seismicisolated bridge substructure. Large values of k2/ k1 imply a rigid layer between m1 and m2. as a function of k2 / k1 is as given in Figure 3.70 kNs2/cm. the expanded form of Equation 1 has real and positive roots for ωn : 2 m1 m2 ω n + (− k 1 m2 − k 2 m2 − k 2 m1 )ω n + k 1 k 2 = 0 4 2 (3.2 kN/cm.4) Consider the structure to have m1=0.
φn .installation of a flexible layer. and k1=280. Tn (s) 2 1 0 0.6 k2 / k1 0. where k2/k1<1. consider the system to have seismic isolation with the objective of achieving three times increase from the nonisolated period of 0. The nth mode shape.2 0.2 Figure 3. is determined from: [k − ω m]φ 2 n n =0 (3.5) The solution of Equation 3.2 kN/cm ).2 First.3.18 kNs2/cm.35 s. It is observed that in the first 35 . mode structural periods as a function of k2/ k1 (m1=0.0 1. To highlight the dynamics of superstructure isolation.5 by assuming unit displacements at the first degreesoffreedom in each of the two modes is given in Figure 3. T2.70 kNs2/cm. 4 k2 m2 u2 u1 k1 m1 T1 T2 3 Period.05 s and T2=0.3 kN/cm in which the first two periods become T1=1. is an effective approach for increasing the fundamental period of the structure with little influence on the second period. and second. m2=0.15 s. T1.2.4 0. From Figure 3.0 0. .8 1. this corresponds to approximately k2=26.
The spatial distribution of the effective earthquake forces are defined by : s = ∑ Γ n mφ n n =1 T T N (3. An important implication of this result is that the first mode forces are essentially the same as the total forces. and M n = φ n mφ n . Vbnst. In the second mode the structural deformation is larger. based on its low participation factor. the isolator undergoes significant deformation compared to the substructure.6) where Γ n = Ln /M n .3 Structural shapes for (a) Mode 1 (b) Mode 2. and i is the influence vector. The second mode components of the static forces are negligible compared to those in the first mode. however.mode shape of the structure. the second mode response which involves the structural deformations is expected to make little contribution to the earthquake response of the structure. this mode is subsequently shown to make insignificant contribution to the earthquakeinduced forces of the structure. and base moment. Mbnst. Ln = φ n mi . Consequently. (a) (b) Figure 3. 36 .4. for the nth mode are given in Figure 3. The modal expansion of these forces and the modal static responses for the base shear.
8) where An is the peak value of An (t ) for a particular Tn provided by a pseudoacceleration spectrum.7) 2 where rn st is the modal static response. t. this effect is dampened 37 . is obtained by combining the contributions of all modes: r (t ) = ∑ rn An (t ) st n =1 N (3. This is obtained from: rn O = rn An st (3.5).2 (Figure 3.1. An (t ) = ωn Dn (t ) is the pseudoacceleration timehistory and Dn (t ) is the displacement of the nth mode.4 Modal expansions of effective earthquake forces and modal static responses for the base. The base shear calculation of each mode and their combination via the squarerootofsumofsquares (SRSS) is given in Table 3.Figure 3. The earthquake response of the structure at any given time. Consider the pseudoacceleration design spectrum of AASHTO in the Memphis. Particular interest is the peak response of the structure over the duration of the earthquakeinduced response. TN region with Type II soil profile and acceleration coefficient A=0. Despite a larger pseudo acceleration that corresponds to the second mode of the response.
3 Bridge Seismic Isolation in Design Codes The American Association of Highway and Transportation Officials (AASHTO) Guide Specifications for Seismic Isolation Design provides the “guide specifications for 38 .5 Pseudoacceleration design spectrum for AASHTO Type II soil profile with acceleration coefficient A=0.5 2.3 0.5 0.50g 0. Vbo (kNs2/cm) 0.2.05 s 0.35 s T=1.due to considerably smaller static response.255 3.0 Period.15 s T=0.073 0.5 1.500 Static Base shear. Table 3.0 T=0.875 0.278 0.2 0.244 0.4 0. Vbst (kNs2/cm) 0.6 0.0 0. The design pseudoacceleration for the nonisolated structure corresponds to 0.1 0. The seismic isolation scheme reduced the base shear by 72%.5g and a peak base shear of 0.145 Peak Base shear. T (s) 1. Sa (g) for ξ=5% 0. Spectral acceleration.438 kNs2/cm.1 Calculation of base shear Mode 1 2 SRSS A/g 0.278g Figure 3.
As new designs became available by 1995 the first Seismic Isolation Guide Specification was essentially rewritten in 1997 to address the advances in the industry. In calculating the Keff. By this time elastomeric bearings were primarily used in bridge seismic isolation (Stanton 1998). This statically equivalent seismic force is: F = C sW (3. The first Seismic Isolation Guide Specification for bridges by the AASHTO was available in 1991.9) where W is the total vertical load for design of the isolation system and Cs is the elastic seismic response coefficient computed from: K eff d W Cs = (3. the configuration. The following is a brief overview of these procedures.the seismic isolation design of highway bridges”.1 Procedure 1: Uniform Load Method The Uniform Load Method is essentially the method of approximating earthquake loads with an equivalent static force.10) where d is the total deck displacement relative to ground and Keff is the sum of the effective linear stiffnesses of all bearings and substructures. Four procedures are available in the Guide Specification for the analysis of SIBs.3. flexibility and individual stiffnesses of the isolator units and substructure shall be taken into account: k sub k eff K eff = ∑ j k sub + k eff (3. 3.11) 39 . This is a supplemental document to the AASHTO Standard Specifications for Highway Bridges Division IA: Seismic Design.
Si is the numerical coefficient per sitesoil profile as defined in Table 51 of the Guide Specification. the method is simplified for SIBs due to the rigid body deformation of the superstructure.6 Single substructure and isolator idealization. Teff is the period of the seismically isolated structure in seconds in the direction under consideration and B is the damping coefficient.12) where A is the acceleration coefficient defined in Table C31 of the Guide Specification.2 Procedure 2: Single Mode Spectral Method The Single Mode Spectral Method is the same approach in article 4.where Σ extends over all substructures. The displacement d (mm) is given by: 250 AS i Teff d= B (3. However. The first three steps of the procedure is devoted to finding the deflection. Figure 3. and ksub is the substructure lateral stiffness and keff is the isolator unit lateral stiffness calculated at maximum displacement capacity (Figure 3. 3. slope and moments in the 40 .6).4 of AASHTO Standard Specifications (Division IA: Seismic Design).3.
∆max (Figure 3. is used for calculating resulting member forces and displacements. 3.5 of AASHTO Standard Specifications (Division IA: Seismic Design) for the specifics on modeling.bridge superstructure via the double integration method. yield force.3. For SIBs the Guide Specification Commentary C7. This bilinear hysteretic model is characterized by the elastic stiffness.3 Procedure 3: Multimode Spectral Method Different from the previous two procedures. The procedure is essentially performing an equivalent linear response spectrum analysis. The bridge model is to be 41 . The AASHTO Guide Specifications (1999) Section 7 on Analysis Procedures states that: “To simplify the nonlinear behavior of the isolator unit. pe(x). Fy.7). The Guide Specifications refers to article 4.2 gives: p e ( x) = w ( x ) C s (3. characteristic strength. pe(x). Isolators are modeled by their effective stiffness based on design displacements. These results are used to establish the intensity of the equivalent static seismic loading applied to represent the primary mode of vibration. This Procedure involves the timehistory analysis of the bridge models with isolation bearings that have nonlinear deformation characteristics. the Multimode Spectral Method requires a detailed model of the bridge in a computer program.13) where Cs is calculated from Equation 10 and w(x) is the deadloadperunit length of the bridge superstructure.4 Procedure 4: TimeHistory Method The Time History Method of analysis is the most sophisticated among the four permitted in the Guide Specification (Stanton 1998). kd. postelastic stiffness. ku.3. Qd. a bilinear simplification may be used”. 3. and maximum bearing displacement. The loading.
The 5%damped response spectrum is established by taking the square root of the sum of the squares of the orthogonal components of the ground motions. and Friction Pendulum System and Eradiquake for sliding. (shown in Figure 3. The maximum response parameter among the three ground motion records determines the design. The ground motion records may be frequencyscaled to match the characteristics of the corresponding site. Figure 3. The Guide Specifications specifically mention two isolator types from each category: LeadRubber and High Damping Rubber for elastomeric.subjected to the orthogonal components of no less than three ground motion records. AASHTO 1999). The development of elastomeric isolators.7 Characteristics of bilinear isolation bearings per AASHTO Guide Specifications.8) is considered an extension of elastomeric bridge bearings and 42 . This method is required if the structure acquires effective periods greater than 3 seconds. If seven or more ground motions are used.4 Descriptions of Common Isolators Isolators can be classified as sliding and elastomeric (Taylor and Igusa 2004. than the average value of response parameter may be used for design. 3.
Sliding isolators have been modified from Teflon sliding bearings that are commonly used in bridge applications to accommodate movements from factors such as thermal expansions. widespread use of the sliding isolators corresponds to early 1990s.bearings for vibration control of buildings (Naeim and Kelly 1996). Naeim and Kelly (1996) noted that a purely sliding system with talc proposed by Johannes Avetican Clanterients in 1909 was the earliest seismic isolation system (Figure 3.8 Typical elastomeric isolator (Taylor and Igusa 2004). However. Figure 3. creep. Elastomeric isolators are fabricated through a process called vulcanization. a general overview of the four systems mentioned in the Guide Specifications are presented for completeness. Despite the emphasis of this study on the sliding isolator FPS. These isolators typically possess two surfaces with different finish to slide over one another (AASHTO 1999). 43 . These systems have become a practical tool for seismic isolation in the late 70s. shrinkage or prestressing (HITEC 1998). which is the bonding of steel plates with rubber (HITEC).9).
44 . The value of Qd is a function of the dynamic friction coefficient of the central sliding bearing. 3.4. It is possible to design the bearing to have different energy dissipation and stiffness characteristics in the inplane orthogonal directions. The Eradiquake System (EDS) The Eradiquake System (EDS) isolator is made up of essentially two components: (1) a sliding multirotational bearing assembly (2) a maintenancefree restoring device called the Mass Energy Regulator (MER) (Figure 3.1.10). The system restores through the MER and simultaneously dissipates energy via the steel and composite sliding bearing at the center.Figure 3. The value of Kd is governed by the properties of the MER (AASHTO 1999).9 Clantarient`s base isolation system using a layer of talc as the isolating medium (Naeim and Kelly 1996).
The values of Qd and Kd are a function of the additives to the rubber (AASHTO 1999). 3. The isolator stiffens and acquires a higher level of energy dissipation at large deformations due to the strain crystallization process in the rubber (Naeim and Kelly 1996). High damping rubber layers deform under shear to reduce earthquake loads and dissipate energy (HITEC 1998). 45 .2.4.11) (Naeim and Kelly 1996).Figure 3. The High Damping Rubber System (HDRS) The High Damping Rubber System (HDRS) is essentially the aforementioned elastomeric bearings with the only difference of having a modified rubber compound that acquires increased damping characteristics (Figure 3.10 The Eradiquake seismic isolation bearing.
1993).3. 10 MPa. natural rubber causes a significant increase in the Qd (AASHTO 1999). As a multilayered elastomeric type bearing. the steel plates force the lead plug bearing to deform in shear (Naeim and Kelly 1996).4. Additionally. A lead core is inserted down the center of the bearing for energy dissipation and stiffness (Priestley et al. 1996).11 (a) High damping rubber bearing used in the earthquake simulator tests with dimensions in mm (b) corresponding forcedeformation hysteresis (Kikuchi and Aiken 1997). rubber and a lead core (Figure 3. The forcedeformation response is typically modeled as bilinear (Naeim and Kelly 1996). and has good fatigue properties (Skinner et al. The Lead Rubber Bearing (LRB) The Lead Rubber Bearing (LRB) consists of steel plates.(a) (b) Figure 3. the LRB is susceptible to a buckling type of instability.12). 46 . 3. The steel layers placed between the rubber serves to limit the edgebulging of the rubber (Tyler 1991). The value of Qd is a function of the lead core and the value of Kd is a function of the rubber (AASHTO 1999). Lead is a feasible option because it yields in shear at relatively low stresses. An important characteristic of the LRB is that the in cold temperatures.
It has been also reported based on information provided by the manufacturers that the cost.12 Typical lead rubber bearing (LRB) (Taylor and Igusa 2004). Figure 3.The LRB is a type of isolator widely used in bridge applications (Buckle and Mayes 1990). The mechanical properties of the LRB is elaborated further in Chapter 6. 47 . Consequently. size and energy dissipation of the FPS and LRB may be comparable in bridge applications (Dicleli 2002). the LRB is used in Chapter 6 to compare the applications of sliding versus elastomeric seismic isolation of bridges.
the Guide Specifications. 48 . An overview of common isolators mentioned in the Guide Specifications has been presented. It is concluded from the modal analysis of a simplified bridge model that the stiffness characteristics of the isolators significantly control the dynamic response of the system. the timehistory method. There are four analysis procedures available in the governing code of SIBs. The most sophisticated one of these procedures. 3.5 Conclusion Seismic isolation of bridges is an effective approach for reducing the forces imparted by earthquakes.13 (a) Lead rubber bearing (LRB) used in the earthquake simulator tests with dimensions in mm (b) corresponding forcedeformation hysteresis (Kikuchi and Aiken 1997).(a) (b) Figure 3. The favorable effects of seismic isolation is achieved essentially by decoupling the response of the structure from the ground motion and shifting the period for lower pseudo acceleration in the design spectrum. will be the basis of the analyses throughout this study.
1987): 49 .1 Introduction This chapter is devoted to the development of a new finite element (FE) model that can represent the inherent nonlinear and coupled response of the Friction Pendulum System (FPS). The implementation of the response of the isolator into OpenSees via developing new C++ classes is explained. Preliminary results on the influence of different modeling assumptions for the FPS are highlighted.2 Simplified Isolator Response Modeling The mobilized response of the FPS is representative of a mass sliding on a perfectly spherical surface with a coefficient of friction µ . 4. and the nonlinear kinematics of the isolator response is developed. The two components of the intrinsic forces of the FPS consists of the pendulum motion of the mass. Assuming small deformations. The FE model is verified with experimental data. the unidirectional forcedeformation response of the FPS is (Zayas et al. The equations and modeling techniques used to represent the nonlinear and coupled response of the isolator is reviewed. and the friction between the mass and the sliding surface.CHAPTER 4 FRICTION PENDULUM SYSTEM (FPS) MODELING 4. The analogy between the simplified bilinear response of the FPS and the bilinear modeling presented in the AASHTO Guide Specifications (1999) is described. and a radius of curvature R. fR. f µ . The general characteristics of the simplified bilinear model of the isolator are explained.
1 are representative of plastic and elastic models. and the pendulum motion response. and sgn(δ ) is the signum & function.1 The signum function.2.1). δ is the sliding velocity. The friction response. respectively (Figure 4. in Equation 4. and (b) pendulum. R is the radius of the concave & & surface. fR. The signum function is equal to +1 or 1 depending on whether δ is negative or positive.6 74 4µ 8 } & N f = N µsgn(δ ) + δ R f fR (4. respectively.2 The intrinsic response components (a) friction. illustrated in Figure 4. f µ .1) where N is the normal force acting on the sliding surface. Figure 4. (a) (b) Figure 4. 50 . δ is the sliding deformation. f µ . fR.
∆max .4). (1990) for conditions relevant to the FPS is considered. Ku.4. If the yield displacements of steelTeflon sliding surfaces reported in the order of 0.The combination of fR and fµ corresponds to a unidirectional rigidplastic hysteretic model given in Figure 4.3 is based on the assumptions that: (1) N is constant.050. 51 . Figure 4.02 cm by Constantinou et al. Qd. Figure 4. (3) the horizontal response is uncoupled in the orthogonal directions. Kd. (2) µ is constant. characteristic strength. This bilinear model for the FPS given in Figure 4.3 takes up the characteristics of the bilinear model in Figure 4.3 Forcedeformation characteristics of the unidirectional rigidplastic response of the FPS. The AASHTO Guide Specifications (1999) Section 7 on Analysis Procedures states that: “To simplify the nonlinear behavior of the isolator unit. postelastic stiffness. Fy. a bilinear simplification may be used” (Figure 4. The following sections elaborate on these aspects and how they may be incorporated into the response. and (4) isolator deformations are small and planar. and maximum isolator displacement. This bilinear hysteretic model is characterized by the elastic stiffness.3. yield force.
N. N changes the magnitude of µ .4 Forcedeformation characteristics of bilinear isolators.0 fg = if δ g > 0. f µ and fR. 1987). An increase in the magnitude of N is indicative of a higher yield force which may delay the mobilization of the FPS under dynamic loads and a higher postyield stiffness which may reduce the flexibility of the isolator.0 (4. 4.5).Figure 4. of the response. 52 . This behavior closely matches the response of a zerolength gap element defined with a force: k g δ g if δ g ≤ 0. however this relationship is discussed subsequently. The conventional FPS does not have resistance in tension and it is approximately rigid in compression (Zayas et al.0 0.3 Normal Force The normal force.2) where kg is a high compression stiffness and δ g is the deformation (Figure 4. Additionally. Modeling the vertical response of the FPS with a gap element allows simultaneously the monitoring of the variations in N and capturing the effects of uplift and impact in the FPS (Almazan and Llera 2003). acting on the FPS is inherent in both resisting force components.
Dfmax is the difference between fmax. 1990).6). having units of time per unit length. Accurate mathematical models have been developed by Constantinou et al.p.3) where. and a is a constant. that controls the variation of the coefficient of friction with velocity. and ε is a constant that controls the variation of fmax between very low and very high pressures (Figure 4. 4.3 53 .(1994).0 − D fmax tanh(εP) (4. The term fmax as a function of P was given as: f max = f max. respectively. is considered in this study as the influence of pressure on fmin and a were shown to be negligible by Tsopelas et al. P. in addition to the material properties of the surface. fmax and fmin are the values of coefficient of friction at large and small sliding velocities. (1990) to capture the & value of µ for a range of δ and N that is of interest to the response of the FPS. The & influence of δ on the µ was approximated via the aid of experimental results as: µ = f max − D f e & ( −a δ ) (4.0 and fmax. Equations 4. Df is the difference between fmax and fmin.4) where fmax. & was found to be primarily a function of δ and N (Mokha et al. µ .Figure 4.0 and fmax. Only the dependency of fmax to pressure.5 Gap element forcedeformation model.4 Coefficient of Friction The coefficient of friction.p are the values of fmax at very low and high pressures respectively.
A simplified approach for modeling the planar frictional response is to consider two independent unidirectional elements according to the Coulomb’s model in the orthogonal 54 . 4.6 Variation of the coefficient of friction with (a) velocity of sliding.4 may be updated for the values of δ and N in an iterative solution scheme to account for the changes in µ . and (b) isolator contact pressure (Roussis and Constantinou 2006).& and 4. 1990). The two important characteristics of the bidirectional sliding motion are that: (1) the inplane forcedeformation response of the sliding is isotropic.5 Bidirectional Coupling Bidirectional motion may commence in the FPS isolator subject to multidirectional excitations. (a) (b) Figure 4. and (2) the behavior shifts from stick and slip conditions (Constantinou et al.
6 is used (Figure 4. the magnitude of  fµ  during sliding ranges between Nµ . 55 . and 2 Nµ .5) & & where. are the components of the velocity in the x and y directions. 1990). f µx and f µx are the components of the friction force. This implies that the interaction surface of f µx and f µx is circular and the resultant sliding friction force magnitude. The planar sliding forcedeformation response of the FPS is isotropic. In this case the planar frictional force is: & sgn(δ x ) f µx fµ = = N µ sgn(δ&y ) f µy (4. However.1. Consequently. if sliding in a path along the 45 degree direction.7). The planar frictional force that satisfies these conditions during sliding is: f µx cos(θ ) fµ = = N µ sin(θ ) f µy (4. the calculation of fµ with two independent Coulomb’s models for plane motion. produces inaccuracies in capturing stickslip conditions and raises complications in numerical computation (Constantinou et al.6) & & where.5.5. is equal to Nµ regardless of the sliding 2 direction.6 are insensitive to the variations in the magnitudes of respective sliding velocities. the components of fµ in Equation 4. if the response of fµ is modeled by Equation 4. if sliding along the x or y axes.  fµ  = f µx + f µy . and δ x and δ y . The shape of the interaction surface between f µx and f µx is circular if Equation 4. θ = tan −1 (δ y /δ x ) defines the direction of the sliding motion. respectively.x and y directions of the horizontal plane based on Equation 4. Unlike the case presented by Equation 4. overestimates the resistance.5 is used and square if Equation 4.
