PB84-234384

REPORT NO.
UCBjEERC-84j03 EARTHQUAKE ENGINEERING RESEARCH CENTER
APRIL 1984

REFINED MODELLING OF
REINFORCED CONCRETE COLUMNS
FOR SEISMIC ANALYSIS

by

S. A. KABA
S. A. MAHIN

Report to the National Science Foundation

----~

COLLEGE OF ENGINEERING

UNIVERSITY OF CALIFORNIA . BerkeleYI California

REPRODUCED BY
NATIONAL TECHNICAL
INFORMATION SERVICE
u.s. DEPARTMENT OF COMMERCE
SPRINGFIElD, VA. 22161
Forsale by the National Technicallnforma-
tion Service, U.S. Department of Commerce,
Springfield, Virginia 22161.

See back of report for up to date listing of

EERC reports.

DISCLAIMER
Any opinions, findings, and conclusions or
recommendations expressed in this publi-
cation are those of the authors and do not
necessari Iy reflect the views of the National
Science Foundation or the Earthquake En-
gineering Research Center, University of
California, Berkeley
 REFINED MODELLING OF

REINFORCED CONCRETE COLUMNS

FOR SEISMIC ANALYSIS

by

Said A. Kaba

and

Stephen A. Mahin

A report on research sponsored by

the National Science Foundation

Report No. UCB/EERC-84/3

Earthquake Engineering Research Center

College of Engineering

University of California

Berkeley. California

APRIL 1984
 ABSTRACT

The report extends the fiber representation of a section to make possible the analysis of rein-

forced concrete or steel members independently or as part of two-dimensional frames. Hence,

slice locations are specified along the length of the element. Each slice is further discretized into

steel and/or concrete fibers. The state of the fiber is monitored and updated assuming plane sec-

tions remain plane. Slice stiffness is calculated from the fiber stiffness and the element stiffness is

arrived at by integration over the length of the element assuming linear flexibility between slices.

Shape functions which relate slice deformations to element displacements are also calculated and

updated.

By monitoring the various fibers and slices, it is possible to obtain the local response at critical

sections including moment-curvature and shear histories. Bond slip and shear effects are not con-

sidered however. Hence the multi-slice fiber element is suitable for modeling beams and columns

in a steel or reinforced concrete frame especially when the effects of axial loads on strength and

stiffness are significant. Furthermore, the element can account for axial load-bending moment

interaction and the pinching of hysteresis loops due to compressive loads acting on the element.

The functioning of the element is tested separately and as part of a general dynamic two-

dimensional frame analysis program (DRAIN-2D2). The results of analyses are compared with

theoretical, experimental, and other analytical results and show the element to be a useful addi-

tional tool for the analysis of reinforced concrete frame behaviour. The limitations and advantages

over some other conventional discrete models are also illustrated through analysis results.
 -ii-

ACKNOWLEDGEMENTS

The work described in this report was done under the sponsorship of the National Science

Foundation. This financial support is gratefully acknowledged.

The assistance of graduate student Finley Charney with the DRAIN-2D runs is particularly

appreciated. Thanks are also due to the University of California's Department of Civil Engineering

for making available the facilities of the Microcomputer Research Laboratory of the Structural

Engineering and Structural Mechanics Division as well as to the Earthquake Engineering Research

Center for the use of their VAX 750 computer.

The report and many of the figures were prepared using the facilities of the Berkeley Campus

Computer Center. The fine efforts of the technical illustrators, especially Miss Gail Feazell, are

appreciated.
 -iii-

TABLE OF CONTENTS

ABSTRACT it

ACKNOWLEDGEMENTS ii

TABLE OF CONTENTS iii

I. INTRODUCTION 1

1.1 Introductory Remarks 1

1.2 Available Models 2
1.3 Simple Models 3
1.3 (a) Multi-Degree of Freedom Systems 3
1.3 (b) Equivalent Single Degree of Freedom Systems 3
1.3 (c) Uses and Limitations of Simple Models 4
1.4 Discrete Element Models 5
1.4(a) Single Component Models 5
1.4 (b) Multi-Component Parallel Models 8
1.4(c) Multi-Spring Serial Models 9
1.4(d) Use of Interpolation Functions 10
1.4 (e) Limitations of Discrete Models 10
1.5 Finite Element Models, Uses and Limitations 12
1.6 The Fiber Model 12
1.6 (a) Basis of the Fiber Model 13
1.6(b) Limitations of the Fiber Model 15
1.7 Objectives 16
1.8 Scope 16

II. DESCRIPTION AND OPERATION OF THE MULTI-SLICE FIBER MODEL 18

2.1 Introduction 18
2.2 Element and Slice Geometry 20
2.3 Theory and Workings of the Multi-Slice Fiber Model 21
2.3 (a) Fiber Stiffness 22
2.3(b) Slice Flexibility 24
2.3(c) Element Stiffness 26
2.3(d) Shape Functions and State Determination 27
2.3(e) Unbalanced Forces 28

III. ELEMENT VERIFICATION 30

3.1 Introduction 30
3.2 Monotonic and Cyclic Loading of a Steel Beam 30
3.3 Dynamic Analysis of a Linear Elastic Column 32
 -iv-

IV. ANALYSIS OF REINFORCED CONCRETE COLUMNS AND FRAMES 33

4.1 Introduction 33
4.2 Dynamic Analysis of a Reinforced Concrete Column 33
4.2 (a) Example Column 34
4.2(b) Influence of ModelIing and Discretization 34
4.2 (c) Influence of Geometric Stiffness 36
4.2(d) Influence of Acceleration Record Amplitude 36
4.2(e) Influence of Applied Axial Load 37
4.2(f) Influence of Analysis Technique 37
4.2 (g) Analysis Using EI-Centro Record 39
4.3 Dynamic Analysis of a Reinforced Concrete Frame 40
4.3(a) Example Frame 41
4.3 (b) Effect of Some Modelling Assumptions 42
4.3(c) Influence of Analysis Technique and Time Step 43
4.3 (d) Effect of Axial Load 45
4.3 (e) Comparison With Other Models 46
4.3(f) Effect of a Varying Axial Load 47

V. CONCLUSIONS 50

REFERENCES 52
FIGURES 55
APPENDIX A Use of the Multi-Slice Fiber Model for Dynamic Analysis 91
APPENDIX B Current Array Size Limitations 96
 -1-

I INTRODUCTION

1.1 Introductory Remarks

Recent research into the behaviour of reinforced concrete structures subjected to seismic

loads has been extensive. Yet due to the many phenomena that have to be considered, there

are still many aspects of the behaviour which are not fully understood and which result in wide

variations in response predictions. The phenomena which affect the response include cracking

and crushing of concrete as well as yielding , strain hardening, slipping, and buckling of the

reinforcing steel. Furthermore, when load and displacement reversals are taken into considera-

tion such phenomena as pinching of hysteresis loops due to shear, bond deterioration, and the

Bauschinger effect become important. In addition, the effect of a varying axial load on the

behaviour of structural elements can be significant, especially in the case of reinforced concrete

columns in frames subjected to seismic loading.

Despite significant advances in numerical methods and computational techniques for solv-.

ing these types of problems [1,2,3,4,5,6,7], there has been only limited improvement with

regard to models for analytically predicting the nonlinear dynamic response of complete rein-

forced concrete frames. Although finite element programs have been developed to treat many

of these phenomena individually, little emphasis has so far been placed on practical methods for

predicting overall structural behaviour. There is a definite need for models which can provide a

reliable indication of the complex inelastic behaviour exhibited by members, but which are sim-

ple and flexible enough to be implemented in general purpose computer programs for studying

the seismic response of reinforced concrete frames. Such models would ideally provide informa-

tion on the local behaviour of critical regions which can aid in checking the adequacy of a

design, the collapse mechanism, and local detailing requirements.

Because of the many different phenomena involved, it is desirable to use different

approaches to model different components of a reinforced concrete structure, e.g. slabs, joints,
 -2-

beams, walls, and columns. While considerable research has been devoted to many of these

components, relatively little analytical research has been directed recently to the problem of

modelling columns. Axial loads acting on a reinforced concrete section affect the section's

stiffness and strength characteristics as shown in Fig. 1.1. Furthermore, if the section is sub-

jected to a cyclic curvature variation, the existence of a compressive axial load results in the

pinching of hysteresis loops as shown in Fig. 1.2. Because of this, the member's capacity for

energy dissipation is also reduced, an important consideration for seismic resistant design. The

multi-slice fiber element, which is the object of this study, is especially suited to model the

behaviour of reinforced concrete columns since it can take into consideration many of the con-

trolling phenomena, including the effect of a varying axial load.

1.2 Available Models

This study is mainly concerned with modelling elements where flexure and axial load

dominate the behaviour. Hence the emphasis is on column (and beam) modelling, while joints, .

slabs, and shear walls subjected to significant shear forces are not treated. Even with this res-

triction in scope, the relevant past research is extensive.

Different approaches have been suggested for modelling reinforced concrete frames.

These include simple or lumped models which reduce a frame to a single degree of freedom or

to a system consisting of a few degrees of freedom with lumped (and often simplified) mass

and stiffness properties. Discrete models are more complex in that they represent each struc-

tural element separately. Even more complex fiber and finite element models, whereby each

structural element is further discretized into a number of subelements, have been developed

for special applications.

The following four sections briefly describe these simple, discrete, fiber, and finite ele-

ment models, review recent studies conducted using these models, and outline their uses and

limitations.
 -3-

1.3 Simple Models

Due to the increasing availability of fast computers with large memories, emphasis has

been shifting towards complex models. Yet simple modelling approaches have not been com-

pletely abandoned. Simple structure models include the shear beam model and various pro-

cedures for reducing a frame to a single degree of freedom. In addition to their low computa-

tional cost, simple models are relatively easy to apply and are very useful in preliminary design

stages.

1.3 (a) Multi-Degree of Freedom Systems

The shear beam is the simplest representation of a frame subjected to lateral forces. Such

models have been used extensively to study the response of multi-story frames subjected to

earthquake motions [7,8,9,10]. The nonlinearity is taken into account in the shear springs

which model the story stiffness. For example, the envelope curve relating the story shear to

lateral drift could be trilinear to account for cracking and yielding at the story level. A hys-

teresis rule must also be included to account for unloading and load reversal. Aziz [7] in his

study of the dynamic inelastic analysis of frames includes an investigation of a ten-story frame

using the shear beam model. Aoyama [10] has developed a slightly more complex model that

incorporates flexural as well as shearing types of structural deformations.

1.3 (b) Equivalent Single Degree of Freedom Systems

By assuming that the structure deforms according to its first mode or some other predeter-

mined deflected shape, a multi-degree of freedom system can be reduced to an even simpler

single degree of freedom representation. This was implemented by Pique [11] in his study of

the use of simple models in nonlinear dynamic analysis. A second alternative called the Q-
 -4-

model was developed by Saiidi and Sozen [Ill. It consists of a massless rigid bar with a hinge

and a rotational nonlinear spring at its base and a concentrated mass and damper at the top as

shown in Fig. 1.3. A similar approach, used by Pique [II], superimposes a set of springs hav-

ing different yield levels in parallel. In this model, the combined stiffness changes continuously,

although the hysteresis function for each spring may be a simple relationship.

l.3k) Uses and Limitations of Simple Models

Most studies using simple models arrived at the conclusion that the maximum response

values predicted by the simple models are satisfactory when compared with results from com-

plex models or tests carried out on scaled down frames. Since the input information required is

quite simple, these methods are well suited for preliminary design and for performing

parametric studies regarding overall performance characteristics.

One must keep in mind, however, the assumptions on which such simple models are

based. For example, in the case of the single degree-offreedom representation, the higher

mode components of displacement are assumed to remain essentially unexcited by earthquake

motions. In the case of the shear beam model it is implicitly assumed that the girders are

significantly more rigid and stronger than the columns. This is true to a certain extent in the

case of weak column-strong girder design. However, this is not the typical case in reinforced

concrete construction and determination of appropriate modelling parameters may be difficult.

The information obtained using simple models is also usually insufficient for final designs,

detailing evaluations and reliability studies. For such purposes as well as for the analysis of

general structures which cannot be adequately represented by simple models, more sophisti-

cated modelling techniques are necessary.

 -5-

1.4 Discrete Element Models

Discrete element models include a one-to-one correspondence between elements of the

actual frame and the idealized system. However, they generally simplify the behaviour of the

member by idealizing its .hysteretic behaviour in terms of a set of predefined phenomenological

rules. Discrete models can be further divided into single component elements, multi-

component elements, elements which depend on interpolation functions, and plasticity based

models. This section briefly reviews these various types of discrete models, their uses, their

limitations, and recent studies which dealt with these models.

1.4(a) Single Component Models

Such models consist of a single element, usually elastic, flanked on either end by concen-

trated, nonlinear, rotational springs as shown in Fig. 1.4. Giberson [13] used this single ele-

ment model with the two concentrated flexural springs at the ends which modelled the inelastic'

deformation of the member. Otani [14] used a similar approach to study the response of a rein-

forced concrete frame subjected to base acceleration. In Otani's model the point of inflection

was assumed to be stationary at the member's mid-length, though this is not a necessary res-

triction in general.

As the effects of inelastic deformations on response become more significant, it becomes

important to use as realistic a hysteresis model as possible. Hence the simpler bilinear model

for reinforced concrete member behaviour was superseded by a model proposed by Clough and

Johnston [15] which included overall stiffness degradation of a structure. Takeda, et al [16]

then proposed a hysteresis model which better conforms with results of cyclic loading tests of

reinforced concrete connections.

To account for the various loading and unloading stages, such models can be quite com-

plex - see Fig. 1.5. Nevertheless, the pinching effects associated with shear, bond deterioration,
 -6-

unequal top and bottom reinforcement in beams, and axial load effects in columns were not

modelled in the original Takeda model. Hence, a somewhat modified model was developed by

Takayanagi and Schnobrich [171. In a study by Emori and Schnobrich [18] a further

modification was introduced to describe the effect of bar slip.

A thorough study using three of the formulations mentioned above as well as two others

was carried out by Saifdi and Sozen [I2l. The structural members were idealized as indicated in

Fig. 1.4. Shear deformations were ignored, infinitely rigid joints were assumed but the gravity

load induced P-a effects were taken into account. The member moment-rotation (M-e) rela-

tion was arrived at by simple numerical integration assuming a linear moment variation along

the length of the element and a trilinear moment-curvature (M-cP) relation (with break points

at cracking and yielding) as shown in Fig. 1.6. Using additional simplifying assumptions, rota-

tions due to bond slip were calculated resulting in a modified M-O curve.

Studies using five different hysteresis models for the concentrated flexural springs were

compared with results of a shaking table test of small scale, 3-bay, 10-story frames subjected to

the N-S component of the 1940 El Centro earthquake. The results of the comparison are sum-

marized below.

The Takeda model [16] consists of sixteen rules applied to a trilinear primary curve. The

rules give the stiffness characteristics for cyclic loading and unloading at various stages of crack-

ing and yielding. The model results in energy dissipation for loads cycling below the yield point

since cracking is a breakpoint; such energy dissipation is observed in tests and is consequently

desirable to include in the model. Pinching, however, is not considered which resulted in an

overestimation of structural stiffness during low-amplitude response. Nonetheless, the results

were very good: relative story displacement, deformed shape, and maximum response values

were in close agreement with experimental values. A refined version of the Takeda model has

been implemented in the DRAIN-2D program [2,4].

The Sina [12] model which is simpler than the Takeda model in that there are only nine

rules, has the additional advantage of including pinching which results in better prediction of

low-amplitude response than the Takeda model. To use this model the user has to specify
 -7-

certain parameters which define the shape of the hysteresis loops including the extent of pinch-

ing. This presupposes the availability of appropriate analytical or experimental results.

Nonetheless, the response relative displacements as well as the absolute maxima were overes-

timated by up to 20 %. The normalized deflected shape, however, was found to be quite close

to the experimental results.

The simple bilinear model, shown in Fig. 1. 7, is used in the analysis of steel structures

and because of its simplicity is relatively economical to use. The stiffness characteristics during

unloading and load reversal are quite different from the observed cyclic behaviour of reinforced

concrete members. The bilinear model overestimates the energy dissipation for high amplitude

deformations and ignores dissipation for low amplitude deformations. Hence, the results were

found to be vastly different from actual tests and the bilinear model is believed to be inade-

quate for modelling reinforced concrete members.

The Otani model [I4] as shown in Fig. 1.8 has a bilinear primary curve and therefore

ignores cracking altogether. This simplification was attempted to reduce the complexity and

computational effort of the Takeda model. Nonetheless it remains relatively complex ( there are

11 rules involved ). In spite of this its performance was deemed unsatisfactory: for low ampli-

tude response the predicted displacements deviated considerably from experimental results; and

for high amplitude response, errors of up to 33 % were calculated.

Finally the Q-hyst model, developed for Saiidi's study [I2l, also has a bilinear primary

curve as shown in Fig. 1.9. It differs from Otani's model mainly by using softened unloading

and load reversal stages and reducing the number of rules to four. Its performance was very

good in regions of low amplitude response. The peak values were overestimated by up to 17 %.

The results using the Takeda and the Q-hyst models were found to be satisfactory, but the

latter has the additional advantages of simplicity and better reliability in low amplitude response

regions. None of these models explicitly accounts for axial load effects. An attempt has been
 -8-

made by Fintel and Ghosh [19] to devise a single component model in which axial load

influences the plastic yield moment. The model is based on a modified Q-hyst model, but nei-

ther pinching nor change in elastic stiffness due to axial load is taken into account.

The most recent and possibly the most flexible concentrated spring model is the one pro-

posed by Chen and Powell [20] for the dynamic analysis of reinforced concrete frames. The

model is capable of representing three dimensional behaviour since it takes into account biaxial

bending, torsion, and axial load effects. This is accomplished by application of plasticity theory

to a four dimensional yield surface. As in other lumped plasticity models, zero-length plastic

hinges are assumed at the members' ends with an elastic portion in between. The hinges are

initially rigid, hence the initial stiffness of the member is identical to the stiffness of the elastic

portion. Furthermore, by having two subhinges at each hinge location, cracking and yielding

can be modelled. The model can represent stiffness degradation. However, it is unclear how

well a model based on plasticity interpretations of yield surfaces can account for the behaviour

of concrete columns where cracking and crushing of concrete may have a crucial role.

1.4(b) Multi-Component Parallel Models

Another type of model is the parallel two-component formulation proposed by Clough and

Benuska [21,22,23]. The model consists of an elastic element in parallel with an elasto-perfectly

plastic element; hence it results in a conventional bilinear hysteresis rule. The elastic portion is

assigned a fraction of the member's assumed stiffness equal to the proposed post-yield harden-

ing stiffness. The elasto-perfectly plastic element has an initial stiffness calculated such that the

combined stiffness of the two components in the elastic range is equal to the initial elastic

stiffness of the actual member.

The parallel two-component model was used in many early studies of reinforced concrete

structures [24] and incorporated into several computer programs [2,4,24,25,26,27]. The basic

bilinear hysteretic formulation would not be expected to result in reliable response prediction
 -9-

for a particular structure and ground motion [12,24]. However, Powell and Row [26J found that

the average response values obtained for bilinear and stiffness degrading models using many

different ground motions are nearly identical. Takizawa [28] modified the two-component bil-

inear model to better reflect the softening observed on unloading from the yield range and used

the modified model to analyze a three-story reinforced concrete frame. Aoyama [lO] extended

the model into a nondegrading trili~ear hysteresis formulation by modifying it into a three

-component element consisting of an elastic element in parallel with two elastoplastic elements.

Klingner [27] uses two or three two-component elements in parallel to achieve the same effect.

Nonetheless, the axial load effects are simplified using this approach.

1.4(c) Multi-Spring Serial Models

In the single and parallel component models plastic hinges are assumed to form only at

the ends of a member.. However, at large deformations or as the result of transverse loads the

length of the plastic hinge regions can no longer be neglected. In addition, the stiffness charac~

teristics of elements vary along their length even in the elastic range. To account for this a

number of simple alternatives is possible. For example, as a simple extension of the single

component model, Hsu [29], and Takayanagi and Schnobrich {I7] proposed a multiple spring

model for analyzing wall members. The member is divided into several sub-elements along its

axis each connected by nonlinear springs, see Fig. 1.10. Emori and Schnobrich {I8] in their

study of frame-wall structures used single component elements to model beams and columns,

and multiple spring elements to model the shear wall. The flexural properties of each of the

subelements is defined by a trilinear moment-rotation primary curve and a set of hysteresis

rules is used to describe the unloading and load reversal stages. Although the multi-spring

serial model can represent the behaviour of a frame element subjected to a relatively general

moment distribution along its length, it still does not directly account for the effect of axial load

on member behaviour.
 -10-

1.4 (d) Use of Interpolation Functions

Another way of achieving variable stiffness along the length of the member is through the

use of interpolation functions. Umemura et al [8,30J suggest such a model. They assume a par-

abolic distribution of flexural flexibility (lIED along the member's axis. They account for

shearing deformations by assuming shear stiffness to be proportional to the flexural stiffness.

Using the parabolic interpolation function the member flexibility can be calculated for a known

moment distribution given the section flexibility at the two ends and point of inflection. The

section flexibilities are obtained using a moment-curvature relationship based on a modified

Takeda model. The element flexibility matrix is inverted to obtain the required member

stiffness matrix.

Meyer et al [31,32] proposed a nonlinear reinforced concrete beam element that accounts

for the spread of plastic regions. The model is also based on a simplified Takeda moment-

curvature relation which needs to be calibrated against experimental test results. ZAP [33] is a

similar model that considers fixed end rotations due to bond slip in addition to plastic zones of

finite length.

