Nonlinear Earthquake Response Analysis

of
Reinforced Concrete Buildings

Lecture Notes

August 2002

Shunsuke Otani
Department of Architecture
Graduate School of Engineering
University of Tokyo
 Preface

This note is intended to introduce the state of the art in the nonlinear response analysis of reinforced
concrete building structures under earthquake excitation to graduate students. The state of the
knowledge on the behavior of reinforced concrete members and structures and the art of nonlinear
response analysis are far form an established state. Therefore, this note will not provide any unique
solution to a problem.

The note was initially prepared for a special lecture on “nonlinear analysis of reinforced concrete
buildings” at Department of Civil Engineering, University of Canterbury, New Zealand, from February
to April, 1994. The note has been revised for use in Department of Architecture, University of Tokyo
since 1996; this course was given in English. The note was extensively revised for a series of lectures
on “nonlinear earthquake response analysis of reinforced concrete buildings” at European School for
Advanced Studies in Reduction of Seismic Risk, Universita degli Studi di Pavia, Italy, from February
to March, 2002.

The use of this note should be limited to personal use.

August 2002

Professor Shunsuke Otani

Department of Architecture, Graduate School of Engineering
University of Tokyo
otani@sake.t.u-tokyo.ac.jp
http://www.rcs.arch.t.u-tokyo.ac.jp/otani/

ii
 Contents of Lecture

1. Introduction

2. Properties of Reinforced Concrete Materials

2.1 Concrete
2.2 Reinforcing Steel
2.3 Bond

3. Behavior of Reinforced Concrete Members

3.1 Behavior of Beams
3.2 Behavior of Columns
3.3 Behavior of Interior Beam-column Connections
3.4 Behavior of Exterior Beam-column Connections
3.5 Behavior of Structural Walls

4. Analysis of Reinforced Concrete Members

4.1 Flexural Analysis of Section
4.2 Moment-Curvature Relation under Reversed Loading
4.3 Flexural Analysis of Members
4.4 Load-deformation Relation of Beams
4.5 Analysis of Structural Walls

5. Structural Dynamics
5.1 Differential Equation of Motion
5.2 Mass of Inertia
5.3 Damping
5.4 Strain-rate Effect
5.5 Properties of Earthquake Ground Motion

6. Numerical Integration Methods

6.1 Introduction
6.2 Nigam-Jennings’ Direct Integration Method
6.3 Linear Acceleration Method
6.4 Newmark Beta Method
6.5 Wilson’s Theta Method
6.6 Runge-Kutta-Gill Method (Fourth Order)

7. Matrix Analysis of Linearly Elastic Plane Frames

7.1 Assumptions
7.2 Member Stiffness Matrix in Local Coordinates
7.3 Coordinate Transformation
7.4 Member Stiffness Matrix in Global Coordinates
7.5 Continuity of Displacement at Joint
7.6 Equilibrium of Forces at Joint
7.7 Formulation of Structural Stiffness Matrix
7.8 Free Joint Displacements and Support Reactions
7.9 Member End Actions

8. Numerical Solution of Linear Equations

8.1 Incremental Formulation
8.2 Modified Cholesky Matrix Decomposition
8.3 Solution of Linear Algebraic Equations
8.4 Static Condensation
8.5 Damping Matrix

iii
9. Formulation of Member Stiffness Matrix
9.1 Introduction
9.2 Formulation of Member Stiffness Matrix
9.3 Member with Rigid Ends
9.4 Member with Flexible Ends
9.5 Simply Supported Member

10. Member Stiffness Models

10.1 Member Stiffness Model
10.2 Fiber Model
10.3 Discrete Element Models
10.4 One-component Model
10.5 Multi-component Model
10.6 Distributed Flexibility Model
10.7 Multi-spring Model (1)
10.8 Multi-spring Model (2)
10.9 Multi-spring Model (3)
10.10 Wall Models

11. Member Hysteresis Models

11.1 Introduction
11.2 Bilinear Model
11.3 Ramberg-Osgood Model
11.4 Degrading Trilinear Model
11.5 Clough Degrading Model
11.6 Takeda Degrading Model
11.7 Pivot Model
11.8 Stable Hysteresis Models with Pinching
11.9 Shear-type Hysteresis Models
11.10 Axial Force-Bending Moment Interaction
11.11 Special Purpose Models

12. Response of Different Models

12.1 Effect of Member Modeling
12.2 Effect of Damping Modeling

13. Response of Different Hysteresis Models

13.1 Analysis Method
13.2 Effect of Initial Stiffness (Takeda Model)
13.3 Effect of Cracking Force Level (Takeda Model)
13.4 Effect of Yield Resistance Level (Takeda Model)
13.5 Effect of Post-Yielding Stiffness (Takeda Model)
13.6 Effect of Unloading Stiffness Degradation Parameters (Takeda Model)
13.7 Effect of Hysteresis Energy Dissipation
13.8 Effect of Parameter of Ramberg-Osgood Model
13.9 Response to Different Earthquake Motions
13.10 Response of Different Models
13.11 Response Waveforms and Hysteresis Relations
13.12 Effect of Hysteresis Shape on Frame Response

1４. Reliability of Nonlinear Response Analysis Methods

14.1 Introduction
14.2 Reinforced Concrete Column
14.3 Frame Structures
14.4 Frame-wall Structures
14.5 Wall Structures

iv
 15. U.S.-Japan Full-scale Test
15.1 Test Program of Full-scale Seven-story RC Building
15.2 Description of Test
15.3 Modeling of Structural Members
15.4 Stiffness of Member Models
15.5 Method of Response Analysis
15.6 Results of Analysis
15.7 Concluding Remarks

Suggested Reading

American Concrete Institute, Earthquake Resistant Concrete Structures - Inelastic Response and
Design, ACI-SP127, American Concrete Institute, Detroit, 1991.
Comite Euro-International du Beton, RC Elements under Cyclic Loading - State of the Art Report,
Thomas Telford, 1996, 190 pp.
Comite Euro-International du Beton, RC Frames under Earthquake Loading - State of the Art Report,
Thomas Telford, 1996, 303 pp.

v
 Chapter 1. INTRODUCTION

Dynamic response of a structure can be caused by different loading conditions such as: (a)
earthquake ground motion; (b) wind pressure; (c) wave action; (d) blast; (e) machine vibration; and
(f) traffic movement. Among these, inelastic response is mainly caused by earthquake motions and
accidental blasts. Consequently, more research on nonlinear structural behaviour has been carried
out in relation to earthquake problems.

1.1 Structural Dynamics

Note that dynamic problems are different from static one in the following points: (a) inertial force;
(b) damping; (c) strain rate effect; and (d) oscillation (stress reversals). These need to be clarified in
order to analyze a structure under dynamic loading.

The equation of motion for a linearly elastic system under horizontal ground motion is normally
expressed in a form below;
z} + [c]{x} + [k ]{x} = {0}
[m]{
where [ m] , [c] , [k ] : mass, damping coefficient and stiffness matrices, {z} : absolute acceleration
vector at mass level, {x} and {x} : velocity and displacement vector at mass level relative to the
structural foundation.

Dynamic characteristics up to failure cannot be identified solely through a dynamic test of a real
structure for the following reasons: (a) difficult to understand the behaviour due to complex
interactions of various parameters; (b) expensive to build a structure, as a specimen, for destructive
testing; and (c) capacity of loading devices insufficient to cause failure. Consequently, dynamic tests
of real buildings are rather aimed toward obtaining data (a) to confirm the validity of mathematical
modeling techniques for a linearly elastic structure; and (b) to obtain damping characteristics of
different types of structures. A specifically designed laboratory test becomes inevitable in order to
complement the weakness of full-scale tests and to study the effect of individual parameters.

Damping: Any mechanical system possesses some energy-dissipating mechanisms, for example:
(a) inelastic hysteretic energy dissipation; (b)
radiation of kinetic energy through foundation;
(c) kinetic friction; (d) viscosity in materials; and
(e) aerodynamic effect. Such capacity or energy
dissipation is vaguely termed “damping,” and is
most often assumed to be of viscous type
simply because of its mathematical simplicity;
i.e., resistance proportional to velocity. It should
be noted that the actual damping mechanism
may not be of viscous type.

Damping capacity is often determined by the

response curve (a plot of response amplitudes
at steady state oscillation with respect to
excitation frequencies) during a sinusoidal
steady-state test. Figure 1 shows such
acceleration response curves for a reinforced
concrete building at different excitation levels
（Jennings and Kuroiwa 1968). The frequency
corresponding to peak response indicates the
natural frequency of the structure, and the band
width of the response curve represent the
damping capacity. Note the shift of resonant
frequencies and the change in amplitudes of Fig. 1: Observed Acceleration amplitudes from
damping with increase of excitation level despite Steady-state test (Jennings and Kuroiwa, 1968

1
low response amplitudes.

Damping capacity, expressed in terms of viscous damping factor, is not a unique value of a
structure, but it depends on the level of excitation. The state-of-the-art does not provide a method to
determine the damping capacity based on the material properties and geometrical characteristics of
a structure.

Strain Rate Effect: It is technically difficult to test structural members under dynamic conditions in a
laboratory. Before force-deformation relations already obtained from thousands of static tests can be
studied for use in a dynamic analysis, the effect of strain rate on the force-deformation relation
needs to be examined.

The speed of loading is known to influence the stiffness and strength of various materials (Cowell
1965, 1966). Some member test results are available (Mahin and Bertero l972). Important findings
from these investigations are as follows: (a) high strain rates increased the initial yield resistance,
but caused small differences in either stiffness or resistance in subsequent cycles at the same
displacement amplitudes; (b) strain rate effect on resistance diminished with increased deformation
in a strain-hardening range; and (c) no substantial changes were observed in ductility and overall
energy absorption capacity.

Note that strain rate (velocity) during an oscillation is highest at low stress levels, and that the rate
gradually decreases toward a peak strain. Cracking and yielding of a reinforced concrete member
reduce the stiffness, elongating the period of oscillation. Furthermore, such damage is normally
caused by the lower modes of vibration having long periods. Therefore, the strain rate is small in the
case of earthquake response, and its effect on the response is small.

Consequently, the static hysteretic behaviour observed can be utilized in a nonlinear dynamic
analysis of reinforced concrete structures.

1.2 Stiffness Properties of Reinforced Concrete Members

It is not feasible to analyze an entire structure using microscopic material models. It is more
important to study the behaviour of isolated members and their subassemblies (beam-column,
slab-column, and slab-wall
connections) so that their
analytical models can be
developed for use in the analysis
of a complete structure.

A typical force-deflection curve

of a cantilever column is shown
in Fig. 2 (Otani et al. 1979). Note
the following observations: (a)
tensile cracking of concrete and
yielding of longitudinal
reinforcement reduced the
stiffness; (b) when a deflection
reversal was repeated at the
same newly attained maximum
amplitude (for example, cycles 3
and 4) the loading stiffness in the
second cycle was lower than that
in the first cycle, although the
resistances at the peak
FIG. 2. Hysteretic characteristics of reinforced
displacement were almost
concrete member (Otani et al. 1979)
identical; and (c) average
2
stiffness (peak-to-peak) of a complete cycle decreased with a maximum displacement amplitude.
For example, the peak-to-peak stiffness of cycle 5, after large amplitude displacement reversals,
was significantly reduced from that of cycle 2 at comparable displacement amplitudes. Therefore,
the hysteretic behaviour of the reinforced concrete is sensitive to loading history. Let us study some
typical stiffness characteristics of reinforced concrete.

Flexural Characteristics: The

flexural deformation index (average
curvature) is obtained from
longitudinal strain measurements at
two levels assuming that a plane
section remains plane. This flexural
deformation index does not
represent the flexural deformation
in a strict sense because a plane
section does not remain plane in a
region where an extensive shear
deformation occurs. However, the
index is useful for understanding
flexural deformation characteristics
qualitatively.

A typical moment-flexural
deformation index curve obtained FIG. 3. Flexural deformation characteristics
from a simply supported beam test (Celebi and Penzien 1973)
(Celebi and Penzien 1973) is
shown.in Fig. 3. Note that the stiffness during loading gradually decreases with load, forming a fat
hysteresis loop, and absorbing a large amount of hysteretic energy. The hysteresis loops remain
almost identical even after several load reversals at the same displacement amplitude beyond
yielding. Consequently, vibration energy can be efficiently dissipated through flexural hysteresis
loops without a reduction in resistance. Many hysteretic models, as discussed later, are currently
available to represent the nexural behaviour.

The increase in axial force decreases the flexural ductility of a reinforced concrete member, but
increases force levels corresponding to (a) tensile cracking of concrete; and (b) tensile yielding of
longitudinal reinforcement.

Shear Characteristics: Similar to

the flexural deformation index, a
shear deformation index is defined
from strain measurements in the two
diagonal directions. Again, this index
does not represent the true shear
deformation because the
interference of shear and flexure
exists.

A typical lateral load-shear

deformation index curve (Celebi and
Penzien 1973) is shown in Fig. 4.
Unlike what occurs in flexure, the
stiffness during loading gradually
increases with load, exhibiting a
“pinching” in the curve. The FIG. 4. Shear deformation characteristics
hysteretic energy dissipation is (Celebi and Penzien 1973)
smaller. The hysteresis loop decays
with the number of load reversals, resulting in a smaller resistance at the same peak displacement in
each repeated loading cycle. Although the curve shows a “yielding” phenomenon, it is important to
3
recognize that the shear force of the member was limited by flexural yielding at the critical section
rather than by yielding in shear. This yielding clearly indicates the interaction of shear and bending.

The pinching in the force-deformation curve is obviously less desirable. The shear span to
effective depth ratio is the most significant parameter. Decreasing the shear span to depth ratio
causes a more pronounced pinching in the curve, and a faster degradation of the hysteretic
energy-dissipating capacity. Considerable improvements in delaying and reducing the degrading
effects can be accomplished by using closely spaced ties. Existence of axial force tends to retard the
decrease in stiffness and resistance with cycles. However, it is hard to eliminate this undesirable
effect when high shear stress exists. Consequently, it becomes important to include this degrading
behaviour in a behavioural model for a short, deep reinforced concrete member. The current state of
knowledge is not sufficient to define the stiffness degrading parameters on the basis of the member
geometry and material properties.

Bar Slip and Bond Deterioration: When a structural element is framed into another element, some
deformation is initiated within the other element. Consider a beam-column subassembly. Bertero and
Popov (1977) reported a significant rotation at a beam end caused by the slippage (pullout) of the
beam's main longitudinal reinforcement within the beam-column joint (Fig. 5). The general shape of
the moment-bar slip rotation curve is similar to that shown in Fig. 4, demonstrating a pronounced
pinching of a hysteresis loop. The contribution of bar slip to total deformation cannot be neglected,
especially in a stiff member (short or deep).

FIG. 5. Rotation due to bar slip (Bertero and Popov 1977).

Biaxial Lateral Load Reversals: During an earthquake, columns of a framed structure must resist
lateral forces simultaneously in longitudinal and transverse directions. Recent tests at the University
of Toronto (Otani et al. 1979) on reinforced concrete columns showed that the columns under lateral
load reversals in two perpendicular directions exhibited the decay of resistance and stiffness at a
faster rate than those under uniaxial lateral load reversals. This topic needs further study.

1.3 Hysteretic Models for Reinforced Concrete

Nonlinear dynamic analysis of a reinforced concrete structure requires two types of mathematical
modelling: (a) modelling for the distribution of stiffness along a member; and (b) modelling for the
force-deformation relationship under stress reversals.
4
 A hysteresis model must be able to provide the stiffness and resistance under any displacement
history. At the same time, the basic characteristics need to be defined by the member geometry and
material properties. The current state of knowledge is sufficient to define flexural hysteresis models.
However, it is not sufficient to determine the degree of stiffness degradation due to the deterioration
of shear-resisting and rebar-concrete bond mechanisms.

Bilinear Model: The elastic-perfectly plastic

hysteretic model was used by many investigators
because the model was simple. The maximum
displacement of an elasto-plastic simple system
was found (Veletsos and Newmark 1960) to be
practically the same as that of an elastic system
having the same initial period of vibration as long
as the period was longer than 0.5 s.

A finite positive slope was assigned to the

postyield stiffness to account for the
strain-hardening characteristic, and the model was
called a bilinear model. The bilinear model does
not represent the degradation of loading and
unloading stiffnesses with increasing displacement
amplitude reversals (Fig. 6), and the model is not FIG. 6. Bilinear hysteresis model
suited for a refined nonlinear analysis of a
reinforced concrete structure.

Clough 's Degrading Stiffness Model: A qualitative model for the reinforced concrete was
developed by Clough (1966), who incorporated the stiffness degradation in the elasto-plastic model :
the response point during loading moved toward the previous maximum response point. The
unloading slope remained parallel to the initial elastic slope. This small modification improved the
capability to simulate the flexural behaviour of the reinforced concrete. Compared with the
elasto-plastic model, less energy is absorbed per cycle beyond yielding by Clough's degrading
model.

From the response analysis of a series of

single-degree-of-freedom systems, Clough (1966)
concluded that (a) the degrading stiffness model did
not cause any significant change in the ductility
demand of long-period structures (period longer
than 0.6 s) compared with the elasto-plastic model;
on the other hand, (b) the degrading stiffness model
required significantly larger ductility from
short-period structures than the corresponding
elasto-plastic systems; and (c) the response
waveform of a degrading stiffness model was
distinctly different from that of an ordinary
elasto-plastic model.

The model is relatively simple, and has been

used extensively in nonlinear analysis with the
inclusion of strain-hardening characteristics (Fig. 7). FIG. 7. Clough’s degrading stiffness model.

Takeda's Degrading Stiffness Model: A more refined and sophisticated hysteresis model was
developed by Takeda et al. (l970) on the basis of experimental observation. This model included
stiffness changes at flexural cracking and yielding, and also strain-hardening characteristics. The
unloading stiffness was reduced by an exponential function of the previous maximum deformation.
Takeda et al. also prepared a set of rules for load reversals within the outermost hysteresis loop.

5
These are major improvements over the Clough
(1966) model.

Failure or extensive damage caused by shear

or bond deterioration was not considered in the
model. The Takeda model, similar to the Clough
model, simulates dominantly flexural behaviour
(Fig. 8). Simplified Takeda hysteresis models
were proposed by Otani and Sozen (1972) and
by Powell (1975), using a bilinear backbone
curve.

To test the goodness of the Takeda model, FIG. 8. Takeda’s degrading stiffness models.
cantilever columns tested on the University of
Illinois earthquake simulator were analyzed (Takeda et al.1970). Calculated waveforms were
favourably compared with the observed waveform as shown in Fig. 9.

FIG. 9. Takeda model applied to column analysis (Takeda et al. 1970)

Takeda-Takayanagi Models: The amplitude of the exterior column axial load varies greatly due to
the earthquake overturning moment, and changes its moment-carrying capacity. Takayanagi and
Schnobrich (1976) incorporated the effect of axial force variation in the Takeda model by preparing
various backbone curves at different axial load levels (Fig. 10a).

A pinching action and strength decay are inevitable in a short and deep member due to bar slip
and deterioration in shear resistance. Takayanagi and Schnobrich (1976) introduced a pinching
action and strength decay in the Takeda model (Fig. 10b). Whenever a response point was located
in the positive rotation-negative moment range or the negative rotation-positive moment range, the
pinching was introduced. After the moment exceeded the yield level, strength decay was
incorporated. The values of guideline for strength decay and pinching stiffness were not related to
the member geometry and material properties.

6
 (a) axial force variation (b) pinching and strength decay
FIG. 10. Takeda-Takayanagi models (Takayanagi and Schnobrich 1976):

Degrading Trilinear Hysteresis Model: A

model that simulates dominantly flexural
stiffness characteristics was developed in
Japan (Fukada 1969). The backbone curve is
a trilinear shape with stiffness changes at
cracking and yielding. Up to yielding, the
model behaves in the same way as the
bilinear model. Once deformation exceeds
the yield point, the model behaves as a
perfectly plastic system. Upon unloading, the
unloading point is treated as a new “yield”
point, and unloading stiffnesses
corresponding to pre- and postcracking are
reduced proportionately so that the behaviour
becomes of the bilinear type in a range
between the positive and the negative yield
points (Fig. 11).

The degrading trilinear model can easily FIG. 11. Degrading trilinear model.
include strain-hardening characteristics. The hysteresis energy dissipation per cycle beyond the
initial yielding is proportional to the displacement, and the equivalent viscous damping factor
becomes constant. The fatness of a hysteresis loop is sensitive to the choice of a cracking point.

Comments
Many other hysteresis models have been proposed and used in the past. Figure 12 shows
attained ductility factors of single-degree-of-freedom systems with any of four flexural hysteresis
models: bilinear; Clough; Takeda; and degrading trilinear models. The four models have the same
backbone curve except the cracking point. The four models show similar variations of attained
ductility factors with periods, but attained ductility factors show a wide scatter from one model to
another, especially in a short-period range.

A reinforced concrete building is normally designed to behave dominantly in flexural mode, brittle
failure modes such as diagonal tension failure in shear being carefully prevented at the design stage.
Thus hysteretic models representing shear behavior were not studied.

7
 Fig. 12. Response of various hysteresis models

1.4 Reinforced Concrete Member Model

Inelastic deformation of a reinforced concrete member does not concentrate in a critical location,
but rather spreads along the member (Fig. 13). Various member models have been proposed to
represent the distribution of stiffness within a reinforced concrete member. The effect of gravity load
on the beam behaviour and the contribution of slabs to the structural stiffness will not be discussed.

FIG. 13. Deformation of beam under gravity and earthquake loads.

One-component Model: An elasto-plastic frame structure was analyzed by placing a rigid plastic
spring at the location where yielding is expected. The part of a member between the two rigid plastic
springs remains perfectly elastic. All inelastic deformation is assumed to occur in these springs (Fig.
14). This one-component model was generalized by Giberson (1967).

8
 FIG. 14. One-component member model.

A major advantage of the model is that inelastic member-end deformation depends solely on the
moment acting at the end so that any moment-rotation hysteretic model can be assigned to the
spring. This fact is also a weakness of the model because the member-end rotation should be
dependent on the curvature distribution along the member, hence dependent on moments at both
member ends. Consider two cases of moment distribution along a member AB with corresponding
curvature distributions as shown in Fig. 15. The inelastic rotations at the A end are given by the
shaded areas. For the same moments at the A end, case II causes larger inelastic rotation at the A
end. Consequently, this simple model does not simulate actual member behaviour. Furthermore, it is
not rational to lump all inelastic deformations at member ends.

FIG. 15. Inelastic rotation of beam: (a) moment; (b) curvature and inelastic rotation.

The stiffness of an inelastic spring is normally defined by assuming an asymmetric moment

distribution along a member with the infection point at midspan. The usage of the initial location of
the inflection point in evaluating spring properties was suggested by Suko and Adams (1971).
However, once yielding is developed at one member end, the moment at the other end must
increase to resist a higher stress, moving the inflection point toward the member centre. At the same
time, a large concentrated rotation starts to occur near the critical section. Despite rational criticisms
against this simple model, the performance of the one-component model is expected to be
reasonably good for a relatively low-rise frame structure, in which the inflection point of a column
locates reasonably close to mid-height.

A special-purpose computer program, SAKE (Otani 1974), for a regular rectangular reinforced
concrete frame structure and recent modifications (Powell 1975) to general-purpose computer
program DRAIN 2D (Kanaan and Powell 1973) used the one-component model.

Multi-component Model: In an effort to analyze frame structures well into the inelastic range under
earthquake excitation, an interesting model was proposed by Clough et al. (1965). A frame member
was divided into two imaginary parallel elements: an elasto-plastic element to represent a yielding
phenomenon, and a fully elastic element to represent strain-hardening behaviour. When the
9
member-end moment reaches the yield level, a plastic hinge is placed at the end of the elasto-plastic
element. A member-end rotation depends on both member-end moments. Aoyama and Sugano
(1968) adapted the two-component model, creating the multicomponent model (Fig.16), using four
parallel beams to account for flexural cracking, different yield levels at two member ends, and
strain-hardening. The deformation compatibility of the imaginary components is satisfied only at their
ends.

FIG. 16. Multi-component member model.

The multi-component model appears to have merit; rotation at one end of a member depends on
both member-end moments. In other words, the moment distribution along a member can be
approximately reflected in the analysis. However, the stiffness of the multi-parallel components must
be evaluated under a certain assumed moment distribution. Therefore, the stiffness parameters are
valid only under such a moment distribution, and are bound to be approximate when the moment
distribution becomes drastically different.

Giberson (1967) discussed the advantages and disadvantages of the one-component and the
two-component models, and concluded that the one-component model was more versatile than the
two-component model because the two-component model was restricted to the bilinear-type
hysteresis characteristics. This two-component model was used in a general-purpose computer
program DRAIN 2D (Kanaan and Powell 1973) for plane structure. The interaction of the bending
moment and the axial force was easily incorporated by simply changing the yield value of the
elasto-plastic component depending on the existing axial force.

Connected Two-cantilever Model: When a frame is analyzed under lateral loads only, the member
moment distributes linearly. From the similarity of moment distribution, the member can be
considered to consist of two imaginary cantilevers, free at the point of contraflexure and fixed at the
member end, and connected at the inflection point, satisfying the continuity of displacement and
rotation (Otani and Sozen 1972).

The flexibility relation of a member was formulated by assuming: (a) the inflection point did not
shift much during a short time increment; (b) free-end rotation and displacement were proportional to
the beam length and the square of the beam length, respectively; and (c) instantaneous stiffness for
shear-rotation and shear-displacement curves of a unit length reference cantilever could be defined
by hysteretic models.

The weakness of this method is that the member flexibility matrix is a function of the location of the
inflection point, which tends to shift rapidly when the sign of a member-end moment changes. This
causes a numerical problem. Consequently, this method cannot be recommended for a general
dynamic analysis. However, the method is useful for incremental static load analysis of a structure.

Discrete Element Model: In order to overcome difficult problems of variable stiffness distribution
along a member, the member can be subdivided into short line segments along the length, with each
short segment assigned a nonlinear hysteretic characteristic. The nonlinear stiffness can be
assigned within a segment, or at the connection of two adjacent segments.
Wen and Janssen (1965) presented a method for dynamic analysis of a plane frame consisting of
elasto-plastic segments. Consequently, the mass and flexibility of a member were lumped at the

10
connecting points on a tributary basis. Powell (1975) suggested using a degrading stiffness
hysteresis model for rigid inelastic connecting springs (Fig. 17a). Shorter segments were
recommended in a region of high moment, and longer segments in a low-moment region.
An alternative method is to divide a member into short segments, each segment with a uniform
flexural rigidity that varies with a stress history of the segment (Fig. 17b). Local concentration of
inelastic action can be easily handled by arranging shorter segments at the location of high
concentration of inelastic deformation (Takayanagi and Schnobrich 1976).
These methods are useful when more accurate results are required, or in the analysis of walls.
More computational effort is required compared with the other simple models.

FIG. 17. Discrete element model: (a) lumped inelastic stiffness; (b) distributed inelastic stiffness.

Distributed Flexibility Model: Once cracks develop in a member, the stiffness becomes
nonuniform along the member length. Instead of dividing a member into short segments, Takizawa
(1973) developed a model that assumed a prescribed distribution pattern of cross-sectional flexural
flexibility along member length. A parabolic distribution with an elastic flexibility at the infection point
was used (Fig. 18). The flexura1 flexibility at member ends was given by a hysteretic model
dependent on a stress history.

FIG. 18. Distributed flexibility model.

This is an interesting concept in analyzing an inelastic member. However, the parabolic flexibility
distribution may not describe the actual concentration of deformation at critical sections (normally at
member ends) due to flexural yielding and deformation attributable to slippage of longitudinal
reinforcement within a beam-column connection. The usage of inelastic springs at locations of
concentrated deformation in conjunction with this model may be a useful solution.

Summary
Various member models are reviewed, and their advantages and disadvantages are discussed.
11
These models have been developed specifically for earthquake response. Development of a simple
model for simultaneous gravity and earthquake situations is desired.
Member stiffness matrix, equilibrium of forces and continuity of displacement at joints are used to
formulate a series of linear equations under a given loading condition. Numerical integration
methods are used to solve the equation of motion under dynamic loading conditions. The response
of a structural model is evaluated by solving the set of linear equation incremental time steps.

1.5 Reliability of Analytical Models

Earthquake simulator tests provide interesting opportunities to examine the goodness of different
analytical models in simulating the observed response of small- to medium-scale highly inelastic
model structures. This section reviews the reliability of different analytical models in relation to the
capability to simulate the observed behaviour.

These test structures were designed to behave dominantly in flexure, being prevented as much
as possible from failing in shear or anchorage because the two types of failure are not desirable in
real construction and are avoided in a design process.

Three-storey One-bay Frames (I): Small-scale three-storey one-bay reinforced concrete frames
were tested on the University of Illinois earthquake simulator, and were analyzed using the
connected two-cantilever member model (Otani and Sozen 1972). The stiffness properties of
individual members were calculated on the basis of the geometry and material properties. The
Takeda model was used to represent the force-deflection of each cantilever model. Member-end
rotation due to bar slip was approximated by the simplified bilinear Takeda model, which did not
simulate the pinching behaviour. The damping matrix was assumed to be proportional to the
instantaneous stiffness matrix.

FIG. 19. Connected two-cantilever model applied to three-storey frame analysis

(Otani and Sozen 1972): (a) measured; (b) calculated (h=0.0); (c) calculated (h=0.02)

The model structure was subjected to a base motion simulating the El Centro (NS) 1940
accelerogram. The first-floor displacement was measured to be as much as four times the yield
displacement calculated under static lateral loads. The analytical models with and without viscous

12
damping favourably simulated the large-amplitude oscillations at 1.0, 2.0, and 5 s from the beginning
of the motion (Fig. 19). The analytical models, however, failed to simulate the medium- and
low-amplitude oscillations. Note that the frequencies at the medium- to low-amplitude oscillations are
higher for the analytical model, which indicates that the test structure was more flexible at low stress
levels than the analytical model. In order to reproduce lower amplitude oscillations of the observed
response waveforms, the pinching behaviour needs to be incorporated in a hysteretic model.

Three-storey One-bay Frames (II): Another set of three-storey one-bay small-scale reinforced
concrete frame structures was tested on the University of Illinois earthquake simulator (Otani 1976).
The base motion is significantly more intense than a design earthquake motion.

A member was represented by the one-component model with two inelastic rotational springs at
each member end: one for the flexural deformation and the other for the member-end rotation due to
bar slip (Otani 1974). Takeda models with trilinear and bilinear backbone curves were assigned to
the two inelastic springs. Two types of damping were used in the analysis: (a) a damping matrix
proportional to the constant mass matrix; and (b) a damping matrix proportional to an instantaneous
stiffness matrix. The first-mode damping factor was 5% of critical at the initial elastic stage.
Observed and calculated third-level displacement waveforms are compared in Fig. 20 (Otani 1976).
The comparison is fair for large-amplitude oscillations, and poor at low-amplitude oscillations. Again
in this analysis, the pinching characteristic was not incorporated. A fair agreement between the
computed and the observed may be attributable to the fact that the yielding was developed at most
member ends at the large-amplitude oscillations, and that the inflection point tended to be near the
midpoint of each member in such a low-rise frame structure.

FIG. 20. One-component model applied to three-story frame analysis (Otani, 1976):
(a) measured; (b) calculated (mass proportional damping);
(c) calculated (stiffness proportional damping).

Two-storey One-bay Frame: A two-storey one-bay medium-scale frame structure with slabs was
tested on the University of California earthquake simulator (Hidalgo and Clough 1974). The structure
was analyzed using the two-component model. In an effort to improve the correlation, the elastic
stiffness of the two parallel components was degraded as a function of the first-mode-response

13
amplitude history. The observed and the calculated second-floor displacement waveforms are
satisfactorily compared in Fig. 21. However, the parameters controlling stiffness degradation could
not be determined from the theory.

FIG. 21. Two-component model applied to two-story frame analysis (Hidalgo and Clough 1974).

Ten-storey Coupled Shear Walls: Ten-storey coupled shear walls were tested on the University of
Illinois earthquake simulator (Aristizaba1-Ochoa and Sozen 1976). Takayanagi and Schnobrich
(l976) divided a wall into short segments of uniform stiffness, and represented connecting beams by
the one-component model. The Takeda-Takayanagi model with changing axial force was assigned to
a wall element, and the Takeda-Takayanagi model with pinching action and strength decay was used
in a beam. It was judged that the usage of two-dimensional plane stress elements for the walls was
less desirable because such an approach might cost more computational effort without any
compensating increase in accuracy.

(a) displacement at level 10, in inches (1 in. = 25.4 mm) (b)Acceleration at level 10, g.
FIG. 22. Analysis of ten-story coupled shear wall (Takayanagi and Schnobrich 1976):

The comparison of the measured and calculated displacement and acceleration is excellent, as
shown in Fig. 22. It is necessary to include the effects of inelastic axial rigidity of the wall section and
pinching action and strength decay of the connecting beams to reproduce the maximum
displacement response and the elongation of the period. Some stiffness parameters for the walls
and connecting beams were defined on the basis of static tests of connecting beam-wall assemblies.

Summary
The favorable comparison of the measured and the calculated response waveforms encourages
the use of correct analytical and hysteretic models. It is desirable in developing a mathematical
model that all parameters of the proposed model should be evaluated on the basis of the geometry
of a structure and the properties of materials.

14
1.6 Three-dimensional Building Analysis

The development of analytical methods has made it feasible to discuss nonlinear behaviour of
reinforced concrete plane structures with a certain confidence. However, we have not yet reached a
point to discuss, with any confidence, the nonlinear behaviour of three-dimensional building
structures. Columns of a framed structure must resist lateral forces in two horizontal directions. The
stiffness is reduced significantly under biaxial lateral load reversals (Otani et al. 1979).

The first nonlinear analysis of a frame structure under horizontal biaxial ground motion was made
by Nigam (1967) with elasto-plastic columns. Multi-storey frames with rigid beams and floor slabs
under horizontal biaxial ground motion were studied by Pecknold (1974). Prager's kinematic
hardening theory as modified by Ziegler (1959) was used. The basic effect of biaxial inelastic
interaction is to produce a softer structure. Pecknold (1974) confirmed Nigam's finding that: (a)
horizontal biaxial ground motion increased ductility demand for stiff structures (initial natural period
less than 0.3 s); and (b) horizontal biaxial ground motion had little effect on the ductility demand of
flexible structures. Neither study included the stiffness degradation property.