5 to plane motion (Younis et al.7 Frictional interaction surface (a) uncoupled (b) coupled response. (1986) and presented bidimensional hysteretic parameters. where one or both of the components of the velocity are zero or very low (Constantinou et al. to evaluate f µ in planar steelTeflon sliding interfaces. where motion commences in both directions. The frictional force vector in this case is defined as: f µ = Nµ η (4.1).7) 56 . There are complications in using Coulomb’s model in numerical solutions because of the low rate of convergence caused by this discontinuity (Feldstein and Goodman 1973) and the difficulty of extending Equation 4. In the Coulomb’s model. and (2) the sticking phase. This was attributed to the significantly more sticking phases that developed in the Coulomb’s model as a result of independence of resistance to the magnitude of the sliding velocity. 1983). 1990). The response of the sliding system may shift between two phases: (1) the sliding phase. Constantinou et al. Constantinou et al. η = [η x η y ]T.Figure 4. (1990) extended the work of Park et al. (1990) reported that the response of Teflonsteel sliding surfaces predicted via the Coulomb’s model contained highfrequency components that did not prevail in the experiments. the transition between the two phases is independent from the magnitude of the sliding velocity and discontinuous (Figure 4.
∆ moments occur at the structural members below or above the isolator (EPS 57 . are the components of the sliding velocity in the x and y directions.0 . and (3) for sticking conditions η < 1. A. The orientation of the FPS controls whether the P. . the solution of Equation 4. respectively. γ are dimensionless constants that control the shape of the hysteretic loops. 4.6 Large Deformation Moments In bridge applications. ∆Y is the yield displacements.8) can be solved via numerical algorithms presented for the common computer languages like C++ (Lee and Schiesser 2004. Constantinou et al. (3) common stability. the FPS is installed above piers and abutments as either the concave dish facing up or down (EPS 2002). β . The solution of the of the coupled differential Equation (4. . 2003). Press et al. the semiimplicit method presented by Rosenbrock (1963) have been utilized as the solution algorithm due to its: (1) relative simplicity. (1990) showed that for A /( β + γ ) = 1 . where θ = arctan(δ y / δ x ) . . (2) ability to handle stiff problems.8) & & where δ x and δ y . common numerical calculation software such as Mathcad (Pritchard 1998) and Matlab (Shampine et al. . (2) for sliding conditions the hysteretic parameters are η x = cosθ and η y = sinθ . 2003) have a variety of builtin functions to solve coupled differential equation systems. and (4) acceptable accuracy (<104 – 105) (Press et al. a y = β + γ sgn(δ y η y ) . In the model being developed for this study.7: (1) describes a circular interaction curve. Additionally. a x = β + γ sgn( δ x η x ) . 2003).where η is dimensionless hysteretic parameters that evolve according to the following coupled set of differential equations: 2 & η x 1 A− a x η x η= = & η y ∆s − a x η x η y & − a y η x η y δ x 2 & A− a x η x δ y (4.
2002). This unique feature of the FPS does not have implications on the inplane forcedeformation response and allows for diverting P ∆ moments from weak elements of the structure (Almazan and Llera 2003). Figure 4.8 is a schematic of the displaced shape of an FPS between a simplified bridge superstructure and the column. The normal force, N, is transmitted through the slider to the concave dish. Assuming that the rotations at the superstructure and the top of the column are negligible, the displaced configuration of the FPS results in an internal moment M=N δ . This internal moment, M, is balanced at the tip of the column if the concave dish is at the bottom and by the superstructure if the concave dish is at the top. Almazan and Llera (2003) presented a nonlinear
transformation matrix for their FPS model to account for this aspect, which is elaborated in the subsequent section.
M
N Superstructure Column
Figure 4.8 Deformed shape of the seismic isolator between the superstructure and the substructure with concave dish at the (a) bottom (b) top.
4.7 Mathematical Model
The exact three dimensional (3D) kinematics equations considering large deformation effects of the FPS were developed by Almazan and Llera (2003). The zero58
length element developed in OpenSees to model the response of the FPS is constructed based on these principles by assuming no nodal rotations. The author does not claim any innovation for implementing this simplification and refer the reader to Almazan and Llera (2003) for a more detailed presentation of these principles. Here, only a brief summary of the mathematical formulation is presented with similar nomenclature as the original equations. The zerolength element is 3D with 6 degrees of freedom (DOF) per node (Figure 4.9). This model accounts for the variations of the N via an inherent gap element described by Equation 4.2 and the variations of the µ via Equations 4.3 and 4.4 at each integration time step. The coupling of the sliding forces are included via the hysteretic parameters, η = [ η x η y ]T, as described in the previous sections. The P ∆ moments are transferred to the nodes of the model via a nonlinear transformation matrix.
Figure 4.9 Schematic view of the model.
59
Assuming that the nodal rotations are negligible, the local slider and global coordinates coincide. The nodal deformations of the element are defined as u=[u(J)
u(I)]T, where u(J)=[ux(J) uy(J) uz(J) rx(J) ry(J) rz(J)]T and u(I)=[ux(I) uy(I) uz(I) rx(I) ry(I) rz(I)]T define
the motions of nodes J and I, respectively. The instantaneous position and velocity of the
& & & & slider is defined by the vectors δ = [ δx δ y δz ] and δ = [ δx δ y δz ], respectively
(Figure 4.10). The slider’s motion is bounded by the spherical surface of the concave dish defined as: G = δ x + δ y + (δ z − R ) 2 = 0
2 2
(4.9)
The unitary vectors in the outward normal direction and the tangential to the trajectory of the slider are:
n=
and
∇G 1 T = δ x δ y (δ z − R ) ∇G R
[
]
(4.10)
& δ ηx s= = cosα & δ η
respectively (Figure 4.10), where,
ηy
cosα sin α η
T
(4.11)
ηy ηx δx +δy η η α = arctan R− δ z
(4.12)
denotes the angle between the frictional force component and the xy plane. The local slider restoring forces for all phases is:
f = fx
[
fy
f z =Nr
]
T
(4.13)
where, r, is the restoring force orientation vector that is constituted from the normal and tangential components of the slider as:
60
normal component r= } n tangential component + 68 7 µ ηs (4. respectively. L .10 Deflections and forces acting on the slider.17) The top and bottom signs in Equation 4. is: ˆ F = LT f ˆ where LT is the transform of the nonlinear transformation matrix defined as: (4. Figure 4. The transformation matrix. respectively. The nodal force vector F=[F(J) F(I)]T.14) and The concavity and frictionbased components of the isolator are f R = n N f µ = N µ η s . 61 .16) ± 1 0 0 0 0 0 ± 1 0 0 0 m δz ±δy ˆ = 0 1 0 0 0 0 0 −1 0 δ L 0 −δx z 0 0 ±1 0 0 0 0 0 −1 m δ y ± δ x 0 (4. The normal force in the isolator is: N= fg rz (4. respectively.15) where rz is the axial component of the vector r.17 are used to differentiate between the ˆ downward and upward positions of the FPS. where F(J)=[Fx(J) Fy(J) Fz(J) Mx(J) My(J) Mz(J)]T and F(I)=[Fx(I) Fy(I) Fz(I) Mx(I) My(I) Mz(I)]T define the forces at nodes J and I.
OpenSees has an open source object oriented architecture in the C++ programming language that maximizes its modularity. The implementation of this procedure requires a combined knowledge of C++ programming.13 becomes: f = fx [ fy fz ] T δy δ + η y µ − 1 = N x + ηx µ R R T (4.∆ moments. Although OpenSees provides a variety of hysteretic uniaxial forcedeformation response models. This powerful attribute of OpenSees allows researchers to analyze a wide range of aspects of innovative materials and elements in larger models.depends exclusively on geometry and is nonlinear to account for the variation of the P. Consequently. The exclusion of the vertical rise in the concave dish corresponds to δ z = 0. Material and element models describing the hysteretic response of new structural members can be developed as C++ classes and inserted into the existing library of OpenSees for analyses.18) 4. thus making it a viable choice for research purposes. 1 Italicized word is used for C++ class 62 . a 3D zerolength element1 class and a complimentary material class that can optionally include/exclude the modeling aspects of the FPS is implemented into OpenSees. In this case Equation 4. and N = fg. OpenSees is chosen as the simulation platform for this study mainly because of its ability to access its source code to incorporate new material and element models without the need to perform changes in the existing solution algorithms.8 Evaluation Platform The Open System for Earthquake Engineering Simulation (OpenSees) is an open source software framework for simulating the earthquake response of structural and geotechnical systems (Mazzoni et al. α = 0. none of them was found to be capable of adequately representing the mathematical model described previously. 2006).
Equation 4. that provides a stepbystep explanation of adding new elements or materials into the existing library of OpenSees. 63 .2.17 at each time integration step (Figure 4. and to construct the L via Equation 4.objectorienteddesign and the definitions and architecture of OpenSees framework. California gives limited yet valuable insight about these procedures (McKenna 2005a. update the values of µ via Equations 4. FPSmaterial.8. N.11). The material class receives nodal displacements and velocities as input from the element class and returns trial force and tangent stiffness values according to a predefined forcedeformation law. The element class has access to all the forces from the materials in different directions. The annually held OpenSees Developer Workshop at Berkeley.13 requires that. solve the parameters of η via the semiimplicit Rosenbrock (1963) method in Equation 4.4 and α ˆ via Equation 4. is responsible for evaluating the corresponding components of f via Equation 4.12.3 and 4.13. the FPSelement is designed to evaluate large deformation effects in N via Equation 4. is designed to interact with FPSmaterial. FPSelement. The element class is responsible for the generation of time dependent stiffness and transformation matrices and equilibrium iterations. This is accomplished by adding these parameters to the corresponding abstract method via overriding (see Appendix A). The C++ script for differential equation solution with the Rosenbrock Methods presented by Press et al. A typical structural member in OpenSees is constructed via the material and element classes. In addition to the typical parent element class in OpenSees. In addition to the strain and velocity. at each integration time step be delivered from the element class. to the author’s knowledge. The new material class developed for the FPS model. the corresponding component of η. There is no source. (2003) has been incorporated into the FPSelement class to solve the parameters of η . The new element class developed for the FPS model. McKenna 2005b). and µ .
Berkeley by Mosqueda et al. (2004) and the influence of modeling assumptions are monitored.85 m (y direction) rectangular rigid mass totaling approximately 290 kN. Structural Properties and Loads The structural model considered herein is the test setup studied at the Earthquake Engineering Research Center (EERC) headquartered at the University of California. The isolators had a concave surface with a 76. The frame was loaded with a 1. 4.9.12).9 Verification The FE model developed for the FPS in OpenSees is verified with respect to data obtained from experimental studies performed on a rigid isolated frame by Mosqueda et al.78 m (x direction) by 2. to produce a target vertical load on each isolator of 72. The objective of the test was to examine the bidirectional response of the FPS isolators.5 kN. 64 .Figure 4. The test setup consisted of four FPS isolators installed under a rigid concrete block (Figure 4. 4.2 cm radius and the µ at low and high velocities.11 FPSelement and FPSmaterial interaction. fmin and fmax. A 3D model of the experimental setup is developed in OpenSees.1. Two loading schemes comprised of unidirectional and bidirectional paths are applied to the model to illustrate the characteristics of the FPS forcedeformation response. (2004).
2 to match the value of fmax=0. the constant. The yield displacement of the isolators. The constant describing the rate of transition from low to high velocities.025 cm. This relationship is scaled by 1. The maximum displacement capacity of the isolators was ± 17. (2004).8 cm. the maximum displacement capacity of the simulator was ± 12. ε . However. that controls the variation of fmax between very low and very high pressures is not provided.13). respectively. (2004).5 by Mosqueda et al.5 kN (Figure 4. was approximately 0.11 in the Mosqueda et al.(2004) study at a normal load level of 72. However.12 Test setup studied by Mosqueda et al.were given as 0. Figure 4. 65 .12 − 0. was found to be 1. a. ∆ y .7 cm.05 and 0.11.07 tanh(εp) . Constantinuo et al. (1993) gave this relationship as fmax = 0.
3 kN 0 100 fmax 0. The loading L1a has a maximum displacement that is limited by the simulator capacity of the test setup and is used to verify only forcedeformation response per tests data. The third load path. has the maximum displacement capacity of the isolators and it is used to observe bounds of the µ on the force deformation response.10 0.7.09 N=72.14).8 and 45.05 0. 17. This path is used to monitor the range which large deformation effects become significant in the response of the isolators. L1c. L1b. (1993).0 cm. 66 . L1b and L1c sinusoidal motions with peak displacement values of 12. The frame is analyzed under displacement controlled unidirectional (L1) and bidirectional (L2) loadings.15 fmax=0.20 Constantinou et al. exceeds the maximum capacity of the isolators and it is purely theoretical. respectively (Figure 4. (1993) Scaled 0. The second loading.13 Scaling of fmax with respect to the relationship presented in Constantinou et al.11 fmax=0.00 200 N (kN) 300 400 Figure 4. The unidirectional loading (L1) is categorized as L1a.0.
± 17. The bidirectional loading.5 5. is used to verify the planar response of the isolators and monitor the influence of the effects of bidirectional sliding interaction in the response. The history of the displacements in each orthogonal direction is defined by Bsin(wt). 67 .15).0 cm 17.5 15.0 cm (L1c). where B=5 is the amplitude of the motion.5 L1a L1b L1c 20.0 2.0 7. t is the time.0 17. and w is the frequency equal to 1 in the x direction and 2 in the y direction (Figure 4.5 10. ± 12.8 cm (L1b).60 40 20 12.7 cm 45.0 t (s) 12. L2.8 cm δx (cm) 0 20 40 60 0.7 cm (L1a).14 Unidirectional load histories with amplitude.0 Figure 4. and ± 45.
85 m δy (cm) 0 10 20 30 20 15 10 5 0 5 10 15 20 δx (cm) Figure 4. 4.78 m z y x 30 20 10 2.15 Combined bidirectional loading path for L2. After applying only the gravity portion of the corresponding load case.1. All analyses are geometrically nonlinear and a NewtonRaphson solution algorithm is used for the solutions.9. The base of the model is fixed at the four corners in all directions. The mass of the rigid block is lumped at the four corners and these corners are connected via rigid beam elements. This allows for the development of the frictional forces in zerolength elements. the gravity loads are held constant and the lateral load is applied to the model. 68 .16).2 Modeling and Analysis A three dimensional (3D) model of the setup is developed in OpenSees via rigid beam elements and the zerolength element developed for the FPS (Figure 4.
(2004) reported that as a result of the rigidity of the supported block.16 Finite element model of the test setup studied by Mosqueda et al. 4.17. minor changes in the vertical alignment of the setup caused significant redistribution of the normal forces acting on the isolators.9. and in some cases complete unloading. The normalized forcedeformation history of Isolator 3 and the overall superstructure computed from the model and the experimental results for L1 is given in Figure 4.3 Results Mosqueda et al. 69 . Given the complexity of the response. (2004).Figure 4. results predicted by the model are in agreement with that reported from the experiment. The authors have also underlined the difficulty of adequately modeling the friction response of the FPS.
00 0.4 0. and (b) total isolator forces.05 0.05 0.4 0.4 0. Despite the assumption of constant N on the FPS.10 0.1 0.05 0.3 FPS fx Model Experiment 0.05 to 0.3 0. the value µ ranged between 0.0.18 shows the variation of µ for loading path L1b.4 0.1 0.2 0.20 δx / R (b) Figure 4.10 0.1 0. Figure 4.1 0.3 0.2 f /N x 0.20 δx / R (a) 0.15 0.0 0.00 0.15 0.10 0.10 0.3 FPS fx Model Experiment 0.17 Comparison of the normalized forcedeformation histories between the model and the experimental results for (a) individual Isolator 3.20 0.2 Σf /ΣN x 0.0 0.05 0.11 due to 70 .15 0.2 0.20 0.15 0.
respectively.19 also reveal that the energy dissipated per cycle under different assumptions for the value of the µ is proportionally variant as in the case of MIF.050 µ = fmin 0. The shape of the forcedeformation histories given in Figure 4.5% when the µ was assumed to be fmin and fmax . 15 20 71 . f µ . time history under loading L1b.075 0.18 Friction coefficient.150 µ = fmax 0.19. µ . The maximum isolator force (MIF) calculated by the model by accounting for velocity and pressure effects in the friction coefficient was 24. This is equivalent to a 110% variation in the frictional force component.125 0.11 is given in Figure 4. The upper and lower bounds of the isolator response as a function of fmin=0.100 µ 0. 0.025 0 5 10 t (s) Figure 4.changes in the sliding velocity. This value was overestimated by 9.05 and fmax=0.6% and underestimated by 10. of the FPS during response.02 kN.
This softening is attributed mainly to the inclusion of the geometric angle α in the calculation of the N at large displacements. Isolator response under L1c with the small deformations model (SDM) and the large deformations model (LDM) are given in Figure 4.3 0.20.05 0.1 0. µ . µ =fmax.0.15 0.00 0. Inclusion of the large deformation effect reduced the MIF at theoretical displacements that exceed the isolator deformation capacity.1 f /N x 0. The difference in the MIF between the LDM and the SDM for this particular case was less than 4% at the maximum displacements of L1c loading.18.2 0.19 Comparison of the forcedeformation histories of the FPS with theoretically exact value of the friction coefficient.2 0.and µ =fmin for loading L1b.4 0.0 0.05 0.20 δx / R Figure 4. It is observed that the difference between the SDM and the LDM is negligible under even the maximum isolator displacement capacities which correspond to δx/R=0.20 0.10 0.10 0.3 µ = exact µ=fmax µ=fmin FPS fx 0. 72 .15 0.4 0.
2 0.22.50.4 0.20 Comparison of the response of the small deformations model (SDM) and the large deformations model (LDM).2 f /N x 0.1 0.5 kN with the coupled modeled was overestimated as 29.6 0.4 kN (20% difference) with the uncoupled model.40.0 0.8 0.18 0.5 0.21. The variation of the components of the instantaneous velocity in the orthogonal directions had significant influence on the force history in the x.8 0.18 . 73 δx / R = 0.60.LDM model SDM model 0.0 0.6 Softening δx / R = 0.6 δx / R Figure 4.3 0. The MIF of 24.30.1 0. The forcedeformation histories and the combined resisting force paths in the two orthogonal directions of the combined isolators under L2 are given in Figures 4. A comparison of this response with an uncoupled model is presented in Figure 4.4 0. It is observed that.2 0.4 0. the uncoupled model deviates significantly from the path of the coupled model as the resisting forces change directions.direction.20. This figure shows that the force paths in the orthogonal directions predicted by the coupled model are in agreement with test results. This is attributed mainly to the variation of the coupled parameters of the η varying continuously as a function of time.
1 0.5 Σf /ΣN x Figure 4.2 0.0 0. and coupled model.5 Σf /ΣN x Figure 4.1 0.4 0.4 0.1 0.5 0.1 0.5 0.1 0.4 0.0 0.3 0.2 0.1 0.3 0. Σf /ΣN y 74 .5 0.2 0.5 0.5 0.4 0.3 0.5 0.21 Comparison of the bidirectional resisting forces for the experimental data.3 0.3 0.4 0.0 0.2 0.2 0.0 0.4 0.2 0.3 0.1 0.4 0. Σf /ΣN y Coupled model Uncoupled model 0.22 Comparison of the bidirectional resisting forces for the coupled model and the uncoupled models.2 0.4 0.3 0.1 0.Model Experiment 0.2 0.3 0.