Umemura and Takizawa [30], use similar modelling techniques for shear walls. The shear

deformations are treated independently from the flexural deformations and the corresponding

shear and flexural flexibility matrices are added to obtain the element flexibility which is

inverted to arrive at the element stiffness. Umemura et al [8] suggest that the axial load-

bending interaction for a fluctuating axial load can be accounted for by modifying the model.

However, this is not implemented in their studies.

l.4(e) Limitations of Discrete Models

Discrete models briefly described in this section can provide more detailed and accurate

information about the behaviour of reinforced concrete members than the simple models
 -11-

discussed in Section 1.3. Nonetheless, discrete models based on simplified hinge or other

representations of inelastic behaviour are incapable of simulating the complex behaviour of

reinforced concrete members under arbitrary loading conditions and histories.

As discussed in Section 1.4(a) the concentrated spring model can be refined by increasing

the number and complexity of the rules that govern the nonlinear moment-rotation relationship

for the springs. The fact that the hysteretic behaviour is predefined results in significantly

reducing the potential of the concentrated spring model to adapt to general loading conditions.

The properties of these concentrated springs are a function of the particular loading developed

in the member which is not generally known a priori. In addition, the hysteretic models usually

incorporated in most programs are not sufficiently elaborate to account for pinching due to

shear stress, axial force, and unsymmetrical steel distribution. However, the model proposed

by Thorn [49] considers the effect of shear on the pinching and degradation of hysteresis loops.

The concentrated spring model may indeed be incompatible with some of these problems

due to the fact that unlike flexural deformations inelastic axial and shearing deformations may

not be concentrated at the ends of the member. In addition, it may be difficult to define an .

appropriate set of phenomenological rules that govern the interaction of flexural and axial

deformations. The use of interpolation functions can account for the spread of plasticity and

cracking. However, models using interpolation functions are also based on a multilinear

moment-rotation relationship which needs to be predefined. In addition, the fact that the form

of the interpolation function is constant (e.g. parabolic) may result in an inexact approximation

of the member's stiffness. Meyer et al [31,32] proposed a nonlinear reinforced concrete beam

element that takes into account plastic deformations over regions of finite length. However, it

is based on a simplified Takeda moment-curvature model which does not include pinching due

to shear and axial loads and which needs to be calibrated against experimental test results. The

similar ZAP [33] model considers fixed end rotations due to slip as well as plastic zones of

finite length. It was developed to model reinforced concrete beams only and therefore does not

include axial load effects nor axial load-bending interaction.

 -12-

1.S Finite Element Models, Uses and Limitations

Unlike the discrete models described in Section 1.4, finite element models idealize each

structural member as an assemblage of a large number of finite elements. These elements can

be of various types such- as truss or beam elements for steel and two dimensional plane stress,

plane strain elements or even three dimensional elements for concrete.

Analyses of reinforced concrete structures using the plane stress inelastic type of finite

elements have been performed [34,35,36]. Such analyses are particularly suited to studies of

wall panels. Such studies include Yuzugullu and Schnobrich's [37] investigation of a shear wall

frame system under monotonically increasing loading, Darwin's analysis of reinforced concrete

shear panels subjected to cyclic loading, and Cervenka's [38] inelastic analysis of reinforced

concrete panels loaded in their own plane. Although the correlation with experimental results

was good, such studies can be extremely costly. This is due to the number of elements involved

and the nonlinear behaviour, especially when loads reach significant levels. In addition to

material nonlinearities, the need to monitor crack propagation and bond deterioration further

complicates the analysis. The cost and time involved become prohibitive in the case of large

structures and when a dynamic analysis is contemplated. Furthermore, there are uncertainties

involved in the material and stiffness formulations of the finite elements resulting in approxi-

mate results in spite of the apparent refinement of these methods.

1.6 The Fiber Model

The fiber model is based on the finite element approach but simplifications are introduced

to increase computational efficiency. It differs from the phenomenological approach to discrete

element modelling in that a relatively detailed analytical description of the geometry and

material properties is used to evaluate the behaviour of critical regions rather than a simplified

and predefined set of hysteresis rules.

 -13-

The fiber model is somewhat more sophisticated than discrete models and can take into

consideration more of the phenomena involved in reinforced concrete behaviour. Typically,

the only data necessary 10 define the fiber model are the geometry of the member and section,

and the stress-strain relationships for the concrete and the reinforcing steel. Such data is gen-

erally available and is more reliable than moment-curvature sectional properties or moment-

rotation relationships which are needed in the case of phenomenological discrete models. The

fiber model, however, is less detailed than finite element models. Hence the computational and

storage requirements are less and the fiber model can be used to analyze the dynamic response

of reinforced concrete frames. However, the simplifying assumptions reduce the generality of

the method.

1.6(a) Basis of the Fiber Model

An extensive study of the fiber model approach to structural dynamic analysis was per-

formed by Mark and Roesset [39]. The purpose of their study was to investigate the applicabil-.

ity of the fiber model to the prediction of the nonlinear dynamic response of reinforced con

crete frames. The procedure started by dividing each element into segments along the axis of

the member. The slices at the ends of each segment were further divided into concrete and

steel fibers. The strains in these fibers were calculated from the centroidal strain and section

curvature by assuming that plane sections remain plane. From the assumed uniaxial stress-

strain relationships, tangent stiffness moduli were calculated for the various fibers. From these

fiber stiffness moduli the slice stiffness can be arrived at by proper summation. Assuming

linear variation of stiffness between slices, the element tangent stiffness matrix was calculated

by integrating the slice stiffness along the length of the member.

Thus, the important first step was the adoption of suitable hysteretic stress-strain relation-

ships for the materials involved. The concrete model adopted by Mark [39J consists of a tri-

linear envelope with linear loading and unloading. Concrete is assumed incapable of carrying
 -14-

tensile stresses and once cracked, a concrete fiber cannot supply compressive strength until the

crack has closed. This simple concrete model was adopted because the details of the model

were found not to affect the section properties significantly. Mark models the steel behaviour,

however, with greater care. Six different analytical models based on three experimental studies

were compared. For small to moderate strains the bilinear elasto-plastic steel model was found

to perform reasonably well.

To solve the dynamic problem, the tangent stiffness corresponding to a current computed

moment and curvature state is used to predict the new curvature state (and hence moment).

This approach avoids iterating on the neutral axis position to achieve equilibrium at the section

level. However, Mark did not impose equilibrium corrections due to nonlinearities occurring

within a step so that small errors accumulated at every step.

While Mark [39] and in a similar fashion, FlRES-RC [40] used the assumptions of plane

sections and uniaxial stresses, Bazant and Bhat [41] in their study of the hysteretic response of

reinforced concrete members used triaxial material properties and the deep beam bending

theory with transverse shear. To describe the steel stress-strain relationship Bazant and Bhat

[41l adopted a simple analytic formulation that takes into consideration strain hardening and

the Bauschinger effecL To describe the triaxial concrete behaviour, they used the Endochronic

Theory [42]. Triaxial stress states exist in reinforced concrete members because of shear

effects and the transverse confinement provided by the ties. Although deflections due to shear

deformations were not significant in the case considered, inelastic shear strains caused volume

dilatancy which resulted in significant forces in the stirrups. To include the effects of

confinement due to the ties, Bazant and Bhat [41] included transverse normal strains as

separate variables. Results for a cantilever beam were found to be in good agreement with

experimental data.

As used in the studies by Mark [39], Bazant and Bhat [41] the fiber model was found to
I

be a valuable tool for studying the hysteretic response of reinforced concrete members. Such

phenomena as the "pinching effect" and the effect of axial load on stiffness and strength can be
 -15-

reproduced automatically. Both studies included comparisons of analytic and experimental

results that showed that the fiber model can provide results which are in general agreement

with experiments and which are far more reliable than phenomenological models.

1.6(b) Limitations of the Fiber Model

Many of the limitations of the concentrated model that were described in Section 1.4(e)

do not apply to the fiber model since the fiber model starts with material stress-strain relation-

ships, the moment-rotation curve is not predefined and the model can represent the pinching of

hysteresis loops due to the a constant or varying axial load [39]. However, other considerations

are only treated approximately. For example, the common assumption of plane sections

remaining plane may not be exactly true, especially when there are significant amounts of

shearing deformations, shear cracks, or bond slip. Most fiber model analyses disregard these

effects. Bazant and Bhat [41] include shear strains which mean that the section (though

remaining plane) need not be normal to the axis of the element. Filippou et al [43] in a recent

study of bond deterioration do not assume compatibility of strains between the reinforcement

and the adjacent concrete, hence bond slip is explicitly considered.

Further simplifying assumptions are made when deducing element stiffness properties

from section properties. For example, a finite number of sections are chosen at various points

along the element and the stiffness is then arrived at by integrating over the length of the ele-

ment. The integration is performed by using a predetermined variation (shape or interpolation

function) which is completely defined once the stiffnesses of the sections being monitored are

calculated. Mark [39] uses twenty such sections while Bazant and Bhat [41] use nine. Both,

however, place those sections at equidistant intervals along the element. Such interpolation

functions cannot reflect the exact stiffness variation along the element especially since this vari-

ation changes as the loading conditions evolve.

 -16-

1.7 Objectives

The fiber model as described in the previous section is particularly suited for modelling

the behaviour of reinforced concrete members and frames. The increasing availability of high

speed computers with large storage capacity has made the use of fiber models a viable option

for the dynamic analysis of reinforced concrete frames.

The objective of this study is to develop and evaluate a multi-slice fiber element for

predicting the inelastic cyclic behaviour of reinforced concrete beam-columns. Inherent in this

development is the need to incorporate automatically those features of reinforced concrete

member behaviour (such as cracking, stiffness degradation and pinching of hysteretic loops,

axial load-bending moment interaction, etc.) that can have an important influence on structural

behaviour, while achieving reasonable computational efficiency. Thus, certain simplifications

will be introduced in order to permit analysis of complete structural systems using modern com-

puter capabilities. Primary assumptions will limit consideration to uniaxial bending, and shear-

ing deformations and bond slip will be disregarded at this stage. To facilitate application of the

model it will be implemented in a general purpose nonlinear dynamic analysis computer pro-

gram. Comparisons with experimental and other analytical results will be made to assess the

accuracy and practicability of this modelling approach.

1.8 Scope

The following chapters discuss the theory of the multi-slice fiber model and its use in the

static and dynamic analysis of steel and reinforced concrete members and frames. Chapter 2

includes a description of the theory and operation of the multi-slice fiber model including the

step by step procedure for calculating and updating element stiffness. Material models are

described in this chapter. The implementation of the element in the DRAIN-2D2 [4] program

is also presented. Chapter 3 is mainly concerned with the verification of the functioning of the
 -17-

element. Simple analyses including the monotonic and cyclic loading of a steel beam as well as

the dynamic analysis of a linear elastic column are carried out and the results compared with

theoretical solutions.

Chapter 4 presents results of dynamic analyses of reinforced concrete columns and frames

subjected to sinusoidal as well as earthquake base motions. The effect of various modelling and

discretization assumptions is investigated in addition to the influence of analysis technique and

size of time step. The use of the multi-slice fiber model in predicting the response of reinforced

concrete members subjected to constant or varying axial loads is illustrated through various ana-

lyses and comparisons with results of analyses using other available models. Finally, conclusions

and suggestions for further research are presented in Chapter 5.

 -18-

II Description And Operation Of The Multi-Slice Fiber -Model

2.1 Introduction

Many approaches have been suggested for improved modelling of reinforced concrete

members for dynamic analyses of frames. One option is the use of many single component ele-

ments connected in series. While this procedure can take into account spreading of yielding

and cracking, it results in additional degrees of freedom and hence computational expense.

More importantly in the non-iterative, step-by-step integration procedures used in most

dynamic analyses, these additional degrees of freedom can develop significant unbalanced loads

that have to be eliminated which frequently results in numerical instabilities. Another option is

to use interpolation functions to distribute the yielding along the member. Such functions,

however, need to be predefined and while they may accurately represent the elastic behaviour

of the element, they would be incapable of representing inelastic behaviour under general load-

ing histories. Moreover, these and other modelling alternatives are usually based on idealized

moment-rotation relationships for the plastic hinge and it would be difficult to include the effect

of varying axial loads.

As adapted by Mahasuverachai and Powell [46] for the inelastic analysis of steel piping

and tubular structures, the multi-slice fiber element combines the advantages of both of these

approaches. Sections at various positions along the length of the member can be monitored and

interpolation functions can be used so that no additional degrees of freedom need to be

included or condensed out. Furthermore, the sections (or slices) are modelled by a fiber

representation so that the limitations associated with predefined phenomenological type hys-

teresis rules are overcome.

The multi-slice fiber element requires as input only the section and member geometry,

and the stress-strain relationships of the various 'fibers'. From these relationships, the slice

properties and subsequently the element stiffness can be deduced for the current load condi-

tions.
 -19-

The interpolation (or more appropriately shape functions) need not be predefined. These

are formulated as matrix transformations relating slice deformations to member end displace-

ments. Such functions depend on the current state of the various slices and need to be updated

whenever yielding or other changes occur. The derivation, use and updating of the shape func-

tions and element stiffness is central to the simplicity of the method and will be described in

some detail in the following sections.

Members to be modelled using the multi-slice reinforced concrete fiber element are

assumed herein to be straight and the locations of the slices along the length of the member

can be arbitrarily located by the user as described in the next section. Each slice must have an

axis of symmetry and the axes of the various slices must be in the same plane. All applied loads

must be in that plane; thus, only uniaxial bending and axial effects are considered. The

detailed theory and workings of the model are described in Section 2.3.

Theoretically the multi-slice fiber model for reinforced concrete can be formulated so as

to be suitable for three dimensional frame analyses. The model as proposed by this study, how-

ever, is intended for evaluation purposes and will be implemented for nonlinear dynamic, two-.

dimensional, frame analyses using the computer program DRAIN-2D2. For a detailed descrip-

tion of the workings of the program as well as input requirements refer to Golafshani [4]. As

part of this program, the multi-slice fiber element can be used with other available elements

such as beam-column, brace, truss and paneL elements. Since it is intended for the analysis of

two-dimensional structures of arbitrary configuration, there are three degrees of freedom per

node which can be fixed or slaved. Masses are lumped at the nodes resulting in a diagonal mass

matrix. Damping can be mass and/or stiffness proportional. Static loads may be applied prior to

the start of the dynamic analysis, but currently such loads should not cause inelastic deforma-

tions. Geometric stiffness (otherwise known as the pea effect) may be included, but is a func-

tion of the axial loads resulting from the static analysis; hence the geometric stiffness of an ele-

ment does not change with variations in the axial load which occur as the dynamic analysis

progresses. Finally, a major advantage of the program is the capacity of performing an event-
 -20-

to-event analysis - i.e. the structural stiffness is modified whenever an element undergoes a

change in stiffness due to a change in state. Depending on the element and the events con-

sidered such a procedure can reduce or eliminate the unbalanced forces which accumulate dur-

ing a time step and have t~ be corrected, or balanced at the end of every step, a process which

necessitates small time steps and may result in numerical instability.

2.2 Element and Slice Geometry

As shown in Fig. 2.1 the slice locations are specified at arbitrary points along the member.

That is, the slices need not be equally spaced. Furthermore, the slices are completely indepen-

dent which permits the modelling of a member with a variable cross section and/or material

properties. Fig. 2.2 shows a slice broken down into its component fibers. The user needs to

specify the fiber coordinates with respect to an arbitrary axis around which the moments are

later calculated. Note that there are no internal degrees of freedom. Hence the slice deforma-

tions and therefore fiber strains are calculated from the three local degrees of freedom which in

turn are calculated from the six global degrees freedom - see Fig. 2.3. Moreover, appropriate

transformations are also included to consider two rigid end zones of arbitrary length. Note,

however, that the member and the rigid end zones have to be on a straight line joining the two

end nodes as shown in Fig. 2.3.

In the current implementation of the element up to 15 slices can be used and their posi-

tions, chosen by the user, remain fixed throughout the analysis. The element flexibility is

assumed to vary linearly between the slices. With this in mind, slice locations should be

chosen to represent as accurately as possible the expected flexibility variation along the member

taking into account cracking, crushing, and yielding.

In the case of pure bending, for example, if the flexibility ( :/) is constant, then only two
slices at the ends of the member would be needed to accurately capture member behaviour. In

the case of reinforced concrete members, however, even for a linearly varying moment
 -21-

distribution the section flexibility varies along the member and changes dramatically when the

reinforcement yields. Generally the moment and, therefore, the curvature variation is unk-

nown a priori which makes the choice of optimum slice locations difficult. This is further com-

plicated by the fact that the flexibility distribution changes due to loading, unloading and load

reversal as well as moment-axial load interaction. In general, however, the model's accuracy

increases with an increasing number of slices. In the case of a reinforced concrete column

where the moment distribution is linear and the largest moments are expected to be near the

ends of the member, it would be reasonable to concentrate the slices near the member's ends

with additional slices near midspan. This is possible since the slices need not be located at reg-

ular intervals along the member's length. However, caution is required to insure that one cap-

tures an adequate picture of the flexibility variation along the member - see Fig. 2.4.

2.3 Theory and Workings of the Multi-Slice Fiber Model

This section describes the theory and the implementation of the multi-slice fiber model

representing the behaviour of reinforced concrete members. The various sub-sections describe

the assumptions and methods used for calculating fiber, slice, and element stiffnesses as well as

determining the shape functions.

Beam theory is the basis of this model. Plane sections are assumed to remain plane,

hence the strains in the various fibers can be easily calculated and the corresponding fiber

stresses and stiffnesses deduced from the cyclic stress-strain material relationships. The ele-

ment flexibility is arrived at by integrating slice flexibilities which are calculated by summation

over the fibers. The tangent element stiffness is calculated by inverting the element flexibility

and expanding from the three local degrees of freedom to the six global ones. The shape func-

tions relating slice deformations to element displacements are derived from the current slice

and element flexibilities.

 -22-

2.3 (a) Fiber Stiffness

Using the assumption of plane sections remaining plane, the strains in the various fibers

can be calculated once the strain at some reference point has been determined along with the

slice curvature. The equation giving the strain (10 i) in the i'th layer is:

(1)

where, strain at reference point (usually

plastic centroid or midheight)

slice curvature

Yi distance from centroid of i'th fiber

to the reference point

The sign convention governing the various quantities appearing in Equation (1) is given in Fig.

2.2. Note that compressive strains and stresses in the fibers are assumed positive.

Once the strains in the various fibers are determined, the stresses are arrived at from the

adopted cyclic stress-strain material relationships discussed later. The accuracy and generally

the complexity of the fiber model are a direct function of the adopted stress-strain relationships.

The tangent stiffness of the fiber is of course the slope of the stress-strain curve for the given

strain. Note that, as in the case of stress calculations, the stiffness (or slope) must be derived

giving due consideration to the previous strain history of the fiber and whether the fiber is load-

ing or unloading.

In the special case when the stress-strain curves are approximated by multi-linear seg-

ments, the stiffness would simply be the slope of the segment corresponding to the given strain.

Hence, the multi-linear stress-strain formulation reduces the amount of calculation involved

since the fiber stiffness is readily available without the need for differentiation. The multi
 -23-

linear formulation also simplifies the definition of an 'event' i.e. a significant change of fiber

stiffness. Such a change in any fiber's stiffness also leads to a change in the overall stiffness of

the member. To obtain 'exact' results the element stiffness should therefore be updated when-

ever any event occurs. This, however, can be computationally unfeasible especially if highly

complex material models are adopted and the element is part of a large reinforced concrete

frame undergoing a dynamic analysis. Because of this, the material models should be as simple

as possible while stilI retaining the significant physical properties.

The bilinear steel model described in Fig. 2.5 was found by Mark [39] to give sufficiently

accurate results provided that the inelastic deformations remained relatively small. For this

model there are only two possible stiffnesses, the initial (or elastic) and the post-yield (or strain

hardening) stiffnesses. Any transition from one to the other is considered an event, i.e. yield-

ing and unloading. More complex material behaviour can be modelled by superimposing

several bilinear steel fibers at the same location. The complexity of the material models must be

consistent with the other assumptions being made (Le. no shear deformations, no bond slip

etc.) and the computational effort. The bilinear model appears to be a reasonable compromise

for this application. The concrete model shown in Fig. 2.6 consists of a monotonic envelope

defined by five linear segments. The number of segments can be easily increased which would

result in a slight increase of storage requirements. As shown in Fig. 2.6 concrete is assumed

incapable of taking any tension. Unloading proceeds elastically and a crack opens once zero

stress is reached. The fiber can carry compressive stresses only upon the closing of the crack,

and reloading to the envelope curve again proceeds elastically.

It can be noted that there are many events (stiffness changes) associated with the concrete

model. It has been observed [39,44,45] that the precise details of the concrete stress-strain

diagram do not greatly affect the overall behaviour of a member, especially in the case of large

cyclic deformations. Concrete properties can often be simplified without adversely affecting

accuracy. Hence, to reduce computational effort this type of simplified concrete model may be

appropriate. Furthermore, the user has the option, of disregarding many of the events associated
 -24-

with the concrete fibers that might occur during a loading increment. Unloading, however, can

be retained as a 'significant' event. This reduction in the total number of events to be con-

sidered can be numerically advantageous, since every significant event requires the element's

and therefore the structure's stiffness to be updated.

By disregarding certain events it is clear that an approximate stiffness is being used. This

results in errors which can accumulate, such that the final results may be quite inaccurate. To

avoid this, an exact state determination for the element is used which does not ignore any of

the events. Once the state of the element is determined at the end of a step, the forces at the

ends of the member are calculated based on (1) the significant events and (2) all events and

any difference between them is corrected on the nodes at the ends of the member. This pro-

cedure is described in Section 2.3 (e).

