by
VOSSI REICHMAN
September 1996
Doctor of Philosophy
ABSTRACT
Current stateofpractice methods used for seismic analysis and design of bridge
structures are based on modified elastic spectral methods. These methods are unable to
address certain critical issues that may be peculiar to bridge structures, namely: cyclic
plastic behavior of structural elements, soilstructure interaction, differential ground
motion, the nonlinear modeling of base isolators and supplemental damping devices. To
address these deficiencies, this research develops two evaluation methods based on (i)
nonlinear time history analysis including the effects of the transient development of dam
age in structural components, and (ii) pushover procedures based on static and dynamic
lateral loading approaches. Using the latter, it is possible to assess the global seismic
capacity of an entire bridge structure which can be used in conjunction with seismic
demand determined from inelastic spectral techniques. Classical pushover analysis uses
lateral loads that are increased in a monotonic fashion until the bridge reaches its failure
limit state, this defines displacement capacity. This approach assumes a firstmode distri
bution of lateral forces and this distribution of forces is apt to miss higher mode effects
which may be important, particularly for long bridges. Therefore, several procedures are
considered to address this issue including: adaptive modal distribution, acceleration
ramps, and acceleration pulses. The latter two, being dynamic procedures, are generally
capable of capturing multimodal and total dynamic effects.
Several new modeling features are addressed in this research by the development of a
3D computational platform (IDARCBRIDGE). First, a new triaxial model is developed
that is capable of accurately developing the behavior of sliding isolators including the
influence of the changing vertical force and velocity on the friction coefficients. Second,
modeling of the connection between the bridge deck and the substructure is advanced.
This includes a macroelement representation of the deck seat with the effect of the spatial
layout of bearings and diaphragms by using features of elastic elements, hinge, spring, end
releases and rigid end block transformations.
Finally, the proposed computational procedures are verified by comparing predictive
results with a combination of closedform analytical solutions, laboratory and field exper
imental results, as well as existing computational software. Several case studies are then
used to explore the utility of the proposed computational framework with a particular
emphasis on capturing observed failure modes.
ii
Acknowledgments
This dissertation could not have been completed without the help of many people
contributing from their valuable time to edit it. I would like to thank my major advisor,
Prof. A.M. Reinhorn, for introducing me to an exciting field of research, and escorting me
along the way. I have special regards for Prof. M.C. Constantinou. His remarkable dedica
tion to teaching and research introduced me to the field of base isolation and passive con
trol and inspired me in my work. I would like to thank Prof. J.B. Mander for introducing
me to the subject of capacity design, which was an important part of my work. Special
thanks to Prof. M.P. Gaus, for his help and support in hard times, and for providing me
I am sure I could have never been able to perform research at this level without the
encouragement and support from Prof!. Sheinman from the Technion, ISRAEL, and other
knowledge.
It is obvious that without true friends who have offered their support through this
roller coaster experience, success would not have been possible. Dr. Roy F. Lobo and Dr.
Chen Li are such friends. I would like to mention also all the other friends I acquired
My arrival to V.B. would not have been possible without the initial support and the
iii
Moshe Lasry. Mrs. Efrat Eshcol, Mr. Dan Levi, and Mr. Shlomo Abramovich who have
Last but not least, I would like to thank my family for encouraging me on this dif
ficult but rewarding path: my parents, Ester and Aharon Reichman, ISRAEL, my brother
Hillel Reichman and his lovely family, and especially my dear wife, Roxana Reichman,
iv
Dedication
v
Table of Contents
1.0 INTRODUCTION .................................................................................................. 1
ART ........................................................................................................................ 32
vi
2.3 Damage Analysis ............................................................................................................. .36
3.1 Bridge Structure and Structural Components  Descriptive Presentations ....................... .40
3.2.2 Inelastic Hysteretic Beam Columns with Nonlinear Shear and Flexure Behavior
3.2.3 Shear and Bending Wall Panel Model (for modeling Abutment) ....................... 65
vii
3.3 Modeling of Bridge Subsystems ........................................................................................ 89
3.5.3.3 Conjugate Gradient Iterative Solver (Kincaid et at. 1990) ............. 118
viii
3.6.2.3 Adaptable (Modal) Loading ........................................................... 129
4.1.1 Case Study #1: Analytical Evaluation of Small Scale Bridge System .............. 140
4.1.2 Case Study #2: East Aurora Bridge  Field Test.. .............................................. 149
4.1.3 Case Study #3: Bridge in Los Angeles  Retrofit Solution ................................ 159
4.1.4 Case Study #4: Bridge over Los Angeles River Retrofit Solution with Base Iso
4.1.6 Case Study 6: Evaluation of Bridge Capacity Using Different Push  Over Proce
ix
List of Figures
Fig # Title Page #
3.2. Connection Between Deck and Bent Using Rigid Arm and End Springs (a) 46
Physical Description (b) Regular Modeling (c) Modeling with Rigid Arm
3.9. Example Frame, for Illustration of Combined Shear and Flexure Model 63
3.12. Shear Wall Definition (a) Shear Wall (b) Shear Wall Modeling for Analysis 67
3.14. Triaxial Sliding Isolator, a) side view, b) sliding force  displacements rela 74
x
3.16. Analytical Loops of Frictional Force and Displacements in 8 Shaped Motion 80
Test
Tests
Springs.
4.2 Input Ground Motion (Japanese code, level 2, ground condition I) 143
xi
4.4 Experimental Derivation of The Sliding Bearings Frictional Characteristic 145
4.14 Modeling of Bridge Deck Connection to Abutment (a) Modeling with Stiff 157
4.15 Comparison of Analytical and Experimental Results for Bridge on Steel 158
Bearing at Abutment.
4.16 Comparison of Analytical and Experimental Results for Bridge on Rubber 158
Bearing at Abutment.
4.17 Bridge Top View and Typical Bent Cross Section 164
xii
4.23 Relative Defonnations in X Direction, Between Ground a9d Deck Disp., at 170
4.24 Relative Defonnations in X Direction, Between Ground and Deck Disp., at 171
4.25 Shear Forces in The Pier's X Direction, for Piers Bl and B5, for Isolated 172
4.26 Shear Forces in X Direction, for Nonisolated Bridge, at Base of piers 173
4.28 Relative Defonnations in Z Direction, Between Ground and Deck Disp., at 175
4.29 Relative Defonnations in Z Direction, Between Ground and Deck Disp., at 176
4.30 Shear Forces in The Pier's Z Direction, for Piers B I and B5, for Isolated 177
4.31 Shear Forces in The Pier's Z Direction, for Nonisolated Bridge, at Their 178
Base
4.32 Shear Forces in The Pier's Z Direction, for Piers Bland B5 179
4.35 Base Isolated Bridge with and without Expansion Joints 187
4.37 Displaced Shape of Typical Frame in the (a) Original Bridge, (b) Isolated 189
Bridge
xiii
4.38 Comparison of Shear Force in Non Isolated Bridge, and Bridge Without 190
Gap
4.39 Comparison of Shear Forces in Original and Retrofitted (Alt.1) Bridge 191
4.40 Comparison of Shear Force in Isolated Bridge, and Bridge Without Gap 192
4.41 Comparison of Deck Displacements in Isolated Bridge, and Bridge Without 193
Gap
4.42 Comparison of Deck Displacements in Isolated Bridge, and Not Isolated 194
Bridge
4.43 Comparison of Deck Displacements in Non Isolated Bridge, and Bridge 195
without Gap
4.53 SR141I5 Separation and Overhead (Southbound)  General Plan and Eleva 208
tion
4.54 SR141I5 Separation and Overhead (Southbound)  Typical Deck Section 209
4.55 SR141I5 Separation and Overhead (Southbound) Pier Sections and Rein 210
forcement.
xiv
4.56 SRI4/I5 Separation and Overhead (Southbound) Structural Model for 211
Analysis
4.57 Distribution of Shear Forces in Piers 27 for Different Analysis Procedures 217
4.59 Response of the Bridge to Non Adaptable Uniform Distributed Load (rela 220
tive to mass). (a) Displaced Shape at Failure, (b) Loading History, (c) Shear
4.60 Response of the Bridge to Ramp Loading. Acceleration Rate: 0.25m! 221
sec3 (a) Displaced Shape at Failure, (b) Loading History, (c) Shear
4.61 Response of the Bridge to Ramp Loading. Acceleration Rate: 0.5 m!sec 3 222
(a) Displaced Shape at Failure, (b) Loading History, (c) Shear History for
Piers 27
4.62 Response of the Bridge to Ramp Loading. Acceleration Rate: 5m!sec3. (a) 223
Displaced Shape at Failure, (b) Loading History, (c) Shear History for Piers
27
4.63 Response of the Bridge to Pulse Loading 0.1 sec. (a) Displaced Shape at 224
Failure, (b) Loading History, (c) Shear History for Piers 27
4.64 Response of the Bridge to Pulse Loading 0.05 sec. (a) Displaced Shape at 225
Failure, (b) Loading History, (c) Shear History for Piers 27
4.65 Response of the Bridge to Dynamic Impulsive Loading. (a) Displaced 226
City Hall (input motion), (c) Shear History for Piers 27
xv
A.I Program General Structure 248
xvi
LIST OF TABLES
4.2. Geometry and Structural Elastic Properties of the Bridge Elements 207
4.3. Transversal Moment Curvature Relations and Shear Capacity of Piers 207
4.4. Comparison of Shear Force Distribution Between the Piers at the Time of 216
4.5. Moment at Bridge Piers due to Different Pushover Procedures at Shear 218
Failure
PushOver Procedures
xvii
1.0 INTRODUCTION
Bridges are an important element in modem urban societies and are vital to normal
event, bridges can pose a significant life hazard if failure should take place and the loss of
bridges may cause severe difficulties in emergency response to the earthquake as well as
methods for the analysis and prediction of performance of designs for new bridges and the
Specific advances made in this dissertation include the development of new struc
tural element models to analyze situations which were previously difficult to model or
which could not be modeled, the development of structural element models for new tech
nologies such as base isolation and improved methods to model and analyze structures
which have loosely connected elements and the assembly of complex nonlinear structural
elements into a working computer program which provides a capability to model the
dynamic 3D behavior of complex bridge structures and their supporting systems. One of
the unique features of the work is the integration of considerations of design and analysis.
This is accomplished by using two evaluation methods in which each of the methods pro
vides a capability for design which is not well handled by the other. The methods used are
nonlinear dynamic damage analysis combined with a pushover analysis specially devel
world as a result of earthquakes over the centuries. Unfortunately, there is not much docu
mentation available for. these occurrences. In addition the design and construction of
bridges has undergone many changes in the past 50 years as a result of new materials and
equipment being available and the large growth in the use of motor vehicles. This is par
ticularly true in the U.S. where the number of motor vehicles has shown large growth and
urban areas and for crosscountry travel and transportation of goods. Similar expansions
have recently been underway in other countries. Unfortunately, the inclusion of earth
quake loadings in the design of many bridges was not a major design consideration for
most of these bridges. Thus many bridges were built without adequate provision for earth
quake loadings nor was the state of knowledge on earthquake loadings and bridge design
The San Fernando Earthquake (1971) in the Los Angeles area resulted in the loss
of many major freeway bridges and the loss of $6.5M. This event helped to draw attention
to the need to utilize the highest level of analysis and design for bridges which may be
subjected to earthquakes.
contained in a report by Imbsen and Penzien (1986). Among the failure mechanisms
1) Tilting of substructure
2) Supports displacements.
2
3) Settlements of abutments.
Examples of these types of failures and resulting damage can be found by examin
ing the Alaska earthquake (1964), in which most of the damage to bridges was caused by
large ground displacements, settlements and loss of the bearing capacity of the main struc
tural systems of the bridges. Dynamic inertia effects were not observed as a primary cause
of the bridge damage. During the period up to 1971 several earthquakes occurred in Japan
and California but these earthquakes caused insignificant amounts of damage to bridges.
For this reason, the design codes for bridges up to that time put very little emphasis on
provisions for dynamic seismic design of bridges, and designs were largely based on
The San Fernando Earthquake (1971) initiated a series of changes in the approach
bridge damage was caused during this earthquake due to dynamic inertia effects, and the
evaluation of the bridges for design started to change to consider these dynamic effects.
This change require the consideration of the ductility of the bridge components which
could be subjected to dynamic inertia effects. Particular emphasis started to be paid to sup
Since the 1971 San Femando event, a number of extremely destructive earth
quakes have occurred in the vicinity of highly populated cities in California and Japan.
The 1989 LomaPriata, the Northridge (Goltz 1994) and Kobe (Aydinoglu et al. 1995)
earthquakes caused a large amount of damage to bridge structures and highlighted the
3
need to improve analysis techniques for new and existing bridges and the need to retrofit
The extensive damage and the surprising levels of ground accelerations (up to 20),
presented the need for more accurate and scientific evaluation and design method of
bridges.
distance
(a) (b)
must be considered carefully to avoid failures of the bridges during earthquake shaking.
b) The bridge may have many supports, not all of which are directly connected to
the deck.
4
c) The supports of the bridge may be spaced a long distance apart and are not gen
erally connected to one another. This may result in differences in the dynamic
loading of each pier due to phase differences in the ground motion waves arriv
d) The geologic conditions under each pier may have significant variations result
of supporting elements.
f) The effects of nonlinear behavior of the bents may result in variation of stiffness
Thus the behavior of bridges is different from the structural behavior of most of
the other types of structures dealt with in the civil engineering profession and require spe
cial attention in analysis and design. Perhaps the most commonly analyzed class of struc
structural unit, which can be characterized as a collection of shear elements, frames and
cantilever beams (Fig. 1.1 (a», all of which are well connected. A gradual variation of
stiffness and mass distribution is reflected in well defined dynamic behavior. When there
are abrupt changes in stiffness, as is the case when the first story of the building is open
without walls (soft first story) this can cause a concentration of dynamic rotations
response at the first story level and has often been observed as a cause of failure of build
ings.
5
In bridge structures (Fig. 1 (b» the dynamic, mass and stiffness properties change
along the height and the length of the bridge. In the vertical direction, the stiffness of the
bridge system is changing from the foundation stiffness to the pier, bent or end wall stiff
ness and then to the deck level which may have a higher level of stiffness. In the horizon
tal direction, the height of the piers (or bents) may vary and therefore their stiffness may
be different. This factor is even more important if nonlinear deformations occur in which
there could be a significant reduction in some of the pier stiffnesses due to plastic behav
ior. The stiffness properties of the bridge system can change during the duration of a
dynamic loading due to relative movement of the bridge sections at the thermal expansion
joints.
record of ground motion acceleration and displacements. The ground motion effects on a
bridge (Fig. 1.1 (b» can be quite different between the supports of the bridge. This can
1) If the bridge supports are located far away from one another, the arrival time of
2) Different types of soil amplify differently the bed rock motion, and causes dif
Both of these phenomena may result in quasistatic and additional dynamic forces
6
1.2 The Use of 3D vs. 2D Models
For purposes of analysis it is necessary to construct a mathematical model of the
physical bridge structure. The subsequent analysis is then carried out using the mathemat
ical model. If the mathematical model can adequately represent the true behavior of the
bridge, a satisfactory prediction of the true behavior of the bridge will be obtained. Thus it
is imperative that careful attention be paid to the generation of the mathematical model
and that the model represent both the topological and physical behavior of the materials of
the bridge into the range of dynamic nonlinear behavior. For complex loosely connected
structures the behavior of the structure is always 3dimensional and the model and analy
Seismic demand as well as the seismic capacity evaluation are dependent on the
structural model and must be fully represented. Thus great care must be exercised in the
bution and special connectivity rules. In bridge structures the deck and the supports form
longitudinal frames which are frequently represented in two dimensional models. Individ
ual supports, piers or bents are also frequently represented by 2d models to describe the
transverse vibration of the bridge. Analyses are then carried out in the 2D directions for
loadings in these directions. If a loading is not in one of the 2D directions an attempt is
even such combinations fail to adequately represent 3D responses which can occur in
Irregular bridges with skewed supports to the deck, with multiple gaps skewed to
the bridge axes, curved deck bridges, etc. cannot be modeled by two dimensional models.
7
Moreover, regular bridges with symmetrical geometrical construction, but with unknown
and also the capacity of the bridge. It should be noted however, that when the bridge is
requires a somewhat arbitrary and careful selection of contributory modes. The selection
tion of the modes. A discussion of this problem can be found in Clough and Penzien
(1975).
As can easily be seen there are several issues which require complex modeling of
the soil and bridge structures using a 3D approach. The spatial variation of propagating
ground motion, the soil supports (springs), expansion joints and seismic isolation systems
in bridges all are influenced by three dimensional motion and therefore three dimensional
considerations should be used. Two dimensional models can provide only limited infor
mation, which in many cases may be nonconservative in predicting the overall behavior of
the bridge.
nerability of bridges to seismic events. These protective systems include base isolation
8
The influence of added protective systems does influence the structural configura
tion of a bridge, such as in the case when a sliding isolation bearing changes from "stick"
to "slip". Initially the substructure and the superstructure act as a single dynamic unit but
when "slip" occurs the bridge behaves as two different dynamic entities. Turkington et. al.
(1989) observed that the equivalent damping of an isolation system is a function of the
type of the earthquake. In an earthquake in which ground shaking continues for a long
period of time the effective damping is higher than in an impulsive earthquake in which
the strong ground shaking is of relatively short duration. Therefore the only way to cap
ture the true behavior of the isolation system and it's effect on the overall response of the
bridge is to use nonlinear dynamic analysis which can take into account timedependent
damping effects.
behavior of isolation elements used for nonlinear dynamic structural analysis is presented
in section 3.0.
The essential features that need to be modeled to capture the behavior of isolation
bearings are:
(i) The appropriate shear stiffness representation in the pre and post yielding
range;
9
Bilinear or trilinear models can be used to model isolation elements like leadrub
ber bearings and mild steel dampers. Lee(1980) has used the bilinear model for modeling
leadrubber bearings. Many Japanese researchers (Yasaka et al. 1988 and others) have
used the bilinear model for modeling leadrubber bearings, and steel dampers. Fujita et
al.(1989) have used the bilinear model with modifications for modeling high damping
elastomeric bearings. The trilinear model has been used by Miyazaki et al.(1988) for mod
eling lead rubber bearings. The RambergOsgood model (1943) has been used for model
ing high damping elastomeric bearings by Yasaka et al.(1988) and Fujita et al.(1989).1t is
A Coulomb model in which the transition from stick to sliding mode and vice
versa is controlled by stickslip conditions was described by Mostaghel et al. (1988) and
Su et al. (1987). This model has been used for modeling sliding bearings. The viscoplastic
model for sliding bearings proposed by Constantinou et al. (l990b) has been used for
modeling sliding bearings. The Viscoplastic model proposed by Ozdemir (1976) has been
used for modeling steel dampers by Bhatti et al.(1980) and by Fujita et al.(1989) with
modifications for modeling high damping elastomeric bearings and leadrubber bearings.
The viscoplastic or the rate model captures most of the essential features of these bearings.
visocplasticity model under certain conditions (Constantinou et al. 1990b) had captured
most of the features. Plasticity based yield surface models have been used to model lead
rubber bearings and high damping elastomeric bearings (Tarics et al. 1984).
10
Experimental evidence in tests on steel dampers and high damping bearings
(Yasaka et al. 1988) reveal the importance of biaxial interaction. Japanese researchers
(Wada et al. 1988, Yasaka et al. 1988, Nakamura et al. 1988) have used the multiple shear
spring model to account for biaxial effects in steel dampers, lead rubber bearings and high
damping elastomeric bearings. This model consists of a series of shear springs arranged in
a radial pattern. The plasticity based yield surface model has been used by Tarics et al.
(1984) for modeling lead rubber bearings and Mizukoshi et al. (1989) for modeling lami
Constantinou et al. (1990b) have used a differential equation model for biaxial
interaction which was proposed by Park et al.(1986), and which is an extension of the
model proposed by Wen(1976) for uniaxial behavior. This model gives accurate modeling
in the elastic and fully plastic zone but not in the transition zone, where the change from
elastic to plastic behavior occurs. To use this model efficiently in an analysis code, the
pseudo force method must be used (Nagarajaiah et al. 1989). This kind of modeling might
tionalproblem is that the influence of the vertical load cannot be efficiently incorporated.
was carried out by Li and Reinhorn(1995), which included several models that can be
used for modeling of complex dampers behavior. A summary of these models is presented
below:
11
The behavior of linear dampers is velocity dependent, in which the damping force F d
(EQ 1.1)
where C is a constant.
When this linear damper is subjected to a harmonic loading of frequency Q the rela
l(~J2
CQu o
+ (.!!:..)2 = 1
oU
(EQ 1.2)
(EQ 1.3)
• Kelvin Model
This model is used to model behavior of an element with both velocity and displace
12
• Maxwell Model
For an element with a strong frequency dependency such as occurs in fluid viscous
(EQ 1.6)
where A = CD/ KD and CD is the damping coefficient at zero frequency and KD is the
• Wiechert Model
MaxwellKelvin model can be used to model damping elements with bituminous fluids.
(EQ 1.7)
Additional models based on Maxwell and Kelvin models with fractional deriva
tives (Bagley and Torvik 1983, Makris 1991 and Kasai et al. 1993) can be used to define
It should be noted that all the models presented are based on properties related to
harmonic loading, this type of modeling might be adequate only if the nature of the earth
13
quake is vibratory. In general for nonlinear analysis, a better quantity to relate the proper
ties of the damper to is the instantaneous velocity. This can be achieved by using nonlinear
stiffness and damping properties ((C(U) and K(u, u)) to represent the behavior of the
element.
In many cases the behavior of a component is very complicated, and sophisticated, com
plex methods are required to obtain sufficiently accurate results. Those methods might be
adequate for some cases but not for others if the other models are very sensitive to the size
of the increment of force, or displacement. In these latter cases very small time steps are
required to accurately follow the force displacements curves which are coupled with Itera
tive solution. While an iterative solution is not difficult to perform when a single element
is considered, in the presence of several nonlinear elements in a structural system, the con
vergence of the entire system is not guaranteed. While the solution for one element is con
verging, the solution for another element might be diverging. This problem is especially
severe when loading and unloading curves are very different, and the search for the equi
In order to overcome the difficulties presented above, the use of iterative proce
dures should be avoided as much as possible. A solution based mostly on components that
contribute directly to the global stiffness and damping matrices is preferred. A contribu
14
tion of pseudo forces to the forcing vector in the governing equations should be avoided as
much as possible. Where necessary, a combination of stiffness damping and pseudo force
should be used.
structural analysis dominate today: (i) Micromodeling  Concrete elements such as beam
columns and wall, are discretize into small elements, in which the actual material proper
ties are introduced. This type of analysis requires the use of very large number of ele
and columns are modeled using hysteretic rules for an entire element. Using this approach
although the accuracy is compromised somewhat the solution is more feasible and
requires much less computational resources as well as providing better significant engi
neering insight.
this dissertation. By using this approach, even though the exact modeling of the compo
nents is slightly compromised, the major characteristics contributing to the overall system
behavior are captured. This way, a good understanding and modeling of system behavior
placement or force history can be extracted from the global analysis, and component
15
1.5 Soil Structure Interaction
The correct evaluation of the dynamic properties (mass, damping, stiffness) of the .
the analysis requires that the ground be included as part of the structural model. At the
present time, even with the advanced computational facilities which are available, this is
Therefore simplified methods are generally used for the representation of soil
structure interaction. Some of these methods utilize a superposition approach in the fre
quency domain, and utilize nonlinear hysteretic models in the time domain.
lent linear stiffness and damping properties to represent soil properties but to use a fully
nonlinear modeling for the superstructure. The use of this simple approximation which
requires a careful evaluation of the soil conditions and rigorous engineering judgment of
the equivalent linear damping and stiffness of the soil, is the best way to model a system
which is expected to have nonlinear structural behavior at the present time for two rea
sons:
(i)Information on soil properties has a much higher uncertainty then the other com
from more rigorous soil modeling is limited and would not improve the accuracy of the
results.
16
1.6 Damage Evaluation in the Inelastic Response of Structures
The design philosophy prevalent over the past 30 years for structures which may
undergo inelastic response with hysteretic energy dissipation and possibly result in perma
nent residual deformations. By allowing relatively large displacements associated with the
nonlinear behavior the total energy input in a structure is reduced due to the dynamic soft
ening of the system. This results in producing smaller force (response) demands on other
elements of the structure. The benefit of the inelastic behavior on the strength require
ments of structural members, may be limited however, by the damage inflicted to the
carrying capacity during other future seismic events. Another result of this large displace
ment may be degradation in the functionality of the structure after the earthquake event.
A survey on the use of damage indexing was done by Valles et al. (1996)
et al. 1985 Powell and Allahabadi 1988) because the future occurrences of earthquakes,
deterministic analysis can provide valuable information on structural behavior. The need
for assessment of condition of structures and their performance during earthquakes led to
the development of damage indicators (surveyed by Chung et al. 1987, Manfredi 1993,
Williams and Sexsmith 1994) provide a way to quantify numerically the seismic damage
Although such indices incorporate many uncertainties resulting from modeling assump
tions, from incomplete knowledge of actual and current material properties, and from
17
future unknown ground motions, the relative ratio of demands and capacities can provide
designed.
lative displacements, or energy dissipation (Williams and Sexsmith 1994, Manfredi 1993,
Powel and Allahabadi 1988, Chung et al. 1987). Numerous indices can characterize the
physical quantities mentioned above at the level of individual members (columns, beams,
walls, etc.), at the level of substructures (bents, abutments), or at the global level.
mum deformations, rotations, or curvatures in respect to the yielding levels (Banon et al.
1981). More recently, the permanent (unrestored) deformation demand (uu y) related to
the ultimate capacity of permanent deformation (uuuy) was suggested (Powel and Allaha
badi 1988; Yao et al. 1985): DI = (u  uy) 1 (uu  uy) . A similar index using the ductil
ity demand (11) versus the ductility capacity (l1 u )' has been used (Darwin and Nmai 1986;
In the above formulations the capacity of the section, or structure, is assumed con
stant during the seismic event. Other indices, relate implicitly deformation and forces
using energy dissipated through hysteretic action. Mahin and Bertero (1981) suggested
design methods based on the determination of the ductility demand, instead of only hys
18
Concern for the influence of the hysteretic energy dissipation, in addition to the
deformation changes, produced two component damage models, one which is related to
deformations and other to energy (Banon et al. 1981 and 1982: Park and Ang 1985, and
Bracci et al 1989). The difference in the above models stem from the method used to com
bine the two influences. It should be noted that, while the two components of the damage
models are assumed independent (Park et al. 1985), or statistically independent (Banon et
al. 1981; Bracci et al. 1989), the influence of hysteretic dissipation on the capacity terms is
neglected.
has been formulated using "cumulative damage theory" (Chung et al. 1987; Powel and
Allahabadi 1988; Cosenza and Manfredi 1992). However, the cumulative formulation
requires calibrated indices and does not explicitly correlate the hysteretic energy dissipa
of deformations were made by Wand and Shah (1987) and by Chung et aI. (1987). Chung
et al. introduced the notion of a failure envelope which links both deformation and
strength capacity limits, based on a fatigue model. However, the deterioration rules in the
without a relation to the work done and to the energy dissipated by the structural members.
fatigue rules produced additional damage indices (McCabe and Hall 1989; Chang and
19
Mander 1994a). However, the damage indices proposed are limited to strength damage,
From experimental data and from shaking table studies (Bracci et al. 1995) it can
be observed that there is a strong correlation between the deterioration of strength and
deformation capacity of structural elements. It is also observed that the damage in most
the elastic limits, but collapse occurs only when those deformations reach their capacity.
energy, can deteriorate the capacity to the level of the applied loading, leading to total col
tic modeling of the nonlinear behavior of its concrete components. Both macro and micro
modeling approaches can be used to model concrete behavior through the nonlinear range
up to failure. Because closed form solutions for complex structures are seldom possible, a
finite element approach in which the structure is divided into discrete (finite) elements for
the purpose of analysis is used. Each element can then be modeled and the total model of
the structure generated from the assemblage of individual elements. In some cases the ele
ments consist of discrete members of the structure but more often individual members or
20
The plastic behavior of a concrete structural element can be assumed to be concen
trated at discrete locations or node points at the elements ends or spread along it's length.