(2004) under both unidirectional and bidirectional loadings.10 Conclusion In this chapter. The influence of neglecting certain modeling aspects of the FPS response has been presented. A new 3D zerolength FE of the FPS has been developed in OpenSees and verified using experimental data. 75 . (2) The FE model developed in OpenSees for the response of the FPS provides good agreement with experimental findings of Mosqueda et al. the simplified bilinear modeling of the Friction Pendulum System (FPS) has been explained. The simplifying assumptions in this approach and how aspects pertaining to these simplifications can be represented in the finite element (FE) model of the FPS has been highlighted. These affects need to be quantified for bridge applications. An FE model to fill this gap has been developed and compiled from existing research and implemented as a new element into OpenSees. The following conclusions can be drawn from this chapter: (1) There exists sufficient theoretical and experimental findings on different modeling aspects of the FPS. (3) It was observed that the different assumptions in modeling the FPS caused significant variations in the paths of the forcedeformation histories of the isolators. Different assumptions pertaining to the modeling of structures seismically isolated with the FPS may lead to over or under design of isolators. however a comprehensive FE model that combines these effects is absent.4.
followed by 76 .∆ effects. P. inplane bidirectional sliding interaction. Seven models of a threedimensional (3D) MultiSpan Continuous (MSC) Steel Girder bridge with different assumptions of the FPS are generated. N.1). The selection and detailed modeling of the bridges considered for seismic isolation with the FPS are presented. 5. and the orientation of the FPS isolators are highlighted. µ . large deformation. Nielson (2005) further performed the fragility assessment of the classes of bridges and concluded that the MSC and MSSS Steel Girder bridges were among the most vulnerable to damage. and coefficient of friction.2 Selection of the Class of Highway Bridges for Seismic Isolation and Analyses A detailed survey of 163. Maximum normalized force (MNF) and deformation (MND) of the isolators and column drifts are used as the parameters to characterize the response of the models. Nonlinear time history analyses (NLTH) are performed for the bridge to examine the effect of modeling parameters of the FPS on the response.1 Introduction This chapter investigates the response of typical highway bridges isolated with the Friction Pendulum System (FPS) as a function of isolator modeling assumptions.433 bridges in the Central and Southeastern US (CSUS) was performed by Nielson (2005). The influence of the variations in isolator normal force.CHAPTER 5 BRIDGE RESPONSE AS A FUNCTION OF ISOLATOR MODELING ASSUMPTIONS 5. The results of this study showed that MultiSpan Continuous (MSC) Steel Girder and MultiSpan Simply Supported (MSSS) Steel Girder bridges and MSC and MSSS Concrete Girder bridges are among the most common classes of bridges found in the CSUS inventory (Figure 5.
This model includes material and geometric nonlinearities. The 3D SIB model was developed in OpenSees.2 and 5.the MSC and MSSS Concrete Girder bridges. the modeling of the steel and concrete bridges are similar. The 77 . 1996). Previous research identified significant vulnerabilities of the steel fixed and rocker bearings employed in these bridges to seismic loads (Mander et al.1 Picture of example MSC Steel Girder Highway Bridge (Nielson 2005). Figure 5. 5.3 Seismically Isolated Bridge (SIB) Modeling The bridge type selected for the NLTH analysis in this chapter is an MSC Steel Girder Bridge seismically isolated with FPS isolators (Nielson 2005). Assuming that the superstructure of SIBs remains within the elastic range. Seismic isolation of these bridges via replacing the existing steel bearings with the Friction Pendulum System (FPS) may be an effective tool for improving the earthquake performance. The geometry and modeling approach for the bridge is illustrated in Figures 5. The seismic isolation of the bridge is achieved via placing FPS isolators under each of the eight girders above the piers and abutments. The bridge has three spans and a continuous slabongirder deck with a total of eight steel girders. Particular emphasis is given to these highway bridges in subsequent sections of this dissertation.3.
The slider diameter has 7. The superstructure is expected to remain within the linear elastic range.103 for the unconfined case and 28.018 is used for this material (Figure 5. thus. ε c . fmax = 0.5 kN/mm and (6. fys = 414 MPa. The unconfined and confined concrete behavior is modeled via the KentScottPark model which utilizes a degraded linear uploading/reloading stiffness and a residual stress.4). The horizontal.025 cm.width of the expansion joints at the abutments is 7. Structural damping is assumed to be 5%. and associated strain. kr.062)103 for the confined case.5 MPa and (2.12 and fmin = 0. and an elastic modulus. 1993). A strain hardening ratio of 0.44 m square and use eight piles.05 (Mosqueda et al. 78 .4).5 second fundamental period which corresponds to R = 99 cm with an inplane displacement capacity of 23 cm. The concrete compressive strength. The isolators are assumed to be positioned as the concave dish at the top. using the composite section properties.7 cm. Es = 200 GPa.6 MPa and 2. The bridge has footings which are 2. kt. respectively (Figure 5. Constantinou et al. and rotational. 2004.02.012 MPa1. are 27. respectively. The characteristic properties of the µ are selected as: a = 59.7 cm to ensure pressures below 275 MPa under gravity and earthquake loads in accordance with the recommendations of the manufacturer. The reinforcing steel is modeled as a bilinear material with a yield strength.06)105 kNm/rad. The FPS isolators are selected to achieve approximately a 2.1 s/m. the deck elements are modeled using elastic beam column elements. fc. The section properties for the columns and the bent beams are created using fiber elements with appropriate constitutive models for both the concrete and the steel reinforcement. stiffnesses of the foundation are 130. ∆s = 0. ε = 0.
3 Pier configuration and bent and column discretization.Figure 5.2 Multispan continuous (MSC) steel girder bridge general elevation and modeling details. Figure 5. 79 .
The fourth model is developed to identify the influence of the bidirectional coupling in estimating 80 .4 FPS Models Seven SIB models are generated with the above properties where the only difference is in the FPS modeling assumptions. with uncoupled bidirectional sliding forces and small deformation assumptions.(a) (b) Figure 5.e. and (b) concrete material.4 Constitutive relationships for the modeling of (a) steel material..∆ effects. the constant value of N is taken as the corresponding value after gravity load analysis and µ as 0. The second model is a simplified bilinear model that is insensitive to the variations in N and µ . accounts for the variations of the N and µ . It is the same model as Model 1 with the only difference of assuming a constant N of the corresponding value after gravity load analysis. The third model is developed to monitor the influence of not accounting for the variations of N on the response of the FPS. The first model is theoretically exact. In Model 2. has bidirectional coupling of the sliding forces and incorporates P. i. 5.07.
The fifth model is developed to monitor the influence of not accounting for the inclination due to the concavity in the FPS.05 to 0. The seventh model.01 81 . Model 7 is established with the same principles as Model 1 with the only difference of having a µ that is constant. is developed to monitor the influence of the assumptions on the value of µ .12 with increments of 0.the response of the FPS. The properties of the models are summarized in Table 5. insensitive to variations in pressure and sliding velocity.1 Summary of model properties Model1 Modeling aspect 1 2 3 4 5 Normal force x x x Bidirectional coupling x x x Large deformations x x x Fiction coefficient x x x x 1 2 62 x x x x 73 x x x ‘x’ denotes exact modeling Concave dish of the FPS at the bottom 3 Seven models with value of µ ranging from 0. Model 6 is same as Model 1 with the only difference being that the FPS isolators are positioned with the concave dish at the bottom ˆ which is accommodated as the corresponding sign shift in the L in Equation 4. This is achieved by computing f with Equation 4.18 .e.12 with increments of 0. It is the same model as Model 1 with the only difference of assuming the orthogonal sliding forces of the FPS isolators to be uncoupled. Table 5.17. Model 7 is discussed separately from the other models and analyzed for a constant value of µ ranging from 0. The sixth model is generated to identify the influence of the FPS orientation. This is & & achieved by assuming η = [ sgn(δ x ) sgn(δ y ) ]T.05 to 0.1.01. i.
The equations of motion were solved The time interval for solving the numerically using the NewtonRaphson method. OpenSees allows the user to select the integration technique and solution algorithm for the analysis. The first three modes of vibration are those involving the isolation system which shows that the characteristics and the design of the FPS isolators govern the dynamic response of the bridge (Figure 5. was used in integrating the nonlinear dynamic equilibrium equations. Figure 5.5).5 Dynamic Analyses The modal properties of the SIB in Model 1 are established by assigning linear effective stiffness to the FPS isolators.22 s.93 s. and 1. the second mode is transverse and the third mode is torsional. A suite of ten earthquake 82 . The first mode is longitudinal. which is an unconditionally stable numerical integration algorithm.005 s.5 Mode shapes of the deck. equations of motion was taken as 0.5. An important recommendation by the bridge engineering community is the use of design earthquakes that have a 2% probability of exceedance in 50 years (an earthquake with a mean recurrence interval of 2475 years) (FEMA 1997). Newmark’s average acceleration timestepping scheme. 2. respectively. Seismically isolated bridge (SIB) models were subjected to NLTH analyses.15 s. The first three modal periods of the SIB are 2.
The geometric mean of the longitudinal and transverse component of each record is scaled to match the spectral value of 0. Canada 1985/12/23 0.767 3.00 1.records from rock sites is used in the NLTH analysis of the bridges (Table 5.2).0 0.654 1.932 0. TN.5 0.75 1.50 1. No.978 1.173 0.552 Spectral acceleration.608 1.118g 2% in 50 years Median Figure 5.263 0.248 0.75 2.453 0.421 1.492 0.00 0.264 N. Montana 1935/10/31 0.6 Response spectrums for the suite of ground motions. 1 2 3 4 5 6 7 8 9 10 Table 5.096 2.75 3.787 0.5 3.785 0.199 0.50 0.25 2. The response spectra of the scaled ground motion records for 5% damping.00 2.718 0.721 0.304 0. and their median are given in Figure 5. Sa=0.5 2.585 1.209 Gazli.229 Whittier Narrows 1987/10/01 0.818 Scale 7.6.5 1.25 0.22 s.2 Ground motion suite Component PGA (g) Earthquake record Longitudinal Transverse Vertical Morgan Hill 1984/04/24 0.092 Northridge 1994/01/17 1.102 Loma Prieta 1989/10/18 0.279 1. 83 .501 0.0 2.00 Period.473 0.22 s corresponding to a 2% probability of exceedance in 50 years hazard level earthquake in Memphis.0 1. T (s) T=2.285 1.086 Landers 1992/06/28 0.50 2. USSR 1976/05/17 0.434 3.227 Loma Prieta 1989/10/18 0.411 0.471 Helena.118g at a period of 2. Palm Springs 1986/07/08 0.15 0.069 0.612 0.507 Nahanni.25 1.0 0.098 0. ξ. Sa (g) for ξ=5% 3.
The SIB models were first analyzed for gravity loads and sequentially subjected to NLTH analyses using simultaneously the longitudinal. The main response quantities monitored for the FPS isolators are the maximum normalized force. and the maximum normalized displacement. the results are presented for one of the isolators on top of the piers and the abutments and one of the columns. MNF = max( f L + f T / N o ). The inplane orthogonal components of the earthquakes are oriented to result in the maximum demands on the columns for all cases. The structural response of the isolators and columns along the same transverse axis were essentially the same. where fL is the longitudinal and fT is the transverse 2 2 isolator force.The three components of the acceleration histories of each scaled ground motion are applied to the SIB models (Figure 5. MND = 84 . transverse and vertical acceleration records of the given earthquake. üL (t) Vertical Longitudinal üT (t) Transverse üV (t) Figure 5. respectively.7). Therefore.7 Orientation of the 3D bridge model. No. of approximately 125 kN and 258 kN respectively (neglecting the normal load variation between the isolators at the exterior and the interior ends at the same pier and abutment). It is found from the gravity load analysis that each isolator above the pier and the abutments carry a gravity load .
uplift took place between the sliding surfaces of the FPS isolators in the vertical direction for all of the records except for the Loma Prieta. Helena and Landers. This ratio was not exceeded during any of the NLTH analyses. This indicates that pounding would occur between the abutment and the deck in the longitudinal direction. 5. The maximum allowable N is limited by the allowable pressure of 310 MPa on the slider.max( δ L + δ T /R ).6 Results It was observed from the NLTH analyses of Model 1 that the maximum allowable displacements at the expansion joints were exceeded in an all records except Morgan Hill. for the FPS above the pier. This substantial increase is indicative of a proportional increase in the postyield stiffness and yield force of the isolator. however.8 that the contact between the two sliding surfaces was lost at least once which resulted in N/No=0. Gazli and Nahanni. It is observed from Figure 5. during the Nahanni earthquake a peak value of N/No=3. This uplift caused instantaneous yet complete loss of stiffness of the isolators during the earthquakes. where δ L is the longitudinal and δT is the transverse isolator 2 2 displacement. Maximum column drifts. The timehistory of the N/No of the Model 1 FPS isolator for the Nahanni earthquake is given in Figure 5. due to the indeterminacy of the model there was no instability.51 was reached. Additionally. dmax. respectively. The impact forces in the deck are difficult to correlate to damage levels and may impede the utilization of the full capacity of the isolators. 85 .4. However. are selected as the response quantity to monitor the structural demands on the SIB. which corresponds to N/No=5.8.
This implies a stiffer isolator response when the 86 .0 0 2 4 max(N/No)=3.5 0. the peak isolator force was overestimated and the peak isolator deformation was underestimated.0 0. bidirectional effects are neglected. Although Model 4 was able to account for the significant variations in isolator forces. These two aspects of the isolator response could not be observed in Model 2.N FPS 4.5 2.0 1.0 6 8 10 12 14 16 18 20 22 24 Time (s) Figure 5.8 Time history of the N/No for the FPS during the Nahanni earthquake NLTH analysis. Model 2 These underestimated both the MNF and the MND in comparison to Model 1.51 N/No min(N/No)=0.0 2.0 3.5 3.5 1.9 shows the normalized forcedeformation (NFND) histories of the FPS isolators on top of the piers among Models 1 to 4 in the longitudinal direction of the bridge during the N. Model 3 was unable to capture peak isolator forces indicating that the normal components of the ground motion were influential in this response quantity. differences between Model 1 and Model 2 NFND histories can be explained by the response observed in Models 3 and 4. Additionally. Figure 5. Model 1 can capture the abrupt changes in isolator force and instances of uplift in the vertical direction. Palm Springs record.
MND and dmax for the suite of ground motions is illustrated via box plots given in Figure 5.0 0.05 0. Palm Springs earthquake record with (a) Model 1 (b) Model 2 (c) Model 3 and Model 4.0 0.10 0.05 0.2 max( δ / R  ) = 0.20 0.2 0.2 max( δ / R  ) = 0.096 L 0.365 f / No L max(  fL / No  ) = 0.2 max( δ / R  ) = 0.180 0.2 f / No L 0.20 0.00 0.05 0.0 0.6 (b) 0.00 0.112 0.20 0.4 max(  fL / No  ) = 0.05 0.15 0.4 0.10 0.4 max(  fL / No  )= 0.10 0.00 0.0 0.6 0.10 0. This information provides an overview of the expected demands on the isolators 87 .195 0.102 L 0.2 0.4 FPS fL 0.15 0.10 0.2 f / No L 0.0. Box plots are a useful way of presenting the graphical description of variability of data (Montgomery 2005).15 0. The influence of the isolator modeling parameters on MNF.6 0.410 FPS fL 0.05 0.15 0.118 L 0.15 0.6 0.20 δL / R δL / R (c) (d) Figure 5.4 max(  fL / No  ) = 0.05 0.6 0.6 0.20 0.2 0.15 0.10.15 0.20 δL / R δL / R (a) 0.4 FPS fL 0.10 0.4 FPS fL f / No L 0.6 0.05 0.00 0.15 0.20 0.4 max( δL / R  ) = 0.10 0.10 0.20 0.9 Forcedeformation history of the FPS in the longitudinal directions on top of the pier for the N.6 0.05 0.
It is concluded that large deformation effects associated with the orientation and exact kinematics were not significant in the response where average MND was smaller than 0. Model 2 attained a general increase in the median of the dmax. The statistical interpretation of the results are presented with numerical values of the median and plots of the 10th. 25th. and 90th percentile cumulative probabilities. 75th. It is observed that Model 2 underestimated the median of the MNF by 20% and the peak MNF as 44% of Model 1. However. MND and dmax approximately the same as Model 1 for the whole suite of ground motion records. This effect is elaborated in the subsequent section.and the structural system as well as the scatter in the results. there was a notable variation in the median values of the MND as a function of the magnitude of the constant value of µ. 88 . The absence of the variability in the dmax with Model 2 stems from the inability to account for normal force variations on the isolators. It was observed that not including the influence of the variations in the normal forces acting on the isolators causes a loss in the variability of the MNF results.20. Another factor that contributed to the increase in the median of the dmax in Model 2 was not accounting for the variation of the µ. which is attributed to the overestimation of the stiffness caused by uncoupled response in the orthogonal directions of the isolator lateral motion. The exclusion of the exact kinematics pertaining to the concavity of the FPS in Model 6 was insignificant since the MND was limited to 0. Model 2 overestimated the median of the dmax by 12% and underestimated the peak dmax as 69% of Model 1.19 for the suite of ground motions. On the other hand. 10th. Models 5 and 6 predicted the MNF. Similar results were observed for Model 3 which indicated that the normal components of the force are influential in design level isolator forces. Peak MND values for Models 2 to 7 had negligible difference with Model 1.
505 0.086 1 2 3 4 Model 5 6 µ=0.12 7 (b) 1.263 0.8 0.8 MNF 0.10 The influence of modeling assumptions on (a) MNF.150 0.6 0.2 0. and (c) dmax.4 0.208 0.260 0.099 MND 0.025 0.098 0.526 0.4 0.263 0. (b) MND.260 0.125 0.2 1.12 7 (a) 0.095 0. 89 .094 0.0 dmax (%) 0.05 µ=0.1.469 0.100 0.513 0.075 0.555 0.200 0.05 µ=0.2 0.000 0.131 0.362 0.260 0.469 0.05 µ=0.0 0.095 0.469 0.495 1 2 3 4 Model 5 6 µ=0.095 0.0 0.182 0.6 0.0 1 2 3 4 Model 5 6 µ=0.12 7 (c) Figure 5.050 0.175 0.
The peak and median of the dmax attain optimal values at µ = 0.09. 90 . Although the median MND generally decreases with the decreasing values of µ. It is observed that both the median and the peak MNF for the suite of ground motions increase consistently with increasing values of µ. the peak MND remains essentially the same.The influence of different magnitudes of µ in Model 7 on MNF. respectively.08 and µ = 0. there is a notable increase in the variability of MND with increasing values of µ. The peak and median of the dmax are for all values of µ overestimated by Model 7 in comparison to Model 1. MND and dmax for the suite of ground motions is illustrated via box plots given in Figure 5. However.11.
504 0.06 0.12 µ (a) 0.488 0.07 0.08 0.06 0.555 0.123 0.6 0.07 0.362 0.150 0.479 0. (b) MND.06 0.495 0.4 0.10 0.2 0.317 0.11 0.086 0.095 0.292 0.514 0. 91 .0 dmax (%) 0.175 0.8 0.0 0.11 0.09 0.09 0.11 The influence of constant value of µ assumptions on (a) MNF.12 µ (b) 1.12 µ (c) Figure 5.05 0.4 0.0 0. and (c) dmax.200 0.05 0.075 0.8 MNF 0.298 0.275 0.11 0.334 0.131 MND 0.08 0.094 0.08 0.490 0.000 0.289 0.0 0.025 0.2 0.100 0.10 0.09 0.556 0.125 0.337 0.6 0.1.111 0.2 1.10 0.07 0.100 0.050 0.05 0.096 0.