2.3 (b) Slice Flexibility

Once the stiffness of the various fibers making up a given slice are determined, the slice

stiffness and therefore flexibility can be easily calculated. Two degrees of freedom per slice

along with the assumption of plane sections remaining plane are sufficient to define the state of

the slice. As shown in Fig. 2.2, the two degrees of freedom can be chosen as the curvature and

the strain at some reference axis, usually the plastic centroid (or simply the middepth of the

section). Slice bending moment (Ms ) around this reference axis and an axial load (Ps ) are

associated with these degrees of freedom. Incremental section forces are calculated using the

following pair of equations which express summations over the fibers in a slice:

dMs = I.A,E;,dE,y,
i
 -25-

where A, = area of fiber i

current stiffness of fiber i

and E" Yi as defined in Eq.(O

Making use of Eq. (0 these equations can be rewritten as:

(4)

(5)

Denoting the slice tangent stiffness by k s we have:

(6)

From Eqs. (4) and (5) it is clear that:

(7)

(8)

(9)

Hence the (2 x 2) tangent slice stiffness matrix is completely defined and the correspond-

ing slice flexibility matrix f s is given by simply inverting the stiffness matrix:

(10)
 -26-

2.3 (c) Element Stiffness

The two end rotations along with the axial displacement shown in Fig. 2.3 constitute the

three local degrees of freedom for the element (OJ, O2 , and A). Associated with these degrees

of freedom are the two end moments (MJ, M 2 ) and the axial load (P). Assuming linear varia-

tion of flexibility between slices, the element flexibility can be calculated by closed form

integration of section flexibility along the length of the element. This procedure is described in

this section.

The first step is to relate the slice forces Ss (Ms , Ps ) to the forces at the end of the

member Sm (MJ, M 2, P); in other words to determine the (2 x 3) matrix b in the equation:

(11)

This can be accomplished by straightforward application of the equations of equilibrium as

shown in Fig. 2.7. One therefore arrives at:

[
-1+ ~ ~ 0]
(12)
b= 0 0 1

The slice flexibility was developed in Section 2.3(b) and by assuming linear variation of flexibil-

ity along the member, the flexibility f s (x) of any section can be calculated by interpolation.

This section flexibility relates incremental slice deformations dVs (dc/J, dE p ) to the incremental
slice forces: dSs (dMS' dPs ) as follows:

(13)

Equations (11) and (13) along with the virtual work principal lead to:

(4)
 -27-

where f m the (3 x 3) element flexibility matrix is given by:

L
fm = Jo b T (x )f
s (x )b(x)dx (15)

By breaking up the integral over the individual slices and using natural coordinates with

the transformation defined in Fig. 2.8, Eq. (15) can be rewritten as:

(16)

where: ns number of slices used

= length of member between slices 'i' and 'i + l'

Hence the (3 x 3) element flexibility matrix relating end displacements to end forces can

be calculated. The element stiffness matrix can be obtained by simple inversion of the flexibil-

ity matrix:

(17)

Finally the element tangent stiffness can be expanded to the complete element (6 x 6) stiffness

matrix which includes rigid body displacements and a transformation applied so that the final

stiffness matrix relates global displacements to global forces. This transformation takes into con-

sideration rigid end zones of arbitrary length.

2.3 (d) Shape Functions and State Determination

One of the advantages of the multi-slice fiber model is the fact that there are no

predefined interpolation or shape functions. The shape functions used to calculate slice defor-

mations from member end displacements are a function of the slice flexibilities and are there-

fore continually updated as the state of the element evolves [46].

 -28-

Combining Eqs. OI), (3), and (4) leads to

(8)

Therefore the shape functions which give slice deformation increments in terms of ele-

ment deformation increments are given by:

(9)

These shape functions can be used to calculate the slice deformations and hence to update the

state of the slices and consequently the element. For a differential increment in slice deforma-

tions the shape functions as derived are exact. For finite deformation increments, however, the

shape functions would only be exact if the state determination and updating phase is divided

into sufficiently small steps so that the shape functions are changed every time a material event

occurs - e.g. the yielding of a steel fiber or the cracking of a concrete fiber, etc. Even if this

procedure is implemented, one should keep in mind that the state determination phase would

be exact only in so far as the adopted linearized material stress-strain relationships represent the

actual material behaviour.

2.3 (e) Unbalanced Forces

If a strict event-to-event approach is adhered to - Le. if the structure's stiffness is

modified whenever any of the elements and therefore fibers changes state - then equilibrium

would always be satisfied. The axial load would be constant along the element and the bending

moment would vary linearly. To save on computations, however, the user may ignore the

stiffness changes during a time step and apply corrective forces at the end of the step. Conse-

quently, equilibrium will not necessarily be satisfied at the end of the step. While iterative stra-

tegies exist to correct for this, they are incompatible with the desired computational efficiency.

An alternative that may be acceptable if the size of the unbalance is relatively small is to add to
 -29-

the next time step's load vector a set of self-equilibrating corrective forces. These corrective

forces should equal the difference between the 'nonlinear' or actual element forces at the end

of the step and the 'linear' forces that result from the assumption of constant element stiffness

throughout the time step. The 'nonlinear' forces are obtained by going through a strict event-

to-event element state determination given the state of the element at the beginning of the

time step and the additional displacements imposed on the element during the time step in

question. This is done in all cases and the 'nonlinear' forces become the starting point for the

subsequent step. The 'linear' forces on the other hand are simply equal to the initial element

forces plus the element stiffness at the beginning of the step multiplied by the additional dis-

placements. Note that in the case of an analysis using 'significant' events, 'linear' forces are

updated whenever a 'significant' event occurs and not merely at the end of a given time step.

If no events occur for a given element during a certain time step then the be no need to apply

corrective forces. If an event does occur a correction is required to prevent a divergence from

the true solution.

 -30-

III ELEMENT VERIFICATION

3.1 Introduction

This chapter deals mainly with the verification of the multi-slice fiber element's operation.

The element can be used separately, in which case it is capable of tracing the slice (as well as

the fiber) deformations and forces given the history of displacements at the ends of a single

member. The model can also be used as part of DRAIN-2D2 [4] to analyze the dynamic

response of two-dimensional frames. Several analyses will be carried out to illustrate these two

uses of the element. The effect of various parameters, such as the number of fibers per slice,

the number and location of slices, and the size of the time step used will be investigated. Sim-

ple analyses will be used and the results compared with theoretical, analytical, or experimental

values. The analyses include static monotonic and cyclic loadings of a steel beam and dynamic

sinusoidal linear elastic analyses of a column. Further verifications and analysis of reinforced

concrete columns and frames will be included in later chapters.

3.2 Monotonic and Cyclic Loading of a Steel Beam

The steel beam used has a five inch by twelve inch rectangular section. As shown in Fig.

3.l(b) the beam analyzed is fixed at both ends and one end is moved vertically relative to the

other. Hence, the load-deflection curve for a beam loaded as shown in Fig. 3.l(c) can be

derived. These are then compared to theoretical results based on an elasto-perfectlY plastic

steel stress-strain formulation.

For the purpose of comparison, the yield moment for the given section is 7200 lb-in. and

the plastic moment is 10,800 lb-in. The corresponding yield load is 288 lb. and the ultimate

load is 432 lb. The theoretically derived load-deflection relation is given in Fig. 3.2 assuming

the geometry and the material properties given in Fig. 3.1 and a steel strain hardening slope of
 -31-

zero - Le. elasto-perfectly plastic steel.

Results of a multi-slice fiber model analysis of this beam are given in Fig. 3.3 which show

the effect of the assumed strain hardening slope. The value of this second branch of the

stress-strain curve is alternately set at 10.0, 100.0, and 1000.0 Ksi. As expected, the pre-yield

behaviour is not affected, but the load required to produce equal displacements beyond first

yield is higher for higher strain hardening rates.

Because of possible numerical problems, it is advisable to use a small second slope instead

of an elasto-perfectly plastic model. In this case a minimum second slope of 10.0 Ksi was used

which amounts to 0.033% of the initial slope - 30,000 Ksi. As expected the results of the

analysis using a 10.0 Ksi second slope are therefore essentially identical to the theoretically

derived curve based on an elasto-perfectly plastic steel model and given in Fig. 3.2.

Fig. 3.4 shows the effect of varying the number of fibers per slice. These analyses all

divide the member into equidistant slices along it's length and use a hardening slope of 1000.0

Ksi. The number of fibers per slice, however, is changed from 6 to 12 and 18. Note that there

is a slight difference between the analyses using 6 and 12 fibers. Since the yield load is a func-

tion of the distance between the neutral axis and the centroid of the extreme fiber, the yield

load drops from 336 lb. to 312 lb. Increasing the number of fibers further - from 12 to 18 per

slice - again results in a decrease in the yield load - from 312 lb. to 304 lb. However, the shape

of the curve does not change appreciably and the curves corresponding to 12 and 18 fibers per

slice nearly coincide.

The number and positions of the slices were held constant in the previous comparisons.

Fig 3.5 indicates the effect of varying the number and position of the slices along the length of

the member while keeping the number of fibers per slice constant at 12. Using the analysis

with eleven equidistant slices as a reference, note that the number of slices used does not make

a difference as long as the beam is behaving elastically. As yielding starts to spread, the larger

the number of slices, the better the results. Concentrating slices in the region of the anticipated

hinging is highly advantageous. However, it is essential to have some slices in the elastic
 -32-

portion of the beam to capture the elastic flexibility distribution.

Finally, Fig. 3.6 indicates the capacity of the model to consider cyclic behaviour. The dis-

placement is cycled between extremes of -4.0 and 4.0, inches and the figure gives the

corresponding P - A variation. Keeping the basic assumptions in mind, the figures presented

so far are "exact" since the model takes into consideration the yielding and unloading of each

fiber in every slice. The steel stress-strain relationship is bilinear. Consequently the Bausch-

inger effect is not in evidence. This absence can be detected in Fig. 3.6.

3.3 Dynamic Analysis of a Linear Elastic Column

A steel column fixed at the base and restrained from rotation at the top is subjected to the

sinusoidal ground acceleration record shown in Fig. 3.7. The amplitude of the ground accelera-

tion is scaled in such a way as to insure that the behaviour of the column throughout its length

remains elastic. This simple analysis is performed, since theoretical results are available for

easy comparison. Furthermore, a parametric study of the effect of the analysis time step can be

performed.

The acceleration record is discretized into steps of 0.025 seconds. The time step used in

the analysis, however, is varied between 0.005 and 0.015 seconds. Noting that the natural

period of the system (T) is 0.15, the analysis time step is thus varied between 3~ and 1~'

The results are given in Fig. 3.8 along with the theoretical curve. This indicates satisfactory

correlation so long as the integration time step is sufficiently small.

 -33-

IV ANALYSIS OF REINFORCED CONCRETE COLUMNS AND FRAMES

4.1 Introduction

Although the multi-slice fiber element can be used to model members of a steel frame,

the main concern of this report is its capabilities for modelling reinforced concrete frames. The

element appears especially well suited for modelling the behaviour of reinforced concrete

columns when the axial load is appreciable or fluctuates significantly with time. Of course,

shear considerations must not be dominant and other elements must be used to model defor-

mations associated with bond slip in connections. The ability of the multi-slice fiber element to

account for the influence of an axial load on a member's strength and stiffness characteristics as

well as on the shape (pinching) of hysteresis loops will be illustrated in the following sections.

Results of dynamic analyses of several reinforced concrete columns and frames are examined.

Comparisons are made with results of other analytical studies to identify the capabilities of the

method.

4.2 Dynamic Analyses of a Reinforced Concrete Column

To check the functioning of the multi-slice fiber model several parametric studies were

run on a reinforced concrete cantilever column subjected to horizontal base excitation. Various

modelling and analysis assumptions were considered to assess their effect on the accuracy of the

element and to develop some insight into the dynamic response of simple reinforced concrete

structures.
 -34-

4.2(a) Example Column

The column selected, shown in Fig. 4.1, has been previously analyzed and tested [39,44].

The geometry and material properties of the column assumed in the analyses are given in Figs.

4.1 and 4.2. The bilinear steel model was used with a yield stress of 48.4 Ksi, an initial slope

of 29000 Ksi and a strain hardening slope of 300 Ksi. The concrete stress-strain curve shown

in Fig. 4.2 closely follows the suggested relationship for the confined portion of the column sec-

tion [39]. Since spalling did not occur in previous analytical and experimental analyses, large

concrete strains were not anticipated and the same concrete model was used to represent the

behaviour of the unconfined portion. This observation was born out by the results of these

analyses. For purposes of later comparisons, the maximum axial load capacity of the section

P max (given by Ae . J'e + As . I y ) is equal to 301 Kips.

Unless otherwise specified, the section was discretized into eight equal concrete layers and

six slices were considered along the length. Since the moment diagram expected for this

column varies linearly from a maximum at the base to zero at the top, slice locations were con-

centrated in the vicinity of the base. Hence for this standard case the slices were spaced at 2.0-

3.0-5.0-5.0-5.0 inches starting at the base.

The excitation used in most of the analyses consisted of a simple sinusoidal base accelera-

tion record with a period of .075 seconds shown in Fig. 4.3. This was used to develop

significant deformation reversals. Some analyses were performed using the N-S component of

the 1940 El Centro acceleration record. The analysis time step used was 0.00125 seconds (or

approximately 1/38 of the apparent period of the column).

4.2(b) Influence of Modelling and Discretization

The first parameter to be considered was the number of slices along the length of the

member. Two runs were made, one using eleven slices uniformly spaced at 2.0 inches and
 -35-

another using the standard six slices. To make sure the steel yielded, the acceleration record

amplitude was set at 2.0g. In both cases the column was subjected to an axial load of 33.5 Kips

or P max / 9 in order to produce relatively complex hysteresis behaviour. The tip displacement

histories are compared in Fig. 4.4(a), while Fig. 4.4(b) compares the moment-curvature his-

tories developed at the base of the column. It is clear from Fig. 4.4(b) that the element is able

to trace the very complex hysteretic loops developed by this column. The global displacement

response and moment-curvature response obtained using only six slices are very close to the

results obtained using eleven equally spaced slices. This is reasonable since the flexibility varia-

tion in regions of lower bending moments would be less dramatic (or even linear) and hence

can be adequately captured by a relatively small number of slices. Slices should be located in

regions where significant variations in flexibilities are likely to occur rather than arbitrarily or

uniformly spaced along the member. Because of the satisfactory results provided by the six-

slice model, the following analyses use this configuration as standard.

The second modelling parameter of interest is the number of concrete fibers considered

across a slice. Fig. 4.5 shows a comparison of two analyses, one discretizing the section into

eight and the other into sixteen layers of equal depth. Again, the axial load applied to the sec-

tion was 33.5 Kips and the sinusoidal acceleration record with an amplitude of 2.0g was used.

The tip displacement histories (given in Fig. 4.5 (a corresponding to the two analyses are in

close agreement. The moment-curvature histories for the base of the column (given in Fig.

4.5 (b differ to a somewhat larger extent. As expected, the sixteen fiber discretization results

in relatively smoother stiffness transitions. The tip displacement history, however, reflects the

overall stiffness of the member and in regions of 10w bending moment - which remain essen-

tially elastic - the number of fibers is not of great consequence. Hence, the variation of tip dis-

placement is less sensitive to the number of fibers per slice than is the moment-curvature his-

tory at the base of the column. It appears that a relatively small number of fibers can be used

to adequately represent the section behaviour.

 -36-

4.2(c) Influence of Geometric Stiffness

As mentioned earlier DRAIN-2D2 [4] has the capacity of including simple geometric

stiffness (P - a effect) based on the axial loads resulting from a static analysis. Fig. 4.6 shows

results of two analyses, one ignoring and the other including geometric stiffness. The

geometric stiffness in this case is a function of the applied axial load which is constant

throughout the analysis and equal to 16.75 Kips or P max / 18. Note that in the initial elastic

phases the two curves show no significant difference. Later, however, the extent of inelastic

behaviour increases as evidenced by the elongation of the period, and the effect of geometric

stiffness correspondingly increases.

4.2 (d) Influence of Acceleration Record Amplitude

Fig. 4.7 compares the tip displacement histories resulting from sinusoidal acceleration

records scaled to 1.0g and 2.0g amplitudes. As expected the larger amplitude record results iri

larger displacements. Furthermore, while the lower amplitude record exhibits periodic motion,

the higher amplitude record does not. This can be explained by the fact that the behaviour of

the column subjected to the I.Og record is essentially elastic, though nonlinear, as shown in Fig.

4.8 (a). This figure shows the moment-curvature history at the base of the column - the loca-

tion of the largest moments. A definite cracking load is observed, since the column is sub-

jected to a compressive axial load. The steel, howe~er, does not yield for this excitation. Upon

reversal of a displacement excursion, open cracks on one side of the section close, while cracks

start opening on the other side. This results in the nonlinear elastic moment-curvature

behaviour indicated. Fig. 4.8(b) shows the moment-curvature history corresponding to the

higher amplitude record. Clearly the steel yields in this case and the behaviour is no longer

elastic. Because of the inelastic behaviour, the response is no longer periodic.

 -37-

4.2(e) Influence of Applied Axial Load

The axial load applied to the column can also significantly change its behaviour. Using

the 1.0g sinusoidal record three runs were made as the axial load varied from 0.0 to 16.75 Kips

and finally to 33.5 Kips. .The mass, however, was held constant. The tip displacement histories

are given in Fig. 4.9(a) and indicate an increase in maximum displacement with a decrease in

axial load. This is essentially due to the fact that for this range of loading the stiffness increases

with the applied load. Furthermore, note that the displacement history corresponding to the

33.5 Kip load is periodic and the behaviour elastic as indicated by the moment-curvature plot

given in Fig. 4.9(b). The moment-curvature plots corresponding to the lower loads, however,

indicate steel yielding and hysteresis loops developing and growing as the axial load decreases -

see Figs. 4.9(c) and 4.9(d). This can be explained by the fact that for this range of loading the

yield moment increases with an increase in axial load.

4.2(f) Influence of Analysis Technique

In the preceding analyses the event-to-event procedures discussed in Chapter II were

used. That is, all stiffness changes occurring during an integration time step are detected and

accounted for in determining the response. Since the member is represented by several slices,

each consisting of several fibers, the number of potential events during a single time step is

very large and could result in excessive computational effort. DRAIN-2D2 [4] has the option

of disregarding events during time steps - i.e., the structure's stiffnesses at the beginning of the

time step is assumed constant throughout the step, irrespective of actual element stiffness

changes. The stiffness is updated at the end of the step and equilibrium errors (unbalanced

forces) are then corrected for at the beginning of the next step.

Two analyses were carried out to illustrate these procedures using a 2.0g amplitude

sinusoidal acceleration record (shown in Fig. 4.3). An axial load of 33.5 Kips was applied to
 -38-

the column. The first analysis proceeded strictly event-to-event, that is the column's stiffness

was updated whenever any of the fibers changed stiffness. In the second analysis such midstep

'events' were disregarded until the end of the time step. The corresponding two tip displace-

ment histories are given in Fig. 4.10(a). Note that the difference is not appreciable for the time

step used and while the analysis disregarding midstep events took two minutes, the one consid-

ering all events took five minutes - these and the following analyses were made using a VAX

750 machine. Fig. 4.10(b) compares the corresponding moment-curvature histories at the base

of the column. Note that updating the stiffness whenever a fiber changes state results in a

smoother curve, but that the maximum values and the overall shape of the hysteresis loops is

not appreciably affected.

As a means of cutting computation effort while still achieving satisfactory results, an addi-

tional option was incorporated into the multi-slice fiber model for use in conjunction with the

event-to-event approach of DRAIN-2D2. This option disregards 'insignificant' events. For the

purposes of this element the only significant event considered for the concrete is unloading.

Any change in a steel fiber's stiffness is considered a significant event, since the behaviour of

the reinforcing steel has a greater influence on section behaviour. Fig. 4.11(b) shows a com-

parison of the tip displacement histories corresponding to a strict event-by event analysis and an

analysis which considers only significant events. The difference is not appreciable while the

analysis times were five minutes and two and. a half minutes, respectively. Fig. 4.11(a) shows

the corresponding moment-curvature histories. Note that the maximum values and overall

shapes are quite similar. It is also interesting to note that including 'significant' events results

in a smoother curve than disregarding all midstep events - compare Figs. 4.10(b) and 4.1l(a) -

without an appreciable increase in computational effort.

 -39-

4.2 (g) Analysis Using EI-Centro Record

To illustrate the capabilities of the element under earthquake type excitations, the exam-

ple column was also subjected to 6 seconds of the 1940 N-S EI-Centro record. A constant axial

load of 16.75 Kips was assumed to act on the column throughout the duration of the excitation.

The analysis time step used was 0.0025 seconds. The analysis was performed considering all

events, only significant events and no midstep events. Figs. 4.12 and 4.13 compare results of

the three analysis techniques. Note that for the small time step used, the tip displacement his-

tories are very close.

Fig. 4.14 shows a typical moment-curvature hysteretic loop from the three analyses.

Although ignoring midstep events results in some discrepancies, the envelope values and

energy dissipation are not greatly affected. Compared to a complete event-to-event analysis,

considering only significant events and ignoring all midstep events reduced the computation

time by 50% and 75%, respectively.

The results of the analysis using the multi-slice fiber model will now be compared with

results of a more conventional analysis. Since the column has a single lateral degree of free-

dom, the program NONSPEC [47] could be used to compute the response. This program con-

siders single degree of freedom systems with a bilinear restoring force envelope. Both bilinear

hysteretic and stiffness degrading [47] hysteretic characteristics may be considered.

The multi-slice fiber model was used .to calculate the monotonic lateral load-displacement

relation of the example column with an axial load of 16.75 Kips. A bilinear envelope curve was

approximated as shown in Fig. 4.15 for use with NONSPEC. As shown in Fig. 4.16 for small

displacements, the bilinear and stiffness degrading models give essentially identical results.