A survey on the different methods of modeling and the spread of plasticity is given below:
Early work on the inelastic analysis of RIC structures was based on elastoplastic or
nondegrading bilinear systems (Velestos and Newmark 1960). The energy dissipation
capacity of such systems, however, did not produce results which were representative of
elements up to the point of failure with cyclic loading. It was noticed that the capacity for
dissipating energy by hysteresis for the bilinear models was lower than that found in
A more representative model that incorporated stiffness degradation into the elastoplastic
model was developed by Clough and Johnston(1966). Another complex hysteretic model
based on experimental data was developed by Takeda et al.( 1970). The model is com
prised of a trilinear envelope curve designed to dissipate energy once the cracking point of
the concrete was exceeded. Thus even at low cycles of loading, energy dissipation was
incorporated. Celebi and Penzien(1973) showed that sections experiencing severe shear
deformations have marked pinching in their hysteretic loops. Thus energy dissipating
capacity of sections is also influenced by the shear deformations through the pinched hys
teresis loops. The Takeda model cannot capture this behavior. Several researchers thus
introduced modifications to this model to account for pinching of slip effect from shear or
bond slip (Takayanagi and Schnobrich 1977, Emori and Schnobrich 1978, and Saiidi and
Sozen 1979). Takayanagi and Schnobrich(1977) incorporated the effect of axial force
21
along with pinching and strength decay into the Takeda model. Strength deterioration, also
reversals was incorporated in other models by Nakata(1978), Aoyama (1968), and Park et
Wall elements are defined in some finite element formulations as panels with four
nodes defining the corners of the panel. They have large stiffness inplane, and small stiff
ness out of plane. In bridge analysis these types of elements are sometimes used to model
One method of modeling walls for in plane loading, using a frame analogy was
developed by MacLeod (1973, 1976,1977). He suggested that the wall properties be spec
ified along the wall vertical centerline in the plane of the wall with nodes at the top and
bottom of this centerline. Rigid body transformations of the degrees of freedom are made
from the wall centerline, to the four comers of the wall. The rotational degrees of freedom
at the corners were eliminated by introducing additional shears thus giving the same
moment as existed at the wall centerline. The connections to the walls were in the form of
stiff coupling beams consisting of only shear and axial forces. This procedure however
resulted in parasitic moments on the overall wall element (Kwan, 1993). He noted that the
end rotation of shear walls does not reflect the influence of shear deformations on walls.
He therefore proposed the use of beam elements with vertical rigid arms (instead of the
conventional beam elements with horizontal rigid arms in the coupling beams). This
method eliminates the errors in beam end rotation and ensures compatibility between ver
tical rotations at the extreme end of the wall fiber and the rotations in the horizontal cou
22
pIing beams. However this procedure cannot model walls connected to outofplane
members. Other procedures for modeling of walls, involved specifying the degrees of
freedom along the wall centerline with transformations of the stiffness from nodes at the
corners to the wall centerline. This procedure has difficulties in modeling interconnected
panels.
major step in the design and retrofit of the structure. The evaluation procedure is aimed at
determining and comparing the estimated seismic demands placed on the structure to the
estimated capacity of the structure. The performance and the level of safety of the struc
ture is derived from this comparison. The capacity of a structure is defined as the level of
forces and deformations that the structure can sustain without reaching a specified limit
state. The limit state is either the level at which the structure reaches collapse or a level
defined by design standards as the maximum deformations which the structure can sustain
and still provide either an acceptable level of life safety or to remain serviceable.
The demand from a structure can be defined as the forces and deformations that the
detailed evaluation of each method will be given below. All the methods presented are
smaller parts of the bridge, to calculated capacities of the entire structure or its parts.
23
TABLE 1.1. Evaluation Methods for Bridges
(EQ 1.8)
Where u t is the total deformations (u g + u), c (t) is the time dependent damping
and k (t) is the time dependent stiffness. EQ. 1.8 can be integrated in a step by step man
ner to derive the deformations ut • The stiffness of concrete element in general is modeled
using nonlinear rules. The protective systems are modeled using nonlinear rules also such
that the stiffness k (t) and the damping c (t) . are changing in time.
The nonlinear deformations of the concrete elements can be compared with the
capacity using the damage index indicator DI (IDARC 2D Valles et aI. 1996).
24
This procedure captures accurately the dynamic behavior of the all structural sys
tern. It is very comprehensive, and gives very good insight into the dynamic behavior of
The non linear behavior of the concrete elements is accurately modeled, and the
influence of the changes in elements properties on the response of the global system is
direct.
Because of the direct interaction in time, between the global and local systems, the
influence of the protective systems such as base isolation, and damping devices on the glo
bal bridge system is direct and no approximations of the influence are required.
The influence of the difference between the ground motion records, at different
•
The damage caused to the concrete components is evaluated instantaneously dur
ing the analysis duration, and since the damage index is related to reduced capacity due to
energy dissipated by inelastic deformations, an accurate indication about the actual reli
be related directly to the reliability of the whole bridge system, since the deformations and
forces in the elements are dependent on the change in the nonlinear properties of all the
elements in the system, and to find the actual capacity of the bridge, a series of analysis
should be performed with increasing ground motion intensity, until failure condition in the
bridge is reached. Relating the design level intensity of the ground motion to its ultimate
25
capacity will provide the reliability of the system. Or through direct evaluation of the
return period of the ultimate intensity of the ground motion the reliability of the bridge can
be evaluated.
It is obvious that this kind of procedure in not feasible today, because of the large
The response of the global bridge system, is sensitive to the modeling of the non
linear behavior of the components. The analysis can be performed only on limited number
of ground motions, which might not be enough to represent accurately the future earth
quake.
This procedure is similar in nature to the procedure in the previous section. The
major difference is that the comparison between capacity and demand in the nonlinear ele
ment is not done instantaneously, and the extreme value of plastic displacement demand is
compared with capacity which is not reduced due to the energy dissipated in previous
reversals and therefore the calculated damage level is lower then the actual one, especially
when subjected to vibratory type earthquakes in which the element experiences large
number of plastic cycles. Most of the software packages used for nonlinear analysis of
reinforced concrete (DRAIN2D Kannan and Powell etc.)can be classified under this defi
nition.
26
  _ .. _ . . .
movement and restrains at important locations. A stiffness matrix and a lumped mass
matrix is used to characterize the bridge properties. The equation of motion under earth
(EQ 1.11)
where
(EQ 1.12)
EQ. 1.11 is a differential equation with one unknown q(t). This equation can be
integrated in time, to find q(t) for each mode. Using the solution of q(t) the maximum
the maximum deformations are increased, and the maximum forced are decreased by
reduction factors related to the rotational ductility of the elements. It should be noticed
27
that the influence of the inelastic behavior is indirect. Thus there is no direct influence on
the analysis procedure (reduced properties such as elements stiffness are considered to
behavior using equivalent damping and stiffness. It should be noted however that those
properties are not placed in the element's specific location, and are "smeared" allover the
sate for this shortcoming, an approximate procedure which includes interpolation between
analysis with open and with closed gap was suggested (Caltrans 1992).
The effect of differential ground motions, cannot be considered using this type of
analysis.
The evaluation of the reliability of a bridge using this method, is subjected to same
limitations, as the previous method. Furthermore the reliability of each element (section),
cannot be derived exactly, and it is unclear, what portion of the ductility is actually used.
As mentioned earlier the common limitation for all the procedures using time his
tory analysis, is the uncertainty in the ability of few ground motions to accurately repre
28
1.9.4 Modal  Response Spectrum Analysis.
tioned, linear time history analysis. The element's maximum forces fmax,i, deformations
umax,i' and total force BSmax,,on the bridge are derived as:
where Sd and Sa are the displacements and acceleration response spectra usually linked by
the relation, Sa = ro2Sd' It should be noted that these spectra are obtained by averaging
multiple spectra from a large number of earthquakes. Therefore this analysis is representa
As compare to the previous procedure, only the maximas of each mode are
obtained, and using conventional SRSS procedure the required values of force and defor
mation are derived. The use of this procedure gives a limited insight into the dynamic
The calculation of the reliability of the all system, has the same limitations as the
previous procedure.
The limitations presented for both the linear time history analysis, and modal spec
tral analysis, in considering the nonlinear behavior, and deriving the reliability of a bridge
29
The method is based on a well known method of plastic analysis, or capacity anal
A short description of the procedure is presented (Priestley et al. 1991): The ele
ments of a concrete frame are assumed to have limited capacity to resist moment. A lateral
load is applied to the frame, until the moment capacity of a critical section is reached. The
section is replaced by a hinge. This procedure continuous until total mechanism is formed.
The maximum deformations are compared to the deformations at first yield. The ratio
between the maximum deformations and deformations at first yield is defined as displace
ment ductility, from which equivalent elastic force capacity can be derived using standard
procedures. This capacity can be compared to elastic force demand derived from response
spectrum curve, using the elastic frequency (or frequencies) of the structure to derive its
reliability. The reliability can be derived directly, by finding the level of earthquake that
this structure can resist, and the probability of occurrence of the earthquake is the reliabil
This method was extended (IDARC2D v:4.0), instead of using failure analysis as
concrete elements modeled using trilinear hysteretic loop and spread plasticity. Failure cri
This procedure emphasis the nonlinear behavior of the concrete elements. The
dynamic response is represented using linear response spectrum, which doesn't consider,
the influence of the nonlinear behavior on the demand (shift in period). The influence of
protective systems, can be accounted for, only through response spectrum curves related
30
to different levels of damping. Expansion joints cannot be modeled accurately. The bridge
cannot be analyzed as one unit, and has to be broken into separate frames with assumed
tributary load to each bent frame. The interaction between the frames through the stiffness
ground motion.
The big advantage of this method is the fact that the reliability of a subsystem can
be derived directly, as compare to the previously presented methods, where only reliability
Table 1.2 below. The procedures numbered from I through 5 are presented in sections
differential
procedure dynamic protective plastic reliabllity
ground
# response systems behavior evaluation
motion
31
2.0 ANALYTICAL TOOLS FQR EVALUATION OF BRIDGES 
STATEOFTHEART
2.1 Introduction
This section describes the stateoftheart of the analytical techniques and compu
tational tools for the seismic analysis and evaluation of bridge structures. First, the various
methods of analysis are outlined pointing out the extent of the applicability for each of
these. This is followed by a description of research needs that highlight the purpose of this
dissertation.
resistance of bridge structures. The following lists most of the methods in approximate
order of complexity:
I) 2D Elastic Analysis. This method is sometimes still used and has its roots in the
early computer programs that were based on frame analysis. The bridge is analyzed in
each of two orthogonal directions. With this approach it is possible to capture various
modes of vibration. The elastic demands on members can also be identified. This approach
2) 2D Plastic Analysis. This approach, although not commonly used in practice
(particularly in North America, but is perhaps more popular in other parts of the world) is
useful in identifying force (strength) capacity, coupled with inelastic spectra displacement
32
3) 3D Elastic Analysis. This method is considered by many practitioners to be the
stateoftheart. Computer programs that are commonly used for such analysis of bridges
(for example, SAP90, GTSTRUDL and SEISAB) are capable of accurately identifying
cial software vendors of these programs are currently exploring ways to extend the pro
4) 2D Inelastic Analysis. Computer programs that can perform inelastic time his
tory dynamic analysis of bridge structures have been available for over 20 years (Kanaan
and Powell, 1973). Inelastic 2D analysis programs have generally been developed by the
research community and are sometimes used by practitioners who wish to do advanced
analysis. The use of such programs, however, has been somewhat limited as they have to
compete with 3D linear analysis methods. In the early 1990s, many of the original nonlin
ear time history dynamic computer programs were extended to enable nonlinear static
(pushover) analysis to be performed. Following the Lorna Prieta earthquake of 1989, there
was heightened awareness within the bridge engineering community of the need for more
advanced analysis tools and bridge structures. As a first step, the inelastic time history pro
grams were enhanced to perform a static pushover analysis to assess the strength and
recommended in the recently published FHWA Seismic Retrofitting Manual (Buckle and
which on the one hand evaluates demand (typically using 3D linear programs by modal
analysis), and on the other hand assesses capacity using the equivalent lateral strength
33
pushover methods. For the latter, there are still few commercially available computer pro
grams that are capable of undertaking such nonlinear analysis, therefore the Retrofitting
6) 3D Inelastic Analysis. The finite element method of analysis has evolved over
the last 35 years and the present stateoftheart permits the use of inelastic time history
analysis of complex systems. There are several advanced computer codes that are capable
of undertaking such a task such as ANSYS, ADINA and ABAQUS. Most civil engineer
ing offices feel uncomfortable using these programs as the analysis requires considerable
skill and experience in utilizing some of the advanced attributes and capabilities of the
software. Although finite element analysis has its roots in civil engineering, it seems that
the mechanical engineering profession has more readily adopted the method and uses
advanced (but mostly linear) analysis techniques for the routine design of manufactured
products such as cars, engines, airplanes, etc. To use some of these stateoftheart finite
element programs for bridge analysis requires special preparation of the input. The major
impediment in using these programs for earthquake engineering is that the nonlinear fea
tures are somewhat simplistic, particularly for analyzing structural concrete elements as
Takeda model resident in the original version of DRAIN2D (Kanaan and Powell 1973)
are quite powerful for analyzing reinforced concrete structures. Such a macro element is
not available in commercial finite element programs. In spite of this, advanced models can
be constructed with a fine mesh of finite elements in commercial software that perhaps
could reasonably emulate reinforced concrete behavior under cyclic loading. However, the
overhead necessary for such a micro mechanics based computational modeling strategy is
34
  
excessive. Moreover, most civil engineering design offices do not have suitable computer
recent paper by Wilson (1994) who outlines the use of substructuring techniques on PC
based computers using the program SADSAP to handle a limited amount of nonlinear
analysis. This program, however, is limited in its nonlinear capabilities and cannot handle
well the highly nonlinear attributes and damage associated with degrading reinforced con
crete elements.
for inelastic 3D analysis of structural systems. Few of these programs have been devel
oped for bridges, most have been developed for buildings. The program DRAIN3DX
(Prakash et ai., 1992) uses a fiber element formulation to model the nonlinear behavior of
the structural concrete elements. Because of the very complex nature of the modeling, it is
not realistic to use such a program for production design office use as the problems them
selves tend to be ill conditioned, particularly when a large number of nonlinear elements
are present.
fiber models is instead to use macro models that have bidirectional modeling capabilities.
Such an approach generally leads to improved problem tractability, but the problem
such a program is IDARC 3D (Lobo, 1994). In this program macro models are calibrated
through a combination of moment curvature analysis parameters that are identified from
35
experimental studies. Alternatively, an offline fiber element analysis of a single column
member can be analyzed using an experimentaltype of load history. This is then used to
more accurately calibrate the macro model. Such an approach has been advocated by
Perhaps the only computational platform that has been already dedicated to nonlin
ear 3D analysis of bridge structures is the software NEABS II (Imbsen and Penzien,
1986). Although this program has many desirable attributes for nonlinear 3D analysis, its
utility is limited by the simplistic nature of the hysteretic models that it uses. For example,
lize such a simplistic approach for highly irregular bidirectional loading. The program
also suffers from lack of flexibility that is given to the analyst as the program was devel
oped to model conditions being characteristic of Californiatype bridges. For bridges that
do not fall into this pattern of construction, the program is difficult to use. The program
also has some limited capability for handing differential ground motion input.
important issue. To date there are no commercially available software packages that com
bine nonlinear time history analysis with damage analysis. It is only the university origi
nated research type software that undertake nonlinear damage analysis. Programs that can
and IDARC2D.
36
2.4 Research Needs
Many features that are desirable for the nonlinear dynamic analysis of bridge sys
'terns already exist in various forms and have been implemented in other areas. There is a
clear need to assemble these features together on to one software platform. For example,
much work has been done over the last few years on the 2 and 3D behavior of seismic
isolation systems with a particular emphasis on building structures. Certain programs that
have resulted from this work (e.g., 3D BASIS Nagarajaiah et al. 1989) have capabilities
that are also desirable in bridge analysis. As mentioned above, the seismic capacity and
demand analysis procedures are presently fragmented and these should be amalgamated
along with damage analysis concepts in such a way that it should be possible to analyze a
bridge that incorporates the primary source of energy dissipation in locations such as
joints and gap hinges in the deck, 3D behavior in column elements and plastification of
these elements in the substructure (the piers), and the nonlinear behavior of the foundation
system including the nonlinear and radiation damping effects of soil structure interaction.
byframe basis and then combined in some way to give the overall effect for the bridge
structure. The problem with this approach is that the combination may not conform to the
dynamic response that could be expected in an extreme earthquake. It is, therefore, desir
able that a 3D computer program be capable of undertaking a pushover analysis of the
bridge both statically and dynamically. In this fashion, some of the uncertainties that per
37
What distinguishes bridges perhaps more than anything else in comparison to
other structural systems is the spatial layout of the structure that gives a high propensity
38
3.0 DEVELOPMENT OF NEW ANALYTICAL PLATFORM FOR
SEISMIC RELIABILITY EVALUATION
The reliability of a bridge for seismic loading is of major concern to ensure proper
operation of road and railway systems following an earthquake. The reliability R or the
D
R = 1 (EQ 3.1)
C
in which D  the demand is the magnitude of forces or deformations that the bridge will
experience, during an earthquake with certain probability of occurrence during the life
time of the bridge; and the capacity C is the magnitude of forces and deformations in
which the bridge loses serviceability or stability. In this chapter several approaches to
evaluate the seismic performance using the reliability definition for bridge structure are
presented. In the first part of this chapter, the important characteristics of bridges as struc
tural system, and each of its major components is presented, as well as the structural
The bridge is modeled as a framed structure, so that matrix analysis using the stiff
ness method can be performed. The structural modeling of each of the structural elements
used in modeling the bridge components, and the way this elements are used to model
each component of the bridge, and subassemblies of the bridge is presented as well as how
mathematicalphysical tool such as rigid body transformations are used to achieve effi
39
In the second part of this chapter different procedures of evaluation capacity and
demand are presented, and the procedures used to evaluate the final reliability of individ
Bridge is a general word to describe a structure that enables traffic to pass over an
obstacle. Bridges types, vary from very small and simple once, that bridge a small creek,
to very large once, that span over large bodies of water. Altogether all of them.have a com
mon typical characteristics and components that differ in size and complexity. A typical
set of a bridge assembly and it's major components is shown in Fig. 3.2
deck abutment
backfill
40
3.1.1 Components and System Definition
A bridge has several major components, as shown in Fig. 3.1. The components
The deck is the surface on which the traffic is moving. The deck is supported by
large bents which provide for both gravity and lateral movement. Several types of struc
tural arrangements of the deck are used: prestressed box girders, monolitically cast beams
and slabs, and concrete or steel slabs supported on steel beams etc.
The bents are the part of the bridge that elevates the road above ground level.
Mostly two types of bents are used: (i) A multiple piers bent with a top beam connecting
them together at their top. (ii) Single walls, which support the deck directly.
The foundations are the system which interface the bridge structure to the support
ing ground soil. The main purpose of the foundation system is to transfer the dead and live
load of the bridge to the ground. This system also transfers the horizontal loads and the
overturning moments resulting from horizontal loads, such as winds and earthquakes. Var
ious types of foundations are used, which are: shallow foundation systems, and deep foun
dations such as caissons and piles. Each provides larger or smaller soilstructure
The abutments are the supports at the two ends of the bridge. They support the ver
tical and lateral forces from the bridge deck as well as the soil behind them. The abutment
is made usually of a central wall and two wing walls, which support horizontally the soil
behind them. The deck is supported vertically on the central wall trough joint (see descrip
tion).
41
In long bridges, the expansion and contraction of the deck above the bents due to
temperature changes might cause unwanted stresses. Therefore expansion joints are intro
The deck sections are supported by bents vertically. To provide horizontal motion
capacity between the deck and the bent and to prevent undesirable forces due to thermal
expansion, bearings are placed between the deck and the bent. Several kinds of bearings
are used for this purpose: elastomeric bearings, teflon steel sliding bearings, steel bearings
etc. Those bearings should also provide rotational capacity, since usually a hinge type of
support is desirable.
In moderate and severe seismic zones, the sections of the bridge are suppose to
move together in the transverse direction. This is achieved by providing shear keys, which
Another important issue relevant to severe and moderate seismic zones, is the
are used to prevent such instability in modem bridge. The restrainers are steel roads or
cables, anchored in the separate deck sections, which limit the relative maximum displace
ments of the deck sections in the longitudinal direction. More modern solutions use non
To reduce the forces in the structure resulting from earthquake, modern protective
systems such as: base isolation systems, damping devices, active control, etc. are also
used. Those devices are placed usually in the expansion joints, between deck sections or
42
3.1.2 Comparison of Buildings and Bridges I Specific Issues
Bridge structures span over large distances in space, and therefore their founda
tions are spread apart in distinct locations, creating several different interface points
between the soil medium and the structure. The ground motion, the supports stiffness and
damping, might vary greatly from one interface point to another, influencing significantly
the structural dynamic response. In most of the other structures i.e building structures, the
interface occurs usually in small area which can be characterized as one point in space.
Although large buildings are supported on a group of individual foundations, usually those
foundations are connected by tie beams, which form a single type of interface, with single
expansion joints. Expansion joints are provided to prevent appearance of thermal stresses
due to bridge elongation. In affect these joints influence the response of the bridge to lat
eral loads by separating the bridge into several substructures loosely connected to one
another.
etc. is given to the basic structural elements used to compose this subassemblies and then
43
a description, how these structural elements are actually used to represent the structural
In three dimensional framed structures analysis, the basic element is the space
frame member (Weaver and Gear 1990), or "stick" element. This element has two general
joints in space, with each joint having translational and rotational degrees of freedom in
all three global directions X, Y, Z all together six Degrees Of Freedom (D.O.F) in each
joint. And for the whole element 12 D.O.F's. The element has it's own coordinate system,
and can be rotated in any direction with regard to the global coordinate system (Weaver
and Gear 1990). The constitutive relations of this element can be described by
(EQ3.2)
where F, M, u and 9 are respectively the three components vectors of forces moments,
(EQ 3.3)
44
where Fc, M c' Ii and e are respectively the three components vectors of damping forces,
damping moments, translational velocities and rotational velocities. Moreover c repre
The element formulation described by Eq. 3.2 and Eq. 3.3 is the so called Kelvin
model formulation.
For types of analysis which include elements that have nonlinear properties (non
linear time history analysis, incremental nonlinear static analysis, pushover analysis) the
stiffness and damping matrices k and c depend on the nonlinear rules specific to elements
used.
The "stick" elements can be provided with special end conditions, which include
Moreover in the dynamic analyses using macromodels the mass is lumped in con
necting joints such that the individual structural models are massless.
interface between the deck surface and the bents. Accurate modeling would require many
elements and joints, to model essentially rigid zones. Using rigid body transformations,
element end springs attached directly to elements, and releases of D.O.F's can provide a
45
~I
deck beams
bearing
r >:1..1 __
theoretica1 joint
(a)
joints
foranal~
'.~~
(b)
spring
isolator
rigid
arm
(c)
FIGURE 3.2. Connection Between Deck and Bent Using Rigid Arm and End Springs (a)
Physical Description (b) Regular Modeling (c) Modeling with Rigid Arm and End Springs
46
A possible connection between the deck and the bent is presented in Fig. 3.2 (a).
To model correctly this connection many elements are required (Fig. 3.2 (b». The deck
is modeled as beam and has only one connection point. The bent however has three con
nection points at the locations of the bearings. By assuming that the deck is rigid for rota
tion around Y axis, and translation in Z direction the corresponding rigid transformation
can be applied to represent the rigid zones. The flexibility of the deck beams for transla
tion in the Z direction, can be modeled by end springs. The use of end springs and rigid
arms enables accurate representation of this area with the use of only three elements and
zones the rigid arm transformation is developed. The elements stiffness matrix is created
with respect to the actual joints (Fig. 3.2 (a» and then transformed, using rigid body
transformation, to the theoretical joints (Fig. 3.2 (c» which are used to assemble the total
structural stiffness matrix. Following the solution of the displacements in the global sys
tem, the displacements in the actual joints and the element's end forces can be found using
The transformation of an action Aj at the actual jointj to the theoretical joint (used
47
y
r j
r
j
x
FIGURE 3.3. Rigid Body Transformation from Joint P to Joint J
(EQ 3.4)
(EQ 3.5)
48
rpjxFj = [I
Zpj
F~ FY F,
J
j k
Ypj Zpj
J
l
J
=cpjFj =
0
Zpj
Ypj x.
PJ
Zpj Ypj
0 x
0
pj
P
J
FY
J
HJ
(EQ 3.6)
The relations presented in EQ. 3.6 are relating a transformation of a single joint of
the "stick" elements. The relations for the whole element (two joints) are:
(EQ 3.8)
D ={ up }; D. ={ uj } (EQ 3.9)
p 0 J 0.
P J
D. = (EQ 3.10)
J
where
i j k 0 a~
ajxrpj = ax
p
ay
p
az
p =Cpjaj = Zpj ',jo Y'j
Xpj
J
a~
J
(EQ 3.11)
49
(EQ 3.12)
The force displacement relation of the element at the actual joints (physical) is:
(EQ 3.13)
(EQ 3.14)
(EQ 3.15)
(EQ3.17)
between elements. For the sake of modularity, a fixed end element stiffness matrix is gen
erated first and the end releases are applied later to generate the released matrix. The num
ber of releases is limited so that no mechanism is created in any direction (such as in the
case when one side fully released and the other side is hinged).
50
ux
uy / possible released
beam end releases uz D.O.Fs
(rotation or translation) rx
~~~~~ x
element start element end
kd = A (EQ 3.18)
The degrees of freedom can be separated into released and non released degrees of
freedom, with the force vector on the released degrees of freedom being zero.
(EQ 3.19)
The stiffness matrix of the element can be related to the nonreleased degrees of
freedom only, when the stiffness of the released degrees offreedom is zero. Thus,
k *  k nn _kT k1k
nr rr nr (EQ3.20)
51
Which also satisfies the equation: A = k: d .
The use ofEq. 3.21 in the assembly of the global stiffness matrix can be done with
relation only to the nonzero terms in the matrix. It should be reminded, as it is well known
that while building a structural model, all the global D.O.F's, should have nonzero stiff
ness terms on the main diagonal, otherwise the stiffness matrix, might be ill conditioned.
The above formulation significantly simplifies the manipulation of the stick ele
ment stiffness matrix since it allows the combination of releases, rigid zones, and end
springs together, in any combination and order. Another important benefit is the fact that
this formulation can be used for any type of element, including inelastic beam elements,
A common way to formulate the stiffness matrix of a beam with end springs is by
using the flexibility method (Weaver and Gear 1990). This method requires different for
mulation for a beam without springs, and for every combination of springs at it's start and
end. To avoid this and achieve a more modular formulation, the stiffness matrix of the
fixed end beam is derived first and the contribution of the end springs is added later.