Aktan et al. (1973) used the finite element technique to include the stiffness degradation through
the degradation of material properties. The biaxial ground motion was found to cause 20-200%
larger response than the uniaxial ground motions from columns for which the calculated deflection
under uniaxial ground motion exceeded approximately twice the yield deflection.

Takizawa and Aoyama (1976) extended the one-dimensional degrading trilinear hysteretic model
into a two-dimensional model on the basis of plasticity theories (Ziegler 1959). The proposed model
was judged to predict the significant trends of the biaxial behaviour of reinforced concrete test
columns. The effect of biaxial response interaction was reported to be significant for degrading
stiffness models, and not so important for nondegrading type hysteretic models.

A brief review of nonlinear analysis of three-dimensional structures indicates the necessity of

further study in this area.

1.7 Summary

The behaviour of reinforced concrete buildings, especially under earthquake motion, was briefly
reviewed. When a structure can be idealized as plane structures, the current state-of-the-art
provides useful and reliable analytical methods.

However, more research is required to understand the effect of slabs, gravity loads, and biaxial
ground motion on nonlinear behaviour of a three-dimensional reinforced concrete structure.

References:
Aktan, A. E., D. A. W. Pecknold, and M. A. Sozen, 1973. Effect of two-dimensional earthquake
motion on a reinforced concrete column, University of Illinois, Urbana, IL, SRS No. 399.
Aoyama, H., and T. Sugano, l968. A generalized inelastic analysis of reinforced concrete structures
based on the tests of members. Recent researches of structural mechanics. Contribution in
Honor of the 60th Birthday of Professor Y. Tsuboi, Uno-Shoten, Tokyo, pp. 15-30.
Aristizabal-Ochoa, J. D., and M. A. Sozen, 1976. Behaviour of ten-storey reinforced concrete walls
subjected to earthquake motion, University of Illinois, Urbana, IL, SRS No. 431.
Bertero, V. V., and E. P. Popov, 1977. Seismic behaviour of moment-resisting reinforced concrete
frames. In Reinforced concrete structures in seismic zones, American Concrete Institute, Special
Publication No. 53, pp. 247-292.
Celebi, M., and J. Penzien, l973. Experimental investigation into the seismic behaviour of critical
region of reinforced concret components as influenced by moment and shear, Earthquake
Engineering Research Center, University of California, Berkeley, CA, EERC 73-4.

15
Clough, R. W., l966. Effect of stiffness degradation on earthquake ductility requirements, Structural
and Materials Research, Structural Engineering Laboratory, University of California, Berkeley, CA,
Report 66-16.
Clough, R. W., K. L. Benuska and E. L. Wilson, l965. Inelastic earthquake response of tall buildings,
Proceedings, 3rd World Conference on Earthquake Engineering, New Zealand, Vo1. II, Session
II, pp. 68-89.
Cowell, W. L., 1965. Dynamic tests of concrete reinforcing steels, U.S. Naval Civil Engineering
Laboratory, Port Hueneme, CA, Technical Report 394.
Cowell, W. L., 1966. Dynamic properties of plain Portland cement concrete, U.S. Naval Civil
Engineering Laboratory, Port Hueneme, CA, Technical Report 447.
Fukada, Y., 1969. Study on the restoring force characteristics of reinforced concrete buildings (in
Japanese), Proceedings, Kanto District Symposium, Architectural Institute of Japan, Tokyo,
Japan, No. 40.
Giberson, M. F., 1967. The response of nonlinear multi-story structures subjected to earthquake
excitation, Earthquake Engineering Research Laboratory, California Institute of Technology,
Pasadena, CA, EERL Report.
Hidalgo, P., and R. W. Clough, l974. Earthquake simulator study of a reinforced concrete frame,
Earthquake Engineering Research Center, University of California, Berkeley, CA, EERC 74-13.
Jennings, P. C., and J. H. Kuroiwa, 1968. Vibration and soil-structure interaction tests of a nine-story
reinforced concrete building, Bulletin of the Seismological Society of America, 58, pp. 891-916.
Kanaan, A. E., and G. H. Powell, l973. DRAIN-2D, A general purpose computer program for dynamic
analysis of inelastic plane structures, Earthquake Engineering Research Center, University of
California, Berkeley, CA, EERC 73-6.
Mahin, S. A., and V. V. Bertero, 1972. Rate of loading effect on uncracked and repaired reinforced
concrete members, Earthquake Engineering Research Center, University of California, Berkeley,
CA, EERC 72-9.
Nigam, N. C., 1967. Inelastic interactions in the dynamic response of structures, Ph.D. thesis,
California Institute of Technology, Pasadena, CA.
Otani, S., 1974. SAKE - A computer program for inelastic response of R/C frames to earthquakes,
University of Illinois, Urbana, IL, Structural Research Series, No. 413.
Otani, S., 1976. Earthquake tests of shear wall-frame structures to failure, Proceedings, ASCE/EMD
(Engineering Mechanics Division) Specialty Conference, Dynamic Response of Structures,
University of California, Los Angeles, CA, Mar. 1976, pp. 298-307.
Otani, S., V. W. T. Cheung and S. S. Lai, 1979. Behaviour and analytical models of reinforced
concrete columns under biaxial earthquake loads, Proceedings, 3rd Canadian Conference on
Earthquake Engineering, Montreal, P.Q., ,pp. 1141-1168.
Otani, S., and M. A. Sozen, 1972. Behaviour of multi-story reinforced concrete frames during
earthquakes. University of Illinois, Urbana, IL, Structural Research Series, No. 392.
Pecknold, D. A. W., 1974. Inelastic structural response to 2D ground motion, ASCE Journal of the
Engineering Mechanics Division, 100(EM5), pp. 949-963.
Powell, G. H., l975. Supplement to computer program DRAIN-2D, Supplement to report, DRAIN-2D
user's guide, University of California, Berkeley, CA.
Suko, M., and P. F. Adams, 197l. Dynamic analysis of mu1tibay multi-story frames, ASCE Journal of
the Structural Division, 97(ST10), pp. 2519-2533.
Takayanagi, T., and W. C. Schnobrich, 1976. Computed behaviour of reinforced concrete coupled
shear walls, University of Illinois, Urbana, IL, Structural Research Series, No. 434.
Takeda, T., M. A. Sozen and N. N. Nielsen, 1970. Reinforced concrete response to simulated
earthquakes, ASCE Journal of the Structural Division, 96(ST12), pp. 2557-2573.
Takizawa, H., 1973. Strong motion response analysis of reinforced concrete buildings (in Japanese),
Concrete Journal, Japan National Council on Concrete, II (2), pp. l0-21.
Takizawa, H., and H. Aoyama, 1976. Biaxial effects in modelling earthquake response of R/C
structures, Earthquake Engineering & Structural Dynamics, 4, pp. 523-552.
Veletsos, A. S., and N. M. Newmark, l960. Effect of inelastic behaviour on the response of simple
systems to earthquake motions, Proceedings, 2nd World Conference on Earthquake Engineering,
Tokyo and Kyoto, Vol. II, pp. 895-912.
Wen, R. K., and J. G. Janssen, 1965. Dynamic analysis of elasto-inelastic frames, Proceedings, 3rd
World Conference on Earthquake Engineering, Wellington, New Zealand, Jan. 1965, Vol. II. pp.
713-729.

16
Ziegler, H. 1959. A modification of Prager's hardening rule. Quarterly of Applied Mechanics, 17(1), pp.
55-65.

17
 Chapter 2. Properties of Reinforced Concrete Materials

Material properties of reinforced concrete are briefly reviewed in this chapter. The following
references are recommended.
1. R. Park & T. Paulay: Reinforced Concrete Structures, John Wiley & Sons, Inc., 1975, 769 pp.
2. Comite Euro-International du Beton, Structural Concrete under Seismic Action, AICAP-CEB
Symposium, Rome, Vol. 1 - State of the Art Report, Bulletin d’Information No. 131, May 1979, 286 pp.
3. Comite Euro-International du Beton, Response of R.C. Critical Regions under Large Amplitude
Reversed Actions, Bulletin d’Information No. 161, August 1983, 306 pp. 4. Comite Euro-International
du Beton: RC Elements under Cyclic Loading - State of the Art Report, Thomas Telford, 1996, 190 pp.

2.1 Concrete
Concrete is a hardened material obtained from a carefully proportioned mixture of (a) cement, (b)
sand, (c) gravel and (d) water in forms of the shape and dimensions of the desired structure. The
compressive strength of concrete is determined by static test on either standard cylinders or cubes.
The strength varies with (a) concrete mix, (b) age of testing, (c) curing method, (d) specimen shape
and size, and (e) loading speed. The water-cement ratio is the main factor that controls the strength
of the concrete.

Compressive strength normally used in building construction ranges from 20 to 60 MPa. Higher
strength concrete is used in (a) columns of the lower part of a building to resist higher axial load, (b)
prestressed concrete members to balance the use of higher strength prestressing steel, and (c)
precast concrete for early removal of casting form.

Uniaxial Stress-Strain Relation in

Compression: The stress-strain relationship of
concrete under short-term monotonically
increasing uniaxial compressive loading shows
gradual deterioration in stiffness with strain
even at a low stress level caused by
development of micro cracks. However, the
curve may be represented by a straight line up
to approximately 70 percent of the compressive
strength The secant stiffness at a point at
one-third of the compressive strength is often
used to represent this linear portion. Maximum
resistance is attained at approximately 0.002
strain, followed by descending branch.

The stress-strain relation of concrete

in compression varies with the strength
of concrete. Higher strength concrete
exhibits high initial stiffness and steep
descending slope after attaining the
compressive strength. The strain at
compressive strength does not change
appreciably with the concrete strength.

It should be noted that the damage

does not distribute uniformly over the
height of a specimen; the damage
normally develops in the middle part of
the specimen because the concrete
strength at the specimen ends is
normally enhanced by the confinement. Stress-strain relation with concrete strengths
The lateral expansion by the Poisson’s
effect is resisted by the friction between the testing machine. The friction provides confining pressure
at the specimen ends, where the strength is enhanced; the damage concentrates in the middle part.
For a given stress level, a larger strain is measured in the middle part, and smaller strain near the
ends. The strain measurement is affected by the choice of gauge length especially in descending part
of the stress-strain curve.

The descending part of the load-deformation relation of a concrete specimen is extremely difficult
to measure because the elastic strain energy stored in the testing machine is released abruptly
during the descending part of the test, causing sudden failure of the testing specimen; a stiff loading
machine is needed for the purpose.

Young’s Modulus: Elastic (Young's) modulus Ec of concrete is normally defined as secant modulus
at approximately one-third of the compressive strength. The value of elastic modulus is often given by
the following empirical formula (Pauw, 1960):
Ec = 1.35 γ 1.5σ B
0.5
GPa
Ec = 33 w1.5 f 'c psi
in which γ : unit mass density of air dried concrete (1000 kg/m3), σ B : compressive strength (MPa)
of concrete, w : air-dry weight of the concrete at testing (pound per cubic feet), and f 'c : concrete
strength at testing (pound per square inches). The unit mass density γ of concrete may be taken as
2.5 (x1000 kg/m3) for normal weight concrete. A wide scatter of data should be observed from the
Pauw's empirical expression.

Reliability of Young’s Modulus Estimate (Pauw, 1960)

ACI364 report on high-strength concrete

Young's modulus of high-strength concrete is influenced by the type of coarse aggregates. An

empirical expression was proposed for the elastic modulus Ec concrete (Tomosawa, Noguchi, and
Onoyama, 1990), taking into account compressive strength and density of concrete, type of coarse
aggregates and mineral admixture;
Ec = k1 × k2 × 3.35 × 104 × (σ B / 60)1/ 3 × (γ / 2.4) 2 (MPa)
in which k1: factor representing type of coarse aggregates, k2: factor representing kind of mineral
admixture, σ B : observed concrete strength (MPa), γ : unit density of concrete (ton/m3). The factor k1
is 0.95 for crushed quartzite, crushed andesite, basalt and clayslate aggregates, 1.0 for other coarse
aggregates, and 1.2 for crushed limestone and calcined bauxite aggregates. Factor k2 is 0.95 for
silica fume, fine powder of blast furnace slag and fly ash fume, 1.00 for concrete without mineral
admixture or with other mineral admixture, 1.10 for fly ashes. Ninety-five percent of test data are
shown to fall within 20 percent of the empirical expression. The modulus is important in defining the
elastic period of a structure. Therefore, if the dynamic analysis procedure is used in the design the
modulus should be controlled, especially in a large and important construction project, within an
acceptable range from the value specified by a structural engineer.

Young's Modulus of High-strength Concrete

(Tomosawa, Noguchi, Onoyama, 1990)

Poisson’s Ratio and Shear Modulus: Poisson's ratio µ of concrete in "elastic" range is
approximately 1/6. Shear modulus Gc of concrete is defined by
Ec
Gc =
2(1 + µ )
The shear modulus is used to estimate shear modulus of structural walls, where shear deformation
may not be neglected in comparison with flexural deformation.
Stress-Strain Models under Monotonic
Loading: The parabolic expression up to the
maximum stress followed by straight σc
descending branch was used by E. Hognestad σo
0.15σo
(1951) to represent the uniaxial stress-strain
relation of the concrete. He suggested the use Linear
of 0.85 f’c for the concrete strength σ o in a
ε ε
structure, where f’c is the compressive strength σc = σo [2( c ) − ( c )2 ]
of concrete obtained from standard cylinder εo ε o
tests. The strength of concrete in a real
structure is smaller than the strength of Ec = tanα 2σo
εo =
concrete cylinders. In the descending branch, Ec εc
the loss of resistance at ultimate strain of
0.0038 was assumed to be 15 % of the 0.0038
εo = 0.002
maximum stress.
εc εc 2 Concrete Model (Hognestad, 1951)
σ c = σ o [2 −( ) ] for ε c ≤ ε o
εo εo
σ c = σ o [1 − Z (ε c − ε o )] for ε c > ε o
where σ c : concrete stress, σ o : compressive strength of concrete, ε c : concrete strain, ε o : strain at
compressive strength of concrete, defined as 2 σ o / Ec .

Smith and Young (1955) proposed an exponential function for the stress-strain relation ( σ c − ε c )
of concrete under monotonically increasing load;
σc ε
= K ( c )m
σo εo
where, σ o : compressive strength obtained from standard cylinders, ε 0 : strain at the peak stress.

Desayi and Krishnan (1964) proposed an expression;

E εc
σc =
εc 2
1+ ( )
ε0
where E : initial tangent modulus, ε 0 : strain at maximum stress.

Saenz (1964) in his discussion to Desayi and Krishnan (1964) suggested the following expression
for ascending branch of the concrete under monotonic loading;
3Eo ε 2E ε
σ c = E ε c [1 − ( − 2) c + (1 − o )( c ) 2 ]
E ε0 E ε0
E εc
σc =
E ε ε
1+ ( − 2) c + ( c ) 2
Eo ε0 ε0
σ0
where ε 0 : strain at maximum stress, Eo : secant modulus at maximum stress (= ), E : initial
εo
tangent modulus.

Saenz (1964) also suggested an expression including descending branch;

 εc
σc =
A + Bε c + Cε c + Dε c
2 3

where
1 RE + R − 2 2R − 1 R
A= B= C=− D=
E RE σ 0 RE σ 0 ε 0 RE σ 0 ε 02
RE ( R f − 1) 1 σ0E εf σ
R= − RE = R Rf =
ε = Eo = o
( Rε − 1) 2
Rε σfEo ε0 ε0
σ f , ε f : stress and strain at failure. The parameters A, B, C , D were selected to satisfy strains and
stresses at the origin (0,0), maximum stress ( ε 0 , σ 0 ) and failure point ( ε f , σ f ), initial tangent
dσ dσ
stiffness (= E ), slope (=0.0) at the maximum stress point ( ε 0 , σ 0 ).
dε ε =0 dε ε =ε o

Kent and Park (1971) modified the model by Hognestad (1951), and proposed to vary the
stiffness of descending branch taking into account the confining effect of concrete by lateral
reinforcement;
εc εc 2
σ c = σ o [2 −( ) ] for ε c ≤ ε o σc
εo εo
σo Unconfined
σ c = σ o [1 − Z (ε c − ε o )] for ε c > ε o
but σ c ≥ 0.2σ o Confined

0.5 0.5σ o
Z=
ε 50u + ε 50 h − ε o ε 50h
0.021 + 0.002σ o 0.2σ o
ε 50u = εc
σ o + 6.89
3 b" ε o = 0.002 ε 50u ε 50c ε 20
ε 50 h = ( ) p s Park and Kent Model, 1971
4 sh
where p s : ratio of volume of transverse reinforcement to volume of concrete core measured to
outside of hoops, b": width of confined core measured to outside of hoops, s h : spacing of hoops. The
strain ε o at maximum stress σ o is taken as 0.002.

Attard and Setunge (1996) proposed a stress-strain curve model for concrete applicable for
concrete strength range from 20 to 130 MPa. The main parameters employed to establish the
equation are Young’s modulus Ec , peak stress σ o , strain at peak stress ε o , and stress σ i and
ε i at the inflection point on the descending branch of the stress-strain curve;
ε ε
A( c ) + B( c ) 2
σc εo εo
=
σ o 1 + ( A − 2)( c ) + ( B + 1)( ε c )2
ε
εo εo
To allow for the difference between the in-situ uniaxial compressive strength and the cylinder strength,
the peak stress σ o may be taken as 0.9 times the cylinder strength. For the ascending branch of the
stress-strain curve,
Ecε o
A=
σo
( A − 1) 2
B= −1
0.55
and for descending branch;
σ i (ε i − ε o )2
A=
ε oε i (σ o − σ i )
B=0
The values Ec , ε o , σ i , ε i may be determined from:
Ec = 4370 (σ o )0.52
ε o = 4.11 (σ o )0.75 / Ec
σi
= 1.41 − 0.17 ln(σ o )
σo
εi
= 2.50 − 0.30 ln(σ o )
εo
where stresses and Young’s modulus are in MPa.

There have been many research works leading to proposals of mathematical or phenomenological
models for concrete under short-term uniaxial monotonic loading; e.g., Sargin (1971), Popovics
(1970), and Buyukozturk et al. (1971).
Creep and Shrinkage: Concrete, subjected
to constant-amplitude long-term loading,
continues to deform with age after
instantaneous elastic strain, phenomenon of
which is called "creep." The rate of strain
decreases with time; creep strain is
stabilized in three to four years. The creep
strain is influenced by (a) stress level, age of
first loading, (b) curing condition, (c) mix of
concrete, (d) shape and size of specimens,
and (e) reinforcement. Concrete without any
loading shrinks with time with loss of water
content. Shrinkage and creep strains cannot
be separated.

Rusch (1960) conducted long term

Creep curve for concrete with
loading tests on unconfined concrete. The
stress-strain relationship is influenced by
the loading rate. The strength of concrete
under long-term loading is less than that
under short tem loading. The tests
showed that sustained load strength of a
concentrically loaded concrete specimen
amounts to at least 75 percent, and on the
average to about 80 percent of the
strength determined in a short term test. If
the axial stress exceeding approximately
80 percent of the compressive strength is
applied over a long period, creep causes
compression failure of the concrete.

Strain- rate effect increases resistance Effect of Load Intensity and Duration (Rusch, 1960)
and stiffness of concrete, but the effect is
relatively smaller compared with that on
steel.
Compressive stress, kgf/cm2

Quasi-static

Strain, mm/mm
Loading Rate Effect on Stress-strain
relation (Rusch, 1960)
Behavior under Stress Reversals: Low cycle fatigue tests at the University of Colorado (Sinha,
Gerstle, and Tulin, 1964) on 6x12-in.
(150x300-mm) standard compression
cylinders led to the following
conclusions;
(a) The stress-strain relationships of
concrete under compressive load
histories possess an envelope curve,
which may be considered unique and
identical with the stress-strain curve
obtained under constantly increasing
strain.
(b) The stress-strain relationships of
concrete subjected to cyclic loading Stress-strain relationship under stress cycles
possess a locus of common points
which are defined as the point where the
reloading portion of any cycle crosses the unloading portion. Stresses above the common points
produce additional strains, while stresses at or below these points will result in the stress-strain path
going into a loop, repeating the previous cycle
without further permanent strain. It was also
observed that the values of the common points
depended on the minimum stress in the cycle;
i.e., the stress amplitude.

Karsan and Jirsa (1969) reported the

cyclic compressive tests on 3x5-in.
(76x127-mm) prisms with flared ends;
(a) The envelope curve of stress strain
paths under cyclic loading coincided with the
stress-strain curve for a specimen under
monotonic loading to failure.
(b) The Smith-Young expression (Smith
and Young, 1955) was found to be a good
Peaks of cycles and envelope curve
approximation of the envelope curves.
(c) The accumulation of strain under (K d Ji 1969)
constant maximum stress levels produced
failure when the envelope curve was reached;
the specimen, however, could be loaded to the
envelope curve regardless of the strain
accumulation prior to a given cycle. The strain
accumulation did not appear to reduce the
strength to a level below the envelope.
(d) The location of the common points
(intersection of unloading and reloading
curves) was dependent on the magnitude of
the maximum stress and strain of the previous
load cycle. The common points for loading
from nonzero levels were identical to the
common points corresponding to load cycles
starting at a stress level of zero.
(e) Examination of the location of the
common points shows that failure would be
produced under repeated loads with stresses
exceeding about 0.63 f’c, the maximum of the
stability limit. This limit was independent of the Variation of common points
minimum stress levels in the cycles.
 (f) Loading and unloading curves starting from a point within the stress-strain domain were not
unique, and the value of stress and strain at the peak of the previous loading cycle must be known to
estimate the response.

Uniaxial Hysteresis Models for

Concrete: Karsan and Jirsa (1969)
proposed a model for the uniaxial cyclic
behavior of concrete based on 46 short
rectangular column tests under cyclical
loading. The envelope curve could be
defined as the stress-strain curve obtained
under monotonic loading to failure. The
monotonic loading curve may be
approximated by the expression proposed
by Smith and Young (1955);
εc
σc ε (1− )
= 0.85 c e ε o

σo εo
where, σ o : compressive strength
obtained from standard cylinders, εo : Calculated stress strain relationship
strain at the peak stress.

The unloading curve is expressed by the second-degree parabola, which passes the following
three points;
(a) Point (σ E , ε E ) on the envelope curve from which the unloading curve or its extension
terminates,
(b) Common point (σ C , ε C ) , and
(c) Plastic strain point (0.0, ε P ) at which the unloading curve or its extension terminates.

The reloading curve is also expressed

by the second order parabola which
passes the following three points;
(a) Plastic strain point (0.0, ε P ) at
which the reloading curve or its extension
starts,
(b) Common point (σ C , ε C ) , and
(c) Point (σ E , ε E ) on the envelope
curve at which the reloading curve or its
extension reaches the envelope curve.

The unloading and reloading curves

intersect at the common point (σ C , ε C ) ,
which lowers the number of loading Calculated stress-strain relationship
cycles. The upper and lower limit curves
of the locus of common points (σ C , ε C ) are called common point limit and the stability limit curves
as follows;
εC εC
( ) (
εo
)
σC εo [1− ]
=β e 0.315+ 0.77 β
σo 0.315 + 0.77 β
where, β = 0.76 for the common point limit curve, and β = 0.63 for the stability limit point curve.
The plastic strain point is defined as follows;
 εP ε ε
= 0.093 ( E )2 + 0.091 ( E ) for loading, and
εo εo εo
εP ε ε
= 0.145 ( E ) 2 + 0.13 ( E ) for unloading.
εo εo εo

Park, Kent and Sampson (1971)

showed a simple model for the uniaxial
stress-strain relationship under cyclic loading
in the analysis of reinforced concrete
members. The stress-strain curve for
concrete is represented by a parabola for
ascending portion and straight line for
descending portion for monotonically
increasing strain (Kent and Park, 1971). A
linear stress-strain curve for concrete in
tension may be assumed to the tensile
strength. The curve under cyclic loading is
represented by straight lines. Upon
unloading from point E on the skeleton curve,
P k K t dS 1974
0.75 of the previous stress is lost without decrease in strain, whereupon a linear path of slope 0.25 Ec
is followed to point G. If the concrete has not cracked, it is capable of carrying tensile stress to point
K; but if the concrete has previously cracked, or if cracks form during this loading stage, the tensile
strains increase but no tensile stress develops. Upon reloading, the strain must regain the value at G
before compressive stress can be sustained again. If reloading commences before unloading
produces zero compressive stress, reloading follows one of the paths IJ. The average slope of the
assumed loop between E and G is parallel to the initial tangent modulus.

Darwin and Pecknold (1974) used Saenz’s equation (1964) for the monotonic ascending branch;
E εc
σc =
E ε ε
1+ [ − 2]( c ) + ( c ) 2
Eo εo εo
where E : tangent modulus of
elasticity at zero stress, Eo :
secant modulus at the point of
maximum compressive stress,
ε o : equivalent uniaxial strain at
the maximum compressive
stress. The falling branch after
attaining the maximum stress
expressed by a straight line
passing through maximum
stress point (σ o , ε o ) and point
(0.2 σ o , 4.0 ε o ) .

Straight unloading stiffness Darwin and Pecknold model of concrete (1974)

changes its slope at the turning
point; the initial unloading stiffness from the envelope curve is equal to the initial elastic tangent
stiffness. Straight reloading stiffness is parallel to the lower unloading stiffness and passes through
the common points. The stress levels of common points and turning points are given below;
 5
σ cp1 = σ en1
6
Region 1:
1
σ tp1 = σ en1
2
1 1
σ cp 2 = σ en 2 − min{ σ en 2 , σ B }
6 6
Region 2:
1 1
σ tp 2 = min{ σ en 2 , σ B }
2 2
1
σ cp 3 = σ en3 − σ B
6
Region 3: σ tp 3 = σ en 3 − 2(σ en 3 − σ cp 3 )
1
= σ en 3 − σ B
3
2
σ cp 4 = σ en 4
3
Region 4:
1
σ tp = σ en 4
3

Other models can be found in literatures by Blakely (1973) and Aoyama (1973).
Tensile Behavior: Tensile strength of concrete is
obtained by (a) splitting test of concrete cylinders
or (b) modulus of rupture test of concrete prism.
The splitting tensile test develops relatively
uniform tensile stress over the section of the
concrete cylinder. Tensile strength varies by
testing methods (pure tension, splitting and
modulus of rupture tests); splitting tensile
strength of concrete is often used in the
laboratory and is approximately 10 % of the
compressive strength for normal strength
concrete. Tensile strength of concrete in a real
structure or specimen is difficult to estimate
because the concrete is subjected to shrinkage
strain and accidental loading.

Stress-strain relation in tension is almost

linear up to the development of cracking; the
elastic strain is developed over the entire height
of the concrete specimen.

Large micro cracks concentrate near the

failure section. Once crack develops at the failure
plane, elastic deformation is released in slightly
damaged regions and crack width continues to
grow at the failure plane (Comite
Euro-International du Beton, 1996).
Development of Tensile Cracks
Reinhardt and his research group presented (Comite Euro-International du Beton, 1996)
detailed description of stress-strain hysteresis
behavior of concrete under tensile stress reversal
(Reinhardt et al., 1986, Yankelevsky and Reinhardt,
1989). Various other hysteresis models have been
presented for concrete in tension (Hillerborg et al.,
1976 and Duda, 1990). These hysteresis models for
concrete in tension are important in the finite element
analysis of reinforced concrete structures.

Concrete under Tensile Stress Reversals

(Yankelevsky and Reinhardt, 1989)
 Confining Effect: The strength
Confining Effect: The strength
and ductility of concrete increases
when subjected to lateral confining
pressure. Based on experimental
research, Richart, Brandtzaeg
and Brown (1928) proposed the
following expression for the
strength of concrete under
confining pressure;
f 'cc = f 'c + 4.1 f l
where f 'cc : axial compressive
strength of confined concrete,
f 'c : uniaxial compressive strength
of unconfined concrete, and f l :
lateral confining pressure.
Axial Stress-Strain Curve with Lateral Confining Pressure

Behavior Under Biaxial Stress Condition: Concrete compressive strength increases with lateral
compressive (confining) stress, but decreases with
lateral tensile stress under bi-directional stress state
(Kupfer, Hilsdorf and H. Rusch, 1969).

The modeling of stress-strain relationship under

biaxial and triaxial loading has been studied
extensively with the development of nonlinear finite
element analysis methods.

Mohr’s theory of failure is often used to estimate

the strength under combined normal and shear
stresses on a plane. The envelope of the failure
Mohr’s circle is often called failure envelope.

Modeling of Concrete Behavior under Multi-axial

Stresses: A mathematical model to represent
inelastic behavior of concrete under stress reversal
should be able to reproduce the following aspects
(Comite Euro-International du Beton, 1996); Bi-axial Strength of Concrete
(a) The capacity of accounting for inelastic
non-proportional unloading and reloading, (Kupfer Hilsdorf and H Rusch 1969)
(b) An adequate non-holonomic relationship between the state of stress and the stiffness of the
material.
(c) The capacity of accounting for the stress degradation as a function of the load history,
including post-peak behavior.
The above features are desired in the mathematical model, but the overall behavior of a reinforced
concrete structure may be insensitive to some aspects of the material behavior. The desirability of a
model is dependent on a type of problem in pursuit.

A large cooperative experimental research project (Gerstle et al. 1980) conducted under unified
testing procedures reported the following conclusions;
(a) Considerable scatter of measured strain were observed for concrete under applied multiaxial
stresses. Probabilistic methods are needed for more meaningful treatment of data.
(b) The behavior of concrete can be represented conveniently in terms of the octahedral normal
(hydrostatic) and shear (deviatoric) stresses σ 0 and τ 0 . The assumption of isotropy is not
substantially violated at a macroscopic level up to stress close to failure.
 1
σ 0 = (σ 1 + σ 2 + σ 3 )
3
1
τ0 = (σ 1 − σ 2 ) 2 + (σ 2 − σ 3 ) 2 + (σ 3 − σ 1 ) 2
3
1
ε 0 = (ε1 + ε 2 + ε 3 )
3
1
γ0 = (ε1 − ε 2 ) 2 + (ε 2 − ε 3 ) 2 + (ε 3 − ε1 ) 2
3
(c) The direct relationships between volumetric and deviatoric stresses and strains are governed
by the bulk modulus Ks and shear modulus Gs.
σ 0 (ε 0 ) E
K s (ε 0 ) = (= : elastic case)
3ε 0 3(1 − 2ν )
τ (γ ) E
Gs (γ 0 ) = 0 0 (= : elastic case)
2γ 0 2(1 + ν )
(4) A coupling effect between octahedral shear stress and volumetric strain is systematically
observed and can be described by introducing a coupling modulus Hs.
τ0
Hs =
ε0
Three stress and strain dependent moduli appear to be sufficient to describe the behavior of
concrete under monotonically increasing loads.

Models based on the theory of elasticity (Elwi and Murray, 1979, Stankowski and Gerstle, 1985,
Shafer and Ottosen, 1985, Buyukozturk and Shareef, 1985), models based on the theory of plasticity
(Han and Chen, 1987), and models based on the mechanism of micro-cracking or elastic damage
(Dougill, 1976, Resende and Martin, 1984, Krajcinovic and Fonseka, 1981, Mazars, 1984) are
introduced in Comite Euro-International du Beton Report (1996).