Later, however, after significant yielding and reversals, the response histories diverge, although

the number of zero crossings remains the same.

Fig. 4.17 compares the response of the multi-slice fiber model with those of the bilinear

and the stiffness degrading models. Table 1 compares the maximum and minimum
 -40-

displacements given by the three models.

Type of Model: Multi-slice fiber Bilinear Stiffness Degrading

Max. Disp. 0.058 0.047 0.052

Min. Disp. -0.075 -0.083 -0.067

Table 1. Extreme displacement values using various models

For the initial low acceleration input phase the bilinear and stiffness degrading models overesti-

mate the response. This is due to the fact that the bilinear envelope curve on which these

models are based does not account for cracking, and thus, as illustrated in Fig. 4.15, a lower

initial stiffness results. Later, when significant yield excursions occur the stiffness degrading

model better approximates the response given by the multi-slice fiber model. Nonetheless, the

results are significantly different, most likely due to the highly pinched hysteretic loops exhi-

bited by the column which are not accounted for in the standard stiffness degrading model

incorporated in NONSPEC. The simple models, moreover, do not provide information on the

local behaviour such as moment-curvature histories. Since the behaviour of concrete elements

in even the 'elastic' range is complex due to cracking and axialload effects, simple models are

not likely to produce reliable results.

4.3 Dynamic Analysis of a Reinforced Concrete Frame

Results of analyses performed on a one-bay, one-story, reinforced concrete frame will be

given in this section. The effects of geometric and material modelling assumptions wiIl be

investigated as well as analysis technique and time step used.

 -41-

4.3 (a) Example Frame

The reinforced frame shown in Fig. 4.18 has been previously used for experimental [48]

and analytical f39J studies. As shown in Fig. 4.18 the frame made up of two 5 in. square

columns and a 10 in. by.5 in. beam. The reinforcement in columns and beam consists of 4 # 3

bars giving a reinforcement ratio of 1. 76% in the columns (based on gross section area). Aver-

age values obtained in material tests [48] were used to idealize the primary steel and concrete

stress-strain curves. The models used are given in Fig. 4.19.

For the experimental study conducted by Gulkan and Sozen [48], plates weighing 4.0 Kips

were bolted to the beam at midspan. For the analyses to follow, this has been idealized as two

2.15 Kip loads (the additional weight accounts for beam weight, etc.) placed at the two joints.

Hence, each column is assumed to be subjected to an initial static axial load of 2.15 Kips and a

corresponding concentrated mass is lumped at the beam-column joints.

The sinusoidal base acceleration record used in the analyses is shown in Fig. 4.20. An

attempt has been made to match as closely as possible the record used in Mark's study of the

same frame [39] - thus, an amplitude of 1.21g was used. Unless otherwise stated, the analyses

are carried out using a time step of 0.00125 seconds. For later comparisons, Fig. 4.21 shows

the result of analyses carried out by Mark using time steps of 0.0005 and 0.001 seconds and 20

slices for each element. Mark ignores stiffness changes during a time step and does not apply

equilibrium corrective forces at the end of the step. This explains the somewhat large difference

between the calculated responses for the two time steps. The following two sections will use

the frame described above to study the effects of modelling and analysis assumptions on the

calculated response.
 -42-

4.3 (b) Effect of Some Modelling Assumptions

As discussed earlier, the multi-slice fiber model avoids the need to define moment-

curvature relations, moment-rotation relations or yield surfaces. Nevertheless, various material

and geometric modelling decisions need to be made. The sensitivity of the calculated response

to variations in these modelling assumptions is investigated in this section. In particular the

influence of the steel strain hardening stiffness, the number of slices used to discretize an ele-

ment and the assumption of perfectly rigid joints are studied.

Fig. 4.22(a) compares the tip of column 1 displacement histories assuming strain harden-

ing slopes of either 300 or 1000 ksi. Note that a higher strain hardening slope does not neces-

sarily result in smaller displacements. In fact the analysis based on the higher value results in a

half-cycle maximum sway of 0.4 in. compared to 0.36 in. for the lower value of strain harden-

ing rate. This may be explained by the fact that changing the stiffness also changes the effective

period which affects the dynamic response. The corresponding moment-curvature plots for the

base of column 1 are compared in 'Fig. 4.22(b). Note that a higher strain hardening rate results

in a stiffer second branch as expected. However, as illustrated in Fig. 4.22(b) a higher strain

hardening rate also results in larger hysteretic loops in this case which explains the larger dis-

placements - compare points C and C' in Figs. 4.22(a) and 4.22(b).

To assess the importance of the number of slices used along the length of the elements,

two analyses were carried out : one using 5 slices per element and the other 9 slices per ele-

ment. In both cases, however, the spacing was not uniform: the first two slices at each end of

an element were placed closer together since that is were the greatest moments and flexibility

variations are expected. In the case of the columns, the end slices were spaced at 3.75 and

0.875 inches apart for the 5-slice and 9-slice idealizations, respectively. Fig. 4.23 (a) compares

the two displacement histories and Fig. 4.23(b) the corresponding moment-curvature histories

at the base of column 1. Again, note that considerable differences in moment-curvature his-

tories do not necessarily result in correspondingly large differences in displacements. For

 -43-

example, this can be seen by comparing the loops ABC and A'B'C' in Figs. 4.23 (a) and

4.23 (b). This is due to the fact that the slice spacings in the two analyses are not the same,

hence comparing the moment-curvature histories of just one slice is not sufficient for compar-

ing the overal1 element behaviour. The displacement histories are not greatly different, how-

ever. Furthermore, compared to the analysis reported by Mark using 20 equal1y spaced slices

and a time step of 0.0005 seconds (see Fig.4.2I(b the analysis using only 5 slices gives very

good results. Final1y, using 5 slices as opposed to 9 considerably reduces analysis time. Hence,

the 5 slice configuration was adopted for the fol1owing analyses.

Another modeJling decision to be made is whether to consider the joints rigid or not. In

fact the joints are never perfectly rigid, but are considered such in certain analyses as a matter

of convenience. Sources of flexibility - such as shear, bond slip etc. - can be partial1y accounted

for by extending the elements to the center of the the corresponding joints. Fig. 4.24 compares

the displacement histories based on the flexible and rigid joint assumptions. Note that assuming

rigid joints results in an increased initial stiffness, reduced displacement and period. Later, how-

ever, after cycling and significant yielding the two histories come closer together in terms of

displacements and periods. A precise analysis of joint deformations including bond slip would

require development of a special element to be included in the model.

4.3(c) Influence of Analysis Technique and Time Step

As discussed and i11ustrated earlier, DRAIN-2D2 [4] has the capacity of carrying out

event-to-event analyses which take into consideration every change of stiffness in any of the

elements. In addition, the multi-slice element model was developed such that certain changes in

fiber stiffness are considered 'significant', namely yielding and unloading of steel fibers, and the

unloading of concrete fibers. Figs. 4.25 and 4.26 compare the results of analyses based on con-

sidering al1 events, 'significant' events, and ignoring events altogether during the time step. The

sinusoidal record given in Fig. 4.20 along with' an analysis time step of 0.00125 seconds are
 -44-

used. Note that the displacement response histories (Fig. 4.25) almost coincide, although the

analysis times were 40, 8 and 6 minutes for the strict event-to-event, 'significant' events, and

no events analyses respectively. Figs. 4.26(a) and 4.26(b) compare the moment-curvature his-

tories at the base of column 1. Clearly the variation is not very significant, although taking the

'significant' events into consideration (Fig. 4.26(a)) gives closer results and a smoother

moment-curvature history than ignoring events altogether (Fig. 4.26(b)).

The preceding discussion of the influence of analysis technique illustrates the fact that

significant computational effort may be saved by taking into account only significant stiffness

changes or even ignoring all stiffness changes during the time step ( and applying equilibrium

corrective forces at the end of the step ). As Fig. 4.25 illustrates this may not result in

significant loss of accuracy. Care must be taken, however, to insure that the corrective forces

do not grow to such an extent that their application leads to numerical instability. Usually, the

larger the time step the larger the unbalance and the greater the risk of numerical instability. In

addition, numerical inaccuracies result from the use of large time steps [4]. To illustrate this,

the same analyses used to produce Fig. 4.25 were repeated using a time step of 0.00625 seconds

( or TI7.6 ) as opposed to the time step of 0.00125 seconds used previously. The results con-

sidering all events, 'significant' events and no events are given in Fig. 4.27. Note that the

analysis ignoring all events diverges, while the analysis considering 'significant' events remains

stable although the accuracy diminishes with the increased time step size - compare Figs. 4.25

and 4.27. Thus, the size of the time step should be chosen to attain the required accuracy and

insure numerical stability. The event-to-event techniques usually mean a larger computational

effort than the no events technique for the same time step size. On the other hand, such tech-

niques permit the use of a larger time step with acceptable accuracy and no numerical problems.
 -45-

4.3 (d) Effect of Axial Load

The action of an axial load significantly changes the behaviour of a reinforced concrete

section. As discussed and illustrated previously, the stiffness, especially before yielding, and the

yield moment are functions of the axial load acting on the section. Furthermore, a compressive

axial load can result in pinched hysteresis loops. To illustrate these phenomena and their effect

on the response of a reinforced concrete frame, the example frame of Section 4.3 (a) was

reanalyzed for the sinusoidal record given in Fig. 4.20 considering a different column axial load.

The same time step (0.00125 seconds), material, and geometric idealizations were used for this

analysis. For the previous runs, an axial load of 2.15 Kips due to gravity effects was assumed to

act on each column. For the new run, the axial load was arbitrarily increased to 25.4 Kips, or

P maxl6. Both of these gravity loads are below th~ balanced axial load which is about 55 Kips

for this column. The mass of the structure was unchanged in the analysis.

The results of the two analyses are shown in Fig. 4.28 which gives the lateral displacement

as well as column base moment-curvature histories. Note that, as expected, the higher axial

load results in a larger initial stiffness and in this case a higher yield moment. These effects are

illustrated in the first branch of the moment-curvature history. The higher initial stiffness and

yield moment corresponding to the higher axial load case also mean that the displacements of

the frame would tend to be lower than for the frame whose columns were subjected to the

higher axial loads. This is apparent in Fig. 4.28(a) - compare the displacements up to point A.

Later, however, the higher axial load results in appreciable pinching of the hysteresis loops as

shown in Fig. 4.28(b) which in turn means reduced stiffness during significant portions of the

response and larger displacements as shown in Fig. 4.28(a).

 -46-

4.3 (e) Comparison with Other Models

The two analyses presented in the previous section, corresponding to gravity axial loads of

2.15 and 25.4 Kips, were also carried out using a more conventional reinforced concrete

member model available in DRAIN-2D [2]. This model is a one component element with con-

centrated end springs having rotational characteristics based on a simplified Takeda stiffness

degrading model. The simplified Takeda model assumes a bilinear moment-rotation relation for

the concentrated springs. It requires as input the axial stiffness (assumed constant), the yield

moment, and the initial as well as strain hardening flexural stiffnesses. Member moment-

curvature relations were computed using UNCOLA [44] and the two required flexural

stiffnesses were established by ignoring the cracking transition. The displacement histories for

an axial load of 2.15 Kips calculated using the multi-slice fiber element and the concentrated

spring element based on the simplified Takeda stiffness degrading model [2] are compared in

Fig. 4.29. As illustrated in this figure, ignoring the initial precracking stiffness does not

significantly affect the response, since the axial load is very low. The stiffness degrading model

gives satisfactory results except in the final phase of the response where, as shown in Fig.

4.28(b), the results from the multi-slice fiber model indicate a small but noticeable pinching of

hysteresis loops.

For the higher axial load case, the required yield moment and stiffnesses were deduced

from the moment-curvature relation corresponding to the specified axial load. The results of

the analyses using the multi-slice fiber model and the concentrated spring model are given in

Fig. 4.30. Note that for this higher axial load (25.4 Kips) ignoring cracking results in a larger

initial discrepancy. The significant pinching of hysteresis loops observed in Fig. 4.28(b) results

in greater difference between the responses than for the case with the lower axial load. The

same analysis was carried out using the two-component parallel model incorporated in DRAIN-

2D. This element represents members exhibiting bilinear hysteretic moment-rotation relations.

The results obtained with this element are shown in Fig. 4.31. Comparing Figs. 4.30 and 4.31
 -47-

indicates that the single component element based on the stiffness degrading model leads to

better prediction of the response than the two-component parallel element based on the simpler

bilinear model.

The response to 4 seconds of the N-S component of the El Centro record were also com-

puted using the multi-slice fiber model and the simplified Takeda model. Since the frame is a

scaled down model, the acceleration amplitudes were magnified by a factor of 4 to produce

yielding. Fig. 4.32(a) compares the results of the analyses. Note that by ignoring the higher ini-

tial precracking stiffness, the concentrated spring model leads to higher calculated displacements

in the initial low amplitude acceleration phase. Although the two response curves are similar in

shape, they differ appreciably as far as extreme displacement values are concerned. It is

interesting to note that at point A in Fig. 4.32 (a) the responses start to diverge significantly and

that the corresponding point on the moment-curvature history - point A Fig. 4.32(b) - is where

the hysteresis loops are most severely pinched.

4.3(f) Effect of a Varying Axial Load

The previous two sections analyzed the dynamic response of the example frame when its

columns were subjected to an essentially constant axial load. This section will demonstrate the

capacity of the multi-slice fiber model to predict the behaviour of reinforced concrete members

when subjected to a varying axial load.

The same example was used for the analysis except that a concentrated mass correspond-

ing to five additional stories was introduced. The mass was placed at the center of gravity of the

additional structure and was assumed to be rigidly connected to the tops of the columns - see

Fig. 4.33. In addition, the static axial loads acting on the columns were increased to 12.75 Kips

to account for the five additional stories. The model was subjected to the same sinusoidal

record - see Fig. 4.20 - and the response was calculated using the multi-slice fiber model as well

as the two-component parallel model and ihe modified Takeda models. The resulting
 -48-

displacement histories are given in Fig. 4.34. Note that results from the two-component model

deviate considerably from those of the fiber model, although the two-component model takes

into account the variation of yield moment with applied axial load. The two-component model

does not, however, account for changes of stiffness in the 'elastic' range that result from varia-

tions in axial load. The stiffness within the yield surface is fixed and was set equal to the initial

cracked stiffness under a load equal to the static axial load. The modified Takeda model results

in closer agreement as far as displacement history is concerned. The Takeda model does not

account for variations of stiffness or yield moment with axial load. The superior results of the

Takeda model when compared to the two-component parallel model are mainly due to the fact

that the modified Takeda model accounts for stiffness degradation.

Figs. 4.35-4.37 give additional results of the analysis using the multi-slice fiber model and

summarize the behaviour of the columns. Fig. 4.35 traces the history of the axial load varia-

tions in the two columns. As expected, the sum of the two axial loads is constant, as equili-

brium would require. Hence, as the compressive load in one column increases the load in the

other decreases which results in significant variations in the bending moments and shears to

which the columns are subjected - see Fig. 4.36. Note that initially at least the column sub-

jected to larger compressive loads receives a larger portion of the base shear. This would be a

consequence of the effect of axial load on stiffness. Fig. 4.37 gives the moment-curvature his-

tories at the bases of the two columns. Note that variations in axial load produce frequent and

significant variations in stiffness. This results in complex moment-curvature relationships which

the multi-slice fiber model can trace.

Fig. 4.38 summarizes the behaviour of the columns as given by the analysis based on the

Takeda model. As Fig. 4.38(a) shows, the sum of the axial loads in the two columns remains

constant. The history of axial load in the individual columns is similar to the history prediced

by the multi-slice fiber model - compare Figs. 4.35 and 4.38(a). There are differences, however,

which are mainly due to the fact that the Takeda model assumes constant axial stiffness while

the multi-slice fiber model updates the axial stiffness to reflect the state of the element.
 -49-

Furthermore, the multi-slice fiber model takes into consideration the axial load-bending

moment interaction. This also explains the significant differences in the predicted shear history

variation in the two columns - compare Figs. 4.36 and 4.38(b). Since the Takeda model

assumes the flexural stiffness to be independent of the axial load and stiffness, the shears in the

two columns are identic.al throughout the analysis. A more realistic shear distribution is given

by the multi-slice fiber model as illustrated in Fig. 4.36. For example, when the maximum base

shear is attained the column in compression carries 70% of it while the column in tension car-

ries the remaining 30%. Finally, it might be worth mentioning that information concerning local

behaviour cannot be obtained using discrete models. The multi-slice fiber model, however, can

provide such information as illustrated by the various moment-curvature histories.

 -50-

V CONCLUSIONS

Various models have been proposed for the analysis of reinforced concrete members and

structures ranging from very simple single degree of freedom models to finite element discreti-

zations. The multi-slice fiber model seems to be a satisfactory alternative for modelling ele-

ments of a reinforced concrete frame. The model requires as input material stress-strain rela-

tionships for the various fibers and provides as output detailed history of local and member

behaviour including curvatures, rotations, axial deformations as well as member forces. The

model can account for axial load-bending interaction, thus variations in yield moment and

flexural stiffness due to the presence and change in axial load can be considered. As i11ustrated,

the model can also trace complex hysteresis loops which show the effects of crack opening and

closing as well as pinching due to a compressive axial load acting on the member. The various

features of the multi-slice fiber model make it ideally suited for modelling reinforced concrete

columns where axial load-bending interaction is important.

The limitations of the multi-slice fiber model arise from the assumptions on which it is

based as well as the adopted concrete and steel stress-strain relations. Hence, the assumption of

plane sections remaining plane ignores the effects of shear and bond slip on section geometry.

The model, moreover, does not consider the shear-flexural interaction and is therefore not suit-

able for modelling members where shear considerations are dominant. The bilinear steel model

implemented for this study has the virtue of simplicity, but does not realistically model strain

hardening and does not account for the Bauschinger effect. Finally, due to the large number of

fibers and the need to trace the history of each of the fibers, the required computational effort

can be considerable especially in the case of dynamic event-to-event analyses.

This study has shown the multi-slice fiber model to be a flexible analysis tool. However,

various improvements and uses can still be investigated. For example, the steel model can be

modified to account for local buckling and the Bauschinger effect, the use of the model in static

load-to-collapse analyses can be studied, and the extension of the model for three dimensional

analyses can be investigated. To reduce computational effort, parametric studies can be

 -51-

performed to establish guidelines for efficiently locating fibers and slices and for judiciously

choosing an analysis time step. Finally, a simplified beam model can be adapted for use in cases

where axial load-bending interaction does not significantly affect behaviour.

 -52-

REFERENCES

0) Otani, S., "SAKE - A computer program for inelastic response of RIC frames to earth-
quakes," Civil Engineering Studies, Structural Research Series No. 413, University of Illinois,
Urbana, November 1975.

(2) Kanaan, A.E. and Powell, G.H., "DRAIN-2D - A General Purpose Computer Program for
Dynamic Analysis of Inelastic Plane Structures - With User's Guide," EERC Reports No. 73/6
and 73/22, University of California, Berkeley, April 1973, Revised September 1973 and August
1975.

(3) Saiidi, M., "User's Manual for the LARZ Family - Computer Programs for Nonlinear Se-
ismic Analysis of Reinforced Concrete Planar Structures," Civil Engineering Studies, Structural
Research Series No. 466, University of Illinois, Urbana, November 1979.

(4) Golafshani, A.A., "DRAIN-2D2 - A Computer Program for Inelastic Seismic Response of
Structures," Ph.D. Dissertation, Dept. of Civil Engineering, University of California, Berkeley,
1982.

(5) Mondkar, D.P. and Powell, G.H., "ANSR-II Analysis of Nonlinear Structural Response,
User's Manual," EERC Report No. 79/17, University of California, Berkeley, July 1979.

(6) Bathe, K.J., "ADINA - A Finite Element Program for Automatic Dynamic Incremental
Nonlinear Analysis," M.LT. Report 82448-1 (Revised 1978).

(7) Aziz, T.S., "Inelastic Dynamic Aaalysis of Building Frames," Publication No. R76-37, Dept.
of Civil Engineering, M.LT., Cambridge, August 1976.

(8) Umemura, H., Aoyama, H., and Takizawa, "Analysis of the Behaviour of Reinforced Con-
crete Structures During Strong Earthquakes Based on Empirical Estimation of Inelastic Restor-
ing Force Characteristics of Members," Proceedings of the Fifth World Conference on Earth-
quake Engineering, Rome, Italy, June 1973, Vol. 2, pp. 2201-2210.

(9) Takizawa, H., "Non-Linear Models for Simulating the Dynamic Damaging Process of Low-
Rise Reinforced Concrete Buildings During Severe Earthquakes," Earthquake Engineering and
Structural Dynamics, John Wiley & Sons, Chichester, Sussex, England, Vol. 4, No.1, July 1975,
pp. 73-94.

(0) Aoyama, H., "Simple Nonlinear Models for the Seismic Response of Reinforced Concrete
Buildings," Proceedings of the Review Meeting of the U.S.-Japan Cooperative Research Pro-
gram in Earthquake Engineering, Honolulu, Hawaii, August 1975, pp. 291-309.

(11) Pique, J.R., "On the Use of Simple Models in Nonlinear Dynamic Analysis," Publication
No. R76-43, Dept. of Civil Engineering, M.I.T., Cambridge, September 1976.

(12) Saiidi, M. and Sozen, M.A., "Simple and Complex Models for Nonlinear Seismic Response
of Reinforced Concrete Structures," Report to the National Science Foundation, Dept. of Civil
Engineering, University of Illinois,. Urbana, August 1979.
 -53-

(3) Giberson, M.F., "The Response of Nonlinear Multistory Structures to Earthquake Excita-
tion," Earthquake Engineering Research Laboratory, California Institute of Technology, Pasade-
na, May 1967.