As shown in Fig. 3.5 four joints are defined for the basic "stick" element and the
two end spring elements. The constitutive relations of the combined system are:
52
kx
ky
kz
possible spring stiffness krx
kry
~~_2_______________3~
element start
/
end spring element end
x
Fi,j are the forces and moments at joint i and j and uiJ are the displacements at
joint i and j. The matrix k is assembled in a standard manner by arranging the element
stiffness matrix K beam , and the stiffness matrices of the end springs K s1 ' Ks2 into the
Since joints 2 and 3 are internal joints, there is no external force acting on them,
53
Introducing Eq. 3.24 into the second part of Eq. 3.23 will yield the final constitu
(EQ 3.25)
kfinal  k 1,4 kT k 1 k
2,31,4 2.32,3 2,31,4 (EQ 3.26)
In the case when only few of the D.O.F's have springs attached, the following con
joints 14 in the previous development) and u i are displacements related to the internal
D.O.F's (similar to joints 23 in the previous development). Employing simple static con
densation, results in
Jma I = kee  kT
k" le k":I.k.
ll le (EQ 3.28)
This expression is general, without any limitations, and can be combined both
with rigid arms, and end releases. The stiffness of the end springs can be linear as well as
nonlinear. This operation requires finding the inverse of the matrix (k;J of the size of the
54
 
3.2.2 Inelastic Hysteretic Beam Columns with Nonlinear Shear and Flexure
Behavior in Beam/Column elements
F = ku (EQ 3.29)
Which is similar to the relations of the basic "stick" element. The stiffness matrix
of the basic "stick" element k as well as the stiffness matrix of the hysteretic beam I col
EA EA
L L
12EI 6EI 12EI 6EI
z _z z
;y L2 ;yz L2
12EI _6Ely 12Ely _6Ely
y
L3 L2 ;y L2
GJ GJ
L L
4Ely 6Ely 2EI
y
L L2 L
4EI 6El 2EI
z z z
k = L  L2 L (EQ 3.30)
EA
L
12EI 6EI
z z
;y  L2
12EI 6Ely
SYM y
L3 L2
GJ
L
4EI
y
L
4EI
z
L
For the hysteretic beam I column element the variables In Eq. 3.30 E(u) the Young
modulus, and J(u) the moment of inertia ofthe section are considered nonlinear functions
55
of curvature. For an inelastic beam, an equivalent EI(u) is used which is assumed to be a
3.2.2.1 Bending
flexibility fonnulation (the procedure is the same one used by Lobo 1994). A linear distri
bution of flexibility over the length ofthe beam(Fig. 3.7(b» is used. The size of the pen
etration of the damage into the element indicated by xl, x2 at Fig. 3.7(b) and designated
as cracked length in Fig. 3.7(a). The size of the cracked (damaged) length is derived
from linear moment distribution over the element length (Fig. 3.7(a». In the process of
the incremental analysis the damaged zones cannot recover after they were damaged pre
viously, so the maximum value reached during the analysis history is used to define the
zones at the ends of element (Fig. 3.7(c», for each local beam direction (xz, xy). The
elastic flexibility at the center of the elementfiexO is lIEI of the elastic (undamaged) sec
tion of the beam. The flexibility at the ends of the element defined as d<j>1 dM is obtained
from the trilinear hysteretic rule (Fig. 3.6) which includes, stiffness and strength degrada
tion and slip (Park et al. 1987). Currently the behavior of the element in bending in its two
56
M
MPultimate
MPyield
MPcrack
4>nultimate unloading
slop
<I> Pultimate
,
/
, ...M'ncrack
,,
/ /
Mnyield
' Mnultimate
,,
/
/ /
/
/,
,
N
Mnyield*U,S,
, /
(a)
first unloading
with slip
unloading ~, ~
with slip ' \ ,..M'Crack*S.F
Mtarg.,*S.F \.. ~ >.. . . . . . . yo \. •.•
(b)
FIGURE 3.6. Trilinear Hysteretic Model, a) General Model, b) Slip Definition
57
cracked length(zone_number(1))
,...:l+,
a)
cracked_length(zone_number(2))
x2 xl
b)
up
I 2
3 4
down
c)
58
3.2.2.2 Combined Bending and Shear
The general behavior of bridge piers under lateral loading is dominated by bending
and shear. The shear behavior is especially important in piers with aspect ratio, less then
To be able to predict the failure mechanism, a combined shear and flexure model is
developed. For the flexure model the moment curvature relation is described by the trilin
ear model, from which the element stiffness is calculated using distributed plasticity (see
section 3.2.2.1) in terms of end forces and displacements. A trilinear shear model is used
The final stiffness of the element is derived by combining the flexural and shear
stiffnesses into the element stiffness matrix. First the flexural stiffness matrix is derived
+klJ
__
k22
~ ~=LO
.
J J
The stiffness matrix of joint J of the beam in Fig. 3.8 which is related to flexure
(EQ 3.31)
59
The flexibility matrix due to flexure can be derived by inverting the stiffness
matrix
F = k I = 1 [k22 k (EQ3.32)
det (k) k k lJ
21 II
where
(EQ3.33)
The influence of shear can be added directly to the term F II such that
(EQ 3.34)
where F II is the flexibility term including the shear deformations, and ks is the shear stiff
ness defined as
k = fL (EQ 3.35)
s GA
where f is the shape factor, L the length of the beam, G the shear modulus, and A the
effective area. It should be noted that the parameters in this equation if, A, G) can depend
on the stresses and strains (J(u), A(u), G(u), and as a result the shear stiffness klu) is also
nonlinear.
l
The flexibility matrix for joint j including shear deformations has the form
F= 1 k22 + det(k)
k k12
(EQ 3.36)
det (k) s
[
k21 kll
Note that when the shear stiffness ks is infinite, (no shear deformations) the origi
60
Inverting the flexibility matrix will give the stiffness matrix which includes shear
deformations contribution.
~
 1 1 kll k12 ]
(EQ 3.37)
k = det (F) det (k) k k + det (k)
21 22 k
s
where
(EQ 3.38)
and
(EQ 3.39)
Using equilibrium the stiffness terms k13 ,. k23 ,. k14 ,. k24 can be derived such
that
        
k13 = kll ,. k14 = kllL  k21 ,. k32 = k 21 ,. k42 = k22 + k12L (EQ3.40)
In the same manner (using equilibrium), the stiffness matrix for the other side of
the beam can be derived. The element stiffness matrix is which includes bending and shear
is defined as:
kll k12 ]
k k +det(kll)
[ 21 22 k
s
(EQ 3.41)
k33 k34 ]
k k det (k22)
[ 43 44 + k
s
where
61
(EQ 3.42)
[: k + d~'(kll)]
22 k
II
[~31 ~3J
k41 k42
k =
[~31 ~3J
k41 k42 [: k" + d~'i~2)]
It should be noted that all the elements of k matrix can change in time in nonlinear
structure. If the element reaches yielding in flexure or in shear, the failure mechanism is
controlled by the lowest capacity of the two mechanisms. When the limit state of either of
As mentioned earlier this formulation was adopted since both the flexure and the
shear stiffness can be highly nonlinear. It is beneficial to derive separately the flextural
stiffness matrix and the shear stiffness for increased modularity in handling and allowing
In order to illustrate the shear or moment dominant failure mechanism, the simple
frame in Fig. 3.9 is analyzed. First the frame is assumed to have only flexural stiffness.
Then both flexural and shear are considered in two cases: (i) The flextural capacity is
assumed to be smaller than the shear capacity. (ii) The shear capacity is assumed to be
62
F
' " rigid
1= 10m
" .""
FIGURE 3.9. Example Frame, for llIustration of Combined Shear and Flexure Model
smaller than flexural capacity. The constitutive relations of shear and flexture for all the
The influence of shear on the capacity of the structure is demonstrated in Fig. 3.11
The first case corresponds to pure bending model. For this case (no shear at
Fig. 3.11 ) yielding in bending can be observed. In the case where shear deformations are
considered but the shear capacity is high, again flextural yielding occurs, with increased
deformations because of the contribution of shear deformations. When the shear capacity
is low, the element is not able to develop it's full flextural capacity and fails in shear. If the
shear model is assumed to be brittle (which is the case in many short RIC bridge piers)
63
2.00
1.50
~
z
~
C 1.00
~
E
0
:;
0.0 I:~'_~....J...~_'~:!
0.00 0.05 0.10 0.15 0:20
Curvature (11m]
Moment curvature
0.4 0.4
0.3  0.3
Z
~
~ 0.2 0.2
oS
iii
V
~
rn
0.1  0.1
0.0 0.0
0:0 0.5 1.0 1':5 0:0 0.5 1.0 1:5
Shear disp.[m] Shear disp [m]
FIGURE 3.10. Constitutive Relations in Shear and Bending of the Frame's Columns
64
0.8
0.7
0.6
0.5
Z
.!!.
iB 0.4
'"
0
~
0
III
0.3
0.2
0.1
Olsp[m)
3.2.3 Shear and Bending Wall Panel Model (for modeling Abutment)
The abutments of a bridge can be modeled by using a four nodded shear wall ele
ments. This element's nonlinear behavior has similar characteristics to the hysteretic beam
described in the previous section which includes flexture and shear. The wall is modeled
by using the a shear/flexure beam. The element has four nodes of which two of them are
coupled to the other two. The shear wall element, shown in Fig. 3.12 (a), can be physically
defined by a flexuralshear beam connected between nodes 2 and 4 Fig. 3.12 (b) with rigid
arm to the center of the wall. The actual properties are related to the theoretical beam
located at the center of the shear wall. The stiffness properties are transformed to the phys
65
ical beam by using rigid arms, as described in section 3.2.1.1 . Joints 1 and 2 are coupled
using the coupling procedure described in section 3.4 to ensure rigid rotation of line 12
such that
Lu2 I
U 1 (EQ3.43)
U I 3_U
 2
3 (EQ 3.44)
(EQ 3.45)
The current formulation of the shear wall is a straight forward extension of the
basic hysteretic beam element into wall element by using existing tools such as rigid arms
Base isolation is a strategy to reduce the seismic hazard for a bridge. An isolation
system combines two elements. The first one is, to provide a soft medium between the
bridge and the ground and thus isolate the bridge from the ground. The other element is to
provide energy dissipating capability between the ground and the bridge. The two ele
ments can be used together, in which case the energy dissipating capability is used to
used to dissipate energy only, and shifting the period of the bridge is not an objective.
Two models are introduced to model the behavior of such devices. The first one is
elastic  plastic constitutive relations with smooth transition. The other model, is based on
triaxial interaction to model the behavior of an isolation system which includes the infiu
66
" \V I
3U
shear Uflexure
++,1 rigid
u/t 3
" \v2
•
rhP~""1~ •
I
2
I I
I I
I I
theoreticaJ beam I
I
¥ I
theoretical beam
I
I
I
I
~:I
I
I beam for
I rigid arm analysis
y 3 4
 4
~
d rigid arm
(b)
x
(a)
FIGVRE 3.12. Shear Wall Definition (a) Shear Wall (h) Shear Wall Modeling for Analysis
ence of the vertical load on the lateral force, and the dependency of the lateral forces on
velocity.
67
Bouc Wen's equation requires solution of a first order differential equation that can
solved by enhanced Runga Kuta method (Nagarajaiah et al. 1989). In order to obtain a
more accurate and faster solution, and to better understand the model, an analytical solu
F ,
~ Kinitial
F .'L
~ ~ ~Y
Kyield __
~ .   ..
..
F
by
U
(a) (b)
F
~
F~
(c)
FIGURE 3.13. Elastomeric Isolator. a) side view, b) force displacements relations, c) top view of
deformed isolator
The basic equation for an isolator with elastic perfectly plastic behavior is (Nagar
68
. 0 . 0
Z = A
U
12111 (ysgn UZ + ~) 
U
y y (EQ 3.47)
force", and its value assumed to be limited 1 ::; Z::; 1 , U is the total displacement, A ~
and y are nondimensional parameters which control the unloading slope of the hysteretic
(EQ 3.48)
or
(EQ 3.49)
where F is the restoring force, 0: is the ratio of initial elastic stiffness to the post yielding
t = g
y
11
{AIZl ((ysgn(UZ) +~)} (EQ3.50)
dZ d(U/U) .

dt
= d y {A
t
12111 (ysgn (UZ) +~)} (EQ 3.51)
69
dZ U
dU = y (EQ3.52)
A IZl11 (y sgn (UZ) + ~)
This equation is time independent and can be solved directly. Integrating both
sides:
A close form solution is possible only for small number of.., values, .., = 1 or
A = a2 (EQ 3.54)
For y>~:
Introducing EQ. 3.54 . and EQ. 3.56 into the integrand of EQ. 3.53 will give
The solution of EQ. 3.57 can be derived for the following cases:
dZ
J
U = Uy a 2 b 2z2 (EQ 3.58)
70
and after integration
u = u _1_10gla+b~ +c (EQ3.59)
Y2ab abZI
This solution is divided into two cases:
1 bZ
U = U atanh+c (EQ3.60)
Yab a
and
Z = batanh (Uab
+c)
U (EQ 3.61)
y
U bZ
U = .2 acoth + c (EQ 3.62)
ab a
and
a
Z = coth (Uab
+c) (EQ 3.63)
b Uy
(B) sgnUZ<O
Uy bZ
U = atan+c
ab a
(EQ3.64)
Z = b~tan( Uab
U
+ c) (EQ3.65)
y
The case y > ~ and b2Z2>1 leads to Z larger then 1 which it invalid, so that the
71
The instantaneous stiffness of the isolator can be derived using the solutions for Z
(EQ. 3.61 . EQ. 3.63 and EQ. 3.65 ). The solution for Z can be used to define directly the
k dF = a+
= dU Fy
U
dZ
(la)F
YdU
(EQ3.66)
y
2
dZ) _ d (a Uab) _ a sech2Uab
( dU  dU btanhU  U U
1 Y Y Y (EQ 3.67)
2
dZ) = .!!...(~ctanh Uab) = _ a csech
( dU 2 dU b Uy Uy
(EQ3.68)
2
dZ) d (a Uab) a sec2 (Uab)
( dU 3 = dU btan U
y
= uy U
y (EQ3.69)
The values of the instantaneous stiffness k are used in the analysis to update the
The parameters required to define this model are: the yield force Fyo the yield dis
placements Uyo the ratio between Kinitial' and Kyield a (see Fig. 3.13(b)), and the parame
ters A ~, and 'Y, which control the shape of the unloading branch. In order to limit the
extreme value of Z = ± 1.0 the ratio alb in EQ. 3.61 , EQ. 3.63 and EQ. 3.65 must be
72
equal to 1.0. One of the ways to achieve it, is by fulfilling the conditions A= 1 and
"( + ~ = 1. When ~ = "( the unloading branch of the hysteresis loop is a straight line
with slope (stiffness) equal to the slope of the loading branch (initial stiffness kinitial=F/
u y ), when ~ is larger than ,,(, the unloading stiffness is higher than Kinitial and the unload
ing curve is convex (as shown in Fig. 3.13(b)). When "( is larger than ~, the initial
unloading stiffness is smaller then Kinitial and the unloading curve is concave.
Triaxial interaction model was developed to model the behavior of sliding isolators
subjected to triaxial loading. Modeling of sliding isolators with biaxial interaction was
treated previously using pseudo force technique in which equivalent forces are added into
the right hand side of the global equilibrium equation, while there are no stiffness or
damping tenns related to these D.O.F's (Nagarajaiah et al. 1989). Iterative procedures are
required to solve this problem. The pseudo force method is used because it allows to keep
the initial stiffness of the structure, and update only the right hand side of the differential
equation of motion, while using iterations to find equilibrium. It allows the use of linear
solution procedures such as mode superposition to solve nonlinear systems (Bathe et al.
1981). However, there are two significant deficiencies to this procedure: (i) In the presence
of nonlinearities in several locations in the structure, many or all the modes of vibration
will have to be considered to achieve accurate modeling. (Bathe et al. 1981) (ii) When
direct integration method is used, in the presence of large number of nonlinear elements,
the search for eqUilibrium is multidirectional. While convergence in one direction might
73
F
~
I
I
p
t ~ ~rf
P I'd'mgmte ace
51
I
I
I
I
(a) (b)
Jl
smooth function
Jl max
bilinear
~ ____L ______________~O
Vlimit
(c)
FIGURE 3.14. Triaxial Sliding Isolator, a) side view, b) sliding force· displacements relation. c)
dependancy of coeficlent of friction on velocity
be reached, divergence might occur in other directions, which result in infinite oscillation
by using first order differential equation to represent the hysteretic behavior of the isola
74
tion systems (Tsopelas et al. 1994). This method is very accurate and gives very good
results for relatively small structural systems, and when no other types of nonlinear ele
ments are present. In the presence of elements with other types of nonlinear rules, the sec
ond order differential equation of motion can be solved using a numerical method
(Newmark 1959). For this kind of procedure a better way is to represent the behavior of
nonlinear elements by their instantaneous stiffness and damping. This type of modeling
provides standard and unified procedure for the different types of nonlinear elements.
The sliding element model developed here has the capability to represent triaxial
behavior of a sliding isolator which includes displacements in the global X and Z direction
and the influence of the variation of vertical force on the sliding force. The friction sliding
isolators have two stages, slip and stick. In the stick stage the element has elastic stiffness,
Kinitia/' After the transition to sliding (slip) stage, at the first sliding (slip) stage, the coeffi
cient of friction is function of velocity (Fig. 3.14 (c). When the velocity exceeds the veloc
ity limit, the coefficient of friction is constant and equal to I1max' The data required for
this model is: P  the initial normal static force on the sliding surface without the contribu
tion of dynamic forces, Kinitiar the sticking stiffness (no sliding), Ksecondary the stiffness of
the recentering spring during the slip mode, VUmit  velocity limit above which the friction
coefficient becomes constant and equal to I1 max ' and, I1 min' I1max ' I1static  the minimum,
a combination of three components (i) a linear rise; (ii) a viscous component; and (iii) a
75
~ ~ ~
F(t) = F(tIlt) +AF(t) (EQ 3.70)
where
(EQ 3.71)
where A indicates the increment from time t  Ilt to time t, N is the variable normal force
)
to the sliding surface, and i is the sliding direction vector
(EQ 3.72)
(EQ 3.73)
where kois the initial stiffness and !1b_ away N (t) is the breakaway friction coefficient.
!1min N (t) ~ IF (t) I ~ !1maxN (t) and Iv (t) I~ Vlirnit (EQ 3.74)
(EQ3.75)
(EQ3.76)
Note that !1max and !1min are the maximum and the minimum friction coefficient
obtained from the inversion of the exponent a in the model by Mokha et al. (1990). This
76
coefficient (in velocity units) can be taken as 24 in.ls. (Tsopelas et al. 1994). It should be
noted that N(t) is time variable, therefore, Ceq is also time variable.
(c) For j = 3 the maximum sliding force achieved ClFtl = llmaxN (t) ) and
(EQ 3.78)
It should be noted that coefficient Sj is an indicator for biaxial effects in the sliding
mode. When the isolator's force, IF(t)I, drops below llminN (t) the system is transferred to
Using the above formulation, the model can be integrated into regular numerical
formulations without loss of generality. The vectorial formulation indicated above for the
directional movement in the plane of the isolator can be conveniently transformed back to
subjected to biaxial motion, with high and low velocity (Mokha et aI. 1993), were used to
The input displacements are defined by the function (see Fig. 3.15):
77
ux = umaxsincot
x
uy = uymax sin2rot
The frequency co for the case with high velocity was taken as 2.22 rad/sec, and for
The analytical modeling was able to predict well (Fig. 3.16) the experimental
results (Fig. 3.17). It should be noted though that the input motion used for the experi
ment did not follow exactly the input motion as described by the functions above because
of the influence of the testing equipment. This caused some discrepancy between the ana
Iytical and experimental results in the behavior in the initial stage (breakaway stage) in
which the actual velocity from the experiment is lower than the velocity obtained theoreti
cally. The general shape of the experimental results is also distorted, because of the influ
78
50
30
E 10
E
........
>
::> 10
30
30 10 10 30 50
Ux(mm)
FIGURE 3.1S. Graph of Theoretical Input Motion of 8 Shaped Test
79
16
12
/'
...
8
4
/.
/
.. " I"".. rngl 'l. .,..,
Ie wvelo ity
./
" '\
" ~'\

Z
...It:
X
0
y
"
"" "\
A
LL. \. " ./ /
4
8 \. I .......... /
" /' "' ./
12
16
50 30 10 10 30 50
Ux [mm]
16
hi~h elocit
12
8
L ":..,....
....
,
........
...
Z
4
'/
wn
,

...It:
>.
LL.
0
4
;.... .....
8 , I'.. ,/
J
1/
12
16
50 30 10 10 30 50
Uy [mm]
FIGURE 3.16. Analytical Loops of Frictional Force and Displacements in 8 Shaped Motion Test
80
Ux (m)

·4
.05 0.0,25 0.05·
Test No.
15
3
2
• 10
,..... 1 5 ,.....
VI
Co Z
3 ~
:i 0
........ 0 ........
x 1 x
lL. 5 lL.
2
10
3
. 15
2 1
Ux (inch)
Uy (m)
(,05 0~25 0.0,..?5 0.0 5
4
Test No. 3~
l 15
3· 6\
,
1
I 10
2· V'
,..... 1 ~
VI · 5
Co
:i 0
........
• o
>1 ·
t
lL. 5
2~.
10
3\.
w 15
4
2 1 0 i
Uy (inch)
FIGURE 3.17. Recorded Loops or Frictional Force and Displacements In S·Sbaped Mollon Tests
81
3.2.5 Three Dimensional Spring. Connection Element Model
An important basic element is the 3D spring, which can have linear or nonlinear
properties in every direction. This element can be viewed as shown in Fig. 3.18 as six
joint 2
kz
ky kx
+ + + rotational stiffness
lax, kry, krz
y
joint 1
82
The stiffness matrix k of this element has the form:
kx 0 0 0 0 0 kx 0 0 0 0 0
ky 0 0 0 0 0 ky 0 0 0 0
kz 0 0 0 0 0 kz 0 0 0
krx 0 0 0 0 o krx 0 0
kry 0 0 0 0 0 kry 0
krz 0 0 0 0 0 krz
k= (EQ 3.79)
kx 0 0 0 0 0
ky 0 0 0 0
SYM. kz 0 0 0
krx 0 0
kry 0
krz
The elements of the matrix k can be linear or nonlinear(k (u (t)) ). In the case
when interaction between the springs exists it can be added directly to the elements stiff
ness matrix in the proper locations (EQ. 3.79 is revised and the proper stiffness terms are
It should be noted that there is certain similarity between the end springs formula
tion presented earlier, and this element. The differences between the two cases are: (i) The
end spring can be connected only to a single element, while the spring element can be con
nected to a joint in which several elements are connected. (ii) for postprocesing purpose,
in the locations where the displacements must be monitored it is easier to use the connec
tion spring element and directly obtain displacements since there is an actual joint at that
location. This is especially important when those springs are connected to the ground,
since the relative displacements between the bridge and the ground are of interest. (iii)
83
when mass or force is connected to the joint between the. "stick" elements and the springs
(mass of the ground), the end spring formulation cannot be used since it assumes that there
The previously described structural elements are contributing to the global stiff
ness matrix k. In this section global modeling of structural damping, and modeling of indi
The global damping matrix is used to describe the inherent damping in the struc
tural system. Since it is difficult to quantify the damping in specific locations, it is consid
frequency of the structure. A common way to model damping is by using Rayleigh pro
(EQ3.80)
Where M is the global mass matrix, k is the global stiffness matrix, u and ~ are
determined so that specific damping ratios are obtained for two selected frequencies, say
1;  uM+ ~k (EQ3.81)
1 2Mco 1
1;  uM+ ~k
2  2Mco2
In the case of inelastic structure, the values of k and co are changing with time.
Several options are available to update the damping matrix (e.g. Fajfar et. el 1994): 1)
84
Keep the initial damping matrix constant for the whole time of analysis 2) Update the
damping matrix such that C = a.M + ~k/ (t) , where k/ is the tangential instantaneous
stiffness. 3) Update the damping matrix considering both the change of stiffness and the
(EQ 3.82)
In the case were elements with individual damping properties such as viscous
dampers are considered, they contribute to the damping matrix in the same manner as the
individual stiffness elements are contributing to the global stiffness matrix. The individ
ual damping elements has damping matrix of the size l2x12 with 6 D.O.F's at each side of
the element.
The basic damping element as mentioned earlier has a l2x12 element damping
matrix. The element represents damping in all degrees of freedom as shown in Fig. 3.19.
joint 2
~z
+ + rotational damping
crx, cry, crz
joint 1
85
And the element damping matrix c has the form
Cx 0 0 0 0 0 cx 0 0 0 0 0
cy 0 0 0 0 0 cy 0 0 0 0
Cz 0 0 0 0 0 cz 0 0 0
crx 0 0 0 0 0 crx 0 0
cry 0 0 0 0 0 cry 0
crz 0 0 0 0 0 crz
c = (EQ 3.83)
Cx 0 0 0 0 0
cy 0 0 0 0
SYM. CZ 0 0 0
crx 0 0
cry 0
crz
The behavior of each of the subelements in the damper element can be nonlinear as
well as linear. The c value can vary with time such that c = c (Ii (t) , u (t) ). In this case
the global damping matrix of the all structure needs to be updated in time.
et. al. 1995). The force in a fluid damper is a result of flow through orifices leading to a
pressure differential across the piston head. Most of the practical devices are built using
power of velocity:
(EQ 3.84)
86
       _ .. _ 
where sgn (Ii) indicated the sign of velocity Ii and IX is a power between 0.5 to
(EQ 3.85)
It should be noted that the values of the instantaneous damping in EQ. 3.85 are:
dF dF
e(t) . ~ 00 for IX < 1.0 and e(t) d. ~ 0 for IX;:: 1.0
duu .... o uu .... o
Fig. 3.20 shows a typical expansion joint detail with a restrainer and a gap in
extension of the size of tgap and gap in compression of the size egap. It may be assumed
that during travel form egap to +tgap no force is transmitted between the deck and the
abutment. When the compression gap closes, the joint exhibits stiffening characteristics
which can be approximated as bilinear stiffening elastic behavior. This is depicted in Fig.
3.21. In tension, the behavior is more complex: (a) the behavior is hysteretic with soften
ing characteristics, as approximately depicted in Fig. 3.21 and (b) the tension gap
increases with each inelastic extension from tgap to tgap' and so on, as depicted in Fig.
3.21.
87
Fig. 3.21 depicts an approximate forcedisplacement relation for an expansion
joint. A one directional element has been developed with these characteristics to represent
the behavior of the tributary section of an expansion joint. The behavior of the whole
soft
material
Beam
Restrainer
Abutment
Bearing
88
3.3 Modeling of Bridge Subsystems
In the previous section the mathematical modeling of the basic structural elements
required for performing analysis of a bridge was presented. In the current section the mod
eling of the bridge components described in section 3.1.1 , using the basic structural ele
ments is described.
The foundation system is the medium that connects the bridge to the ground. In
general, the bridge and the ground form a single structural system, and therefore it should
be analyzed as a single unit. Realistic limitations does not allow integral modeling and
analysis. Therefore each one of the systems is modeled and analyzed separately with sub
sequent interaction defined as soil  structure interaction. The methods used to analyze the
behavior of the ground by itself are: finite elements, boundary elements and several other
approximation methods. The use of the finite elements method for analyzing large volume
like the one of the soil, requires descritization of large portion of the soil to avoid creating
refraction of the propagating waves from the artificial boundaries of the model.
The use of boundary elements is much more appropriate for analyzing infinite
diamines, such as the ground medium Benerjee (1994). The boundary elements method
however has also significant drawbacks in approaching the problem: the problem of 3D
structural dynamics cannot be solved, since the required Green function to solve this kind
of problem is not readily available. Furthermore, this method can be viewed as a linear
superposition method, and therefore it cannot be used straightforward for nonlinear analy
89
3.3.1.1 Elastic Linear Models for Soil Structure Interaction
The simplest models used for analyzing soil structure interaction are the once
using equivalent linear springs and dampers to model surface or embedded foundations.
According to this method the soil behavior is represented by a spring and a dashpot. Their
properties are depending on the size of the foundation interface and the soil properties.
However simple, these model was found to be very suitable to represent the behavior of
the foundation interface as compare to more complicated models (Cofer et al. 1994).