00 0. This indicates that the deck engaged into torsional vibration. ΣMNFabutment.Smaller No developed in the FPS isolators at the abutments than at the piers due to the difference in the corresponding tributary mass of the superstructure. MNDabutment for the suite of ground motions. On the other hand MNDpier were approximately 16% less than MNDabutment on the median.00 1.25 0.00 2. MNDpier. Figure 5. ΣMNFpier.00 0.75 1. and MND on top of the pier.50 0.75 0.50 0. Abutments may be further engaged into resisting earthquake induced loads in SIB by designing the FPS isolators at the abutments with higher stiffness properties. and the total MNF transferred to the abutment. (b) MND on top of the pier.50 2.75 1.25 1. and abutments.50 1.25 1.25 2.50 2.75 2.12 shows the comparison of the total MNF transferred to the pier. ΣMNFabutment.00 ΣMNFpier / ΣMNFabutment MNDpier / MND abutment 0. MNDabutments. It is observed that the isolators transferred almost twice as large as the force to the piers in comparison to the abutments on the median.25 0. 3.12 Comparison of the (a) total MNF transferred to the pier.50 1.75 0. ΣMNFpier.84 (a) (b) Figure 5.97 2.00 1. 92 . and abutments. MNDpier. and the total MNF transferred to the abutment.25 2.00 1.
bidirectional coupling and the variable magnitude of the friction coefficient. (4) The incorporation of the effects of orientation and the exact concave geometry of the FPS in to the response had negligible affects. Abutment forces were twice of those at the piers in the 93 . (2) The uplift and pounding of the deck in the vertical direction had notable affects in the response of the FPS that in one case caused an increase of up to 3. the modeling of a typical highway bridge seismically isolated with the FPS has been presented. However. (3) Excluding the bidirectional coupling of the FPS isolators generally resulted in overestimating the isolator maximum normalized forces (MNF) and underestimating the isolator maximum normalized displacements (MND). effects in a seismically isolated multispan continuous (MSC) steel girder bridge has been highlighted via nonlinear timehistory (NLTH) analyses. N. This indicates an overestimation of the stiffness of the isolators. The influence of FPS modeling assumptions on normal force.7 Conclusion In this chapter. This model underestimated the maximum column drifts (dmax) by up to 31%. This was mainly a result of not accounting for the effects of vertical components of ground motions. (6) The structural demands transferred by the isolators to the abutments and the piers were significantly different. The following conclusions are made: (1) The simplified bilinear idealization of the FPS response was unable to capture the variability in the results. P. orthogonal coupling and large deformation.51 times in the initial gravity load acting on the isolators (No). This is mainly a result of the MND remaining under 0. (5) The peak MND of the isolators among the suite of ground motions acquired negligible variations among all the modeling assumptions.20 for the suite of ground motions. the median MND was influenced by the assumptions in the magnitude of µ .∆ . and friction coefficient.5. µ .
median and the isolators acquired 18% larger deformations in the median at the abutments in comparison to those at the piers. 94 . This is a results of the uneven distribution of isolator stiffness properties along the bridge as a function of deck tributary mass.
Maximum isolator forces and displacements.1 Introduction This chapter compares sliding versus elastomeric seismic isolation of a typical MultiSpan Continuous Concrete Girder (MSCCG) bridge with advanced isolator models. isolated in one case with the LRB and in another case with the FPS. In spite of existing research findings on the dependency of LRB inplane response to the magnitude of inplane deformation and normal load. and column drifts are selected as response quantities. however there exists considerable differences in their mechanisms.CHAPTER 6 COMPARATIVE ASSESSMENT OF SLIDING VERSUS ELASTOMERIC SEISMIC ISOLATION FOR TYPICAL MULTISPAN BRIDGES 6. is performed for a hazard level of 7% in 75 years using a nonlinear threedimensional (3D) analytical model. Particular emphasis is given to the distinct vertical load dependency modeling of the isolators.2 The Lead Rubber Bearing (LRB) The Lead Rubber Bearing (LRB) was invented in April 1975 by W H Robinson (Skinner et al. respectively. A seismic evaluation of the bridge. The LRB is one of the most commonly used elastomeric isolator 95 . existing nonlinear models do not account for these effects. 1993). Isolators serve the common objective of lengthening the period of the structure and providing additional energy dissipation. The Friction Pendulum System (FPS) and the Lead Rubber Bearing (LRB) are selected as representative examples of sliding and elastomeric isolators. A detailed isolator model for the LRB that can account for the inplane and vertical coupling of the response is developed in OpenSees. 6.
It has been also reported based on information provided by the manufacturers that the cost. Analytical and experimental research and observed performances of bridges during earthquakes isolated with the LRB showed that the LRB may have a substantial impact on improving the structural performance of bridges prone to seismic loads (Jangid 2004. Ghobarah 1988). Additionally. 2002. Lee et al. size and energy dissipation of the FPS and LRB may be comparable in bridge applications (Dicleli. International 2000). M 2002).types in bridges (Buckle and Mayes 1990) (Figure 6. both isolators have been incorporated into the design codes (AASHTO 1999. Kelly and Buckle 1986. DIS 1996. 96 .1).
Mexico (courtesy of Dynamic Isolation Systems). Califonia (b) Patria Acueducto.(a) (b) Figure 6.1 Examples of LRB applications (a) Rio Vista Bridge. 97 .
Additionally. Figure 6.2). rubber and a lead core (Figure 6. 1996).3 Effects of geometrical variations of the LRB on the forcedeformation response (Priestley et al. The steel layers placed between the rubber serves to limit the edgebulging of the rubber (Tyler 1991). the steel plates force the lead plug bearing to deform in shear (Naeim and Kelly 1996). 1993). and has good fatigue properties (Skinner et al. 10 MPa. AASHTO 1999. The LRB provides the advantage of attaining versatile forcedeformation characteristics via the geometrical variations of both the lead core and the rubber (see Figure 6. Naeim and Kelly 1996). The forcedeformation response of the LRB is typically idealized as bilinear (Ghobarah 1987.The LRB consists of steel plates. 1996).3) (Priestley et al.2 The Lead Rubber Bearing (LRB) interior elevation. Lead is a feasible option because it yields in shear at relatively low stresses. A lead core is inserted in the center of the bearing for energy dissipation and stiffness (Priestley et al. 98 Shear Force . 1996). Increasing lead core diameter Deformation Figure 6.
tr is the total thickness of the rubber. A is the crosssectional area of the isolator and (EI)eff is the effective rigidity.5) where stiffness kpo is the nominal (meaning absent of N effects) postyield stiffness.3 ForceDeformation Characteristics of the LRB As a multilayered elastomeric type bearing.2) and Pe = π 2 ( EI ) eff tr 2 (6. The critical buckling load capacity of the LRB reduces as the isolator deforms (Buckle and Liu 1993): δ Pcr = Pcro 1 − D (6. The LRB response softens and yield force increases with increasing level of N (Ryan and Chopra 2005).3) where G is the shear modulus of the rubber.LRB N = k po 1 − Pcr 2 (6. The critical buckling load established from the elastic theory in undisplaced position is (Naeim and Kelly 1996): Pcr o = Ps Pe (6. If the large deformation (P∆) effects described in Chapter 4 are neglected.6. the LRB is susceptible to a buckling type of instability (Kelly 1997).4) where D is the diameter of the isolator and δ is the lateral bearing displacement. the post yield stiffness of the FPS is: 99 . The postyield stiffness of the LRB as a function of the N is (Buckle and Kelly 1986): k p.1) where Ps = GA (6.
LRB = FY o 1 − e (− N/Po ) [ ] (6. δe is the elastic deformation with a () and (+) sign for tension and compression. is typically estimated as 10kp.6) where FY o is the nominal yield strength of the isolator which can be computed from the yield stress of the lead core. The yield strength of the FPS is (Earthquake 2003): FY.k p.FPS = N R The yield force of the LRB was observed to not achieve the theoretical strength under low N (Hwang and Hsu 2000). Mori et al.LRB. This behavior closely matches the response of a zero length symmetrically elastic element bound by Pcr under compression and ε c under tension: k δ for buckling < δ e ≤ 0 fe = e e − k e δe for 0 < δ e < cavitation (6. The stiffness of the LRB remains elastic under tension with the same magnitude as in compression until reaching cavitation at strains ε c = 1 / ( 2 S 2 ) . The initial elastic stiffness of the LRB.LRB (Naeim and Kelly 1996). respectively. where S is the shape factor of the isolator (Kelly 2003. The yield force of the LRB is approximated as (Ryan and Chopra 2005): FY. and Po is the normal load corresponding to approximately 63% of nominal strength.9) where ke is the vertical stiffness of the elastic element equal to that of the isolator.8) where Ec is the compression modulus and h is the total isolator height. 100 .7) The LRB compression stiffness is (Naeim and Kelly 1996): kv = Ec A h (6. ki. 1996). The yield displacement of elastomeric isolators is typically larger compared to sliding isolators (Matsagar and Jangid 2004).FPS = N µ (6.
6. respectively. In retrospect.10) 101 . y.LRB is constant. The only difference is in evaluating the postyield stiffness. It is possible to extend the model developed in Chapter 4 for the FPS to represent the nonlinear and coupled response of the LRB using the relationships described in the previous section. and z directions for the FPS and the LRB are: f = fx [ fy fz ] = [k T p δ x + FY η x k p δ y + FY η y N ]T (6. yield force. and (4) isolator deformations are small. Consequently. the general forcedeformation relationship in the x. The hysteretic forcedeformation response of the LRB is modeled by implementing a 3D zerolength element class and a complimentary material class in OpenSees.4 Modeling of the Isolator Response The bilinear idealization of the LRB forcedeformation response is based on similar simplifications adopted for the bilinear idealization of the FPS response: (1) N is constant. (3) the horizontal response is uncoupled in the orthogonal directions.This is a similar approach adopted for modeling of the FPS vertical response in Chapter 4.5). The large deformation affects are assumed to be negligible (as described in Chapter 5). It has been shown that excluding the inplane coupling of the orthogonal response for the LRB may result in significant underestimation of the displacements and forces (Jangid 2004). The FPS model developed in Chapter 4 is used in the subsequent analyses with the only additional assumption of neglecting largedeformation effects.5) and (6. this interaction is included in the response of the LRB using the methodology described in Chapter 4. and vertical response from Equations (6. (2) FY.9). The procedure for this approach is essentially the same as the FPS modeling described in Chapter 4. (6.
4 Isolator model.6 GPa. The total weight of the superstructure is W=12272 kN with a moment of inertia about the strong and weak axes as Iy=75.03 m4 and Ix=0. Figure 6. Additionally.12 m4 and modulus of elasticity of Ec=25. 6. The superstructure is comprised of a continuous slabongirder concrete deck with a total of eight concrete girders. the pounding between the deck and 102 .where f and δ are the isolator forces and displacements in the direction denoted with the subscript.5 Bridge Model The MultiSpan Continuous Concrete Girder (MSCCG) bridge considered in this Chapter is essentially the same with the one used in Chapter 5 with the only difference in the geometric and material properties of the superstructure.4). Fy is the yield force described by corresponding Equations in the previous section for the two isolators (Figure 6. The superstructure is expected to remain within the linear elastic range and modeled as a beam element. kp is the postyield stiffness.
The LRB design properties are chosen by considering: (1) a vertical load capacity of at least three times the initial gravity load. Assuming an equivalent distribution of superstructure weight on 32 isolators. According to AASHTO (1999) the isolation period is to be determined from the isolation system effective stiffness based on maximum design displacement.1 kN/cm (4) a yield force of approximately the same value to the average FPS to acquire a comparative seismic isolation scheme. lead core diameter. the stiffness of a single isolator is k i = 4. No. Additionally. It is assumed that a single isolator size is used throughout the bridge to save on the cost of an extra mold.6 cm. An approximate FPS design to match this ki has R=99 cm and µ = 0. the kp of the isolators are more influential for the seismic response of bridges compared to the initial stiffness (Dicleli and Buddaram 2005).8 cm.1 kN/cm with a static weight of Wi = 387 kN.the abutments are neglected to focus on the differences in the response of the two isolators. D=35.6 Bridge Seismic Isolation The seismic isolation of the bridge is achieved via placing isolators under each of the eight girders above the piers and abutments. 6. (2) a shape factor S>8 (3) a postyield stiffness of approximately 4.7 kN/cm. The following are the properties selected to achieve these design objectives: bearing diameter. Dc=5. single 103 . Due to the absence of maximum design displacement the period is determined with the kp of the isolation system. Details of the substructure modeling of the bridge are given in Chapter 5. The slider diameter is Ds=12.05 . To achieve a target isolation period of approximately T=2 seconds for a onedegreeoffreedom mass of W=12272 kN the stiffness is 123. The effective stiffness and kp are correlative and typically close (Jangid 2004).7 cm.
Each isolator above the pier and the abutments carry a gravity load. n=17. The variation of the Pcr as a function of δ is given in Figure 6. using the design parameters described above. t=0.585 cm respectively. as a function of isolator inplane deformation.5 Variation of the buckling load.layer thickness for rubber.026 cm and 0. for the LRB. This design has an S=8. of approximately 258 kN and 104 . Fy.6 cm Pcro=1600 kN δ (cm) Figure 6. The tributary mass supported by each isolator varies once they are installed into the bridge and subjected to gravity loading.76 MPa of rubber.6 and 6. The two systems possess a yield force of approximately Fy=19 kN for N=Wi. of the two isolators as a function of the N are given in Figures 6. G=0. δ. The yield displacements for the FPS and the LRB are assumed to be at 0. shear modulus. 2000 1750 1500 Pcr (kN) 1250 1000 750 500 250 0 0 8 16 24 32 40 D=35.7 respectively. No.7 and a Pcro=1600 kN. Pcr.953 cm. The yield force.5. number of layers. kp. and postyield stiffness. Two models of the bridge are generated and one is isolated with the LRB and the other is isolated with the FPS.
512 kN respectively (neglecting the load variation between the bearings at the exterior and the interior ends at the same pier and abutment). The corresponding bilinear forcedeformation idealization of the two isolation systems after gravity loading is given in Figures 6.8 and 6.9 respectively. It is observed that the FPS sustains a considerable variation of the Fy and kp throughout the longitudinal axis of the bridge due to corresponding tributary gravity load from the superstructure. The Fy and kp of the FPS isolators above the piers become twice in magnitude of those above the abutments. Consequently, the isolators above the piers become stiffer and the ones atop the abutments become more flexible compared to the initial design with Wi. Since the rate of dependency of the LRB response to N is weaker compared to FPS, the variation of the Fy and kp located at different parts of the bridge are smaller. Another notable distinction is that the LRB isolators become stiffer at the abutments and more flexible at the piers compared to design with Wi because the kp is inversely proportional to N in Equation 6.5.
LRB FPS
75 Yield force, Fy (kN) 60 45 30 15 0 0
Force
Fy
Deformation
Fy(N=387 kN) = 11.2 kN 300 600 900 1200 1500 1800
Normal force, N (kN)
Figure 6.6 Variation of isolator yield force, Fy, as a function of applied compressive axial load, N.
105
LRB FPS
14 Postyield stiffness, kp (kN/cm) 12
Force
kp
Deformation
10 8 6 4 2 0 0 300 600 900 1200 1500 1800 Normal force, N (kN) kp(N=387 kN) = 4.1 kN/cm
Figure 6.7 Variation of isolator postyield stiffness, kp, as a function of applied compressive axial load, N.
90 75 Force, F (kN) 60 45 30 15 0 0 2
Isolators at abutments Isolators at piers
kp
17 = 5.
kp
cm kN/
25 F y=
.8 k
ip
∆y=0.26 cm
3 = 4.0
m kN/c
.6 kN = 19 Fy
1 k p= 2.6
kN/cm
.9 F y= 12
kip
Design (N=387 kN) Piers (N=512 kN) Abutments (N=258 kN)
4
6
8
10
12
Deformation, d (cm)
Figure 6.8 Bilinear idealizations of the FPS forcedeformation characteristics after gravity loading in the bridge.
106
90 75 Force, F (kN) 60
Isolators at abutments Isolators at piers
kp= 3.68 kN/cm Fy= 20.24 kN kp= 4.03 kN/cm Fy= 19.58 kN
kp= 4.30 kN/cm
45 Fy= 17.80 kN 30 15 0 0 2 4 6 8 10 12 Deformation, d (cm)
∆y=054 cm
Design (N=387 kN) Piers (N=512 kN) Abutments (N=258 kN)
Figure 6.9 Bilinear idealizations of the LRB forcedeformation characteristics after gravity loading in the bridge.
6.7 Dynamic Analysis
The structural periods of the two bridges utilizing either FPS or LRB were established per kp of the isolators. The first three structural periods for the FPSisolated and LRBisolated bridge were T1=2.38, T2=2.26, T3=2.00 and T1=2.30, T2=2.16, T3=1.64, respectively. The first three mode shapes were those involving the seismic isolation and were essentially the same for the two bridges (Figure 6.10). The modal characteristics of the two bridges, isolated with the FPS in one case and with the LRB in the other, are considered to be sufficiently close for comparative assessment. A notable distinction
among the vibration characteristics of the two bridges arises for the torsional mode, T3. The FPSisolated bridge acquired a higher period due to the more flexible isolators at the abutments compared to the LRBisolated bridge.
107
1). ξ.44 s (arithmetic average of the T1 of the two bridges) corresponding to a 7% probability of exceedance in 75 years hazard level earthquake in Memphis. TN. The response spectra of the scaled ground motion records for 5% damping.10 Mode shapes of the bridge deck from plan. 108 . and their median are given in Figure 6. to examine the effects of N variations on the response of the bridge and the isolators. The three components of the acceleration histories of each scaled ground motion are applied to the models using NLTH analyses. The inplane orthogonal components of the earthquakes are oriented to result in the maximum demands on the columns for all cases. The geometric mean of the longitudinal and transverse component of each record is scaled to match the spectral value of 0. The bridge models were also analyzed without the vertical && acceleration history component of the ground motions. A suite of ten earthquake records from rock sites is used in the NLTH analysis of the bridges (Table 6.11.056g at a period of 2. u (t ) v .Figure 6.
721 0.056g 7% in 75 years Median Figure 6.00 0.290 1.285 1.718 0.978 1.453 0. Sa=0.227 Loma Prieta 1989/10/18 0.50 1.818 Scale 5.00 2.25 1.34 s.229 Whittier Narrows 1987/10/01 0.00 0. Palm Springs 1986/07/08 0. Sa (g) for ξ=5% 1.473 0. Canada 1985/12/23 0.086 Landers 1992/06/28 0.617 0.50 0.50 0.No. 109 .092 Northridge 1994/01/17 1. USSR 1976/05/17 0.507 Nahanni.75 1.25 0.845 0.096 2. 1 2 3 4 5 6 7 8 9 10 Table 6.814 0.585 1.264 N.304 0.102 Loma Prieta 1989/10/18 0.492 0.25 1.612 0.50 1.50 2.25 2.608 1.820 0.471 Helena.135 2.595 2.199 0.11 Response spectrums for the suite of ground motions.75 1.302 0.500 0.209 Gazli.069 0.173 0.25 0.1 Ground motion suite Component PGA (g) Earthquake record Longitudinal Transverse Vertical Morgan Hill 1984/04/24 0.00 Period.75 0.501 0. T (s) T=2.320 Spectral acceleration.785 0. Montana 1935/10/31 0.00 1.75 2.15 0.75 3.411 0.00 0.098 0.
12. 2 2 2 2 respectively. The LRB on top of the piers were observed to buckle (N>Pcr) under the && Northridge earthquake record when the N increase due to u (t ) v and Pcr reduction due to isolator deformation was considered in the models. (2) the maximum isolator displacement. the buckling condition (N>Pcr) of the LRB did && not prevail. It is observed that the N/No makes a notable drop to 0. respectively. The fL . Therefore. where δ L is the longitudinal and δT is the transverse isolator displacement. where fL is the longitudinal and fT is the transverse isolator force. dmax.8 Results The structural response of the isolators and columns along the same transverse axis were essentially the same. in a given earthquake timehistory analysis. This effect can be observed in Figure 6. and (3) maximum column drift.FPS.14 which is proportional to the decrease of the kp. respectively.6. the results are presented for one of the isolators on top of the piers and the abutments and one of the columns. The main response quantities monitored for the isolators are the: (1) maximum isolator force. MIF = max( f L + f T ). It is concluded that neglecting the effects of u (t ) v in the isolator model may result in overlooking a fundamental failure mode of the LRB.12a as deviation from the idealized bilinear forcedeformation path && of the isolator. The analysis without the effects of u (t ) v on the isolator models can not account for the increased flexibility of the isolator and consequently underestimated the 110 .13. The timehistory of the N/No for the same earthquake is given in Figure 6.δ L history of the && FPS with and without the u (t ) v effects on the isolator located on top of the pier for the Northridge earthquake record is given in Figure 6. When the bridge was analyzed && without the u (t ) v effect on the isolator. MID = max( δ L + δ T ). Maximum and minimum values of other response quantities are denoted by ‘max(response quantity)’ and ‘min(response quantity)’.