(4) Otani, S. and Sozen, M.A., "Behavior of Multistory Reinforced Concrete Frames during
Earthquakes," Civil Engineering Studies, Structural Research Series No. 392, University of Illi-
nois, Urbana, November 1972.

05) Clough, R.W. and Johnston, S.B., "Effect of Stiffness Degradation on Earthquake Ductility
Requirements," Proceedings, Japan Earthquake Engineering Symposium, Tokyo, October 1966,
pp. 195-198.

(6) Takeda, T., Sozen, M.A., and Nielsen, N.N., "Reinforced Concrete Response to Simulated
Earthquake," Journal of the Structural Division, ASCE, Vol. 96, ST12, December 1970, pp.
2557-2573.

(7) Takayanagi, T. and Schnobrich, W.e., "Computed Behavior of Reinforced Concrete Cou-
pled Shear Walls," Civil Engineering Studies, Structural Research Series No. 434, University of
Illinois, Urbana, December 1976.

(8) Emori, K. and Schnobrich, W.e., "Analysis of Reinforced Concrete Frame-Wall Structures
for Strong Motion Earthquakes," Civil Engineering Studies, Structural Research Series No.457,
University of Illinois, Urbana, December 1978.

(9) Fintel, M. and Ghosh, S.K., "Case Study of Seismic Resistance of a 16-Story Coupled Wall
Structure Using Inelastic Dynamic Analysis and an Energy Dissipation Approach," Portland Ce-
ment Association, Skokie, Illinois, April 1979, 67.

(20) Chen, P.F. and Powell, G.H., "Generalized Plastic Hinge Concepts for 3D Beam-Column
Elements," EERC Report No. 82/20, University of California, Berkeley, November 1982.

(21) Clough, R.W., Benuska, K.L., and Wilson,E.L., "Inelastic Earthquake Response of Tall
Buildings," Proceedings of the Third World Conference on Earthquake Engineering, Auckland
and Wellington, New Zealand, January 1965, Vol. II, pp. 68-84.

(22) Clough, R.W., Benuska, K.L., and T.Y. Lin and Associates, "F.H.A. Study of Seismic
Design Criteria for High-Rise Buildings," HUD TS-3, U.S. Housing and Urban Development,
Federal Housing Administration, Washington D.C., August 1966.

(23) Clough, R.W. and Benuska, K.L., "Nonlinear Earthquake Behavior of Tall Buildings,"
Journal of the Engineering Mechanics Division, A.S.e.A., Vol. 93, No. EM3, June 1967, pp.
129-146.

(24) Mahin, S.A., "An Evaluation of the Seismic Response of Reinforced Concrete Buildings,"
Ph.D. Dissertation, Dept. of Civil Engineering, University of California, Berkeley 1974.

(25) Anderson, J.C. and Bertero, V.V., "Seismic Behavior of Multistory Frames by Different
Philosophies," EERC Report No. 69/11, University of California, Berkeley, October 1969.

(26) Powell, G.H. and Row, D.G., "Influence of Analysis and Design Assumptions on Comput-
ed Inelastic Response of Moderately Tall Frames," EERC Report No. 76/11, University of Cali-
fornia, Berkeley, April 1976.
 -54-

(27) Klingner, R.E. and Bertero, V.V., "Infilled Frames in Earthquake-Resistant Construction,"
EERC Report No. 76/32, University of California, Berkeley, December 1976.

(28) Takizawa, H., "Notes on Some Basic Problems in Inelastic Analysis of Planar RIC Struc-
tures (Parts I and II)," Transactions of the Architectural Institute of Japan, No. 240, February
1976, pp. 51-62; No. 241, March 1976, pp. 65-77.

(29) Hsu, L.W., "Behavior of Multi-Story Reinforced Concrete Walls During Earthquakes,"
Ph.D. Thesis, Dept. of Civil Engineering, University of Illinois, Urbana 1974.

(30) Umemura, H. and Takizawa, H., "Strong Motion Response Analysis of RIC Frames,"
Transactions of the Architectural Institute of Japan, Extra, 1972.10.

(31) Meyer, C, Arzoumanidis, S.G., and Shinozuka, M., "Earthquake Reliability of Reinforced
Concrete Buildings," Proceedings of the Symposium on Probabilistic Methods in Structural En-
gineering, Edited by Shinozuka, M. and Yao, IT.P., ASCE, New York 1981, pp. 378-398.

(32) Arzoumanidis, S.G. and Meyer, c., "Modeling Reinforced Concrete Beams Subjected to
Cyclic Loads," Dept. of Civil Engineering and Engineering Mechanics, Columbia University,
New York, Report No. NSF-PFR-79-24695-CU-l, 1981.

(33) Soleimani, D., Popov, E.P., and Bertero, V.V., "Nonlinear Beam Model for RIC Frame
Analysis," Seventh Conference on Electronic Computation, ASCE, St. Louis, Missouri, August
6-8, 1979.

(34) Agrawal, A.B., Jaeger, L.G., and Mufti, A.A., "Crack Propagation and Plasticity of Rein-
forced Concrete Shear Walls under Monotonic and Cyclic Loading," Conference on Finite Ele-
ment Methods in Engineering, Adelaide, Australia, December 1976.

(35) Schnobrich, W.C., "Behavior of Reinforced Concrete Structures Predicted by the Finite
Element Method," An International Journal, Computers and Structures, Vol. 7, No.3, June
1977, pp.365-376.

(36) Suidan, M. and Schnobrich, W.C., "Finite Element Analysis of Reinforced Concrete,"
Journal of the Structural Division, ASCE, Vol. 99, No.STlO, Proc. Paper 10081, October 1973,
pp. 2109-2122.

(37) Yuzugullu, O. and Schnobrich, W.C., "A Numerical Procedure for the Determination of
the Behavior of a Shear Wall Frame System," Journal of the ACI, Proc., Vol. 70, No.7, July
1973, pp. 474-479.

(38) Cervenka ,V., "Inelastic Finite Element Analysis of Reinforced Concrete Panels Under
In-Plane Loads," Ph.D. Dissertation, Dept. of Civil Engineering, University of Colorado,
Boulder, 1970.

(39) Mark, K.M.S., and Roesset, J.M., "Nonlinear Dynamic Response of Reinforced Concrete
Frames," Publication R76-38, Department of Civil Engineering, M.LT., August 1976.

(40) Iding, R., Bresler, B., and Nizamuddin, Z., "Fires-RC II, A Computer Program for the
Fire Response of Structures-Reinforced Concrete Frames - Revised Edition," Fires Research
Group, University of California, Berkeley, July 1977.

(41) Bazant, Z.P., and Bhat, P.D., "Prediction of Hysteresis of Reinforced Concrete Members,"
Journal of the Structural Division, ASCE, Vol. 103, No. STl, January 1977, pp. 153-167.
 -55-

(42) Bazant, Z.P. and Bhat, P.D., "Endochronic Theory of Inelasticity and Failure of Concrete,"
Journal of the Engineering Mechanics Division, ASCE, Vol. 102, No. EM4, Proc. Paper 12360,
August 1976, pp. 701-722.

(43) Filippou, F.C., Popov, E.P., and Bertero, V.V., "Effects of Bond Deterioration on Hys-
teretic Behavior of Reinforced Concrete Joints," EERC Report No. 83119, University of Cali-
fornia, Berkeley, August 1983.

(44) Kaba, S.A. and Mahin S.A., "Interactive Computer Analysis Methods for Predicting the
Inelastic Cyclic Behaviour of Structural Sections," EERC Report No. 83118, University of Cali-
fornia, Berkeley, July 1983.

(45) Ma, S.M., Bertero, V.V., and Popov, E.P., "Experimental and Analytical Studies of the
Hysteretic Behavior of Reinforced Concrete Rectangular and T-Beams," EERC Report No.
76/2, University of California, Berkeley, May 1976.

(46) Mahasuverachai, M. and Powell, G.H., "Inelastic Analysis of Piping and Tubular Struc-
tures," EERC Report No. 82/27, University of California, Berkeley, November 1982.

(47) Mahin, S.A. and Lin, J. ,"Construction of Inelastic Response Spectra for Single-Degree-
of-Freedom Systems," EERC Report No. 83117, University of California, Berkeley, July 1983.

(48) Gulkan, P. and Sozen, M.A., "Response and Energy Dissipation of Reinforced Concrete
Frames Subjected to Strong Base Motions," Civil Engineering Studies, Structural Research
Series No. 377, University of Illinois, Urbana, May 1971.

(49) Thorn, C.W., "The Effect of Inelastic shear on the Seismic Response of Structures," Ph.D.
Dissertation, Dept. of Civil Engineering, University of Auckland, New Zealand, March 1983.
 320

P=2P max/9
240

.-..
.SI
C.
J2
--E 160 P=P max/18
Q)

E
o
~ ,
VI
0"-
,
p=o

80

o' ' , , , , !

o 0.6 1.2 1.8 2.4 3.0 3.6

Curvature 00- 3 radlin)

Fig. 1.1 Influence of axial load on section strength and stiffness

 240
P .. - Pmal 18 <tension)
c:
-.- 120
I
a.
~

t-
o
Z
w
:::E -120
o
:::E
-240
-1.0 0 -3 1.0
CURVATURE (10 rod/in)

240 ,
P .. P ma/18
c:
-I
120
a.
~

- O
t-
Z
W
~ -120
0
:::E
-240
-1.0 0 1.0
-3
CURVATU RE (10 rod/ in)

Fig. 1.2 Effect of axial load on cyclic moment-curvature behaviour

 ~}--QM.q
Ceq 8~+A8~
Nonlinear Spring Elastic" I ~
Ma +6M a
L eq 8A +68A --
)
Fig. 1.3 The Q-Model

3 Fig. 1.4 Single component or concentrated spring model

l
.
VI
00
I

Primary Curv.

I
I
Deformation I
.. I
I
I I
I
I
I
I
I
Be Br By
u' Rotation

Fig. 1.5 Takeda hysteresis model Fig. 1.6 Trilinear model - primary curve
 .
~

/ J_ I Deformation '? V { Deformation

> y >

Fig. 1.7 Bilinear hysteresis model Fig. 1.8 Otani hysteresis model
I
V>
p '0
I
J u", I
(a) Multiple spring model
Deformotion
-

u;.
(b) Idealized curvature distribution

Fig. 1.9 Q-Hyst model Fig. 1.10 Multiple spring element for modelling wall members
 -60-

slices to be located by user

Fig. 2.1 Slice locations along the element length

ep,M
Reference Axis _-t-_-+-...._ E p , p

Fig. 2.2 Typical slice showing fibers and sign convention

5
Node J

(a) Local D.O.F.'s (b) Global D.O.F.'s

Fig. 2.3 Local and global degrees of freedom

 123 567 2 5 6

4 3 4

(a) Actual flexibility variation (b) Flexibility variation (c) Superior flexibility variation
using 7 slices using only 6 slices

Fig. 2.4 The importance of a suitable selection of slice locations

I
0'\
I
80 -
I I
E sh
4
40
~
~
.-.
'iii 'r;;
... ~
~ '-"
l/l 0
l/l ~ 2
W u.l
It: ~
...
l/l
I-< I
V)
-40/- /' I .--
01 '~lJ 1
f,
-80 1
, I -I' I , "

-0.004 -0.002 0 0.002 0.004 0.006 0.008 -0.002 o 0.002 0.004

STRAIN (in/in) STRAIN On/in)

Fig. 2.5 Bilinear steel model Fig. 2.6 Concrete model

 -62-

p'=p
S
M s = (M 1 + M2 ) x/L - M 1

(M 1 +M 2 )/L
Ms
(II J,j-P :j-.PS
MIl'-
M2
P
-E1'I
M,
L 1 I x

Fig. 2.7 Relating section forces to member forces

SLICE i+ I

x-o.s (Xi-I+ Xi)

e= Lj

Fig. 2.8 Transformation to natural coordinates

 -63-

r
h=12"
60 ksi

El =30,000 ksi

E
0.002
b =5"
(a) Section geometry and assumed stress-strain relation

j--~t A=--pf---:Ar
(b) Displacement controlled loading
l.--- L =100" I
..

(c) Equivalent loading

Fig. 3.1 Section geometry, material properties and loading configuration

600

--:e 400

" t:)
ctl
0
....:l
200

o
o 2 3 4 5
Displacement (in.)

Fig. 3.2 Theoretical load-displacement curve assuming EPP steel

 -64-

600
.-.---
-_ ..... --
.-_ .. -- -- --
500
-- -- -- -- --
.. -- -- -----------------
.. ,""::.""_-a.=:=::.::.:~F==:..:;.::;;:.;..-----
400 compare Fig. 3.2
,-..
=?
'-'
"0
ell 300
0
...J Strain Hardening Rate
of Bilinear Steel Model
200
- - - 2 = 10.0 Ksi
2 = 100.0 Ksi
100
2 = 1000.0 Ksi

o
o 2 3 4 5
Displacement (in,)

Fig. 3.3 Influence of rate of strain hardening (second slope of stress-strain curve)

600

500

400
,-..
:e
'-'
"0
ell 300
0 Number of Fibers per Slice
,.J

------- N=6
200
_. _.- N = 12

- - - N -= 18
100

o
o 2 3 4 5
Displacement Gn.)

Fig. 3.4 Influence of number of fibers per slice used to discretize the section
 -65-

600

500

400

-go 300
....:l Number and Placement of Slices
5 Equidistant slices
200
7 Equidistant slices
11 Equidistant slices
100 5 Slices, 3 near midspan

o
o 2 3 4 5
Displacement Gn,}

Fig. 3.5 Influence of number and position of slices along the member

400

200
,-..
=i
'-'
"0
~
0
0
....:l

-200

-400

-5 -3 -I 0 I 3 5
Displacement Gn.)

Fig. 3.6 Cyclic load-displacement curve using bilinear steel model

 1.0
.....
~ 0.5
c::
o
';j 0
e
~

'8 0.5
< 1.0

o 0.3 0.6 0.9 1.2 1.5

Time (sec.)

Fig. 3.7 Acceleration record used to analyze elastic column

- - Analysis --- - Theory I

0'>
0'>
I

41=0.005 sec. (41/T=0.033) 41=0.010 sec. (41/T=0.067) At =0.015 sec. (4t/T=0.D

1.20 j I

...... 0.12 ,:
.: ..:.
,
,, ,
...
~
..:. :~
; \1 : ~
,
~ 0.24
~ .: ~, .f \
..
: 'I

~
~
Q.
0
-0.24
,,
,,
II
en
is .
-0.12 , ,
..
~ :
\~.
.. ','
-1.20 I ! I I I , I I I

o 0.4 0.8 1.2 o 0.4 0.8 1.2 o 0.4 0.8 1.2

Time (sec.)

Fig. 3.8 Influence of the analysis time step used

 -67-

m
--,...-
4#4BARS - ~

~

,
#2(ci 2.0' ........
...... 20"
8"

1.25"
.L .
,----
11 1+-, ---:..:.-=..=:---..
4.83" I
(a) Section Geometry (b) Column Showing Standard Slice Spacing

Fig. 4.1 Example column

Stress (Ksi) Stress (Ksi)

6.95

5.21
48.4

3.04

1.39

0.135 0.27
Strain On/in)
Strain (xlO- 2 in/in)

(a) Concrete (b) Steel

Fig. 4.2 Idealized material properties of example column

 -68-

1.0

--
'00
c::
o
0.5
.~

~ 0
"i>
< -0.5
t.l

-1.0

o 0.45 0.91 0.37 1.83 2.28

Time (l0- 1 sec.)
Fig. 4.3 Acceleration record used to analyze reinforced concrete column

0.4
Number of Slices
=
-c:: 0.2 ------ N
N
=
6
11
-
-.::;
cQ)::
E
Q)
t.l 0
~
C.
rIl
is
-0.2

-0.4 L.-_ _- l - L.-_ _-.L..._ _---' --L..._ _......J

o 0.05 0.10 0.15 0.20 025 0.30

Time (sec.)
(a) Tip Displacement History
270 . . . - - - - - - - - - - - , - - - - - - - - - - - - ,
Number of Slices
180 --N=6

-
.5
I
90
------ N .. 11

--
C.
~
o
..,
c::
Cl)

,,
~ -90


-180

,
i:L_------lI""o:::;...-~
-270 L--_ _........_ _---'- L--_ _....L-_ _- - L_ _---J

-1.2 -0.8 -0.4 0 0.4 0.8 1.2

Curvature 00- 2 rad/in)
(b) Base moment-curvature history

Fig. 4.4 Analysis of reinforced .concrete column using 6 and 11 slices

 -69-

0.4
Number of Fibers
--N=8

0.2 N = 16
-
~
c

-0.2

-0.4
o 0.05 0.10 0.15 0.20 0.25 0.30
Time (sec.)
(a) Tip Displacement History

270
Number of Fibers I
I
--N=8 I
180 ,
I

----- N = 16
I

,,
, I

,,
-C
'i'
.9-
90
,
,,
I
I

-
'-'
o --'
cQ,) ,--- --------
,
E
,,,
~ -90 . ,
,
I
I

,,,
I
,
I

, ,,,
,,, ,
-180 ,
,,,
-270
----------
-1.2 -0.8 -0.4 0 0.4 0.8 1.2
Curvature 00- 2 radlin)

(b) Base moment-curvature history

Fig. 4.5 Analysis of reinforced concrete column using 8 and 16 fibers per slice
 -70-

0.06 r-------------------------,

0.04

0.02

-
cG)::
E
G)
0
~
Q.
en
o -0.02

-0.04 Ignoring Geometric Stiffness

Including Geometric Stiffness
-0.06
o 0.05 0.10 0.15 0.20 0.25 0.30
Time (sec.)
Fig. 4.6 Including geometric stiffness in the analysis of a reinforced concrete column

0.4
P-P max/9

0.2
,-..

-.5-
cG)::
E
G) 0 .. --
j
Q.
en
0
-0.2

Acceleration Amplitude := 1.0g

Acceleration Amplitude := 2.0g .
-0.4
o 0.05 0.10 0.15 0.20 0.25 0.30
Time (sec.)
Fig. 4.7 Analysis of reinforced concrete column using different acceleration amplitudes
 -71-

150
P=P ma J9
100

.-.,
c 50
"
c..
~

-
'-'

c
eu
E
0
0
~
-50

-100

-150
-0.3 0.2 -0.1 0 0.1 0.2 0.3
Curvature 00- 3 radlin)
(a) Acceleration Amplitude = l.Og

208
P==P ma J9

- .5I
c..
104

-
'-'

ceu
E
0
0
~

-104

-208

-1.0 -0.6 -0.2 0 0.2 0.6 0.1

Curvature (10- 2 radlin)

(b) Acceleration Amplitude == 2.0g

Fig. 4.8 Moment-curvature history using different acceleration amplitudes

 0.2 I I 150, i I
Axial Load
----- P==O.O
," ...\ 100
,, - P==P ma J9
-; 0.1
...
,,, _.-. P=Pmax/ 18 '"'t:
.S
'-"
,, 50
.. "6.
..... I""", ~
:.Q
c:: I, \ ~
" , , '-'
E
~
0 i5
Cl)
0
~ eo
Q.
II)
~ -50
is -0.1
-100

-0.2 . I , I I I I
-150
o 0.05 0.10 0.15 0.20 0.25 0.30 -0.3 -0.1 0 0.1 0.3
Time (sec.) Curvature 00- 3 rad/in)
I
(a) Displacement histories for various axial loads (b) Moment-curvature history for axial load = P maJ9 -...l
IV
I

140 120

"2 70 '"'
.9 60
" I
Q,
c.
~ :.Q
'-" '-'

c= 0 i5 0
E
o
e
o
~ -70 ~ -60

-140 -120

-1.0 -0.5 0 0.5 1.0 -4 -2 0 2 4

Curvature (10- 3 radlin) Curvature 00- 3 rad/in)
(c) Moment-curvature history for axial load = Pma J18 (d) Moment-curvature history for axial load = 0.0

Fig. 4.9 Analysis of reinforced concrete column subjected to different axial loads
 -73-

0.4

0.2

-c
;,:::.
-c(I)
E 0
(I)
u
CIl
C.
VJ
(5

-0.2 Analysis Technique

Event-to-Event
- - - - Ignoring Events

-0.4
o 0.05 0.10 0.15 0.20 0.25
Time (sec.)
(a) Tip displacement history

270

180

-c 90 , I

'6- ,
I

~
"-'
0 ,--
I
I
I
,
I

, I

-90 I
,
,,,
,, Analysis Technique
-180 I - - - Event-to-Event
,
I

I
I - - - - - Ignoring Events
I

-270
-0.0/2 -0.008 -0.004 0 0.004 0.008 0.012
Curvature (rad/in)

(b) Base moment-curvature history.

Fig. 4.10 Effect of ignoring all changes in fiber stiffness between steps
 -74-

270
I
I

180 ,
I

, I

I
I
I
I
I
,-... 90 I
.5 I

c..
I , I

-....
32
r::
Q)
0
.,
I

E
0
I
I

:; I
I
-90 I
I

I
I
I
I
I
I
-180 I
I
- - - Event-to-Event

I Ignoring Certain Events
-270 "---_ _I . -_ _---IL...-._ _...--I"--_ _---" ---" --'

-0.012 -0.008 -0.004 0 0.004 0.008 0.012

Curvature (rad/in)
(a) Base moment-curvature history
0.4

0.2
,-...
d
0
....r::
Q)

E
Q)
0
Co)
tU
Q.
II)

is
-0.2
Event-to-Event
Ignoring Certain Events

-0.4
o 0.05 0.10 0.15 0.20 0.25
Time (sec.)