A more advanced method for analyzing soil structure interaction with dynamic
effects, considers the soil medium as a combination of equivalent springs and dampers in
which their main properties are function of frequency. The soil is modeled therefore, as a
combination of spring and damper elements, described earlier. These elements however
taneous frequency of the nonlinear system. If the frequencies of the bridge with changing
stiffness are calculated at each time step of analysis, then the stiffness and the damping
properties of soilfoundation medium can be updated at each time step too. A more effi
cient computation scheme however provides such an adjustments only when substantial
The abutment is usually made of three walls as shown in Fig. 3.22, two wing walls
and one central load bearing wall, on which the deck is supported. The soil is supported
90
laterally by all the walls and apply passive pressure on them. A major roll is played by the
abutment in supporting lateral loads which are applied to the bridge during seismic events.
soil spring
bearing
wall
FIGURE 3.22. Typical Arrangement of Bridge Abutment. The Soil Stiffness Modeled By Springs.
The abutment can be modeled in two ways: the first one is more rigorous, in which
the soil is represented by several spring and damper elements, and the walls are modeled
by several "wall" elements. The other way is to lamp the all abutment system into few
springs and dampers, similarly to the way the foundations system is modeled.
Since the deck of a bridge is more critical and expensive to repair in seismic
design, permanent damage due to seismic loading is allowed only in piers and bents.
91
"".j1..
The bents are the parts of the bridge which ls..the most susceptible to damage dur
ing earthquakes. The bents and piers are designed so that they will experience damage but
will not collapse when subjected to seismic loading. Therefore the behavior of the bent
can be modeled by one or more of elements of the type of hysteretic beam I column. In
short piers and walls, the failure mechanism is dominated by shear mechanism. Therefore
the combined flexture shear model (see flexturalshear model in section 3.2.2.2) should be
used for modeling of these elements. In frame type bents, the top beam is also modeled as
an inelastic concrete element since it is also a part of the energy dissipation mechanism.
The deck is usually made of multiple beams supporting a plate or of box girders.
The deck can be modeled using very fine finite elements. From past experience however it
appears that this approach is very expansive computationaly, especially for nonlinear anal
this task even if some uncertainties are involved in deriving the correct properties. There
fore approximation methods are preferred for modeling of the deck. Two approximation
According to the first modeling method the deck is represented by a single beam in
the centroid ofthe deck. Thus the structural properties of the deck are collapsed to it's cen
terline. For this purpose the basic elastic beamcolumn element can be used. Since in
many cases the ratio between the length and width of the deck is smaller than 4, shear
deformations in the horizontal plane should be considered. The deck is usually supported
by bearings at its ends. Since the deck can be assumed to be rigid at its ends around the
vertical axis, the connection between the deck which is modeled as an elastic beam and
92
the bearings can be done using suitable rigid body transformations. Those rigid body
transformations can be performed using the coupled motions method (explained in the
Fig. 3.23 ,where the top joints of the bearings Goints I to 4) are coupled to the single joint
The other method of modeling the deck behavior for the case where irregular set
ting of the deck exist is shown as an example of the use of the coupled motions in section
3.4.
93
bearings
deck edge
4
"\
\
deck end
joint
c;
deck modeled by stick elements
rx
z
kJdzl
,: ,
.
_y_
dS 1
_ ...:...:...:~ •
 ,  1  •  •.=::::::_C":'"_....       •
d54 • a.."
,, 9
' ~'""
: dx 4
94
3.3.5 Expansion Joints and Special Connections
bearings
A typical expansion joint on a bent is shown in Fig. 3.24. The left deck section is
supported directly by the bent, the right section, is supported by the bent through bearings.
The two sections of the deck can be connected together by restrainers (see Fig. 3.25), that
prevent the right section of the deck from dropping from it's support, while moving in the
longitudinal direction
9S
~~
H restrainer
r~~
bearing
F
tension in restrainer
compression in a
concrete
96
A possible model of an expansion joint is shown in Fig. 3.25. In principle the joint
should be modeled using several elements. Since the deck edge can be assumed to move
as a rigid body around the vertical axis, all the nodes along the joint can be represented by
two nodes only. The bearings and restrainers, can be transformed from their actualloca
tion to the two nodes at the centerline of the joint using rigid body transformation, by
of freedom representing motions / movements of rigid bodies in the structures are coupled
by rigid'links. Such cases are customary in bridges with decks which can be considered
rigid inplane. The rigid links provide large stiffness terms into the structure global stiff
ness matrix as compare to the other more flexible elements of the structure, leading to
using appropriate geometrical relations. The solution procedure suggested herein is auto
matically processing and reducing the number of active degrees of freedom and the stiff
ness matrices for effective processing. This is done using an optimal scheme as outlined
The conditions on the relative motions are specified using the variational formula
tion of the structural system, by introducing Lagrange multipliers, which include the con
(EQ3.86)
97
Where II is the energy, U is the displacements vector, K is the stiffness matrix, R
is the load vector, and A is. the Lagrange multiplier; c and c' are coefficients of the con
Taking the variation of the expression in Eq. 3.86 and equating it to zero leads to:
(EQ3.87)
Since the terms associated with 0 are arbitrary, will get a set of linear equations
cU +c' =0 (EQ3.89)
Introducing EQ. 3.89 into EQ. 3.88 and taking c' as zero the number of equations
that needs to solves will be reduced from the initial number of Degrees Of Freedom
(D.O.F) to reduced number, equal to the original number of D.O.F's minus the number of
coupled equations. The terms in the structure global stiffness matrix will change such that
where Ci is an index on all the D.O.F's coupled to the i'th dof, Cj is an index on all
the D.O.F's coupled to the j'th dof. While the coupled (slaved) D.O.F's are eliminated
from the stiffness matrix. The terms in the global coupled load vector are defined with
98
Example:
The deck section in Fig. 3.26 is rigid in plane. The behavior of this deck section
can be represented by three D.O.F's only, i.e. displacements in two directions and rotation.
The procedure for performing the transformation using coupled motion is presented
below.
J4
Jl
Since the rotation of the whole deck is identical for the all joints, the coupling
read:
(EQ 3.92)
(EQ 3.93)
99
(EQ 3.94)
The first tenn in each equation is the coupled tenn, and the second is the master
term. For example in Eq. 3.92 the master D.O.F is 3 and the coupled D.O.F is 6. The cou
(EQ 3.95)
from which the coefficient c36 =1, (see Eq. 3.90 Eq. 3.91 ). Rewriting the next two cou
Defining dX ij as the distance in the x direction between joint i and j and similarly
dYij as the same distance in the y direction, the additional relations between the coupled
In the same manner by rewriting the equations into the form of Eq. 3.95 , the coef
lOO
Flexible Deck modeling
that the deck is composed of several parallel hysteretic beams, assuming that the deforma
tions in the vertical and horizontal direction are related through shear interaction.
A deformed section of such a bridge is shown in Fig. 3.27 . Assuming that the deck
cross section remains plan and vertical to the beam axis (Bernoulli assumption) the defor
mations in the horizontal X direction of all the beams (bj)can be coupled to a single beam
Ubi
x
= u xbj+6 y B IJ.. (EQ 3.102)
where the index bi indicate that the deformations of the i'th beam are coupled to
the displacements of the j'th beam and Bij is the distance between beam i and beam j.
Assuming that the inplane stiffness of the deck is infinite the following coupling equation
can be written:
Using the same approach for the vertical direction, the coupling between the
Ubi
y
= u ybj +6X B I).. (EQ3.104)
Where u;i are the deformations in the vertical direction of the i'th beam.
Since the height of the deck is substantial, the distance between the theoretical
beam center line and the actual support should be considered when modeling the connec
tion between the deck and bearings, which are placed on the top beam of a bent. To con
101
sider the offset, the beams can be modeled with offset connections (rigid arms).
Furthermore, in order to consider flexibility in the support connections, beam end springs
can be added.
•
x
z 1
r
!.:..................................... b:.~ __ ~~
I
102
3.5 Evaluation of Bridges via Approximated Methods
The evaluation of a bridge implies determining the internal strain resultant in the
bridge elements as well as the relative deformations in the elements and of the entire
bridge in case of load disturbance. In the case of earthquakes when the system responds
inelastically, the internal force distribution influences the deformations and accelerations
rately the force (strain resultants) and deformation distribution without a time history anal
show that the maximum response has the tendency to fit more simplified, independently
monotonic loading and the second by determining the expected maximum response for a
(a) Monotonic structure's capacity will be defined as the above mentioned force
(b) Seismic demand will be defined as the maximum expected response for a
the one describing the structural system for evaluation. This is defined using
A plot of the structural capacity, [see Fig. 3.28 (a)] indicating possible loading
paths, can be combined with a composite spectrum (Reinhorn et. al 1996) [see Fig. 3.28
(b)] to define the response at the intersection of the curves [see Fig. 3.28 (c)].
103
elastic response
Salg=F/w
F/w ... elastic Salg
.
•
inelastic
d Sd Sd=d
(a) capacity (b)demand spectrum (c) response
The current section attempts to show various techniques to determine the response
by determining the capacity and the seismic demand separately, as well as it explains more
It should be noted that a good evaluation can be obtained if the monotonic "capac
ity curve" is determined for increasing inertial forces best simulating those that might
develop in a single large cycle during an earthquake. At the same time the calculation can
104
3.5.1 Methods for Determining Seismic Demand
The seismic demand can be defined in terms of the expected forces and deforma
tions which will develop in the structural system due to seismic event. This can be done by
determining the response specific for a single degrees of freedom (s.d.o.f) or using pre
defined code averaged data for such spectra. However, the forcedeformation demands in
m.d.oJ's require farther evaluation as will be outlined below. Several methods currently
Current seismic codes including AASHTO suggest the use of modal analysis pro
cedure, which transfers the structural problem from the physical space domain to the
domain defined by the eigenvectors of the homogeneous equation of motion. In the fol
lowing a review of the method is presented along with the presentation of efficient compu
and EQ. 3.105 can be written for each mode i using its eigen vector <Pi as
(EQ 3.107)
The eigenvectors of this equations are derived as the roots of the polynomial
105
(EQ 3.108)
(EQ 3.109)
In order to solve this problem efficiently it is suggested to use the Lanczos method.
(EQ 3.110)
where Tn is a tridiagonal matrix of the size of n. The basic problem (EQ. 3.109 ) can be
rewritten as
(EQ 3.111)
The transformation
$ = X$ (EQ 3.112)
 1
T$
n
=$
co 2 (EQ 3.113)
The eigenValues of Tn can be found using one of the iterative techniques. The
eigenvalues of Tn are the reciprocal of the eigenValues of the original problem and the
106
The big advantage of using this method is that finding the roots of the polynomial
of tridiagonal system which is significantly reduced is much easier than working on much
The model of the structural system of the bridge is assembled from individual
matrices of elements to get the system matrices k, c, m, and the load vector F.
The eigenvectors and eigenvalues of the bridge are calculated using the aforemen
tioned procedure. It is assumed that the bridge can be modeled by a limited number of
modes. The site response spectrum is used to derive the equivalent forces on the bridge
using the expected acceleration profile to derive the distribution of the inertial forces on
the bridge:
(EQ 3.114)
where Fj is the force at thej D.O.F for mode i, Mj the mass at thej D.O.F, <Pij is the
ordinate at the j D.O.F of the i'th mode, and rSj the response spectrum value related to the
Using this force profile an equivalent static approach can be used to determine the
internal forces in the elements and deformations. The deformations can be also determined
directly from displacements spectra if available. If only few of the mode shapes need to be
considered, the results from the static analysis for the various modes can be combined
Fi = JU 5 (EQ 3.115)
107
where Fj is the end force in element j F ij is the j element end force due to the response in
the i'th mode. The indices for the displacements u are defined similarly to the once for the
forces.
It should be noted that this method can be used also for inelastic response evalua
The simplest way to evaluate the demand from a bridge, is by representing the
dynamic behavior of the bridge (explained in the previous section) by equivalent static
one. This approach is widely used by the design codes (UBC 94, AASHTO, etc.).
This procedure is based on solving the static equilibrium equations of the bridge in
which the forces f are determined however using simplified empirical formulation to dis
tribute the total force. The distribution of the forces is obtained according to the force fi
relative to mass M i, and to the expected total force I.F j as defined in design codes:
(EQ 3.117)
ku = f (EQ 3.118)
The matrix k can be separated into four submatrices related to the free and con
strained D. O.F' s
108
(EQ3.1J9)
In the case where Uc equals to 0, the problem can be viewed as force input prob
(EQ3.120)
(EQ 3.121)
where kif is the tangential stiffness matrix, l1u the incremental displacements of the free
f
The reactions on the restrained D.O.F's can be calculated from the equation
(EQ 3.122)
The same procedure can be used to apply displacements input in which case the
(EQ 3.123)
Time history analysis provides the most accurate information about the response of
bridges to seismic loading. If such loading is well known, structural parameters well
defined and the model is complete. Two procedures can be used to perform time history
analysis (Clough and Penzin 1993) i.e. (i) mode superposition and (ii) direct integration.
109
The mode superposition procedure even though much more efficient numerically can be
used only for linear analysis. Differential support excitation cannot be considered ade
quately by this method. The direct integration procedure is the one that is recommended in
the current approach due to its versatility to handle elastic and inelastic structures.
Bridges span over large distances, and therefore the ground deformations, acceler
ations and velocities at one support might be significantly different than at other supports,
in magnitude and phase due to several reasons: (i) When the bridge is located over a fault
the excitation on one side of the fault differs significantly from the other side. (ii) Different
soil types amplify differently the magnitude of traveling shear waves, and cause excitation
magnitude difference between different foundations. (iii) The difference in arrival time of
the shear waves to largely distanced footings causes phase difference between the ground
excitations.
All this affects the response of the bridge both by changing the dynamic excitation
of the bridge, and by applying quasistatic relative deformations which result in quasistatic
forces.
The solution procedures for both, uniform support excitation, and differential sup
The governing equation for nonlinear time history analysis can be expressed in
uo
Mdu + Cdu + Kdu = dF for ti ::; t::; ti +!:J.t (EQ 3.124)
in which the nonlinear properties C (u, u) and K (u, u) are assumed constant during the
time step interval. The excitation which is a uniform ground motion, in which the acceler
dF = Mdug (EQ3.125)
This equation is solved using an implicit scheme based on the Newmark ~ method
(Bathe 1982).
du  utdt  dt 2 (112  a) ut
ut + dt = =d:t2:a.: (EQ3.126)
du ut U (112  ex) ..
du = =.:,:.:;,
(dt) 2ex adt ex
I
u
I
(EQ 3.129)
Introducing EQ. 3.128 and EQ. 3.129 into EQ. 3.124 ,will give the numerical
111

kdu = F
 (EQ 3.130)
where

k=M 1 +C+K
(dt) 2ex.
°
ex.dt
(EQ3.131)
 = FM[,,1
F + u  +u  .IJC[(OI)"
+ u +  0(1  ex. )"O'J
u + u
t2ex. tdtex. dt ex.dt 2 ex.
The global equation of motion in matrix form, for the acceleration input is:
(EQ 3.132)
in which mf efand kf are the superstructure mass, damping and stiffness matrices respec
tively. Those matrices are related only to the free degrees of freedom. The relative dis
The equation can be written explicitly with respect to the total displacements u
t
such that:
CEQ 3.133)
112
where the sub matrices with the c index corresponds to the restrained degrees of freedom
or the degrees of freedom where ground displacements are applied. The first block of
equations in Eq. 3.133 represents the behavior of the superstructure and can be written as:
(EQ 3.134)
where the acceleration, velocity and displacements marked f are the total displacements
of the structure, while the displacements and velocity marked g are those related to the
ground. The solution of these equations is similar to the solution for the uniform ground
motion with the proper adjustments for the forcing vector, i.e. the right hand side of EQ.
3.134.
son between deformations resulting from analysis with acceleration input motion and
analysis with displacements input motion is performed. The verification model shown in
Fig. 3.29 is analyzed using displacement input shown in Fig. 3.30 (b) and acceleration
input shown in Fig. 3.30 (a). The displacements results extracted from the base displace
ment analysis are the total displacements ufin Fig. 3.30 (b) which includes the ground dis
placements. The results from the acceleration input analysis are the relative deformations
(u in Fig. 3.30 (a)). To compare the relative displacements u for both cases, the ground dis
placements are subtracted from the total displacements to get the relative displacements
between the base and the structure (Fig. 3.30  b). A comparison of the results is shown in
Fig. 3.30 . It is apparent that the relative displacements are the same. However it should be
noted that the acceleration input does not require any restriction on the initial conditions of
113
\.
the ground motion. However with displacements input the initial conditions should be
properly adjusted so that the first and the second derivative will equal to zero.
it
I
,
I
t+ ,
I
,
I
I I I I
I I I I
I I I I
I I
I
(a) ~U
g
(b)
114
\
11O
C
~ 0
II
g
~!f·10
'6
1"500
·201r,~,_,\, •
time (sec]
·'00c0'n.,:....,.,,\,0
time [sec]
20r_ _ _~ro~m~l~d~is~l~ac~e~m~en~t_ _ __,
I'0
w
~ 0
~!f·10
is
·20 ir:::__,\,
time [sec]
20r_ _;rr~e~lm~lv~e~d~lso~lla~c~em~en~t_ _ _, 20r_ _;rr~el~at~lv~e~d~IISSO~lla~~~m~en~t_ _ _,
I'o f I 10 ~
c
~ 0t\!
;10 I
'6
·201r.L.":~_...lL._ _.,\, .201r.L.:":~...lL.,b
time [sec] time [sec]
115
3.5.3 Numerical Solution for Bridge Systems
kx =/ (EQ3.135)
 
kx = / (EQ 3.136)

where k is the reduced stiffness matrix considering the coupled or slaved motions, and /

is the reduced force vector.
(1) Usually the stiffness matrix k is a banded matrix, for efficient solution of the
by the bandwidth and the length of the diagonal. This type of storage is not efficient in
many cases. In order to improve this storage, most of the commonly used solution proce
dures perform reordering of elements to achieve smaller band width. This issue might
especially be significant when coupled motions are used, because the band width might
increase in the presence of coupled motions, since the D.O.F's that didn't have interaction
terms initially, will be interacting after the modification related to coupled motions.
The global solution procedure used for solving EQ. 3.136 is usually the Gaus elim
ination or frontal solver. These solvers usually have problems of round off error when the
stuffiness matrix has terms with drastically different values. This might happen when the
structure has stiffness discontinueties, in which soft and stiff elements are present in the
116
In order to reduce the difficulties mentioned above, a sparse matrix storage com
bined with a solution using the Conjugate Gradient Iterative Solver are used. In the fol
Iterative solvers are based on vector matrix multiplication. This can be done very
efficiently by using sparse format storage of the stiffness matrix. The sparse format used
A symmetric matrix K can be represented by three vectors A, JA and IA. All the
nonzero elements are stored in the vector A. JAG) contains the column number at which
the term AU) is located in the K matrix. The IA vector denotes the elements serial number
on the diagonal, and used to define the number of nonzero elements at each row. JA has
the size of number of nonzero elements, and IA has the size of number of rows + 1.
The matrix K is
14. O. O. 44. O.
15. O. O. O. 55.
JA =[1 4523455]
117
IA=[1 4 5 6 8 9)
An iterative solver was chosen for two major reasons: (i) The accuracy of the solu
tion is not influenced by large differences between the values of the terms in the stiffness
matrix. (ii) The solution from one step can be used as a first guess for the solution at the
next step which is useful in incremental static analysis, and in step by step dynamic analy
sis where there are no significant changes expected in the incremental solution between
CG is the most popular iterative method for solving large systems of linear equa
tions. It is effective on systems of the form
Ax = b (EQ 3.138)
where A is a known symmetric, positive definite matrix. Iterative methods are suit
able for use with sparse matrices because they usually use matrix vector multiplications
operations.
(EQ 3.139)
For positive definite matrices the surface defined by f(x) has the shape of parabola
for the case where the size of vector x is 2.
(EQ 3.140)
and if A is symmetric
118
f (X) = Axb (EQ 3.141)
Setting the gradient to be zero will retrieve EQ. 3.138 . If A is positivedefinite and
symmetric, this solution is the minimum off.
In this method we start with an initial guess Xo and take a series of steps to values
of x: x j>x2'" until we are close enough to the solution x.
The direction of the step is chosen to be the direction of the gradient to the function
fwhich is also the residual
(EQ 3.142)
(EQ 3.143)
Moving from one point to another in the search for the minimum point will be
done in this manner.
(EQ 3.144)
After finding the direction of the step ri' to find a line search is performed along
the direction of ri' such that a minimize f where
dx. 1 d(x.+ar.)
 0  fT ( )  '+  fT ( ) , ,
  xi+l da  xi+l da
(EQ 3.145)
Using the last term and EQ. 3.143 and EQ. 3.144 the value of a is derived as
rTr.
a='' (EQ 3.146)
rTAri
119
(EQ 3.147)
(EQ 3.148)
~2
and k = rAl2
;~ = ~
':>1
where Al ' A2 are the eigen vector of matrix A and ~I ; ~2 are the length's of the projec
It shows that the smaller 0) is, the faster is the convergence. The term k relates to
the ratio of the eigen values and the term ~ relates to the ratio of properties related to the
initial guess. The drawback of steepest descent is that steps are taken in the same direction
several times, and the number of iterations might be large.
This method is based on the idea of choosing the search directions so they will be
A orthogonal, in this way the search in each direction will be done only once and after
moving n times in each direction, the solution is found.
(EQ 3.149)
;1
d; = uj + L ~jkdk (EQ 3.150)
k=o
and
120
~ _
uTAd.
, J
(EQ 3.151)
ij  dTAd.
J J
The drawback of the method of cojugate directions is that to derive each new vec
tor, all the old vectors should be used and kept, this requires large number of operations
O(n 3).
set of vectors simplifies the solution considerably because of the special properties of r ,
j
(EQ 3.152)
rTr.
a. =  '  ' (EQ 3.153)
, ,
dTAd.,
(EQ 3.154)
where
and
(EQ 3.156)
(EQ 3.157)
(EQ 3.158)
"'min "'max are the minimum and maximum eigenValue of the matrix A,
121
and the rate of convergence of CO (or the decrease of the error norm) is
(EQ 3.159)
Preconditioning
(EQ3.160)
(EQ 3.161)
(EQ 3.162)
r!MI r .
ex. = I I
(EQ 3.163)
, ,
d!Ad. ,
(EQ 3.164)
(EQ 3.165)
(EQ 3.166)
(EQ 3.167)
122
3.6 Methods of Analysis to Evaluate Bridge Capacity
Capacity is a limit state which can be defined as: (i) The maximum force or defor
mation that a system or subsystem can experience before it reaches collapse. (ii) A system
failure is reached when it cannot play it's designated roll. A failure of a system occurs
Each element in the system by it self can fail in different ways: (i) Section failure,
such as failure in flexure (ii) Zone failure, such as failure in shear. (iii) A failure of mecha
The reliability of a system can be derived in two basic ways. (I) By first deriving
the reliability of each individual member and then combining them into reliability of the
global system. For example by using "Fault Tree" approach, in which the probabilities of
each component of the system are arranged in a system "Fault Tree". And the critical path
in the tree defines the probable failure path of the system. (2) By using analysis procedure
Similarly the capacity of the system, can be assessed by deriving first the capacity
of the basic elements directly from the constitutive relations of the individual element and
expending it to the global system, or by assessing the capacity of the whole system
directly.
In the following sections, a brief review of the commonly used techniques to eval
uate ultimate state or capacity of a section are presented, followed by presentation of pro
cedures to evaluate the structural capacity of the whole bridge system, namely by
monotonic pushover or dynamic analysis procedures in which the capacity of the struc
123
" 
ture as a whole is derived by applying simple loading to the bridge and detecting condi
The derivation of the moment curvature relations for a section in flexure is well
known. Two methods are used usually for this purpose: (i) the approximate one based on
stress block and, (ii) the more rigorous one: fiber analysis. Both methods are based on the
following assumptions: plane sections remain plane, so the concrete strain distribution can
be defined just by two variables (e.g., the strain at the top face and the strain at the bottom
face) or by the strain at the centroid, Ecen and the curvature $. The curvature is equal to
the change in slope per unit length along the member, and is also equal to the strain gradi
(EQ 3.168)
The strain in a reinforcing bar at any level y is equal to the strain in the surrounding
concrete; hence
(EQ 3.169)
124
In which!c,! s are stresses in concrete and steel, Ac. A s are the areas cross sections
directly by assuming that the stresses in the concrete are concentrated in a rectangular
The moment curvature relations are derived by iterating on the eqUilibrium equa
tions, for a given axial force and moment using the stress strain relationships for steel and
concrete.
Instead of assuming distribution of the stresses over the height of the section, the
section is divided into number of layers where the concrete and the steel micro mechanics
are used. The integral of the stresses of the layers over the sections height should satisfy
the equilibrium conditions presented earlier. Iterative procedure is used to find the stresses
at the layers, and the related curvature. The ultimate state is defined as a condition where,
system capacity to sustain forces and deformations before reaching a point of failure is
presented.
125
The capacity of a single element is usually defined as the extreme value on the
constitutive relation curve of the element, when the failure mechanism is expected to be
flexure, the failure happens in a single section. However when shear failure occurs it can
The commonly recommended way in the design codes to evaluate the strength
demand of a single element is by reducing the forces derived using elastic analysis for
seismic loading Felastic by a reduction factor R which relates to the assumed capacity of
F _ Felastic
design  R (EQ3.172)
Using this procedure it is assumed that the inelastic behavior increases the seismic
capacity of the elements. However the inelastic behavior doesn't influence the distribution
of the forces in the bridge due to inelastic behavior which might transfer significantly
larger force to certain elements. In order to give more emphasis to the plastic behavior
(Seismic Retrofit Manual for Highway Bridges  Buckle and Friedland 1995), a mono
tonic incremental static lateral analysis or dynamic analysis with a predefined dynamic
load, defined also as "PushOver" analysis is used to evaluate the capacity of the whole
structural system, or major parts of it such as frame bents. In such a case a single demand
to capacity ratio, which represents the reliability of the whole system can be defined. The
derivation of single system demand to capacity ratio, is desirable for accurate road sys
terns evaluation.
In order to derive the capacity of a structural system: (i) the geometry, (ii) the con
stitutive law ofthe elements and (iii) the loading distribution should be known. Assuming
126
that the geometry can be well defined by geometric modeling and the constitutive law can
be also well defined, the main difficulty is to define an appropriate distribution of the total
force F. The distribution of the total force F between the elements of a bridge should be
The distribution of the mass in building structures can be considered in the vertical direc
tion only, assuming that the floor diaphragms are stiff in there plan. Using this assumption
EQ. 3.173 may not be simplified as easy. In bridges of significant length where the number
of spans is more then two, the modes of vibration are much more complex and so is the
bridge response to earthquakes. To get better evaluation of the bridge response, several
types of "PushOver analyses" are suggested. Two of them, i.e. the incremental force load
ing and incremental displacements loading, were previously defined in various versions
for analysis of buildings (Valles et. el. 1996). To be able to represent fue accurately the
influence of the changing structural properties of the bridge due to inelastic behavior adap
tive (modal) loading is suggested. The other two options are dynamic versions of the static
"PushOver procedure", in which the expected dynamic loading and the adaptable distri
127
Both the previously developed push over procedures (arbitrary force loading) and
the newly developed procedures (displacement quasistatic, modal adaptable and the quasi
The simplest and most commonly used method for "PushOver analysis" is per
formed by applying incrementally lateral forces to the bridge until failure criteria is
reached. The incremental static analysis with force input presented earlier is used. Several
criteria (Valles et. el 1995) are used as a definition of failure: 1) first component reached
failure, 2) gravity load carrying element reached failure, 3) mechanism is formed, 4) ser
As mentioned earlier the distribution of the forces along the bridge is an important
issue and two types of distribution which are similar to the once which are used in build
ings (Valles et. al. 1995) are suggested: 1) uniform load distribution 2) load distribution
There are several drawbacks in using this procedure: 1) if the modes of vibration
represent the displacement response of the bridge, the acceleration response will be pro
In the general case the force distribution is the fourth derivative of displacements:
128
d ( 1d
[(x) = dx2 Eldx2
2
u) (EQ 3.177)
In the case of complex modes of vibration of the bridge. The deformation u and the
external force [are distributed as two different function in space. 2) usually it is consid
ered that only few first modes of vibration are enough to represent the actual displacement
response (Clough and Penzien 1993) however it does not apply for acceleration response.