References:

Attard, M. M., S. Setunge, “The Stress Strain Relationship of Confined and Unconfined Concrete,”
Material Journal, American Concrete Institute, Vol. 93, No. 5, 1996, pp. 432 - 442.
Blakeley, R. W. G., et al., “Prestressed Concrete Sections with Cyclic Flexure,” Journal, Structural
Division, ASCE, Vol. 99, No. ST8, August 1973, pp. 1717 - 1742.
Buyukozturk, O., et al., “Stress-strain Response and Fracture of a Concrete Model in Biaxial
Loading,” Journal, American Concrete Institute, Vol. 68, No. 8, August 1971, pp. 590 - 599.
Buyukozturk, O. and S. S. Shareef, “Constitutive Modeling of Concrete in Finite Element Analysis,”
Computers and Structures, Vol. 21, No. 3, 1985.
Comite Euro-International du Beton, “Concrete under Multiaxial States of Stress, Constitutive
Equations for Practical Design, CEB, Lausanne, 1983, Bulletin d’Information No. 156.
Comite Euro-International du Beton: RC Elements under Cyclic Loading - State of the Art Report,
Thomas Telford, 1996, 190 pp.
Darwin, D. and D. A. W. Pecknold, “Inelastic Model for Cyclic Biaxial Loading of Reinforced
Concrete,” Structural Research Series No. 409, Department of Civil Engineering, University of
Illinois, Urbana-Champaign, July 1974.
Desayi, P., and S. Krishnan, “Equation for the Stress-strain Curve of Concrete“, Journal, American
Concrete Institute, Vol. 61, No.3, March 1964, pp. 345 - .
Dougil, J. W., On Stable Progressively Fracturing Solids,” Zeitschrift fur Angewandte Mathematik und
Physik, Vol. 27, Fasc. 4, 1976, pp. 423 - 437.
Duda, H., “Bruchmechanincsche Verhalten von Beton unter monotoner und zyklischer
Zugbeanstruchung,” Doctoral Thesis, Technical Hochschule Darmstadt, 1990.
Elwi, A. A., and D. W. Murray, “A 3D Hypoelastic Concrete Constitutive Relationship, Journal,
Engineering Mechanics Division, ASCE, Vol. 105, August 1979.
Gerstle, K. H., H. Aschl, R. Bellotti, P. Bertacchi, M. D. Kotsovos, H. Y. Ko, D. Linse, J. B. Newman, P.
Rossi, G. Schickert, M. A. Taylor, L. A. Traina, H. Winkler and R. M. Zimmerman, “Behavior of
Concrete under Multiaxial Stress States,” Journal, Engineering Mechanics, ASCE, Vol. 106, No.
6, December 1980, pp. 1383 - 1404.
Han, D. J., and W. F. Chen, “A Nonuniform Hardening Plasticity Model for Concrete Materials,”
Journal, Mech. Mat., Vol. 4, 1985.
Hillerborg, A. et al., “Analysis of Crack Formation and Crack Growth in concrete by Means of Fracture
Mechanics and Finite Elements,” Cement and Concrete Research, Vol. 6, 1976, pp. 773 - 782.
Hognestad, E., "A Study of Combined Bending and Axial Load in Reinforced Concrete Members,"
Bulletin No. 399, Engineering Experimental Station, University of Illinois, 1951.
Karsan, I. D., and J. O. Jirsa, “Behavior of Concrete under Compressive Loadings,” Journal,
Structures Division, ASCE, Vol. 95, No. ST12, December 1969, pp. 2543 - 2563.
Kent, D. C., and R. Park, "Flexural Members with Confined Concrete," Journal, Structural Division,
ASCE, Vol. 97, ST 7, July 1971, pp. 1969-1990.
Krajcinovic, D., and G. U. Fonseka, “The Continuous Damage Theory of Brittle Materials,” Journal,
Applied Mechanics, ASME, Vol. 48, 1981.
Kupfer, H., H. K. Hilsdorf, and H. Rusch, "Behavior of Concrete Under Bi-axial Stress," Journal,
American Concrete Institute, Vol. 66, No. 8, pp. 656-666, August 1969.
Mander, J. M., N. M. Priestley and R. Park, “Theoretical Stress-Strain Model for Confined Concrete,”
Journal, Structural Division, ASCE, Vol. 114, No. 8, 1988, pp. 1804 - 1826.
Mazars, J., “Description of the Multiaxial Behavior of Concrete under Multiaxial Conditions,”
INSA-OPS Toulouse, May 1984.
Park, R., and T. Paulay, Reinforced Concrete Structures, John Wiley & Sons, 1975, 769 pp.
Pauw, A., "Static Modulus of Elasticity of Concrete as Affected by Density," Journal, American
Concrete Institute, No. 57, No. 6, December 1960, pp. 679-687.
Popovics, S., “Stress-strain Relations for Concrete under Compression,” Journal, American Concrete
Institute, Vol. 67, No. 3, March 1970, pp. 243 - 248.
Popovics, S., “A Review of Stress-Strain Curve of Concrete,” Cement and Concrete Research, Vo. 3,
No. 4, September 1973, pp. 583 - 599.
Reinhardt, H. W. et al., “Tensile Tests and Failure Analysis of Concrete,” Journal, Structural
Engineering, ASCE, Vol. 112, No. 11, November 1986, pp. 2462 - 2477.
Resende, L., and J. B. Martin, “A Progressive Damage Continuum Model for Granular Materials,”
Comp. Meth. Appl. Mech. Engng, Vol. 42, 1984.
Richart, F. E., A. Brandtzaeg, and R. L. Brown, “A Study of the Failure of Concrete Under Combined
Compressive Stresses,” University of Illinois Engineering Experimental Station, Bulletin No. 185,
1928, 104 pp.
Rusch, H., “Researches Toward a General Flexural Theory for Structural Concrete,” Journal,
American Concrete Institute, Vol. 57, No. 1, July 1960, pp. 1 - 28.
Saenz, L. P., “Discussion of the paper by Desayi, P., and S. Krishnan entitled ‘Equation for the
Stress-strain Curve of Concrete’,” Journal, American Concrete Institute, Vol. 61, No. 9,
September 1964, pp. 1229 - 1235.
Sargin, M., “Stress-strain Relationships for Concrete and the Analysis of Structural Concrete
Sections,” Study No. 4, Solid Mechanics Division, University of Waterloo, Ontario, 1971, 167 pp.
Sinha, B. P., K. H. Gerstle, and L. G. Tulin, "Stress-Strain Relationships for Concrete Under Cyclic
Loading," Journal, American Concrete Institute, Vol. 61, No. 2, pp. 195-211, February 1964.
Smith, G. M., and L. E. Young, “Ultimate Theory in Flexure by Exponential Function,” Journal,
American Concrete Institute, Vol. 52, No. 3, November 1955, pp. 349 - 359.
Shafer, G. S., and N. S. Ottosen, “An Invariant-based Constitutive Model, Structural Research Series
No. 8506, Department of Civil Environmental and Architectural Engineering, University of
Colorado, Boulder, 1985.
Stankowski, T., and K. H. Gerstle, “Simple Formulation of Concrete Behavior under Multiaxial Load
Histories,” Journal, American Concrete Institute, Vol. 82, No. 2., March-April 1985.
Tomosawa, F., T. Noguchi, and K. Onoyama, "Investigation on Fundamental Mechanical Properties of
High-strength and Super High Strength Concrete (in Japanese)," Summaries of Technical Papers
of Annual Meeting of Architectural Institute of Japan, Vol. A, 1990, pp. 497-498.
Yankelevsky, D. Z. and H. W. Reinhardt, “Uniaxial Behavior of Concrete in Cyclic Tension,” Journal,
Structural Engineering, ASCE, Vol. 115, No. 1, 1989, pp. 166-182.
2.2 Reinforcing Steel

Stress-strain relation of a reinforcing bar under monotonically increasing tensile load shows a
linearly elastic region, yield plateau, strain-hardening region before fracture. Elastic modulus Es of
steel is approximately 210 GPa and Poisson's ratio µ is approximately 1/3. High strength steel or
work-hardened steel does not exhibit clear yield point nor yield plateau; a yield point is defined as the
intersection of a straight line parallel to the initial stiffness at 0.2 % strain and the observed
stress-strain curve.

Stress-strain relation of reinforcing steel

In reinforced concrete building construction,

yield strength of reinforcing bars ranges from
300 to 500 MPa, but higher strength of the
order of 700 to 1200 MPa is used for the lateral
reinforcement. Higher strength steel is
commonly used in prestressed concrete
construction.

With the strength of steel, yield plateau

becomes narrower, strain at fracture becomes
smaller, and the yield ratio (the ratio of the yield
stress to the tensile strength) decreases.
Properties of steel in compression are similar to
that in tension unless buckling of a bar does not
occur.

High rate of loading can significantly

increase the upper yield stress especially at a
rate of loading for blast. The stress increase in
the strain-hardening is relatively small. The
strain-rate properties were studied for blast
loading cases. Until the servo-controlled
actuators were developed, the well-controlled
testing and instrumentation was difficult for
reliable data.

When stress is reversed after plastic deformation in a direction, the range of linearly elastic
behavior is reduced, and stress-strain relation departs from linearly elastic response. This
phenomenon is called "Bauschinger effect." The stress after a stress reversal beyond initial yield
stress is increased compared to the stress at the same strain; the phenomenon is called “isotropic
strain hardening.”.
 Tensile stress, kgf/cm2
Room temperature

Strain, mm/mm

The range of cyclic strain history, to which reinforcing bars are likely to be subjected, differs
significantly from that of structural steel members; i.e., compressive strains are not as large as tensile
strains. Under a flexural condition, reinforcing bars resist tensile stresses after concrete cracking,
while compressive stress is resisted by concrete. When concrete cover spalls after crushing of
concrete, reinforcing bars may yield in compression. The Bauschinger effect is important in
simulating the behavior of reinforced concrete especially when the compression stresses are resisted
solely by reinforcing bars during crack opening immediately after unloading.

Any stress-strain relation under

repeated and reversed loading can be
decomposed into three parts, (a) skeleton
part, (b) unloading part and (c) softening
part. Kato Akiyama and Yamanouchi
(1973) demonstrated that if a series of
progressively larger hysteretic loops are
known for a given material, a monotonic
stress-strain curve can be constructed
with a reasonable degree of accuracy. by
plotting the stress-strain relations for
positive and negative directions
separately. In a cluster of cyclic hysteresis
loops, the bold lines are defined as the
skeleton parts, which are the portions of
the curves at stresses of the same sign
larger than the ones during the previous
cycles. Fine straight lines are the
unloading parts and the dashed lines are
softening parts in which the Bauschinger Reloading stiffness after unloading from the skeleton
effect is dominant. Connecting the
skeleton lines end to start, either above or below the horizontal axis given curves similar to those
found from monotonic experiments except for the first yielding region; i.e., any skeleton line is always
a part of the monotonic stress-strain curve. Any unloading line can be approximated by a straight line.
Reloading curve after unloading from the skeleton curve in the opposite direction can be expressed
by
 σ E
( − a)( B ε + a − 1) − a(1 − a) = 0
σs σs
E
a=
E − EB
where E : Young’s modulus, EB : secant modulus on the monotonic loading curve at stress level
σ s and strain ε B as shown below. The expression for a was obtained by equating the tangent
stiffness at zero stress level to be equal to the Young’s modulus. The secant stiffness EB can be
empirically expressed as
E
EB = − log10 10ε s
6
where ε s : accumulated skeleton strain.

The stress-strain relationship of metal is simulated by Ramberg and Osgood (1943) for the
skeleton and reloading cases;
r −1
ε σ σ
= (1 + )
εy σ y σy
r −1
ε − ε0 σ −σ 0 σ −σ0
= (1 + )
2ε y 2σ y 2σ y

Park, Kent and Sampson (1972) proposed the modification of the Ramberg-Osgood model as
follows:
r −1
σs σ
ε s − ε si = (1 + s )
Es σ ch
0.744 0.71
σ ch = σ sy { − + 0.2411}
ln(1 + 1000ε ip ) 1 − e1000εip
 4.49 6.03
r= − n + 0.297 for odd n
ln(1 + n ) ( e − 1)
2.20 0.469
r= − n + 3.04 for even n
ln(1 + n ) ( e − 1)

where ε s : steel strain, ε si : steel strain at zero

stress, σ s : steel stress, Es: elastic modulus of
steel, σ ch : stress dependent on the yield
strength and plastic strain in the steel produced in
the previous loading run, and γ : parameter of
the Ramberg-Osgood model, ε ip : plastic strain
in steel produced in previous loading run, n: Stress strain relation under monotonic loading
number of post yield loading runs with n = 0 for the first yielding. Unloading stiffness was taken equal
to the initial elastic stiffness.

A simple nonlinear stress-strain model of reinforcing bars is proposed by Zulfiqar and Filippou
(1990). The proposed steel stress-strain relation under monotonically increasing load consists of
three regions; (a) a linearly elastic region, (b) a plastic yield plateau, and (c) a nonlinear
strain-hardening range.
σ s = E0 ⋅ ε s for εs ≤ ε y
σ s = σ y1 + E1 ⋅ (ε s − ε y ) for ε y < ε s ≤ ε sh
σ s = σ y 2 + E2 ⋅ (ε s − ε sh ) for ε sh < ε s
The behavior under load reversals is expressed as
follows;
*
σ − σ S 1 − e − λε
σ =*
=
σ E − σ S 1 − e− λ
ε − εS 1
ε* = = − ln[1 − (1 − e − λ )σ * ]
εE − εS λ
ε −ε λ
E * = E0 E S =
σ E − σ S 1 − e−λ
where, parameter λ may be calculated for the
third equation for given stresses σ S , σ E and
strains ε S , ε E at the two outer most unloading
points.
Stress-strain relation under stress reversals
A summary of steel behavior is reported in
Bulletin d’Information No. 161 from Comite Euro-International du Beton, August 1983. Additional
model can be found in literatures (Peterson and Popov, 1977, Ma, Bertero and Popov, 1976).

References:

Aktan, A. E., B. I. Karlsson, and M. A. Sozen, “Stress-strain Relationships of Reinforcing Bars

subjected to Large Strain Reversals,” Civil Engineering Studies, Structural Engineering Series No.
397, University of Illinois at Urbana-Champaign, 1973.
Comite Euro-International du Beton, “Response of RC Critical Regions under Large Amplitude
Reversed Actions,” Paris, Bulletin d’Information No. 161, August 1983.
Kato, B., H. Akiyama and H. Yamanouchi, “Predictable Properties of Material under Incremental
Cyclic Loading,” Symposium on Resistance Ultimate Deformability of Structures Acted on by
 Well-Defined Repeated Loads, Reports of Working Commissions, Vol. 13, International
Association for Bridge and Structural Engineering, Lisbon, 1973.
Kent, D. C., and R. Park, “Cyclic Load Behavior of Reinforcing Steel,” Strain, British Society for Strain
Measurement, Vol. 9, No. 3, July 1973, pp. 98 - 103.
Ma, S. Y., V. V. Bertero and E. P. Popov, “Experimental and Analytical Studies on the Hysteretic
Behavior of Reinforced Concrete Rectangular and T-Beams,” Report No. EERC 76-2, Earthquake
Engineering Research Center, University of California at Berkeley, May 1976.
Park, R., D. C. Kent and R. A. Sampson, "Reinforced Concrete Members with Cyclic Loading,"
Journal, Structural Division, ASCE, Vol. 98, ST 7, pp. 1341-1360, July 1972.
Peterson, H., and E. P. Popov, “Constitutive Relations for Generalized Loadings,” Journal,
Engineering Mechanics Division, ASCE, August 1977.
Ramberg, W. and W. R. Osgood, "Description of Stress-Strain Curves by Three Parameters,"
Technical Note No. 902, National Advisory Committee for Aeronautics, July 1943.
Zulfiqar, N., and F. C. Filippou, “Models of Critical Regions in Reinforced Concrete Frames under
Earthquake Excitations,” Earthquake Engineering Research Center, Report EERC 90-06,
University of California, Berkeley, May 1990.
2.3 Bond

A mechanism of stress transfer between a reinforcing bar and surrounding concrete in the form
shear stress on reinforcement surface is called "bond," developed by (a) chemical adhesion between
mortar paste and bar surface, (b) friction on rough steel surface and (c) mechanical interlocking of
ribs with surrounding concrete.

Adhesion bond is known to break at an earlier

stage of relative deformation (slippage) of a bar
and concrete. Major source of resistance after
initial bar slippage is the mechanical interlocking.
At the inclined face of bar deformation, stress is
transferred by direct contact (bearing) stress
normal to the face and friction stress along the
face.

When the ribs on a deformed bar is high and

spaced too closely, the shear failure occurs along
the cylindrical concrete surface connecting the
top of adjacent ribs.

When the rib spacing is larger, the concrete crushes at the rib face, and crushed concrete forms a
wedge in front of the rib; hence surrounding
concrete is pushed outward by force in the
reinforcing bar, and tensile force, called "ring (or
circumferential) tension", is developed in the
circumferential direction of the bar face.
Therefore, surrounding concrete is split by this
ring tension stress.

Bond stress (average shear stress over bar

surface) free-end slip relation is shown below
for a pull-out test, in which concrete strength
was 25 MPa. Deformed bar D16 and round bar
16 φ were used in the test. Maximum bond
stress is developed at approximately 0.1 to 0.3
mm slip deformation.

The bond strength may be affected by (a)

concrete strength, (b) deformation of a
reinforcing bar surface, (c) bar size (Poisson's
effect), (d) bar stress (tension or compression),
(e) bar location, (f) concrete cover depth over a
bar, (g) lateral reinforcement, (h) lateral
pressure, and (i) anchorage length.
 Eligehausen et al. (1983) suggests bond stress-bar slip relation under monotonically increasing
loading;
s
τ = τ 1 ( )α for s ≤ s1
s1
τ = τ1 for s1 < s ≤ s2
τ 3 − τ1
τ = τ1 + ( s − s2 ) for s2 < s
s3 − s2
The parameters are defined in the figure and table.

Bond stress- slip relation under

monotonic loading

Cycles with reversed loading produce degradation of bond strength and bond stiffness. The rate of
degradation is more pronounced under reversed loading than under uni-directional loading.
Degradation depends on the previous peak slip in either direction, the number of cycles and the
difference of the peak slips in the two directions. Significant strength decay is observed with slip
amplitude under reversed loading. If the peak bond stress during cycling does not exceed
approximately 70 percent of monotonic bond strength, the bond stress-slip relation is stable.

Bond stress-slip relation under one-sided slip

 The deterioration of stiffness under
monotonically increasing loading is
described by Eligehausen et al.
(1983). At low bond stress (Point A),
inclined cracks begin to propagate
from the top of the ribs; their growth
and size are controlled by the
confining pressure provided by
transverse reinforcement. Transfer of
forces is more by bearing with a
shallow angle of inclination.
Increasing the load, local crushing of
concrete in front of the lugs produces
reduction of the tangent of the bond
stress-slip curve (Point B). When the
maximum bond stress is attained the
concrete key is sheared off, forming a
cone with a length of about four times
the lug height. With increasing slip the
bond stress begins to drop slowly. As
the bond shear cracks reach the
bottom of the adjacent lug (Point D)
the bond stress begins to drop, and by
the time the slip reached the lug
spacing only the frictional component
remains (Point E).

The deterioration of stiffness under

reversed loading is also described by
Eligehausen et al (1983). The initial
loading follows the monotonic curve,
but the cyclic load behavior is
sensitive to the level of slip at which Bond stress-slip relation with respect to damage
the reversal occurs. (Eligehausen, et al., 1983)
Case 1: If a slip reversal is imposed before inclined cracking, unloading branch exhibits high
stiffness (path AF) because only a small part of the slip is caused by inelastic concrete deformation.
As soon as slip in the opposite direction is imposed, the friction branch is reached (path FH). The
slope of this portion of the curve is small because the surface of the concrete surrounding the bar is
smooth. As soon as the cracks close, the stiffness differs little from that of the monotonic envelope
(point I). Unloading from point I, where the slip in the two directions is about equal, the curve (path
IKL) is very similar to that from the initial unloading curve (path AFH). The major difference is that due
to previous cracking and crushing of the concrete in front of the ribs, the point where the bond
stresses begin to pick up again (point L) will be shifted to the right of the origin. The lug will not be
bearing fully until point M is reached. Further loading follows the bond-slip curve up to the monotonic
envelope.

Case 2: If unloading occurs after the inclined cracks were formed, and therefore near the slip at
which ultimate bond stress has been attained, the unloading path is similar to that of the first case up
to point F. Since there is more damage to the concrete, a higher frictional resistance is mobilized
(point G). When the loading is reversed the lug presses against a key whose resistance has been
lowered by inclined cracks over a part of its length that were induced by the first half-cycle. The
splitting cracks created in the first half-cycle close at a higher load than those of the first case (point
H), and lead to an earlier formation of splitting cracks in the opposite directions. Splitting cracks,
combined with the existing inclined cracks along the bar, result in a reduced envelope (path HI) and a
reduction of bond capacity in the second direction (point I). Unloading from this peak (path IKLMN)
and reversing the load results in a reduced stiffness and strength because only the remaining
uncrushed concrete between the lugs must be sheared off. The bond strength (point N) is
substantially lower than that of point C, and lower than that of point I.
 Case 3: If unloading occurs after the
slip reached a value much larger than
the slip at maximum strength (point C),
the behavior is very poor. Since more
damage has occurred, the friction
resistance (point G) is larger than for
either of the previous cases. However,
since the concrete between the lugs is
completely sheared very little force can
be transmitted by bond when direction
of loading is reversed (path HIJ).
Unloading and reloading in the opposite
direction (path JKLMN) results in very
little additional bond capacity beyond
that provided by friction since most of
the mechanical anchorage is lost.

The first analytical model of the bond

stress-slip relationship for reversed
cyclic loading was presented by Morita
and Kaku (1973). The monotonic
loading envelopes of bilinear type were
varied for loading in compression and
tension. The deterioration of bond
resistance at peak slip and frictional
bond resistance with the number of
cycles is not considered in the model.

Bond stress-slip relation with respect to damage

(Reversed loading) (Eligehausen, et al., 1983)
 Additional models for bond stress-bar slip hysteresis models can be found in Refs. Viwathanatepa,
S. and et al. (1979), Hawkins et al. (1982) and Balaz (1989).

References:
Balazs, G. L., “Bond Softening under Reversed Load Cycles,” Stui e Ricerche - Corso Flli. Presenti,
Pollitecnico di Milano, No. 11, 1989, pp. 503 - 524.
Eligehausen, R., V. V. Bertero, and E. Popov, “Local Bond Stress-slip Relationships of Deformed
Bars under Generalized Excitations,” Earthquake Engineering Research Center, Report No.
EERC 83-23, University of California, Berkeley, October 1983.
Hawkins, H. M., et al., “Local Bond Strength of Concrete for Cyclic Reversed Loadings,” Bond in
Concrete, P. Bartos (ed.), Applied Science Publishers Ltd., London, 1982, pp. 151 - 161.
Morita, S. and T. Kaku, “Local Bond Stress-slip Relationship under Repeated Loading,” Proceedings,
IABSE Symposium on Resistance and Ultimate Deformability on Structures Acted on by Well
Defined Repeated Loads, Lisbon, 1973, pp. 221 - 227.
Viwathanatepa, S. and et al., “Effect of Generalized loadings on Bond of Reinforcing Bars embedded
in Confined Concrete Blocks,” Earthquake Engineering Research Center, Report No. EERC 79/22,
University of California, Berkeley, August 1979.
 Assignment No. 1
20020222
S. Otani

When a concrete specimen is tested under monotonically increasing deformation, the descending
part of the stress-strain relationship is difficult to obtain due to the sudden release of elastic energy by
the frame of a testing machine.
In order to understand this phenomenon, consider the following mathematical model, in which a
concrete cylinder is subjected to forced displacement through a linearly elastic spring representing
the stiffness of the testing machine.
The forced displacement is applied to the system at a constant rate of 0.1 mm/sec. The
force-deformation of the concrete specimen is given below. Plot the load-deformation relation of the
concrete specimen and of the loading point at 0.1 sec interval.
Discuss the force deformation relation of the concrete for the two cases for the stiffness of the
testing machine; elastic stiffness of (a) 100 kN/mm, (b) 150 kN/mm, and (c) 300 kN/mm.

Constant
displacement

300
Force, kN

Testing
Machine

Concrete
Specimen
0.5 1.0 3.0

Displacement, mm
Solution:

1) Equilibrium of forces: Forces in the concrete cylinder Pconcrete and the frame of the testing
machine Pspring should be the same, and equal to the force P in the testing machine.
P = Pconcrete = Pspring
2) Displacement compatibility: The applied displacement D by the testing machine should be equal
to the sum of displacement Dconcrete in the concrete cylinder and displacement Dspring of the frame.
D = Dconcrete + Dspring

The stress-strain relation of the concrete is piece-wise linear and the stiffness of the testing machine
is linearly elastic. Therefore, we need to consider the three points.

Point A:
P = Pconcrete = Pspring = 300 kN
Dconcrete = 0.5 mm
A B
Dspring = 300 kN / kspring 300
D = 0.5 + (300 / k spring )
Point B:
Force, kN

P = Pconcrete = Pspring = 300 kN

Dconcrete = 1.0 mm
Dspring = 300 kN / kspring Concrete C
D = 1.0 + (300 / k spring )
0.5 1.0 3.0
Point C:
P = Pconcrete = Pspring = 0.0 kN Displacement, mm
Dconcrete = 3.0 mm
Dspring = 0.0 kN / kspring
D = 3.0 mm

Displacement at the loading head.

Point A B C
Spring 100 kN/mm 3.5 mm 4.0 mm 3.0 mm (not controllable)
Spring 150 kN/mm 2.5 mm 3.0 mm 3.0 mm (sudden failure)
Spring 300 kN/mm 1.5 mm 2.0 mm 3.0 mm
 Chapter 3 Behavior of Reinforced Concrete Members

3.1 Behavior of Beams

A beam member, normally subjected to bending and shear without axial force, can develop large
deformation capacity and dissipate ample hysteresis energy during an earthquake well in the inelastic
range as long as brittle shear and bond splitting failures are prevented. Four modes of failure should
be considered when the behavior of reinforced concrete members is discussed; i.e., (a) flexural
failure caused by crushing of concrete or buckling of longitudinal reinforcement in compression or
tensile fracture of longitudinal reinforcement, (b) shear failure in diagonal tension mode or shear
compression mode or shear tension mode, (c) bond splitting failure along the longitudinal
reinforcement and, and (d) anchorage failure of longitudinal reinforcement at the member end. The
building code places design requirements to prevent these modes of failure.

The behavior of beams is influenced by (a) dimensions of section, (b) member length, (c) amount
of longitudinal reinforcement, (d) anchorage of longitudinal reinforcement at member ends, (e)
concrete strength, (f) amount and detailing of lateral reinforcement, (g) contribution of slabs, and (h)
loading (one-way monotonic, one-way cyclic, or reversal).

The beam is normally designed to develop flexural yielding at the member ends under earthquake
loading. The amount of tensile reinforcement and the effective depth control the flexural resistance of
reinforced concrete section. The use of compressive reinforcement does not increase the flexural
strength, but it contributes to the deformation capacity after flexural yielding. The concrete strength
has little influence on the flexural strength of girders without axial forces. A large deformation capacity
and stable hysteresis energy dissipation can be expected from the beam dominantly behaving in
flexure if a member is properly designed to prevent shear failure, bond splitting failure and anchorage
failure of longitudinal reinforcement at the formation of flexural yield hinges at the member ends; the
gravity load should be included in the evaluation of design forces. The use of large diameter bars as
longitudinal reinforcement may result in bond splitting failure along the longitudinal reinforcement.

Lateral reinforcement is effective (a) improving shear resistance, (b) confining core concrete to
enhance resistance and deformation capacity, (c) improving bond splitting resistance, and (d)
improving resistance against buckling of longitudinal reinforcement. The spacing and detailing of
lateral reinforcement is important. The concrete strength is important in defining shear strength of a
reinforced concrete member. It should be noted that the shear strength is not a unique value, but that
shear resistance deteriorates with damage in concrete after flexural yielding.

The slab monolithically cast with a girder

contributes to the stiffness and flexural resistance
of the girder. This increase in flexural resistance
should be considered in determining the design
shear of a member in preventing shear failure.

An uncountable number of beam specimens

have been tested, under monotonically increasing
loading and under load reversals, in structures
laboratories throughout the world. These
specimens are generally tested under simulated
earthquake loading without vertical gravity loading
acting on slabs; i.e., the inflection point was
assumed to locate at mid-span of the beam.

In old days, specimens were tested under a

simple support condition or in a cantilever
configuration, but more recently, beam specimens Examples of Anti-symmetric Bending Test
have been tested under anti-symmetric bending
conditions. Some specimens were tested with adjacent slabs.
Flexural Deformation and Shear Deformation: The flexural deformation is associated with the
Bernoulli’s hypothesis that the plane section normal to the member axis before bending remains
plane after bending. The flexural deformation is attributed to the normal stress acting on the section.
Shear force causes deformation in the direction normal to the member axis. There exists the
interaction of flexural and shear deformation, and the two deformations cannot be clearly separated
after the formation of cracking. Additional rotational deformation of a beam is resulted near the critical
section of bending from the pull-out of the longitudinal reinforcement from its anchorage zone.

The flexural deformation (average curvature) is obtained from longitudinal strain measurements at
the top and bottom chords assuming that a plane section remains plane after bending. Similar to the
flexural deformation, a shear deformation index is defined from strain measurements in the two
diagonal directions. This index does not represent the true shear deformation because the
deformation cannot be simply divided into the classical flexure and shear deformations once a shear
cracking occurs in a member.

Flexural deformation index Shear deformation index

(A '1 − A1 ) + (A '2 − A 2 )
θ= : flexural deformation index
h
δ5 − δ6 (h 21 + A1 )
2

γ= : shear deformation index

2 h1 A1
h1 + h2
h1 =
2
A + A7
A1 = 6
2

A typical moment-average curvature curve obtained from a simply supported beam test (Celebi
and Penzien, 1973) shows that the stiffness during loading gradually decreases with load, forming a
fat hysteresis loop and absorbing a large hysteresis energy. The hysteresis loops remain almost
identical even after several load reversals at the same displacement amplitude beyond yielding. A
typical lateral load-shear deformation index curve obtained in the same specimen shows the stiffness
during loading gradually increases with load, exhibiting a "pinching" in the curve. The hysteretic
energy dissipation is smaller. The hysteresis loop decays with the number of load reversals, resulting
in a smaller resistance at the same peak displacement in each repeated loading cycle. Although the
curve shows a "yielding" phenomenon, it is important to recognize that the shear force of the member
was limited by flexural yielding at the critical section rather than by yielding in shear. This yielding
clearly indicate the interaction of shear and bending.
Flexure Dominant Behavior: In the virgin loading branch, flexural cracking near the cross-section of
maximum moment reduces the initial elastic stiffness. The specimen continues to soften with loading
after cracking due to the spreading of cracking along the specimen length, crack opening associated
with tension softening of concrete and bond-slip of the reinforcement between cracks. Yielding of
tensile reinforcement causes an abrupt and sharp reduction in stiffness. Even after the flexural
yielding, the resistance continues to increase due to the shift of neutral axis and later due to strain
hardening of reinforcement. Spalling of the concrete in compression has a negative effect on the
resistance.

Upon unloading after post-yielding, the

unloading stiffness is generally high, but gradually
softens at lower loading level. A significant residual
deformation exists even after the removal of loads
caused by permanent strain in longitudinal
reinforcement and residual bar slip. Cracks remain
open at the removal of loads due to the residual
bar slip. The overall unloading stiffness degrades
with increasing plastic deformation amplitudes.

Reloading stiffness immediately after load

reversal is generally low until opened cracks close; Dominant flexural behavior of beam
the compression by bending moment must be (Celebi and Penzien, 1973)
resisted by the compressive reinforcement. The
reloading stiffness gradually recovers with the closing of cracks. Although the first post-elastic
excursion may be considered as virgin loading in the reloading direction, the softening is more
gradual than in the initial yielding direction partly attributable to the Bauschinger effect of the steel.
The resistance at the previous maximum displacement reaches the level of the previous maximum
resistance.

When the reloading branch reaches the previous maximum response point, further loading
proceeds along the continuation of the virgin loading branch. Flexural failure of beams due to cyclic
loading is gradual, controlled by progressive deterioration in the compressive zone such as spalling of
concrete and local buckling of the longitudinal reinforcement. The reinforcement sometimes fracture
in tension in the subsequent half cycle after bucking.

Clear definition of failure is difficult in flexure dominated members unless tensile fracture of
longitudinal reinforcement is observed. Therefore, failure is often defined in the experiment as a point
where the resistance can not recover approximately 80 percent of the maximum resistance. It should
be noted that this definition of failure point is affected by the loading history.

Member deformation capacity is defined as the ratio (ductility ratio) of the deformation at failure to
the deformation at flexural yielding. The definition of yield point is sometimes difficult to determine
especially when the tensile reinforcement is placed in double layers.

Behavior of flexure dominated beam under load reversals

Role of Compressive Reinforcement:

The compressive reinforcement placed in
section will not contribute to the flexural
strength of the section. Gaston, Siess and
Newmark (1952) tested five beams of
152x274 mm under monotonically
increasing load. Main parameter in the test
was the amount of tensile reinforcement.
The tensile reinforcement ratio ρ was
increased from 0.34% (Specimen T1La) to
1.90% (Specimen T4La) in the first four
specimens. The test results showed that
the bending resistance increased
proportional to the tensile reinforcement
ratio, but the deformation capacity
decreased with increasing amount of
tensile reinforcement. Concrete must resist
compressive force equal to the yielding
force in tensile reinforcement, and failed at
early stage when larger compressive forces Effect of tensile and compressive reinforcement
were demanded by the tensile Gaston, Siess, and Newmark, 1951
reinforcement.

In the last specimen (Specimen C4xna), the tensile reinforcement ratio was 1.90%, but 0.98%
compressive reinforcement was added in the section. The result showed small increase in flexural
resistance of Specimen C4xna compared with the resistance of Specimen T4La, but the deformation
capacity was significantly improved by the use of compressive reinforcement.

Flexural Member with Unsymmetric Cross Section: Beams are often not symmetrically reinforced
at the top and bottom. Even if the top and bottom are equally reinforced, the contribution of slab
reinforcement to beam flexural resistance will result in bending resistances different in positive and
negative bending. During the reloading in the weak direction with open cracks on the compression
side, tensile yielding is not enough to cause the compression bars to yield and cracks to close.
Therefore, during the reloading in the weak direction after yielding in the strong direction, cracks
remain open over full-depth near the critical section, the compression stress under bending must be
resisted by the compression reinforcement. The pinching of hysteresis shape occurs upon reloading
in the strong direction because high compression stress must be resisted by small amount of
longitudinal reinforcement before cracks closing.

Failure of unsymmetrically reinforced section occurs in two forms; (a) gradual progressive
compressive failure of the weaker side, and (b) abrupt fracture of tensile reinforcement in the weak
side.

Test of full-scale seven-story RC structure (Hiraishi et al., 1985)

Effect of Shear-span-to-depth Ratio: Shear-span to depth ratio is the most significant parameter
that influences the shear deformation characteristics. In a beam of small shear-span-to-depth ratio,
shear deformation becomes appreciable compared with bending deformation. Dominant shear
response causes a more pronounced pinching in the
force-deformation (hysteresis) curve, and a faster
degradation of the hysteresis energy dissipating capacity.
Considerable improvements in delaying and reducing the
degrading effects can be accomplished by using closely
spaced ties. The state of the knowledge is not sufficient to
define the stiffness degrading parameters on the basis of
the member geometry and material properties.

When the shear span-depth ratio becomes less than

two, the effect of shear becomes important. Garstka et al.
(1993) demonstrated the effect of shear-span-depth ratio
on the force-deformation curves under monotonically
increasing loading. The compression zone of the concrete Effect of shear span-depth ratio on
is severely damaged with a decreasing shear span-depth behavior (Garstka et al., 1993)
ratio by the diagonal strut action.

Shear Failure Modes: Shear failure of reinforced concrete members takes place in the from of (a)
diagonal tension failure, (b) shear compression failure, and (c) shear tension failure.

Shear failure became notorious by the experience of abrupt shear failure of members in diagonal
tension mode. This mode of shear failure occurs abruptly in a relatively slender member with light
amount of lateral reinforcement; i.e., the amount of lateral reinforcement is not sufficient to resist
tensile forces previously carried by concrete before diagonal shear cracking.

Failure in shear compression mode is relatively ductile, in which concrete in diagonal direction fails
in compression after tensile yielding of lateral reinforcement.

Failure in shear tension mode takes place in the form of bond splitting failure along longitudinal
reinforcement after formation of diagonal shear cracks.

Shear strength may be affected by (a) tensile reinforcement ratio, (b) shear span-to-depth ratio, (c)
shear reinforcement ratio, and (d) arrangement of shear reinforcement.