(b) Tip displacement history

Fig. 4.11 Effect of ignoring certain changes in fiber stiffness between steps
 -75-

0.06

0.03 .
~
I-

.
-....
c
c:: 0
~)VJ IN
~'
"V
V"
,
c::
~
E
.
\
~
t,)
ell -0.03 l-
is..
en
C5 . I

. I

Analysis Technique
-0.06 I- ,I
- All Events
-- - --- Significant Events
I I I I I
-0.09
o 2 3 4 5 6
Time (sec)
Fig. 4.12 Response to EI-Centro considering all events and significant events

0.06

.
0.03 - ~
.-..
~
c::
0
~JW iN
~I
V"
c::
~
E
g . I
~ -0.03 -
C5
I

Analysis Technique
-0.06 -
-- All Events
-- .. -- No Events
I I I I I
-0.09
o 2 3 4 5 6
Time (sec)

Fig. 4.13 Response to EI-Centro considering all events and no events

 -76-
180 __
Analysis Technique
AlI Events
---h~
" , ,/
120
_._._. Significant Events "
, ,
/
,,' I'
----- No Events
, ,/ "
-
"
c:
c..
60
"
, ,/
, ,/
"
--
~

cQ,):
0
~
,~
,

,/
I
I
/

E
0
~

-60

-120

-180
-1.5 -1.0 -0.5 0 0.5 1.0 1.5
Curvature 00- 3 rad/in)

Fig. 4.14 Hysteretic loop from response to El Centro using various techniques

12.5.-----------------------------,
P=Pmax/18

10.0 -
_ 7.5
~

V]J-j~b.
~
'-'

>
5.0 I I
/ I 20"

2.5
- - - - Multi-Slice Fiber Model
- - - Bilinear Idealization

o
0.0 0.1 0.2 0.3 0.4 0.5
~ (in)

Fig. 4.15 Load-displacement relation and corresponding bilinear idealization

 -77-

0.06 r-, --------~-----I

0.03 -
-
c
c
o

-0.03 -
\
Type of Model '1' ~ i,;
-0.06 - B.I. I
'I --- 1 mear l t

- - - - - Stiffness Degrading Ii
-0.09 L - i_ _- ' - - - -_ ---L-
_ . . l - -_ _. L -_ _'----_ _

o 3 2 4 5 6
Time (sec)
Fig. 4.16 Response to EI Centro using bilinear and stiffness degrading models

-...
c
c
c
~
E
~
U
te
c..
til
(b)
is
0.03 :-

o
-0.03

4 5 6

Fig. 4.17 Response to EI Centro using simple models and multi-slice fiber model
 -78-

( I) (2) 26"

'f'
5
", .. I
,

54
..
"',"
~I "
I
5
.1

(a) Frame dimensions

5"
5" J---....I ---.,._
- 'J

II
7.95 4#3

(b) Column section

-~
--
(c) Beam section

Fig. 4.18 Geometry of Gulkan frame - series F

6,.--------------. 60 ~------------,

,....
'Vi
~
40
"-"

~
rtl

-~
t I'}
2 t I'} 20

0.008 0.016 0.024 0.006 0.012

Strain Gn/in) Strain Gn/in)

(a) Steel stress-strain relation (b) Concrete stress-strain relation

Fig. 4.19 Idealized material properties for analysis of Gulkan's frame

 -79-

1.21g

c
.2
~
I-
0
Qj
0
u
u
-<
1.21g

0 0.275
Time (sec)

Fig. 4.20 Sinusoidal record used to analyze Gulkan's frame

0.3

--. 0.2
.5
-e
'-"

c
0 0.1
0
u
=
Q. 0
ell

is
-0.1

-0.2
0 0./ 0.2 0.3 0 0.1 0.2 0.3
Time (sec)
(a) A t = 0.001 seconds (b) A t = 0.0005 seconds

Fig. 4.21 Mark's analyses of Gulkan's frame using two different time steps
 -80-

0.300 . . - - - - - - - - - - - - - - - - - - - - - - - . ,
Strain Hardening Rate B

2 = 300 ksi
0.225 _.-.- 2 = 1000 ksi

___ 0.150
c:
=-
-
t:
Q)
F'

~ 0
~
Q.
II)

is
-0.150

A,A'

-0.300 L -_ _--L -L ...l.- I - -_ _---l- .....

o 0.05 0.\0 0.\5 0.20 0.25 0.30

Time (sec)
(a) Tip displacement history

60
Strain Hardening Rate
2 = 300 ksi
40 ----- 2 = 1000 ksi

,-...
c 20
"
Q.

-- ~

c 0 ,
,,,
u
E
0
,
~
-20 I
,,
I
,
I

,
I

I
-40 , I

B .. -.--- .- .. -
:.. -------
-60 B'
L..-_......JI...-_--"_ _--.J._ _--'-_ _-L.._ _...J..._ _...&.-_----l
-0.004 -0.002 o 0.002
Curvature (rad/in)

(b) Moment-curvature history

Fig. 4.22 Influence of steel strain hardening rate on calculated response

 -81-

0.300
Number of Slices B
N = 9
_._.- N = 5
0.225

-
c
0.150
-E
~

cCl)
Cl)
u
to:! 0.075
Q.
Vl F'
is
0

-0.075

-0.150
o 0.05 0.10 0.15 0.20 0.25 0.30
Time (sec)
(a) Tip displacement history
60r----------------.
Number of Slices
. . . .- --------.,
E A
---N=9
-'-'- N = 5
40
,
-
"
c
c.
20
I
/
I
/

--
~

cCl)
o t-----t'----/f------+----f-,rH---f:'---t--------t /
,/

E
o I

~ I
/
-20 I
/
I
I
I
/
-40 /
I
8'
o
B
-60 I - -_ _. . . -_ _...-._ _........_ _......_ _ ~ _ ___&.._ _---l-_ __ _ I

-0.004 -0.002 0 0.002

Curvature (radlin)
(b) Moment-curvature history

Fig. 4.23 Influence of number of slices used on calculated response

 -82-
0.3 , - - - - - - - - - - - - - - - - - - - - - - - - - - - ,
Joint Idealization
- - - Flexible
- - ' - ' Rigid
0.2

\ I

\

-0.1 \
\

\
\ /
\,

-0.2 '-----~----'------...-..----'------'------'
o 0.05 0.10 0.15 0.20 0.25 0.30
Time (sec)

Fig. 4.24 Influence of joint idealization on calculated response

0.300 ~------------------------,
Analysis Technique
- - - All Events
_. - ' - Significant Events
0.225
------ No Events

--
.5 0.150
-E
cQ):

~ 0.075
Q.
fI}

i5
l~ 0

-0.075

- 0.150 '---:--_---'- --I.- ..J.- ' - -_ _--L. -I

o 0.05 0.10 0.15 0.20 0.25 0.30

Time (sec)

Fig. 4.25 Influence of analysis technique on displacement response (A t = .00125 sec)

 -83-

60 r---------------,---------,
Analysis Technique
All Events
40 _._._. Significant Events

.,c..
C 20

-c::
.Jw:

Q)
0
E
o
::E
-20

-40

-60 '--_----.lL....-._--L_ _--L_ _--L._ _-1.._ _- l -_ _ ....l-_--I

-0.004 -0.002 0 0.002
Curvature (rad/in)
(a) Considering all events vs. only significant events

60 r--------------r---------.,
Analysis Technique
All Events
40 _._.- No Events

---
c 20
.6-
-c
:.;:

Q)
0
I
E I
o I
::E I

-20 /
I
I
I
/
-40
\
/ '
-60 ' - - _ - - - I_ _--L._ _--L-_ _.....l..-_ _.L-_--IL...-_--'-_ _--I

-0.004 -0.002, 0 0.002

Curvature (radlin)

(b) Considering all events vs. no events

Fig. 4.26 Influence of analysis technique on calculated moment-curvature response

 -84-
0.300 . - - - - - - - - - - - - - - - : " " - - - - . . ,
Analysis Technique
- - All Events r""
0.225 _._. Significant Events /-'.\
, \'
---- No Events "~
-
c:
Q)

E
0.150
\\
Q)

~
0.075 ~\
',I
C.
til
'.j
','
i5 o
-0.075

-0.150 L-_----J_ _--L_ _---.L_ _. . . . J - _ ' - - - - - ' - - _ - '

o 0.1
Time (sec)
0.2 0.3
Fig. 4.27 Influence of large time step size on calculated response
0.4
Axial Load
- p = 2.15 Kips
---- P = 25.4 Kips
0.2
-cQ):

i
c:.
til
o
i5
-0.2

-0.4 O"-----L--O-.L..,--...l.----0--.J.-:----"----7"'O.3
i
Time (sec)
(a) Displacement History
100 r - - - - - - - - - , r - - - - - - :=::::=-.=1
-------------~~~::::7-------~:
.--------- ,,
: ,:
,,.
,,
- c:
I
Co
50 ,, ,
,,

... .,,,.
,
,
:;2 ,.
'-'

-cQ):
E
o
o
.. .--
,1
.4
---~
,
.. --------
,

" "
~
-50 ..""
II

.'
,..'
I'
Axial Load
- - p = 2.15 Kips
//
!.~ =::- ..
--,)'
---- P = 25.4 Kips
-100 "---.....I----J.-_...l-_-L .l.----'----:-::'
-0.30 o Q30 O~O
Curvature (10- 2 radlin)
(b) Moment-curvature history
Fig. 4.28 Dynamic response of Gulkan frame for two values of axial loads
 -85-
0.30
P=2.15 Kips - - Multi-Slice Fiber
---- Concentrated Spring

- s::

-
'=
sll)::
E
0.15

g
0.
til
Q 0

-0.15
0 0.1 ) 0.2 0.3
Time (sec
Fig. 4.29 Response using multi-slice fiber model and concentrated spring model
0.4 r - - - - - - - - - - - - - - - - - - - - - - - ,
P=25.4 Kips

C 0.2
c

-0.2
- - Multi-Slice Fiber
---- Concentrated Spring
-0.4 l - -_ _.L.--_ _- ' - -_ _- ' - -_ _- ' -_ _......!-_ _- - '

o 0.1
Time (sec)
0.2 0.3
Fig. 4.30 Response using multi-slice fiber model and concentrated spring model
0.4 r - - - - - - - - - - - - - - - - - - - - . ,
P = 25.4 Kips

--
.5
0.2

- s::
ll)

E
8Cl:l o
Q.
til
Q
-0.2
- - Multi-Slice Fiber
---- Two-Component Parallel
-0.4 l - - _ _.L.-_ _- ' - -_ _- ' -_ _- ' -_ _--L._ _- - '

o 0.1 0.2 0.3

Time (sec)
Fig. 4.31 Response using multi-slice fiber model and two-component parallel model
 -86-

0.30

0.15

.
-...
c:
;;
0
,\
,, "
, ' ""
t: ",' a , I ,\
I1l
E -0.15
.,, :",
I,.' f I
I
II 1. " c
I ., It I ' "
, ""."'t ,I,
'I " ' , ,t,""
I1l
g
is.
"" . I'"
I "'"
'I I I ,
I ' \I \,
f
J

,, .
I I , '.
11, II
"

" ~ ~I,
C5'" -0.30
I "" I:

:, I'" I, ~ ,

Type of Model
-0.45
------ Multi-Slice Fiber
- - Concentrated Spring
-0.60
o 2 3 4
Time (sec)
(a) Displacement History

75

50

,-...
c:
"
c. 25

--
~
c:
u
E 0
0
~

-25

-50
-0.3 o 0.3 0.6
Curvature (10- 2 rad/in)

(b) Moment-curvature history

Fig. 4.32 Response to EI-Centro using multi-slice fiber and concentrated spring models
 -87-

77.5 "

12.75 kips 12.75 kips

COLUMN (I) COLUMN (2) 31.0"

r- 29.5"
~..
29.5"
.1

Fig. 4.33 Example frame with concentrated mass to simulate additional stories

0.1
- - Multi-Slice Fiber
_._.- Takeda
------ 2-Component Parallel
o

-0.1 ,, , ./
, . ,, /

f ,/-\ , ., ,,
,
,, .,
,, ,, ,
-0.2 " , ,
. , , ,,
,
" , ,, "
, ..
..... _,
:
,
' "

-0.3 '----_ _'----_ _' - - - - _ - - - - ' ' - - - - _ - - - - - ' C - - - _ - - ' ' - - _ - - - '
o 0.1 0.2 0.3
Time (sec)

Fig. 4.34 Displacement history of frame shown in Fig. 4.33

 -88-
6

-g
..2 -12
';j
'x
<:: -18

-24 --Column 1
----- Column 2
-30
o 0.1 Time (sec) 0.2 0.3
Fig. 4.35 Axial load history in the two columns

4,...-.-------------------,
2
-:;2
Q.

0
...
'-'

'"
.c
Cl)

rn -2

-4 - - Column 1
----- Column 2
-6
0 0.1 0.2
Time (sec)
Fig. 4.36 History of shear variation in the two columns
60 ,...------------r---------,

-.E
I
Q.
o
:;;: ,
-
'-'
cCl)
,,
E ,,
o ,'.. -....
:E -60
- - Column 1
-.--- Column 2

-120 L--- ---L --'- ~

-0.4 -0.2 0 0.2

Curvature 00- 2 radlin)
Fig. 4.37 Moment-curvature history at the base of the two columns
 -89-

5
I
"
I, I " I
I I I

I
I
,
I

0 I , I
" I

, I

,
I
' ,,
-5
,,
I

,, ,
I

---.e- ,,
I \

,
-
I \ /'

,,
I
::t: "
,,
I I

"'0
-10 , I

ce ,I I
0 - I
......J I I
I
: -15 I
I , I

<:
X
,
I
I
I
I
I
,
I
I
-20 1
I
'\
I, I
I

I
\ I
I I
, I
I
I

\,' ,
\
I

-25 - - Column 1
I
,
\ I
I

----- Column 2
-30
0 0.1 0.2 0.3
Time (sec)
(a) History of axial load variation in the two columns

---~ Co

-...
.c
ce
Cl)
0

en
-2

-4

-6
o 0.1 0.2 0.3
Time (sec)
(b) History of shear variation in the two columns

Fig. 4.38 Analysis results of frame of Fig. 4.33 using Takeda model
 -90-

Appendix A
Use of the Multi-Slice Fiber Model for Dynamic Analysis

The multi-slice fiber model has been adapted to work as part of DRAlN-2D2 - a program

for the analysis of two-dimensional frames subjected to base acceleration for which input data is

described in [4]. The following is a description of the part of the input data relating to the

multi-slice fiber element. The numbers in parenthesis refer to the notes at the end of the

appendix. The limits on the size of the various arrays and parameters are given in Appendix B.

A. Group control card (315,lX,16A4):

Element type identification number (= 3 ), number of elements, event code for the

group (KGREP) (l), analysis title.

B. Group material data

1. Group control card (215): number of concrete types (2), number of steel types (2)

2. Concrete description; two cards for every type (delete if there aren't any concrete

fibers): The first card (5FIO.0) contains five strain values and the second (5FIO.0) the

corresponding five stress values. Note that an initial point is assumed to be (0.,0.). The six

pairs of values describe the multi-linear concrete stress-strain relationship - see Fig. A.l. (3)

3. Steel description (3FIO.O); one card for every type (delete if there aren't any steel

fibers): Initial or elastic slope, second or strain hardening slope, and the yield stress ify).

c. Element data : for each element in the group data should be prepared conforming to the

following specifications:

1. Element control card (715)

 -91-

I-node number, J-node number, number of slices, geometric stiffness indicator (KGEOM) (4),

output printout control (KOUT) (5), event type indicator (KVNT) (6), detailed output file

number (KOUT3) (7), maximum allowed number of events per time step (IND3) ,

default=50. If this number is exceeded the analysis moves on to the next time step.

2. Slice spacings (8FIO.O): One or two cards to accommodate the number of spacings

starting from the one closest to node I ( one less than number of slices specified). (8)

3. Rigid end zones' lengths (2F10.0): respectively) to be considered perfectly rigid - e.g.

parts of the element within the joints. (2) (8)

4. Slice description; the following date should be included for every slice (unless

automatic slice generation is used):

4.1 Slice control card (215): NGC = number of concrete groups of fibers (2)(9), NGS =

number of steel groups of fibers (2) (9). Note if slice is identical to the previous one then set

NGC = -1 and delete rest of slice description.

4.2 Concrete configuration (delete if NGC 0): the following for every group of con-

crete fibers:

4.2.1 Concrete group control card (215, F10.0, 15):

Number of fibers in group, concrete material type associated with group, area of concrete fiber

in group, coordinate generation index (IY) = 0 to generate fiber coordinates and '= 1 if coordi-

nates are to be explicitly specified.

4.2.2 Fiber coordinates (8FlO.0) (10): If IY = 0, give coordinates of centroids of first and
last fibers. If IY = 1, list coordinates of all fibers in the group.

4.3 Steel configuration (delete if NGS = 0): the following for every group of steel fibers:
 -92-

4.3.1 Steel group control card (215, FIO.O, 15):

Number of layers in group, steel material type associated with group, area of steel fiber in

group, coordinate generation index (IY) = 0 to generate fiber coordinates and = 1 if coordi-

nates are to be explicitly specified.

4.3.2 Fiber coordipates (8FI0.0)(10): If IY = 0, give coordinates of centroids of first and

last fibers. If IY = 1, list coordinates of all fibers in the group.

NOTES

(l) If any type of midstep event is to be considered for any of the elements in the group,

KGREP should be set equal to one. If midstep events are to be ignored for all elements in the

group, KGREP should be set equal to zero - see 6 below.

(2) Can be zero.

(3) Note that abrupt breaks in the concrete stress-strain relation as well as too steep a des-

cending branch might result in equilibrium errors and/or numerical instabilities. For strain

values larger than the last specified strain ( the sixth ), the concrete is assumed crushed and a

zero stress is assigned to the fiber henceforth. Note that computational savings can be achieved

if the concrete stress-strain relationship can be adequately described in fewer than six points -

in this case specify appropriate strains and zero stress for the superfluous points as shown in

Fig. A.2.

(4) Set KGEOM = 1, if geometric (p -~) stiffness based on static analysis is to be

included; set equal to 0 to ignore geometric stiffness.

 -93-

(5) KOUT is used to control element output as follows:

Set = -1 for no output and no element description

o for no output

1 to print only

2 to print and save

3 to save only

4 to save and print 'reorganized' time history

Note that this output, along with other program output, is written to file IOU, and is indepen-

dent of the individual element output files referred to in (7) below.

(6) To include event-to-event calculations, KGREP (see 1 above) must be set equal to

one. KVNT should be set equal to zero to include all events. It should be set equal to one to

include steel fiber events and concrete fiber unloading events only. Note that if KGREP is set

equal to zero (i.e., ignore midstep events), KVNT should be set equal to one to calculate and

correct for unbalanced forces - see Section 2.3(e).

(7) Set KOUT3 = 0, if no detailed output for the element is required, otherwise indicate
number of file in which the output is to be written. These files are opened and named in the

MAIN program. Detailed output consists of the history of element forces and deformations

written in 10 columns which include : time, curvature (j), rotation (j), moment (j), curvature (j),

rotationG), momentG), axial displacement (.~), axial load, and shear force. Note that curva-

tures and moments are those of the slice at or closest to the indicated node.

(8) A warning is given by the program if the sum of the slice spacings and the two rigid

end zones' lengths is not equal to the distance between the nodes which the element connects.

(9) Fibers in the same group should have the same material type and area.

(0) The coordinates with respect to an axis chosen by the user, usually at the section's

midheight or at its plastic centroid. Note that this is the axis around which the moments are

calculated.
 -94-

C/)
C/)
w
a::
~
C/)

STRAIN

Fig. A.I Concrete stress-strain envelope showing required input parameters

STRAIN

Fig. A.2 Simplified concrete stress-strain curve

 -95-

Appendix B

Current Array Size Limitations

The array sizes as currently implemented in the program are given below. The limits can

be relaxed by modifying the array sizes found in the various COMMON blocks in the multi-

slice fiber element subroutines. Note, however, that such modifications must be relayed to the

main program by updating the values of variables NINFG and/or NINFE see [4].

B.I Limitations on Size of Group Arrays

Total Number of Concrete Types = 12

Total Number of Steel Types = 9

B.2 Limitations on Size of Element Arrays

NSLIC = Number of Slices 15

NFC = Number of Concrete Groups of Fibers 30

NFS = Number of Steel Groups of Fibers 15

Number of Concrete Fibers per Element = 150

Number of Steel Fibers per Element = 50

 -97-

~~THQUAKE ENGINEERING RESEARCH CENTER REPORTS

NOTE: Numbers in parentheses are Accession Numbers assigned by the National Technical Information Service; these are
followed by a price code. Copies of the reports may be ordered from the National Technical Information Service, 5285
Port Royal Road, Springfield, Virginia, 22161. Accession Numbers should be quoted on orders for reports (PB --- ---)
and remittance must accompany each order. Reports '....ithout this information were not available at time of printing.
The complete list of EERC reports (from EERC 67-1) is available upon request from the Earthquake Engineering Research
Center, University of California, Berkeley, 47th Street and Hoffman Boulevard, Richmond, California 94804.