Because of the limitations mentioned in the previous section (the forces in the
bridge cannot be accurately described by the modes of vibration) analysis based on the
previously described incremental displacement input can be performed. The analysis pro
cedure for this case is similar to the one used in the previous section for arbitrary force
input. However instead of applying forces on the bridge, a certain displacements pattern is
In the previous two sections, the distribution of the load is not a function of the
damage the structure sustain due to inelastic behavior. To provide more compatibility
between the load distribution and the change in the structure stiffness, adaptable modal
loading push over procedure is developed (Reinhorn et al. 1996). This procedure applies
both to the force input and displacement input procedures. In this procedure at each time
increment the mode shapes are calculated using the instantaneous stiffness of the structure
which changes due to plastic behavior, the incremental force at location i along the bridge
129
(EQ 3.178)
in which <l>i,j (u) is the ordinate of the j'th mode vector at location "i";, rj (u) is the
modal participation factor for the j'th mode; Ftolal is the total force, or total displacements)
and F11d is the quantity level at location "i" in the previous loading step. With the load
In the previous section the basic assumption is that the effect of a random dynamic
loading can be represented by static equivalent loads and the effect of the dynamic behav
A better approximation for the behavior of the bridge under dynamic loading can
One type of such analysis is the dynamic loading using a base "Ramp Accelera
130
Ground Acceleration
Application rate
Time
time as shown in Fig. 3.31 until the "failure" condition is reached. The rate in which the
acceleration is increased influence the results. When the acceleration is increased slowly,
the behavior is similar to applying static loading. Significant dynamic effects appear when
the rate of application is high. A range of application rates should be used, to estimate reli
In many of the recent strong earthquakes the major damage to bridges was attrib
uted to a single strong short impulse. This pulse has the effect of one strong "Push" on the
structure. A way of simulating the expected response of a bridge near failure is to apply a
single pulse with high magnitude of acceleration to the structure. The shape of this pulse is
not significant, if it is short enough relative to the structures natural period (Clough and
Penzin 1993).
131
mx+kx = mxg (EQ 3.179)
Integrating the equation on both sides will give the impulsemomentum relation
ship as
I,
'I
mllV;: fp (t) dt (EQ 3.181)
o
Therefore an initial velocity analysis can be performed.
Since the level of energy that will cause the structure to fail is not known appriori,
an iterative procedure is used. The initial energy required to bring "failure condition" is
estimated for the first iteration, and then increased, until failure is observed.
In order to evaluate the reliability of the system, the demand and capacity of the
During nonlinear time history analysis, the component's capacity and demand can
be compared instantaneously, at each time instant during the analysis, since this compari
132
son represents the amount of damage sustained by the element it is defined as "damage
(EQ 3.182)
where Ud (t) is the maximum displacement attained in case of shear or the maximum cur
Eh (t) is the hysteretic energy dissipated until time t, and Qy is the yielding force
(EQ 3.183)
(EQ 3.184)
is the displacement capacity. It should be noted that the capacity is reduced as the hyster
etic energy Eh (t) increases during the inelastic excursion (see EQ. 3.184 ). Therefore the
damage index DI is a function of time, and it is not possible to derive it by simply compar
The final damage index DI is the largest value attained during the analysis
f
133
3.6.3.2 Evaluation of Bridges in ModeratelLow Seismicity Zones· Procedure
In order to evaluate bridges with low risk, costly and complex procedures are not
required. Simplified equivalent elastic procedures are enough to provide adequate evalua
tion, especially since other load cases might control the design. The procedures described
hereafter are described in the AASHTO code, and brief summery of them is presented for
sack of completness.
The static equivalent procedure is very simple and easy to implement. The
response of the bridge is assumed to be similar to the response of a single mass, which is
the attributed weight of the deck and the upper part of the bents, on a single spring, in the
transverse direction. In the longitudinal direction the deck is supported by the bents later
iJ
ally, so the response can be approximated by single mass of the deck and upper part of the
A
In the transverse direction, the bridge can be divided into separate frames, where
each frame represents the bent with the tributary section of the deck. By doing that, the
The lateral force on the frames is derived, using code formula, considering the
mass the stiffness, and the PGA (Peak Ground Acceleration). This procedure is very sim
pie although not very accurate, and can be used, as any other static loading case. The
resulting element's forces are compared to the nominal strength of the element increased
by ductility factor which represents it's ability to dissipate energy. The deformations of the
134
bridge are compared to design requirements in the code, which represent stability and ser
viceability requirements.
Modal spectral analysis is a more rigorous procedure than the static one. In this
procedure the dynamic behavior of the structure is considered, and therefore the engineer
The bridge modeled as framed structure, represented by it's stiffness and mass.
Modal analysis that was presented in the previous sections is performed to derive the
As it was described earlier using the mode shapes and the response spectrum the
loading for each mode can be derived. The structural analysis performed on the bridge
The comparison of the forces demand and capacity, is similar to the one performed
for the previously presented static procedure. The element forces are compared to there
nominal strength, and the displacements are compared to the once in the code.
In severe seismicity zones, the response of the bridge to earthquake will be usually
the lateral loading case that dominates the design of several components of the bridge.
Usually performing design with the methods presented for the low/moderate seismicity
135
The appropriate methods to be used in severe seismicity zones are suppose to con
sider more rigorously both the dynamic behavior of the bridge and the resulting design
forces, and the element's nonlinear capacity under cyclic dynamic loading. The methods
presented in the following sections address this issues. The first procedure uses the spec
tral method presented in the previous section to derive the demand. The capacity however
is addressed more rigorously using "push over" techniques. The second procedure, evalu
ates directly and instantaneously both demand and capacity, and "damage index" is used
Comparing the demand derived from the spectral procedure, and the capacity from
the "push over" procedure, enables direct calculation of the reliability of the bridge. The
• The nonlinear force displacement curve (Fig. 3.32) is derived using one of the proce
dures presented earlier. The choice of the procedure should be done according to the
single pulse, such as Northridge, which excite many modes, the appropriate procedure
would be "dynamic pulse". While "adaptive modal" procedure is not appropriate, since
only few modes are used to represent the load. On the other end when the expected
earthquake has vibratory nature "adaptive modal" or "ramp loading" would be appro
priate for the same reasons. In the longitudinal direction, there is single displacement
might be different in different locations along the bridge, and equivalent displacements
136
The limit state that defines failure, is defined according to the performances expected
from the bridge. If the bridge is expected to remain serviceable following an earth
quake, the service limit state should be used. While if the bridge is not essential the lim
Total force
Equivalent dlsp.
u.
u.
11. = "ii" (EQ 3.186)
y
. • Using the displacements ductility factor 11., the reduction factor Rtis derived using
equivalent displacements approach for long period bridge, (Priestley et al. 1992)
137
and equivalent energy approach for short period structures
. T
forT<1.5sec Rf = 1+0.67(l1u l)1'""<l1 u (EQ 3.188)
o
Where T is the dominant period, or equivalent period, and To is the period at which
• The inelastic capacity of the bridge Fu is converted into elastic capacity Fe' such that
(EQ 3.189)
• The total force demand F demand is derived using the regular modal spectral procedure
described earlier
R = 1 Fe (EQ 3.190)
Fdemand
The total reliability of the bridge system is a very important parameter, and can be
cv
used to evaluate the reliability of bridge system asA lifeline following an earthquake.
Although this procedure compromise the accuracy by using procedure which only
approximate the dynamic response, and the capacity evaluation procedure is not address
ing the issue of capacity reduction due to cyclic loading, the advantages of this procedure
are obvious.
In all the previously described procedures the capacity and the demand are derived
separately, both are only approximations of the actual response, and there is no direct
138
U sing time history analysis provides a much more accurate tool, for the evaluation
of bridge reliability. The dynamic properties of the bridge such as the mass and the stiff
ness, are directly introduced into the analysis procedure, as well as the nonlinear elemen~
properties which include the variation of force and stiffness as function of displacements,
Using the directly defined properties and the variation of the properties during the
The damage to the nonlinear elements is evaluated instantaneously using the pro
cedure described previously, and the final damage to the element is taken as the maximum
of the damage values calculated at every time interval during the duration of the analysis.
Although the analytical procedure by itself is accurate, the earthquake that the
bridge will experience in the future is unknown. Therefore the analysis should be per
formed several times, using set of several earthquakes that can reliably represent the site
The damage index or the reliability derived by the analysis apply only to individ
ual elements or sections of elements. The desirable reliability index however is the one of
0
the system a~whole. To achieve this target, engineering judgment should be used as of the
contribution of each element to the reliability of the whole system with the different per
into the reliability of a global system is the "fault tree" method (Melchers 1987).
139
4.0 CASE STUDIES FOR BRIDGE EVALUATIONS
4.1.1 Case Study #1: Analytical Evaluation of Small Scale Bridge System
Results of an experimental study performed by Tsopelas et. al. (1994) are used to
verify the capacity of the computational models to evaluate the response of structural sys
terns which include "base isolation systems" placed at arbitrary locations in the structure
A shaking table test was performed on a simple scaled bridge model (1:4) (a sche
matic of the bridge on the shaking table is shown in Fig. 4.1 ). The bridge consisted of a
deck of 140kN weight, supported by four flexible columns. On top of the columns were
bM.I,,~ 1\\ ~~
four Friction Pendulum System (FPS) bargains (Zayas et. al. 1987). The bridge was sub
jected to ground motion defined by the Japanese seismic design code as "level 2" and
Fig. 4.3 shows the geometry of the FPS bearings. Each of these bearings carried a
load of 35kN. The frictional characteristics of the bearings were experimentally deter
The bridge was modeled as a single frame as shown in Fig. 4.5. The deck was
modeled using a very stiff elastic beam element, and the weight of the deck was concen
trated at joints 5 and 6 (deck ends). The isolators were modeled using a continuous kine
matic model, known as Wen's model (see Sect. 3.2.4). It should be noted that the model
used for the bearings is not velocity dependant although the bearings friction characteris
tics are velocity dependent. The dominant coefficient of friction as observed from the hys
140
teretic loops obtained from the shaking table test experimental results was found to be 9%.
This value is related to the behavior of the bearing in low velocity (Fig. 4.4). The
required parameters for this model, Le.: the initial stiffness kinirial, the secondary stiffness,
kyie/d, the yielding force Fy and the post yielding ratio n are derived using the geometrical
properties of the bearing, the friction coefficient of the bearing and the supported weight.
The yielding force is Fy = WIl = 35 x 0.09 = 3;'..J5kN. The other parameters, the yield
ing displacements uy was determined from the component testing hysteretic loop as
0.255mm. Using Fy and uy the initial stiffness is kinirial =13059kN/m. The post yield stiff
ness of the isolation system is calculated from the radius of surface R and the weight W:
kyield = (j)2m = (~)41t2 :.~ = 62.4 kN/m which gives post yield ratio nof 0.006. The
simulated ground motion is applied to the frame in the longitudinal direction. The results
The stiffness of the columns was determined from their tabulated structural prop
erties. Because of flexibility of the shaking table the supports to the bridge cannot be con
sidered fixed. Therefore the degree of fixity was determined from the experimental results,
using the free vibration characteristics of the bridge in very low amplitudes when the bear
One of the parameters required for the analysis procedure is the global structural
equivalent damping. Evaluating this parameter has significant influence on the results
obtained from the analysis. As mentioned in the previous section, the global structural
141
1 _.
I
I
of the two (Rayliegh) damping. In order to observe the influence of global structural
damping and types of damping on the analytical solution, several analyses with different
levels of damping were performed. A comparison between the analytical and experimental
results for different levels of damping, is presented in Fig. 4.6 and Fig. 4.7 .
ments of the deck (Fig. 4.6) for most types of damping. However In the case where there
maximum peak.
The displacements at the top of the piers (point 3, 4) (Fig. 4.7), are related
directly to the moments and shear forces at the piers. Those displacements are more sensi
tive to the global damping representation. In most cases the analytical results overestimate
the experimental, the best results are achieved with 3.4% mass proportional damping, and
a small contribution from stiffness proportional damping. This happens probably because
the high modes of the piers are damped, and therefore it is more difficult for sliding to
occur
The change in damping representation influence the results because the of the
change in the stiffness matrix when the bearing are softening. Tsopelas et al (1994)
applied damping. terms only to the degrees of freedom which has only linear elements con
The presented comparison shows that using careful representation of global damp
ing, the behavior of isolated bridges with flexible piers can be adequately evaluated using
142
r;====....;,,___....,...; II AISC
WI4.VO
0.!5 ..
~ ~. ~.
 I AISC
1'S ...../11
I
3.....
4.10...
'.0...
0,2
~
c
'~ 0,0
.£
~ ·0,2
, Time [sec,)
143
...
'"
units:mm
IUII sc"
1'.l!'!L....,
144
z
0.15 r.. . C.."O....M:=P""O""S=IT='E":":M,...A=TE="R::':'IA..:L"""N,..,..O....,.3,....,
Q fmax0.120, PRESSURE 17.2 MPa (2.5 ksl)
t o.12 ·  o •• • •
•
IE
IS
~ 0.09
V •
~ 0.06.!.e......:..;....._.:.
w fmax0.062, PRESSURE 275.8 MPa (40
___ksl)
....:....:j
8
~ 0.03 o IDENTIFICATION TEST
uj • SEISMIC TEST
145
Deck,L = 6.0 m
3 ~ Isolator
Column, L =l.4m
1
:;::;
146
Eltpcriment
Analysis
1.7% mass prop. damp. 1.7% mass + small stiff prop. damp.
0.01 0 r'"T"'""':"""::"'""T"=~' 0.010
0.005 ~ 0.005
~~
]:
~
~O.ooo
i<>0.005
.0.0100.L;;0"~5.':.0:;';;1O!c.0;";}15.0
3.4% mass prop. damp. damp, + small stiff prop. damp
0.010 ,r,...:.,...:,.....:..,
0.005 0.005
]:
e O.OOO 0.000
i
&0.005 0.005
0.005 0.005
]:
•
~ 0.000 0.000
i 0.005 0.005
.0.011O~'::';';;'~7t;;""7!15.0 .0.0100'.'::0;''5''.0;;';;~'~~
Time (sec.) Time (sec.)
147
 Experiment
1.7% mass pro. damp.  Analysis 1. 71"'%_m_as..,s..:p_ro_._da,.m,p._+_s~m_a1_I_S_til"'f..:;.p_ro.:;.p..,.d;.:am~p__,
0.10 ,,,..,.....;.i., 0.10
g 0.05 0.05
!
~
0.00 0.00
is 0.05 0.05
I5 .0 0·100."'0'c5.1;;0';;101;.0;';;I15.0
0.100."'0'c51;;.0';;101;.0;'d
3.4% mass prop. damp. 3.4% mass prop. damp. + small slif. prop. damp.
0.10 ,...::..:..:.:r=::.:.!.;.:::::==r:., 0.10 ,'';....,..'0';,
0.05 0.05
I
g
0.00
0.00
is 0.05 0.05
0.05 0.05
g
t: 0.10 '"..1':'::''
0.00
0.05
0.10 '::",f;:'~;__~;;
M ~ 1M 1~ M ~ 1M 1~
Time (sec) Time (sec)
148
4.1.2 Case Study #2: East Aurora Bridge  Field Test
A full scale field test performed by Wendichansky (1996) on the southbound high
way bridge crossing Cazanovia Creek and laboratory component experiments (Mander et.
al (1996) were used to evaluate the capability of the analytical platform to adequately pre
dict the transient dynamic response behavior of bridges with different types of bearings
during the experiment. The bridge is located in East Aurora, New York, on Route 400. It is
a three spans continuous structure, supported on two abutments at its ends, and two piers
in the interior (Fig. 4.8). The south bound bridge deck is supported on seven girders,
which are connected laterally by stringers (Fig. 4.9, Fig. 4.10 and Fig. 4.11 ). For the
original condition of the bridge, the girders were supported on steel bearings. Three of the
supports had expansion bearings, and one pier support provides the longitudinal fixity.
The bridge was retrofitted and the steel bearings were replaced by lead rubber bearings at
the abutments, and laminated rubber isolation bearings at the piers. Several snap back tests
were performed on the bridge with both types of bearings. The bridge was pulled by two
jacks connected to the bridge at points on the deck above the bents and then quick released
to observe its free vibration response. The history of the force applied to bridge measured
The bridge was analyzed using the analytical platform for both types of bearing
settings. The structural model used for the analysis is presented in Fig. 4.13. The deck
and the bents are modeled as elastic beam with shear deformations modeling capability
because of the low length to width ratio of this elements. The soil at the bent's foundation
level and at the abutments is modeled as elastic foundation element. The participating
149
Although each deck section between the bents is modeled as a single beam ele
ment, and the deck and the bent are connected only at one point, there is no loss of accu
racy (see Fig. 4.11 ), since the rigid body motions at this connection are represented by
rigid arms (see section 3.2.1.1). The finite stiffness of the stringers in the lateral direction
can be modeled by using beam end springs (see section 3.2.1.3). The linear properties of
Both types of bearings were modeled as Wen's model type isolator element (see
section 3.4.2.1). The required properties of the bearings were retrieved form laboratory
The displacements measured during the experiment for the steel bearings (Fig.
4.15 ) and the elastomeric bearings (Fig. 4.16) were compared to the results from the
It should be noted that although the bridge response is primarily governed by first
mode behavior, due to the angle of the applied load and the skew inherent in the bridge, all
modes are excited by the snap test. It is therefore essential to use 3D nonlinear modeling
to capture all of the behavior attributes. A 2D nonlinear program can be used to achieve
reasonable results, in an approximate fashion, but not without great difficulty as demon
ISO

H
11"0 • 2f213
151
 _. . . _
f.l.711~
,
~
,).,
I I
rr .., ' 686
rr"" rr~
~L,' rr'.., ;
'" ~'0450r ~
f 
N
'" on
on
'"'"
t
lIlI'!l
, "
_ 1753 __
•

N
on
152
".
13869
StIffeners 178x 16
..
ii !I ',
~
i
... JoL ;~
Ii
11 L
~
ABUTMENT
153
r:.
,II.
147U
l
I
I
~
7;l91
,II.

80475
12850
+ ..~
I
I
\ ~ &~ ,II. JI
,
 "''L.271.93m
~
JI.
".
154
.".
100.0
~ 50.0
If
0.0
".
50.0 '"..L..''"""',,,~
0.0 2.0 4.0 6.0 8.0 10.0
Time [Sec.]
155
FIGURE 4.13 Analytical Model or the Bridge, Used ror Analysis
1$6
~~c
bearing element
soil mass
soil springs
(longltudmal and lateral)
stiff support
(a) Modeling of connection between bridge deck and abutment without rigid body
transformatIOn
deck modeled as
shear and ~arn
bearing
rigid arm
soil mass (rotational
~;~~~~~;~~~~~~~~~~~~~~tr:anslational)
soil spring
(longitudinal transverse
and rotational) stiff support
(b) Modeling of conq,,\,tion between bridge deck and abutment using rigid body
transformatIOn (ngld arms)
FlGURE 4.14 Modeling of Bridge Deck Connection to Abutment (a) Modeling witb Stiff
Elements (b) Modeling witb Rigid Body Transformation
157
10
 an$fis
ex ment
~
l'i
'Q 0
]
·5
·10
Time (Sec]
FIGURE 4.15 Comparison of Analytical and Experimental Results for Bridge on Steel
Bearing at Abutment.
~r~~~rrr~~
30
20 / experimental
E
.§.
~ 10
·100."'0':1.'::0......,2::\.0,........;3:\;.0:..............,.J4.0
Time (Sec]
FIGURE 4.16 Comparison of Analytical and Experimental Results for Bridge on Rubber
Bearing at Abutment.
158
4.1.3 Case Study #3: Bridge in Los Angeles· Retrofit Solution
A bridge model based on a prototype bridge in Los Angeles was evaluated to dem
onstrate the 3D capacity of the computational model to: (i) evaluate a bridge subjected to
differential ground motion and (ii) to evaluate the efficiency of a retrofit scheme using
Using a simplified evaluation technique, the bridge was found to have deficient .
The bridge (see Fig. 4.17 ) has seven spans, ranging between 30m to 39m in
length. The deck is supported on eight supports. Five of them (BlB5) are bents of the
type shown in Fig. 4.17 (b). Support Al is a very stiff abutment. Supports A2 and A3
are also very stiff but have some flexibility due to their flexible foundations. Abutment
AI, piers Bl, B2 and support A2 are skewed at 48°,54°,61° and 67° respectively.
The deck's vertical load is supported by concrete arches through ribs (see Fig.
4.17 b, Fig. 4.18 a). The horizontal forces from the deck are transferred through a wall
1.2m thick, at the support centerline (Fig. 4.17 b). The arches are supported by a shear
wall type bent (Fig. 4.17 b). The bents are supported on spread footings.
The bridge is modeled as a space frame system (see Fig. 4.19). The combined
section of the deck, the ribs and the arches is modeled using four equal longitudinal seg
ments of beam type elements in each span. The segments have different section properties,
relating to the arch depth. Since the ratio of length to width of the deck is low (1: 1.4 
159
1:2.0) shear deformations are considered. The bents are modeled as beam colunms, with
the adequate rigid body transformations, to account for stiff connection between the deck
and the bent. The foundations system is modeled as elastic springs, using information pro
A2 and A3 are modeled as stiff elements with flexible foundations, using the given soil
properties.
The soil conditions are varying along the bridge, from very stiff soil under bent B 1
Because of the low seismic capacity of the bents and foundations, a retrofit solu
tion using PTFE bearings system is considered. According to this solution the bents (B 1
B5) are cut at a level of 2m below the arches spring line. Eight isolators with elastomeric
restoring springs are inserted in the space created by the cut, elevating the deck on the iso
lation system. It is assumed that by isolating the deck, the force transferred to the bents
will be reduced, by transferring some of the horizontal forces to supports AlA3 while
also reducing the seismic demand, by shifting the bridge period and increasing the dissipa
The group of isolators was modeled as single isolator element for bents B3B5.
Several isolation elements were used however to model the isolation system of bents B 1
and B2, to be able to capture the behavior related to the skewed bents. The model of the
original bridge is presented in Fig. 4.19 and the model of the isolated bridge is presented
in Fig. 4.20.
160
(c) Spatial Ground Excitation
Since the soil conditions under the supports vary significantly, so does the ground
motion. The ground motion (Fig. 4.21) under the supports on the soft soil (Bents B2B5
and abutment A2) is amplified and it is much larger then the ground motion (Fig. 4.22)
In order to evaluate the seismic demand of the bridge, before and after retrofit, and
the influence of the variable ground motion, four different analysis cases were performed:
original bridge with variable and uniform ground motion and the same for the isolated
bridge.
and Z (transversal) directions. The period of the original bridge is about 0.2 sec (Fig. 4.23
and Fig. 4.28). The period is increased to about 0.4 sec for the isolated case (the bridge is
only isolated at bents BIB5 and at A2 in X direction). The responses in X and Z direc
tions are interacting trough the skew of bents B 1, B2. which leads to similar natural peri
while floating (sliding) over the other supports. In Z direction the bridge is bending like a
161
In X direction there is no significant reduction in shear forces in bents I and 5 due
to isolation solution (Fig. 4.27) for the case of uniform motion. The displacements, how
ever, are significantly larger in the isolated bridge due to change of structural system from
which are straight (not skewed). This translates into reduction of 50% in shear forces in
bent B5 (straight), and a reduction of only 20% in bent BI (skewed) (see Fig. 4.32  uni
form motion).
The most significant influence obtained using base isolation for this bridge, can be
observed in the case when differential motion is applied to the bridge. Supports AI, BI
and A3 experience smaller displacements than the other supports, which result in differen
tial motion between the supports. This differential motion introduce large stresses in the
bridge components. The bents of the original bridge subjected to differential motion have
significant deformation (relative movement between their top and bottom) in X direction
(Fig. 4.24), this results in significantly larger shear forces as compared to the "uniform"
ground motion case (Fig. 4.26). The isolated bridge however experiences much smaller
shear forces in X direction (Fig. 4.27) due to the "floating behavior" mentioned earlier.
The deformations of the piers are reduced at the expense of the relative deformations of
the isolators.
motion is very significant in Z direction. While the increase of shear forces between the
162
uniform and differential ground motion for the original bridge is of the order of ten (Fig.
4.31 ) the increase for the isolated case is of the order of 2.5 (Fig. 4.30). And the signifi
cant differences between the shear forces of the isolated and original bridge are shown in
Fig. 4.32.
The case study presented herein demonstrates the ability of the new computer plat
form to model a three dimensional problem, using condensed macro elements, including
triaxial isolation models (see also section 3.2.4), and soil springs (see also section 3.3.1).
The alternative models, i.e. planer models could not capture the influence of skewness of
supports, or general purpose three dimensional models such as ANSYS require extremely
The case study presented also the ability to determine influence of differential
ground motion.
The analytical results show that each of the modeling details, i.e. isolators, soils,
skewness, differential motion, has indeed substantial influence on the force and deforma
163
AI BI B2 A2
30 30 30 36
,
B3
39
,
B4
39
85
, 36
A3
,
1 "
"' "'
~,
\~ '} '~rJ ~lo7f ~  J: !L  ci
~
/j 8
ribs
arch
isolator
location '.hoI~'_ bent
... foundation
3.0
6.0
FIGURE 4.17 Bridge Top View and Typical Bent Cross Section
164
deck
21m
rib /
arch
inserted
isolator
21.0
165
J
•
".
•
• 1
,
,
•
!
,
•
•
•
166
,•
!
•
,o •
•
167
x direction
20.0
10.0
0.0
·10.0
·20.0 0
O. 5.0 10.0 15.0 20.0
z direction
40.0
30.0
..
~
§ 20.0
P. 10.0
.....'"
0
0.0
.10.0
0.0 5.0 10.0 15.0 20.0
vertical
10.0
0.0 \."""'"'
·1 0.00.':0'''':5"'0,......."'':1:!0'O::''''1::5'::0'''':::!20 0
. Time Lsec.]· .
168
x direction
40.0
30.0
20.0
10.0
0.0
10.0
20.0
30.0
40.0
50.0 .
00 5.0 10.0 15.0 20.0
z direction
60.0
50.0
40.0
'8
()
30.0
20.0
''
10.0
.....~ 0.0
Q
10.0
20.0
30.0 .
00 5.0 10.0 15.0 20.0
vertical
20.0
10.0
0.0
10.0
20.0 .
00 5.0 15.0 20.0
T'lme10[0sec. ]
169
A2 B5
0.10 0.10
0.05 0.05
0.00 0.00
0.05 0.05
~O.IO 0 0.10
2 4 6 8 10 0 2 4 6 8 10
B2 B4
0.10 0.10
0.05 0.05
.
§
~
l5
0.00 0.00
0.05 0.05
0.10 ·0.10
0 2 4 6 8 10 0 2 4 6 8 10
Bl B3
0.10 0.10
0.05 0.05
0.00 0.00
0.05 0.05
0.10 ·0.10
0 2 4 6 8 10 0 2 4 6 8 10
Time [Sec]
  isolated
  not isolated
FIGURE 4.23 Relative Deformations in X Direction, Between Ground and Deck Disp., at Piers
Centerline for Uniform Ground Motion
170
I •
A2 B5
0.20 ,r''.,r,,...., 0.20
O.lS O.IS
OW OW
O.OS O.OS
0.00 1_"""''''''''.... 0.00 L..J"'="'"
.().05 0.05
O.W CliO
O.lS O.lS
.(l.20 .(l.20
0.2S 0.2S
.(l.30 OLL2L....L.....J4L,.....J6,.1S.&iW .(l.30 OLL2L....L.....J4L,.....J6'.....JS'....IW
B2 B4
0.20 0.20 ,.,,.,..,...