4 fL 135 21 180 21 14 7 0 7 FPS 14 14 21 δL (cm) δL (cm) (a) (b) Figure 6.0 0.13 Time history of the N/No for the FPS during the Northridge earthquake record. effect is (a) included (b) not included. 180 135 90 f (kN) L f (kN) L 45 0 45 90 135 180 21 max( δL ) = 11.max( δ L ).12 Forcedeformation history of the FPS in the longitudinal direction on top of the && pier for the Northridge earthquake record where the vertical component.0 0 2 4 6 min(N/No)=0.0 N/No 1.5 1.5 2. u (t ) v . N FPS 2. The maximum allowable pressure stress of 275 MPa at the slider of the FPS was not exceeded in any of the analyses.5 14 7 0 7 fL 180 max(  fL  ) = 74.3 135 90 45 0 45 90 FPS max(  fL  ) = 72. 111 .14 Figure 6.13 8 Time (s) 10 12 14 16 max(N/No)=2.1 max( δL ) = 9.5 0.
5 45 0 45 90 135 180 21 14 7 0 7 LRB fL fL LRB 14 21 180 21 14 7 0 7 14 21 δL (cm) δL (cm) (a) (b) Figure 6. vary during the earthquake response of the isolator.7 max( δL ) = 13. 180 135 90 f (kN) L max(  fL  ) = 32.&& The fL.δ L history of the LRB with and without the u (t ) v effect on the isolators located on top of the pier for the Whittier Narrows earthquake is given in Figure 6. kpf = 1 − ( N/Pcr ) . An 2 && increase of the N by the random pulses of the u (t ) v at large deformations compounds the reduction of the Pcr and results in smaller kpf. u (t ) v . effect is (a) included (b) not included.14 Forcedeformation history of the LRB in the longitudinal direction on top of && the pier for the Whittier Narrows earthquake record where the vertical component.14. 112 . which increases the flexibility and the MID of the isolator.27. The reduction factor for the postyield stiffness.15. It && is observed that excluding the u (t ) v resulted in an overestimation of the max( fL ) and underestimation of the max( δ L ).5 180 135 90 f (kN) L 45 0 45 90 135 max(  fL  ) = 59. The value of kpf reaches a minimum of 0.3 max( δL ) = 15. This result can be explained via the timehistory of the Pcr and the N of the isolator given in Figure 6.
respectively.15 Time history of the LRB buckling load. && The ΜΙD of the LRB may also be overestimated by not including the u (t ) v effect on the isolator models. The max( δ T ) with and without the u (t ) v effect on the isolator was 4. Pcr.4 cm and 8.27 2 15 20 25 Time (s) Figure 6. N. and normal force. there was a considerable difference in the && transverse direction.16 gives the fL. Figure 6. It is observed that although the longitudinal response of the LRB isolator was essentially && the same with and without the u (t ) v effect.δ T of the LRB with and && without the u (t ) v effect on the isolator located on top of the pier for the Gazli earthquake. The reduction in the kp from higher N corresponded to a more flexible response in the transverse direction thus attracting smaller earthquake induced displacements on the isolators in this direction.δ L and fT. . 113 .N Pcr 1800 Force (kN) 1350 900 450 0 0 5 10 N LRB min[1(N/Pcr) ]=0.7 cm.
17.6 LRB fL 180 21 14 7 0 7 14 21 δL (in) δT (cm) (a) (b) && u (t ) v = 0 40 30 20 f (kips) L 10 0 10 20 30 40 8 6 4 2 0 2 4 6 8 LRB fL max(  fL  ) = 10.4 10 0 10 20 30 40 8 6 4 2 0 2 4 6 8 max(  fL  ) = 10.6 f (kN) T 180 135 90 45 0 45 90 135 180 21 14 7 max(  fT  ) = 45. the max( fL ) was underestimated by approximately 47%. and (b). It is observed that the exclusion of the u (t ) v resulted in a slight overestimation of max( δ L ) . however. && The fL.8 135 90 45 0 45 90 135 max(  fT  ) = 29.0 LRB fT max( δT  ) = 4.8 max( δL ) = 3.16 Forcedeformation history of the LRB on top of the pier for the Gazli earthquake in the (a). (d) transverse directions.&& u (t ) v included 40 30 20 f (kips) L f (kN) T 180 max( δL ) = 3. (c) longitudinal.δ L history of the FPS with and without the u (t ) v effect on isolator the isolator located on top of the pier for the Whittier Narrows earthquake is given in Figure && 6.9 max( δT  ) = 8. The 114 .7 fT LRB 0 7 14 21 δL (in) δT (in) (c) (d) Figure 6.
17 Forcedeformation history of the FPS in the longitudinal direction on top of && the pier for the Whittier Narrows earthquake record where the vertical component. In addition to acquiring a higher initial and postyield stiffness. The forcedeformation histories of the LRB and the FPS given in Figures 11(a) and 14(a) show that there are notable differences in the response of the two isolators.&& increase of the N by inertial forces from the u (t ) v resulted in instantaneous but considerable stiffening of the FPS response. the N reached a maximum of 837 kN and a minimum of 290 kN. effect is (a) included (b) not included.1 45 0 45 90 135 180 21 14 7 0 7 max( δL  ) = 10. the FPS hysteresis has short duration wriggles compared to the LRB. which corresponds to a 61% increase and a 56 % decrease from No respectively.0 FPS fL f 135 14 21 180 21 14 7 0 7 14 21 δL (cm) δL (cm) (a) (b) Figure 6.0 135 90 45 0 45 90 FPS fL max(  fL  ) = 88. u (t ) v . It is concluded from the aforementioned results that excluding the normal loaddependency of the isolators in modeling the forcedeformation response may produce considerable errors in MIF and MID and mislead to similarities between the FPS and the LRB.7 L max( δL  ) = 11. 180 135 90 f (kN) L (kN) 180 max(  fL  ) = 129. Although at no instance was there separation between the two surfaces of the FPS or was the maximum allowable pressure exceeded. This is attributed to the stronger dependency of 115 .
It is observed from Figures 6.17(b) that the forcedeformation response of the two isolators is smoother and similar to the bilinear && idealization when the u (t ) v effects in the isolator models are excluded. It is observed that the forcedeformation hysteresis of the FPS gets shorter in width and deeper in height on top of the pier compared to the response on top of the abutment.18 illustrates the differences in the forcedeformation responses of the FPS and LRB located on the same beam line but on top of the pier and the abutment for the Helena earthquake.FPS to N.&& the FPS response to u (t ) v . 116 . This is an attribute of the strong fluctuation of the kp. It is observed that the longitudinal deformation of both types of isolators is larger at the abutments compared to those at the piers. Figure 6.FPS and FY. This indicates a larger torsional effect in the response of the FPSisolated bridge. This is attributed mainly to the torsional vibration of the bridges deck. However.14(b) and 6. the difference between the abutment and pier isolator deformations is greater for the FPS compared to the LRB.
180 135 90 (kN) 45 0 45 90 135 180 21 max(  fL  ) = 77. MNDpier. The FPS inflicted more than twice the amount of force to the piers than the abutments on the median. and abutments.18 Forcedeformation history of the FPS and LRB for the Helena earthquake (a).9 14 7 0 7 LRB fL 14 180 21 14 21 δL (cm) δL (cm) (c) (d) Figure 6.19 shows the box plots for the ratio of the total MIF transferred to the pier.(d) on top of the abutment.4 max( δL ) = 11. ΣMIFabutment. MNDabutment for the 9 earthquake NLTH analyses (the Northridge earthquake excluded from the initial 10 due to LRB buckling condition).2 14 7 0 7 LRB fL 14 180 21 14 21 δL (cm) δL (cm) (a) 180 135 90 f (kN) L (b) 180 135 90 (kN) f L 45 0 45 90 FPS fL 45 0 45 90 135 180 21 max(  fL  ) = 40.7 max( δL ) = 10. and the total MIF transferred to the abutment.(b) on top of the pier and (c).5 14 7 0 7 135 21 max(  fL  ) = 63.0 90 f (kN) L 45 0 45 90 135 21 f L max(  fL  ) = 54.9 14 7 0 7 FPS fL 180 135 max( δL ) = 7. and MND on top of the pier.7 max( δL ) = 10. A notable difference with the LRB is that the difference in the isolator forces transmitted to the abutments compared to the piers were closer in magnitude compared to 117 . ΣMIFpier. Figure 6.
50 1. However.25 2. 118 .87 2. and abutments.20 gives the total energy dissipated by FPS and LRB located on the same beam line but on top of the pier and the abutment for the 9 earthquake timehistory analyses. the LRB transmitted larger forces to the abutments compared to the piers.18 0.the FPS.50 1.93 (a) (b) Figure 6. ΣMNFpier. MIDpier.25 1.80 0.00 0.75 0.50 2. this effect was observed to a lesser extent with the LRBisolated bridge because the stiffness distribution of the isolators along the longitudinal axis of the bridge was more uniform compared to the FPSisolated bridge. FPS LRB 90th percentile 75th percentile Median 25th percentile 10th percentile 2.00 1.25 0.50 0. ΣMNFabutment. and the total MNF transferred to the abutment.25 0. The MID for both isolator types were larger on top of the abutments compared to the piers. (b) MID on top of the pier.00 MIDpier / MID abutment 0. Contrary to the FPS.75 0. This is attributed to the torsional motion of the bridgedeck.00 ΣMNFpier / ΣMNFabutment 2.75 1.00 0. MIDabutments.50 0.25 2.19 Comparison of the (a) total MNF transferred to the pier.00 1. It is observed that the median energy dissipation of the FPS at the abutment is approximately 55% of that at the abutment.75 1.25 1. Figure 6. The energy dissipation capacities of the LRB on top of the pier and the abutment are closer compared to the case in the FPS.50 2.
Furthermore. unlike the case in the FPS.22.21 to 6.20 Total energy dissipated for by the isolators (a) on top of the pier (b) on top of the abutment. The statistical interpretation of the results are presented with numerical values of the median and plots of the 10th. dmax. The MIF of the FPS for 119 . the energy dissipation in the LRB is higher at the abutment compared to the pier.21 and 6. are given in Figure 6. 25th. respectively.22 shows that the FPS acquired smaller MID and larger MIF. respectively. 10th.cm) FPS LRB 90th percentile 75th percentile Median 25th percentile 10th percentile 6000 5400 4800 4200 3600 3000 2400 1800 1200 600 0 1785 1735 997 2084 (a) (b) Figure 6. The MIF and MID acquired on top of the pier for the 9 earthquake timehistory analyses are given in Figures 6. and 90th percentile cumulative probabilities. Above pier Above abutment Total energy dissipated (kN. 75th. The maximum column drifts. It is observed from Figure 19 that the median MIF in the FPS are approximately 46% larger than the LRB. This difference reaches a peak of 139% for the Whittier Narrows earthquake where the MIF in the FPS and LRB were 116 kN and 50 kN.23. and placed larger demands on the columns compared to the LRB. A general comparative assessment from the median values in Figures 6.
u (t ) v . && u (t ) v included 150 125 FPS LRB 90th percentile 75th percentile Median 25th percentile 10th percentile && u (t ) v = 0 MIF (kN) 100 75 64.&& the suite of ground motions is more scattered then the LRB.21 Maximum isolator forces (MIF) for the suite of ground motions on top of the && pier with vertical component. It is observed that the && absence of u (t ) v in MID prediction resulted in negligible difference in the FPS and slight overestimation in the LRB on the median. similar to the results obtained for the MIF. In particular. The median of the dmax in the FPSisolated bridge is approximately 17% larger than the LRBisolated bridge.6 cm.1 50 25 0 34. the difference between the FPS and LRB was maximum for the Nahanni earthquake with 8. The median of the MID of the LRB was approximately 32% larger than the FPS.1 (a) (b) Figure 6. The median of the dmax for the suite of ground motions did not change significantly by && excluding the u (t ) v for either isolator. Exclusion of the u (t ) v in the timehistory analyses resulted in the underestimation of the median of the MIF in the two systems by approximately 15%.2 cm and 4. effect is (a) included (b) not included.4 48. 120 . For example the dmax=0.78% in the FPSisolated bridge for the Whittier Narrows earthquake is approximately 82% higher than the LRBisolated bridge. the scatter for the FPSisolated bridge dmax is higher. However.8 45.
u (t ) v . effect is (a) included (b) not included.2 (a) (b) Figure 6. 121 .36 0. u (t ) v .33 (a) (b) Figure 6.42 0.22 Maximum isolator deformations (MID) for the suite of ground motions on && top of the pier with vertical component.2 6. dmax.8 FPS LRB 90th percentile 75th percentile Median 25th percentile 10th percentile && u (t ) v = 0 dmax (%) 0.23 Maximum column drifts.1 6.&& u (t ) v included 18 15 && u (t ) v = 0 MID (cm) FPS LRB 90th percentile 75th percentile Median 25th percentile 10th percentile 12 9 6 3 0 3.4 9. effect is (a) included (b) not included.6 0.2 0.4 0. for the suite of ground motions on top of the && pier with vertical component. && u (t ) v included 1.0 0.42 0.0 0.
Figure 6. This is attributed to lower elastic stiffness of the LRB compared to the FPS. The LRB acquired a larger MID compared to the FPS.24(a) and (b) reveal the fundamentally different responses of the two types of isolators. This can be explained further by Figure 6.9 cm. respectively.24 gives the deformation histories of the FPS and LRB on top of the pier for the Morgan Hill earthquake. Consequently. The LRB deformation history has a higher frequency content compared to the FPS. It is concluded that the LRB is more sensitive to the frequency content of the ground motion compared to the FPS.24(c)(d).1 cm and 47. 122 . The total sum of the deformation of the LRB and the FPS are 29. the LRB engages into motion at smaller dynamic vibrations compared to the FPS. The displacement trajectories in Figure 6.
5 15.0 2.56 4 3 2 1 0 1 2 3 4 MID=4.0 17.0 t (s) δL (cm) (c) 4 max( δT  ) =3.0 27.0 17.30 3 2 1 0 1 2 3 4 0.5 20.0 2.5 30.21 3 2 1 0 1 2 3 4 0.0 27.0 t (s) δL (cm) (d) 4 max( δT  ) =3.0 7.5 30.0 7.5 10.5 5.0 2.0 12.0 7.5 10.0 22.5 10.5 20.03 3 2 1 0 1 2 3 4 0.0 22.0 12.5 25.0 27.5 20.0 17.4 3 2 1 0 1 2 3 4 4 3 2 1 0 1 MID=3.0 17.5 5.0 t (s) (b) 4 max( δL  ) =3.5 30.5 10.5 20. (d).0 12.0 22.5 15.0 2.5 5.5 25.0 t (s) δT (cm) (e) δT (cm) (f) Figure 6.5 5.0 12.24 Deformation histories for the Morgan Hill earthquake of (a).5 30.5 15. (e) FPS and (b).0 7.0 22. (c).5 15.5 25. (f) LRB located on top of the pier.5 25.0 27. 123 .32 δT 2 3 4 δT 4 3 2 1 0 1 2 3 4 δL δL (a) 4 max( δL  ) =2.72 3 2 1 0 1 2 3 4 0.
Nonlinear timehistory analyses using a suite of 10 ground motions that corresponded to a hazard level of 7% in 75 years was performed for the two bridges. the distinctions of the response characteristics of sample elastomeric and sliding isolator types. have been highlighted. The FPS stiffened on top of the piers while the LRB was stiffer on top of the abutments. the effects of the vertical components of the ground motion on the isolator response had negligible effect on maximum column drifts. These distinctions arise from the variant dependency of the two isolators to normal loads. (4) The LRB had a more uniform distribution of lateral stiffness throughout the bridge compared to the FPS.9 Conclusion In this paper. Two models of a threedimensional (3D) MultiSpan Continuous Concrete Girder (MSCCG) bridge. An LRB model to account for both the coupling in the inplane and vertical directions has been developed. The median maximum isolator forces and displacement values were underestimated for both types of isolators in the absence of vertical components of the ground motions in the analyses. Following conclusions are made: (1) Despite attaining similar seismic isolation periods. the choice between the LRB or the FPS resulted in considerable differences in the response of the bridge and the isolator forces and displacements. 124 . one isolated with the FPS and the other with the LRB. was generated. however. placed smaller demands on the columns compared to the FPS. However. (3) The LRB acquired higher displacements. the Lead Rubber Bearing (LRB) and the Friction Pendulum System (FPS) respectively. (2) Excluding the vertical component of the ground motion in the analyses dampened the distinctions that exist between the forcedeformation response of the FPS and the LRB.6.
Excluding the vertical component of the ground motion in the modeling and analysis phases of the LRB may result in overlooking a fundamental failure mode.(5) The FPS is capable of accommodating vertical components of the ground motion records that might result in a buckling failure of the LRB. 125 .
2 Influence of Bridge Design Parameters To ascertain the effects of the design parameters of the typical highway bridges on the system’s seismic response. In the first section. In the second section. the “singlefactor“ method and regression analyses are performed with nonlineartime history (NLTH) analyses to quantify the significance of bridge design parameters on the system’s response quantities. A further understanding of the effects of design parameters will provide insight on both the level of complexity required in modeling and the seismic response for a wider range of bridge designs. the design characteristics of the FPS may be modified to alter the seismic response of the bridge. a new seismic isolation strategy that involves a practical design modification to the FPS at the abutments is presented. This chapter investigates the effects of bridge design parameters and an innovative isolation design strategy on bridge seismic response. an experimental design is constructed. The implications of possible design variations in this new seismic isolation strategy are illustrated. 7. This approach aims to improve the seismic response of the bridge by creating a more uniform response between the isolators located at the piers and the abutments. Additionally.CHAPTER 7 ASSESSMENT OF THE INFLUENCE OF DESIGN PARAMETERS ON THE RESPONSE OF BRIDGES SEISMICALLY ISOLATED WITH FRICTION PENDULUM SYSTEM (FPS) 7.1 Introduction Seismic isolation is applicable to a wide range of bridges that differ in geometric and material properties. The term “experimental design” can be defined as: “the experimental structure used to generate 126 .
is selected for this study (Gardiner and Gettinby 1998). The details of this model are not repeated in this chapter. considered. The design parameters of the base model are denoted by a subscript ‘o’. The “singlefactor design”.2.1 Analyses The threedimensional bridge model with FPS isolation developed in Chapter 6 is used as the base model to which the variation of the design parameters is compared. The span length is varied in two different ways: (1) mass adjusted to increasing length.1. Reinforcement ratio and distribution is assumed to remain the same for the range of column height. which is a form of experimental design to analyze the influence of different levels of a controllable parameter on a measured response. (2) mass kept constant at all lengths. 127 . superstructure sections with the same structural characteristics may be constructed for constant mass. 7. the results from this analysis are handled separately since this variation is associated with modeling assumptions as opposed to a design parameter in a bridge. Ld. The NLTH analyses are performed using the suite of ground motions described in Chapter 6 for four equally spaced values of selected bridge design parameters. The influence of pinned and fixed modeling assumptions for the base is illustrated to provide the bounds of the response (Figure 7. Given the wide range of material properties available. These parameters are detailed in Table 7. Ld*. The upper and lower bounds of the selected design parameters in Table 7.2).practical data for interpretative purposes” (Gardiner and Gettinby 1998).1 and illustrated in Figure 7. Adjusting the mass for the same crosssection in the superstructure provides insight on the influence of additional dead weight. Increasing the span length by keeping the mass and crosssection properties constant provides the opportunity to monitor the effects of superstructure flexibility. Lc.1 are selected from the bridge inventory analysis performed by Nielson (2005) for the Central and Southeastern United States. However.
128 .3 (MPa) Pinned.67.2 Base conditions for the substructure modeled as (a) pinned (b) partiallyfixed (c) fixed.1 Bridge design parameters.5 (m) 2060 (m) 2060 (m) 20. elastic.748.Table 7.1 Parameter variation ranges Parameter Abbreviation Range Degree of skew 040 (degrees) α Column height Span length Span length* Concrete nominal strength Base conditions *Mass adjusted in accordance with increased length Lc Ld Ld* fc  3. (a) (b) (c) Figure 7. fixed Figure 7.