The shear reinforcement is inactive up to the occurrence of inclined cracking. Shear reinforcement,
upon yielding, develops large plastic strain. During unloading stage, large residual strain remains due
to relative movement along inclined shear cracking. Unloading and reloading after yielding of lateral
reinforcement leads to a gradual build-up of permanent tensile strain in lateral reinforcement. During
reloading, slippage along the inclined cracking takes place at low stiffness exhibiting a pinching effect.
Bond Splitting Failure: Another failure mode is bond
splitting failure along the longitudinal reinforcement.
The stress in the longitudinal reinforcement must be
transferred to surround concrete within the beam. Specimen No 3
試験体
This is a critical when large diameter bar or high
strength bars are used for longitudinal reinforcement.

Koda, Otani and Aoyama (1987) tested T-shape

beam with large-diameter bars (2-D25 and 1-D19,
tensile reinforcement ratio pt =0.0144). The lateral
reinforcement ratio at the end was 0.98 % (3-D10 at
87 mm) in the two specimens; Specimen No. 9 had a
constant lateral reinforcement but bond along the Specimen No.9
longitudinal reinforcement was removed by placing
wax around the longitudinal reinforcement. The
lateral reinforcement ratio was reduced to 0.65 %
(2-D10 at 87 mm) in the middle span in Specimen No.
3.

Specimen No. 9 without bond resistance along the

longitudinal reinforcement clearly shows pinching
Deformation, mm
hysteresis with lower initial stiffness; significant decay
in the second loading cycle at each deflection
amplitude is observed. Bond splitting cracks
developed in Specimen No. 3 at a member deflection
angle of 4/100.

Effect of Lateral Reinforcement on Bond Splitting

Failure: Reinforced concrete beams are normally
designed to develop flexural yielding at the member
ends, and resistance sufficient to prevent shear
failure or bond splitting failure is provided. A series of
tests on one-half scale reinforced concrete beams
(175 x 270 mm, effective depth was 243 mm) with
slab were conducted by Fujisawa et al. (1988) to
study the effect of lateral reinforcement under load
reversals. Shear span-to-(overall) depth ratio was 2.5 for all specimens. Tensile reinforcement ratio
pt was 1.20 % (4-D13) in all specimens. Slab thickness was 50 mm, and width was 500 mm on
each side reinforced with 6-D6 bars.

The test results of four specimens are compared to study the effect of lateral reinforcement on the
bond-splitting behavior. The lateral reinforcement ratio pw was 0.30 % in Specimen No. 1
(2-D6@123 mm), 0.60 % in Specimen No. 2 (2-D6@61 mm), and 0.90 % in Specimen No.3
(3-D6@61 mm). In Specimen No. 4, lateral reinforcement ratio within distance D (overall beam
depth=270 mm) from the end was 0.60 %, but the ratio was reduced to 0.30 % in the middle region.

Specimen No. 1 with light lateral reinforcement ratio failed in bond splitting mode along the
longitudinal reinforcement before flexural yielding. With an increase in lateral reinforcement ratio in
Specimen No. 2, the resistance after flexural yielding was maintained to a larger deformation.
Although significant damage occurred near the member ends, Specimen No. 3 with high lateral
reinforcement ratio could develop flexural yielding and stable hysteresis loops, and maintained the
yield resistance to deflection angle R of 5/100 rad. The lateral resistance decayed due to the
compression failure of concrete and buckling of longitudinal reinforcement. Specimen No. 4 failed in
bond splitting mode in the middle part where the amount of lateral reinforcement was reduced to the
level the same as Specimen No. 1.
 pw=0.3% pw=0.6%

pw=0.9% pw=0.6% pw=0.3%

pw=0.3% pw=0.6%

Displ. Displ.

pw=0.9%

Displ. Displ.

Member deflection angle R is defined as lateral deflection divided by the member length. A
reinforced concrete member designed in accordance with Japanese seismic force is known to yield in
flexure approximately at a member deflection angle of 1/200 rad. Shear dominated specimens
exhibited thin and S-shaped hysteresis loops compared with flexure dominated specimens.
Effective Width of T-beam: A girder and slabs, cast monolithically, act integrally as a T-beam; a part
of the slab acts as the flange of the girder under bending, increasing stiffness and flexural resistance
when the girder is subjected to negative bending. Therefore, slabs on either side of the girder should
be considered in evaluating the stiffness and flexural resistance of the girder.

Under positive bending causing, compression stress at the top fiber, the stress at extreme
compressive fiber of the slab decay with distance from the girder face due to shear deformation in the
flange (shear lag) (Park and Paulay, 1975). The slab increases flexural stiffness, but does not
increase the resistance appreciably.

Under negative bending, slab reinforcement parallel to the girder increases the bending resistance
of the girder. Width of slab in which the longitudinal reinforcement is effective to girder flexural
resistance, increases with lateral deformation.

Suzuki et al. (1984) tested half-scale three-dimensional beam-column sub-assemblages under

bi-directional horizontal load reversals (. The beam dimensions were 200 x 300 mm, reinforced by
4-D13 bars at the top and 4-D13 bars at the bottom. The column dimensions were 300 mm square,
reinforced by 8-D13 bars. The
yield stress of D13 bars was 366
MPa, and concrete strength in
beams was 19.5 MPa. The slab
was 70 mm thick and 2,440 mm
wide, reinforced by D6 bars at
200 mm on centers.

The strain in slab longitudinal

reinforcement was shown to
increase with lateral deformation
applied to the specimen, and the
region of yielding slab
reinforcement became wider. The
resistance at the final stage was
close to the resistance calculated
using entire slab tensile
reinforcement. The spread of the Strain in slab reinforcement
effective width of slab affects the Suzuki et al., 1984
stiffness after yielding.

A seven-story full-scale building specimen was tested under pseudo-dynamic loading (Yoshimura
and Kurose, 1985). Strains in slab reinforcement were measured during the test. The sum of slab
stress divided by the yield stress of slab reinforcement gave equivalent number of yielded slab
reinforcing bars. They indicated that the slab reinforcement in a width equal to three to four times the
normally assumed effective slab width (slab width equal to one-tenth of the girder span) yielded at a
story drift angle of 1/75 rad.

References:

Celebi, M., and J. Penzien, "Experimental Investigation into the Seismic Behavior of Critical Region of
Reinforced Concrete Components as Influenced by Moment and Shear," Report EERC No. 73-4,
Earthquake Engineering Research Center, University of California at Berkeley, 1973.
Comite Euro-Internationa du Beton, “RC Frames under Earthquake Loading - State of the Art
Report,” Thomas Telford, 1996, 303 pp.
French, C. W., and A. Boroojerdi, "Contribution of RC Floor Slabs in Resisting Lateral Loads," Journal,
Structural Engineering, ASCE, Vol. 115, No. 1, January 1989, pp. 1-18.
Fujisawa, M., et al., "Study on Ductility of Girders, Pilot Tests (in Japanese)," Report on Development
of High-rise Frame Wall Structures, Building Research Institute, Ministry of Construction, 1988.
Garstka, B., et al., “Damage Assessment in Cyclically Loaded Reinforced Concrete Members,”
Cyclically Loaded Reinforced Concrete Members, Structural Dynamics, Balkema, Rotterdam, Vol.
 1, 1993, pp. 121 - 128.
Hiraishi, H., S. Nakata, Y. Kitagawa and T. Kaminosono, “Static Tests on Shear Walls and
Beam-column Assemblies and Study on Correlation between Shaking Table Tests and
Psuedo-dynamic Tests,” ACI SP-84, Earthquake Effects n Reinforced Concrete Structures, -
U.S.-Japan Research, American Concrete Institute, 1985, pp. 11 - 48.
Koda, S., S. Otani and H. Aoyama, “Reinforcement Details of T-shape beams and Ductility (in
Japanese),” Transactions, Architectural Institute of Japan Annual Convention, October 1987, pp.
209 - 210.
Ma, S. M., V. V. Bertero and E. P. Popov, "Experimental and Analytical Studies of the Hysteretic
Behavior of Reinforced Concrete Members under Bidirectional Reversed Lateral Loading," Report
No. EERC 76-2, University of California at Berkeley, 1976.
Park, R. and T. Paulay, Reinforced Concrete Structures, John Wiley & Sons, Inc., pp. 99-100, 1975.
Suzuki, N., S. Otani and H. Aoyama, "Three-dimensional Beam-column Sub-assemblages under
Bi-directional Earthquake Loading," Proceedings, Eighth World Conference on Earthquake
Engineering, Vol. VI, San Francisco, July 1984, pp. 453-460.
Yoshimura, M., and Y. Kurose, "Inelastic Behavior of the Building," ACI SP-84, Earthquake Effects on
Reinforced Concrete Structure, US-Japan Research, American Concrete Institute, Detroit, 1985,
pp. 163-202.
3.2 Behavior of Columns

A column is an important structural element that supports the weight of a structure and resists
earthquake story shear. The column height is normally determined by the story height of a building.
Section dimensions are controlled by the amount of axial force and shear force used in design.
During an earthquake, exterior columns and corner columns are subjected to varying axial load in
addition to bi-directional lateral load reversals, while interior columns are subjected to almost constant
axial loads.

Yielding is generally permitted at the top of top-story columns and at the base of first-story
columns, and often at the ends of exterior columns subjected to tension force under earthquake
induced overturning moment. The other columns are normally provided with flexural strength
sufficient to prevent flexural yielding using the capacity design consideration. The failure of columns
may lead to the collapse of a building; hence, the brittle failure is carefully avoided in the design of
columns. Brittle failure of a column causes an increase in shear in the other columns of the story, and
might lead to the progressive failure of all columns in the story. Furthermore, the axial force carried by
the failing column must be transferred to the adjacent columns and walls by girders.

The behavior of independent columns (without wing walls nor structurally attached non-structural
elements) is influenced by (a) dimensions of section, (b) amount of longitudinal reinforcement, (c)
amount of lateral reinforcement, (d) level of axial load, (e) concrete strength, and (f) loading (one-way,
cyclic, or reversal). Four modes of failure should be considered when the behavior of reinforced
concrete members is discussed; i.e., (a) flexural failure caused by crushing of concrete or buckling of
compressive reinforcement, (b) shear failure, (c) bond splitting failure along the longitudinal
reinforcement, and (d) anchorage failure of longitudinal reinforcement at the member end.

Columns are normally tested (a) simple support conditions, (b) cantilever form, and (c)
anti-symmetric bending conditions.

Hysteresis Characteristics of Columns under Flexure: A typical force-deflection curve of a

cantilever column is shown with no axial force (Otani et al. 1979). The column section was 305x205
mm, and reinforced with 8-No. 7 bars (yield stress of 438 MPa). The concrete strength was 23 MPa.
The height of loading point from the column base was 1,372 mm.

Note the following observations:

(a) tensile cracking of concrete and
yielding of longitudinal
reinforcement reduced the
stiffness; (b) when a deflection
reversal was repeated at the same
newly attained maximum amplitude
(for example, cycles 3 and 4) the
loading stiffness in the second
cycle was lower than that in the first
cycle, although the resistances at
the peak displacement were almost
identical; and (c) average stiffness
(peak-to-peak) of a complete cycle
decreased with a maximum
displacement amplitude. For
example, the peak-to-peak
stiffness of cycle 5, after large
amplitude displacement reversals,
was significantly reduced from that Column lateral force deflection relationship
of cycle 2 at comparable Otani et al., 1979
displacement amplitudes. Therefore, the hysteretic behaviour of the reinforced concrete is sensitive
to loading history.
 Higashi et a. (1977) showed the influence of loading history on the response of shear failing
beams under one-way loading and reversal loading. The reduction in resistance after yielding is
significant when the specimen fails in shear. It is difficult to generalize the hysteresis relation for this
type of failure.

Monotonicaly loading

Cyclic or reversed loading

Effect of Loading History on Shear Failing Members

Higashi et al., 1977
Columns with Confining Reinforcement: Lateral
reinforcement is known to confine the core concrete and
enhance the compressive strength and deformation capacity
of the confined concrete. The effect of confining reinforcement
is important; a column with closely spaced stirrups and
well-distributed longitudinal reinforcement shows very little
strength decay even subjected to high axial force above the
balance point.

Rabbat et al. (1986) showed the improvement of

deformation capacity with additional lateral reinforcement. The
axial force level was 0.30 times the compressive strength of
concrete area. The lateral reinforcement ratios of the two
Lateral reinforcement ratio = 0.0074
specimens were 0.74% and 1.60%. The column with small
amount of lateral reinforcement failed before flexural yielding,
while the column with large lateral reinforcement maintained
stable hysteresis. The lateral reinforcement can confine the
core concrete and delay the compression failure of concrete
and also increase shear resistance.

Sugano et al. (1985) tested a series of columns of 210 x

210 mm with different amount of lateral reinforcement
subjected to monotonically increasing axial loads. Lateral
reinforcement used were round lateral reinforcement (yield
stress σ wy = 421 MPa) used in Specimen S06 (lateral
reinforcement ratio pw = 0.56 %) and Specimen S12 (pw =
1.12 %), plain φ 4 welded wire mesh lateral reinforcement Effect of lateral reinforcement
( σ wy = 554 MPa) in Specimen M06 (pw = 0.60 %) and (Rabbat et al., 1986)
Specimen M12 (pw = 1.20 %), high-strength lateral reinforcement ( σ wy = 1,397 MPa) in Specimen
U06 (pw = 0.56 %) and Specimen U12 (pw = 1.12 %), and band plate of 25 x 2.3 mm at 50 mm on
centers in Specimen BP12. Longitudinal reinforcement was common among the specimens; 16-D10
bars ( σ wy = 400 MPa) were used in a section with gross reinforcement ratio of 2.58 %. Concrete
strength was 25 to 28 MPa.

Welded Wire Fabric Rectangular Spiral Band Plate

Welding Failure
Welding Failure

Tie Fracture
Load

Tie Fracture
No Reinf.

Strain

Sugano et al., 1985

 Deformation of each column was measured over middle 375 mm length on opposite two faces.

All specimens reached maximum resistance at strain between 0.3 to 0.5 %, and the resistance
deteriorated by fracture of lateral reinforcement. The use of lateral reinforcement is seen to enhance
the strength and deformation capacity of a reinforced concrete column. The following can be
observed in the figure;
(a) For the same type of lateral reinforcement, the axial deformation capacity can be observed to
increase with increasing amount of lateral reinforcement.
(b) High strength lateral reinforcement (Specimens U06 and U12) can develop larger deformation
capacity than normal strength lateral reinforcement (Specimens S06 and S12).
(c) Steel band plate did not perform well compared.
(d) Welded wire fabric appears to be most effective in the series.
(e) Thin but closely placed lateral reinforcement is more effective than large but far spaced lateral
reinforcement.

Another example is shown to demonstrate the effect of lateral (confining) reinforcement on the
deformation capacity of reinforced concrete columns. The two specimens have the same dimensions
and the same amount of longitudinal reinforcement. The amount of lateral reinforcement was varied
from 1.2% (Specimen A2) to 2.0% (Specimen A4) in the two specimens. Specimen A2 could develop
flexural yielding and maintain the resistance to a story drift angle of 1.5/100, while Specimen A4 could
maintain the resistance to a story drift angle of 4/100.
Column subjected to Axial Load and Bending: The yield and ultimate moments under monotonic
loading increase with the compressive axial force below the balanced load level. The stiffness in
virgin loading, unloading and reloading increases with the level of axial force. It is generally known
that a column subjected to compression force above the balanced point cannot develop large plastic
deformation beyond yielding because compressed concrete deteriorates at a faster rate. Axial force
closes cracks at a low lateral force in a column. This phenomenon is different from the behavior of a
beam.

The strength degrades considerably with cycling when the axial force level is near or above the
balance point. Rabbat et al. (1986) tested columns with different levels of axial force and
demonstrates the effect of axial force level on the column deformation capacity. The lateral
reinforcement ratio was 1.74 percent in the specimens. The axial force was 10, 20 and 30 percent of
the compressive strength of the concrete area. The column subjected to low axial force level
developed flexural yielding and exhibited reasonable deformation capacity under lateral load
reversals until failure occurred in compression side. The column subjected to higher axial load, on the
other hand, did not develop flexural yielding.

Effect of axial force levels (N/Ac f’c=0.10, 0.20 and 0.30)

A small scale column was tested under constant tensile axial force and lateral load reversals. The
column was 225 x 225 mm, reinforced with grade SD395 (nominal yield stress of 395 MPa)
longitudinal reinforcement and grade SBPD1300 high-strength lateral reinforcement. The tensile
reinforcement ratio pt was quite large and 4.7%. Lateral reinforcement ratio pw was 0.7%.
Concrete strength was 60.7 MPa. The axial force level was 0.18 times the compressive strength of
concrete area. Note a stable hysteresis behavior to a large lateral deformation under tensile axial
force.

Column subjected to tensile axial force

Column subjected to Varying Axial Load: Exterior columns and corner columns are subjected to
varying axial load and bi-directional lateral load reversals during an actual earthquake, while interior
columns are subjected to almost constant axial loads. The level of varying axial forces is limited by
the formation of yield hinges at the beams in the structure; the upper limit is the sum of shears in the
exterior beams at the formation of yield hinges at the two ends. Variation of axial force is more closely
related to shear acting in the first story column.

Sakaguchi et al. (1985) tested two columns of 450 x 450 mm section subjected to varying axial
load (0.6 Nuc to 0.7 Nut for an exterior column) or constant axial load (0.35 σ B Ae for an interior
column) and lateral load reversals, where σ B : compressive strength of concrete (=43 MPa), Ae:
transformed section area of column, Nuc: compressive strength of column (= 0.85 σ B Ac + Ag σ sy ),
Nut: tensile strength of column (= Ag σ sy ). Gross reinforcement ratio was 2.29 % (12-D22, yield stress
σ sy = 411 MPa). Lateral reinforcement ratio was 1.27 % (exterior and interior spiral reinforcement of
4-D10 at 50 or 60 mm spacing, yield stress σ wy = 390 MPa). Clear height was 1,640 mm. Steel
H-section was embedded at the centroid of section to resist high tensile force. During the loading, the
inflection was maintained at column mid-height.

Both specimens showed good performance, dominated by flexure, up to member rotational angle
(lateral deformation divided by clear member length) of 1/50 rad. However, the resistance started to
decay at a larger deformation in the exterior column due to high level of axial force. The loading
condition of constant axial force is severer to a specimen than that of varying axial load as long as the
axial force amplitudes of the two loading cases are comparable.

Maximum Load Maximum Load

Calculated
Calculated Strength
Strength

Maximum Load

Maximum Load

Li et al. (1986) tested a cantilever column (200 x 200 x 570 mm) under varying axial load and
uni-axial lateral load reversal. Eight D10 bars were used as longitudinal reinforcement (yield stress
σ sy = 426 MPa). Lateral reinforcement ratio was 0.64 % (2-D6@50 mm, yield stress σ wy = 394 MPa).
Concrete strength was 27.6 MPa.

Axial stress was varied by 2 MPa from the static value of 2 MPa proportional to lateral resistance
of the column. As expected, the flexural resistance increased with increased axial load in positive
direction: "interaction of axial force and flexural resistance". The resistance started to decay after
buckling of longitudinal reinforcement in the positive direction.
 Test Specimen (unit: mm) Lateral load-deformation relation
Li et al., 1986

Additional test results were reported by Gibertsen and Moehle (1980), Kreger and Linbeck (1986),
Abrams (1987) and Ristic et al. (1986, 1988).
Columns subjected to Bi-directional Lateral Load Reversals: Columns in a structure are
subjected to constant or varying axial forces and bi-directional shear and bending moment reversals
during an earthquake. The test results under bi-directional lateral load reversals are rather scarce.
The first column test was conducted by Fujii, Aoyama and Umemura (1974).

Otani and Cheung (1981) reported tests of concrete columns subjected to bi-directional loading.
No axial force was applied to specimens. Column section was 305x305 mm, and reinforced by 8-No.
7 deformed bars (yield stress of 438 MPa). Lateral reinforcement was No. 2 square outer ties and No.
2 (yield stress of 236 MPa and 300 MPa) diamond inner ties both spaced at 88 mm. Three pairs of
specimens were tested. Compressive strength was 34 MPa in specimens SP-3 and Sp-4, 23 MPa in
specimens SP-5 and SP-6, and 27 MPa in specimens SP-7 and SP-8.

Specimen SP-4 was loaded NS direction first and then EW direction, separately, and finally in NS
direction. The influence in loading in the orthogonal direction can be clearly observed in the
load-displacement relation in NS direction. Cycles 13 and 22 were loaded to the same displacement
amplitude before and after loading in the orthogonal direction. Significant reduction in resistance was
caused by the loading in the orthogonal direction. Specimens failed after heavy flexural and inclined
cracking, spalling of the concrete, disintegration of the core concrete, mainly due to the propagation
of inclined cracks and grinding along them, followed by buckling of longitudinal reinforcement.

Effect of transverse loading, specimen SP4

Otani, Cheung, 1981

Specimen SP-5 was loaded only in NS direction, while specimen SP-6 was initially loaded in NS
direction and then in EW direction. The total displacement history was made same in the two
specimens. Specimen SP-5 subjected to loading in NS direction failed at smaller deflection.
 Effect of orthogonal loading
Otani and Cheung, 1981

Cantilever column specimens,

constructed under the same specifications,
were tested under constant and varying axial
load and bi-directional lateral load reversals
(Li et al., 1986).

Column section was 200x200 mm, and

the height from the base to the lateral loading
point was 600 mm. A column section was
reinforced with 8-D10 deformed bars (gross
reinforcement ratio of 1.43 %, yield stress of
417 MPa); lateral reinforcement was square
D6 bars spaced at 50 mm (lateral
reinforcement ratio of 0.64 %, yield stress of 386 MPa). Concrete strength at testing was 27 to 58
MPa.

For Forced displacement path under bi-directional loading is shown right; uni-directional
displacement was applied to a specimen at a time to clarify the effect of loading in each direction.
Initial axial force level was 0.07 times bDf’c, where b and D: width and depth of section, and f’c:
compressive strength of concrete.

The level of axial force was maintained constant in Specimen B8-0. An interaction of resistance
was observed in the test; i.e., when the displacement amplitude was kept at peak value in one
direction and displacement was increased in the orthogonal direction, the resistance in the first
direction decayed with displacement in the orthogonal direction: "interaction of bi-directional
resistance."

Lateral load-biaxial horizontal displacement relation under constant axial force

Li et al. 1986

Interaction of bi-directional resistance

Li et al., 1986

Column subjected to Varying Axial Load and Bi-directional Lateral Load Reversals: An exterior
column is subjected to constant axial force and varying axial force depending of the loading direction
in a structure. A corner column is subjected to varying axial load and lateral load reversals in the two
direction.

Specimen B8-1 was subjected to varying axial force proportional to shear acting in EW direction.
Compressive strength of concrete was 31.4 MPa, otherwise the specimen was fabricated under the
same specification as the previous two specimens. The axial stress was varied from the static level of
0 to 0.13 times bDf’c, proportional to lateral resistance in EW direction.
 Exterior column specimen subjected to varying axial force in EW direction and
constant axial force in NS direction
Li et al., 1986

Corner column specimen B8-2 was loaded varying axial force in the two directions. As the axial
force was varied proportional to the lateral resistance in each direction, the axial force was varied
from 0.21 Ag fy to 0.19 bDf’c. The decay in resistance after reaching maximum resistance was faster
under bi-directional lateral load reversals.

Corner column specimen subjected to varying axial stress proportional

to the lateral resistance in each direction
Li et al., 1986

References:

Abrams, D. P., “Influence of Axial Force Variation on Flexural Behavior of Reinforced Concrete
Columns,” Structural Journal, American Concrete Institute, Vol. 84, May-June 1987, pp. 246 - 254.
Abrams, D. P., "Laboratory Definitions of Behavior for Structural Components and Buildings
Systems," ACI-SP127, Earthquake Resistant Concrete Structures - Inelastic Response and
Design, American Concrete Institute, Detroit, 1991, pp. 91-152.
Bousias, S. N., et al., “RC Columns in Cyclic Biaxial Bending and Axial Load,” Proceedings, Tenth
World Conference on Earthquake Engineering, Madrid, 1992, pp. 3041 - 3046.
Bousias, S. N., “Experimental and Analytical Study of Reinforced Concrete Columns in Cyclic Biaxial
Bending with Axial Force, “ Ph. D. Thesis, Department of Civil Engineering, University of Patras,
 Greece, 1993.
Fujii, S., “Study on Reinforced Concrete Columns subjected to Biaxial Flexure (in Japanese),” M. eng.
Thesis, Department of Architecture, University of Tokyo, 1974.
Gibertsen, N. and J. P. Moehle, “Experimental Study of Small Scale R/C Columns subjected to Axial
Load and Shear Reversals,” Structural Research Series No. 481, Department of Civil Engineering,
University of Illinois at Urbana, 1980.
Higashi, Y., M. Ohkubo and M. Ohtsuka, “Influence of Loading Excursions on Restoring Force
Characteristics and Failure Modes of Reinforced Concrete Columns,” Paper No. 11-23, Sixth
World Conference on Earthquake Engineering, New Delhi, January 1977.
Kreger, M., and L. Linbeck, “Behavior of Reinforced Concrete Columns subjected to Lateral and Axial
Loading Reversals,” Proceedings, Third U.S. National Conference on Earthquake Engineering,
Charleston, South Carolina, Vol. “, pp. 1475 - 1486, 1986.
Li, K.-N., et al., "Behavior of Reinforced Concrete Columns subjected to Varying Axial Load and
Bi-directional Horizontal Earthquake Loads (in Japanese)," Proceedings, Eighth Annual
Conference, Japan Concrete Institute, pp. 489-492, 1986.
Low, S., and J. P. Moehle, “Experimental Study of Reinforced Concrete Columns subjected to
Multi-axial Cyclic Loading,” Earthquake Engineering Research Center, University of California at
Berkeley, EERC Report 87-14,1987.
Otani, S, "Nonlinear Dynamic Analysis of Reinforced Concrete Building Structures," Canadian Journal
of Civil Engineering, Vol. 7, National Research Council of Canada, pp. 333-344, 1980.
Otani, S., and V. W.-T. Cheung, “Behavior of Reinforced Concrete Columns under Biaxial Lateral
Load Reversals, - (II) Test without Axial Loads,” Department of Civil Engineering Publication 81-02,
University of Toronto, February 1981, 127 pp.
Rabbat, B., et al., “Seismic Behavior of Light-weight and Normal-weight Concrete Columns,” Journal,
American Concrete Institute, Vol. 83, Jan.-Feb. 1986, pp. 69 - 78..
Ristic, D., et al., “Effect of Variation of Axial Forces to Hysteretic Earthquake Response of Reinforced
Concrete Structures,” Proceedings, Eighth European Conference on Earthquake engineering,
Lisbon, Vol. 4, 7.4, 1986, pp. 49 - 56.
Ristic, D., et al., “Inelastic Stress-strain based Seismic Response Prediction of RC Structures
considering Dynamically Varying Axial Forces,” Proceedings, Ninth World Conference on
Earthquake Engineering, Tokyo-Kyoto, 1988, Vol. VI, pp. 531 - 536.
Saatcioglu, M., "Modeling Hysteretic Force-Deformation Relations for Reinforced Concrete
Elements," ACI SP-127, Earthquake-Resistant Concrete Structures - Inelastic Response and
Design, American Concrete Institute, Detroit, 1991, pp. 153-198.
Saatcioglu, M., and G. Azcebe, “Response of Reinforced Concrete Columns to Simulated Seismic
Loading,” Structural Journal, American Concrete Institute, Vol. 86, Jan.-Feb. 1989, pp. 3 12.
Sakaguchi, N., et al., "Study on Structural Characteristics of High-rise Reinforced Concrete
Residential Buildings (Part 2: Strength Test of Columns) (in Japanese)," Summary Report, Annual
Meeting, Architectural Institute of Japan, Structures 2, pp. 153-154, October 1985.
Sugano, S., et al., "Experimental Study on High-rise Reinforced Concrete Buildings (Part 1: Outline of
Study and Compression Test of Columns) (in Japanese)," Summary Report, Annual Meeting,
Architectural Institute of Japan, Structures 2, pp. 145-146, October 1985.
3.3 Behavior of Interior Beam-column Connections

A beam-column connection (joint), a region common for columns and girders, is often treated as a
rigid panel or a shear panel in a structural analysis; shear deformation may be considered in the
frame analysis. The beam-column connections are classified into (a) interior beam-column
connection, (b) exterior beam column connection and (c) corner beam-column connection, depending
on the location of the column in a three-dimensional frame. The beam-column connection is designed
to sustain maximum resistance of all members connected to the joint.

Interior and exterior beam-column connections

The essential requirements for the satisfactory performance of a beam-column joint in a reinforced
concrete structure can be summed up as follows (Park and Paulay, 1975);
(a) A joint should exhibit a service load performance equal in quality to that of the members it
joints,
(b) The strength of the joint should not govern the strength of the structure, and its behavior should
not impede the development of the full strength of the adjoining member, and
(c) Ease of construction and access for depositing and compacting concrete.

Failure Modes: The beam-column joint is subjected to normal

stresses and high shear stresses at the boundary of adjacent Vc1
Cs2 Ts1
beams and columns in addition to bond stress acting along the
beam and column longitudinal reinforcement. Concrete flexural
compression from beams and columns develops at opposite Cc2
corners in laterally loaded frames, suggesting the formation of a Vjh=Ts1+Ts2-Vc1
diagonal compression strut. Cc2
Ts2

Beam and column reinforcement passing through joints is Cs1

Vc2
subjected to tension at one boundary and compression at the
other; significant amount of forces must be transmitted from
reinforcement to concrete through bond resistance within the Actions in interior beam-column
connection. connection

Parameters to affect the joint performance are (a) amount of joint hoop, (b) column axial force, (c)
input intensity of joint shear, (d) concrete compressive strength, (e) bond demand along beam bars
through joint, and (f) presence of transverse beams.

The beam-column sub-assemblage may fail in three modes; i.e., (a) shear failure in the connection
before flexural yielding at girder ends, (b) shear failure after flexural yielding at girder ends, (c)
flexural yielding at girder ends, and (d) bond failure along the beam longitudinal reinforcement. There
exist significant interaction of shear resistance and bond resistance along the beam longitudinal
reinforcement.

Examples of crack patterns and shear-shear distortion relations in the interior beam-column
connection are shown. In the first specimen (a), large distortion (large diagonal extension but small
diagonal compression) is observed in the beam-column connection with the reduction in shear
resistance; extensive damage was observed in the connection. In the second specimen (b), the
distortion in the beam-column connection was small before beam flexural yielding, but the connection
distortion increased significantly after the beam yielding. When the connection is reinforced with
sufficient lateral reinforcement, the shear distortion of the connection becomes small (the third
specimen). The beam-column joint sometimes fails after developing flexural yielding at beam ends
although the shear input into the connection was controlled by the beam yielding. The shear strength
is not a unique value of the joint, but the resistance deteriorates with the damage within the joint.

Shear stress/Concrete strength

Story drift

Crack pattern -2
Shear distortion angle x10 rad
Shear stress-distortion relation
(a) Shear failure of joint

Shear stress/Concrete strength

Crack Pattern
Shear distortion angle x10-2 rad
Shear stress-distortion relation
(b) Shear failure after beam flexural yielding
Shear stress/Concrete strength

Crack pattern Shear distortion angle x10-2 rad

Shear stress-distortion relation
(c) Beam flexural yielding

Shear Resisting Mechanisms: Two major

shear resisting mechanisms are generally
considered; (a) diagonal compression strut
mechanism and (b) truss mechanism. The
diagonal compression strut mechanism is
formed by normal concrete stresses at the
boundary of the connection. The truss
mechanism is formed by the bond stress
acting along the beam and column
longitudinal reinforcement, the tensile stress
in lateral reinforcement, and compression Diagonal compression strut Truss
struts uniformly distributed in the connection. Shear resisting mechanisms of interior
Therefore, the truss mechanism is strongly beam-column connection
related to the condition of bond resistance
along the beam longitudinal reinforcement especially after flexural yielding at the beam ends.

Paulay et al. (1978) postulated that a diagonal compression strut develops at the initial load stage
before significant cycles of flexural hinging cause residual reinforcement strain and full-depth
cracking so that shear forces are introduced to the joint by normal stresses acting in the compression
zones of the framing members. A minimum amount of joint hoop reinforcement is necessary for
confinement purpose at this stage. Following bar yield penetration and bond deterioration during
repeated inelastic excursion, the shear input is resisted by a self-equilibrating truss mechanism that
consists of a network of small compressive struts in the core concrete and of tensile forces in the
horizontal and vertical joint reinforcement (including the longitudinal reinforcement of the column).

Kitayama et al. (1991) compared the behavior of two beam-column assembly specimens J1 and
C1. The overall dimensions and loading methods of the two specimens are the same. Beams
(200x300 mm) of Specimen J1 were reinforced with 8-D13 deformed bars (cross sectional area
As=1016 mm2, and yield stress fy=402 MPa) and 4-D13 bars at the bottom (As=508 mm2), while those
of Specimen C1 were reinforced with 12-D10 bars (As= 856 mm2 and fy= 323 MPa) at the top and
6-D10 bars (As=428 mm2). The amount of lateral reinforcement in the connection was the same
(lateral reinforcement ratio of 0.38%). Note that the tensile reinforcement ratio was quite high
compared to that commonly used in a frame structure. Both specimens were designed to yield at the
end of beams. Bond stress transfer along the beam reinforcement within the connection was severer
in Specimen J1 by the use of larger diameter and higher strength reinforcement.

Specimen J1 was judged

to fail in shear in the joint at a
story drift angle of 1/23 rad by
crushing and spalling of shell
concrete; X-shaped cracks
gradually opened along the
main diagonal with
deformation. Specimen C1
developed beam yielding with
fine diagonal cracks uniformly
developed in the connection.

Specimen J1 exhibited a
pinching hysteretic shape
especially after a story drift
angle of 1/46 rad, while
Specimen C1 developed a
good spindle-shape
hysteresis. It should be
pointed out that the shear
stress level in Specimen J1 was approximately 1.25 times larger than that in Specimen C1.