0.15 0.15
O.W O.W
O.OS O.OS
0.00 0.00 i.J""""''''''''""
O.OS .(l.OS
0.10 .(l.lO
O.lS .(l.IS
0.20 0.20
0.25 0.25
Bl B3
0.20 0.20 '',....,.,,,.,
0.15 0.15
0.10 0.10
O.OS O.OS
0.00 0.00 1""""""'",
O.OS .(l.OS
0.10 0.10
O.lS .(loiS
0.20 .(l.20
.(l.25 0.2S
FIGURE 4_24 Relative Deformations in X Direction, Between Ground and Deck Disp., at Piers
Centerline for Variable Ground Motion
171
BI B5
above isolator
3000
2000
1000 1000
o !11m
1000
2000 2000
1000 1000
1000 1000
2000
bottom
3000
2000
1000 1000
1000 1000
2000 2000
FIGURE 4.25 Shear Forces in The Pier's X Direction, for Piers Bl and B5, for Isolated Bridge, at
Different Heights Along the Pier
172
A2 85
7000 7000
6000 6000
5000 5000
4000 4000
3000 3000
2000 2000
1000 1000
0 0
1000 1000
2000 2000
3000 3000
4000 4000
5000 5000
6000 6000
7000 7000
0 2 4 6 10
82 84
7000 7000
6000 6000
5000 5000
4000 4000
3000 3000
C 2000 2000
~ 1000 1000
~ 0 0
.E 1000 1000
~ 2000
3000
2000
3000
4000 4000
5000 5000
6000 6000
7000 7000
0 2 4 6 8 10 0 2 4 6 8 10
81 83
7000 7000
6000 6000
5000 5000
4000 4000
3000 3000
2000 2000
1000 1000
0 0
1000 1000
2000 2000
3000 3000
4000 4000
5000 5000
6000 6000
7000 7000
6 10
time (sec]
  differential motion
_. uniform motion
FIGURE 4.26 Shear Forces in X Direction, for Nonisolated Bridge, at Base of piers
173
uniform motion differential motion
B5 B5
4000 4000
3000 3000
2000 2000
1000 1000
0 0
1000 1000
2000 2000
3000 3000
4000 4000
'0 5000
g 0 2 4 6 8 10
5000
0 2 4 6 8 10
~
.2
~ 4000
BI
4000
BI
3000 3000
2000 2000
1000 1000
0 0
1000 1000
2000 2000
3000 3000
4000 4000
5000 5000
0 2 4 6 8 10 0 2 4 6 8 10
orne [sec]
 isolated
 not isolated
174
A2 B5
0.10 I I I 0.10
0.05 f  0.05
.k
0.00 0.00
"I
·0.05 f  ·0.05
I I I I
·0.10 0 10 ..().10 0
2 4 6 8 2 4 6 8 10
B2 B4
0.10 0.10
0.05 0.05
:§:
Q. 0.00 0.00
.:!l
Q
·0.05 .0.05
.0.10 10 .0.10 0
0 2 4 6 8 2 4 6 8 10
Bl B3
0.10 0.10
0.05 0.05
0.00 0.00
·0.05 0.05
.0.10 10 .0.10 0
0 2 4 6 8 2 4 6 8 10
Time [Sec)
  isolated
  not isolated
FIGURE 4.28 Relative Deformations in Z Direction, Between Ground and Deck Disp., at Piers
Centerline for Uniform Ground Motion
175
A2 B5
0.10 0.10
0.05 O.OS
0.00 0.00
0.05 0.05
0.10 10 0.10 0
0 2 4 6 8 2 4 6 8 10
B2 B4
0.10 0.10
0.05 0.05
:§:
0.00 0.00
~
0.05 0.05
0.10 0 10 0.10 0
2 4 6 8 2 4 6 8 10
Bl B3
0.10 0.10
0.05 0.05
0.00 0.00
0.05 0.05
0.10 10 0.10 0
0 2 4 6 8 2 4 6 8 10
Time [Sec]
  isolated
  not isolated
FIGURE 4.29 Relative Deformations in Z Direction, Between Ground and Deck Disp., at Piers
Centerline for Variable Ground Motion
176
bl b6
3000 3000
2000 2000
1000 1000
o 0
1000 1000
2000 2000
3000 L'L'~.L'L'_L'..J_ 3000 L..I..L,__L,...L,'L,.J
o 2 4 6 8 10
bottom bottom
6000 6000 '~'""""T.r~,~,r~,
5000 5000
4000 4000
3000 3000
2000 2000
1000 1000
o 0
1000 1000
2000 2000
3000 OLI1L:.L,..LL.l.113OOO0!,.J2l..JL4,.J6l.I!SL..JIO
timelsec)
  differential motion
.. __. uniform motion
FIGURE 4.30 Shear Forces in The Pier's Z Direction, for Piers Bl and B5, for Isolated Bridge,
at Different Points Along the Pier
177
                        
A2 BS
20000 20000
15000 IS000
10000 10000
SOOO SOOO
0 0 ~,.
SOOO 5000
10000 10000
15000 IS000
20000 20000
0 2 4 6 8 10 0 2 4 6 8 10
B2 B4
20000 20000
IS000 IS000
10000 10000
sooo
I8 SOOO
0 0
.E
~ SOOO SOOO
10000 10000
15000 IS000
20000 20000
0 2 4 6 8 10 0 2 4 6 8 10
B1 B3
20000 20000
IS000 IS000
10000 10000
SOOO SOOO
0 0
SOOO SOOO
10000 10000
IS000 IS000
20000 20000
0 2 4 6 8 10 0 2 4 6 8 10
time [sec]
  differentia1 motion
  umfonn motion
FIGURE 4.31 Shear Forces in The Pier's Z Direction, for Nonisolated Bridge, at Their Base
178
uniform motion differential motion
BS BS
4000 20000
3000 1S000
2000 10000
1000
·1000
·2000
·3000
~
4000
0 2 4 6 8 0 2 4 6 8 10
~
.2
BI
~
BI
4000 20000
3000 1S000
2000 10000
1000 sooo
1\'\
\
0 0
·2000
·3000
·4000
0 2 4 6 8 2
V 4 6 8 10
time (sec]
 isolated
 Dot isolated
FIGURE 4.32 Shear Forces in The Pier's Z Direction, for Piers Bl and BS
179
4.1.4 Case Study #4: Bridge over Los Angeles River ·Retrofit Solution with Base
Isolation and Restrained Expansion Joints.
A large size bridge modeled from a prototype bridge in Los Angeles with deficient
evaluate the efficiency of retrofit of a bridge using base isolation, when the deck has mul·
tiple expansion joints restrained, or unrestrained. The bridge suffered minor damage dur
ing past earthquakes, but using simplified methods, it was found that the seismic capacity
of the bridge is smaller then the predicted demand in future events. The original bridge
was analyzed and a selected retrofit scheme based on introduction of seismic isolation sys
Several techniques of retrofit were initially considered. Among them the conven
tional techniques of strengthening the piers with steel jackets. It was found that this tech
nique cannot provide a good retrofit solution to the problem, since this type of retrofit
increases the bearing demand on the foundations due to the additional weight, and
increase the seismic forces to a level of which the foundations will not be able to resist.
For this type of retrofit the foundations will have to be retrofitted also, at large cost, proba
bly prohibitive. On the other hand, retrofit using base isolation, doesn't change the capac
ity, but is expected to reduce the demand, and therefore increase the reliability of the
bridge.
The bridge, shown in Fig. 4.33a is a mUltiple concrete arches bridge with 12
spans of an average opening of 24.0m (between 17.5m and 35.0m) supported by bents of
changing height, of 8.0m to 19.0m. The bridge supports 4 traffic lanes, and has width of
about 20m, and height of O.5m. It is supported by 5 parallel arch girders in each span. The
180
original bridge has thermal expansion joints, 20 rnrn in size, in span B3B4 near bent B3
(shown in Fig. 4.33b), and similar joints near bents B6 and B9. The two abutments are
very rigid, and can be considered as rigid supports. The foundations of the bents, sup
ported by soft soil, are considered as flexible. The passive pressure on the piers below
For evaluation, the bridge is modeled as a space frame (see Fig. 4.34). In the cur
rent analysis the bridge is assumed to be elastic, and the nonlinear behavior is concen
trated, in the base isolation system and in the expansion joints only. The deck sections are
modeled as a single beam elements. Since the ratio of length to width is small, the shear
deformations of the beam should be considered. The deck elements are connected to the
bents, using rigid body transformations (coupled motions  see section 3.2.1). The expan
sion joints are modeled using several "gap" elements (see section 3.3.5), which are trans
formed from the bridge center line by using rigid body transformations (rigid arms see
section 3.2.1).
1) The bridge is fully isolated as shown in Fig. 4.35(a) friction (teflon and stain
less steel) devices are inserted in bents 3, 6 and 9 just under the bents crown, and
above ground level in the rest of the bents. At the abutments, a horizontal and vertical
cut is created and the isolation system is inserted between the bridge deck and the
abutment. Springs in the form of elastomeric springs are suggested to be inserted near
the sliding supports (see Fig. 4.35 (c». Those springs are not transferring any vertical
load because they are softer then the isolation system. The natural period of the iso
181
lated bridge was chosen to be T = 2.5 sec. The ratio between the initial vertical force
P initial on the isolator and the restoring spring stiffness kisolated is in this case:
= 4rt 2Pinitiai
k
isolated r g (EQ4.1)
The friction coefficient was assumed to change, from 5% at 0 velocity to 10% at veloc
ity of 0.1 mlsec and above. The isolation system is modeled using the triaxial isolation
2) The retrofit according to the second alternative, has a similar arrangement to the
first alternative, except that the connection between the deck and the abutments
remains rigid (not isolated) and can transfer forces without relative m~vement. This
type of solution is considered because of the significant cost involved in isolating the
abutment
The two retrofit schemes are compared to each other and to the original unretrofit
Since the deck of the bridge is sloped from the elevation of 8.3 m at the west (left)
abutment to 19.2 m at the east (right) abutment (Fig. 4.33a), the bridge is supported by
columns with changing height. The longer columns are much more flexible then the
shorter columns. Under seismic excitations, in the original bridge, there is transfer of force
";1'
through the deck from the long columns to the short ones. The transverse shape of the
deformed bridge at time 2.0 seconds is shown in Fig. 4.1 (a), shows that bents b7 b8 and
b9 experience the largest displacements while bent b I hardly moves. At the same time the
182
displacements of the retrofitted bridge according to alternative 1 (see Fig. 4.1(b» are
almost uniform along the bridge, showing that the deck is floating over the isolators. As a
consequence the forces of the isolated bridge, at the tall part of the bridge at bents b9 and
b 10 are up to three times smaller then the forces in the original bridge, as shown in Fig.
4.39.
The deformations of a typical frame can be observed for the original bridge in Fig.
4.37(a) and for the bridge retrofitted according to alternative I in Fig. 4.37(b). The defor
mations in the original bridge resulting from the bending of the piers, while in the isolated
case, the movement is mostly due to relative displacements in the isolation system and the
For the retrofit with alternative II, the behavior of the structure is predominantly
Although the displacements at the two ends are much smaller, as shown in Fig. 4.40.
Therefore the forces in the bents are close to the forces in alternative I of the isolated
bridge, except for the end bents, which transfer some of their force, directly to the abut
In both retrofit alternatives, there was significant reduction in shear force demand
especially in the long columns, and as a result reduction in bending moments, which
183
The modeling <!f the bridge included transversely distributed isolators and flexibil
ity of transversely distributed piers with soil springs influence. This modeling could be
achieved alternatively either with large general purpose computer codes i.e ANSYS,
ADINA, etc.) inefficient in carrying the computations, or with much simplified coeds like
(DRAIN 2DX, IDARC2D etc.) but with restrictive assumptions for isolators behavior
and spacial behavior of the overall structure. The solution was obtained in presence of
highly nonlinear expansion joints models (gap. elements) with stiffness changing from
zero to extremely high values. The solutions converged for long time records as shown in
184
+19.2
A
... ,
[ ....,
. I.!
'" "" 18,0 ,.... "., n.' ".,
l!1.~
'" 19.5
'" '"
a) Bridge Side View (As Built)
PI'JoInr
811' joint 'tt'B
...., ppJoint
r ....
'" ,.... 18.0
'" "., '" '" 19.5 n.'
'" '" ".,
• •
.j,!' ~
SecBB Sec AA
Section through joint Cross Section
elastomeric or sliding
bearings
SecCC
185
.....
"
186
+19.2
..., A
1
1 lI I ,
...
....,
'" ,,~
18.0 ,,~ ,,~
'" 35.0
GGGGG8G8GB8
a) Base Isolated Bridge With Expansion Joints
~.:~.:,~<
, ',,'
g g g
Sec AA
..., .,., +
r ....,
'" , . 18.0
"" ,,~
'" ".0
bridge deck
bridge pier
FIGURE 4_35 Base Isolated Bridge with and without Expansion Joints
187
~. ·. . ~·t.. ·4·......J,..........""·t';1
.,
,..
~~. }, . ·. . t:z"... ·~\
. . . .,
\1 \
\ . 
,"
\1
t=.
\ . . .i=
. .1\........ ~==t:::=".I
. ,
(a)
(b)
188
~~~" ,
I
I
I
I
I
(a)
(b)
FIGURE 4.37 Displaced Shape or Typical Frame In the (a) Original Bridge, (b) Isolated Bridge
189
  no gap
  not isolated
b2 bll
1000 1000
800 800
600 600
400 400
200 200
0 0
200 200
400 400
600 600
800 800
1000 1000
0 2 4 6 8 10 0 2 4 6 8 10
bl·n blO
100 1000
80 800
60 600
40 400
C 20
g 200
0
0 0
"
.s
"
0
Ii
20
40
200
400
60 600
80 800
100 1000
0 2 4 6 8 10 0 2 4 6 8 10
bloc b9
100 1000
80 800
60 600
40 400
20 200
0 0
20 200
40 400
60 600
80 800
100 1000
0 2 4 6 8 10 0 2 4 6 8 10
time {sec)
FIGURE 4.38 Comparison of Shear Force in Non Isolated Bridge, and Bridge Without Gap
190
  isolated
.... _ not isolated
b2 bll
1000 1000
SOO SOO
600 600
400 400
200 200
0 0
200 200
400 400
600 600
800 800
1000 1000
0 2 4 6 S 10 0 2 4 6 8 10
bl·n blO
100 1000
SO SOO
60 600
40 400
"g~ 20
0
200
0
O!
1 20
40
200
400
60 600
SO soo
100 1000
0 2 4 6 S 10 0 2 4 6 S 10
bloc b9
100 1000
SO SOO
60 600
40 400
20 200
0 0
20 200
40 400
60 600
SO 800
100 1000
0 2 4 6 8 10 0 2 4 6 8 10
time [sec)
FIGURE 4.39 Comparison of Shear Forces in Original and Retrofitted (AIt.1) Bridge
191
  isolated
  nogap
b2 bll
1000 1000
800 800
600 600
400 400
200 200
0 0
200 200
400 400
600 600
800 800
1000 1000
0 2 4 6 8 10 0 2 4 6 8 10
bID blO
100 1000
80 800
60 600
40 400
'0 20
g 200
~ 0 0
oS
~ 20 200
'Ii•
40 400
60 600
80 800
100 1000
0 2 4 6 8 10 0 2 4 6 8 10
bloc b9
100 1000
80 800
60 600
40 400
20 200
0
20 200
40 400
60 600
80 800
100 1000
0 2 4 6 8 10 0 2 4 6 8 10
time (sec)
FIGURE 4.40 Comparison of Shear Force In Isolated Bridge, and Bridge Withont Gap
192
 isolated
nogap b5 blO
0.10 0.10
0.05 0.05
0.00 0.00
0.05 0.05
·0.10 ·0.10
0.15 0 10..0,15
2 4 6 8 0 2 4 6 8 10
b4 b9
0.10 0.10
0.05 0.05
0.00 0.00
0.05 0.05
0.10 0.10
.0,15 100,15
0 2 4 6 8 0 2 4 6 8 10
b3 b8
0.10 0.10
0.05 0.05
S~
0.00 0.00
~ 0.05
._ 0.05
Q
0.10 0.10
·0.15 0 0.15
2 4 6 8 10 0 2 4 6 8 10
b2 b7
0.10 0.10
0.05 0.05
0.00 0.00
0.05 0.05
0.10 0.10
0.15 0 0.15
2 4 6 8 10 0 2 4 6 8 10
bI b6
0.10 0.10
0.05 0.05
0.00 0.00
0.05 0.05
0.10 0.10
0.15 100. 15 0
0 2 4 6 8 2 4 6 8 10
Time [Sec)
FIGURE 4.41 Comparison of Deck Displacements in Isolated Bridge, and Bridge Without Gap
193
 isolated
 Dot isolated b5 blO
0.10 0.10
0.05 0.05
0.00 0.00
0.05 0.05
0.10 0.10
0.15 10 0.15 0
0 2 4 6 8 2 4 6 8 10
b4 b9
0.10 0.\0
0.05 0.05
0.00 0.00
0.05 0.05
0.10 0.10
0.15 0 100.15 0
2 4 6 8 2 4 6 8 10
b3 b8
0.10 0.10
0.05 0.05
E 0.00 0.00
0.
~
0.05 0.05
is
0.10 0.10
0.15 ~
0 2 4 6 JO ·0.15 0 4
8 2 6 8 10
b2 b7
0.\0 0.10
0.05 0.05
0.00 0.00
0.05 0.05
0.10 0.10
0.15 100.15 0
0 2 4 6 8 2 4 6 8 10
bI b6
0.10 0.10
0.05 0.05
0.00 0.00
0.05 0.05
0.10 0.10
0.15 100.15 0
0 2 4 6 8 2 4 6 8 10
Time [Sec]
FIGURE 4.42 Comparison of Deck Displacements In Isolated Bridge, and Not Isolated Bridge
194
nogap
 not isolated b5 blO
0.10 0.10
0.05 0.05
0.00 0.00
0.05 .0.05
0.10 0.10
0.15 0 10 0.15 0
2 4 6 8 2 4 6 8 10
b4 b9
0.10 0.10
0.05 0.05
0.00 0.00
0.05 0.05
0.10 0.10
0.15 10 0.15 0
0 2 4 6 8 2 4 6 8 10
b3 b8
0.10 0.10
0.05 0.05
E 0.00 0.00
0.
0.05 0.05
is'" ·0.10 0.10
O.IS 0 100.15 0
2 4 6 8 2 4 6 8 10
b2 b7
0.10 , 0.10
0.05 0 0.05
0.00 ~
_L:I
V
L'
" 0.00
0.05 0 0.05
0.10 E
," 0.10
1 ,
0.15 0
8 10 0.15 0 2 4 6 8 10
2 4 6
bi b6
0.10 , I I I 0.10
0.00 0.00
0.05 ~ : 0.05
0.10 ~ : 0.10
I 1 1 1
..0.15 0 10 0.15 0
2 4 6 8 2 4 6 8 10
Time [Sec]
FIGURE 4.43 Comparison of Deck Displacements In Non Isolated Bridge, and Bridge without Gap
195
4.1.5 Case Study 5: Miyagawa Bridge
and inelastic behavior of base isolated bridges. The bridge was used as a pilot project for
base isolated bridges in Japan. Those bridges are called in Japan Menshin bridges. The
Menshin design is aimed toward increasing the energy dissipation capacity of the bridge
Bridl:e Description
The Miyagawa bridge is a three span continuous bridge, the total length of the
bridge is 105.5m, the width is lO.5m and it was erected across the Keta river as part of
National Highway No. 326. A view of the bridge is shown in Fig. 4.44 and a geometric
description is given in Fig. 4.45 (Buckle 1992). Lead rubber isolators were used to isolate
the deck from the abutment and piers in the longitudinal direction only; the bearings are
locked in the transverse direction. A geometrical description of the lead rubber bearings is .
shown in Fig. 4.46 for the abutment bearings and in Fig. 4.47 for the pier bearings. The
The cross section and reinforcement of the bridge piers is shown in Fig. 4.48. As
shown in Fig. 4.45 the cross section of the pier is changing along it's height and so is the
reinforcement as shown in Fig. 4.48. The force displacement curves used for the bearings
The structural model of the bridge used in the analysis is shown in Fig. 4.50. The
supporting pier was modeled by several sections in order to take into account the changing
stiffness and the changing mass along its height. The pier foundations are represented by
196
two infinitely stiff elements (elements 1,2,3 and 4) and the weight of the foundations of
300 tf is applied in between them (joints 5 and 6 in Fig. 4.50). The inelastic pier is split
into several elements (elements 5 to 11 and 6 to 12) for the following two reasons: (i)The
geometric section of the pier and the hysteretic properties are changing over the height of
the pier (ii) It enables the application of the mass of the column in intermediate locations
along the height of the pier. The isolator elements are located above the piers, and above
the abutment. The deck weight is concentrated at the joints between the deck and the piers
and between the deck and the abutment. The deck is assumed to act as an elastic beam.
The inelastic properties of the piers segments shown in Table 4.1 . Those values
were calculated from the section geometry, reinforcement, and the applied axial force.
(Fig. 4.50) MeCtm) cI> (11m) My Ctm) cI> (11m) MuCtm) cI> (11m)
5 819.75 0.9389E4 1509.50 0.1I29E2 1632.42 0.3005El
7 859.89 0.8832E4 1568.16 0.II09E2 1687.06 0.3207El
9 886.02 0.8279E4 1126.48 0.1057E2 1217.59 0.3939E1
II 1001.87 0.7714E4 1135.92 0.1029E2 1225.71 0.4402El
Two input ground motions were used for the analysis, as it is required in the Japa
nese design code. According to the Japanese code, it is required that the bridge remains in
the elastic range for a moderate earthquake described as level 1 design earthquake, and
survive a strong earthquake designated as the level 2 design earthquake (see Fig. 4.2)
197
without collapse. The ground conditions of the site is stiff soil which are under the cate
Two types of analysis were performed: (i) elastic analysis was performed using the
initial stiffness of the piers as presented in Table 4.1 . (ii) nonlinear analysis were the stiff
Since in the current study, the interest is concentrated in analysis of structures sub
jected to severe earthquakes, the presented results are for the level 2 earthquake only. The
the structure is subjected to large displacements at about 2.0 sec from the beginning of the
excitation, and the base of the pier undergoes large inelastic deformation in which the cur
 ~
vature is about 116 of the ultimate curvature, so that the structure can survive this excita
tion. The displacements of the deck are larger then the design displacements which control
the size of the required gap in the abutment, the relative displacements of the pier from the
analysis where about 230mm compare to 130mm which are the displacements calculated
In order to compare inelastic analysis, where the strength of the element is limited
to elastic analysis, where the strength of the element is infinite, both type of analysis were
performed on the isolated bridge. As mentioned earlier, the properties presented earlier
and marked on the bridge model, were used. The results of those analysis are presented in
198
The displacements of the isolated deck are hardly influenced, by the limited capac
ity analysis, but the displacements at the top of the bent under the isolator are much
The moments at the bottom ofthe pier are kept low in the case of the inelastic anal
ysis while for the case of the elastic analysis they are 50% higher.
..;:~?~
1" " '~.' '
, "
,
199
.. ..................
.... ............ ;:
, ., , ~
,
.' : '""f.. :
t ... ~

~~:'~.
"
! .... ,
.,
..... f
,
, .. "
~ ... ....r ......
, . •
! ...~ ....... ..Jll
' . ""'r.'
"'~:
.:(:.'.
.~.
'
. . .. ................. .....   l
, .
~ i":r .".. L_
L4l . Lf1
• Side View
At suppon
....
AI Center of span
 
too
,

•.Il00
AUI\" • ••
'"
C..c.,n ...,
1100
....,..". "''C.......
_If_
•
.
\!wh.
 I 1
,

'
,
j
: . ".:! .. '
,~,

,"
• • ,.oo_tUO UO
. ! ~ !
II
I
'10
I
I
."
Supersaucrure Bridge Pier
200
·AI
i' •
'1$ ~ •
IS 41+SUSlII
122
;'1
AA
J:. . ~;..~·.fA+:143:[,
:=J
1000
201
At Mid·height At Bottom
3100
2@300
600 10@300=SOOO 20@150=3000
.,...
  i r t      ,       .         H 
.
."
3989.5 20@150=3000
3100
4833 4000
D22
FIGURE 4.48 Cross Section or Plen
....""'" 70 • • '! r! ,
~ ~F  U curve'uled for dellgn
.. 50 ·····f·····r . ···;·····,····· ····:·····f·· .0; :
..... .0 ·····'·····1··· . ·;·····,,· . ··· .... ":...... ..: ...:.....
.~ 30 ·····i·····i·····i·····i·····
"
20 ·····'·····1·· .. •I..... j.....
. ....
.. !o.. .. ..··i·····
I......
..
o
10 ••••• i···•. i· ..
O.
·.i..... . .. .
} .........
::
i. ·.·i.·· ..
 • • • I •
10 ..... : .... : .......... .:..... . ..., ..... ; ..... : .... .
20 ••••• !•..... t ••• :.. • .0' •••• l .... ~.:.• .....•:. ..... :• .... .

..
• I
 30 ...•. !.. .
I t..
··1····· ····7·····;·····:·····!·····
•
II .0 ········1·
. ·i·· ::'s·····,····
·l·····:····..
tOO.
. ,,····· ····T·····:·····r·····!·····
.... .,. .....,.....,.....•.....

o so
H
:; JYSOI20eo
......
&0 .•.•• , ....... ,.................. •.•..;. ............; ...•• i .... .
~ 30 0 lIO ~ eo
.. .
120 ISO
:c Horizontal dhpllcelDent (IDIII) ,
202
t
+9.0 . = 1126t~
Y:
I
y = 1126t~
IS 16
8
+4.0 . M '=
Y'
co y = 1570t~
y: w=114t
5 6
+2.0
Y:
+1.0 . r ' 8
Y: , . 4
6 . . . .
• • • ' ••• J • •• ;•• ,..,326l' . . . . • . . , ,
+0.0 3 4
Y
5
kT = 2.21xlO tim
6
k, = 2.77xlO tmlTad
203
moment at bottom of foundation
2000
1000 I
itII
E
6
"c
0 ~~
rf" ~ ~
'11
~ f
IJ I tI A
IJV
j
~
1000 ,
2000
3000
o 5 10 15 20
Time (Se<;.)
  inelastic
elastl.c
moment at bottom of pier
2000
1000
E
6
"~
c 0
~k
rr
~h\ K~
V
IAt./\
pv
Q
::;

1000
~
2000
o 5 10 15 20
Time (Sec.)
204
Pier displacement above isolator
fI
0.05
I
!
fJ
.().l5
·0.25
o 5 to 15 20
Time (Sec.)
elastic
"'' inerastic
Pier displacements, under isolator
0.03
om
~A ~
!
"~
c
0.00
l!. Id ~I~ , W ~ 11'1111 IJ 1/1
~
~
,f\/\
11'1
1"
~ l' I I M
sis ·0.01
·0.02
HI
·0.03 0
5 10 15 20
lime (Sec.)