00 s (see Figure 6. for the median values of the response quantities are developed. Figure 7.3 summarizes the absolute values of the slope. The first three vibration periods for the base model were T1=2. The variation of these response quantities is plotted against the selected design parameters for the ten ground motion records used in Chapter 6. T2=2. 7. Linear regression curves in the form: y(x) = ax + b. maximum column drifts.2 Results Linear regression curves for the bridge response quantities as a function of the variation of the design parameters are given in Appendix B. the ratio of the total MNF transferred to the pier and the total MNF transferred to the abutment. MNF and MND.7). The variation of these two parameters caused the largest shift in the fundamental vibration period of the structure (Table 7. a’response quantity’ of the regression curves in Appendix B.3 that the two design parameters that have made the highest impact on the response quantities were Lc and Ld*. respectively. Although the shape of the first three modes of vibration remained the same with the base model.38 s. and T3=2. ΣMNFpier / ΣMNFabutment. a notable change in the first period.2. MNDpier / MNDabutment.Performance of the bridge is monitored via the maximum normalized force and displacement of the FPS isolators located on top of the pier. a‘response quantity’.26 s. It is observed from Figure 7. was observed in the bridge as a function of Lc and Ld*. is used to quantify the significance of the design parameters. The exact values of the slopes are denoted on top of the bars with the corresponding signs. the ratio of the MND on top of the abutment and MND on top of the pier. Positive and negative signs indicate an increase and decrease of the linear regression curves as the design parameter magnitude increases. T1. The slope of these regression curves.10). 129 . dmax.
01 0.06 0.08  aMNF  0.002 0.044 0.02 0.0.00 fc / fco Lc / Lco Ld / Ldo Ld */ Ldo α (b) 1.6 0.351 0.04  aMND  0.093 0.05 0.2 0.016 0.06 0.03 0.00 fc / fco Lc / Lco Ld / Ldo Ld */ Ldo α (a) 0.024 0.12 0.001 0.04 0.001 0.0 0.647 0.2 1.001 0.0 fc / fco Lc / Lco 0.02 0.4 0.001 Ld / Ldo Ld */ Ldo α (c) 130 .10 0.0 0.042 0.080 0.8  ad  max 0.0 0.
211  aMND pier 0.056 0.2 0.4 0.2 0.8 0.0 fc / fco Lc / Lco Ld / Ldo Ld */ Ldo α (e) Figure 7.3 Slopes.’ for the regression lines of the median design parameters.6 0.792 0.497 0.106 0. 131 .051 0.0 0.4 0.8  / MNDabutment 0.002 fc / fco Lc / Lco Ld / Ldo Ld */ Ldo α (d) 0.643  aΣMNF pier 0. a’response quantity.0 0.1.0 0.222 / ΣMNFabutment  0.6 0.
the bridge and attracted lower seismic forces. The 132 .519 2.98 Ld*/Ldo 1.296 2.274 1.2 2.019 1.327 The increase of Lc created a more flexible structure with increased fundamental vibration.2 Vibration periods of the bridges as a function of design parameters Lc/Lco Period T1 T2 T3 0.551 2. Larger Lc/Lco resulted in flexible supports for the isolators located on top of the piers. Consequently.427 2. the fixity of the supports for the FPS located on top of the abutments remained the same.343 2.354 2.654 2. magnitude of dmax increased with larger substructure flexibility.Table 7. the terms indicating the difference in the responses between isolators at the piers and abutment. It is observed that the increased flexibility of the columns resulted in higher structural displacements of the substructure at the isolator level and dampened the effect of seismic isolation. However.4 shows the longitudinal normalized forcedeformation response of the FPS on top of the pier and the history of the column tip displacements.6 2.75 2.8) and flexible (Lc/Lco=1. MNDpier / MNDabutment and ΣMNFpier / ΣMNFabutment increased considerably as the substructure flexibility increased.268 1.018 0.32 2.008 1.999 1. for the Whittier Narrows earthquake record for the cases of stiff (Lc/Lco=0.454 2.307 2.461 2.328 2.8 2.999 1.303 1.008 2.25 1.75 2. Figure 7.592 2.6) substructure. δcolumn.5 2. Consequently.4 2.
05 0. Although the larger gravity load has stiffened the response of the isolators and resulted in larger MNF.2 0.10 0.29 0.05 0.2 0.0 0.1 0.3 max( fL / N ) = 0. (c) timehistory of the longitudinal column tip deformations.05 0.4 Forcedeformation history of the FPS in the longitudinal direction on top of the pier for the Whittier Narrows earthquake record with (a) Lc/Lco=0. It is observed that larger N increased the stiffness and postyield values of the 133 .15 0.8 (b) Lc/Lco=1.75 and Ld*/Ldo =1.0.2 0.10 0.05 0.00 0. Ld*.20 max( δL / R ) = 0.3 0.13 0.5 shows the longitudinal normalized forcedeformation of the FPS on top of the pier for the Whittier Narrows earthquake record for the cases of Ld*/Ldo=0.20 0.1 0. The influence of increased mass simulated via adjusted span length.0 0. is unique.75).1 0.6. it has also increased the fundamental period of the structure.15 0.3 0.3 0.10 0.20 0.14 f / No L f /No L max( fL / N ) = 0.8 δcolumn (cm) 5 0 5 10 0 2 4 6 8 10 12 14 16 18 min( δcolumn  )=1.7 Lc / Lco=1.00 0.6 Lc / Lco=0.20 max( δL / R ) = 0.05 δL / R δL / R (a) (b) 10 max( δcolumn  )=6. Figure 7.2 0.15 0.15 0.10 0.8 20 22 24 Time (s) (c) Figure 7.1 0.
No.13 f / No L f /No L max( fL / N ) = 0.2 0.2 0.10 0.05 0.15 0.05 0. However.00 0. This is attributed to the reduced vertical flexibility gained by larger Ld.10 0.6 shows the longitudinal normalized forcedeformation and the N history normalized with the initial gravity load.24 0.75) superstructure.05 0.75 (b) Ld*/Ldo =1. Figure 7. Larger MNF was acquired with increased superstructure mass and this resulted in higher structural demands in the columns observed from increased values of dmax.5 Forcedeformation history of the FPS in the longitudinal direction on top of the pier for the Helena earthquake record with (a) Ld*/Ldo=0.15 0.2 0.2 0.1 0.30 0. of the FPS on top of the pier for the Northridge earthquake record for the cases of stiff (Ld/Ldo=0.20 0.FPS and caused the isolators to deform less.3 0. It is observed that the stiff superstructure caused considerably larger MND and MNF and resulted in the uplift of the isolator when compared to the flexible superstructure.20 0.20 max( δL / R ) = 0.75) and flexible (Ld/Ldo =1. and ΣMNFpier / ΣMNFabutment were moderately affected. 0.3 0.1 0.1 0.0 0. As Ld increases the vertical flexibility of the structure increases and dampens the vertical effects of the ground motions.10 0. Consequently.00 0.15 0.08 δL / R δL / R (a) (b) Figure 7.20 max( δL / R ) = 0.15 0.0 0. it is 134 . the MNF.10 0.3 max( fL / N ) = 0.05 0.1 0. MNDpier / MNDabutment. Increased flexibility of the superstructure had negligible effect on dmax and MND.3 0.75.
10 f /No L 0.observed that the max(N/No) for the flexible superstructure is approximately 60% of the stiff superstructure.15 0.5 0.00 0.15 0.0 1.1 0.21 max( δL / R ) = 0.1 f / No L 0.2 max( fL / N ) = 0.5 2.75 (b) Ld/Ldo =1.5 1.0 0.10 0.05 0.3 0.35 (c) (d) Figure 7.6 Forcedeformation history of the FPS in the longitudinal direction on top of the pier for the Northridge earthquake record with (a) Ld/Ldo=0.20 δL / R δL / R (a) (b) 2.15 0.05 0.20 0.10 0.19 N/No 2.5 0.77 8 Time (s) 10 12 14 16 max(N/No)=1.2 0.15 0.0 0.2 0.1 0.15 0.0 0.05 0.0 0.3 max( fL / N ) = 0.5 2.10 0.1 0.0 0 2 4 6 min(N/No)=0.10 0.0 8 Time (s) 10 12 14 16 max(N/No)=2.05 0. This stiffening affect increases MNF and dmax.20 max( δL / R ) = 0.5 1.12 0. 0.75.3 0. 135 .0 0 2 4 6 min(N/No)=0.00 0.2 0.0 N/No 1. This indicates that the vertical components of the ground motion are less influential for bridges with large superstructure flexibility.20 0.3 0.
01 s. This is attributed to the concentration of the seismic response at the isolators and the uncoupling of the superstructure from the substructure. partiallyfixed.9 shows the longitudinal and transverse normalized forcedeformation history of the FPS on top of the pier for the Loma Prieta earthquake record for the pinned and fixed base conditions of the bridge model. T3=2. Figure 7. Figure 7.26 s. In the pinnedbase model. It is observed that the pinnedbase modeling Similarly. It is observed that in the pinned base model. This is a similar phenomenon to that observed with the increased values of L c. the influence of fc on assumption results in substantial overestimation of the T1 and a general underestimation of the response quantities compared to partiallyfixed and fixed base modeling assumptions. the structural response quantities was negligible.31 s. α. The influence of pinned.80 s.00 s and T1=4. T3=2. The transverse action of the isolators is due to the frame rigidity of the piers in these directions. T2=2.31 s. respectively. It is concluded that the pinned modeling assumption may result in considerable underestimation of isolator design parameters and 136 . The first three vibration periods of the fixedbase and pinnedbase were T1=2. the effect of the isolators at the piers was reduced because the substructure flexibility allowed for accommodation of more of the seismic deflections compared to the fixedbase condition.8 illustrates this aspect using approximate deflected shapes of the bridge where the bases are modeled as either pinned or fixed.Bridge skew. The isolators located on top of the piers in the pinnedbase model attained negligible deformations due to the deformability of the substructure. The base bridge model response quantities are in general closer to the fixedbase modeling assumption. Seismic deformations of the substructure were essentially resisted by the isolators on top of the abutments.7. T2=2. and fixed base modeling assumptions on the bridge response quantities is given in Figure 7. had negligible effect on the response quantities of the bridge. the longitudinal response of the isolators located on top of the piers become negligible in the longitudinal direction.
reduce the effectiveness of seismic isolation. 137 . Increased substructure flexibility contributes to the difference in the response of the isolators located at the piers and the abutments.
0.06 0.12 0.00 MND MNF 0.08 0.2 1.10 0.20 0.04 0.068 (a) 0.318 0.04 0.08 0.6 0.10 0.10 0.0 (e) Figure 7.20 0.067 0.02 0.20 0.16 0.5 (c) 1.0 ΣMNFpier / ΣMNFabutment 2.318 2.4 0.25 0.16 0.18 0.0 0.7 Median values of the response quantities of the bridge as a function of base modeling assumptions.032 0.146 MND 0.06 0.849 (d) 0.801 0.4 MNDpier / MNDabutment 1.133 2.30 0.02 0.14 0.068 1.067 0.0 0.00 0.032 0.8 0.5 2.12 0.2 0.18 0.05 0.14 0.0 1. 138 .5 1.087 0.140 0.135 0.00 (b) 3.15 0.
139 .8 Structural bridge responses as a function of base modeling assumptions.Figure 7.
05 0.00 0. R (Naeim and Kelly 1999).10 0.1 0.15 0.04 0.2 max( fL / N ) = 0.2 0.05 0.05 0.05 0.20 f / No T f /No L 0.02 max( fT / N ) = 0.05 0. It has been shown in Chapters 4 and 5 that achieving constant prescribed friction forces is difficult due substantial changes in the magnitude of the friction coefficient as a function of sliding 140 .20 δL / R δT / R (a) 0.10 0.10 0. (b) fixed base conditions.00 0.00 0.0 0.2 0.09 0. (c).2 0.3 Influence of a Modified Seismic Isolation Strategy The FPS design parameters that can be investigated for a range of values are limited.1 0.20 0.3 0.00 0.0 0.05 0.2 (b) max( δT / R ) = 0.1 0.1 0.15 0.11 max( δL / R ) = 0.3 0.0.9 Forcedeformation history of the FPS in the longitudinal and transverse directions on top of the pier for the Loma Prieta earthquake record with (a).10 0. (d) pinned base conditions.06 f / No T f /No L 0.2 max( fT / N ) = 0.05 0.20 δL / R δT / R (c) (d) Figure 7.3 0.20 0. 7.10 0. A notable disadvantage of the FPS is that it is essentially a oneparameter system based on the radius of its concave dish.2 0.15 0.3 0.0 0.15 0.3 0.1 0.20 0.10 0.15 0.05 0.20 0.3 0.1 0.3 0.3 0.15 0.09 max( δT / R ) = 0.15 0.0 0.15 0.10 0.10 0.1 0.1 0.2 0.20 0.
It has been shown in Chapters 5 and 6 that the vibration characteristics of bridges are essentially a function of the isolator properties. ∆max. 224. of the isolator. and R = 76 cm (Mosqueda et al. Standard radii in manufacture for the FPS are 99. and 620 cm. Specifically. 2004). 396. 305. Kasalanati and Constantinou (2005) showed that the FPS isolators can be effectively prestressed to achieve higher N. however. However. the isolators at the abutments acquired smaller yield force and postyield stiffness compared to those located at the piers. Additionally. 1993).10. This lead to torsional modes of vibration of the superstructure during earthquake induced loads. using larger R for the FPS without additional measures may shift the fundamental period of the structure to a region out of interest in the design spectra. N. tendons must be configured to sustain a limited amount of tensile strain as shown in Figure 7. A more favorable response may be obtained by acquiring stiffer FPS forcedeformation response at the structurally stronger abutments compared to the piers. Additionally. It has been shown in Chapters 5 and 6 that there exists a notable variation in the stiffness and yield force properties of the FPS isolators located above the piers and abutments. 155. Isolators with smaller R have been manufactured. R = 56 cm (Constantinou et al. larger displacement demands and lower transfer of seismic forces were observed at the isolators located at the abutments compared to those located at the piers. The smallest R of the FPS in standard manufacture by the Earthquake Protection Systems is 99 cm (Earthquake 2003). Hence. Examples include R = 48 cm (Erdik and Uckan 2004). An important implication of smaller R is the reduction of maximum lateral deformation capacity. One option to increase the stiffness and yield force of the FPS located above the abutments is to increase the magnitude of the normal force. for experimental research purposes. This setup may not be practical for the configuration of the isolators located at the abutments 141 . the primary objective of this approach is the prevention of tensile force and uplift of the isolators.velocity and pressure.
(a) (b) Figure 7. The objective of the design is to: 1) Contribute more to reducing the structural demands on the columns by achieving smaller dmax 2) Create a more balanced ΣMNFpier / ΣMNFabutment and MNDpier / MNDabutment The primary constraint is to maintain the same R for all the isolators in the bridge to avoid additional mold costs.11.10 Prestressing the FPS (Kasalanati and Constantinou 2005). a modified approach for the seismic isolation of the bridge is proposed.3. The proposed bridge seismic isolation approach consists of increasing the R throughout the isolators in the bridge and introducing a modified design to the isolators at the abutments. The objective of this supplemental design approach is to generate a stiffer force 142 .12). 7. The conceptual drawing of the proposed design modification for the FPS located at the abutments is given in Figure 7. This simple modification consists of the introduction of extension spring elements around the perimeter of the isolator (Figure 7.1 Proposed Design Considering the limitations on the design parameters of the FPS explained above.
11). the vertical position of the springs also provides tensile restraint in the event of uplift. Extension springs are proposed for the task because of their existing widespread use in a vast variety of industries. this is outside the scope of this study (see Göbel and Brichta 1974). capability to undergo deformations that are of interest to isolators. The number of springs may be varied as a function of the required amount of supplemental stiffness and energy dissipation. such as rubber springs. and wide range of design parameters (Carlson 1978. Other alternatives. 143 . Additionally. however.deformation response for the isolators at the abutments. The springs are pinconnected to the upper and lower steel plates of the isolator and provide supplemental source of stiffness and energy dissipation as the isolator displace (Figure 7. may further be explored. Chironis 1961).
(a) (b) Figure 7.11 Proposed modified design for the FPS above the abutments in (a) undeformed (b) deformed state. 144 .
this property is neglected. Figure 7. The force deformationresponse of the springs is assumed to be elastic.Figure 7. The influence of the supplemental spring in the forcedeformation response of the FPS is analyzed for the displacementcontrolled load history given in Figure 7. spring and the combined FPS and spring response.13.14 gives the forcedeformation response of the FPS. It is observed that the addition of the elastic spring increases the postyield stiffness and the energy dissipation capacity of the isolator. 145 . is introduced in the supplemental springs that have a stiffness. ks=1.286 kN/cm. Although it is possible to design extension springs for pretensioned loads. It is assumed that a single FPS isolator with the design properties used in the analyzed bridge and N=258 kN.12 Sample spring element to be used in the proposed modified design of the FPS at the abutments.
5 0.10 0.2 0.00 0.20 0.3 Spring 0.05 0.5 20.5 10.0 δ (cm) 2.3 0.15 0.0 t (s) 12.5 5.15 0.10 0.0 0.25 20 15 10 5 0 5 10 15 20 25 0.13 Displacementcontrolled load history 0.2 0.20 f / No δ/R Figure 7.5 FPS 0.14 Forcedeformation response of the isolator with modified design.1 0.4 0.05 0.1 0. 146 .0 Figure 7.0 7.4 Spring + FPS 0.0 17.5 15.
15 shows the configuration and the modeling of the modified FPS design in the 3D bridge model. 7.3.Figure 7. This is an additional advantage of the proposed design because it allows for the assignment of separate stiffness characteristics in different directions to give more control over the seismic response over the isolator. Figure 7. y and z directions. The supplemental springs are modeled in OpenSees as zerolength elements with elastic uniaxial material object in the x. The response of the springs in the three main directions is uncoupled.2 Analyses The bridge seismically isolated with the proposed modified design is analyzed with the same methodology used in the previous sections for a range of R and ks. The spring element essentially acts in parallel in all three principal directions to the response of the FPS isolator. The 147 .15 Modified isolator design modeling at the abutments.
176 0. 1. the period of the third mode shapes of the bridges had notable differences compared to the base model. This indicates that the torsional vibration characteristics of the modified designs are stiffer than the base model. the torionsal stiffness increases. Table 7.objective is to quantify the effects of the modified bridge seismic isolation approach.38 and 1. The value of the T3 of the bridges for the five values of the R ranging from 155 cm to 620 cm in Table 7. 1.3 were.62.34 s. The fundamental period of the base bridge model was T1=2.3 Modified isolator design properties to achieve a fundamental period of T1=2.640 0. It is also observed that although the postyield stiffness of the isolators is reduced with increasing R.48.120 0. This is an attribute of the supplemental stiffness provided at the abutment 148 .38 s R 99 155 224 305 396 620 kp / ks 0. The stiffness of the supplemental springs in the isolators located at the abutments is selected to result in the same fundamental seismic isolation period of the bridges (Table 7.38 s. 1. Five new modified designs are introduced.42. respectively. 1. Each design is attained by one of the standard R’s of the isolators provided by the manufacturing company.287 0.070 Although the shape of the vibration modes in the modified isolation designs remained the same.3).
3 Results The median of the response quantities as a function of the R of the isolators are given in Figure 7. Additionally. for approximately R = 620 cm. This is the fundamental principle underlying the objective of the new design. The ratio ΣMNFpier / ΣMNFabutment. the MNDpier / MNDabutment is increased with the new isolation strategy. However. implying a completely uniform MND distribution throughout the isolators.3. indicating more energy and force transfer to the abutments. Figure 7. This ratio becomes approximately equal to 1. of the FPS on top of the pier and abutment for the Helena earthquake record for the cases of conventional design with R = 99 cm and the new design with R = 620 cm. It is observed that the new design allows for considerable reduction in the MNF and stiffness on top of the isolator located on top of the pier. The increased R implies more flexible isolators at the piers and consequently the MND slightly increases.isolators to acquire the same fundamental period in the bridges isolated with the modified design approach. However.17 shows the longitudinal forcedeformation. This allowed higher isolator forces to be transferred to the abutments. Additionally. deformations of the abutment slightly reduced. the stiffness of the isolator on top of the abutment has increased. which are typically structurally stronger components compared to the piers.16. Similar uniformity in the MNF is captured at approximately R = 244 cm. The median values of the MNF and dmax reduces and reaches a flatter plateau as the R increases. It is observed that the new seismic isolation strategy may result in considerable improvements in the response of the bridge. 7. the isolator 149 . reduces. the highest value of the dmax among the suite of ground motions reduces with the new seismic isolation strategy.