The crack patterns in the joint support the modeling of the shear resistance of a beam-column
connection as the diagonal strut and truss mechanisms. Note that Specimen J1 developed shear
cracks initially caused by the truss mechanism, but the shear cracks in the mail diagonal became
dominant at a larger deformation; i.e., the truss mechanism deteriorated with the bond deterioration
along the beam longitudinal reinforcement, and the principal stress concentrated along the main
diagonal strut to cause shear failure. On the other hand, the diagonal strut and truss mechanisms
were maintained in Specimen C1; diagonal compression stresses distributed uniformly in the panel
concrete. The diagonal strut mechanism can exist without any bond stress transfer along the beam
reinforcement within the connection, while the truss mechanism can exist only when a good bond
stress transfer is maintained. The bond along the beam reinforcement inevitably deteriorates
especially after the beam flexural yielding unless the strength and size of the reinforcement are
strictly controlled.

Liande and Jirsa (1982) discussed the shear resistance of interior beam-column joints as the
diagonal compressive strut mechanism. This concept is accepted in ACI-ASCE Committee 352 report
(1985).

Otani (1991) suggests that the truss mechanism can be effective only when the bond resistance
along the longitudinal reinforcement is maintained. After load reversals of flexural hinging, it is difficult
to maintain the bond resistance along the beam reinforcement; hence, the shear must be resisted by
the main diagonal strut. Lateral reinforcement is necessary to confine cracked core concrete.

Lateral Reinforcement: Lateral reinforcement participates in the truss mechanism and confines the
core concrete in the beam-column connection. Noguchi and Kurusu (1987) reported that the strain in
lateral reinforcement was much larger if the bond resistance along the longitudinal reinforcement was
better. The lateral reinforcement may confine the core concrete in the connection, enhancing the
concrete compressive strength and ductility in the diagonal compression strut mechanism.

Otani et al. (1986) reported three half-scale beam-column sub-assemblage tests. Beams were
200x300 mm, and were reinforced with 12-D10 bars (tensile reinforcement ratio pt =1.59%, yield
stress σ y =333 MPa) at the top and 6-D10 bars ( pt =0.79%). Columns were 300x300 mm, and
reinforced with 16-D13 bars ( σ y =439 MPa). The amount of lateral reinforcement was varied in the
three specimens; 2-D6 bars (lateral reinforcement ratio pw =0.27%, σ y =337 MPa) in Specimen C1,
4-D6 ( pw =0.90%) and 4-D10 bars ( pw =2.01%) in Specimen C3. The concrete strength was 26.6
MPa. The column was subjected to a constant axial stress of 20 kgf/cm2.

Specimen C1 provided with less joint lateral

reinforcement developed diagonal shear cracks wider
than specimen C3 at a story drift angle of 1/23 rad.
The shell concrete of the connection swelled out
Story Shear, tonf

slightly at a story drift angle 1/15 rad. On the other

hand, specimen C3 did not widen diagonal shear
cracks after a story drift angle of 1/46 rad. The amount
of joint lateral reinforcement affected the states of
cracking. However, it is important to note that the
difference was small up to a story drift angle less than
1/46 rad.
Story Shear, tonf

The story shear-story drift relations showed fat

spindle-shaped hysteretic shape without a decay in
resistance even at a story drift angle of 1/15 rad. The
resistance at peak deflection amplitudes was almost
the same for the three specimens. The hysteretic loop
area in the second cycles at the same displacement
amplitude was slightly greater for specimen C3 than
for specimen C1. From the fat hysteretic loop, it is 100.0 -50.0 0.0 50.0 100.0
concluded that the bond of beam bars within a Story Drift, mm
connection did not deteriorate much with the number Story shear-story drift relation
of load reversals.

The story shear is compared with diagonal deformation of a joint panel. The diagonal deformation
was measured over a gauge length of 333 and 310 mm for specimens C1 and C3, respectively. Little
shear distortion of a joint panel was observed in two specimens up to a story drift angle of 1/46 rad.
Specimen C1 started to increase the width of shear cracks at a story drift angle of 1/46 rad, while
specimen C3 did not increase the diagonal deformation even at a story drift angle 1/15 rad. Observe
that the amount of joint lateral reinforcement little affected the shear distortion for a deflection range
expected from an intense earthquake motion. Beyond such deflection level, the lateral reinforcement
started to show the difference in confining the joint core concrete.
 Diagonal Deformation of Beam-Column Connection

The strain of joint lateral

reinforcement is studied at peaks of each
positive loading cycle. Strains in
Specimen C1 exceeded the yield strain at
a story drift angle of 1/92 rad when the
beam started to yield. Strains in
Specimen C3 stayed less than 0.1
percent; in other words, the amount of
joint lateral reinforcement might not be
required as much as that in Specimen C3,
but the amount could be reduced to one
half even if the structure is expected to
deform at a story drift angle of 1/23 rad.

Bond and Beam Bar Slip: Additional Strains in Joint Lateral Reinforcement
source of deformation in the connection
is deformation caused by bar slip within the connection. Consider a girder reinforcement passing
through a connection; girder reinforcement is subjected to high tensile stress at one end and
compression stress at
the other end, and the
stresses must be
transferred to concrete
within the connection.
Vc1
However, the width of
Cs2 Ts1
column may not be
sufficient to allow the
stress transfer. This Cc2
situation is much easier Vjh=Ts1+Ts2-Vc1
in the exterior
Cc2
beam-column Ts2
connection.
Cs1
The bond Vc2
deterioration of beam
bars within a joint is said Bond stress
Cs2 Ts1
to be undesirable
because (a) the energy
dissipation at beam ends
is reduced by pinching in Rotation due to bar slip
a hysteresis shape, (b) (Bertero and Popov 1977)
the diagonal
compressive stresses increase with a change in the joint shear transfer mechanism after beam
 Story shear, tf

Strain, %
Column
Story drift, mm
Strain distribution in beam bars
(a) Specimen with Small column width/bar diameter ratio
Story shear, tf

Strain, %

Column
Story drift, mm
Strain distribution in beam brs
(b) Specimen with large column width/bar diameter ratio (low yield stress bars)
Story shear, tonf

Strain,%

Column
Story drift, mm
Strain distribution in beam bars

(c) Specimen with large column width/bar diameter ratio (high yield stress bars)
Effect of column width to bar diameter ratio on bond resistance deterioration
Kitayama et al.
yielding, and (c) the beam deformation increases due to the bar slip within a joint.

Examples of lateral load-deformation relation of beam-column sub-assemblages are shown with

strain distribution of girder reinforcement within the connection.

If the column width to bar diameter ratio is small as shown in case (a), the bond resistance along
the reinforcement within the connection deteriorates at a faster rate. The stress supposedly in
compression in flexure changes its sign to tension, and beam bars slip in the connection. The
hysteresis of the beam-column assembly exhibits the pinching shape even at a low stress stage. If
the column width to bar diameter ratio is large and low strength steel is used for beam bars, bond
stress can be kept low along the beam bars within the connection as shown in case (b) , and fat
spindle-shape hysteresis shape is observed even after beam yielding. If the column width to bar
diameter ratio is large, but the strength of the beam bars is high as shown in case (c), bond stress
increases near the beam yielding, and bond resistance deteriorates within the connection, developing
a pinching hysteresis shape.

It is, therefore, desirable to use low-strength small-diameter bars for the beam longitudinal
reinforcement to develop good performance in the beam-column joint. However, it is not practical
from the construction point of view. The use of large-diameter and high-strength bars is demanded to
lower the construction cost. A reasonable compromise is necessary.
Transverse Beams: Joint shear failure accompanies the increase in the volume of the connection.
To the extent that transverse beams can restrain this volume increase, the transverse beams are
expected to enhance the joint shear resistance.

Kitayama et al. (1991) tested a three-dimensional

interior beam-column joint and compared with the

Concrete Strength
Joint shear stress/
joint strength of a planer joint. Both specimens
have the same dimension and reinforcement. The
three-dimensional specimen was pre-loaded in the
transverse direction causing flexural cracks at the
faces of the column before loading in the principal
direction. The three-dimensional specimen yielded
in flexure and no failure was observed in the
beam-column connection. The planer specimen
failed in shear compression in the connection. The
transverse beams confine the connection and
enhance the strength of the joint even after the
formation of cracks at the joint boundary. Story drift

Stroy drift angle, rad

Slab Effect: Floor slabs, whose concrete is cast
monolithically with beam concrete, have two Effect of transverse beams on joint strength
effects; (a) slab reinforcement contributes to the Kitayama et al., 1989
flexural resistance of the beam, and (b) slabs
confine the interior joint.

Column Axial Force: Park and Paulay (1975) expect

beneficial effects of axial force on the joint performance and
joint shear reinforcement. A steeper diagonal compression
strut may form as a result of an enlarged compression block
across the column section. The horizontal bond force along
the beam bars can develop within the wider diagonal
compression strut.

Kitayama, Otani and Aoyama (1987) studied the

influence of column axial load on the bond stress transfer
along the beam reinforcement in a joint and reported the Effect of axial force on joint behavior
relation of column compressive stresses normalized by the concrete compressive strength and
energy dissipation expressed in the form of equivalent viscous damping ratio heq at a story drift angle
of 1/50 rad for planar interior beam-column joint specimens tested in Japan. Solid circles represent
specimen with beam bar bond index ub/f’c less than 4.5. Test results are scattered widely regardless

Effect of column axial force on energy dissipation (Kitayama, Otani and Aoyama, 1987)
of column axial stress level. Therefore,. it is considered that column axial stress smaller than 0.3 f’c
does not exhibit beneficial effect on the bond resistance along the beam bar within a joint.

Column axial stress level is compared with the maximum joint shear stress normalized by concrete
compressive strength for plane beam-column joint specimens, failed in joint shear. These specimens
were tested in Japan and U.S. Column axial load does not seem to influence the joint shear strength.
High axial compression load in a column, however, accelerates the strength decay in the diagonal
compression failure of the joint core concrete after beam flexural yielding.

Effect of column axial force on joint shear resistance

Kitayama, Otani and Aoyama, 1987

References:

ACI-ASCE Committee 352, “Recommendations for Design of Beam-column Joints in Monolithic

Reinforced Concrete Structures, Journal., American Concrete Institute, Vol. 82, No. 3, May-June
1985, pp. 266 - 283.
Bertero, V. V., and E. P. POPOV, “Seismic Behaviour of Moment-resisting Reinforced Concrete
Frames,” ACI Sp-53, Reinforced Concrete Structures in Seismic Zones, American Concrete
Institute, 1977, pp. 247-292.
Bonacci, J and S. Pantazopoulou, “Parametric Investigation of Joint Mechanics,” Structural Journal,
American Concrete Institute, Vol. 90, No. 1, January-February 1993, pp. 61 - 71.
Cheung, P. C., et al., “Mechanism of Slab Contribution in Beam-column Sub-assemblages,”
ACI-SP123, Design of Beam-column Joints for Seismic Resistance, American Concrete Institute,
Detroit, Michigan, 1991, pp. 259 0 289.
Durrani, A. J., and J. K. Wight, “Behavior of Interior Beam-to-column Connections under Earthquake
Type Loading,” Journal, American Concrete Institute, Vol. 82, No. 3, May-June 1985, pp. 343 -
350.
Ehsani, M. R., and J. K. Wight, “Exterior Reinforced Concrete Beam-to-column Connections
subjected to Earthquake-type Loading,” Journal, American Concrete Institute, Vol. 82, July-August
1985, pp. 492 - 499.
Fujii, S., and S. Morita, “Comparison between Interior and Exterior RC Beam-column joint Behavior,”
ACI-SP123, Design of Beam-column Joints for Seismic Resistance, American Concrete Institute,
Detroit, Michigan, 1991, pp. 145 - 165.
Hanson, N. W., and H. W. Connor, “Seismic Resistance of Reinforced Concrete Beam-column
Joints,” Journal, Structural Division, ASCE, Vol. 93, No. 5, October 1967, pp. 533 - 560.
Joh, O., et al., “Influence of Transverse Joint and Beam Reinforcement and Relocation of Plastic
Hinge Region on Beam-column Joint Stiffness Deterioration,” ACI-SP123, Design of Beam-column
Joints for Seismic Resistance, American Concrete Institute, Detroit, Michigan, 1991, pp. 187 - 223.
Kitayama, K., S. Otani and H. Aoyama, “Development of Design Criteria for RC Interior Beam-column
Joints,” ACI-SP123, Design of Beam-column Joints for Seismic Resistance, American Concrete
Institute, Detroit, Michigan, 1991, pp. 97 - 123.
Kurose, Y., “Recent Studies on Reinforced Concrete Beam Column Joints in Japan,” Report 87-8, Phil
M. Ferguson Structural Engineering Laboratory, Department of Civil Engineering, University of
Texas at Austin, Texas, 1987.
Liande, Z., and J. O. Jirsa, “A Study of Shear Behavior of Reinforced Concrete Beam-column Joints,”
Report No. 82-1, P. M. Ferguson Structural Engineering Laboratory, Department of Civil
Engineering, University of Texas at Austin, February 1982.
Meinheit, D. F., and J. O. Jirsa, “Shear Strength of R. C. Beam-column Connections,” Journal,
Structural Division, ASCE, Vol. 107, No. 11, November 1981, pp. 2227 -2244.
Noguchi, H. and Kurusu, “Experimental Study on Seismic Resistance of Reinforced Concrete
Beam-column Connections (in Japanese),” Report C-II, Annual Meeting, Architectural Institute of
Japan, Kyusyu, October 1997, pp. 627 - 628.
Otani, S., “The Architectural Institute of Japan Proposal of Ultimate Strength Design Requirements for
RC Buildings with emphasis on Beam-column Joints,” ACI-SP123, Design of Beam-column Joints
for Seismic Resistance, American Concrete Institute, Detroit, Michigan, 1991, pp. 125 - 144.
Pantazopoulou, S., and J. Bonacci, “Consideration of Questions about Beam-column Joints,”
Structural Journal, American Concrete Institute, Vol. 89, No. 1, January-February 1992, pp. 27 -
36.
Paulay, T., “Equilibrium Criteria for Reinforced Concrete Beam-column Joints,” Journal, Structural
Engineering Division, ASCE, Vol. 86, No. 6, Nov.-Dec. 1989, pp. 635 - 643.
Paulay, T., and R. Park, “Joints in Reinforced Concrete Frames Designed for Earthquake
Resistance,” Research Report 84-9, Department of Civil Engineering, University of Canterbury,
Christchurch, New Zealand.
Paulay, T., et al., “Reinforced Concrete Beam-column Joints under Seismic Actions,” Journal,
American Concrete Institute, Vol. 75, No. 11. November 1978, pp. 585 - 593.
Uzumeri, S. M., “Strength and Ductility of Cast in Place Beam Column Joints,” ACI-SP53, Reinforced
Concrete Structures in Seismic Zones, American Concrete Institute, Detroit, Michigan, 1977, pp.
293 - 350.
Sugano, S., et al., “Behavior of Beam-column Joints using High-strength Materials,” ACI-SP123,
Design of Beam-column Joints for Seismic Resistance, American Concrete Institute, Detroit,
Michigan, 1991, pp. 359 - 377.
Zhang, S., and J. O. Jirsa, “A Study of shear Behavior of Reinforced concrete Beam-column Joints,”
Report No. 82-1, P. M. Ferguson Structural Engineering Laboratory, Department of Civil
Engineering, University of Texas at Austin, 1982.
Zhu, S., and J. O. Jirsa, “A Study of Bond Deterioration in Reinforced Concrete Beam-column Joints,”
Report No. 83-1, P. M. Ferguson Structural Engineering Laboratory, Department of Civil
Engineering, University of Texas at Austin, 1983.
3.4 Behavior of Exterior Beam-column Connections

The shear acting in an exterior beam column joint is normally smaller than that in an interior beam
column joint because only one beam is connected to the joint. The anchorage detail of beam
longitudinal reinforcement affects the performance of the exterior joint such as (a) horizontal
development length, radius of 90-degree hook and vertical length after the hook.

Park and Paulay (1975) pointed out that diagonal tension and compression stresses are induced
in the exterior beam-column joint panel. The diagonal tension may be high when the ultimate capacity
of the adjoining members is developed, and this can lead to extensive diagonal cracking. The severity
of diagonal tension is influenced by the amount of flexural steel and the magnitude of the axial force
on the column.

The beam top reinforcement is anchored in relatively weak concrete; the surrounding concrete is
subjected to sedimentation. A splitting crack can easily form along these bars at a relative early stage
of loading, and the bond resistance may be lost in the horizontal anchorage portion. Major anchorage
resistance arises from the high bearing stress in the corner of the bent. Therefore, the horizontal
portion of anchorage length may be ignored in design. The condition of the beam bottom
reinforcement is better. The outer column bars are also subjected to high bond stress due to the
change of stress required to resist bending at the top and bottom boundary. of the joint.

The role of lateral reinforcement is important in the exterior beam column joint in resisting diagonal
tension and confining the core concrete. When the transverse shear reinforcement across diagonal
cracks of joints commences to yield, disintegration of the concrete begins because of the repeated
opening and closing of cracks and sliding movement along the diagonal shear cracks. Full shear
expected in the joint must be resisted by the lateral reinforcement. Column width is important effect in
shear resistance of the exterior joint; i.e., the column width is necessary to increase the angle of
diagonal strut.

Actions in exterior beam-column joint

(Park and Paulay, 1975)

The level of axial force is important on the performance of an exterior joint.

References:

Paulay, T., and R. Park, “Joints in Reinforced Concrete Frames Designed for Earthquake
Resistance,” Research Report 84-9, Department of Civil Engineering, University of Canterbury,
Christchurch, New Zealand.
Seckin, M., and S. M. Uzumeri, “Examination of Design Criteria for Beam-column Joints,”
Proceedings, Sixth European Conference on Earthquake Engineering, Dubrovnik, 1978.
3.5 Behavior of Structural Walls

Introduction: A structural wall is a stiff element and attracts large lateral load, especially shear,
during an earthquake. Therefore, a structural wall was often called as a shear wall. The structural wall
provides large lateral load resistance to a structure. The effectiveness of structural walls in
earthquake resistance has been demonstrated in many earthquakes; e.g., the 1985 Mexico
Earthquake, the 1985 Chile Earthquake, the 1992 Erzincan, Turkey, Earthquake. The basic roles of
structural walls in earthquake resistant structures are listed by Paulay (1981);
(a) To provide adequate stiffness to protect the structure against damage particularly in
non-structural components during moderate seismic disturbances.
(b) To provide adequate strength to limit structural damage to superficial level under
code-specified design forces even though some non-structural damage is expected.
(c) To provide adequate structural ductility and capability to dissipate energy under the largest
disturbance expected in the region.
In order to fulfill the roles, Paulay strongly advocates the design of structural walls to develop flexural
yielding at the base.

A structural wall, even of rectangular cross section, has boundary elements, where large amount
of vertical longitudinal reinforcement is placed to resist bending moment. The longitudinal
reinforcement in a boundary element must be confined by properly detailed lateral reinforcement. The
role of the lateral reinforcement is (a) to maintain the effectiveness of core concrete to a large plastic
strain, (b) to support the vertical reinforcement against inelastic buckling, and (c) to enhance shear
carrying capacity of the boundary element.

The existence of structural walls sometimes causes eccentricity of stiffness relative to the mass
center. Proper arrangement of structural walls must be considered at the structural planning stage.

Failure Modes: The failure of structural walls take place in the following mode, (a) flexure, (b)
diagonal tension, (c) sliding shear, (d) hinge sliding, (e) inadequate anchorage or splicing of
reinforcement, (f) buckling of compressive reinforcement. Concrete, being a relatively brittle material
that show s rapid strength degradation in both compression and shear when subjected to repeated
inelastic strains and multidirectional cracking, should not be considered in structural walls as a
significant source of energy dissipation. An obvious source of hysteretic energy dissipation should be
the yielding of the principal flexural reinforcement (Paulay, 1981).

Failure modes of structural walls （Paulay, 1981）

A structural wall in a low-rise building tends to fail in shear, and more research was conducted on
the shear behavior of structural walls. With a development of high-rise reinforced concrete
construction, flexural behavior of a structural wall was studied. Typical crack patterns at failure are
illustrated for a first-story structural wall. Shear crack pattern in (a); shear dominant failure in (b) to
(d); flexural crack pattern in (e); and flexure dominant failure in (f) and (g).
 (a) Shear cracks (e) Flexural crakcks

(b)Shear compression failure (f) Flexural failure

(c)Shear compression failure (g) Sliding shear failure

(d) Sliding shear failure

Failure modes of structural walls

Although it is desirable for a structural wall to behave in dominantly the flexure mode, it is difficult
to expect such behavior from a squat wall with height to width ration less than 2.0. A significant
portion of horizontal shear is transmitted directly to the foundation by diagonal compression.

In a shear dominated failure, diagonal cracks are spaced regularly at an angle of approximately 45
degrees. When the diagonal shear cracks develop in a structural wall, the entire shear must be
resisted by the horizontal shear reinforcement. If the amount of shear reinforcement is not sufficient,
a diagonal tension failure plane may develop, and the widened diagonal shear cracks lead to
diagonal tension failure. When the average shear stress in the wall section is large and when
adequate horizontal shear reinforcement has been provided, failure may be cause by crushing of
concrete under diagonal compression. The diagonal compression failure is observed when the
flexural capacity is large or when the diagonal shear cracks develop under load reversals. Diagonal
compression failure sometimes develops suddenly with a rapid propagation of the crushing zone of
the concrete. A few cycles of reversed loading sometimes cause significant yielding of flexural
reinforcement; horizontal cracks may penetrate full depth of the wall base section. Sliding
displacement can develop along the flexural cracks. The process of this sliding shear mechanism is
described by Paulay (1981). In the first cycle, involving large flexural yielding, the major part of the
shear force at the base of a cantilever squat wall is transmitted across the flexural compression zone.
This compression zone is small due to light axial load in the low-rise wall. Upon load reversal, cracks
develop across the previous flexural compression zone, while bars are subjected to compressive
stresses. Until the base moment reaches a level sufficient to yield these bars, in compression, a
continuous wide crack develop along the base. The shear must be resisted by dowel action of the
vertical reinforcement. Because of the relatively flexible nature of this mechanism, large sliding
displacement takes place until yielding of the compression steel occurs, closing the crack at the
compression end. With subsequent inelastic load reversals, further deterioration of the shear friction
mechanism along the planes of potential sliding develops. Due to a deterioration of the bond transfer
along the vertical bars and due to the Bauschinger effect, the stiffness of the dowel shear mechanism
also reduces. Eventually the principal mode of shear transfer along the base is by kinking of the
vertical bars.

Shear failure modes of squat wall Sliding shear failure of squat wall
Paulay (1981)

In a flexure dominated failure, initially horizontal flexural cracks change their direction toward
flexural compression zone at the base of the structural wall, forming a fan-shape crack pattern. Note
that compression force acts in the direction of cracks. Compression struts are formed uniformly at an
angle of 45 degrees in a shear dominated wall, whereas compression struts are formed toward
flexural compression zone. Failure may result either by yielding of vertical web reinforcement or by
crushing of concrete.
Deformation: Lateral deformation of a member is often divided into flexural and shear deformation,
similar to elastic problem. However, the two deformation cannot be clearly separated after cracking;
e.g., a diagonal crack developed by shear stress causes rotational deformation.

Flexural deformation can be evaluated by integrating rotational deformation along a member.

Therefore, shear deformation is often defined as the difference of the total deformation and flexural
deformation.

Testing of a structural wall in a laboratory is difficult because the resistance is large. Most
laboratory test data have been obtained for single-story single-span walls to study shear
characteristics.

Shear Stress-deformation Relation: A typical average shear stress-shear deformation relation

shows stiffness reduction at the formation of large inclined crack along the main diagonal; and further
stiffness reduction with opening of the main diagonal crack and formation of additional inclined cracks.
The structural wall fails when the main diagonal crack penetrates through the boundary columns,
leading either shear compression failure in the compression zone of the wall panel and the boundary
column or sliding failure along a horizontal plane of the wall panel accompanied by shear failure of the
boundary columns. The average shear stress may be calculated for the area defined by
center-to-center distance of the boundary columns and thickness of the wall.

Shear Cracking: A diagonal shear crack may be assumed to develop when the principal tensile
stress of section reaches the tensile strength of concrete in the elastic analysis.

Shear stress reported in a structural wall test does not agree with the result from the elastic theory
by the following reason;
(a) the stress-strain relationship of concrete is not linearly elastic before the formation of a crack,
(b) shrinkage stress exists in a reinforced concrete member before the test,
(c) flexural crack may form prior to shear crack, and
(d) a crack is normally reported after it becomes visible.

Sugano (1970) developed an

empirical formula to estimate average
shear stress at shear cracking for wall
test data using shear-type loading:
τ cr = ( 4.3 p g + 0.051)σ B
where τ cr : average shear stress at
shear cracking, σ B : compressive
strength of concrete, pg: gross vertical
reinforcement ratio of boundary columns.
The coefficient of correlation is 0.73.

Axial stress is known to increase the

resistance against cracking; The shear
Column gross reinforcement ratio, %
cracking stress τ cr * may be modified to
include the effect of axial stress acts Shear cracking stress and column reinforcement ratio
(Sugano, 1970)
σ o in a wall section;
τ cr * = (τ cr 2 + σ oτ cr )1 / 2

Deformation at shear cracking may be calculated on the basis of the elastic theory using shear
modulus Gc
Ec
Gc =
2(1 + µ )
where Ec: elastic (Young's) modulus of concrete, µ : Poisson's ratio (approximately 1/6 for concrete).
Secant modulus is normally used for the elastic modulus of concrete.

Shear deformation (angle) γ is given as

κτ c
γ=
Gc
where κ : shape factor for shear deformation. The shape factor κ may be obtained by equating the
external work done by external shear V and the internal work done by shear stress obtained by the
elastic theory. The shear angle is defined as lateral deformation at wall centerline divided by the wall
height.

Ultimate Shear Strength:

Cracking dilates a wall panel, Shear stress observed at shear failure, kgf/cm2
but boundary girders above and
below the panel and boundary
columns on both side give
confinement to the panel.
Therefore, the shear strength
increases with the amount of
longitudinal reinforcement in
the girders and boundary
columns.

Sugano (1970) plotted the

average shear stress τ u
observed at failure in cantilever
wall tests and calculated at
ultimate flexural strength.
Ultimate flexural strength was
Calculated shear stress at flexural resistance, kgf/cm2
calculation at an extreme
compressive fiber strain of
0.004 assuming plane section Average shear stress observed at shear failure
to remain plane after and calculated flexural resistance (Sugano, 1970)
deformation. Different symbols
were used for the amount of wall reinforcement ratio pw (%). A strong correlation can be observed in
the figure; flexural yielding of boundary members reduce the confining effect on the panel and
triggers shear failure in the boundary columns. A lightly reinforced shear panel tends to fail at
approximately 80 percent of average shear stress τ mu calculated at flexural strength.

Sugano (1970) also plotted the relation between the average shear stress observed at shear
failure and the product of lateral reinforcement ratio of a wall and yield stress of wall reinforcement.
The ultimate shear stress increases with the amount of lateral reinforcement, but the rate of increase
in shear resistance is smaller than the amount of shear reinforcement (pw σ wy ).

Hysteresis Behavior: The hysteresis shape of reinforced concrete structural walls, which behavior is
dominated by flexure, can be stable and fat (Paulay and Spurr, 1977).

Reinforced concrete members can be relatively ductile in shear during monotonic loading, but they
generally do not develop suitable performance under inelastic load reversals. The hysteresis shape
exhibits pinching phenomenon. The shear resistance can be attained only when the subsequent
largest displacement is attained and the stiffness decreases with displacement amplitudes. Note that
the envelope curve of the hysteresis curves follows closely the load-displacement curve under
monotonically increasing displacement. Inelastic tensile strains in lateral reinforcement cannot
recovered at complete unloading and accumulates with reversed loading. Special reinforcement to
suppress inelastic shear distortion is desired such as diagonal shear reinforcement.

Flexure dominated wall (Paulay and Spurr, 1977) Shear dominated wall (Oesterle et al, 1976)

Three-dimensional Effect of Wall: Flexural

behavior of a structural wall is similar to a girder or
a column; i. e., flexural crack and flexural yielding
take place before failure in compression zone. Due
to the flexural cracking, the neutral axis of a
section shifts well to compression side, causing a
significant elongation in the boundary column in
tension, while deformation in the boundary column
in compression is small. Consequently, a girder
end connected to the tensile boundary column is
lifted upward, and girders in the plane of the wall
and orthogonal to the wall resist the upward
movement by introducing downward force to the
boundary column. Due to this additional vertical
force in the tensile boundary column, the yielding
of longitudinal reinforcement in the tensile
boundary column is delayed. This behavior was
first reported in the test of a full-scale seven-story
reinforced concrete building (Yoshimura and
Kurose, 1985).

Base Uplifting Rotation at Wall Base: A

structural wall attracts large lateral load due to its
high lateral stiffness. A large overturning moment
at the base of a structural wall sometimes causes
tensile stress at its footing and uplifts the tensile Three-dimensional effect of structural walls
end. The rocking at the base of structural walls is Charney, 1991
known to be beneficial in dissipating kinetic energy of the structure without much damage (Priestley et
al., 1978).

Three fifth-scale two-story three-bay reinforced concrete frame with a structural wall in the central
bay were tested. The base of the wall was supported on footing foundation. Variables in the test were
failure modes (flexural yielding of shear failure) of connecting girders and support conditions (flexible
support or rigid support) at the wall base. Vertical loads were assumed to act on columns and lateral
loads were concentrated at the second floor level. The wall was designed to uplift at the tension side
under the seismic forces prior to the formation of yield mechanism of the system.

Vertical loads were applied at the top of columns and wall simulating the gravity loads. Horizontal
reversal force was applied at the top of the wall simulating seismic actions.
 The lateral load-displacement relation of the three specimens is compared. The three specimens
developed “yielding behavior” by the formation of a yield mechanism of the system. A steep stiffness
upon unloading and small residual displacement at complete unloading was observed. After the
unloading was completed, the high stiffness was recovered during the reloading. Small energy was
dissipated by the specimen with shear failing girders; energy dissipation of the system was controlled
by the frame. The uplifting of the wall commenced at a story drift as small as 1/400. Shear input to the
wall was limited by the uplifting and the damage in the wall was relatively small compared with a wall
with the fixed base.

References:

Bertero, V.V., E. P. Popov, T. Y. Wang, and J. Vallenas, “Seismic Design Implication of Hysteretic
Behavior of Reinforced Concrete Structural Walls,” Proceedings, Sixth World Conference on
 Earthquake Engineering, New Delhi, 1977, Reprints 5, pp. 159 - 165.
Cardenas, A. E., and D. D. Mugura, “Strength of High Rise Shear Walls - Rectangular Cross
Sections,” ACI SP-36, Response of Multistory Concrete Structures to Lateral Forces, American
Concrete Institute, 1973, pp. 119 - 150.
Charney, F. A., "Correlation of the Analytical and Experimental Inelastic Response of a 1/5-Scale
Seven-Story Reinforced Concrete Frame-Wall Structure," ACI SP-127, Earthquake-Resistant
Concrete Structures, Inelastic Response and Design, American Concrete Institute, Detroit, 1991.
Kato, D., S. Otani, H. Katsumata and H. Aoyama, "Effect of Wall Base Rotation Behavior on
Reinforced Concrete Frame-Wall Building," Proceedings, Third Pacific Regional Conference on
Earthquake Engineering, Victoria University of Wellington, New Zealand, May 1983.
Oesterle, R. G., A. E. Fiorato, L. S. Johal, J. E. Carpenter, H. G. Russell, and W. G Corley,
“Earthquake Resistant Structural Walls, - Tests of Isolated Walls,” Report to National Science
Foundation, Portland Cement Association, Illinois, November 1976, 44 pp.
Paulay, T., “The Design of Reinforced Concrete Ductile Shear Walls for Earthquake Resistance,”
Research Report 81-1, Department of Civil Engineering, University of Canterbury, New Zealand,
February 1981, 72 pp.
Priestley, M. J. N., R. J. Evison, and A. J. Carr, “Seismic Response of Structures Free to Rock on their
Foundation.” Bulletin, New Zealand National Society for Earthquake Engineering, Vol. 11, No. 3,
September 1978, pp. 141 - 150.
Sugano, S, "Experimental Study on Restoring Force Characteristics of Reinforced Concrete
Members (in Japanese)," Thesis submitted to fulfill the requirements of Doctor of Philosophy,
University of Tokyo, March 1970.
Synge, A. T., T. Paulay, and M. J. N. Priestley, “Ductility of Squat Shear Walls,” Research Report 80-8,
Department of Civil Engineering, University of Canterbury, 1980, 150 pp.
Yoshimura, M. and Y. Kurose, "Inelastic Behavior of the Building," ACI SP-84, Earthquake Effects on
Reinforced Concrete Structures, U.S-Japan Research, American Concrete Institute, Detroit, 1985,
pp. 163-202.
 Home Assignment No. 2

2001-10-24 Otani, S.

Discuss the following topics on typical behavior of reinforced concrete members:

(1) Load-deflection relation of a beam with different amount of longitudinal reinforcement at the top
and at the bottom, typically seen in a T-shaped beam
Reference: Ma, S. M., V. V. Bertero and E. P. Popov, "Experimental and Analytical Studies of the
Hysteretic Behavior of Reinforced Concrete Members under Bidirectional Reversed Lateral Loading,"
Report No. EERC 76-2, University of California at Berkeley, 1976.

(2) Effect of slab on the stiffness and flexural resistance of a beam.

Reference: French, C. W., and A. Boroojerdi, "Contribution of RC Floor Slabs in Resisting Lateral
Loads," Journal, Structural Engineering, ASCE, Vol. 115, No. 1, January 1989, pp. 1-18.

(3) Role of compressive reinforcement on flexural resistance and deformation capacity.

 Chapter 4 Analysis of Reinforced Concrete Members
The minimum modeling element of nonlinear analysis of a structure can be selected as (a)
material level modeling (Finite Element Method), (b) member level modeling (frame modeling), (c)
story level modeling (mass-and-spring modeling), and (d) one-degree-of-freedom model for an entire
structure. The finite element method (FEM) is used for material level modeling. The story level
modeling by a mass-and-stick model was extensively used to understand the overall behavior of a
structure. Recent capacity and demand spectra method idealizes a structure by a
single-degree-of-freedom system.