205
4.1.6 Case Study 6: Evaluation of Bridge Capacity Using Different Push· Over
Procedures and Dynamic Analysis
The concept of using pushover procedures for the entire bridge assembly instead
of using this procedure on each frame separately where the interaction between the frames
is neglected to evaluate the capacity of a bridge was presented in section 3.6.2. In this sec
tion a bridge that was subjected to the Northridge earthquake and experienced collapse
was evaluated using various pushover procedures: (1) uniform incremental static load. (2)
modal adaptable incremental load. (3) acceleration ramp input (4) pulse input. The
general plan and elevation of the bridge is shown in Fig. 4.53 , a typical cross section of
the box girder deck is shown in Fig. 4.54 and the geometry and reinforcement of the piers
The bridge was modeled using combination of elastic and nonlinear hysteretic
concrete beam column elements. The prestressed deck and the top part of the piers was
modeled as elastic beams. The bottom part of the piers were inelastic behavior is expected
was modeled as concrete hyste,etic beam column element. The piers are supported on
shafts. In order to consider the flexibility of the shafts, the height of the piers is increased
for the analysis the increased height is shown in Table 4.2 . The analytical model of the
bridge is shown in Fig. 4.56 . The heights of the piers and the elastic properties of all the
elements are presented in Table 4.2 . The moment  curvature relations and the shear
206
capacities of the nonlinear elements are presented in Table 4.3 . Since the current analysis
focus on ~he behavior of the bridge in the transverse direction, the properties related to the
behavior of the bridge in this direction are presented only. The stiff part at the top of the
TABLE 4.2. Geometry and Structural Elastic Properties of the Bridge Elements
Moment orinertla
Height (for
Element analysis) Area gross longitudinal transversal torsional
# m m2 m4 m4 m4
pier 2 . 13.4 4.2 0.49 4.23 1.74
pier 3 13.7 4.2 0.49 4.23 1.74
pier 4 17.5 4.2 0.49 4.23 1.74
pier 5 18.0 4.2 0.49 4.60 1.74
pier 6 29.2 3.7 0.85 
5.20 2.37

TABLE 4.3. Transversal Moment Curvature Relations and Shear Capacity of Piers
207
•
1582'.()·
/' < Ii
II~ II (I II i f »
c
"
ELEVATION
,. 53'..Q· II
T
~
d I. :I
AI
P4 P5
PLAN
P6 P7
~ P10
All
FIGURE 4,53 SR1411·5 Separation and Overhead (Southbound) • General Plan and Elevation
208
PRESTRESSED BOX GIRDER SECTION •
I REINPORCED BOX GIRDER SECTION
•
TYPICAL !ECTIONS
I
~
,;
•I
•I ~
r ~
,;
~ ~ 0.305 I
•
~ ]r
~
•I TYP. ~
~
Ill'
~
,; :I .,
CII'
   ~  
 1.118    
13.810 ,;
11.118
0 p5 •I O. 5
1.118 ..2.,892 2.743 . 1.372 ~ 1.372 2.743 2.743 1.118
0.152 •
0.102
18.154
DIMENSIONS IN 1m)
FIGURE 4.54 SR1411·5 Separation and Overhead (Southbound) • Typical Deck Section
209
".
tt· ,
  ..:~'!""4!":::'::.J
...", ..
FIGURE 4.55 SR1411.5 Separation and Overhead (Southbound) .Pler Sections and Relnrorcement.
210
FIGURE 4.56 SR141I·5 Separation and Overhead (Southhound) ·Structural Model for Analysis
211
Evaluation procedures
The bridge was subjected to several pushover procedures: (1) Modal adaptable:
static force distributed relative to the changing mode shapes due to change in the stiffness
of the elements when experiencing inelastic behavior. (2) Uniform pseudo acceleration:
the static incremental load is distributed relative to the mass (the distribution doesn't
change during the analysis). (3) Ramp loading (0.25 mlsec3, 0.5 mlsec3, 5.0 mlsec 3): lin
early increasing acceleration (Fig. 4.60 b, Fig. 4.61 b, Fig. 4.62 b), is applied in differ
ent rates to the bridge until failure is reached. (4)Pulse loading (0.1 sec, 0.05 sec): a short
duration acceleration pulse (Fig. 4.63 b, Fig. 4.64 b) with two different durations is
Santa Monica City Hall, increased to 150% was also used to analyze the bridge for earth
quake loading.
Results
The limit state of the bridge was defined by two conditions: (i) flexure failure,
where one of the piers, reached its ultimate flexture capacity (see Table 4.3 ). (ii) shear
failure, where one of the piers reached its shear capacity (see Table 4.3 ). The loading his
tory, the shape of the bridge at time of failure, and the history of shear forces in piers 2 to
7 is presented in Fig. 4.58 to Fig. 4.65 for the various analysis cases. The distribution of
the shear forces at failure is presented in Table 4.4 and graphically in Fig. 4.57 . The dis
of the bridge subjected to the various loading procedures are presented in Table 4.6 .
212
~ 
Results evaluation
Significant deformations were observed at the time of failure when the bridge was
analyzed for the first five pushover procedures (modal, uniform, three ramp loading pro
cedures). The failure mechanism for all these procedures was found to be shear domi
nated. However the moments at the time of failure in al these cases were very close to the
ultimate moment capacity. The failure occurred in all cases in short piers located in prox
imity to long piers (pier 3: 13.7m/pier4: 17.5m;pier5: 18m/pier 6, 7: 29m, 41m) due to
transfer of shear forces from the flexible support to the stiffer support (Fig. 4.57 ).
The external inertia forces on the bridge F can be related to its global acceleration
F = wiilolal (EQ4.2)
where w is the weight. The total acceleration can be divided into two parts: the ground
(EQ4.3)
The forces developed in the bridge in the case of modal adaptable analysis are related
directly to relative acceleration ii only. Higher forces are distributed to the softer parts of
the bridge and very low forces are applied to the stiffer parts.
The distribution of the forces in the cases of uniform and ramp loading is related to
the distribution of the ground acceleration iig' The distribution in all this cases is similar.
However, the force in pier 2 are higher by 30% in the case of steep ramp loading (5.0 mI
213
sec 3) then the other cases because of significant dynamic amplification for stiff elements.
F = ku+mii (EQ4.4)
The part of the force that control the failure is ku, therefore higher ground acceler
ation is required in the dynamic cases (Table 4.6 ) then in the static and low acceleration
cases.
The case of pulse loading provides very different behavior then the previous cases,
the failure is dominated by shear at the stiff piers, where most of the shear force is concen
trated (Fig. 4.57 ) which results in failure of pier 2. The maximum deformations are very
low (0.058m, 0.046m) and all the piers have almost the same displacements, because at
this stage only the inertia effect influence the response, while the effect of the piers stiff
ness was not developed yet. Very high ground acceleration is required in a very short
period of time to bring the bridge failure. The moments in the piers are relatively low at
A very impulsive ground motion was chosen as input to perform nonlinear time
history analysis. This kind of ground motion is characteristic to near fault seismic events.
The ground motion was increased to 150% to be able to bring the bridge to failure during
the first large pulse of the earthquake. The response of the bridge at failure is characterized
by a higher mode (see Fig. 4.65 a). The larger shear forces are concentrated in piers 2 and
3. relatively low maximum deformations (0.07m) are reached at failure. The failure is
reached before the ground acceleration reached its maximum, therefore the ground accel
214
eration at failure (O.35g) is lower than the maximum ground acceleration of the earth
quake.
Conclusion
It is evident that two different types of behavior were observed: The first one is
dominated by first mode response, which is typical to vibratory earthquake. The second is
kind of behavior can be represented relatively well by the pulse loading in which the mag
nitude of deformations and the concentration of shear force was close to the one observed
in the pulse pushover procedure. The shape of the bridge at the time of failure however
was different. It should be noted that the only procedure that was able to predict the actual
failure of pier 2 during the northridge earthquake is the pulse pushover procedure.
215
TABLE 4.4. Comparison of Shear Force Distrihution Between the Piers at the TIme of Fallure of
Critical Pier In Shear
;'l®ftotal•.
Ramp 0.5 mlsec 3
%bftotal
Ramp 5.0 mlsec 3
%oftotal
Pulse 0.1 sec
%of total
2160
%oftotal
Ground motion 4830 14242
(THA)
%oftotal
216
40
30 I 
20 I  Modal
10
r 
40
30 
20
 Uniform
10

....9'"
,......., 40
30 
.....0 20
 Ramp 0.25
10 
~
........ 40 I
I:

...
.....
0 30
20
I
I 
..6
..... 10 I 
Ramp 0.5
.....~
"0
40 I
~
30

~ 20 I 
.s 10 I 
Ramp 5.0
~ 40
I
~
..c:
'" 30 I 
20  Pulse 0.1
10
I
·
40
30
·
20
 Pulse 0.05
10 
40
30 l ·
I 50%
20
() THA
10 l ·
0
pI.r2 plcr3 P1cr4
FIGURE 4.57 Distribution of Shear Forces in Piers 27 for Different Analysis Procedures
217
TABLE 4.5. Moment at Bridge Piers due to Different Push·over Procedures at Shear FaUure
TABLE 4.6. Comparison of Performances of The SR141I5 Bridge Subjected to Different Push·
Over Procedures
Critical Maximum
Type of FaUure FaUedPier Acceleration Deformation
G m
Modal adaptable flexure/shear p5 0.21 0.590
Unifonn pseudo aeeel· flexure/shear p3 0.29 0.550
eration
Ramp 0.25 mlsee2 shear p5 0.34 0.470
218
I
!
; !/
) /'
p7
(a)
8000
/pS
6000
L
/ po
/' /.,... ..... p4

I'
/ ~ .,,/ I p3
./ V~
V I .... 'p7
 /" ....
)~
o
p2
·2000
Step number [#]
(c)
FIGURE 4.58 Response of the Bridge to Adaptable Proportional to Instantaneous Mode Shape
Distributed Load (relative to mass). (a) Displaced Shape at Failure, (b) Loading History, (c) Shear
History for Piers 27
219
force 40000
liiiiiiiil ~3~
j
".' .... /
~~
p2 p3
~
FI~
(b)
8~ ~'rT~T~~'
rooo ~+++4~~~~~~~
2
"~ 4000
<E p4
t;;
.c " p7
'"
10 20 30 40
Loading step [#]
(c)
FIGURE 4.59 Response of the Bridge to Non Adaptable Uniform Distributed Load (relative to
mass). (a) Displaced Shape at FaUure, (b) Loading History, (c) Shear History for Piers 27
220
p6 p7
p4 P~.<····· ".
p3
~y2. ,
Time [sec]
(b)
(a)
8000 r,.,,,..r,,
p3
p5
rooo~~~44~~~4
~
"~~r__+~~~~_++_~~
,£
R
2000 i+T:;,f:::;;o"'3.....""++I
p7
Time [sec.]
(c)
FIGURE 4.60 RespoDSe oftbe Bridge to Ramp Loading. Acceleration Rate: o.lSmlsecl
(8) Displaced Shape at FaDure, (b) Loading HIstory, (c) Shear HIstory for Piers 2·7
221
4r__~_,
p6 p7
p4 }.S/
i' .. ~~...,::..._~3~..._,_,,,,_',''~
o~~~~~
o 5 10
Time [sec)
(b)
(a)
8000.0
p3
6000.0 ~ pS
j V
~ t?'
~
~ ~V 1/"""..... '/'" p6
2000.0
ibv k? ~ ./'
V/~ ~ . 10 p7
0.0
~~  Time
(c)
FIGURE 4.61 Response of the Bridge to Ramp Loading. Acceleration Rate: 0.5 mfsec 3 (a) Displaced
Shape at Failure, (b) Loading History, (c) Shear History for Piers 27
222
o~~~~~~~
0.0 0.5 1.0 1.5
Time [sec]
(b)
(a)
8000
p3
6000 I' 7ps
/ V/ , nO
VJ"'"V / l7
/ ~1// V p4
p6
2000
//J V/
VI/ V,.
~ p7
io"'"
o0.0
b ~Z ~
FIGURE 4.62 Response of the Bridge to Ramp Loading. Acceleration Rate: 5m1se~. (a) Displaced
Shape at Failure, (h) Loading History, (c) Shear History for Piers 2·7
223
15 ...._ ........_ _......._ _..._......,
14
13
12
II
p2 p3 p4 p5 p6 p7 _ 10
9
i
N
" ....
/) 8
 7
~ 6
'" 45
3
2
I
(a) O~~~~~~~~~
0.00 0.05 0.10 0.15 0.20
Time [Sec.]
(b)
8000
p2
6000 /
7 _0
1/ /
/ V
2000
/ v:
~
"'
p4
n<
p7
F'" .......
%.)0 0.05
TIme [sec.]
v. IV
" p6
(c)
F1GURE 4.63 Response of the Bridge to Pulse Loading 0.1 sec. (a) Displaced Shape at Failure, (b)
Loading History, (c) Shear History for Piers 2·7
224
24
22
p2 p3 p4 p5 p6 p7
...... ,.".. ". ,
.. 20f.
: 18
~ 16
!14
] 12
~ T <10
8
6
4
2
(a) 8~~"~,,,~~u~
. ,"~~,,~,,,~~~,,.0
Time (Sec.)
(b)
8000
pI
6000
p2
~
"1: 4000
.s
~
p5
'""
..c
'"
p4
2000 p3
p6
O~~~~~~~~~~~~~~~~~
0.00 0.Q2 0.04 0.06 0.08 0.10
time [sec.]
(c)
FIGURE 4.64 Response of the Bridge to Pulse Loading 0.05 sec. (a) Displaced Shape at FaHure, (b)
Loading History, (c) Shear History for Piers 2·7
225
1.2
0.9
0.6
i .'
..
\,: ... ;' /p7 0.6
p6
0.9
(a) 1.2
time sec.]
(b)
10000
8000
_.
6000
. I!M
4000
~
'" 0
~\I1fJI\ I!~
p6
2000 v
\j
4000
\
6000
7
Time [sec.l
(c)
FIGURE 4.65 Response of the Bridge to Dynamic Impnlsive Loading. (a) Displaced Shape at Failure,
(b) Northridge Earthquake Record From Santa Monica City HaIl (input motion), (c) Shear History
for Piers 27
226
5.0 DISCUSSIONS AND CONCLUDING REMARKS
• The current work, laid up the ground work for the analysis of bridges were the nonlin
ear behavior of elements (concrete or steel), and protective systems is considered. Sev
eral types of analysis, such as time history, monotonic pushover and modal were
I
introduc~ to extract information about the safety of new design or retrofit design of
bridges. Since every one of the analysis methods has strong and weak points, a good
evaluation procedure should be combined of the different analysis types presented. The
time history analysis can be used as indicator to the global behavior of the bridge dur
ing severe earthquake, and approximate procedures such as pushover analysis, and
• Several case studies were presented in the current work, which were used to demon
The special end connections and the coupled motions were used in almost all the
case studies and proved to provide very efficient tools to model complicated connec
tions between the deck and the substructure. This contribution is especially apparent in
the modeling of the East Aurora bridge in which very simple and efficient modeling of
the connection was used as compare to the very sophisticated modeling used for the
The modeling of base isolation systems placed in arbitrary locations in the bridge
was verified with experimental results. It was observed however, that care should be
taken when using stiffness proportional damping in structural systems which include
227
The modeling of the triaxial behavior which includes dependency of the friction
coefficient on velocity of sliding isolator, was modeled by the new triaxial isolator ele
The expansion joint element was developed to model efficiently the behavior of
The nonlinear damping concept is introduced and used for the modeling of the tri
axial element. The modeling of this element is verified using biaxial experiment as
mentioned earlier.
The use of displacements / velocity input, which allow direct representation of dif
ferential ground motion was introduced. The use of this method was verified by com
All the models mentioned above which include: special connections, different
types of base isolation, expansion joints, differential ground motion and nonlinear
damping were used in analysis of large scale real bridges. Parametric studies were per
• A first step in the direction of developing evaluation methods based on time history
nonlinear procedures which include damage indexing combined with pushover analy
sis was made. A bridge in Los Angeles county was subjected to several types of push
over procedures as well as real ground motion exciton. The influence of the type of the
earthquake excitation on the distribution of forces in the bridge was studied using sev
eral pushover procedures and compared to the response of the bridge to actual impul
228
 ,
The modal adaptable push over procedure seems to be deficient in modeling the
bridge response. The results from the uniform force (relative to mass distribution) and
the slower ramp loading are similar in nature, and seems to provide reasonable repre
sentation of the response of the bridge to vibratory earthquake. The ramp push over
procedure, with faster application rate introduce dynamic effect to the distribution of
the forces along the bridge. Larger part of the seismic load is transferred to stiffer and
shorter piers. Because the dynamic effect results in additional inertia forces, larger
acceleration is required to bring the bridge to failure than the static and slower proce
dures.
bridge in different manner than the procedures related to vibratory earthquakes. Large
transfer of forces to the stiffer and shorter piers can be observed while experiencing
The actual failure mechanism of the bridge during the Northridge earthquake was
• The program provides basic approach based on the idea that the modeling of a struc
tural system should be done as much as possible considering physical models there fore
the nonlinear behavior of elements will be mostly modeled by nonlinear stiffness and
damping properties, and not by defining the nonlinear forces to the forcing vector. This
type of element is the main prototype element in the program IDARCBRIDGE. Other
programs (ANSYS, DRAIN2DX) are using only the stiffness properties and the force
229
• A close form solution of the governing differential equation is used to represent the pla
nar behavior of base isolation systems. This solution gives insight and understanding
on how the system works and explain why a threedimensional additional element was
necessary to be developed.
• Several techniques which were developed previously such as: rigid end connections,
flexible connections, and free connection (Weaver and Gear 1990) and coupled motions
(Bathe (1982» were modified to be more versatile such that they can be used each one
• A new triaxial base isolation element that can be incorporated in a general computer
program was developed. This element allows for a more realistic modeling of sliding
base isolation system and can be easily extended to model other types of isolation sys
• A modern iterative solver was incorporated in the program to provide faster and more
efficient solution of the equilibrium equations. A solver for large eigenvalue problems
• The procedure developed by Priestley et. al. (1992) for the use of Lateral Strength
Method, was expanded. The original method provides solution to a simple planar
frames. The current work extends the idea to model the whole bridge system as an inte
gral unit. Several patterns of loading were developed to enhance the efficiency of this
method.
230
5.1 Further Research Suggestions
In order to develop further the use of the evaluation procedure, additional verifica
tion studies should be performed. The results of the different analysis procedures should
be compared with experimental results, with measured response of bridges which experi
enced real earthquakes and with analysis results from other programs, which were verified
The concept of bridge design using the combined evaluation method should be
evaluated by using case studies on existing bridges, and comparing the design with the
The current evaluation methodology can be used to develop simplified design pro
base isolation systems on the behavior of bridge structures, especially the size of the
device and the influence of adding damping devices, with variety of locations in the
bridge.
The influence of gap elements on the seismic behavior of the bridge should be
studied and the gap element should be further verified in the program IDAReBRIDGE.
The efficiency of the analysis program should be evaluated and if possible the
numerical procedures should be upgraded, to accelerate the speed of the analysis, using
231
The influence of biaxial interaction for concrete elements and protective systems
A very important issue that should be considered is the influence of the limitations
on input properties of the elements, both linear and nonlinear, on the response of the
bridge system. The procedure of evaluation of this part, should be done using initially,
bridges with a small number of elements. Then extend it to more complex systems, and try
to find out if there are limiting values that can be used for design. The verifications and the
time history as a primary evaluation tool. Although it is a gloryless large afford, this step
should be taken to depart from the elastic equivalent design procedure to a more rigorous
design method. The tools developed in this work can play ar roll in such an afford.
Additional structural elements can be incorporated into the program to enlarge the
elements library of the program, and provide more flexibility and accuracy in modeling
bridge structures. The program can be used as a tool, for system identification, and for
232
6.0 RFERENCES
"Bridge retrofit Procedures.", Caltrans Interim Memo to Designers, 204, April 1992.
"Code of Practice for the Design of Concrete Structures.", NSZ 3101:1982, Standards As
sociation of New Zealand, 1982
"Equivalent Static Analysis of Restrainers. ",Caltrans Bridge Design Aids, 1411 to 1425,
October 1989.
"Restrainers at Hinges and Bearings," Caltrans Memo to Designers 203 November 1989.
"Tentative Provisions for the Development of Seismic Regulations for Buildings", Applied
Technology Council, Report ATC306, Redwood City, June, 1978.
"Standard Specifications for Highway Bridges.", 15th Ed., American Association of State
Highway and Transportation Officials, 1992, Including Interim Specifications for 1993,
1994, and 1995.
Aoyama, H., and Sugano, T., (1968), "A Generalized Inelastic Analysis of Reinforced Con
crete Structures Based on the Tests of Members", Recent researches of structural mechan
ics. Contribution in honor of the 60th Birthday of Professor Y. Tsuboi, UnoShote, Tokyo,
pp. 1530.
Arya, A.S. (1984), "Sliding Concept for Mitigation of Earthquake Disaster of Masonry
Buildings." , 8tyh World Conf. On Earthquake Eng., San Francisco, 5, 951958.
Asher, J.W. et al. (1990), "Seismic Isolation Design of the USC University Hospital." Proc.
of Fourth U.S. National Conf. on Earthquake Eng., California, Vol. 3, 529538.
Aydinoglu, M.N, and Erdik, M., (1995), "17 January 1995 Hyogoken Nanbu (Kobe)
Earthquake Reconnaissance and Assessment Report", Kandilli Observatory and Earth
quake Research Institute, BebekIstanbul, Turkey, 1995
Bagley, R.L. and Torvik, P.J. (1983). "Fractional Calculus  A Different Approach to the
233
Analysis of Viscoelastic Damped Structures", AIAA Jouranl, 21(5), pp. 741748 .
Banon, H., and Vneziano, D. (1982), "Seismic Safety of Reinforced Concrete Members and
Structures." Earthquake Engineering and Structural Dynamics, 10, 179193.,
Banon, H., Biggs, J., and Irvine, H. (1981), "Seismic Damage in Reinforced Concrete
Frames.", J.Struct.Eng., ASCE, 107(ST9), 17131728.
Bathe, K J. and Gracewski, S., (1981), "On Nonlinear Dynamic Analysis Using Substruc
turing and Mode Superposition", Computers and Structures, Vol. 13, pp. 699707, 1981.
Betti, R, AbdelGhaffar, AM. and Niazy, AS., (1993), "Kinematic SoilStructure Interac
tion for LongSpan CableSupported Bridges", Earthquake Engineering and Structural Dy
namics, Vol. 22, pp. 415430.
Bhatti, M.A., Pister, K S. and Polak, E. (1981). "Optimization of Control Devices in Base
Isolation Systems for Aseismic Design". Structural control: proceedings of the Intemation
aIlUTAM Symposium on Structural Control held at the University of Waterloo, Ontario,
Canada, June 47,1979. pp. 127138
Boardman, P.R, Wood, B.J. and Carr, AJ. (1983). "Union House  A Cross Braced struc
tures with Energy Dissipaters.", Bull. of the New Zealand National Society for Earthquake
Eng., 16(2),8397.
Bouc, R, (1971), "Model Mathematique D'hysteresis", (in French), Acustica v.24, pp. 16
25
Bracci, 1. M., Reinhorn, AM., Mander, J.B., and Kunnath, S. K (1989), "Deterministic
model for seismic damage evaluation of reinforced concrete structures." Rep. No. NCEER
890033, National Center for Earthquake Engineering Research, State University of New
York, Buffalo, NY.
Buckle, KE. and Kelly, J.M. (1985). "Equivalent Linearizations for Practical Hysteretic
Systems." Int. J. NonLinear Mech., 23(4), 238.
234
Buckle, lG., Mayes, RL. and Button, M.R,(1987) "Seismic Design and Retrofit Manual
for Highway Bridges.", Report FHWAIP876, Federal Highway Administration, pp.290
Buckle, I.G. and Mayes, RL (1990). "Seismic Isolation: History, Application, and perfor
mance  A world Overview.", Earthquake Spectra, 6(2),161202.
Buckle, lG.,(1991) "Screening Procedures for the Retrofit of Bridges," Proceedings of the
Third U.S. Conference on Lifelines Earthquake Engineering, ASCE, Monograph No.4,
Technical Council on Lifeline Earthquake Engineering, 1991, pp. 156165.
Buckle, lG. editor (1992), "Proceedings from the First U.S.  Japan Workshop on Earth
quake Protective Systems for Bridges", Report No. NCEER920004, National Center for
Earthquake Engineering, State University of New York, Buffalo, New York.
Buckle, lG, (1994), "The Northridge, California Earthquake of January 17, 1994: Perfor
mance of Highway Bridges", Report No. NCEER940008, National Center for Earthquake
Engineering, State University of New York, Buffalo, New York
Buckle, lG. (ed) (1994), "The Northridge California Earthquake ofJanuary 17, 1994, Per
formance of Highway Bridges." Technical Report 940008, National Center for Earthquake
Engineering Research, Buffalo, NY, 1994.
Buckle, lG., and Freidland, lM.,(1995) "Improved Screening Procedures for the Seismic
Retrofitting of Highway Bridges.", Fourth International Bridge Engineering Conference,
Transportation Research Board, San Francisco, 1995.
Buckle, I. G. and Friedland, 1 M. (editors) (1995), "Seismic Retrofitting Manual for High
way Bridges" McLean, Va.
Built, S.M. (1982). "LeadRubber Dissipaters for the Base Isolation of Bridges Struc
tures.", report No. 289, Dept. of Civil Eng., University of Auckland, New Zealand.
Caspe, M.S. and Reinhorn. A.M. (1986), "The Earthquake Barrier, A Solution for Adding
Ductility to Otherwise brittle Buildings." Proc. of ATC17 Seminar on Base Isolation and
Passive Energy Dissipation, Applied Technology Council, Palo Alto, Calif. 331342.
Celebi, M. and Penzien, J., (1973). "Experimental Investigation into the Seismic Behavior
of Critical Region of RIC components as Influenced by Moment and Shear.", Report No.
EERC 734, University of California, Berkeley.
Chalhoub, M.S. and Kelly, J.M. (1990). "Sliders and Tension Controlled Reinforced Elas
tormeric Bearings Combined for Earthquake Isolation.", Earthquake Eng. and Struct. Dyn.,
19(3). 333344.
235
Chang, C.J, and Mohraz, B.,(1990), "Modal Analysis of Nonlinear Systems with Classical
and NonClassical Damping", Computers and Structures, Vol. 36, No.6, pp. 10671080.
Chang, G. A and Mander, J.B.(1994a), "Seismic Energy Based Fatigue damage Analysis
of Bridge Columns, PART I, EVALUATION OF SEISMIC CAPACITY.", Rep. No.
NCEER940013, National Center for Earthquake engineering Research, State University
of New York, Buffalo, NY.
Chang, G. A and Mander, J.B.(1994b), "Seismic Energy Based Fatigue damage Analysis
of Bridge Columns, Part II: Evaluation of Seismic Demand.", Rep. No. NCEER940006,
National Center for Earthquake engineering Research, State University of New York, Buf
falo,NY.
Charleson, AW., Wright, P.D. and Skinner, RI. (1987), "Wallington Central Police Sta
tion Base Isolation of and Essential Facility.", Pacific Conf. on Earthquake Eng., New
Zealand 377388.
Chen, C.C and Robinson, AR, (1993), "Improved Time History Analysis for Structural
Dynamics. I: Treatment of Rapid Variation of Excitation and Material Nonlinearity", Jour
nal of Engineering Mechanics, Vol. 119, No. 12, pp. 24962513
Cheng, M. (1988). "Convergence and Stability of Step by Step Integration For Model with
Negative Stiffness."Earthquake Eng. Struct. Dyn., 16(1),227244.
Chopra, A.K., Clough, D.P., and Clough, RW. (1973). "Earthquake Resistance of Build
ings with a Soft First Story." Earthquake Eng. and Struct. Dyn., 1(4),347355.
Chung, Y. S., Meyer, C. and Shinozuka, M. (1987), "Seismic Damage Assessment ofRe
inforced Concrete Members." Rep. No. NCEER870022, National Center for Earthquake
Engineering Research, State University of New York, Buffalo, NY.