05 0.04 0.30 0.0 0 100 200 300 R (cm) 400 500 600 700 (c) 150 .08 0.20 0.10 0.00 0 100 200 300 R (cm) 400 500 600 700 (a) 0.2 0.0 dmax (%) 0.18 0.15 0.14 0.16 0.0.02 0.12 0.00 0 100 200 300 R (cm) 400 500 600 700 MND (R = 99cm) (b) 1.10 0.20 MNF 0.25 0.8 0.4 0.2 1.6 0.06 0.
3.0 ΣMNFpier / ΣMNFabutment 2.5 2.0 1.5 1.0 0.5 0.0 0 100 200 300 R (cm) 400 500 600 700
(d)
MNDpier / MNDabutment (R = 99cm) 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 0 100 200 300 R (cm) 400 500 600 700
(e)
Figure 7.16 Variation of bridge response quantities as a function of FPS design parameters.
151
150 100 50 f (kN) L 0 50 100 150 16 12 8 max( fL ) = 63.5 max( δL ) = 8.5
150 100 50 f (kN) L 0 50 max( fL ) = 38.9 100 8 12 16 150 16 12 8 4 0 4 8 12 16 max( δL ) = 11.1
4
0
4
δL (cm)
δL (cm)
(a)
150 100 50 f (kN) L 0 50 100 150 16 max( δL ) = 10.0 f (kN) L max( fL ) = 38.5 150 100 50 0 50 100 0 4 8 12 16 150 16 12 8
(b)
max( fL ) = 74.1 max( δL ) = 10.2 4 0 4 8 12 16
12
8
4
δL (cm)
δL (cm)
(c)
(d)
Figure 7.17 Forcedeformation history of the FPS in the longitudinal direction for the Helena earthquake (a), (b) conventional design with R = 99 cm; (c), (d) new design with R = 621 cm; (a), (c) on top of the pier; (b), (d)on top of the abutment.
7.4 Conclusion
In this Chapter the influence of the common bridge design parameters on the system’s response quantities have been investigated. A new seismic isolation strategy using a modified FPS design at the abutments has been presented. The influence of the design parameters in this new seismic isolation approach has been parametrically investigated. The following conclusions have been made: 152
1) The two design parameters that had the highest effect on the response quantities of the SIB were column height, Lc, and superstructure dead weight which was imposed via additional superstructure length, Ld*. Both design parameters also had a notable effect on the fundamental period of the bridge 2) Larger substructure flexibility may result form base modeling assumptions or longer piers. SIBs with flexible substructures acquire higher vibration periods thus lower earthquake forces. However, flexible substructure reduces the
effectiveness of seismic isolation. As the flexibility of the substructure of SIBs increase, the capacity to utilize the isolators at the piers tend to reduce and the deviation between the responses of the isolators located on top of the piers and abutments increase. 3) SIBs with larger superstructure mass acquire higher vibration periods. Increased superstructure mass, reduces the isolator deformations and increases the demands on the substructure components. 4) Increased superstructure flexibility acquired via longer span length dampens the inertial loads caused by the vertical components of the superstructure. This
results in lower isolator forces and structural demands on the substructure. However, increased superstructure flexibility contributed to the variation between the responses of the isolators located on top the piers and the abutments. 5) Bridge skew has negligible effect on SIBs. This is mainly due to the uncoupling of the superstructure from the superstructure. The substructure nominal concrete strength had negligible effect on the response of the bridge. 6) The modified isolator design effectively increases the postyield stiffness and energy dissipation capacity of the FPS utilizing springs widely found in the industry. This new design can be tailored to achieve different isolator forcedeformation properties by changing the design, material and number of springs.
153
8) When compared to the conventional seismic isolation approach. the new strategy reduced the isolator deformation demands by utilizing the isolators more evenly at the abutments and piers. This was achieved using the standard isolator dimensions manufactured by the parent company. 154 .7) The new seismic isolation strategy allowed for a more balanced distribution of isolator forces and displacements throughout the bridge.
The Fiction Pendulum System (FPS) was selected as the representative isolator and given particular emphasis in the analyses. The review of the currentstateoftheart on bridge seismic isolation studies showed that there are three issues that need further clarification: 1) The level of accuracy required for modeling the forcedeformation behavior of isolators. Despite being considered a relatively mature technology. CONCLUSIONS.CHAPTER 8 SUMMARY. Isolators typically govern the dynamic characteristics of bridges and may be a powerful tool to calibrate the system’s seismic response to a desired level by the designer. Currently. One reason for this is the absence adequate analytical models that can capture the highly nonlinear behavior of the isolators.1 Summary and Conclusions Seismic isolation is an effective tool for improving the structural performance of bridges susceptible to earthquake induced loads. This study was intended to provide support for seismic risk mitigation and insight for the analysis and design of SIBs by quantifying response characteristics. sliding seismic isolation has been incorporated in bridge design codes only in 1997 and has not found widespread use particularly in the highway bridge community. AND FUTURE RESEARCH 8. The objective of this study is to enhance the understanding of the structural response of bridges utilizing sliding seismic isolation via models that can capture the complex behavior of the isolators. The dynamic response of bridges may be shifted towards higher periods via isolators to attract lower seismic forces. seismic isolators are classified as sliding and elastomeric. In addition. 155 . current seismic isolation design codes do not provide any guidance about the selection of isolator types. Examination of the affects of bridge and isolator design parameters on the system’s seismic response has been limited.
e. and base and other boundary conditions. The bridge was modeled with a high degree of detail of the substructure components. sliding and elastomeric).2) The comparative response of the two main types of seismic isolators (i. Unlike previous models in the literature. The influence of the modeling assumptions on the force deformation response of the FPS was monitored on the developed for the verification study. The influence of seismic isolation in a simplified bridge model was examined via modal analysis. (3) the characteristics of the isolators are an important determinant of the dynamic response of the bridge. It was shown that: (1) dynamic loads are substantially reduced via the insertion of a flexible element between the substructure and the superstructure. The FE model was validated using experimental data from shake table tests of structures seismically isolated with the FPS. The new 156 . New C++ material and element classes were developed and compiled into the open source framework of OpenSees to model the response of the FPS. 3) The influence of bridge and isolator design parameters on the system’s response. Threedimensional (3D) model of the bridge isolated with the FPS was developed. (2) vibration modes are governed by those involving the isolators. A MultiSpan Continuous Steel Girder (MSCSG) bridge was used for this purpose. The nonlinear timehistory (NLTH) analysis was selected as the method for further examination of SIB response. The response of typical highway bridges isolated with the FPS was investigated as a function of isolator modeling assumptions. this model can account simultaneously for the variation in the normal force and friction coefficient. and the coupling of the vertical and horizontal response during motion. large deformation effects. This model makes use of the existing research findings on the response components of the FPS. A new finite element (FE) model of the FPS was implemented into the existing library of OpenSees.
Pounding between the deck and the abutments may occur due to increased deformations at the isolator level.FE model developed for the FPS was incorporated into the bridge model. It was concluded that. The influence of vertical components of the ground motions on the two isolators was assessed. and bidirectional coupling. Seven other models of the SIB were generated with the only difference in the FPS modeling assumptions. This occurrence limits the effectiveness of the isolators and may lead to unanticipated structural damage. Large deformation effects of the isolators were found to be negligible for the systems considered in this study. Two models of a MultiSpan Continuous Concrete Girder (MSCCG) bridge were generated and one was isolated with the LRB and the other was isolated with the FPS with approximately the same seismic isolation periods. In addition to accounting for bidirectional coupling effects available in models found in the literature. The two bridge models were subjected to NLTH with a suite of 7% probability of exceedance in 75 years hazard level earthquake records. It was concluded that unlike the FPS. this model can account simultaneously for the variation in the normal force and large deformation effects. These parameters may have considerable effects on isolator design. the most important modeling aspects of the FPS in bridge applications are the normal force and friction coefficient variations. the LRB 157 . the effects of the accuracy of these modeling parameters on bridge structural response are weak to moderate. Each model Isolator and bridge response characteristics were monitored under NLTH analyses that utilized 2% in 50 years hazard level earthquakes. The FPS and the LeadRubber Bearings (LRB) were selected as representative isolators for sliding and elastomeric seismic isolators types. respectively. A new FE model of the LRB has been implemented into the existing library of OpenSees using a similar approach adopted for the FPS. Isolator and bridge response characteristics were monitored for the two seismic isolation schemes. The seismic response characteristics of a MultiSpan Continuous Concrete Girder (MSCCG) bridge with: (a) sliding (b) elastomeric seismic isolation was compared using advanced isolator models. However.
The choice of elastomeric or sliding seismic isolation of bridges may have considerable effects despite attaining similar vibration periods.forcedeformation response is significantly affected from large deformations. The implications of the proposed design on the FPS forcedeformation The seismic response of the system was investigated as a response were examined. The LRB acquired higher displacements. Linear regression curves for the median values of the response quantities were developed from the results of the analyses. The range of these design parameters were established based on previous inventory analyses performed in the Central and Southeastern United States. The influence of normal load variations on the forcedeformation characteristics of the LRB was weak. function of different parameters of this new seismic isolation strategy. The FPS is capable of accommodating vertical components of the ground motion records that might result in a buckling failure of the LRB. A new bridge seismic isolation strategy that involves a practical modification in the design of the FPS was proposed. The LRB possesses a more uniform response among isolators located at the abutments and piers in comparison to the FPS. The previously established MSCCG bridge models were used for this task. The slope of these regression curves were used to quantify the significance of the design parameters on the system’s response. There is a notable loss of lateral stiffness in LRB at large deformations that may lead also lead to buckling. Nonlinear time history analyses were performed by changing the design parameters within the four equally spaced values of their established bounds. It was concluded 158 . Five bridge design parameters were selected. Previously selected suite of 7% probability of exceedance in 75 years hazard level earthquake records were used in the analyses. The objective was to increase the effectiveness of seismic isolation with the FPS by achieving a more uniform response among isolators located at the piers and the abutments. The influence of the design characteristics of the bridge and the isolators on the system’s seismic response was examined. placed smaller demands on the columns compared to the FPS. however.
The influence of ground motion characteristics on the seismic response of SIBs should be examined. The structural and dynamic characteristics of these bridges may have significant variations.that seismic isolation with the FPS is less effective for bridges that have flexible substructure characteristics. Bridges with large superstructure mass acquires less isolator displacements but place larger demands on the load carrying elements in the substructure. Unlike nonisolated bridges.2 Future Research The following are possible areas which this research can be extended to: • Ground motions have a wide range of variability in their characteristics stemming from proximity. The effectiveness of isolators on different bridge types should be quantified. It was possible to reduce structural demands on the load carrying elements of the bridge by creating a more uniform response among isolators located at the piers and abutments via the new design. Base conditions and column height were identified as the two parameters to impact substructure flexibility the most. This modification involves installing supplemental springs around the isolator to provide latitude for alternate forcedeformation characteristics. the seismic response characteristics of bridges seismically isolated with the FPS were not affected by skew. • Seismic isolation is applicable to virtually any type of bridge. 8. The two disadvantages of the FPS are: (1) the limitation of the design parameters to attain desired forcedeformation response characteristics and. (2) the significant variation in the forcedeformation responses characteristics of the isolators located at the piers and the abutments. Bridges with large superstructure flexibility tend to be effected less from the effects of the vertical components of the ground motions. frequency content and directivity. compared with those that are less flexible. 159 . A new design approach for the FPS is proposed to overcome these limitations.
• A costbenefit analysis should be performed for a range of isolator and bridge types. This will contribute to the wider application of the seismic isolation technology by making these elements a catalog commodity for design. In addition to considering different isolator types. The influence of isolator modeling assumptions and the selection of the type of seismic isolator on building response should be investigated. Additionally. 160 . and other protective devices such as dampers. • Buildings typically possess larger overturning moments compared to bridges. • Possible improvements with hybrid seismic isolation schemes should be explored. the design objectives and the placement of isolators in buildings have significant differences from bridges. these scenarios should include retrofit techniques based on strengthening such as jacketing.
McKenna 2005b). and a zerolength element class. OpenSees has an open source object oriented architecture in the C++ programming language that maximizes its modularity. the constitutive relationship defined for the FPS in 161 . The setTrialStrain method defined in the parent UniaxialMaterial class is pure virtual base class thus no objects of it's type can be instantiated. A material class. thus making it a viable choice for research purposes. has been added to the existing library of OpenSees to model the forcedeformation response of the FPS (McKenna 2005a. Despite the availability of versatile material and element models inside the library of OpenSees. Classes derived from the UniaxialMaterial class must implement the setTrialStrain method to avoid a compiler error. (2) strain rate (strainRate) at step i of the analysis and update the values of trial stress (Tstress) and trial tangent stiffness (Ttangent) values based on the predefined constitutive relationship for step i+1. This framework has been used to simulate the complex forcedeformation response of the FPS and the LRB isolators. However. The parent UniaxialMaterial class has a method called: setTrialStrain which is designed to take in two parameters: (1) strain (strain) and. none has the capability of effectively modeling the complex forcedeformation response of the isolators described in chapters 4 and 6. These classes were derived as child classes of the existing generic UniaxialMaterial and ZeroLength parent classes. A critical approach used to achieve this task was ‘overriding’. FPSmaterial. 2006). FPSelement.APPENDIX A METHOD OVERRIDING IN C++ Open System for Earthquake Engineering Simulation (OpenSees) is an open source software framework for simulating the earthquake response of structural and geotechnical systems (Mazzoni et al.
double Alpha.diff ).strain.strainRate). mat++) { if(mat==0){ strain = this>computeCurrentStrain1d(0.3).diffv).} } 162 .strainRate). the setTrialStrain is kept inside the code but designed to not function by: int FPSMaterial::setTrialStrain(double strain. Alpha is α (Equation 4.Alpha.strain.diffv). double strainRate) {return 0.diff) strainRate = this>computeCurrentStrain1d(1. double strain. ret += theMaterial1d[mat]>setTrialStrain(frictionC_x. double Alpha. strainRate = this>computeCurrentStrain1d(0.Alpha. double n_modified.Chapter 4 requires additional parameters to the default ones.n_modified. double strain. double n_modified.} A new setTrialStrain method to override the existing default one in FPSmaterial is generated from: int FPSMaterial::setTrialStrain(double frictionC. ret += theMaterial1d[mat]>setTrialStrain(strain.8).diffv). To achieve this. double zz. n_modified is N (Equation 4.strainRate).y1. ret += theMaterial1d[mat]>setTrialStrain(frictionC_y. mat<3.} else if(mat==1) { strain = this>computeCurrentStrain1d(1. double strainRate) {return 0.} else { strain = this>computeCurrentStrain1d(mat. zz is the corresponding parameter of η (Equation 4.12).} The utilization of the new following: setTrialStrain method inside the FPSelement is as for (int mat=0.diff).2). double strainRate) where frictionC is µ (Equation 4.n_modified.y2. strainRate = this>computeCurrentStrain1d(mat. This overriding method is also added to the header file of the parent UniaxialMaterial class: virtual int setTrialStrain (double frictionC. double zz.
degree of skew. The term R2 is the coefficient of multiple determination used to measure the amount of reduction in the variability of the response quantity median by the design parameter variables. ΣMNFpier / ΣMNFabutment. MNDpier / MNDabutment. the ratio of the total MNF transferred to the pier and the total MNF transferred to the abutment. Ld*. MNF and MND. the ratio of the MND on top of the abutment and MND on top of the pier. length of the superstructure with adjusted mass. The term a denotes the slope of the linear regression curve established from the median values of the response quantities as a function of the design parameters. Lc. The bridges design parameters considered are column height. Figures a re given as a ratio of the design parameters to the base model design values denoted by a subscript ‘o’ of the design parameter. fc. Ld. maximum column drifts. and the nominal concrete strength of the substructure. 163 .APPENDIX B REGRESSION LINES FOR THE EFFECTS OF BRIDGE DESIGN PARAMETERS ON SYSTEM RESPONSE The following figures present the variation of maximum normalized force and displacement of the FPS isolators located on top of the pier. length of the superstructure with constant mass. for the earthquakes used in chapter 7. dmax. a.
4 0.067 0.8 1.2 1.4 fco = 34.6 0.04 0.140 0.6 (c) Figure B.6 0.00 0.013 MND 0.8 1.071 0.00 0.20 0.4 1.08 0.140 0.068 0.140 0.434 0.065 0.0 0.463 0.2 1.8 dmax (%) fco = 34.05 0.137 0.2 1.4 1.02 0.080 2 R = 0.0 fc / fco 1.16 0.10 0.002 2 R = 0.4 1.5 MPa a = 0.1 The influence of the pier concrete compressive strength. on (a) MNF (b) MND (c) dmax (d) ΣMNFpier / ΣMNFabutment (e) MNDpier / MNDabutment 164 . fc.12 0.473 0.18 0.988 0.6 (a) 0.2 0.495 MNF 0.10 0.445 0.0 fc / fco 1.15 0.409 0.0 fc / fco 1.30 0.0.20 fco = 34.066 0.5 MPa a = 0.5 MPa a = 0.139 0.0 0.001 2 R = 0.06 0.6 0.8 1.14 0.4 0.6 0.6 (b) 1.25 0.4 0.
4 1.4 0.4 1.5 1.912 2.056 2 R = 0.133 2.0 1.0 ΣMNFpier / ΣMNFabutment 2.6 0.8 0.0 fc / fco 1.6 (d) 1.139 fco = 34.5 MPa a = 0.8 1.789 0.2 1.087 2.810 fco = 34.779 0.801 0.6 0.0 0.1 continued 165 .125 2.2 1.896 (e) Figure B.4 2.760 0.6 0.053 0.106 2 R = 0.0 0.5 0.8 1.4 0.3.0 fc / fco 1.0 0.4 MNDpier / MNDabutment 1.5 2.6 0.2 1.2 0.5 MPa a = 0.
on (a) MNF (b) MND (c) dmax (d) ΣMNFpier / ΣMNFabutment (e) MNDpier / MNDabutment 166 .071 0.8 (a) 0.12 0.965 dmax (%) 0.120 0.05 0.067 0.563 0.18 0.30 0.0 1.6 Lco = 4.25 Lco = 4.2 0.4 1.6 1.6 m a = 0.15 0.16 0.04 0.8 (b) 1.053 0.2 Lc / Lco 1.105 0.10 0.20 MNF 0.2 Lc / Lco 1.10 0.00 0.127 0.8 0.572 0.6 0.4 0.8 1.4 1.8 0.6 1.02 0.972 MND 0.351 2 R = 0.6 m a = 0.2 1.140 0.386 0.137 0.906 0.0.0 1.6 0.042 2 R = 0.2 The influence of the column length.8 1.4 1.0 0.06 0.2 Lc / Lco 1.6 0.6 m a = 0.20 0.0 1. Lc.0 Lco = 4.00 0.063 0.14 0.040 0.671 (c) Figure B.8 1.6 1.044 2 R = 0.445 0.08 0.
4 0.954 1.6 m a = 0.2 continued 167 .6 0.607 0.6 0.5 0.8 0.4 MNDpier / MNDabutment 1.726 1.6 0.6 1.495 Lco = 4.4 1.898 0.3.5 1.0 ΣMNFpier / ΣMNFabutment 2.0 0.8 (d) 1.6 1.6 m a = 0.8 1.0 0.0 0.801 0.792 2 R = 0.0 2.861 1.675 0.133 1.0 1.2 Lc / Lco 1.0 1.2 1.540 Lco = 4.994 (e) Figure B.0 0.8 1.2 Lc / Lco 1.2 0.4 1.5 2.8 0.117 2.497 2 R = 0.
75 2.12 0.00 (b) 1.04 0.16 0.6 0.30 0.3 m a = 0.024 2 R = 0.10 0.459 0.0 Ldo = 30.3 m a = 0.50 1.50 1.08 0.00 0.25 Ld / Ldo 1.065 0.25 Ldo = 30.3 m a = 0.4 0.473 0.918 0. Ld.10 0.0 MND 0.2 1.00 (c) Figure B.060 0.00 1.0.127 0.75 1.455 0.15 0. with constant mass on (a) MNF (b) MND (c) dmax (d) ΣMNFpier / ΣMNFabutment (e) MNDpier / MNDabutment 168 .50 Ldo = 30.06 0.50 1.001 2 R = 0.001 dmax (%) 0.00 1.0 2 R = 0.20 MNF 0.75 2.50 0.135 0.130 (a) 0.00 1.05 0.00 0.153 0.140 0.18 0.8 0.02 0.447 0.75 2.065 0.20 0.060 0.064 0.3 The influence of longitudinal deck length.75 1.00 0.2 0.0 0.14 0.50 0.445 0.75 1.25 Ld / Ldo 1.25 Ld / Ldo 1.