UCB/EERC-77/01 "PLUSH - A Computer Program for Probabilistic Finite Element Analysis of Seismic Soil-Structure Inter-
action," by M.P. Romo Organista, J. Lysmer and H.B. Seed - 1977 (PB81 177 651)A05

UCB/EERC-77/02 "Soil-Structure Interaction Effects at the Humboldt Bay Power Plant in the Ferndale Earthquake of June
7, 1975," by J.E. Valera, H.B. Seed, C.F. Tsai and J. Lysmer - 1977 (PB 265 795)A04

UCB/EERC-77/03 "Influence of sample Disturbance on Sand Response to Cyclic Loading," by K. Mori, H.B. Seed and C.K. r
Chan - 1977 (PB 267 352)A04

UCB/EERC-77/04 "Seismological Studies of Strong Motion Records," by J. Shoja-Taheri - 1977 (PB 269 655)AIO

UCB/EERC-77/05 Unassigned

UCB/EERC-77/06 "Developing Methodologies for Evaluating the Earthquake Safety of Existing Buildings," by No.1 -
B. Bresler; No.2 - B. Bresler, T. Okada and D. Zisling; No. 3 - T. Okada and B. Bresler; No.4 - V.V.
Bertero and B. Bresler - 1977 (PB 267 354)A08

UCB!EERC-77/07 "A Literature Survey - Transverse strength of Masonry Walls," by Y. Ornote, R.L. Mayes, S.W. Chen and
R.W. Clough - 1977 (PB 277 933)A07

UCB/EERC-77 /08 "DRAIN-TABS: A Computer Program for Inelastic Earthquake Response of Three Dimensional Buildings," by
R. Guendelman-Israel and G.H. Powell - 1977 (PB 270 693)A07

UCB!EERC-77/09 "SUBWALL: A Special Purpose Finite Element Computer Program for Practical Elastic Analysis and Design
of Structural Walls with Substructure Option," by D.Q. Le, H. Peterson and E.P. Popov - 1977
(PB 270 567)A05

UCB/EERC-77/10 "Experimental Evaluation of Seismic Design Methods for Broad Cylindrical Tanks," by D.P. Clough
(PB 272 280) AU

UCB/EERC-77/11 "Earthquake Engineering Research at Berkeley - 1976," - 1977 (PB 273 507)A09

UCB!EERC-77/12 "Automated Design of Earthquake Resistant Multistory Steel Building Frames," by N.D. Walker, Jr. - 1977
(PB 276 526)A09

UCB/EERC-77/13 "Concrete Confined by Rectangular ,loops Subjected to Axial Loads," by J. Vallenas, V. V. Bertero and
E.P. Popov - 1977 (PB 275 165)A06

UCB/EERC-77/14 "Seismic Strain Induced in. the Ground During Earthquakes," by Y. Sugimllra - 1977 (PB 284 201JA04

UCB/EERC-77/15 Unassigned

UCB/EERC-77/16 "Computer Aided Optimum Design of Ductile Reinforced Concrete Moment Resisting Frames," by S.W.
Zagajeski and V.V. Bertero - 1977 (PB 280 137)A07

UCB/EERC-77/17 "Earthquake Simulation Testing of a Stepping Frame with Energy-Absorbing Devices," by J.M. Kelly and
D.F. Tsztoo - 1977 (PB 273 506)A04
UCB/EERC-77/1B "Inelastic Behavior of Eccentrically Braced Steel Frames under Cyclic Loadings," by C.W. Roeder and
E.P. Popov - 1977 (PB 275 526)A15

UCB/EERC-77/19 "A Simplified Procedure for Estimating Earthquake- Induced Deformations in Dams and Embankments," by F. I.
Makdisi and H.B. Seed - 1977 (PB 276 820)A04

UCB/EERC-77/20 "The Performance of Earth Dams during Earthquakes," by H.B. Seed, F.I. Makdisi and P. de Alba - 1977
(PB 276 821)A04

UCB/EERC-77/21 "Dynamic Plastic Analysis Using Stress Resultant Finite Element Formulation," by P. Lukkunapvasit and
J.M. Kelly - 1977 (PB 275 453)A04

UCB/EERC-77/22 "Preliminary Experimental Study of Seismic Uplift of a Steel Frame," by R.W. Clough and A.A. Huckelbridge
1977 (PB 278 769)A08

UCB/EERC-77/23 "Earthquake Simulator Tests of a Nine-Story Steel Frame with Columns Allowed to uplift," by A.A.
Huckelbridge - 1977 (PB 277 944)A09

UCB/EERC-77/24 "Nonlinear Soil-Structure Interaction of Skew Highway Bridges," by M.-C. Chen and J. Penzien - 1977
(PB 276 176)A07

UCB/EERC-77/25 "Seismic Analysis of an Offshore Structure Supported on Pile Foundations," by D.D.-N. Lieu and J. Penzien
1977 (PB 283 180JA06

UCB/EERC-77/26 "Dynamic Stiffness Matrices for Homogeneous Viscoelastic Half-Planes," by G. Dasgupta and A.K. Chopra -
1977 (PB 279 654)A06

---~eding page blank

 -98-

UCB/EERC-77/27 "A practical Soft Story Earthquake Isolation System," by J .~. Kelly, J .~\. Eidinger and C. J. Derham _
1977 (PB 276 814)A07

UCB/EERC-77 /28 "Seismic safety of EXisting BUildings and Incentives for Hazard 1-litigation in San Francisco: An
Exploratory Study," by .'\oJ. Meltsner - 1977 (FB 281 970)A05

UCB/EERC-77/29 "Dynamic Analysis of Electrohydraulic Shaking Tables," by D. Rea, S. Abedi-Hayati and Y. Takahashi
1977 (PB 282 569)A04

UCB/EERC-77/30 "An Approach for Improving Seismic - Resistant Behavior of Reinforced Concrete Interior Joints," by
B. Galunic, V.V. Eertero and E.P. Popov - 1977 (PE 290 870)A06

UCB/EERC-78/01 "The Development of Energy-Absorbing Devices for Aseismic Base Isolation Systems," by J.M. Kelly and
D.F. Tsztoo - 1978 (PB 284 978)A04

UCB/EERC-78/02 "Effect of Tensile Prestrain on the Cyclic Response of Structural Steel Connections, by J.G. Bouwkamp
and l'.. Mukhopadhyay, - 1978

UCB/EERC-78/03 "Experimental Results of an Earthquake Isolation System using Natural Rubber Bearings," by J. 11.
Eidinger and J.M. Kelly - 1978 (PB 281 686)A04

UCB/EERC-78/04 "Seismic Behavior of Tall Liquid Storage Tanks," by A. Niwa - 1978 (PB 284 017)A14

UCB/EERC-78/05 "Hysteretic Behavior of Reinforced Concrete Columns Subjected to High Axial and Cyclic Shear Forces,"
by S.W. Zagajeski, V.V. Bertero and J.G. Bouwkamp - 1978 (PB 283 858)A13

UCB/EERC-78/06 "Three Dimensional Inelastic Frame Elements for the ANSR-I Program," by A. Riahi, D.G. Rowand
G.H. Powell - 1978 (PB 295 755)A04

tiCB/EERC-78/07 "Studies of Structural Response to Earthquake Ground !'lotion," by O.A. Lopez and A.K. Chopra - 1978
(PE 282 790)A05

UCB/EERC-78/0S "A Laboratory Study of the Fluid-Structure Interaction of Submerged Tanks and Caissons in Earthquakes,"
by R.C. Byrd - 1978 (PB 284 957)A08

UCB/EERC-78/09 Unassigned

UCB/EERC-78/10 "Seismic performance of Nonstructural and Secondary Structural Elements," by I. Sakamoto - 1978
(PB81 154 593)A05

UCB/EERC-78/11 "Mathematical Hodelling of Hysteresis Loops for Reinforced Concrete Columns," by S. Nakat:a, T. Sproul
and J. Penzien - 1978 (FB 298 274)A05

UCB/EERC-78/12 "Damageability in Existing Buildings," by T. Blejwas and B. Bresler - 1978 (PB 80 166 978)A05

UCB/EERC-78/13 "Dynamic Behavior of a Pedestal Base Multistory Building," by R.M. Stephen, E.L. Wilson, J.G. Bouwkamp
and M. Button - 1978 (PB 286 650)A08

UCB/EERC-78/14 "Seismic Response of Bridges - Case Studies," by R.A. Imbsen, V. Nutt and J. Penzien - 1978
(P8 286 503)AIO

UCB/EERC-78/15 "A Substructure Technique for Nonlinear Static and Dynamic Analysis," by O.G. Rowand G.H. Powell -
1978 (PB 288 077)AIO

UCB/EERC-78/16 "Seismic Risk Studies for San Francisco and for the Greater San Francisco Bay Area," by C.S. Oliveira -
1978 (PB 81 120 115)A07

UCB/EERC-78/17 "Strength of Timber Roof Connections Subjected to Cyclic Loads," by P. GUlkan, R.L. !1ayes and R.W.
Clough - 1978 (HUD-OOO 1491)A07
UCB/EERC78/18 "Response of K-Braced Steel Frame Models to Lateral Loads," by J.G. BOUWkamp, R.M. Stephen and
E.P. Popov - 1978

UCB/EERC-78/19 "Rational Design Methods for Light Equipment in Structures Subjected to Ground Motion," by
J.L. Sackman and J.M. Kelly - 1978 (PB 292 357)A04

UCB/EERC-78/20 "Testing of a Wind Restraint for Aseismic Base Isolation," by J.M. Kelly and D.E. Chitty - 1978
(PB 292 833) A03

UCB/EERC-78/21 "APOLLO - A Computer Program for the Analysis of Pore Pressure Generation and Dissipation in Horizontal
Sand Layers During Cyclic or Earthquake Loading," by P.P. Martin and H.B. Seed - 1978 (PB 292 835)A04

UCB/EERC-78/22 "Optimal Design of an Earthquake Isolation System," by M.A. Bhatti, K.S. Fister and E. Polak - 1978
(PB 294 735)A06

UCB/EERC- 78/23 "MASH - A Computer Program for the Non-Linear Analysis of Vertically Propagating Shear Waves in
Horizontally Layered Deposits," by P.P. Martin and H.B. Seed - 1978 (PB 293 1011A05

UCB/EERC-78/24 "Investigation of the Elastic Characteristics of a Three Story Steel Frame Using System Identification,"
by I. Kaya and H.D. McNiven - 1978 (PB 296 225)A06

UCB/EERC-78/25 "Investigation of the Nonlinear Characteristics of a Three-Storv Steel Frame Using System
Identification," by I. Kaya and H.D. McNiven - 1978 (PB 301 363)A05
 -99-

UCB/EERC-73/26 "Studies of Strong Ground ~otion in Taiwan," by Y.~. Hsiung, B.A. Bolt and J. Penzien - 1978
(PB 298 436)A06

UCB/EERC-78/27 "Cjclic Loading Tests of ;tasonry Single Piers: volume 1 - Height to Width Ratio of 2," by P ..;. Hidalgo,
R.L. :1ayes, H.D. :1cNiven and R.i'. Clough - 1978 (PB 296 2il)A07

UCB/EERC-78/28 "Cyclic Loading Tests 0: Masonrj Single Piers: volume 2 - Height to Width Ratio of 1," by S.-W.J. Chen,
P.A. Hidalgo, R.L. Mayes, R.W. Clough and '1.0. McNiven - 1978 (PB 296 212)A09

UCB/EERC-78/29 "Analytical Procedures in Soil Dynamics," by J. Lysmer - ~978 (PB 298 445)A06

UCB/EERC-79/01 "HystereUc Behavior of Lightweight Reinforced Concrete Beam-Column Subassernblages," by B. Forzani,

E.P. Popov and V.V. Bertero - Apcil 1979(PB 298 267)A06

UCB/EERC-79/02 "The Development of a ~thematical ~del to Predict the Flexural Response of Reinforced Concrete Bealns
to Cyclic Loads, Using System Identification," by J. Stanton & H. ~cNiven - Jan. 1979(PB 295 875)AIO

uCB/EERC- 79/03 "Linear and Nonlinear Earthquake Response of Simple Torsionally Coupled Systems," by C.L. Kan and
A.K. Chopra - Feb. 1979(PB 298 262)A06

UCB/EERC-79/04 "A ~athematical ~del of Mascnry for Predicting its Linear Seismic Response Characteristics," by
Y. Mengi and '1.0. McNiven - Feb. 1979(PB 298 266)A06

UCB/EERC-79/05 '''~lechanical
Behavior of Lightweight Concrete Confined by Different Types of Lateral Reinforcement,"
by M.A. Manrique, V.V. Bertero and E.P. Popov - ~ay 1979(PB 301 114)A06

UCB/EERC-79/06 "Static Tilt Tests of a Tall Cylindrical Liquid Storage Tank," by R.W. Clough and A. Niwa - Feb. 1979
(PB 301 167) A06

UCB/EERC-79/07 "The Design of Steel Energy Absorbing Restrainers and Their Incorporation into Nuclear Power Plants
for Enhanced safety: Volume 1 - SUlmnary Report," by P.N. Spencer, V.F. Zackay, and E.R. Parker-
Feb. 1979 (UCB/EERC-79/07) A09 .
UCB/EERC-79/08 "The Design of Steel Energy Absorbing Restrainers and Their Incorporation into Nuclear Power Plants
for Enhanced Safety: Volume 2 - The Development of .;r.alyses for Reactor System Piping, ""Simple Systems"
by 14.C. Lee, J. Penzien, A.K. Chopra and K, Suzuki "Complex Systems" by G.H. Powell, E.L. Wilson,
R.W. Clough and D.G. Row - Feb. 1979(UCB/EERC-79/0B)AIO
UC9/EERC-79/09 "The Design of Steel Energy Absorbing Restrainers and Their Incorporation into Nuclear Power Plants
for Enhanced safety: Volume 3-Evaluation of Commercial Steels," by W.S. Owen, R.M.N. PelloL1x,
R.O. Ritchie, M. Faral, T. Ohhashi, J. Toplosky, S.J. Hartman, V.F. Zackay and E.R. Parker -
Feb. 1979(UCB/EERC-79/09)A04
UCB/EERC-79/10 "The Design of Steel Energy Absorbing Restrainers and Their Incorporation into Nuclear Power Plants
for Enhanced safety: Volume 4 - A Review of Energy-Absorbing Devices," by J. M. Kelly and
M.S. Skinner - Feb. 1979(UCB/EERC-79/10)A04
UCB/EERC-79/ll "Conservatism In Summation Rules for Closely Spaced ~lodes," by J.M. Kelly and J.L. Sackman - May
1979(PB 301 328)A03

UCB/EERC-79/12 "Cyclic Loading Tests of ~'lasonry Single Piers; Volume 3 - Height to Width Ratio of 0.5," by
P.A. Hidalgo, R.L. Mayes, H.D. McNiven and R.W. Clough - May 1979(PB 301 321)AOB

UCll/EERC- 79/13 "Cyclic Behavior of Dense Course-Grained Naterials in Relation to t.'1e Seismic Stability of Dams," by
N.G. Banerjee, H.B. Seed and C.K. Chan - June 1979(PB 301 373)A13

UCB/EERC-79/14 "Seismic Behavior of Reinforced Concrete Interior Bea'll-Column Subassernblages," by S. Viwathanatepa,

E.P. Popov and V.V. 8ertero - June 1979(P8 301 326)AIO

UCB/EERC-79/15 "Optimal Design of Localized Nonlinear Systems with Dual Performance Criteria Under Earthquake
Excitations," by M.A. Bhatti - July 1979 (PB 80 167 109) A06
UCB/EERC-79/16 "OPTDYN - A General Purpose Optimi zation Program for Problems with or wi thout Dynamic Constraints,"
by M.A. Bhatti, E. Polak and K.S. Pister - July 1979(PB 80 167 091)A05

UCB/EERC-79/17 "At,SR-II, Analysis of Nonlinear Str'.lctural Response, Users Manual," by D.P. Mondkar and G.H. Powell
July 1979(PB 80 113 301)A05
UCB/EERC-79/18 "Soil Structure Interaction in Di fferent Seismic Environments," A. Gomez-Masso, J. Lysmer, J. -C. Chen
and H.B. Seed - August 1979(PB 80 101 520)A04
UCB/EERC-79/19 "ARHA M::ldels for Earthquake Ground Motions," by M.K. Chang, J.W. Kwiatkowski, R.F. Nau, R.M. Oliver
and K.S. Pister - July 1979(PB 301 166)A05
UCB/EERC-79/20 "Hysteretic Behavior of Reinforced Concrete Structural Walls," by J. M. Vallenas, V. v. Bertero and
E.P. Popov - August 1979(PB 80 165 905)Al2
UCB/EERC-79/21 "Studies on High-Frequency Vibrations of Buildings - 1: The Column Effect," by J. Lubliner - August 1979
(PB 80 158 553) A03
UCB/EERC-79/22 "Effects of Generalized Loadings on Bond Reinforcing Bars Embedded in Confined Concrete Blocks," by
S. Viwathanatepa, E.P. Popov and V.V. Bertero - August 1979(PB 81 124 018)A14
UCB/EERC-79/23 "Shaking Table Study of Single-Story Masonry Houses, Volume 1: Test Structures 1 and 2, " by P. GUlkan,
R.L. Mayes and R.W. Clough -Sepf!. 1979 (HUD-OOO 1763)A12
UCS/EERC-79/24 "Shaking Table Study of Single-Story Masonry Houses, Volume 2 : Test Structures and 4, " by P. Gulkan,
R.L. Mayes and R.W. Clough -Sept. 1979 (HUD-OOO 1836)A12
UCB/EERC-79/25 "Shaking Table Study of Single-Story Masonry Houses, Volume 3: Summary, Conclusions and Recommendations,"
by R.W. Clough, R.L. ~yes and P. Gulkan - Sept. 1979 (HUO-OOO 18371A06
 -100-

UCB/EERC-79/26 "Recommendations tor aU. S. -Japan Cooperative Research Program Utilizing Large-Scale Tes ting Faci l i ties ...
by U.S.-Japan Planning Group - Sept. 1979(PB 301 407)1'.06
UCB/EERC79/27 "Earthquake-Induced Liquefaction Near Lake Amatitlan, Guatemala," by H.B. Seed, I. Arango, C.K. Chan,
A. Gomez-Masso and R. Grant de Ascoli - Sept. 1979(NUREG-CR134l)A03
UCB/EE;\C-79/28 "Infill Panels: Their Influence on Seismic Response of Buildings," by J.W. Axley and V. '1. Bertero
5ept. 1979(PB 80 163 37l)A10
UCB/EERC-79/29 "3D Truss Bar Eleir.ent (Type 1) for the ANSR-n Program," by D.P. Mondl<ar and G.H. Powell - NOV. 1979
(PB 80 169 709)1'.02
UCB/EERC--79/30 "2D Beam-Column Element (Type 5 - Parallel Element Theorf) for the !\llSR-n Program," by D.G. Row,
G.H. Powell and D.P. Mondkar - Dec. 1979(PB 80 167 224)1'.03
UCB/EERC-79/3l "3D Beam-Column Element (Type 2 - Parallel Element Theory) for the ANSR-II Program," by A. Riahi,
G.H. Powell and D.P. Mondkar - Dec. 1979(PB 80 167 2l6)A03
UCB/EERC-79/32 "On Response of Structures to StationarJ Excitation," by A. Der Kiureghian - Dec. 1979 (PB 80166 929) 1'.03
UCB/EERC-79/33 "Undisturbed Sampling and Cyclic Load Testing of Sands," by S. Singh, H.B. Seed and C.K. Chan
Dec. 1979(ADA 087 298)A07
UCS/EERC-79/34 "Interaction Effects of Simultaneous Torsional and Compressional Cyclic Loading of Sand," by
P.M. Griffin and \-I.N. Houston - Dec. 1979 (.1\DA 092 352l.'\15

UCB/EERC-80/0l I1Earthquake ..Response of Concrete Gravity Dams Including Hydr8dynamic and Foundation Interaction
Effects," by A.K. Chopra. P. Chakrabarti and S. Gupta - .:'an. 1980lAD-A087297)AlO
UC8/EERC-SO/02 "Rocking Response of Rigid Blocks co Earthquakes," by C.S. '{im, A.K. Chopra and J. Penzien - Jan. 1980
(PB80 166 002)A04
UC8/EERC-80/03 "Optimum Inelastic Design of Seismic-Resistant Reinforced Concrete Frame Structures," by 5. \-1. Zagajeski
and V.V. 8ertero - Jan. 1980(PB80 164 635)A06
UCB/EERC-SO/04 "Effects of Amount and Arrangement of Wall-panel Reinforcement on Hysteretic Behavior of Reinforced
Concrete 'llalls," by R. Iliya and V.V. Bercero - Feb. 1980(PB8l 122 525)A09
UCB/EERC-BO/05 "Shaking Table Research on Concrete Dam ~10dels," by A. Niwa and R. W. Clough - Sept. ~980 (PB8l 122 368) A06
UCB/EERC-80/06 "The Design of Steel Energy-Absorbing Restrainers and their Incorporation into Nuclear Pcwer Plants for
Enhanced safety (Vol lA): Piping with Energy Absorbing Restrainers: Parameter Study on Small Systems,"
by G.H. Powell, C. Oughourlian and J. Simons - June 1980
UCB/EERC-SO/07 "Inelastic Torsional Response of Structures Subjected to Earthquake Ground ~1otions," by Y. Yamazaki
April 1980(PB8l 122 327)A08
UCB/EERC-80/08 "Study of X-Braced Steel Frame Structures Under Earthquake Simulation," by Y. Ghanaat - April 1980
(PB8l 122 335)All
UCB/EERC-80/09 "Hybrid tlOdelling of Soil-Structure Interaction," by S. Gupta, T.W. Lin, J. Penzien and C.S. Yeh
May 1980(PB8l 122 3l9)A07
UC8/EERC-BO/10 "General Applicability of a Nonlinear Model of a One Story Steel Frame," by 3. I. Sveinsson and
H.D. ~lcNiven - "lay 1980(PB8l 124 877)A06
UCB/EERC-80/ll "A Green-Function Method for Wave Interaction with a Submerged Body," by W. Kioka - ;'.pril 1980
(PB81 122 269)A07
UCB/EERC-80/12 "Hydrodynamic Pressure and Added Mass for Axisymmetric Bodies," by F. Nilrat - ,1ay 1980 (PB8l 122 343) A08
UC3/EERC-80/13 "Treatment of Non-Linear Drag Forces Acting on Offshore Platforms," by B. V.. Dao and J. Penzien
May 1980(PB8l 153 413)1'.07
UCB/EERC-80/l4 "2D Plane/Axisymmetric Solid Element (Type 3 - Elastic or Elastic-perfectly Plastic) for the ANSR-II
Program," by D.P. Mondkar and G.H. Powell - July 1980(PB8l 122 350)1'.03
UCB/EERC-BO/15 "A Response Spectrum Method for Random Vibrations," by A. Der Kiureghian - June 1980 (PB8l 122 30l)A03
UCB/EERC-80/l6 "Cyclic Inelastic Buckling of Tubular Steel Braces," by V.A. Zayas, E.P. Popov and S.A. Hahin
June 1980(PB8l 124 385)A10
UCB/EERC-80/l7 "Dynamic Response of Simple Arch Dams Including Hydrodynamic Interaction," by C.S. Porter and
A.K. Chopra - July 1980(PB8l 124 000)A13
UCB/EERC-80/l8 "Experimental Testing of a Friction Damped Aseismic Base Isolation System with Fail-Safe
Characteristics," by J.M. Kelly, K.E. Beucke and M.S. Skinner - July 1980(PB8l 148 595)A04
UCB/EERC-80/l9 "The Design of Steel Energy-Absorbing Restrainers and their Incorporation into Nuclear power Plants for
Enhanced Safety (Vol lB): Stochastic Seismic Analyses of Nuclear Power Plant Structures and Piping
Systems Subjected to Multiple Supoort Excitations," by M.C. Lee and J. Penzien - June 1980
UCB/EERC-80/20 "The Design of Steel Energy-Absorbing Restrainers and their Incorporation into Nuclear Power Plants
for Enhanced Safety (Vol lC): Numerical Method for Dynamic Substructure Analysis," by J.M. Dickens
and E.L. Wilson - June 1980
UCB/EERC-BO/2l "The Design of Steel Energy-Absorbing Restrainers and their Incorporation into Nuclear Power Plants
for Enhanced Safety (Vol 2): Development and Testing of Restraints for Nuclear Piping Systems," by
J.M. Kelly and M.S. Skinner - June 1980
UCB/EERC-80/22 "3D Solid Element (Type 4-Elastic or Elastic-Perfectly-Plastic) for the ANSR-II Program." by
D.P. Mondkar and G.H. Powell - July 1980(PB8l 123 242)A03
UCB/EERC-80/23 "Gap-Friction Element (Type 5) for the ANSR-II Program," by D.P. Mondkar and G.H. Powell - July 1980
(PB8l 122 285)1'.03
 -101-