The FEM has been developed for a half century, and is popularly used in the analysis of
reinforced concrete member and sub-assemblage even in a nonlinear range to failure. The nonlinear
FEM generally requires (a) modeling of constitutive relations for plain or reinforced concrete and
reinforcement under generalized multi-axial loading and loading reversals, (b) modeling of bar slip
with respect to surrounding concrete, (c) modeling of crack initiation and propagation, and (d)
modeling of stress transfer at crack interface. The method can represent the detailed geometry of the
structure, the history of stresses and strains at every point. The constitutive modeling of materials
developed significantly in the past decades. The FEM, however, is not practical for use in the
nonlinear earthquake response analysis of an entire building structure due to the limitation in
computational and memory requirements.

The member level modeling may not represent the detailed geometry of reinforcement in a
section, but it can reflect the basic features of member characteristics. The stiffness analysis of a
member is normally based on flexural response, and ignores the shear and bond stiffness
degradation because the two modes of response are prevented in design process. The method can
describe the damage distribution and intensity within a structure. A fiber model of section can be
combined with the member level modeling to represent the geometry of section to a limited extent.

A close-coupled shear-beam model or far-coupled bending beam model is used to represent

stiffness at each story in the mass-and-stick model analysis. This type of modeling is sometimes
useful to estimate the story drift of a structural system from different earthquake motions.

Therefore, the nonlinear earthquake response analysis of a building is normally carried out on the
basis of member-by-member modeling. This chapter reviews the analysis of reinforced concrete
section and members.

1
4.1 Flexural Analysis of Section

The moment-curvature relation of a cross section provides the basis of understanding the
nonlinear behavior of reinforced concrete members, such as cracking, crushing and spalling of the
concrete, and yielding and buckling of longitudinal reinforcement.

Assumptions: Reinforced concrete section under bending is normally analyzed using the following
simplifying assumptions;
(a) Plane section before bending remains plane and normal to the member axis after bending,
(b) Normal stress-strain relation (constitutive model) of materials is known, and
(c) External forces are equal to internal forces of the section.

The first assumption, often called the Bernoulli's hypothesis, simplifies the analysis and gives
linear distribution of longitudinal strain across the section with null strain at the neutral axis. The
location of the neutral axis of the section is determined by the equilibrium of axial force acting in the
section.

Bernoulli’s Assumption: Eivind Hognestad (1951)

reported the results of strain measurement in a
reinforced concrete section near failure region of tied
columns (254 mm square) and spiral columns (305
mm round) under the combined axial load and
bending moment. Strains in the reinforcement and on
the concrete surface were measured by SR-4 wire
strain gages.

He noted some deviation of observed strains from

the linearity due to inaccuracy in individual strain
measurements and to small errors in the location of
gage lines. He concluded that a reasonable
agreement existed between strains measured in
reinforcement and on the concrete surface. He noted E. Hognestad (1951)
that the departures from linearity appeared to be
inconsistent, indicating accidental or local rather
than systematic variations. Therefore, he εo + φy z
assumed that “Bernoulli’s hypothesis is valid.”

The neutral axis of a section is defined as a φy

level where no extension nor compressive
Strain

deformation takes place. It should be noted that εo

the neutral axis moves to the compression side
with loading especially after initial cracking. The
slope of linear strain distribution is equal to the
curvature at the section, which is the second
derivative of deflection. y
ε ( y, z )
Local bond stresses between the
reinforcement and concrete and local bar slips ε o − φz y
relative to the surrounding concrete exist near z
tension cracks. Therefore, the Bernoulli’s
assumption is not valid, in the exact sense, near a
crack, shifting the neutral axis away from the
geometrical centroid. The curvature also varies εo φ
along a member with the varying depth of the z
neutral axis. However, it is believed that the
assumption holds on the average over a finite
Section Strain
region. Slip between concrete and reinforcing
steel may be ignored if deformed bars are used as Coordinate system and strain in section

2
the longitudinal reinforcement.

Let us take the right-hand coordinate system with x-axis in the direction of a horizontal member,
and y-axis in the vertical direction and z-axis in the horizontal direction.

Strain: For the small deformation, the normal strain ε ( y, z ) at coordinate ( y, z ) in the section can
be expressed as
ε ( y, z ) = ε 0 − y φ z + z φ y
where, ε 0 : normal strain at the geometrical centroid of section, φ y , φ z : curvature about the
centroidal axes y and z .

Stress: Using the stress-strain relations of constitutive

materials, normal stress σ x ( y, z ) can be determined

Stress
for a given normal strain ε x ( y, z ) at coordinate
( y, z ) from the origin. σ = f (ε )
σ x ( y, z ) = f (ε x ( y, z ))

Equilibrium of Forces: Knowing stresses within a

section, resultant axial force and bending moments
about the geometric centroid are calculated by Strain
integrating the stress and moments of stress over the Stress-strain relation of material
section:
N ( x) = ∫
Cross Sectional Area
σ x ( y, z ) dA

M z ( x) = ∫
Cross Sectional Area
−σ x ( y, z ) y dA

M y ( x) = ∫
Cross Sectional Area
σ x ( y, z ) z dA

 M y ( x) 
 
S R ( x) =  M z ( x) 
 N ( x) 
 

It is important to note that all structural members are represented by line elements at the
geometric centroid section in the frame analysis; i.e., moment of a member in the structural analysis
and bending moment resistance of the section must be calculated about the common axis.

Moment-Curvature Relation under Monotonically Increasing Load: Uni-axial moment-curvature

relationship of section under a constant axial load N is calculated for a given curvature φ z rather
than for a given moment M z . Calculation of moment for a given curvature requires an iterative
procedure because the strain at the center of the section is not known.

The following procedure is normally taken to evaluate moment M z of a section under existing
axial load N and a given curvature φz ;
(a) Assume strain ε cent at the center of the section; which enables us to define strain profile
across the section;
3
 ε ( y ) = ε cent − φ y
in which ε ( y ) : strain at distance y from the section center.
(b) Determine stress profile across the section on the basis of stress-strain relation of materials;
σ ( y ) = σ (ε ( y ))
(c) Resultant axial force N cal is estimated by integrating normal stress over the section;

N cal = ∫ σ ( y ) dA
Cross Sectional Area
(d) This resultant axial force must be equal to the existing axial load of the section if the choice of
the assumed strain ε o at the center is correct. Trial and error method with interpolation is used to
iterate steps (a) to (c) until the resultant axial load becomes practically equal to the existing axial load.
(e) Bending moment M z is calculated for the correctly assumed strain at the center of the
section;
Mz = ∫
Cross Sectional Area
−σ ( y ) y dA

Stress-strain Relation of Concrete: Various stress-strain relations have been used in the flexural
analysis of reinforced concrete section.

Kent and Park (1971), for example, used the following expression for concrete stress-strain
relation under monotonically increasing compression load;
εc εc 2
σ c = σ o [2 −( ) ] for ε c ≤ ε o
εo εo σc
σ c = σ o [1 − Z (ε c − ε o )] for ε c > ε o σo Unconfined

but σ c ≥ 0.2σ o
Confined
where σ c : concrete stress, σ o : compressive
strength (MPa) of concrete, ε c : concrete strain, 0.5σ o
ε o : strain at compressive strength of concrete ε 50h
(=0.002). The slope of descending part was 0.2σ o
controlled by a parameter Z in the following εc
manner;
ε o = 0.002 ε 50u ε 50c ε 20
Stress-strain relation for concrete
under monotonic loading
Park and Kent Model, 1971

4
 0.5
Z=
ε 50u + ε 50 h − ε o
0.021 + 0.002σ o
ε 50u =
σ o + 6.89
3 b"
ε 50 h = ( ) p s
4 sh
where p s : ratio of volume of transverse reinforcement to volume of concrete core measured to
outside of hoops, b": width of confined core measured to outside of hoops, s h : spacing of hoops.
Stiffness of descending branch was varied taking into account the confining effect of concrete by
lateral reinforcement.

Tensile stress of the concrete may be considered before the first cracking in the section, but the
tensile stress may be ignored after the initial cracking.

Stress-strain Relation of Reinforcing Steel:

Stress-strain relation of reinforcing steel is normally
represented by elastic-perfectly plastic relation
with/without strain hardening branch. Young’s
modulus Es of steel is approximately 205 GPa.
σ s = Es ε s ε s ≤ ε sy
σ s = σ sy ε sy ≤ ε s ≤ ε sh
σ s = σ sy + Esh (ε s − ε sh ) ε sh ≤ ε s
where, Es : elastic modulus of steel, ε y , ε sh : strains
of reinforcement at yielding at the initiation of strain
hardening. The stress-strain relation is normally
assumed to be the same in tension and
compression.

Numerical Procedure: The bending moment-curvature relation is analytically obtained by a

numerical method. It is not practical to calculate the curvature for a given bending moment. Bending
moment is normally calculated for a given curvature.

For a given curvature and axial force in the section, the strain at the geometrical centroid of the
section is assumed. With the curvature and the strain at the geometrical centroid, the strain
distribution is uniquely defined across the depth of the section. Stress at each point in the section is
determined from the stress-strain relations of materials for the strain at the point. Normal stresses are
summed up to calculate an axial force corresponding to the assumed strain at the geometrical
centroid. If the calculated axial force is not equal to the existing (given) axial force, the assumed strain
at the geometric centroid is not correct. An iterative procedure is used until the assumed strain gives
the resultant axial force equal to the existing axial force in the section. It is convenient to plot the
resultant axial force in the vertical axis and the assumed strain in the horizontal axis, and use linear
interpolation or extrapolation to estimate the strain in the next iteration.

Depth to the neutral axis may be chosen as a variable instead of the strain a the geometrical
centroid.

Moment-curvature Relation: Moment- curvature relationship of a reinforced concrete section

changes its stiffness at (a) tensile cracking of concrete, (b) tensile yielding of longitudinal
reinforcement.
5
 Before tensile cracking of concrete, the
neutral axis of the section without axial load Ultimate
Yielding
lies at the geometric centroid of the section,

Bending Moment
and area of concrete in tension cannot be
neglected for bending resistance.

After initial cracking, the neutral axis

shift to compression side, and the
contribution of uncracked concrete in
tension to bending resistance becomes Cracking
negligible. For a reinforced concrete section
satisfying design requirements, the EcIe Curvature
concrete in compression can be assumed
to act linearly elastic even after cracking, Moment-curvature relation
and cracked transformed section may be
used after cracking.

Significant stiffness change occurs at the tensile yielding of longitudinal reinforcement, and very
small increase in resistance takes place after flexural yielding.

In design, the concrete compressive strain at the ultimate stage is normally used as 0.003 to 0.004.
However, this ultimate strain is not intended for use in evaluating ultimate deformation (curvature,
rotation nor deflection), but is used to estimate the ultimate moment, which is not affected by the
amplitude of ultimate strain. The concrete strain at the flexural failure is believed to be much larger
especially when the concrete is properly confined by lateral reinforcement.

Resisting moments at cracking and yielding

are significantly influenced by existing axial
force in the section. This is called the
"interaction of axial force and bending moment".
Compressive force delays the tensile cracking
Compression
of concrete and tensile yielding of longitudinal
Axial Force

Balance
Failure
reinforcement. Point

The balance point is defined as a point in the

axial force-moment diagram where the Yielding Tension
compression failure of concrete develops at the
Failure
same time as the tensile yielding of longitudinal
bars in the section. A compression failure
Bending Moment
occurs if the axial force is larger than the axial
force at the balance point; no tensile
reinforcement develops before failure. Interaction of bending moment and axial force
Therefore, there is no ductility. A tension failure
occurs if the axial force is lower than the axial force at the balance point, where the concrete fails in
compression after the tensile yielding in the longitudinal reinforcement. The presence of axial force
significantly reduce the ductility of the section.

References:

Blume, J. A., N. M. Newmark and L. H. Corning, “Design of Multi-story Reinforced Concrete Buildings
for Earthquake Motions, Portland Cement Association, Chicago, 1961, 318 pp.
Hognestad, E., "A Study of Combined Bending and Axial Load in Reinforced Concrete Members,"
Bulletin No. 399, Engineering Experimental Station, University of Illinois, 1951.
Kent, D. C., and R. Park, "Flexural Members with Confined Concrete," Journal, Structural Division,
ASCE, Vol. 97, ST7. pp. 1969-1990, July 1971.
Park, R., and T. Paulay, Reinforced Concrete Structures, Wiley-Inter-Science Publication, 1975.

6
Pfrang, E. O., C. P. Siess and M. A. Sozen, “Load-Moment-Curvature Characteristics of Reinforced
Concrete Cross Sections,“ Journal, American concrete Institute, Vol. 61, No. 7, July 1964, pp. 763
- 778.

7
 Example of Flexural Analysis

Standard Dimensions and Material Properties

- Dimensions
Width b=400 mm
Depth D=800 mm
Effective depth d= 750 mm
dc=50 mm (concrete cover depth to center of longitudinal reinforcement)
- Reinforcement
Tensile reinforcement area As=1435 mm2 (5-Ｄ19)
Compressive reinforcement area A’s=861 mm2 (3-Ｄ19)
- Material Properties
Yield stress of longitudinal reinforcement σ y =345 N/mm2 (SD345)
ε y =0.00167
Concrete strength σ B =20.6 N/mm2 (C20)

Case studies:
- Case 1: Tensile reinforcement area As was reduced to 2871 mm2 (3-D35).
- Case 2: Compressive reinforcement was totally removed (A’s =0.0 mm2).
- Case 3: Yield stress of steel was reduced to σ y =235 N/mm2 (SD235).
ε y =0.00114
- Case 4: Concrete strength was increased to 31 N/mm2 (C31).
- Case 5: Width was reduced to 200 mm.
- Case 6: Effective depth was reduced to 400 mm.
- Case 7: Concrete cover depth was increased to 100 mm.

Yield Moment and Sectional Parameters

The flexural yield point is defined at the tensile yielding of longitudinal reinforcement. The
stress-strain relation of concrete is assumed to be linear with concrete elastic modulus (secant
stiffness at one-third compressive strength). It should be examined if the concrete stress σ c max at
the extreme compressive fiber should be less than 70 percent of the compressive strength σ B . If the
stress exceeds the limit value, more realistic stress-strain curve should be used for the concrete. It
was also examined if the strain in compressive reinforcement exceeds the yield strain. Modular ratio
(Modular ratio =Es/Ec) n=15 was assumed in the analysis.

Standard Case 1 Case 2 Case 3 Case 4 Case 5 Case 6 Case 7

kd , mm 217 268 235 217 217 274 147 213
φ y , x10-6/mm 3.13 3.46 3.24 2.15 3.13 3.51 6.58 3.11

M y , kN-m 335 625 331 230 335 330 172 326

I cr , x105 mm4 7.81 13.2 7.44 7.81 7.81 6.85 1.92 7.66
σ c max / σ B 0.45 0.62 0.51 0.31 0.30 0.64 0.64 0.44
ε sc / ε y 0.31 0.45 --- 0.36 0.31 0.47 0.38 0.21

kd : Depth of the neutral axis from the extreme compressive fiber;

φ y : curvature at yielding;
M y : yield moment;
I cr : moment of inertia of cracked transformed section.

8
Ultimate moment and section parameters
The ultimate moment of sections was calculated using equivalent rectangular stress block when
the strain at the extreme compressive fiber reached 0.003. The depth and amplitude of the
rectangular stress block were 0.85 times the depth c to the neutral axis and 0.85 times the
compressive strength of concrete σ B . The resultant of compressive stress was assumed to act at
the center of the rectangular stress block.

Standard Case 1 Case 2 Case 3 Case 4 Case 5 Case 6 Case 7

c , mm 63.6 116 82.7 52.9 53.2 88.2 63.6 91.3
φu , x10-6/mm 47.2 25.9 36.2 56.7 56.4 34.0 47.2 32.9
M u , kN-m 353 690 352 249 358 348 180 353
ε st / ε y 19.4 9.85 14.0 34.5 23.5 13.5 9.53 13.0

ε sc / ε y 0.39 1.02 ---- 0.14 0.11 0.78 0.39 -0.2

c : Depth of the neutral axis from the extreme compressive fiber;
φu : curvature at yielding;
M u : yield moment.

9
 4.2 Moment-curvature Relation under Reversed Loading

Analytical procedure is identical to the flexural analysis of section under monotonically increasing
moment. Three basic assumptions are used; i.e., (a) Bernoulli's hypothesis, (b) normal stress-strain
relations of the concrete and reinforcing steel, and (c) equilibrium of internal and external forces.
Stress-strain hysteresis relation must be prepared for the concrete and reinforcing steel under stress
reversals. Iterative procedure is normally used to determine the strain at a pre-selected point of the
section for a given curvature until the equilibrium of axial forces is reached. The integration of normal
stresses over the section is not simple because the stress distribution cannot be defined by a simple
function. Therefore, the lamina or fiber model is generally used in the analysis.

The first effort to calculate the moment-curvature relation for reinforced concrete beams under
reversed loading was made by H. Aoyama (1964), followed by Agrawal et al. (1965), Bertero and
Bresler (1969), and Brown and Jirsa (1971). Aoyama (1964) assumed elasto-plastic stress-strain
relation for both steel and concrete in the analysis of reinforced concrete section under constant axial
force and reversal bending. He reported that the level of axial force and the plastic deformation in the
previous loading made drastic change in the moment-curvature relations.

Lamina Model Analysis: A reinforced section is sliced to horizontal pieces parallel to the neutral axis.
For each layer of concrete and reinforcement, stress and strain are represented by the values at the
mid-height of the layer and stress is assumed to be uniform within each layer element.

Section Strain Stress Resultants

Lamina model

Park, Kent and Sampson (1972) analyzed a rectangular girder section under moment reversals.
Stress-strain relations of concrete and reinforcing steel are modeled from the observed behavior in
the laboratory. The amount of top and bottom reinforcement is quite different in the section.

Loading part of stress-strain relation of reinforcing bars is represented by modified

Ramberg-Osgood Model (1943):
r −1
σs σ
ε s − ε si = (1 + s
Es σ ch
0.744 0.71
σ ch = σ sy { − + 0.2411}
ln(1 + 1000ε ip ) 1 − e1000ε ip
4.49 6.03
r= − n + 0.297 for odd n
ln(1 + n ) ( e − 1)
2.20 0.469
r= − n + 3.04 for even n
ln(1 + n ) ( e − 1)
where ε s : steel strain, ε si : steel strain at zero stress, σ s : steel stress, Es: elastic modulus of steel,
10
σ ch : stress dependent on the yield strength and plastic strain in the steel produced in the previous
loading run, and γ : parameter of the Ramberg-Osgood Model, ε ip : plastic strain in steel produced
in previous loading run, n: number of post yield loading runs with n = 0 for the first yielding. Unloading
stiffness was taken equal to the initial elastic stiffness.

The stress-strain curve for concrete is represented by a parabola for ascending portion and
straight line for descending portion for monotonically increasing strain (Kent and Park, 1971). A linear
stress-strain curve for concrete in tension may be assumed to the tensile strength. The curve under
cyclic loading is represented by straight lines. Upon unloading from point E on the skeleton curve,
0.75 of the previous stress is lost without decrease in strain, whereupon a linear path of slope 0.25 Ec
is followed to point G. If the concrete has not cracked, it is capable of carrying tensile stress to point
K; but if the concrete has previously cracked, or if cracks form during this loading stage, the tensile
strains increase but no tensile stress develops. Upon reloading, the strain must regain the value at G
before compressive stress can be sustained again. If reloading commences before unloading
produces zero compressive stress, reloading follows one of the paths IJ. The average slope of the
assumed loop between E and G is parallel to the initial tangent modulus.

The stress-strain curve for the cover concrete in compression may be assumed to follow the curve
for the confined core concrete at strains less than 0.004. The cover concrete at strains greater than
0.004 may be considered to have spalled and to have zero strength.

An iterative technique may be used to calculate the depth of the neutral axis at each loading
stage.

Doubly reinforced concrete beams had a rectangular section of 125 x 203 mm, and simply support
span of 1830 mm. Strains were measured on the top and bottom reinforcement over a 51 mm gauge
length in the critical region and the curvature was calculated from the strain measurement.

The comparison of the observed and the calculated was reasonably good. Note that the observed
curvature history was given to the analysis model and that the resistance after yielding is limited by
yield moment. Therefore, the behavior prior to yielding needs be carefully examined. The calculated
hysteresis loops before flexural yielding were fat compared with the observed. General behavior
during crack opening and closing was simulated well. When the cracks were open in the theoretical
curves, the moment is carried by a steel couple alone.

11
 The filament model was used in the analysis of section and members under bi-directional
bending; i.e., Monegotto and Pinto (1973), Aktan et al. (1974), and Zeris and Mahin (1988).

References:

Agrawal, G. L., L. G. Tulin and K. H. Gerstle, “Response of Doubly Reinforced Concrete Beams to
Cyclic Loading,” Journal, American concrete Institute, Vol. 62, No. 7, July 1965, pp. 823 - 836.
Aktan, A. E., et al., “R/C Column Earthquake Response in Two Dimensions,” Journal, Structural
Division, ASCE, Vol. 100, No. ST10, October 1974, pp. 1999 - 2015.
Aoyama, H., “Moment-Curvature Characteristics of Reinforced Concrete Members subjected to Axial
Load and Reversal of Bending,” Proceedings, International Symposium on the Flexural
Mechanics of Reinforced Concrete, ASCE-ACI, Miami, November 1964, pp. 183 - 212.
Bertero, V. V., and B. Bresler, “Seismic Behavior of Reinforced Concrete Framed Structures,”
Proceedings, Fourth World Conference on Earthquake Engineering, Vol. 1, Session B-2, Chile,
1969, pp. 109 - 124.
Brown, R. H., and J. O. Jirsa, “Reinforced Concrete Beams under Reversed Loading,” Journal,
American Concrete Institute, Vol. 68, No. 5, May 1971, pp. 380 - 390.
CEB, “RC Elements under Cyclic Loading,” Bulletin d’Information 210, Thomas Telford, London,
1991.
Fujii, S., H. Aoyama and H. Umemura, “Moment-curvature Relation Calculated on the Basis of
Material Properties (in Japanese),” Report (Structural Engineering), Annual Meeting, Architectural
Institute of Japan, October 1973, pp. 1261 - 1262.
Kent, D. C., and R. Park, "Flexural Members with Confined Concrete," Journal, Structural Division,
ASCE, Vol. 97, ST7. pp. 1969-1990, July 1971.
Menegotto, M., and P. E. Pinto, “Method of Analysis for Cyclically Loaded RC Plane Frames including
Changes in Geometry and Non-elastic Behavior of Elements under Combined Normal force and
Bending,” Preliminary Report, IABSE, Vol. 13, 1973, pp. 15 - 22.
Park, R., D. C. Kent and R. A. Sampson, "Reinforced Concrete Members with Cyclic Loading,"
Journal, Structural Division, ASCE, Vol. 98, ST 7, pp. 1341-1360, July 1972.
Ramberg, W. and W. R. Osgood, "Description of Stress-Strain Curves by Three Parameters,"
Technical Note No. 902, National Advisory Committee for Aeronautics, July 1943.
Zeris, C., and S. A. Mahin, “Analysis of Reinforced Concrete Beam-columns under uniaxial
excitation,” Journal, Structural Division, ASCE, Vol. 114, No. ST4, April 1988, pp. 804 - 820.

12
4.3 Flexural Analysis of Members

In the analysis of prismatic reinforced concrete members, it is convenient to work with the member
axis and with cross sections normal to the member axis. The resultant axial and shear forces and
bending and torsional moments of the normal and shear stresses acting on the section are
considered. The interaction between shear and bending deformations as well as between the shear
and bending moment is important especially under nonlinear loading reversals. However, the
interaction is generally neglected because the shear dominated behavior is generally prevented in
the design process.

The local behavior of a member is described by the moment-relative rotation relations, in which
the relative rotation θ AB of two neighboring cross-sections A and B is the integral of curvature
between the two sections,
xB
θ AB = ∫ φ ( x) dx
xA

where φ ( x) : curvature at x. The relative rotation is defined as the angle between their planes of
cross section or the corresponding tangents to the member axis.

The chord rotation is the angle between the

chord connecting the member axis and the tangent
to the member axis; for member AB, chord rotations θB B
θ A and θ B are A
θA
xB x −x
θ A = ∫ φ ( x) B dx θ AB
xA xB − x A
xB xA − x
θ B = ∫ φ ( x) dx
xA xB − x A
The relative rotation θ AB , therefore, is expressed as
θ AB = θ A − θ B

The deflection of point A from the tangent to the member axis at point B due to curvature between
the two points is calculated by
xB
∆ AB = ∫ x φ ( x) dx
xA

where x : distance from point A.

The following effects should be considered in evaluating member deformation; (a) the increase in
stiffness of members due to tension carried by the concrete between the cracks (tension stiffening) ,
(b) deformation caused by diagonal tension cracks, and (c) deformation caused by bond slip of the
reinforcement.

When diagonal tension cracks are present in a member, the tension in the flexural reinforcement
at sections away from the section of maximum moment may be larger than that computed from the
bending moment diagram. The internal tension remains nearly constant at the maximum value over a
distance nearly equal to the effective depth from the critical section. The region of a yield hinge zone
is wider than the bending moment diagram implies.

Bond slip of reinforcement in anchorage zones increases the deformation.

13
Deformation of Simply Supported Member:
Member end displacement expressed for a simply
supported member can be transformed to θA
displacements at member ends in the local d x1 θB
coordinate system. Consider member end R d y2
displacements of member 1-2 in a plane; i.e.,
member end displacements in the member axis d y1 2 dx2
direction, orthogonal to the member axis and 1
rotation. The member end displacements of a simply d − d y1
supported member are given by member end R = y2
rotations θ A , θ B and axial deformation e These l
displacements are related in the following; Deformation of Member in Local Coordinates
e = − d x1 + d x 2
θ A = θ z1 + ( d y 1 − d y 2 ) / L A B
θA
θ B = θ z 2 + ( d y1 − d y 2 ) / L θB

e
in a matrix form; Deformation of simple Beam
   
 e  − 1 0 0   d x 1  1 0 0  d x 2 
   1    1  
θ A  =  0 1  d y 1  + 0 − 0 d y 2 
θ   L  θ   L  θ 
 B  1  z1 
0 − 1  z2 
0 0 0
 L   L 

It is often convenient to study the deformation relation of simply supported beams under mid-point
loading and anti-symmetric loading. The loading cases will cause triangular bending moment
distribution in one-half member, but maximum moment at member end under anti-symmetric bending
and maximum moment at the mid-span in mid-point loading. The chord rotation θ A of the
anti-symmetric loading beam is defined as the angle between the tangent at the end and the straight
line connecting two member ends, which is the ratio of deflection δ at the mid-span measured from
the tangent line divided by one-half member length l / 2 . The same angle can be obtained in the
mid-span loading beam by dividing the deflection at mid-span by one-half member length.
mA P

mA
mB = mA

θA
θB = θ A δ
θ A = RAB RAB = δ /(l / 2)
Relation between chord rotation at beam ends of anti-symmetric bending and
member rotation of simply supported beam

14
Simply Supported Beams:
Consider a simply supported straight
member i subjected to member end
moments mA and mB at the two ends
A and B. No intermediate loads act m( x )
mA
within the member. Moment mB
distribution within the member can
be determined from the member end
forces mA and mB; curvature φ (x)
at point x along the member can be
determined for the existing bending φ ( x)
moment m(x) at the point on the
basis of moment-curvature relation.

Member-end rotations θ A and θ B

θB
may be calculated by using the A B
dummy (unit) load method; bending
θA
moment muj(x) due to unit moment mj
= 1.0 (j= A or B) applied at j-end, x
then member-end rotation θ j is Member-end Moment and Member-end Rotation
calculated as
L
θ j = ∫ φ ( x) muj ( x) dx
0
In other words,
L mA m( x)
x mB
θ A = ∫ φ ( x) (1 − )dx
0
L
L
x
θ B = ∫ φ ( x)(− )dx φ ( x)
0
L x

Note that a member end rotation is

not solely dependent on the moment at mA = 1.0
the end.

The elongation e of the member is

calculated by integrating axial strain at
the geometrical centroid of the section.
No elongation occurs under bending at
initial elastic stage. However, a mB = 1.0
significant elongation occurs after initial Calculation of Member-end Rotation
cracking even at the centroid due to
bending because the neutral axis of a section shifts to compressive side.
L
e = ∫ ε ( x)dx
0

The member-end moment-rotation relation is calculated for anti-symmetric moment distribution

having the inflection point at the mid-span; rotations at the two ends are the same. The member-end
moment-rotation relation under monotonically increasing load may be approximated by a tri-linear
relation with stiffness changes at initial cracking and at tensile yielding at the member ends.

At an elastic stage with elastic modulus Ec and equivalent moment of inertia I e of transformed
section is given as

15
 6Ec I e
mA = θA
L Yielding
in which L: member length. Ultimate

Moment
α yS
mA Cracking

6EI
S=
l
mB = mA
Rotation, θA
Anti-symmetric moment distribution
Member-end moment-rotation relation

Cracking Moment: Cracking moment may be estimated as

ND
mcr = c σ t Z e +
6
where c σ t : tensile strength of concrete, Z e : section modulus of transformed section, N : axial load
(positive in compression), D : overall depth of section. Sugano (1970) evaluated the tensile strength
of concrete in test specimens from observed cracking moments and section modulus of concrete. A
wide scatter can be observed in data, partially attributable to shrinkage strain, partially to late noticing
the cracking in a specimen during the test.

c σt ( kgf / cm 2 )

σB (kgf / cm2 )
Concrete tensile strength evaluated from test results (Sugano, 1970)

Member end rotation at flexural yielding and ultimate stages can be calculated from the curvature
distribution along the member;
my
mc
mc mu
my
φu mu
φy mc
mc my φu
my
φy φc φc
φc φc
φy
First Yieldng Ultimate Stage
φy
φu
Moment and curvature distribution at critical stages
16
Yield Deformation: The member-end rotation at flexural yielding calculated from curvature
distribution is known to underestimate the yield rotation observed in the test because the analysis
does not consider (a) shear deformation, (b) deformation caused by bar slip within anchorage zone,
and because the Bernoulli's hypothesis holds only in an approximate sense. Therefore, an empirical
expression α y was formulated for the ratio of secant stiffness at yielding to the initial stiffness
(Sugano, 1970):
a N d
α y = {0.043 + 1.64n pt + 0.043 + 0.33 }( ) 2
D bDσ B D
where n : modular ratio of steel to
concrete, p t : tensile reinforcement
ratio calculated for overall cross
sectional area, a / D : shear
span-to-depth ratio, N : Axial force,
b D : cross sectional area of section,
σB : compressive strength of
concrete, d: effective depth of section,
D : overall depth of section.

The expression is applicable for

the following range of parameters:
p t = 0.4 to 2.8 %, a / D = 2.0 to 5.0,
N / b D σ B = 0.0 to 0.55.
Beams
Columns
Observed

Although the stiffness degradation

ratio has been used in the nonlinear
earthquake analysis of buildings
under design in Japan, the yield
D
stiffness is determined with respect Calculated α y ( )2
to the initial stiffness, which cannot d
be determined with accuracy; e.g., Calculated and Observed Stiffness Reduction Factor at Yielding
note the reliability of an expression Sugano, 1970
for the elastic modulus Ec of
concrete. my

Another simplifying method to estimate a

member end rotation at yielding of
anti-symmetrically loaded member is to my
assume a triangular distribution of curvature φy
with a calculated yield curvature at a
member end and null curvature at the
inflection point. This sometimes gives a
reasonable value. This method sometimes
underestimate the deformation for a short
deep beam.

φy
Ultimate Deformation: The deformation at
the ultimate stage is one of the most
important deformation indeces of a member. Simplified method to evaluate member end rotation
However, the method of estimating this Triangular curvature distribution
deformation has drawn little attention in the
past. The plastic rotation in a yield hinge region at the ultimate stage may be estimated by assuming
17
the plastic curvature ( φu − φ y ) distributes uniformly over
φu
hinge width l p ;
θ p = (φu − φ y ) l p
φy
Baker (1964) purposes an expression for plastic hinge
length l p as
z lp
l p = 0.8 k1 k2   c
d 
where k1 =0.7 for mild steel and 0.9 for cold-worked steel,
k2 =0.6 for f c ' =35 MPa and 0.9 for f c ' =12 MPa, z :
distance of critical section to the inflection point, d :
Plastic hinge length
effective depth of section, c : neutral axis depth at the
ultimate moment with ultimate strain of concrete given in the following form;
d
ε cu = 0.0015{1 + 150 ρ s + (0.7 − 10 ρ s ) } ≤ 0.01
c
where ρ s : ratio of the volume of the transverse confining reinforcement to the volume of the core
concrete. Concrete stress-strain relation is a parabola for ascending region and a constant value to
ultimate strain. The maximum concrete stress f c " is given by
d
f c " = (0.8 + 0.1 ) f c ' ≤ f c '
c
The test results show a significant scatter, but Baker claims that the expression furnishes a
reasonable safe prediction.

Corley (1964) proposed the following expressions for the equivalent plastic hinge length l p and
the maximum concrete strain ε cu at the ultimate stage;
z
l p = 0.5d + 0.2 d  
d 
2
b ρ f 
ε cu = 0.003 + 0.02 +  s y 
z  20 
where z : distance of critical section to the inflection point, b : width of beam in inches (=25.4 mm),
d : effective depth of section, ρ s : ratio of the volume of the transverse confining reinforcement and
compressive longitudinal bars to the volume of the core concrete, f y : yield strength of confining
steel in kips (=6.89 MPa)

References:

Baker, A. L., and A. M. N. Amarakone, “Inelastic Hyperstatic Frame Analysis,” Proceedings,

International Symposium on the Flexural Mechanics of Reinforced Concrete, ASCE-ACI, Miami,
November 1964, pp. 85 - 142.
Corley, G. W., “Rotational Capacity of Reinforced Concrete Beams,” Journal, Structural Division,
ASCE, Vol. 92, No. ST5, October 1964, pp. 121 - 146,
Sugano, S, "Experimental Study on Restoring Force Characteristics of Reinforced Concrete
Members (in Japanese)," Thesis submitted to fulfill the requirements of Doctor of Philosophy,
University of Tokyo, March 1970.