708 2.801 0.50 0.3 continued 169 .3 m a = 0.956 1.148 2.0 ΣMNFpier / ΣMNFabutment 2.00 (d) 1.466 0.5 0.75 1.3.952 0.746 0.727 Ldo = 30.3 m a = 0.222 2 R = 0.25 Ld / Ldo 1.4 0.0 0.00 (e) Figure B.774 0.0 0.939 Ldo = 30.00 1.5 2.4 MNDpier / MNDabutment 1.2 0.6 0.2 1.0 0.25 Ld / Ldo 1.051 2 R = 0.50 1.75 2.133 1.5 1.75 2.0 1.738 0.00 1.75 1.50 1.8 0.50 1.
4 The influence of longitudinal deck length.573 0.50 1.14 0.116 0.18 0.25 Ldo = 30.25 Ld / Ldo 1.10 0.15 0.16 0.50 1.796 MND 0.00 1.061 0. with adjusted mass on (a) MNF (b) MND (c) dmax (d) ΣMNFpier / ΣMNFabutment (e) MNDpier / MNDabutment 170 .12 0.623 0.50 0.3 m a = 0.30 0.25 Ld / Ldo 1.50 0.772 1.00 0.75 1.3 m a = 0.156 0.052 0.072 0.75 1.351 0.50 0.219 0.2 1.061 0.75 2.20 MNF 0.02 0.987 dmax (%) 0.062 0.178 0.3 m a = 0.25 Ld / Ldo 1.75 1.674 2 R = 0.20 0.06 0.975 0.75 2.04 0.00 (c) Figure B.08 0.093 2 R = 0.05 0.00 1.00 (b) 1.0 Ldo = 30.445 0.75 2.140 0.6 0. Ld.00 (a) 0.10 0.0 0.50 Ldo = 30.8 0.2 0.00 1.00 0.016 2 R = 0.0.4 0.
50 1.809 1.75 2.50 1.25 Ld / Ldo 1.724 0.5 2.75 1.331 2.00 (d) 1.5 0.3.0 2.2 1.978 1.50 0.0 ΣMNFpier / ΣMNFabutment 2.0 0.701 0.3 m a = 0.00 (e) Figure B.0 0.5 1.75 2.4 continued 171 .6 0.979 0.3 m a = 0.801 0.00 1.211 R2 = 0.25 Ld / Ldo 1.840 0.877 1.628 Ldo = 30.00 1.2 0.643 R2 = 0.50 0.4 0.673 Ldo = 30.133 1.0 0.75 1.8 0.4 MNDpier / MNDabutment 1.
05 0.067 0 10 20 30 40 50 α (b) 1.530 0.140 0.0 MND α o = 0o 0.0 dmax (%) αo = 0 o a = 0.20 0.140 0. on (a) MNF (b) MND (c) dmax (d) ΣMNFpier / ΣMNFabutment (e) MNDpier / MNDabutment 172 .15 0.445 0.139 0.200 0. α.127 0.001 2 R = 0.10 0.0.00 10 a = 0.30 0.12 0.2 0.067 0.04 0.0 10 0 10 20 30 40 50 α (c) Figure B.0 2 R = 0.4 0.075 0.401 0.02 0.001 2 R = 0.06 0.16 0.00 10 0 10 20 30 40 50 α (a) 0.147 0.25 MNF α o = 0o a = 0.6 0.8 0.14 0.455 0.20 0.18 0.5 The influence of skew angle.08 0.431 0.074 0.10 0.2 1.075 0.397 0.
6 0.0 2 R = 0.842 α o = 0o a = 0.5 10 2.8 0.801 0.0 1.133 2.852 0.002 2 R = 0.0 0.121 1.0 0.4 10 0.761 0.4 MNDpier / MNDabutment 1.5 2.944 α o = 0o a = 0.3.2 1.978 1.157 2.762 0.003 0 10 20 30 40 50 α (e) Figure B.5 1.0 ΣMNFpier / ΣMNFabutment 2.5 continued 173 .112 0 10 20 30 40 50 α (d) 1.
“Stability of elastomeric seismic isolation systems. H (2006).” Journal of Structural Engineering. 247269 Button. RL (1990). A (2004).” Earthquake Engineering and Structural Dynamics. AASHTO (1999). New York 174 . 32.C. Redwood City.” Earthquake Spectra. Cronin. H (1993). “Physical model for dynamic analysis of structures with FPS isolators. M. Mayes. IG and Mayes. Standard specifications for highway bridges.” ATC Seminar on Seismic Isolation. JM (1986). 128(12). Vancouver. D. application and performance – a world review. MC.” 13th World Conference on Earthquake Engineering.C. B. Marcel Dekker. 15511564 Carlson.” Earthquake Engineering and Structural Dynamics. State University of New York at Buffalo Buckle. “Effect of vertical ground motions on seismic response of highway bridges. JC and Llera. Spring designer’s handbook. JC and Llera. IG and Liu.” Joint Sealing and Bearing Systems for Concrete Structures (ACI). “Seismic isolation: history. EL. 27. Constantinou. 305332 Almazan. Guide specifications for seismic isolation design. Technical Report MCEER06SP07. H (1978). California Buckle. IG. Vol. “Analytical model of structures with frictional pendulum isolators. Multidisciplinary Center for Earthquake Engineering Research. Canada Buckle. JCDL (2002). 31.REFERENCES AASHO (1965). 1. Llera. 11571184 Anderson. “Properties of slender elastomeric isolation bearings during shake table studies of a largescale model bridge deck.” Earthquake Engineering and Structural Dynamics. 845867 Almazan. CJ. JCDL (2003). “Modeling aspects of structures isolated with the frictional pendulum system. Washington. JA (1998). JCDL. IG and Kelly. Seismic isolation of highway bridges. Almazan. Passive Energy Dissipation and Control. “An evaluation of bidirectional earthquake shaking on the provisions of the AASHTO Guide Specifications for Seismic Isolation Design. 6(2) Buckle.C. Washington D. Dicleli. and Ghasemi. MR. JL. RL (2002). and Inaudi.. and Mahin.
4th Ed. and Witt. Prentice Hall.. Nimis. California 175 . “The Northridge: progress made. National Center for Earthquake Engineering Research. NP (1961). Kim.” Engineering Structures. 521539 Constantinou.Chironis. Mokha. 5(4). Technical Report NCEER930020. State University of New York at Buffalo Cook. M. New York Chopra. IG. RB. Dynamics of structures theory and applications to earthquake engineering.” Public Roads. Buckle. Vallejo. Friedland. P. A (1990). S (2006). 455474 Constantinou. 11391156 Dicleli. AK (2001). S and Mansour. J. McGraw Hill. RJ (2002). and Okamoto. 10(1). New York Cooper. “Seismic retrofitting of bridges in Illinois using friction pendulum seismic isolation bearings and modeling procedures. A. Kohn Wiley & Sons. lessons learned in seismic design of bridges. Albhaisi. CJ and Elnashai. “Seismic design of lifeline bridge using hybrid seismic isolation. Concepts and applications of finite element analysis. “Static soilstructure interaction effects in seismicisolated bridges” Practice Periodical on Structural Design and Construction. AS (2001). 233250 Dicleli. New Jersey Collier. MY (2003). YS. 2233 Dicleli. S (1993). Tsopelas. RD. 94103 Earthquake Protection Systems (2003).” Journal of Bridge Engineering. 116(2). Plesha. II: modeling. Technical Characteristics of Friction Pendulum Bearings. Lafayette. “A procedure for combining vertical and horizontal seismic action effects.” Journal of Earthquake Engineering. “Teflon bearings in base isolation. “Effect of isolator and ground motion characteristics on the performance of seismicisolated bridges. Spring designer and application. MC. 25. DS. 2nd Ed. Malkus. M (2002). NM (1994). M and Mansour. 7(2). Earthquake performance of structures on leadrubber isolations system. 58(1) DIS (1996).” Earthquake Engineering and Structural Dynamics. M and Buddaram.” Journal of Structural Engineering. NCEERTaisei corporation research program on sliding seismic isolation systems for bridges: experimental and analytical study of a friction pendulum system (fps). MC. and Bobb. Reinhorn.. 35. California Dicleli. ME. MY (2005). IM.
A and Constantinou. and Uckan (2004). “Base isolation for nearfault motions. SK (2005). M.” OpenSees Developer`s Workshop. Y and Jain. 9.” Engineering Structures. AM (1978). Berkeley. GL (2005). 17771797 International Code Council (2000).” Engineering Structures. “Experimental study of isolated building under triaxial ground excitations. “Effect of variation of normal force on seismic performance of resilient sliding isolation systems in highway bridges. 857866 176 .” Journal of Bridge Engineering. Part 1: Provisions. TY (2000). 34. “Brief notes on objectoriented software design and programming with C++. Austria FEMA (1997). Falls Church. NewnesButterworths.” Canadian Journal of Civil Engineering. EF and Brichta. Horwood Publishing Limited. International building code. G (1998). Vienna. 691707 Kasalanati. MC ( 2005). 30. JS and Hsu. 879886 Iemura. A (1988). “Testing and modeling of prestressed isolators. Takahashi. Federal Emergency Management Agency Fenves. “Earthquake response of a 1:4 scaled three story steel structure seismically isolated by the FPS type sliding isolation system. “Seismic response of isolated bridges.” ASCE Journal of Structural Engineering.Erdik. “NEHRP Recommended Provisions for Seismic Regulations for New Buildings and Other Structures. Experimental design techniques in statistical practice.” ASCE Journal of Structural Engineering. RS (2004). and Gettinby. West Sussex. 156166 Jangid. 7278 Göbel. “Optimum friction pendulum system for nearfault motions. “Seismic behavior of highway bridges with base isolation. H.” Earthquake Engineering and Structural Dynamics. Vol. 27(3). HM (1988). 157166 Ghobarah. RS (2004). London Hwang. JM (2001). WP. T. Taghikhany. RS and Kelly. FEMA 302. Virginia Jangid.” Earthquake Engineering and Structural Dynamics. “Seismic performance of highway bridges. California Gardiner. England Ghobarah. 10. 15. A and Ali. 126(8). 131(6).” Report No.” Proceedings of the Third European Conference on Structural Control. No. Rubber spring design. 349359 Jangid. 2.
7.berkeley.” Research Progress and Accomplishments: 20012003. MCEER.” ATC Seminar on Response Modification Technologies for PerformanceBased Seismic Design. Chem. JB.” Journal of Vibration and Control. Advanced engineering mathematics. IG. DK. GJ (1996). and Lee. 26. WE (2004). Loh. University at Buffalo State University of New York Liao. Earthquakeresistant design with rubber. Buffalo. BH (2004).edu/ ~silvia/OpenSees/manual/html/> 177 . FT. Scott. “Influence of isolator characteristics on the response of baseisolated structures. California Mazzoni. “Impediments to the implementation of seismic isolation.” Report No. Kim. Buckle.” Engineering Structures. University of California. S. J (1997). JM. GC. “Open System for Earthquake Engineering Simulation User Manual. RS (2004). Fortran. John Wiley & Sons. MH and Fenves. Los Angeles. “Effects of diaphragms on seismic response of skewed bridges. and Premus. IG (2002). Ordinary and partial differential equation routines in C.” Journal of Engineering Mechanics. December. CH.” <http://peer. 935952 Maleki. VA and Jangid. S (2001a).” 1First MIT Conference on Computational Fluid and Solid Mechanics. Berkeley Kreyszeig. London Kelly. Report No. New York Lee. J (2003). Tsai. 8th Ed. Boca Raton Lee. 21732183 Maleki.” Engineering Structures. and Matlab. HC (1986). 13631368 Kelly. NCEER 960014. Y and Buckl. Kitane. 26(2004). HJ. SS. Java. and Schiesser. W. NY Matsagar. NCEER. C++. Earthquake simulator testing of a baseisolated bridge deck. 17351749 Mayes. Maple. “Tension buckling in multilayer elastomeric bearings. GL (2007). Chapman & Hall/Rc. “Free vibration of skewed bridges. McKenna. SpringerVerlag. E (1999). “Comparison of dynamic response of isolated and nonisolated continuous girder bridges subjected to nearfault ground motions. R (2002). “Response of steel bridge bearings to the reversed cyclic loading. S (2001b).Kelly. June 1215 2001 Mander. “Literature review of the observed performance of SIBs. EERC 198509..
130(3). Iemura.” OpenSees Developer`s Workshop.” 12th World Conference on Earthquake Engineering. Berkeley. GL (2004). 240261 Mokha. G.” Journal of Structural Engineering. 455475 Mokha. H. F (2005a). California McKenna. AS. Constantinou. John Wiley & Sons. “Characterization and modeling of friction pendulum bearings subjected to multiple components of excitation. 117(2). Constantinou. Ogawa. B (2005). 116(2). Ordonez. Reinhorn. Whittaker. MC. Wiley. Design and analysis of experiments. Fenves. K. (2000). Reinhorn. AM (1990). Vancouver. F (2005b). 119(1). 1st Ed. “Pseudo dynamic tests and implementation of sliding bridge isolators with vertical motion. “Comparative study of the inelastic response of base isolated buodings. II: modeling. Constantinou. AS. 433442 Muthukumar. 439454 Montgomery. 622626 Mokha. Constantinou.” Journal of Structural Engineering. AS. “Verification of friction model of Teflon bearings under triaxial load. 151164 178 . (2003). Georgia Institute of Technology. Design of seismic isolated structures. Mosqueda.S. Takahashi. Georgia Institute of Technology. AS. AM (1991). “Adding an element into OpenSees. New York Nakajima. DC (2005). S. MC. 6th Edition. L (2003). AS.” PhD thesis. Naeim. MC. U. K. Canada Nielson. MC. “Introducing a new uniaxial material into OpenSees.” PhD thesis. 32.” ASCE Journal of Structural Engineering. “Analytical Fragility Curves for Highway Bridges in Moderate Seismic Zones.” Journal of Structural Engineering. I: testing.McKenna.” Earthquake Engineering and Structural Dynamics.” Journal of Structural Engineering. Berkeley. “Teflon bearings in base isolation. “Further results on frictional properties of teflon bearings. Reinhorn. Reinhorn. “Teflon bearings in base isolation. AM (1990). F and Kelly. D.. J (1996). D and Bozzo.A. Foti.” OpenSees Developer`s Workshop. 116(2). “A Contact Element Approach with Hysteresis Damping for the Analysis and Design of Pounding in Bridges. AM (1991). Y. California Mokha.
Cambridge University Press. “Dynamic behavior of R/C highway bridges under the combined effect of vertical and horizontal earthquake motions. KL and Chopra. New York Pritchard. 2007 from http://peer. McGrawHill Science Rosenbrock.berkeley. Gaithersburg. Amin.” Proceedings of the FHWA/NCEER Workshop on the National Representation of Seismic Ground Motion for New and Existing Highway Facilities. 329330 Ryan. B. 25.Palfalvi. CA. and Flannery. and Futch. P (1993). and Thompson. San Francisco. GM (1996). Gladwell. Guidelines for prequalification. EERC 200501. 11091137 PEER Strong Motion Database.” Earthquake Engineering and Structural Dynamics. Retrieved October 10. Maryland Shampine. HW (1996). AJ and Elnashai. DA (1991). H. John Wiley and Sons. A. MathCad: a tool for engineering problem solving. HH (1963). AS (1996). Report No. New York Silva. “Characteristics of vertical strong ground motions for applications to engineering design. S (2003). 5(4). Passive Energy Dissipation and Active Control. and Lee. WH. Berkeley Saadeghvaziri. Technical Report NCEER970010 179 . Mokha. PJ (1998). University of California. I. MJN. National Institute of Standards and Technology. prototype and quality control testing of seismic isolation systems. Vetterling. (2000). F and Calvi. BP (2003). Solving ODEs with Matlab. rocking and axialload effects. SA. Estimating the seismic response of baseisolated buildings including torsion. 20. 535549 Shenton. March. Numerical recipes in C++: the art of scientific computing. Cambridge Priestley. Cambridge University Press. “Some general implicit processes for the numerical solution of differential equations” The Computer Journal. Teukolsky. A. LF. Fatehi. Papazoglou.edu/smcat/ Press. WT. “Implementation issues in seismic isolation retrofit of government buildings.” Earthquake Engineering and Structural Dynamics. WJ (1997). AK (2005). “Analytical and field evidence of the damaging effect of vertical earthquake ground motion. Seismic design and retrofit of bridges. MA.” ATC171 Seminar on Seismic Isolation. Seible.
17(2). Fundamentals of seismic protection for bridges.. 5th Edition. Vancouver. 38 Tyler. Univ. T (2004). MC (1996). “Performance estimates in SIB structures. 7(2). MJ (2003). 251274 Tyler.” Proceedings of the U. P and Constantinou. Vol. “The 1997 AASHTO seismic isolation guide specification. “Development of design procedure for bridges on leadrubber bearings”. An introduction to seismic isolation.. John Wiley & Sons. Virginia Thakkar. of MissouriRolla. N. Rolla. Yanagawa. 1st Ed. Third Int. M and Karshenas. SK and Maheshwari R (1995). on Recent Advances in Geotechnical Earthquake Engineering and Soil Dynamics. GP and Whittaker. Iemura. Robinson. J and Roeder. C (1991).” Bulletin of the New Zealand National society for Earthquake Engineering.” Engineering Structures. Earthquake Engineering Research Institute. 1.Italy Workshop on Seismic Protective Systems for Bridges. CC (2001). AJ (1989).” 13th World Conference on Earthquake Engineering. “Experimental study of FPS system in bridge seismic isolation. “Shaking table test for frictional isolated bridges and tribological numerical model of frictional isolator. 90105 Vlassis. RI. Vol. 24. DH. “Rubber bearings in baseisolated structures – a summary paper. AW and Igusa. Missouri. California 180 .S. S. “Advantages and limitations of seismic isolation. PJ and Carr. WH (1984). H. “SIB piers on shallow soil stratum with soilstructure interaction. New York Stanton. “Highstrain tests on leadrubber bearings for earthquake loadings. Engineering Structures. 79. Henry Holt & Co. Primer on seismic isolation. Hibi. GH (1993). AG and Spyrakos.” Computers and Structures. 11.” Earthquake Spectr. “Study of seismic base isolation of bridge considering soil structure interaction”. Cooke. 301323 Stevens. 1st Ed. AS (2004). M (2004). Moss. 26. Technical Report MCEER980015. 397400 Tsopelas.. New York Takahashi. Teach yourself C++. RG (1991). Canada Taylor.” Earthquake Engineering and Structural Dynamics. New York Stanton.” Bulletin of the New Zealand National society for Earthquake Engineering. RG and Robinson. Y. A (1997). and McVerry. Conf. JF (1998). 6578 Turkington. WH. 28472861 Warn. American Society of Civil Engineers. 25.Skinner. 12611278 Yashinsky.
VA and Low. University of California. SA (1987).” 8th Canadian Conference on Earthquake Engineering. Report No. EERC 198701. Berkeley 181 .” Journal of Bridge Engineering. VA and Low. M (1998). “Performance of bridge seismic retrofits during Northridge earthquake. Canadian Association for Earthquake Engineering. 3(1). The fps earthquake resisting system experimental report. Vancouver. S (1999). and Mahin. 114 Zayas. “Seismic isolation for extreme cold temperatures. SS. 391395 Zayas.Yashinsky.
1980 in Susurluk.S. Turkey in 2002. degree in the same department. Mechanics. He continued to pursue a Ph. Istanbul.VITA Murat Eröz was born on November 3. 182 . degree in Civil Engineering from Istanbul Technical University.D.S. with a minor in Structural Mechanics. He earned a B. Georgia Institute of Technology. Turkey. degree in 2003 from the Structural Engineering. He earned an M. His major concentration was in the area of earthquake engineering and seismic isolation of structures. and Materials group of Civil Engineering.