UCB/EERc-a0I24 "U-Bar Restraint Element (Type 11) for the ANSR-II Program," by C. Oughourlian and G.H. Powell
July 1980(2881 122 293).'103 .
UCB/EERC-aO/25 "Testing of a Natural Rubber Base Isolation System by an Explosively Simulated Earthquake," by
J.M. Kelly - August 1980(PB81 201 360).'104
UCB/EERC-30/26 "Input Identification from Structural Vibrational Response," by Y. Hu - August 19S0(PBSl 152 30S)A05
UCB/EERC-BO/27 "Cyclic Inelastic Behavior of Steel Offshore Structures," by V. A. Zayas,S. A. ~lahin and E. P. Popov
August 19S0(PBS1 196 lS0lA15
UCB/EERC-80/2S "Shaking Table Testing of a Reinforced Concrete Frame with 8iaxial Response," by /'1. G. Oliva
October 19S0(PB81 154 304).'110
UCB/EERC-SO/29 "Dynamic Properties of a Twelve-Story Prefabricated Panel Building," by J.G. Bouwkamp, J.P. Kollegger
and R.M. Stephen - October 1980(PBS2 117 12S)A06
UCB/EERC-80/30 "Dynamic Properties of an Eight-Story Prefabricated Panel Building," by J.G. Bouwkamp, J.P. Kollegger
and R.M. Stephen - October 1980(PBSl 200 313).'105
UCB/EERC-80/31 "Predictive Dynamic Response of Panel Type Structures Under Earthquakes," by J.P. Kollegger and
J.G. Bouwkamp - October 1980(PB81 152 316).'104
UCB/EERC-aO/32 "The Design of Steel Energy-Absorbing Restrainers and their Incorporation into Nuclear Power Plants
for Enhanced Safety (Vol 3): Testing of '::::ommercial Steels in Low-Cycle Torsional Fatigue r" by
P. ~~r~~~r, E.R. Parker, E. Jongewaard and M. Drory

U~~/SRc-aO/33 "The 0esl;n of Steel Energy-.~sorbing Restrainers and t~eir Incorporation into Nuclear POwer Plants
for Enhanced Safety (Vol 4): Shak~ng Table Tests of Piping Systems with Energy-Absorbing Restrainers,"
by S.F. Stiemer and W.G. Godden - Sept. 1980
uCB/E:ERC-80/34 "The Design of Steel E:nergy-Absorbing Restrainers and their Incorporation into Nuclear Power Plants
for Enhanced safety (Vol 5): Summary Report," by P. Spencer
0C3/EERC-80/35 "Experimental Testing of an Energy-Absorbing Base Isolation System, II by J. ~-'1. Kelly I H.:3. Skinner: and
K.E. Seucke - October 19S0(PB81 154 072).'104
UCS/E:ERC-80/35 "Si!llulating and .'\na1yzing Artificial Non-Stationary Earthquake Ground ~!otions," by R.F. Nau, R.N. Oliver
and K.S. Pister - October 1980(PE81 153 397)A04
UCS/E:ERC-80/37 "Earthquake E:ngineering at Berkeley - 1980." - Sept. 1980(PB81 205 874).'109
UC8/EERC-80/3S "Inelastic Seismic .'\nalysis of Large Panel Buildings," by V. Schricker and G.H. Powell - Sept. 1980
(PB81 154 338).'113
UCB/EERC-80/39 "Dynamic Res!?onse of Embankment, Concrete-Gravity and Arch Dams Including Hydrodynamic Inte"action,"
by J.P. Hall and A.K. Chopra - October 19S0(PBSl 152 324)All
UCB/EERC-80/40 "Inelastic Buckling of Steel Struts Under Cyclic Load Reversal," by R.G. Black, W. A. Wenger and
E.P. Popov - October 1980(PBSl 154 312).'108
UCB/EERC-80/41 "Influence of Site Characteristics on Building Damage During the October 3, 1974 Lima Earthquake," by
P. Re!?etto, I. Arango and H.B. Seed - Sept. 1980(PB81 161 739).'105 .
UCB/EERC-80/42 "E"laluation of a Shaking Table Test Program on Response Behavior of a Two Story Reinforced Concrete
Frame," by J.M. Blondet, R.\'i. Clough and S.A. Mahin
UCB/E,ERC-80/43 "t'lodelling of Soil-Structure Interaction by Finite and Infinite E:lemer.ts," by F. ~edina -
December 1980(PB81 229 270).'104

DCB/EERC-81/01 "Control of Seismic Response of Piping Systems and Other structures by Base Isolation," edited by J.M.
Kelly - January 1981 (PB81 200 735)A05

UCB/EERC-81/02 "OPTNSR - An Interactive Software System for Optimal Design of Statically and Dynamically Loaded
Structures '"ith Nonlinear Response," by M.A. Bhatti, V. Ciampi and K.S. Pister - January 1981
(PB81 218 851).'109

UCB/EERC-81/03 "Analysis of Local Variations in Free Field Seismic Ground Motions," by J.-C. Chen, J. Lysme" and H.B.
Seed - January 1981 (AD-A099508)A13

UCB/EERC-81/04 "Inelastic Structural Modeling of Braced Offshore Platforms for Seismic Loading," by V.A. Zayas,
P.-S.B. Shing, S.A. Mahin and E.P. Popov - January 1981(PB82 138 777).'107

UCB/EERC-81/05 "Dynamic Response of Light Equipment in Structures," by A. Der Kiureghian, J.L. Sackman and B. Nour-
amid - April 1981 (PB81 21S 497).'104

UCB/EERC-81/06 "Preliminary Experimental Investigation of a Broad Base Liquid Storage Tank," by J.G. Bouwkamp, J.P.
Kollegger and R.M. Stephen - May 1981(PB82 140 385)A03

UCB/EERC-81/07 "The Seismic Resistant Design of Reinforced Concrete Coupled Structural Walls," by .'I.E. Aktan and V.V.
Bertero - June 1981(PB82 113 358).'111

UCB/EERC-81/08 "The Undrained Shearing Resistance of Cohesive Soils at Large Deformations," by M.R. Pyles and H.B.
Seed - August 1981

UCB/EERC-81/09 "Experimental Behavior of a Spatial Piping System with Steel Energy Absorbers Subjected to a SimUlated
Differential Seismic Input," by S.F. Stiemer, W.G. Godden and J.M. Kelly - July 1981
 -102-

UCB/EERC-81/10 "Evaluation of Seismic Design Provisions for Masonry in the United States," by B.I. Sveinsson, R.L.
Mayes and H.D. McNiven - August 1981 (PB82 166 07~)A08

UCB/EERC-8l/ll "Two-Dimensional Hybrid Modelling of Soil-Structure Interaction," by T.-J. Tzong, S. Gupta and J.
Penzien - August 1981 (PB82 142 118)A04

UCB/EERC-81/12 "Studies on Effects of Infills in Seismic Resistant RiC Construction," by S. Brokkcn and V.V. Bertero -
September 1981 (PBB2 166 190)A09
UCB/EERC-81/13 "Linear Models to Predict the Nonlinear Seismic Behavior of a One-Story Steel Frame," by H. Valdimarsson,
A.H. Shah and H.D. McNiven - September 1981(PB82 138 793)A07

UCB/EERC-Bl/14 "TLUSH: A Computer Program for the Three-Dimensional Dynamic Analysis of Earth Dams," by T. Kagawa,
L.H. Mejia, H.B. Seed and J. Lysmer - September 1981(PB82 139 940)A06

UCB/EERC-Bl/15 "Three Dimensional Dynamic Response Analysis of Earth Darns," by L.H. Mejia and H.B. Seed - September 1981
(PBB2 137 274)A12

UCB/EERC-81/16 "Experimental Study of Lead and Elastomeric Dampers for Base Isolation Systems," by J.M. Kelly and
S.B. Hodder - October 19B1 (PB82 166 l82)A05

UCB/EERC-81/17 "The Influence of Base Isolation on the Seismic Response of Light Secondary Equipment," by J.M. Kelly -
April 19B1 (PBB2 255 266)A04

UCB/EERC-81/18 "Studies on Evaluation of Shaking Table Response Analysis Procedures," by J. Marcial Blondet - November
19B1 (PB82 197 278)AIO

UCB/EERC-81/19 "DELIGHT.STRUCT: A Computer-Aided Design Environment for Structural Engineering," by R.J. Balling,
K.S. Pister and E. Polak - December 1981 (PB82 218 496)A07

UCB/EERC-8l/20 "Optimal Design of Seismic-Resistant Planar Steel Frames," by R.J. Balling, V. ciampi, K.S. Pister and
E. Polak - December 1981 (PB82 220 l79)A07

UCB/EERC-82/0l "Dynamic Behavior of Ground for Seismic Analysis of Lifeline Systems," by T. Sato and A. Der Kiureqhian -
January 19B2 (PBB2 21B 926)A05

UCB/EERC-82/02 "Shaking Table Tests of a Tubular Steel Frame Model," by Y. Ghanaat and R. W. Clough - January 1982
(PBB2 220 l6l)A07
UCB/EERC-B2/03 "Behavior of a Piping System under seismic Excitation: Experimental Investigations of a Spatial piping
system supported by Mechanical Shock Arrestors and Steel Energy Absorbing Devices under Seismic
Excitation," by S. Schneider, H.-M. Lee and W. G. Godden - May 1982 (PB83 172 544)A09
UCB/EERC-82/04 "New Approaches for the Dynamic Analysis of Large Structural systems," by E. L. Wilson - June 1982
(PB83 148 080)A05
UCB/EERC-82/05 "Model Study of Effects of Damage on the Vibration Properties of Steel Offshore Platforms," by
F. Shahrivar and J. G. Bouwkamp - June 1982 (PBB3 14B 742)AlO

UCB/EERC-82/06 "States of the Art and Practice in the Optimum Seismic Design and Analytical Response Prediction of
Ric Frame-Wall Structures," by A. E. Aktan and V. V. Bertero - July 19B2 (PB83 147 736)A05
UCB/EERC-82/07 "Further Study of the Earthquake Response of a Broad Cylindrical Liquid-Storage Tank Model," by
G. C. Manos and R. W. Clough - July 1982 (PB83 147 744)All

UCB/EERC-B2/0B "An Evaluation of the Design and Analytical seismic Response of a Seven Story Reinforced Concrete
Frame - Wall structure," by F. A. Charney and V. V. Bertero - July 1982(PB83 157 62B)A09

UCB/EERC-B2/09 "Fluid-Structure Interactions: Added Mass Computations for Incompressible Fluid," by J. S.-H. Kuo -
August 1982 (PBB3 156 281)A07

UCB/EERC-B2/l0 "Joint-Opening Nonlinear Mechanism: Interface Smeared Crack Model," by J. S.-H. Kuo -
August 19B2 (PBB3 149 195)A05

UCB/EERC-82/ll "Dynamic Response Analysis of Techi Dam," by R. W. Clough, R. M. Stephen and J. S.-H. J<uo -
August 1982 (PBB3 147 496)A06

VCB/EERC-B2/l2 "Prediction of the Seismic Responses of R/C Frame-Coupled Wall Structures," by A. E. Aktan, V. V.
Bertero and M. piazza - August 19B2 (PBB3 149 203~09

VCB/EERC-B2/l3 "Preliminary Report on the SMART 1 strong Motion Array in Taiwan," by B. A. Bolt, C. H. Loh, J.
Penzien, Y. B. Tsai and Y. T. Yeh - August 1982 (PBB3 159 400)AlO

UCB/EERC-B2/l4 "Shaking-Table Studies of an Eccentrically X-Braced Steel Structure," by M. S. Yang - september

1982 (PB83 260 77B)Al2

UCB/EERC-B2/15 "The Performance of stairways in Earthquakes," by C. Roha, J. W. Axley and V. V. Bertero - September
1982 (PBB3 157 693)A07

UCB/EERC-B2/16 "The Behavior of Submerged Multiple Bodies in Earthquakes," by W. - G. Liao - Sept. 1982 (PB83 158 -
 -103-
UCB/EERC-ll2/l7 "Effects of Concrete Types ilnd Loading Conditions on Local Bond-Slip Relationships," by A. D. Cowell,
E. P. Popov and V. V. Bertero - September 1982 (PB83 153 577)A04
UCB/EERC-82/l8 "Nechanical Behavior of Shear Wall Vertical Boundary Members: An Experimental Investigation," by
M. T. Wagner and V. V. Bertero - October 1982 (PB83 159 764)A05

UCB/EERC-82/l9 "Experimental Studies of Multi-support Seismic Loading on Piping Systems," by J. M. Kelly and
A. D. Cowell - November 1982

UCB/EERC-82/20 "Generalized Plastic Hinge Concepts for 3D Beam-Column Elements," by P. F.-S. Chen and G. H. Powell -
November 1982 (PB03 217 ~81)A13
UCB/EERC-82/2l "ANSR-III: General Purpose Computer Program for Nonlinear Structural Analysis," by C. V. Oughourlian
and G. H. Powell - November 1982 (PB83 251 330)A12

UCB/EERC-82/22 "Solution Strategies for Statically Loaded Nonlinear Structures," by J. W. Simons and G. H. Powell -
November 1982 (PB83 197 970)A06

UCB/EERC-82/23 "Analytical Model of Deformed Bar Anchorages under Generalized Excitations," by V. Ciampi, R.
Eligehausen, V. V. Bertero and E. P. Popov - November 1982 (PB83 169 532)A06

UCB/EERC-82/24 "" Mathematical Model for the Response of Masonry Walls to Dynamic Excitations," by H. Sucuoglu,
Y. Mengi and H. D. McNiven - November 1982 (PB83 169 011)A07

UCB/EERC-82/25 "ftJfJ:l'~u~te2r~\sIO'crse Considerations of Broad Liquid storage Tanks," by F. J. Cambra - November 1982

UCB/EERC-82/26 "Computational Models for Cyclic Plasticity, Rate Dependence and Creep," by B. Mosaddad and G. H.
Powell - November 1982 (PB83 245 829)A08

UCB/EERC-82/27 "Inelastic Analysis of Piping and Tubular Structures," by M. Mahasuverachai and G. H. Powell - November
1982 .(PB83 249 987)A07

UCB/EERC-83/0l "The Economic Feasibility of Seismic Rehabilitation of Buildings by Base Isolation," by J. M. Kelly -
January 1983 (PB83 197 988)AOs

UCB/EERC-83/02 '(~~~~"'1:5~T~rR04onnections for Moment-Resisting Steel Frames," by E. P. Popov - January 1983

UCB/EERC-83/03 "Design of Links and Beam-to-Colurnn Connections for Eccentrically Braced Steel Frames," by E. P, Popov
and J. O. Malley - January 1983 (PB83 194 8l1)A04

UCB/EERC-83/04 "Numerical Techniques for the Evaluation of Soil-structure Interaction Effects in the Time Domain,"
by E. Bayo and E. L. Wilson - February 1983 (PB83 245 60s)A09

UCB/EERC-83/0s "A Transducer for Measuring the Intemal Forces in the Columns of a Frame-Wall Reinforced Concrete
Structure," by R. Sause and V. V. Bertero - May 1983 (PB84 ll9 494)A06

UCB/EERC-83/06 "Dynamic Interactions between Floating Ice and Offshore structures," by P. Croteau - May 1983
(PB84 ll9 4861Al6

UCB/EERC-83/07 "Dynamic Analysis of Multiply Tuned and Arbitrarily Supported Secondary Systems," by T. Igusa
and A. Der Kiureghian - June 1983 (PB84 ll8 2721All

UCB/EERC-B3/0fl "A Laboratory study of submerged Multi-body Systems in Earthquakes," by G. R. Ansari - June 1983
(PB83 26l 842)Al7

UCB/EERC-B3/09 "Effects of Transient Foundation Uplift on Earthquake Response of structures," by C.-S, yin\ and
A. K. Chopra - June 1983 (PB83 26l 396)A07

UCB/EERC-83/l0 "Optimal Design of Friction-Braced Frames under seismic Loading," by M, A. Austin and K, S. Pister -
June 1983 (PB84 119 288,A06

UCB/EERC-83/11 "Shaking Table study of Single-story Masonry Houses: Dynamic Performance under Three Component
Seismic Input and Recommendations," by G. C. Manos, R. W. Clough and R. L. Mayes - June 1983

UCB/EERC-B3/l2 "Experimental Error propagation in Pseudodynamic Testing," by P. B. Shing and S. A. Mahin - June 1983
(PB84 119 270)A09

UCB/EERC-83/l3 "Experimental and Analytical Predictions of the Mechanical Characteristics of a LIS-scale Model of a
7-story RIC Frame-Wall Building Structure," by A. E. Aktan, V. V. Bertero. A. A. Chowdhury and
T. Nagashirna - August 1983 (PB84 119 2l3)A07

UCB/EERC-83/l4 "Shaking Table Tests of Large-panel Precast Concrete Building System Assemblages," by ~I. G. Oliva and
R. w. Clough - August 1983

UCB/EERC-83/l5 "seismic Behavior of Active Beam Links in Eccentrically Braced Frames," by K. D. Hjelmstad and E. P.
Popov - July 1983 (PB84 ll9 676)A09

UCB/EERC-83/l6 "System Identification of Structures with Joint Rotation," by J. S. Dimsdale and H. D. McNiven -
July 1983

UCB/EERC-83/l7 "Construction of Inelastic Response Spectra for Single-Degree-of-Freedom Systems," by S. Mahin and
J. Lin - July 1983
 -104-

UCB/EERC-83/18 "Interactive Computer Analysis Methods for Predicting the Inelastic Cyclic Behaviour of Structural
Sections," by S. Kaba and S. Mahin - July 1983

UCB/EERC-83/19 "Effects of Bond Deterioration on Hysteretic Behavior of Reinforced Concrete Joints," by F.C. Filippou,
E.P. Popov and V.V. Bertero - August 1983

UCB/EERC-83/20 "Analytical and Experimental Correlation of Large-Panel Precast Building System Performance," by M.G.
Oliva, R.W. Clough, M. Velkov, P. Gavrilovic and J. Petrovski - November 1983

UCB/EERC-83/21 "Mechanical Characteristics of Materials Used in a 1/5 Scale Model of a 7-Story Reinforced Concrete
Test Structure," by V.V. Bertero, A.E. Aktan, H.G. Harris and A.A. Chowdhury - September 1983

UCB/EERC-83/22 "Hybrid Modelling of Soil-Structure Interaction in Layered Media," by T.-J. Tzong and J. Penzien -
October 1983

UCB/EERC-83/23 "Local Bond Stress-Slip Relationships of Deformed Bars under Generalized Excitations," by R. Eligehausen,
E.P. Popov and V.V. Bertero - O~tober 1983

UCB/EERC-83/24 "Design Considerations for Shear Links in Eccentrically Braced FraJres," by J.O. Halley and E.P. Popov -
November 1983

UCB/EERC-84/01 "Pseudodynamic Test Method for Seismic Performance Evaluation: Theory and Implementation," by P.-S. B.
Shing and S. A. Mahin - January 1984 (PB84 190 644) AD8

UCB/EERC-84/02 "Dynamic Response Behavior of Xiang Hong Dian Dam," by R. W. Clough, K.-T. Chang, H.-Q. Chen, R. M.
Stephen, G.-L. Wang, and Y. Ghanaat - April 1984
UCB/EERC-84/03 "Refined Modelling of Reinforced Concrete Columns for Seismic Analysis," by S.A. Kaba and S.A. Mahin
- April, 1984