18
4.4 Load-Deformation Relation of Beams

Data Base of RC Beams: Test data of beams, yielding in flexure before failure, were searched from
literature (Ref. 1-22) between 1982 and 1992 in Proceedings of Japan Concrete Institute (JCI),
Summaries of Technical Papers of Annual Meeting of Architectural Institute of Japan (AIJ), Annual
Reports of the New RC Project, and reports of research institutes of Japanese construction
companies. The specimens must satisfy the following conditions; (a) rectangular cross section, (b)
same amount of top and bottom longitudinal reinforcement, (c) width wider than 150 mm, and (d)
overall depth deeper than 225 mm. Among 105 beam specimens obtained, the concrete strength
ranged from 26 to 98 MPa; the shear span-to-depth ratio from 1.0 to 3.6; the tensile reinforcement
ratio from 0.44 to 2.70 percent; the yield strength of longitudinal reinforcement from 261 to 976 MPa.

The specimens were tested under two types of loading methods (Fig. 1) simulating the conditions
during earthquake excitation; i.e., (a) Type-A test (31 specimens): statically indeterminate beams with
two stiff end stubs subjected to lateral displacement at the two ends maintaining the end stubs in
parallel during loading, and (b) Type B test (74 specimens): simply supported beams, normally with
two loading stubs, subjected to two-point loading causing the point of inflection at the center of the
middle span. In Type A test, moment distribution is not known due to static indeterminacy; linear
moment distribution is assumed with an inflection point at the mid-span. In Type B test, the damage
within the test span tends to concentrate at an end during the test. Simply supported specimens
subjected to mid-span loading were not selected for the study.

(a) Type A test (b) Type B test

Fig. 1 - Loading methods in laboratory
The load-deformation relation curves under reversed cyclic loading were obtained from the generous
researchers, and were digitized at the University of Tokyo for the sequence of parts where the load
exceeded the maximum load of the previous loading cycles. The error of digitization is less than 0.3
percent, on the average 0.2 percent, of the full scale.

Moment-Rotation Relation: The

member-end moment was obtained
from shear force measured in the test
span by assuming the inflection point
at the mid-span. The member-end
rotation of Type A specimens was
calculated from the measured relative
lateral displacement divided by the
clear span, but the member end
rotation of Type B specimens was
used as the researchers reported.
The observed moment-rotation curve
was idealized into a trilinear
relationship (Fig. 2).

Although many specimens failed in

shear or bond-splitting modes after
flexural yielding, the initial stiffness Fig. 2 - Idealization of observed moment-rotation relation
19
may not be affected by the failure modes. Therefore, all 105 specimens were used in the study of
initial stiffness and flexural cracking moment. On the other hand, the yield deflection is increased by
the damage associated with the failure modes. Therefore, those specimens failing in shear or
bond-splitting modes within a deflection equal to three times flexural yielding deflection were
excluded from the examination of yield and ultimate points; 38 specimens were used to study the
yield deflection.

The initial stiffness in a test was defined as a secant slope at a load equal to one-half of the reported
cracking load. The cracking load was not reported in five specimens; hence the cracking moment was
evaluated by assuming the tensile strength of concrete to be 0. 56 σ B (Ref. 23). Note that the
cracking load is normally reported at a loading step when cracking is detected for the first time; i.e.,
the reported cracking load is normally higher than the actual cracking load. Therefore, the cracking
point was determined from the shape of force-deformation relation by the method described in the
following paragraph.

The stiffness of reinforced concrete section changes drastically at the yielding of tensile
reinforcement. If tensile reinforcement is placed in double layers in the section, the stiffness changes
at the yielding of the outer layer reinforcement and then of the inner layer reinforcement. In order to
select a single yield point, the yield point and cracking moment were defined such that the energy
stored at the ultimate deformation should be the same for the test and the model making the absolute
difference in the energy to be minimum (Fig. 2).

Resistance at the ultimate point was taken as the observed maximum resistance. The determination
of an ultimate deformation is an important but impossible issue; the ultimate deformation is not a
unique value but is highly dependent on the progress of concrete deterioration dictated by loading
history and failure modes. Small stiffness after yielding will not change appreciably by the choice of
an ultimate deformation. Therefore, the ultimate deformation was selected to be an arbitrary
deformation at a deformation ductility factor of four.

An iterative procedure was used to define the yield point and cracking moment for the established
initial stiffness and ultimate point.; i.e., (a) a trial yield displacement was assumed, (b) an ultimate
displacement was selected at four times yield deformation, (c) post-yield stiffness was determined by
connecting the ultimate point and a point on the observed curve at 2.5 times yield deformation, and
(d) the cracking moment and yield deformation were determined for equal absorbed energy at the
ultimate deformation and minimum absolute difference.

Initial Elastic Stiffness: The methods to evaluate stiffness parameters and their reliability with
respect to the observed values are discussed. The initial elastic stiffness K E was evaluated by the
elastic theory of a lineal member considering flexural and shear deformation;

1 1 1
= + (Eq. 1)
KE KB KS
where K B : flexural stiffness (= 6 Ec Ie / L), K S : shear stiffness (= Gc A L/ 2κ), Ec: elastic modulus of
concrete, Ie: moment of inertia of uncracked transformed section, L: member length, Gc: shear
modulus of concrete, A: cross sectional area, κ :shape factor for shear deformation (=1.2). The
shear modulus of concrete was estimated from the elastic modulus Ec and assumed Poisson's ratio
of 0.20. Elastic modulus Es of steel was assumed to be 206 GPa.

The initial stiffness was calculated using the observed elastic modulus of concrete and the clear span.
The observed initial stiffness (Fig. 3) was notably low and, on the average, 0.53 times that calculated
with a large coefficient of variation (=0.51) for 73 specimens with reported concrete elastic moduli.
The large coefficient of variation and discrepancy between the test and calculation was probably
attributed to (a) technical difficulty in measuring accurate initial stiffness in the test and (b) formation
of accidental and shrinkage cracks prior to the test. In a real structure, flexural cracks under gravity
20
loading, shrinkage cracks, cracks after
medium intensity earthquake excitation
may exist, and the initial stiffness for the
analysis is difficult to estimate. However,
the initial stiffness and cracking force level
of a single-degree-of-freedom system do
not influence the maximum response
amplitude as long as the attained
response ductility factor reached more
than 4 (Ref. 24). The response in this
range is more sensitive to the secant
stiffness and the resistance at yielding
rather than the initial stiffness.

Furthermore, the actual elastic modulus in

a structure, as built, is not known at the
time of structural design although the initial Fig. 3 - Reliability of calculated initial stiffness
stiffness of a member is directly dependent
on the modulus. The elastic modulus of concrete is normally difficult to control in construction.
Therefore, the initial stiffness of a structure can be significantly different from the value assumed by a
structural engineer.

An empirical expression was proposed for the elastic modulus Ec (Ref. 25), taking into account
compressive strength and density of concrete, type of coarse aggregates and mineral admixture;

Ec = k1 × k2 × 3.35 × 104 × ( σ B / 60 )1/ 3 × ( γ / 2. 4 )2 (MPa) (Eq. 2)

in which k1: factor representing type

of coarse aggregates, k2: factor
representing kind of mineral
admixture, σ B : observed concrete
strength (MPa), γ : unit density of
concrete (ton/m3). The factor k1 is
0.95 for crushed quartzite, crushed
andesite, basalt and clayslate
aggregates, 1.0 for other coarse
aggregates, and 1.2 for crushed
limestone and calcined bauxite
aggregates. Factor k2 is 0.95 for silica
fume; fine powder of blast furnace
slag and fly ash fume, 1.00 for
concrete without mineral admixture or
with other mineral admixture, 1.10 for
fly ashes. Ninety-five percent of test Fig. 4 - Elastic modulus and strength of concrete (Noguchi)
data are shown to fall within 20
percent of the empirical expression (Fig. 4). The modulus should be controlled in construction within
an acceptable range from the value specified by a structural engineer.

Cracking Moment: Cracking moment Mcr is calculated on the basis of the observed splitting tensile
strength σ t of concrete and the section modulus Ze of the uncracked transformed section. The ratio
of the reported to the calculated cracking moments is compared in Fig. 5 for 59 specimens; the
average ratio was unexpectedly good at 1.03, with a significantly large coefficient of variation of 0.50.

Cracking tensile strength σ cr of concrete was determined by dividing the cracking moment of the
21
trilinear idealization of the moment-rotation relation by the section modulus Ze of uncracked
transformed section. The cracking tensile strength and compressive strength of concrete are
compared in Fig. 6 for 68 specimens. A wide scatter of data can be observed, but a tendency is
observed for the cracking tensile strength to increase with the compressive strength. Following
empirical relation was derived:

σ cr = 1. 26 × σ B 0.45 (Eq. 3)

The cracking tensile strength should not be used as the tensile strength of concrete, but is intended to
evaluate a moment level at which the initial elastic stiffness of the trilinear idealization changes in the
member-end moment-rotation relationship.

Fig. 5 - Reported to calculated cracking moment Fig. 6 - Cracking tensile strength and
compressive strength of concrete

Yield Moment and Rotation: Yield moment at the critical section was calculated for the yielding at an
imaginary centroid of tensile reinforcement. The amount of tensile reinforcement is normally limited
well below the balanced tensile reinforcement ratio; hence, the stress-strain relation of the concrete
was assumed to remain linearly elastic when the tensile reinforcement first yielded under bending. In
addition, the following assumptions were made in calculating yield moment; i.e., (a) plane section
remained plane after deformation, and (b) the concrete in tension did not resist tensile stresses. The
elastic modulus of concrete was determined by Eq. 2 with k1=k2=1.0 and γ =2.4 ton/m3, while the
compressive strength of concrete and the yield stress of reinforcement were obtained from the
reported material tests. In 2 specimens out of
38, the calculated stress at the extreme
compressive fiber exceeded the compressive
strength of concrete; these specimens were
removed from the study.

The calculated yield moment and the estimated

yield moment of the trilinear idealization are
compared in Fig. 7 for 36 specimens. The
average ratio of the estimated to the calculated
yield moment is 1.12 with a coefficient of
variation of 0.078; only one estimated yield
moment was smaller (0.94) than the calculated
value. The yield moment at which the stiffness
of an RC member changes drastically may be
calculated conservatively by the flexural Fig. 7 - Calculated and observed yield moment
analysis.

22
Member end rotation at flexural yielding was calculated using cracked transformed section for flexural
deformation ( θ f = M y L / 6 Ec I cr ), elastic stiffness for shear deformation ( θ s = 2κ ⋅ M y / GAL ) and
pullout deformation θ slip of longitudinal reinforcement from the anchorage. The pullout deformation
δ slip was calculated by the following expression (Ref. 26):

δ slip = ε y ( 2 + 3, 500ε y ) ⋅ d b / ( σ B / 20 )2/ 3 (Eq. 4)

where, ε y : yield strain of longitudinal reinforcement, db: diameter of longitudinal reinforcement, σ B :

concrete strength (MPa). The center of rotation at the critical section was assumed to be at the
centroid of compressive reinforcement; i.e., θ slip = δ slip / ( d − d c ) , where d: effective depth and d c :
depth to the centroid of compressive reinforcement.

The calculated yield rotation and the estimated yield rotation of the trilinear idealization are compared
in Fig. 8.a for 36 specimens. The estimated rotation was, on the average, 2.54 times larger than the
calculated rotation with a coefficient of variation of 0.22. Calculated yield rotation significantly
underestimates the estimated rotation. This discrepancy is attributable to the additional rotation
caused by shear cracking and the error in evaluating the pullout deformation of longitudinal
reinforcement from the anchorage.

An empirical expression was derived for the yield deformation by assuming the yield rotation at a
member end consists of the rotations from flexural deformation, θ f , shear deformation, θ s and
deformation, θ slip due to the pull-out of longitudinal reinforcement from the anchorage zone. A
regression analysis with respect to the estimated yield rotation was carried out to determine
coefficients;

θ y = 115
. θ f + 12. 6θ s + 3.89θ slip (Eq. 5)

The calculated yield deflection and the observed yield rotation of the trilinear idealization are
compared in Fig. 8.b for 36 specimens. A coefficient of variation of the ratio was 0.20 with a mean of
1.0.

(a) Calculated yield rotation (b) Empirical expression

Fig. 8 - Calculated and observed yield rotation

Ultimate Moment: Flexural strength of beam section is not sensitive to the shape of stress-strain
relationship nor the compressive strength of concrete because the neutral axis depth is so small that
the distance between the resultant compressive and tensile forces cannot change appreciably within
the section. The ultimate moment was calculated using the plasticity theory suggested by Eberhard
23
and Sozen (Ref. 27), in which the flexural mechanism was assumed to form by the yielding of tensile
reinforcement followed by the compressive failure of concrete. Instead of strain compatibility,
equilibrium conditions for axial force and bending moment of section were used based on the lower
bound theorem. The maximum bending resistance was sought by satisfying the yield criteria of the
materials and is given by

M u = k ⋅ σ B ⋅ b ⋅ d c + at ⋅ σ y ( d − d c )
2
(Eq. 6)

in which,σ B : compressive strength of concrete, b: width of section, dc: distance from the extreme
compressive fiber to the centroid of compressive reinforcement, at: area of tensile reinforcement, σ y :
yield stress of tensile reinforcement, d:
effective depth of section. Values of k are
(3/8) and (1/2) for triangular and rectangular
stress distribution of compressive concrete
with maximum stress of σ B , respectively.
The stress in the compressive reinforcement
must be checked not to exceed the yielding
stress.

The ratio of the observed to the calculated

ultimate moments is compared with respect
to compressive strength of concrete in Fig. 9
for 38 specimens; the average ratio is 1.15
with a coefficient of variation of 0.074 for the
triangular concrete stress block, and the
average of 1.13 with a coefficient of variation
of 0.069 for the rectangular stress block. The Fig. 9 - Observed and calculated ultimate moment
shape of a stress block shape should be (Triangular stress block)
carefully selected in the evaluation of
ultimate moment of a column.

References:
1. Korenaga, T., T. Mogami, et al., "Test of structural members and frames in tall buildings utilizing the
R. C. layered construction system, Part 1: Test on short beams," Taisei Technical Research
Report, No. 18, March 1986, pp. 111-126.
2. Tanaka, N., N. Sakaguchi, et al., "Flexural and shear strength of short span beams using ultra high
strength reinforced concrete, (Part 1: Flexural behavior of beams, Part 2: Shear behavior of
beams)," Summaries of Technical Papers of Annual Meeting, Architectural Institute of Japan,
Structures, 1987, pp. 67 - 70.
3. Kobayashi, J., K. Kanada, S. Yoshizaki and T. Yamada, "Test of structural members and frames in
tall buildings utilizing the R. C. layered construction system, Part 4: Test of shear reinforcing
methods of beams," Taisei Technical Research Report, No. 18, 1987, pp. 73 - 88.
4. Taga, A., K. Kawasaki, et al., "Development of MAEDA high-rise reinforced concrete building
system (MARC system), (Part 4: Experimental study on structural members and
subassemblages)," Report of Technical Research Institute, Maeda Corporation, Vol. 29-2, 1988,
pp. 31 - 55.
5. Yanagisawa, N., Y. Shimizu, K. Tsumura and M. Fujiwara, "Strength and ductility of reinforced
concrete T-beam with high strength concrete," Proceedings, Japan Concrete Institute, Vol. 10, No.
3, 1988, pp. 681 - 684.
6. Matsutani, T., J. Ishida, et al., "Experiments for development of high-rise reinforced concrete
structure," Technical Research Reports, Konoike Construction Co., Ltd., 1988, pp. 71 - 84.
7. Honda, Y., T. Iwakura, S. Hakuto and H. Maie, "Experimental study on RC beams with web
openings, ultimate shear strength and deformability of RC beams reinforced using ring materials
of high tensile strength steel," Technical Reports, Tokyu Construction, No. 15, 1989, pp. 67 - 72.
8. Sugano, S., T. Nagashima, H. Kimura and A. Ichikawa, "Experimental study on high strength
concrete beam using high strength main bar," Proceedings, Japan Concrete Institute, Vol. 12, No.
2, 1990, pp. 215 - 220.
24
9. Taga, A., K. Kawasaki, T. Watanabe, and K. Tsujita, "Study on a seismic design of a high-rise
reinforced concrete building (Flexure-shear loading tests and shear loading tests on beams with
web openings)," Report of Technical Research Institute, Maeda Corporation, Vol. 30, 1990, pp.
131 - 144.
10. Nakamura, M., S. Bessho, T. Kato and A. Zan, "Bending-shear test of beams with high strength
concrete and rebars for high rised R/C building," Proceedings, Japan Concrete Institute, Vol. 14,
No. 2, 1992, pp. 529 - 534.
11. Ishikawa, Y., M. Hamamoto, S. Otani and H. Aoyama, "Experimental study on deformation
capacity of reinforced concrete beams," Proceedings, Japan Concrete Institute, Vol. 14, No. 2,
1992, pp. 255 - 260.
12. Kamura, T., T. Ohmizu, S. Otani and H. Aoyama, "Experimental study on deformation capacity of
reinforced concrete beams," Proceedings, Japan Concrete Institute, Vol. 15, No. 2, 1993, pp. 335
- 340.
13. Sumi, A., T. Segawa, et al., "An experimental study on flexural performance of reinforced concrete
beams using high tensile strength shear reinforcement," Summaries of Technical Papers of
Annual Meeting, Architectural Institute of Japan, Structures, 1984, pp. 1681 - 1682.
14. Muguruma, H., A. Sumi, T. Segawa and T. Hisatoku, "An experimental study of reinforced
concrete beams laterally confined by high strength reinforcement," Summaries of Technical
Papers of Annual Meeting, Architectural Institute of Japan, Structures, 1988, pp. 229 - 230.
15. Sumi, A., K. Masuo, et al., "An experimental study on flexural performance of short span RC
beams using high strength shear reinforcement," Proceedings of Architectural Research Meetings,
Kinki Chapter, Architectural Institute of Japan, 1992, pp. 73 - 76.
16. Muguruma, H., F. Watanabe, "Study on shear design of R/C ductile beams subjected to combined
bending and shear (Part 1, 2)," Summaries of Technical Papers of Annual Meeting, Architectural
Institute of Japan, Structures 2-C, 1988, pp. 183 - 184.
17. Iwai, Y., Y. Kakita, F. Watanabe and H. Muguruma, "Study on shear design of RC ductile beams
subjected to combined bending and shear (Part 3)," Summaries of Technical Papers of Annual
Meeting, Architectural Institute of Japan, Structures 2-C, 1990, pp. 305 - 306.
18. Minami, K., H. Kuramoto, N. Tsukamoto and A. Nakazawa, "Shear and bond strength behavior of
R/C beams with grade 13,000kgf/cm2 shear reinforcement under cyclic bending and shear,"
Proceedings, Japan Concrete Institute, Vol. 12, No. 2, 1990, pp. 221 - 226.
19. Fujisawa, M., T. Kaminosono, M. Takeuchi and H. Murakami, "Study on bond splitting failure of
beams after flexural yielding," Report of Structures Committee, New RC Project, Japan Institute of
Construction Engineering, March 1990, pp. 4-4-1 - 4-4-15.
20. Fujisawa, M., T. Kaminosono, M. Takeuchi and H. Murakami, "Study on ductility of high-strength
reinforced concrete beams," Summaries of Technical Papers of Annual Meeting, Architectural
Institute of Japan, Structures, 1990, pp. 277 - 278.
21. Fujisawa, M., T. Kaminosono, et al., "Study on effect of slab on flexural performance of beams,"
Report of Structures Committee, New RC Project, Japan Institute of Construction Engineering,
March 1991, pp. 4-4-1 - 4-4-19.
22. Fujii, S., H. Fujitani, et al., " Study on bond splitting failure of beams after flexural yielding," Report
of Structures Committee, New RC Project, Japan Institute of Construction Engineering, March
1992, pp. 4-1-1 - 4-4-10.
23. Sugano, S., "Experimental study on restoring force characteristics of reinforced concrete
members (in Japanese)," a thesis submitted to the University of Tokyo for a partial fulfillment of the
requirements for doctor of engineering degree, University of Tokyo, December 1970.
24. Otani, S., "Hysteresis models of reinforced concrete for earthquake response analysis," Journal
(B), Faculty of Engineering, University of Tokyo, Vol. 36, No. 2, 1981, pp. 125-159.
25. Tomosawa, F., T. Noguchi, and K. Onoyama, "Investigation on fundamental mechanical
properties of high-strength and super high strength concrete," Summaries of Technical Papers of
Annual Meeting of Architectural Institute of Japan, Vol. A, 1990, pp. 497-498.
26. Shima, H., L.-L. Chou, and H. Okamura, "Micro and macro models for bond in reinforced
concrete," Journal, Faculty of Engineering, University of Tokyo, Series (B), Vol. 39, No. 2, 1987, pp.
133 - 194.
27. Eberhard, M. O., and M. A. Sozen, "Behavior-based Method to determine design shear in
earthquake-resistant walls," Journal, Structural Engineering, ASCE, Vol. 119, No. 2, February,
1993, pp. 619 - 640.

25
4.5 Analysis of Structural Walls

T. Paulay (1981) presents analysis procedures of reinforced concrete structural walls for design
purpose. The following deformation should be considered for a structural wall; (a) flexural
deformation, (b) shear deformation after diagonal cracking, (c) anchorage deformation, (d)
deformation of foundation and supporting ground.

The equivalent second moment I e of area may be taken as 60 % of the value I g based on the
uncracked gross concrete section. A more accurate estimate of flexural deformations may be made if
the ratio of the moment causing cracking to the maximum applied moment is evaluated;
 M cr 
3
  M 3 
Ie =   I g + 1 −    I cr
cr
(1)
M
 a  M
  a  
where, M cr : cracking moment, M a : maximum moment at which deflection is computed, I g :
second moment of area of the gross concrete section, I cr : moment of inertia of cracked transformed
section. The contribution of longitudinal reinforcement can be neglected because the reinforcement
ratio is normally less than 0.1 % in a wall.

The elongation of the vertical bars within the foundation structure and slip due to high local bond
stresses along the development length will result in an apparent pull-out of such bars at the base of
the wall. This can significantly increase the wall deflection.

For cantilever walls with aspect ratios, hw / l w , larger than 4, the shear deformation may be
neglected. After diagonal cracking, however, the shear stiffness is reduced to 10 - 30 % of the
uncracked stiffness. The lateral reinforcement influences the shear deformation after diagonal
cracking.

When the aspect ratio is less than 4, the second moment of area for a structural wall may be
assumed that
Ie
Iw =
1.2 + F
(2)
30 I
F= 2 e
hw bw l w
In this expression, some allowance is made for shear distortions and deflections due to anchorage
(pull-out) deformation at the base of a wall.

The structural wall tests at Portland Cement Association were examined. The aspect ratio of the
specimens was 2.4. Cross sections were flanged (H-shape), barbell (with boundary columns) and
rectangular. Equation (1) tends to overestimate the initial stiffness.

For a cantilever wall, the wall may be represented by a straight line passing through the centroid
of the gross section..

Reference:

Paulay, T., “The Design of Reinforced Concrete Ductile Shear Walls for Earthquake
Resistance,“ Research Report 81-1, Department of Civil Engineering, University of Canterbury,
New Zealand, February 1981, 72 pp.

26
 Home Assignment No. 2

2002-02-27 Otani, S.

A simply supported beam was tested at two point loading. The cross section was 102 x 152 mm;
support distance was 2.60 m with each load point at 1.00 m from the support. The section was
reinforced singly by 2-D12 bars at the bottom (cover thickness of 15 mm). Concrete strength was
22.5 MPa; load at yielding of a D12 reinforcing bar was 37.1 kN. The beam was reinforced laterally by
φ 6 bars at 100 mm spacing.

Assume concrete and reinforcing bars behave linearly elastic until flexural yielding. The concrete
in tension carries tensile stress before cracking (tensile strength of concrete is one-tenth of
compressive strength). Ignore the stress carried by tensile region of concrete after cracking. Use
rectangular stress block for concrete in calculating the moment and curvature at an ultimate state,
where a strain of concrete at the extreme compression fiber reaches a limiting value of 0.004 and
stress of tensile reinforcement reaches the yield stress.

Calculate (a) moment-curvature relation of section at cracking, flexural yielding and ultimate
stages, (b) load-deformation relation of the specimen on the basis of curvature distribution at cracking,
flexural yielding and ultimate stages at the loading point.

Compare the calculated and the observed load-deformation relation. The total load (kN) and
displacement (mm) at mid-span relation is shown below:

Total Load, kN
20.0

10.0

Deflection at mid-span, mm

10.0 20.0 30.0

27
(1) Linearly elastic stage:
Young’s modulus of concrete
Ec = 1.35 × γ 1.5 × σ B0.5
= 1.35 × 2.5 × 103 × 9.8 × (22.5)0.5
= 24.6 GPa
Young’s modulus of steel
Es = 210 GPa
Total area of longitudinal reinforcement
As = 2π r 2 = 2 × 3.14 × (0.006) 2 = 2.26 × 10−4 m 2
Effective depth
d = 125 mm
Modulus ratio
Es
n= = 8.55
Ec
Moment of inertia of transformed section
I 0 = 3.35 ×10−5 m 4
Neutral axis from the extreme tensile fiber
xn = 80.9 mm
Cracking moment
Io
M c = σ ct = 1061 Nm
( D − xn )
Cracking curvature
Mc
φc = = 1.29 × 10−3 (1/ m)
Ec I 0

(2) Moment and curvature at yielding

Tensile reinforcement ratio
As 2.26 × 10−4
ρt = = = 0.01776
bd 0.102 × 0.125
Depth of neutral axis kd of cracked section
k = ( ρt n) 2 + 2 ρt n − ρ t n = 0.420
Yield moment
k
M y = 2 × Py × d × (1 − )
3
0.420
= 2 × 37.1× 103 × 0.125 × (1 − )
3
= 7980 Nm
Moment of inertia of cracked section
b(kd )3
I cr = + nAs (1 − k ) 2 d 2
3
0.102 × (0.420 × 0.125)3
= + 8.55 × 2.26 × 10−4 × (1 − 0.420) 2 × 0.1252
3
= 1.506 ×10−5 m 4
Yield curvature

28
 My
φy = = 0.0215 (1/ m)
Ec I cr

(c) Ultimate moment and curvature

The tensile reinforcement ratio is less than the balanced reinforcement ratio.
From the equilibrium of axial forces,
0.85 f c' ab = As f y
As f y
a=
0.85 f c'b
37.1× 2 × 103
= = 0.038 m
0.85 × 22.5 ×106 × 0.102
Ultimate moment
a
M u = As f y (d − )
2
0.038
= 37.1× 103 × 2 × (0.125 − )
2
= 7870 Nm
Depth of neutral axis
a
c= = 0.0447 m
0.85
Ultimate curvature
0.004
φu = = 0.0894 (1/ m)
c

29
 Assignment No. 2
20020227
S. Otani

A cantilever reinforced concrete beam was tested under lateral load reversals. The shear span
(distance between loading point and the end of the beam specimen) was 750 mm. The section is
200x300 mm and is reinforced with 4-D13 bars (SD345, nominal area of a bar= 127 mm2, tensile
reinforcement ratio of 0.85%) at the top and bottom. The lateral reinforcement was 2- φ 4 plain bars
placed at 40 mm on centers (lateral reinforcement ratio of 0.32 %, nominal area of a bar=12.6 mm2).
The concrete cover to the center of the longitudinal reinforcement was 30 mm.

The standard tensile test shows that

the yield stress of Grade SD345 steel 80
was 361 N/mm2, and tensile strength 60
was 500 N/mm2. Yield stress of a plain
40
φ 4 bar was 478 N/mm2 and the tensile
Resistance, kN
20
strength was 509 N/mm2. The 0
compressive strength of concrete was -30 -20 -10 -20 0 10 20 30
28.2 N/mm2, tensile strength 2.50 -40
N/mm2, and secant elastic modulus
-60
28.6 GPa.
-80
Displacement, mm
The load-deformation relation of the
specimen is shown below;

Estimate the resistance and deformation under monotonically increasing loading at flexural
cracking, yielding and ultimate points from the given information and compare the calculated results
with the test results. The observed relation is enlarged below.

80

60

40
Resistance, kN

20

0
-8 -6 -4 -2 0 2 4 6 8
-20

-40

-60

-80
Displacement, mm

30
 Chapter 5. Structural Dynamics

5.1 Differential Equation of Motion

Newton's Law of Motion: The relationship between force and motion was formulated by Sir Isaac
Newton (1642-1727) in "Principia (1687)."
Law 1: The absolute velocity {v} of a particle remains constant if there is no net external force
applied to the particle.
Law 2: When a force { f } acts on a particle, the absolute acceleration {a} of the particle is
directly proportional to the force.
{a} = c{ f }
Law 3: When two particles A and B are in contact, the force applied to particle A by particle B, at
the contact point, is equal in magnitude but opposite in direction from the force applied to particle B
by particle A.

"Newton's second law of motion" expresses the motion of a particle under a given force history:
d2
{x} = c{ f }
dt 2
By solving the differential equation, the displacement {x} of the particle can be uniquely defined for
a given set of initial conditions.

D'Alembert's Principle: Newton’s law of motion can be rewritten in the form:

{ f } = m{a}
d2
=m {x}
dt 2
If a particle moves with an absolute acceleration of {a (t )} at time t , force { f (t )} must act on the
particle in the direction of the acceleration. The proportionality constant is called "mass of inertia."

A fictitious force commonly known as the "inertia force" was introduced by D'Alembert (Traite de
Dynamique, 1743) in order to express a state of dynamic equilibrium:
d2
{ f '} = − m {x}
dt 2

The equilibrium of (static) forces was extended to a dynamic problem by the introduction of an inertia
force:
{ f } + { f '} = {0}

Equation of Motion for Mass-Spring-Damping System: x

A system consisting of a mass and a spring oscillates with
a constant period. For simplicity, let us consider a system
k
consisting of one translational mass of inertia m and one
linear spring k , where the particle can move in the m
direction of the lineal spring.

Writing an equilibrium equation of forces (D'Alembert's

Principle) at a given displacement;
Mathematical Model of Mass-Spring System

1
 d 2x
−m − kx + f (t ) = 0
dt 2
d 2x
m 2 + kx = f (t )
dt

The solution x(t ) for a differential equation is the sum of a particular solution x p (t ) , which
satisfies the original equation at any time, and a complementary solution xc (t ) , which is a general
solution for the differential equation for the right hand side of the equation to be null (the state of free
vibration).
x ( t ) = x p (t ) + x c (t )
d2
m x p (t ) + k x p (t ) = f (t )
dt 2
d2
m 2 x c (t ) + k x c (t ) = 0
dt
The particular solution x p (t ) must be found for a given forcing function f (t ) . The particular
solutions have been found for simple forcing functions f (t ) , but a general closed form solution is
not available.

The complementary solution xc (t ) can be found as follows;

x c (t ) = A cos ω n t + B sin ω n t
= A2 + B 2 sin(ω n + φ )
A
sin φ =
A + B2
2

B
cos φ =
A2 + B 2
k
where ωn = . A and B are integration constants dependent on the initial conditions. For given
m
initial displacement xo and velocity v o at t = 0 : the complementary solution is derived as
v
xc (t ) = xo cos ω n t + o sin ω n t
ωn
2π m
Note that the system oscillates with a constant period of Tn = = 2π .
ωn k

This period Tn of oscillation is called

"natural period" of the system; ω n is
called "circular (or angular) (natural)
frequency" of the system. Natural
frequency f n is the reciprocal of the
natural period.

2
5.2 Mass of Inertia

The dynamic equilibrium of forces is normally formulated for each displacement degree of
freedom. The inertia force is associated with mass and acceleration; the mass of a particle is
constant, but the mass must be defined for each degree of freedom. It should be noted that there
exists mass associated with rotational acceleration.

Distributed Mass: The mass distributes in a solid body, the displacement is not uniform in a
deformable body. The system of a deformable body has an infinite number of degrees of freedom.
For a practical analysis, the number of degrees of freedom must be reduced to a finite number.

Consider a lineal member of length L . The (axial/ lateral/ rotational) displacement is y ( x, t ) at

coordinate x at time t . Suppose the member oscillates in a single shape φ (x ) with the
time-varying amplitude (coordinate) Y (t ) ;
y ( x, t ) = φ ( x )Y (t )
The kinetic energy T of the entire member under this oscillation is defined as;
L
1 dy ( x, t ) 2
T =∫ ρ ( x){ } dx
0
2 dt
L
1 dY (t ) 2
=∫ ρ ( x){φ ( x) } dx
0
2 dt
L
1 dY (t ) 2
=
20∫ ρ ( x){φ ( x)}2 dx {
dt
}

where, ρ (x ) : mass (translational or rotational) per unit length at location x .

Using the displacement coordinate Y (t ) , the kinetic energy is defined as

1 * dY (t ) 2
T= M { }
2 dt
where M*: equivalent lumped mass associated with the deformed shape. Therefore, the equivalent
mass is defined as
L
M = ∫ ρ ( x ){φ ( x )}2 dx
*

The shape function φ (x ) is sometimes called a generalized coordinate, and M

*
is a generalized
mass associated with the coordinate.

Consistent Mass: Mass coefficients corresponding to the nodal coordinates of a beam element may
be defined by a procedure similar to the determination of element stiffness coefficients; the mass
coefficient mij is the force at nodal coordinate i due to a unit acceleration at nodal coordinate j
while all other nodal coordinates are maintained at zero acceleration.

It is assumed that the deflection resulting from unit dynamic displacement d i = 1.0 at the nodal
coordinate i of a beam element is given by the consistent function φi (x ) obtained from static
consideration.