Clough, RW., and Johnston, S.B., (1966), "Effect of Stiffness Degradation on Earthquake
Docility Requirements." Proceedings, Japan Earthquake Engineering Symposium, Tokyo,
Japan, Oct. 1966, pp.195198.
Clough, RW. and Penzien, J., (1993), "Dynamics of structures", McGrawHill, inc.
Cofer, W.F., MaLean, D.1. and McGuiere, J.W., (1994), "Analytical Modeling of Founda
tions for Seismic Analysis of Bridges", Technical report number WARD 328.2, Washing
ton State Department of Transportation.
Collins, M.P. and Mitchell, D. (1991), "Prestressed Concrete Structures." Prantice Hall,
Englewood Cliffs, NY.
236
Base Isolation Systems," Thesis Presented to the Rensselear Polytechnic Institute, New
York, in partial fulfillment of the degree of the Doctor of Philosophy.
Constantinou, M.C (1987). " Simplified Analysis Procedure for Base Isolated Structures on
Flexible Foundations." Earthquake Eng. and Struct. Dyn., 15(8),963983.
Constantinou, M.C., Mokha, A. and Reinhom, A.M. (1990a). "Experimental and Analyti
cal Study of a Combined Sliding Disc Bearing and Helical Steel Spring Isolation System."
Report no. NCEER9232, National Center for Earthquake Engineering, State Univ. of
New York, Buffalo, New York
Constantinou, M.C., Mokha, A. and Reinhom, A.M. (1990b). "Teflon Bearings in Base
Isolation IT: Modeling." J. Struct. Eng. ASCE, 116(2),455474.
Darbre, G.R. and Wolf, J.P. (1988), "Criterion of Stability and hnplementation Issues of
Hybrid FrequencyTime Domain Procedure for Nonlinear Dynamic Analysis." Earth
quake eng. and Struct. Dyn., 16(4),569582.
Darwin, D., and Nmai, C. K. (1986), "Energy Dissipation in RC Beams Under Cyclic
Load.",1. Struct. Eng., ASCE, 112(8), 18291846.
Den Hartog, J.P.(1931), "Forced Vibration with Combined Coulomb and Viscous Fric
tion." Transactions of ASME, APM539, 107115.
237
Der Kiureghian, A. (1995), " A Coherency Model for Spatially Varying Ground Motion.",
Earthquake Engineering and Structural Dynamic, in press.
Dynamic Isolation Systems, Inc. (1983). "Seismic Base Isolation Using LeadRubber Bear
ings.", berkeley, Calif.
Elms, D.G., and Martin, G.R., (1979)"Factors Involved in the Seismic Design of Bridge
Abutments.", Proceedings of a Workshop on Earthquake Resistance of Highway bridges,
Applied Technology Council, Report No. ATC61, November, 1979.
Emori, K. and Schnobrich, W.C. (1978), "Analysis of RIC FrameWall Structures for
Strong Motion Earthquakes", Civil Engineering Studies, SRS 457, University oflllinois,
Urbana, December 1978.
Fajfar, P. and Vidic Tomaz, (1994), "Consistent Inelastic Design Spectra: Hysteretic and
Input Energy", Earthquake Engineering and Structural Dynamics, Vol. 23, 523537.
Fang, H. V., (1991), "Foundation Engineering Handbook" 2nd edition, Van Nostrand Rein
hold.
Fujita, T., Suzuki, S. and Fujita, S. (1989), " Hysteretic Restoring Force Characteristics of
High Damping Rubber Bearings for Seismic Isolation." ASME Pressure Vessels and Pip
ing Conference, PVPVol. 1981, Hawaii, 2328.
Fung, G.G., LeBeau, R.J., Klein, E.E, Belvedere, J., and Goldschmidt, A.F, (1971), "Field
Investigation of Bridge Damage in the San Fernando Earthquake, State of California, Di
vision of Highways, Bridge Department, 1971.
Gear, C.W. (1971), "The Automatic Integration of Ordinary Differential Equations." Nu
merical Mathematics, Communications of ACM, 14(30, 176190.
Ghobarah, A. and Ali, H.M, (1988), "Seismic Performance of Highway Bridges", Journal
of Engineering Structures, Vol. 10, July. 157166.
Gluck, N., Reinhorn, A.M., Gluck, 1. and Levy, R., (1995), "Design of Supplemental
Dampers for Control of Structures", Submitted to ASCE Journal of Structural Engineering,
March, 1995.
Goltz, G.D. (1994), "The northridge, California Earthquake of January 17, 1994: General
Reconnaissance Report.", Report No. NCEER940005, National Center for Earthquake
Engineering, State University of New York, Buffalo, New York.
Gulkan, P., and Sozen, M.A., (1974),"Inelastic Responses of Reinforced Concrete Struc
tures to Earthquake Motions.", ACI Journal, Proc. 71(12): 604610, December, 1974.
Hisano, M. et al. (1988), "Study on a Slidingtype Base Isolation System: TriAxial Shak
ing Table Test and its Simulation." Proc. of the Ninth World Conf. on Earthquake Eng., Ja
238
pan, V, 741746.
Huffmann, G.K,(1986), "Full Base Isolation for Earthquake Protection by Helical Springs
and Viscose dampers.", Nuclear eng. and Design, 84, 331338.
Ikonomou, AS. (1985), "Alexismon Isolation Engineering for Nuclear Power Plants." Nu
clear Eng. Design, 85, p201
Imbsen, R.A, Nutt, R.V., and Penzien, 1.,(1978), "Seismic Response of Bridges Case
Studies", Report No, UBCIEERC 78/14, Earthquake Engineering Research Center, Uni
versity of California, Berkeley, June, 1978.
Imbsen, R.A, Lea, 1., Kaliakin, V., Perano, K, Gates, J., and Perano, S.,(1982), "SESI
SABI User Manual, Engineering Computer Corporation, October, 1982
Iwasaki, T., Penzien, 1., and Clough, R., (1972) "Literature Survey  Seismic Effects on
Highway Bridges", Report No. 7211, Earthquake Engineering Research Center, Universi
ty of California, Berkeley, November, 1972.
Jagdish, KS., Raghuprasad, B.K and Rao, P.V.(1979). "The inelastic Vibration Absorber
Subjected to Earthquake Ground Motion." Earthquake Eng. and Struct. Dyn., 7(4),317
326.
Kan, C.L. and Chopra, A.K.( 1977), "Elastic Earthquake Analysis of Torsionally Coupled
Multistory Building.", Earthquake Eng. and Struct. Dyn., 5(4), 395412.
Kannan, AM. and Powell, G.H. (1975), "DRAIN2D as General purpose Computer Pro
gram for Dynamic Analysis of Inelastic Plane Structures with Users Guide.", report No.
UCBIEERC 73/22, Earthquake Engineering Research Center, University of California,
Berkeley, Calif.
239
Kasai, K., Munshi, J.A., Lai, M.L. and Maison, B.F., (1993), "Visocoelastic Damper Hys
teretic Model: Theory, Experimental and Application.", Proceedings of Seismic Isolation,
Passive Energy and Active Control, V.2, ATCI71, San Francisco, Ca, pp. 521532.
Kelly, J.M., and Beucke, K.E,(1983). "A Friction Damped Base Isolation System with Fail
Safe Characteristics." Earthquake Eng. and Struct. Dyn., 11 (1), 3356.
Kelly, J.M., Buckle, I.G. and Tsai, H.C. (1986a), "Earthquake Simulator Testing of a Base
Isolated Bridge Deck." Report no. UCBIEERC85/09, Earthquake Engineering. Research
Center, University of California, Berkeley, Calif.
Kelly, J.M. (1986b), "Aseismic Base Isolation: Review and Bibliography." Soil Dyn. and
Earthquake Eng., 5(4), p 202
Kincaid, D.R., Respess, J.R Young, D.M. and Grimes, RG.,(1990), "ITPACK 2C: A For
tran Package for Solving Large Sparse LInear Systems by Adaptive Accelerated Iterative
Methods", (downloaded from netlib)
Kunnath, S.K., Reinhorn, A.M., and Park, YJ., (1990), "Analytical Modeling of Inelastic
Seismic Response of RIC Structures", Journal of Structural Engineering, ASCE, 116,4,
9961017.
Kunnath, S.K., Reinhorn, A.M. and Lobo, RF. (1992), "IDARC version 3.0: A Program
for the Inelastic Damage Analysis of Reinforced Concrete structures. "Rep. No. NCEER
920022, National Center for Earthquake Engineering Research, State University of New
York, Buffalo, NY.
Kwan A.K., (1993), "Improved WideColunmFrame Analogy for Shear/Core Wall Anal
ysis", Journal of Structural Engineering, Vo!. 119 No.2 February 1993.
Lam, I. P., Martin, G. R, Imbsen, A.R "Modeling Bridge Foundation for Seismic Design
and Retrofitting" Third Bridge Engineering Conference; Denver, March 1013, 1991.
Transportation Research Board, Washington, volume 2, pp. 113126.
Lee, D.M. (1980), "Base Isolation for Torsion Reduction in Asymmetric Structures Under
Earthquake Loading." Earthquake Eng. Struct. Dyn., 8(3), pp.349359.
Leger, P. and Dussalut, S., (1992), "Seismic Energy Dissipation in MDOF Structures",
Journal of Structural Engineering, Vo!. 118, No.5, pp. 12511269.
240
Li, C. and Reinhorn, AM., (1995)"Experimental and Analytical Investigation of Seismic
Retrofit of Structures with Supplemental Damping: Part II  Friction Devices.", Technical
Report NCEER950009, National Center for Earthquake Engineering Research, Buffalo,
NY.
Liu, W.D., Ricles, J.M., Imbsen, R.A, Priestley, MJ.N., Seible, F., Nobari, F.S. and Yang,
R., (1990), "Response of a Major Freeway Bridge During the Whittier Earthquake", Pro
ceedings for the Fourth U.S. National Conference on Earthquake Engineering, Vol. 1, pp.
9971006.
Liu, W. D., Nobari, F. S., and Imbsen, R. A.,"Dynamic Response Prediction for Earthquake
Resistance Design of Bridge Structures", Seismic engineering research and practice: pro
ceedings of the sessions related to seismic engineering at Structures Congress '89, spon
sored by ASCE; San Francisco, May 15, 1989. KircherCharlesA, ed. American Society
of Civil Engineers, New York, 1989, pages 110.
Liu, W.D., Tseng, W.S., Chang, C.Y., Imbsen, R.A, Lilhanand, K., and Li, F., (1993),
"Seismic Evaluation of BeniciaMartinez Bridge.", Proceedings of the First US Seminar on
Seismic Evaluation and Retrofit of Steel Bridges; Hyatt Regency Hotel, San Francisco, Oc
tober 18, 1993. University of California at Berkeley and the California Department of
Transportation, (Sacramento, 1993).
Lukose, K., Gergely, P., and White, R.N., (1982), "Behavior of Reinforced Concrete
Lapped Splices for Inelastic Cyclic Loading.", ACI Journal, SeptemberOctober, 1982.
MacLeod, LA, and Hosny, H.M., (1977), "Frame Analysis of Shear Wall Cores.", Journal
of the Structural Division, ASCE, 193(10),20372047.
MacLeod, LA, (1973), "Analysis of Shear Wall Buildings by the Frame Method.", Pro
ceedings Institution of Civil Engineers, 55 (Part 2; Sept.), 593603.
MacLeod, LA.,( 1976), "General Frame Element for Shear Wall Analysis." Proceedings In
stitution of Civil Engineers, 61 (Part 2; Dec), 785790.
Mahin, S., and Bertero, V. (1981), "An Evaluation of Inelastic Seismic Design Spectra.",
J. Struct. Eng., ASeE, 107(9), 17771795.
241
Makris, N., Badoni, D., Delis, E. and Oazetas, O. (1995), "Prediction of Observed bridge
Response With SoilPileStructure Interaction.", Journal of Structural Engineering, ASCE
Mander, J.B., Kim, lH. and Chen, S.S., (1993), "Some Implications Regarding Different
Bearing types on Seismic Performance of Bridges in the Eastern United States", 10th An
nual International Bridge Conference, Pittsburgh, Pennsylvania. mC9350.
Mander, J.B., Kim, J.H., Chen, S.S. and Premus, 0.1., (1996), "Response of Steel Bridge
Bearings to Reversed Cyclic Loading. ", Report No. NCEER96xxxx, National Center for
Earthquake Engineering, State University of New York, Buffalo, New York.
Manfredi, O. (1993), "Modeling of Damage in Seismic Analysis of Structures. " Ph.D. Dis
sertation. Department of Civil Engineering, University of Naples, "Federico IT", Italy (in
Italian).
Maroney, B., Romstad, K. and Chajes, M., (1990), "Interpretation of Rio Dell Freeway Re
sponse During six recorded Earthquake Events.", Proceedings of the Fourth U.S. national
conference on Earthquake Engineering, Vol. 1, ppl0071016.
McCabe, S.L., and Hall, W.1., (1989), "Assessment of Seismic Structural Damage. "J.
Struct. Div, ASCE, 115(9).
Miyazaki, M. et al. (1988). "Design and its Performance Verification of a Base Isolated
Building Using Lead Rubber Bearings in Japan." Proc. of Ninth World Conf. on Earth
quake Eng. Japan, V, 717722.
Mizukoshi, K. et al. (1989). "Torsional response Behavior of Base Isolated FBR Reactor
Building during Earthquake Excitation." ASME Pressure Vessels and Piping Conf., Ha
waii, p 127.
Mokha, A. S., (1990), "Experimental And Analytical Study Of Sliding Isolation Systems",
Ph.D dissertation, State University of New York, Buffalo, New York.
Mokha, AS., Constantinou, M.C. and Reinhom, A. M., (1993), "Verification Of Friction
Model Of Teflon Bearings Under Triaxial Load.", Journal of structural engineering, vol
ume 119, number 1, pp. 240260.
Nagarajaiah, S., Reinhom, A.M. and Constantinou, M.C., (1989), "Nonlinear Dynamic
Analysis of ThreeDimensional Base Isolated Structures (3DBASIS).", Report No.
NCEER8900l9, National Center for Earthquake Engineering, State University of New
York, Buffalo, New York.
242
Nakamura, T. et al. (1988), "Study on Base Isolation for Torsional Response Reduction in
Asymmetric Structures Under Earthquake Motion." Proc. of Ninth World Conf. on Earth
quake Eng., Japan, V, 675680.
Nakata, S., Sproul, T. and Penzien, J., (1978), "Mathematical modeling of Hysteretic Loops
for Reinforced Concrete Columns", Report No. UCBIEERCI78/11, University of Califor
nia, Berkeley. Park, R., and Paulay, T.,(1975), "Reinforced Concrete Structures", A Wiley
Interscience Publication, pp 577579.
Newmark, N. M., (1959), "A Method of Computing for Structural Dynamics" Journal of
the Engineering Mechanics Division, Vol. 85, No. EM 3, pp. 6794, July, 1959
Orangun, e.O., Jirsa, J., and Breen, J.E.,(1975), "The Strength of Anchor Bars: A Revalu
ation of Test Data on Development Length and Splices.", University of Texas at Austin,
research Report 1543F, January, 1975.
Park, R., Priestley, M.J.N., Gill, W.D., and Potangaroa, R.T.,( 1980 "Ductility and Strength
of Reinforced Concrete Columns with Spirals or Hoops Under Seismic Loading." Proceed
ings of the Seventh World Conference on Earthquake Engineering, Istanbul, Turkey, Sep
tember 813,1980.
Park, Y.J. and Ang, A HS., (1985), "Mechanistic Seismic Damage Model for Reinforced
Concrete." J. Struct. Eng., ASCE, 111(4),740757.
Park, Y.J., (1992), "Equivalent Linearization for Seismic Responses. I: Formulation and
Error Analysis", ASCE Journal of Engineering mechanics, Vol. 118, No. 11, November,
1992.
Park, Y.J., Wen, Y.K. and Ang., AH.S.,(1986), "Random Vibration of Hysteretic Systems
Under Bidirectional Ground Motions." Earthquake Eng. Struct. Dyn., 14(4),543557.
Park, Y.J., Reinhorn, AM. and Kunnath, S.K. (1987), "IDARC: Inelastic Damage Analysis
of Reinforced Concerete Frame  ShearWall Structures.", Report No. NCEER870008,
National Center for Earthquake Engineering, State University of New York, Buffalo, New
York
Paskova, T. and Meyer, C. (1994), "Damage of Plain and Fiber Reinforced Concrete Under
Lowcycle Fatigue Load." Department of Civil Engineering and Engineering Mechanics,
Columbia University, New York, NY.
Paulay, T., Zanza, M. and Scarpas, A, (1981), "Lapped Splices in Bridge Piers and in Col
umns of EarthquakeResisting Reinforced Concrete Frames.", Research Report 816, De
partment of Civil Engineering, University of Canterbury, Christchruch, New Zealand.
243
Penzien, J., (1979) "Seismic Analysis of MultiSpan reinforced Concrete Bridges.", Pro
ceedings of a Workshop on Earthquake Resistance of Highway Bridges, Applied Technol
ogy Council, Report No. ATC61, November, 1979, pp. 423432.
Prakahs, S., (1989), "Analysis and Design of Pile Foundation Under Dynamic Condition",
Bulletin ofIndian Society for Earthquake Engineering Technology, Paper No. 269, Vol. 26,
No.1, March 1989, p. 161.
Prakash, V., Powell, G.H., and Filippou, F.C. (1992). "DRAIN3DX: Base Program User
Guide" Report NO. UCB/SEMM92129 Department of Civil Engineering, University of
California, Berkeley, CA.
Priestley, M.J.N., Seible, F., and Chai, Y.H.,(1992) "Design Guidelines for Assessment
Retrofit and Repair of Bridges for Seismic Performance.", Report No. SSRP92/01, Dept.
applied Mechanics and Engineering Sciences, University of California, San Diego, 1992,
p.266.
Priestly, M.J.N., and Seible, F., (1991)," Seismic Assessment and Retrofit of Bridges. ", Re
search Report SSRP91103, Department of Applied Mechanics and Engineering Sciences,
university of California, San Diego, 1991.
Ramberg, W. and Osgood, W.R., (1943), "Description of Stress Strain Curve by Three Pa
rameters.", Technical note 902, National Advisory Committee on Aeronautics.
Reinhorn, AM., Kunnath, S.K. and Mander, J.B. (1992), "Seismic Design of Structures for
Damage Control. ", Monograph On Nonlinear Seismic Analysis of RC Buildings, P. Pajfar
and H. Krawindler Editors, Elsevier Applied Science.
Reinhorn, A M. and Kunnath, S.K. (1994), "IDARC: Computer Program for Inelastic
Damage Analysis of Reinforced Concrete Structures." in Research Accomplishments
19861994: A Selection of Technical Achievements, National Center for Earthquake Engi
neering Research, State University of New York, Buffalo, NY (in press).
Reinhorn, AM., Li, C., Constantinou, M.C., (1995), "Experimental and Analytical Inves
tigation of Seismic Retrofit of Structures with Supplemental Damping: Part 1  Fluid Vis
cous Damping Devices" Report No. NCEER950001, National Center for Earthquake
Engineering, State University of New York, Buffalo, New York Report No. NCEER95
0001, National Center for Earthquake Engineering, State University of New York, Buffalo,
New York.
244
Robinson, A.R and Chen, C.C, (1993), "Improved Time History Analysis for Structural
Dynamics. II: Reduction of Effective Number of Degrees of Freedom.", Journal of Engi
neering Mechanics, Vol. 119, No. 12, pp. 25142530
Ross, G.A., Sed, H.B., and Migilicaccio, R.R., (1973), "Performance of Highway Bridge
Foundations," The Great Alaska Earthquake of 1964Engineering.", Committee on the
Alaska Earthquake ofthe Division of Earth sciences, National Research Council, National
Academy of Sciences, Washington, D.C., 1973.
Saiidi, M. and Sozen, M.A., (1979), " Simple and Complex Models for Nonlinear Seismic
Response of RIC Structures", Civil Engineering Studies, SRS 465, University of lllinois,
Urbana, August 1979.
Shames, I.H. and Cozzarelli, F.A., (1992), "Elastic and Inelastic Stress Analysis.", Prantice
Hill Inc., Englewood Cliffs. NJ.
Shewchuk, J.R., (1994), "An Introduction to the Conjugate Gradient Method Without the
Agonizing Pain. ", CMUCS94125, School of Computer Science, Carnegie Mellom Uni
versity, Pittsburgh, PA.
Sivakuman, B., Gergely, P., and White, R.N., (1983),"Suggestions for the Design of RIC
Lapped Splices for Seismic Loading.", Concrete International: Design and Construction,
February, 1983.
Su, L., Ahmake, G. and Tadkbakhish, I.G., (1987), "A Comparative Study of Base Isolation
Systems", Report No. MIE150, Clarkson Univ., Potsdam, N.Y.
Takayanagi, T. and Schnobrich, W.e., (1977). " Computed Behavior of Coupled Shear
Walls", Proceedings of 6th WCEE, New Delhi.
Takeda, T., Sozen, M.A. and Nielsen, N.N., (1970), "Reinforced Concrete Response to
Simulated Earthquakes." Proceedings, ASCE, Vol.96, No.Stl2, Dec. 1970, pp. 25572573.
Tarics, A.G., Way, D. and Kelly, J.M., (1984), "The Implementation of Base Isolation for
Foothill Communities Law and Justice Center.", Report to the National Science Founda
tion.
Tsopelas, P., Okamoto, S., Constantinou, M C., Ozaki, D., Fujii, S. (1994), " Nceertaisei
Corporation Research Program On Sliding Seismic Isolation Systems For Bridges: Exper
imental And Analytical Study Of Systems Consisting Of Sliding Bearings, Rubber Restor
ing Force Devices And Fluid Dampers, volume ii.", Report No. NCEER940002, National
Center for Earthquake Engineering, State University of New York, Buffalo, New York
Turkington, D.H., Carr, AJ., Cooke, N. and Moss, P.J, (1989), "Seismic Design of Bridges
245
on LeadRubber Bearings", Journal of Structural Engineering, Vol. liS, No. 12.
Uang, C. and Bertero, V.V., (1990), "Evaluation of Seismic Energy in Structures", Earth
quake Engineering and Structural Dynamics, Vol. 19, pp. 7790
Valles, R.E., Reinhom, A.M., Madan, A., Barron, R., (1996), "Sesimic Evaluation of a Low
Rise RC Building in the Vicinity of the New Madrid Seismic Zone",Report No. NCEER
96xxxx, National Center for Earthquake Engineering, State University of New York, Buf
falo, New York.
Valles, R.E., Reinhom, A.M., Kunnath, S.K., Li, C. Madan, A., (1996), (IDARC 2D Ver
sion 4.0: A Computer Program for the Inelastic Analysis of Buildings,", Report No.
NCEER960010, National Center for Earthquake Engineering, State University of New
York, Buffalo, New York.
Velestos, A.S. and Newmark, N.M., (1960), "Effect of Inelastic Behavior on the Response
of Simple Systems to Earthquake Motions.", Proceedings, 2nd World Conference on Earth
quake Engineering, Tokyo and Kyoto, Vol. II, pp. 895912.
Villaverde, R. and Melad, H.M., (1992), "Efficient Mode Superposition Algorithm for
Seismic Analysis of NonLinear Structures.", Earthquake Engineering and Structural Dy
namics, Vol. pp 849858.
Wada, A. et al. (1988), "Dynamic Analysis of Base Isolated Structure Using various Nu
merical Models." Proc. of Ninth World Conf. on Earthquake eng., Japan, Vii, 403408.
Wang, M.L. and Shah, S.P., (1987), "Reinforced Concrete Hysteretic Model Based on the
Damage Concept.", Earthquake Engineering and Structural Dynamics, 15,9931003.
Weaver, W. and Oere, J.M., (1990), "Matrix analysis of framed structures", Van Nostrand
Co.
Wen, Y.K. (1976), "Method of Random vibration of hysteretic Systems." J. of Eng. Mech.
Div. ASCE, 102(EM2),249263.
Williams, M.S., and Sexsmith, R.O., (1994), "Review of Methods of Assessing Seismic
Damage in Concrete Structures.", Technical Report # UBC 9402, Earthquake.Engineering
Research Facility, University of British Columbia.
Wilson, E. L., (1994) "The Static and Dynamic Analysis of Structures With Limited Num
ber of Nonlinear Elements", Structural Analysis Programs, Inc., Draft, March 14, 1994.
246
Cliffs, N.J.: Prantice Hall, 1988.
Yao. J.T.P.,(1985), "Safety and Reliability of Existing Structures. ", Pitman Advanced Pub
lishing Program.
Yasaka, A. et al. (1988). "Biaxial Hysteretic Model for Base Isolation Devices." Summa
ries of Technical Papers of Annual Meeting  Architectural Institute of Japan, 1,395400.
Yasaka, A., Mizukoshi, K., Lizuka, M., Takenaka, Y., Maeda, S. and Fujimoto, N., "Biax
ial Hysteretic Model for Base Isolation Devices", Internal Report, Kajima Institute of Con
struction Technology, Kajima Corporation, Tokyo, Japan.
Zayas, V.A., Low, S.S. and Mahin, S.A., (1987), "FPS Earthquake Resisting System Ex
perimental Report.", Report No. 87/01. Earthquake Engineering Research Center, Univer
sity of California, Berkeley.
247
A. APPENDIX A
L_ID_A_R_C__B_R_ID_G_E_li I Input
I Preprocessing
I Processing
The description of the major functions of the input routines is provided in Table
A.I
248
TABLE A.1 Input Routines Description
geccoupled_motions
define ele gecelemencproperties The main routine  gecelemencproperties calls the
ments proper geCbiJ..gap other routine to retrieve the required properties of the
ties elements.
get_e_b
geCfoundation
geCh_b
geCh_b_s
geUso_sli
geUsolator_1
geClin_damper
define ele gecmember_connectivity This group of routines defines the elements. Including
ments assign_elements_property the connectivity of the elements to the nodes, the prop
erty number of the element, and its type.
assign_elements_types
define ele get_member_boundarY30 The element matrices are created first assuming they
ments end nditions are fixed at the ends. Releases, end springs, and rigid
conditions gecmember_springs arms as described in chapter 3 are applied later for eas
ier modeling
geuigid_arms
loadings get.,joinUoad Some of the parameters defining the right hand side of
get.,joincweight the equation of equilibrium are defined in this group of
routines. The joint load (for incremental static analy
gecmembecuniformJoad sis) and the number of steps to which the load is
gecnumber_oUoad_steps divided to.
Joint weights are used to define the mass for dynamic
analysis
pushover get_usecpushover Those conditions are used to indicate the stopping con
conditions get_romp_pushover ditions for the adaptable pushover procedures, and the
pseudo dynamic procedures
dynamic and gecnumber_oCmodes The parameters required for: modal analysis (number
modal analy geUotal_analysis_duratio of modes, displacements velocity (excitation groups).
sis parameters n and some general parameters like global damping and
integration time are read
gecanalysis_time_step
get_damping coefficients
geCdirection of excitation
gecexcitation_file_name
gecexication..groups
geCpealcground_accelera
tio
geUype_oCexciation
read_number_oCmodes
249
The general description of the routines included in the preprocessing portion is
Fig. A.2. A detailed description of each one of the blocks in the processing section is pre
250
Processing update elements
solve
process results
251
TABLE A.3 Update Elements Properties
element general update_elements!leneral Element of the type "general" are elements that
update_iso_sli contributes to three global matrices: (1) stiffness
matrix (2)damping matrix (3) load vector. An
example of this type of element is the isolator
slider element
252
TABLE A.S Global Matrices Building Description
253
I
254
Гораздо больше, чем просто документы.
Откройте для себя все, что может предложить Scribd, включая книги и аудиокниги от крупных издательств.
Отменить можно в любой момент.