3
 For a linearly elastic prismatic uniform p3 p6
beam element, the deflection due to a p2 p5
member end displacement is expressed by a d3 d6
cubic function if no external load acts within d2 d5
the member. The deflected shape φi (x )
due to unit displacement d i at an end is p4
p1
given in Reference (Paz, 1985).
d1 d4
(1) Axial displacement φ1 ( x ) due to unit
displacement d 1 = 1.0 at the starting end, Coordinate Systems for Forces and Displacements
(2) Lateral displacement φ 2 ( x ) due to unit displacement d 2 = 1.0 at the starting end,
(3) Lateral displacement φ 3 ( x ) due to unit displacement d 3 = 1.0 at the starting end,
(4) Axial displacement φ 4 ( x ) due to unit displacement d 4 = 1.0 at the starting end,
(5) Lateral displacement φ5 ( x ) due to unit displacement d 5 = 1.0 at the starting end,
(6) Lateral displacement φ 6 ( x ) due to unit displacement d 6 = 1.0 at the starting end,

θ1z θ2z
x d1y d2y
φ1 ( x ) = 1 − ( ) for d1x = 1.0
L d1x d2x
x x d1x=1.0
φ 2 ( x) = 1 − 3( ) 2 + 2( ) 3 for d1 y = 1.0
L L
x
φ 3 ( x) = x{1 − ( )}2 for θ 1z = 1.0
L
d1y=1.0
x
φ 4 ( x) = ( ) for d 2 x = 1.0
L θ1z = 1.0
x x
φ 5 ( x) = 3( ) 2 − 2( ) 3 for d 2 y = 1.0
L L d2x=1.0
x x
φ 6 ( x) = x( ){( ) − 1} for θ 2 z = 1.0
L L

If acceleration is assumed to be proportional to d2y=1.0

displacement, the inertia force f i (x) per unit
length due to a member end acceleration ai (t ) is
f i ( x ) = − ρ ( x )φi ( x ) ai (t ) θ 2 z = 1.0

The inertia force f ji in j -direction caused by member end acceleration ai (t ) in i -direction

using a consistent mass m ji is
f ji = −m ji ai

For a virtual displacement δ j at member

end in j -direction, the virtual work WE of
the external force at coordinate i is
WE = − m ji aiδ j
and the corresponding virtual work WI of
internal forces is

4
 L
WI = − ∫ ρ ( x )aiφi ( x )δ jφ j ( x )dx
0
L
= − aiδ j ∫ ρ ( x )φi ( x )φ j ( x )dx
0
Therefore,
L
mij = ∫ ρ ( x )φi ( x )φ j ( x )dx
0
The consistent mass matrix associated with the member end acceleration is given (Ref. 1) by
 p1 x  140 0 0 70 0 0  d 1xx 
p   0 156 22 L 0 54 − 13L  d 
 1y     1y 
 m1z  ρL  0 22 L 4 L2 0 13L − 3L2  d 2  θ1z 
 =    
 p 2 x  420  70 0 0 140 0 0  dt 2 d2x 
 p2 y   0 54 13L 0 156 − 22 L d2 y 
     
 0 − 13L − 3L 0 − 22 L 4 L2  θ2z 
2
m 2 z 
The deflection shape function changes in an inelastic stage. It is normally difficult to evaluate the
distributed mass and the deflected shape consistent with member stiffness matrix. The member
mass matrix must be reformulated consistent with the stiffness matrix.

The lateral deflection shape functions for a member with a 1

rotational plastic hinge at the start end are given (Ref. 1) as 2

3 x 1 x
φ 2 ( x) = 1 − ( ) + ( ) 3 for d1 y = 1.0
2 L 2 L
φ 3 ( x) = 0 for θ 1z = 1.0
3 x 1 x
φ 5 ( x) = ( ) − ( ) 3 for d 2 y = 1.0
2 L 2 L
1 x
φ 6 ( x) = x{( ) 2 − 1} for θ 2 z = 1.0
2 L
The consistent mass matrix for this case is given (Ref. 1) by
 p1x  140 0 0 70 0 
0  d1x 
p   0 204 0 0 − 16.5 L 
58.5 d 
 1y    1y 
 m1z  ρL  0 0 0 0 0 0  d2 θ 1z 
 =    
 p 2 x  420  70 0 0 140 0 0  dt 2 d 2 x 
 p2 y   0 58.5 0 0 99 − 36 L  d 2 y 
     
m2 z   0 − 16.5 L 0 0 − 36 L 8 L2  θ 2 z 

The deflection shape functions for a member with a

1 2
rotational plastic hinge at the terminal end are given (Ref. 1) as

5
 3 x 1 x
φ2 ( x ) = 1 − ( ) + ( ) 3 for d1 y = 1.0
2 L 2 L
3 x 1 x
φ3 ( x ) = x − x ( ) + x ( ) 2 for θ1z = 1.0
2 L 2 L
3 x 1 x
φ5 ( x ) = ( ) 2 − ( ) 3 for d 2 y = 1.0
2 L 2 L
φ6 ( x ) = 0 for θ 2 z = 1.0
The consistent mass matrix associated with the member end acceleration is given (Paz, 1985) by
 p1x  140 0 0 70 0 0  d1x 
p   0 99 36 L 0 58.5 0 d 
 1y    1y 
 m1z  ρL  0 36 L 8 L 0 d 2 θ 1z 
2
0 16.5 L
 =    
 p 2 x  420  70 0 0 140 0 0 dt 2 d 2 x 
 p2 y   0 58.5 16.5 L 0 204 0 d 2 y 
     
m2 z   0 0 0 0 0 0 θ 2 z 

The lateral deflection shape functions for a member with 1 2

rotational plastic hinges at the two ends are given (Ref. 1) as
x
φ 2 ( x) = 1 − ( ) for d1 y = 1.0
L
φ 3 ( x) = 0 for θ 1z = 1.0
x
φ 5 ( x) = ( ) for d 2 y = 1.0
L
φ 6 ( x) = 0 for θ 2 z = 1.0
The consistent mass matrix associated with the member end acceleration is given by
 p1x  2 0 0 1 0 0  d1x 
p  0 2 0 0 1 0 d 
 1y    1y 
 m1z  ρL 0 0 0 0 0 0 d 2 θ 1z 
 =    
 p 2 x  6 1 0 0 2 0 0 dt 2 d 2 x 
 p2 y  0 1 0 0 2 0 d 2 y 
     
m2 z  0 0 0 0 0 0 θ 2 z 

It should be noted that (a) the displacement functions due to unit member end displacement is not
the same as the deflected shape due to member end displacement and inertia force acting along the
member, (b) the stiffness distribution changes along a member in the nonlinear response due to the
progress in damage during an oscillation. Therefore, the deflected shape must be evaluated at each
instance during response analysis. The consistent mass is not constant with time, but is affected by
the stiffness (damage) distribution.

Lumped Mass: A simple method to consider the inertial properties for a dynamic system is to
assume that the mass of the structure is lumped at the nodal coordinates where translational
displacements are defined. The inertia effect associated with any rotational degree of freedom may
be assumed to be zero.

For a member AB of length L and distributed mass ρ (x) per unit length, the lumped mass may
be defined by considering the rigid body motion;
6
 L
x
m A = ∫ {1 − ( )}ρ ( x)dx
0
L
L
x
m B = ∫ ( ) ρ ( x)dx
0
L
The mass matrix may be formed by adding the contribution of lumped masses at the nodal
coordinates defined for translations.

A regular building structure oscillates in a horizontal direction under a horizontal earthquake

motion, causing horizontal inertia forces associated with the mass of floors and beams. The floor
diaphragm may be assumed to be rigid in its own plane in a cast-in-situ reinforced concrete building,
yielding the same horizontal displacement at a floor. Therefore, the mass of a building may be
assumed to concentrate at the floor level; the mass of inter-story structural and non-structural
elements may be included in the floor mass.

The total mass of a structure may be assumed to concentrate at the levels of the floors, and no
other mass may be ignored as the secondary effect. The mass matrix becomes diagonal with
non-zero elements only at the locations associated with horizontal floor displacements. For
horizontal inertia forces PXi at floor level i , the lumped mass matrix may be written as follows;
 PX 1   M 1 0 0 K 0 0   DX1 
P   0 M2 0 L 0 0  D 
 X2    X2 
 PX 3   0 0 M3 L 0 0  d2  D X 3 
 =   
 M   M M M O M M  dt 2  M 
 PXn −1   0 0 0 L M n −1 0   D Xn −1 
     
 PXn   0 0 0 L 0 M n   D Xn 
2
d
where {D X } : horizontal acceleration at floor level i , and M i : total mass of floor level i .
dt 2

Reference:

1. Mario Paz, Structural Dynamics, Theory & Computation, Van Nostrand Reinhold Co. Ltd.,
1985.

7
5.3 Damping

A mass-spring (oscillatory) system under free vibration does not oscillate forever, but the
amplitude of oscillation in a real system is known to diminish with time. In other words, kinetic energy
of motion must decrease with time indicating there exists an energy dissipating mechanism in a real
oscillatory system. Such a mechanism of energy dissipation is vaguely termed as "damping."

"Damping" is a mechanism to dissipate kinetic energy; some mechanisms have been suggested
for damping such as;
(a) inelastic hysteresis energy dissipation,
(b) radiation of kinematic energy through foundation,
(c) kinetic friction,
(d) viscosity in materials, or
(e) aerodynamic effect.

Unfortunately, the state-of-the-art cannot define the characteristics of damping on the basis of
material properties and geometrical configuration of a structure.

"Because damping mechanism is not clearly understood, we may use a crude model to represent
the energy dissipating feature."

Looking at the equation of motion, there is not a linear term associated with velocity. Therefore, it
is mathematically convenient and beautiful to introduce a velocity related term; i.e., a resistance
proportional to velocity called "viscous damping":
d 2x dx
m 2
+ c + kx = f (t )
dt dt
where c: damping coefficient.

Solution (complementary solution) for free vibration with damping yields

xc = C1 e p1t + C 2 e p2t
where
− c ± c 2 − 4mk
p1, 2 =
2m

Critical damping coefficient c cr is defined as

2
c cr − 4mk = 0
c cr = 2 mk
for which damping is so large that the oscillation will not be developed. In a normal structural system,
damping coefficient is much smaller than the critical value.

Ratio of a damping coefficient c of a system to the critical damping coefficient c cr is called a

damping factor (or ratio) h;
c
h=
2 mk

The value of a damping factor is normally determined by dynamic tests of a structure rather than
calculated from the material properties and geometry of the structure. Damping capacity is often
determined by the bandwidth at the half power level of the response curve during a sinusoidal
"steady-state test" or by "logarithmic decrement" of the response record during a free vibration test.

Note that the damping capacity is not a unique value of a structure, but is known to vary with the
level of excitation. Acceleration response amplitude was plotted with respect to exciting frequency in
a steady-state test of a reinforced concrete building at different excitation levels (Jennings and
8
Kuroiwa, 1968). The resonant frequency
decreased with the level of excitation and
the damping value increased despite low
response amplitude.

Damping factor is known to be

approximately 0.03 to 0.05 for reinforced
concrete buildings. Radiation damping is
dominant source of energy dissipation in a
heavy reinforced concrete building.

Free vibration or steady-state vibration

test data show a discrepancy from the
theoretical response of systems derived
for viscous damping naturally because the
mechanism of damping in a real structure
is not of viscous type.
1 f 2 − f1
h≅
2 fn

The complementary solution xc (t )

using damping coefficient is
xc = e − hω nt ( A cos ω d t + B sin ω d t )

where ω d : circular frequency of a damped system:

ωd = ωn 1 − h2
and A, B: integration constants dependent on the initial conditions.

For a given set of initial displacement xo and velocity v o at t = 0.0 , the solution under free
vibration is
v o + x o hω n
x(t ) = e − hω nt [ xo cos ω d t + sin ω d t ]
ωd

Damping factor h may be determined by the logarithmic decrement, defined as the natural
logarithm of the amplitude ratio of two consecutive peaks in free vibration;
1 x
h= ln m
2π xm +1
9
 Viscous damping is the only means to dissipate kinetic energy in a linear system, and damping
has a considerable influence on response amplitude of a linear system.

Damping Matrix: The mechanism of damping in a structure is not of viscous type. The damping
matrix cannot be formulated by the material properties and geometry of a structure. Therefore, it is
not reasonable to assume complicated damping. The damping matrix may be formulated for each
story rather than for each member.

Rayleigh type damping（Lord Rayleigh, 1842-1919）of the following form is often assumed in the
response analysis of a multi-degree-of-freedom system;
[C ] = a o [ M ] + a1 [ K ]
where [C ] : damping matrix, [M ] : mass matrix, and [K ] : stiffness matrix, and a0 and a1 are
proportionality constants. The Rayleigh damping is used because the general coordinates (mode
shapes), obtained for an undamped system, are also orthogonal with respect to the damping matrix.
For n -th mode, damping factor hn is expressed as
ao
hn = + a1ω n
ωn
If the damping matrix is made proportional to the mass matrix, damping factors corresponding to
vibration modes decreases with mode number; i.e., higher mode response can be more easily
excited. If the damping matrix is proportional to stiffness matrix, damping factors increases with
mode number; i.e., higher mode response can be suppressed in the response. The constants a o
and a1 are determined by damping factors and frequencies of vibration in arbitrary two modes.

In general, the mode shapes can be made orthogonal if a damping matrix is selected in the
following form (Caughey damping);
[C ] = [ M ]∑ a i ([ M ] −1 [ K ]) i
i

The Rayleigh damping is a special case of the Caughey damping, in which i = 0 and i = 1 are
considered. A damping factor hn of n -th mode is given as
1
hn =
ωn
∑a ω
i
i n
2i

Note that damping factors are normally estimated from a series of steady-state dynamic tests of
a structure, but the accuracy of the values is limited due to, first of all, unknown mechanism of
damping, and difficulty in obtaining higher mode modal characteristics.

With the reduction of stiffness caused by damage in en inelastic response, the constant damping
matrix (proportional to constant mass matrix or proportional to initial elastic stiffness matrix) tends to
increase effective damping factors. The additional energy dissipation by damping with damage is
hard to rationalize because hysteretic energy dissipation is considered in the hysteresis of stiffness
models. Therefore, it is recommended to make the damping matrix proportional to instantaneous
stiffness in an in elastic response analysis. Damping matrix proportional to instantaneous stiffness is
reported to be favorable in simulation of earthquake response of test structures (Otani and Sozen,
1972, Omote and Takeda, 1974).

References:

Jennings, P. C., and J. H. Kuroiwa, "Vibration and Soil-Structure Interaction Tests of a Nine-story
Reinforced Concrete Building," Bulletin, Seismological Society of America, No. 58, 1968, pp.
891-916.
Omote, Y., and T. Takeda, "Study on Elasto-plastic Response of Reinforced Concrete Chimney
10
 (Part 1: Model Test) (in Japanese)," Transactions, Architectural Institute of Japan, No. 215,
January 1974, pp. 21-32.
Otani, S., and M. A. Sozen, "Behavior of Multistory Reinforced Concrete Frame Structures," Civil
Engineering Studies, Structural Research Series No. 392, University of Illinois, 1972.

11
5.4 Strain-rate Effect

The speed of loading has been known to influence the stiffness and strength of various materials
since 19-th century; however, the difficulty in testing materials under controlled loading rate and in
instrumenting the response delayed the understanding of the characteristics.

Material, subjected to fast loading (Cowell, 1965, 1966),

(a) increases initial stiffness,
(b) increases resistance at yielding, but
(c) does not change hysteresis characteristics, and
(d) does not change the behavior after strain hardening,
including ductility and energy dissipation characteristics. For example, strain rate of 0.3 mm/mm/sec
was reported to increase the upper yield point stress by 1.60 times from the yield stress at
quasi-static loading; such loading rate was intended for explosion rather than for earthquake
response. The strain rate expected in concrete material during an earthquake is in the order of 0.001
to 0.10 mm/mm/sec (Mahin and Bertero, 1972).
Tensile stress, kgf/cm2

Compressive stress, kgf/cm2

Quasi-static

Strain, mm/mm Strain, mm/mm

Strain rate effect on material stress-strain relations

(Manjoine,1944, and Mahin and Bertero, 1972)

Reinforcing bars were tested under constant

strain rate by Mahin and Bertero (1972).
Stress-strain relation under stress reversals in
Tensile stress of steel, kgf/cm2

plastic region will not be affected by the strain

rate.

Average bond stress-bar slip relation (gauge

length of 30 mm) of deformed and plain bars was
obtained under impact pullout loading by Vos and
Reinhardt (1982); the bond stress rate was
varied from 0.3 MPa/sec to 105 MPa/sec. The
loading rate did not influence the bond
characteristics of plain bars; bond resistance
remained constant after 0.02 mm bar slip.
Deformed bars increased the initial stiffness and Strain, mm/mm
average bond stress with loading rate in
deformed bars; the bond resistance increased Stress-strain relation of steel under load reversals
with bar slip. The loading rate effect decreases
12
with strength of concrete. The difference
in the loading rate effect on deformed Bond stress speed

Average bond stress, kgf/cm2

bars and plain bars might be interpreted
that chemical adhesion bond resistance
is not influenced by loading rate, but
mechanical interlocking bond resistance
is influenced by the loading rate.

Tests of reinforced concrete members

at constant velocity of 0.002 rad/sec and
0.2 rad/sec of member rotational velocity
(Mahin and Bertero, 1972) indicated that Bar slip, mm
(a) high strain rates increased the
initial yield resistance, but caused small
differences in either stiffness or

Average bond stress, kgf/cm2

resistance in subsequent cycles at the
same displacement amplitudes,
(b) strain rate effect on resistance
diminished with increased deformation in
a strain-hardening range, and
(c) no substantial changes were Bond stress speed
observed in ductility and overall energy
absorption capacity.

Earthquake simulator tests of

reinforced concrete members and
structural models have been successfully Bar slip, mm
Bar slip
simulated by analysis on the basis of
hysteresis model developed on the basis
of force-deformation relationship Average bond stress-bar slip relation for D10 bars
observed in static laboratory tests (vos and Reinhardt, 1982)
(Takeda, Nielsen and Sozen, 1970,
Otani and Sozen, 1972, Koike, Omote and Takeda, 1980,).
Resistance, tonf

Resistance, tonf

Displacement, cm Displacement, cm

Note that strain rate (velocity) during an oscillation is highest at low stress levels, and that the rate
gradually decreases toward a peak strain (displacement). Damage in the reinforced concrete
reduces the stiffness, elongating the period of oscillation. Furthermore, such damage is caused
normally by lower modes of oscillation having longer periods. Therefore, the strain rate effect can be
judged small on the earthquake response of a normal reinforced concrete structure.

13
References:
Cowell, W. L., "Dynamic Tests of Concrete Reinforcing Steels," Technical Report No. 394, U.S.
Naval Civil Engineering Laboratory, Port Hueneme, California, 1965.
Cowell, W. L., "Dynamic Properties of Plain Portland Cement Concrete," Technical Report No. 447,
U.S. Naval Civil Engineering Laboratory, Port Hueneme, California, 1966.
Koike, K., Y. Omote and T. Takeda, "Reinforced Concrete Wall-Frame Structures subjected to
Dynamic and Static Loading," Proceedings, Eighth World Conference on Earthquake Engineering,
Vol. 6, September 1980, 419-426.
Mahin, M. A., and V. V. Bertero, "Rate of Loading Effect on Uncracked and Repaired Reinforced
Concrete Members," EERC No. 72-9, Earthquake Engineering Research Center, University of
California at Berkeley, 1972.
Manjoine, M. J., "Influence of Rate of Strain and Temperature on Yield Stresses of Mild Steel,"
Journal of Applied Mechanics, Transactions, American Society of Mechanical Engineers, Vol. 11,
December 1944, pp. A211-A218.
Otani, S., and M. A. Sozen, "Behavior of Multistory Reinforced Concrete Frames during
Earthquakes," Civil engineering Studies, Structural Research Series No. 392, University of Illinois,
November 1972.
Takeda, T., N. N. Nielsen and M. A. Sozen, "Reinforced Concrete Response to Simulated
Earthquake," Proceedings, Structural Engineering Division, ASCE, Vol. 96, No. St12, December
1970, pp. 2557-2573.
Vos, E., and H. W. Reinhardt, "Influence of Loading Rate on bond Behavior of Reinforcing Steel and
Prestressing Strands," Materials and Structures, Vol. 15, No. 85, March 1982, pp. 3-10.

Further References on Strain Rate Effect

1. Atchley, B. L., H. L. Furr, "Strength and Energy Absorption Capabilities of Plain Concrete under
Dynamic and Static Loadings," Journal, American Concrete Institute, Vol. 64, No. 11, November
1967, pp. 745-7556.
2. Birkimer, D. L., and R. Lindemann, "Dynamic Tensile Strength of Concrete Materials," Journal,
American Concrete Institute, Vol. 68, No. 1, January 1971, pp. 47-49.
3. Clark, D. S., and P. E. Duwez, "The Influence of Strain Rate on Some Tensile Properties of Steel,"
Proceedings, American Society of Testing Materials, Vol. 50, 1950, pp. 560-575.
4. Criswell, M. E., "Static and Dynamic Response of Reinforced Concrete Slab-Column Connections,
ACI SP-42, Shear in Reinforced Concrete, Vol. II, 1974, pp. 721-746.
5. Crum, R. G., "Tensile Impact Tests for Concrete Reinforcing Steel," Journal, American Concrete
Institute, Vol. 56, No. 1, July 1959, p. 59.
6. Kaplan, S. A., "Factors Affecting the Relationship between Rate of Loading and Measured
Compressive Strength of Concrete," Magazine of Concrete Research, Vol. 32, No. 111, June
1980, pp. 79-88.
7. Manjoine, M. J., and A. Nadai, "High Speed Tension Tests at Elevated Temperatures, Part 1,"
Proceedings, American Society of Testing Materials, Vol. 40, 1940, pp. 822-837.
8. Seabolt, R. H., "Dynamic Shear Strength of Reinforced Concrete Beams, Part II," Technical
Report R-502, U. S. Naval Civil Engineering Laboratory, California, January 1967, pp. 52-56.
9. Shiga, T. and J. Ogawa, "The Experimental Study on the Dynamic Behavior of Reinforced
Concrete Frames," Proceedings, Fifth World Conference on Earthquake Engineering, Santiago,
Chile, 1969, B-2, pp. 166-176.
10. Sparks, P. E., and J. R. Menzies, "The Effect of Rate of Loading upon the Static and Fatigue
Strengths of Plain Concrete in Compression," Magazine of Concrete Research, Vol. 25, No. 83,
June 1973, pp. 73-80.
11. Takeda, J, and H. Tachikawa, "Deformation and Fracture of Concrete subjected to Dynamic
Load," Proceedings, International Conference on the Mechanical Behavior of Materials, Kyoto,
Vol. IV, 1971.
12. Watstein, D. "Effect of Strain Rate on the Compressive Strength and Elastic Properties of
Concrete," Journal, American Concrete Institute, Vol. 49, No. 8, April 1953, pp. 729-744.
13. Zielinski, A. J., H. W. Reinhardt and H. A. Kormeling, "Experiments on Concrete under Uniaxial
Impact Tensile Loading," Materials and Structures, Vol. 14, No. 80, March-April 1981, pp.
103-112.

14
5.5 Properties of Earthquake Ground Motion

The earthquake risk is not uniform on the earth, but varies significantly from a region to another.
The recent development in seismology is fascinating; plate tectonics developed since the 1960s can
explain the occurrence of earthquakes along the boundaries of tectonic plates. Most major
earthquakes occur along the boundaries of tectonic plates due to their relative movement, which can
be monitored with the use of the global positioning system (GPS). Seismically blank regions where
next large earthquakes may occur can be identified by the observation and historical records. These
plate boundary earthquakes occur at a relatively uniform interval (50 to a few hundred years) for a
given region with accumulation of strain energy.

Some faults have been identified on ground surface, but others are buried under the ground.
Some earthquakes of lesser magnitudes occur by the fracture of active faults within a tectonic plate
caused by stresses developed by the plate movement. Some active faults have been identified on
ground surface, but others are buried under the ground. An active fault in a tectonic plate may
fracture once in a few thousand years.

It is not possible at this stage to accurately predict the time, location and magnitude of an
earthquake occurrence.

Earthquake Ground Motions: A seismometer to measure ground displacement during an

earthquake was developed in late nineteenth century. The seismometer has been used by
seismologists to understand the source mechanism of earthquakes, but it does not provide
acceleration records necessary for engineering purpose. The seismologist believed that the
acceleration signal was affected by accidental phenomena such as local geology.

A strong motion accelerograph to record ground acceleration was developed in the early 1930s.
The characteristics of earthquake ground motions were studied through the observed records; i.e.,
common features of acceleration records were abstracted and general shapes of response spectra
were established for design purpose taking the local effect of soil into account.

Historical records about earthquake occurrences are studied to estimate the probability of the
maximum earthquake intensity in a region. A large uncertainty exists in the estimated maximum
ground acceleration attributable to the inaccuracy and the limited period of the historical
documentation.

Engineering Seismology: Earthquake ground motions specific at a construction site is influenced

by the geometry of active faults, dynamic rupture process of earthquake sources, and the
transmission of earthquake motions from the earthquake source to the construction site. There have
been efforts by engineering seismologists to estimate the characteristics of future earthquake
motions.

The global parameters (fault length, width and seismic moment) of future earthquakes can be
estimated by the seismic history, geological investigation and source modeling of active faults near
the construction site. The local source parameters (slip heterogeneity on fault plane) are important to
characterize the fault movement along the slip plane, especially the slip and slip velocity.

The local parameters, such as the transmission characteristics of earthquake motion from
earthquake source to the construction site, cannot be evaluated theoretically, but must be
determined by the source inversion of past major events with the use of statistical analysis or by the
observation of minor earthquakes. The transfer function from the source to the construction site may
be estimated by an empirical Green’s function; i.e., the transfer function of past small earthquakes in
the region. The intensity of earthquake wave decays with distance.

The earthquake wave is generally transmitted from the fracture fault through hard rock layers and
then to the ground surface through relatively soft surface soil. The characteristics of ground motion
are significantly modified by the properties of surface soil layers, such as the properties and
15
geometry of the subsurface soil layers, surface topography, and depth and properties of the
underlying bedrock. Soft soil layers consisting of river deposits tend to amplify long period
components of an earthquake motion, causing serious damage to houses and buildings.

It is important to recognize that seismic design of a structure is based on a large uncertainty

about the characteristics, especially intensity, of a design earthquake motion.

Reference:

Architectural Institute of Japan: Seismic Loading - State of the Art and Future Developments (in
Japanese), November 1988, 438 pp.

16
 Chapter 6. Numerical Integration of Equation of Motion

6.1 Introduction

The equation of motion of a structure (dynamic equilibrium of horizontal forces) under horizontal
earthquake motion may be expressed;
d2 d2
[M ] { x ( t i )} + { D ( t i )} + { R ( t i )} = −[ M ]{e} y (ti )
dt 2 dt 2
[ M ]{&x&}i + {D}i + {R}i = −[ M ]{e} &y&i
d2
in which [M]: mass matrix, {x (ti )} = {&x&}i : acceleration vector relative to the structure’s base at
dt 2
time ti, {D (t i ) = {D}i : damping force vector at time ti, {R (t i )} = {R} i : resistance vector at time ti,
d2
y (ti ) = &y&i : horizontal ground acceleration at time ti, {e} : vector with elements equal to 1.0 for
dt 2
horizontal degrees of freedom and zero for the rest of degrees of freedom.

Assuming that resistance and damping force vary linearly with displacement and velocity relative
to the base, respectively, over a short time increment ∆t from t i to t i +1 , the equation of motion at
the new time step t i +1 may be written;
x}i +1 + {D}i +1 + {R}i +1 = −[ M ]{e}{&&
[ M ]{&& y}i +1
x}i +1 + {D}i + {∆D}i +1 + {R}i + {∆R}i +1 = −[ M ]{e}{&&
[ M ]{&& y}i +1
or expressing incremental resistance {∆R}i +1 as the product of instantaneous stiffness [ K ]i at
time ti and incremental displacement {∆x}i +1 , and incremental damping force {∆D}i +1 as the
product of instantaneous damping matrix [C ]i at time ti and incremental velocity {∆x&}i +1 ;
x}i +1 + {D}i + [C ]{∆x&}i +1 + {R}i + [ K ]{∆x}i +1 = −[ M ]{e}{ &&
[ M ]{&& y}i +1
x}i +1 + [C ]{∆x&}i +1 + [ K ]{∆x}i +1 = −[ M ]{e}{ &&
[ M ]{&& y}i +1 − {R}i − {D}i
{&x&}i +1 , {∆x&}i +1 and {∆x}i +1 are solved by a numerical integration procedure and numerical
solution of a set of linear algebraic equations.

It is important to evaluate {R}i +1 and

Ri+1
{D}i +1 on the basis of calculated displacement
{x}i +1 and velocity {x&}i +1 at time t i +1
considering nonlinear stiffness and damping
K ∆xi +1
characteristics. The damping force and
resistance should not be calculated as ∆Ri +1
{R}i +1 = {R}i + [ K ]{∆x}i +1
{D}i +1 = {D}i + [C}{∆x&}i +1 Ri
because the stiffness and damping may not be
proportional to incremental velocity and
displacement during the time increment ∆ti +1 ∆xi +1
xi xi+1
due to nonlinearity.

1
6.2 Nigam-Jennings' Direct Integration Method

Consider a linearly elastic single-degree-of-freedom (SDF) system, with mass m, damping

coefficient c and stiffness k, subjected to linearly varying load p(t):
d 2x dx
m 2
+c + kx = p (t )
dt dt
p(t)
m&x& + cx& + kx = p = at + b
The solution of a differential equation is expressed as
the sum of a particular solution xp(t), which satisfies
the original differential equation, and a complementary a
solution xc(t), which is a general solution of the b
differential equation for right hand function to be zero.
There is no systematic way to search for a particular t
solution; the complementary solution is given as the
solution for free vibration.
mx&&p + cx& p + kx p = at + b
mx&&c + cx&c + kxc = 0
A particular solution x p for this differential equation is assumed to be of the following form;
x p = c1t + c2
then, differentiating the relation with time,
x& p = c1
&x&p = 0
The above relations are substituted into the equation of motion to determine the constants c1 and c2
of the assumed particular solution:
cc1 + k ( c1t + c2 ) = at + b
or
( kc1 − a )t + ( cc1 + kc2 − b) = 0
The particular solution must satisfy the differential equation for any time t; hence
( kc1 − a ) = 0
( cc1 + kc2 − b) = 0
Solving for c1 and c2, we obtain
a
c1 =
k
a
c2 = ( b − c ) / k
k
or
a bk − ac
xp = t+
k k2

The complementary solution is given as

xc = e − hωnt ( A cos ω d t + B sin ω d t )
where
ωd = ωn 1 − h2
k
ωn =
m
c
h=
2 mk
2
ω d : damped circular frequency, ω n : undamped circular frequency, h: damping factor. A and B are
integration constants dependent on the initial condition.

Therefore, the complete solution x(t) for the loading is

x = x p + xc
a bk − ac
= t+ 2
+ e −hωnt ( A cos ω d t + B sin ω d t )
k k

The velocity is given by differentiating the displacement with time;

a
x& = + e −hωnt {− hω n ( A cos ω d t + B sin ω d t ) + ω d ( − A sin ω d t + B cos ω d t )}
k

For given initial displacement xo and velocity vo at t = 0,

bk − ac
xo = A +
k2
a
vo = − hω n A + ω d B
k
The integration constants A and B are solved from the initial conditions;
bk − ac
A = xo −
k2
1 bk − ac a
B= ( hω n xo + vo − hω n − )
ωd k2 k
Namely, the response at any time t (> 0) is expressed as a function of the system constants h, k, ωn
and ω d , the loading function p(t)=a t + b, and the initial conditions xo and vo at time t = 0.

In general, if a linearly varying loading function is given from t=ti to t=ti+1 as

t − ti
p(t ) = pi + ( pi +1 − pi )
ti +1 − ti
∆pi +1
= pi + ( t − ti )
∆ti +1
= ai +1τ + bi +1
where
∆pi +1
ai +1 =
∆ti +1
bi +1 = pi
τ = t − ti
The response at any time t ( t i < t < t i +1 , 0 < τ < ∆t i +1 ) can be expressed by known initial conditions
xi and x& i at time t = t i ( τ = 0.0). Therefore, the response velocity x& i +1 and displacement xi+1 at
time ti+1 are determined.

3
 ∆pi +1 ∆ti +1 1 ∆p
xi +1 = + 2 ( pi k − i +1 c ) + e −hωn ∆ti +1 ( A cos ω d ∆ti +1 + B sin ω d ∆ti +1 )
∆ti +1 k k ∆ti +1
1 ∆p c
= ( pi +1 − i +1 ) + e −hωn ∆ti +1 ( A cos ω d ∆ti +1 + B sin ω d ∆ti +1 )
k ∆ti +1 k
1 ∆pi +1
x& i +1 = + e −hωn ∆ti +1 {− hω n ( A cos ω d ∆ti +1 + B sin ω d ∆ti +1 )
k ∆ti +1
+ ω d ( − A sin ω d ∆ti +1 + B cos ω d ∆ti +1 )}
1 ∆p
A = xi − 2 ( pi k − i +1 c )
k ∆ti +1
1 hω n ∆p 1 ∆pi +1
B= {hω n xi + x& i − ( pi k − c i +1 ) − }
ωd k 2
∆ti +1 k ∆ti +1
The procedure may be successively applied as long as the loading function is given in a piece-wise
linear form over a short time increment.

The response can be calculated in the following form:

xi +1 = A1 xi + A2 x& i + A3 pi +1 + A4 pi
x& i +1 = B1 xi + B2 x& i + B3 pi +1 + B4 pi
p − cx& i +1 − kxi +1
&x&i +1 = i +1
m
where,
h
A1 = e − hω n ∆t { sin ω d ∆t + cos ω d ∆t}
1 − h2
1
A2 = e − hω n ∆t sin ω d ∆t
ωd
1 2h 2 − 1 2h 1 2h
A3 = e − hω n ∆t { sin ω d ∆t + cos ω d ∆t} + 2 {1 − }
ω n ∆t ω d
2
ωn ωn ω n ∆t
1 h 2h 2 − 1 2h 2h
A4 = e − hω n ∆t {( + ) sin ω d ∆t + (1 + ) cos ω d ∆t} + 3
ωn 2
1 − h2 ω d ∆t ω n ∆t ω n ∆t
ωn
B1 = −e − hω n ∆t sin ω d ∆t
1 − h2
h
B2 = e − hω n ∆t (cos ω d ∆t − sin ω d ∆t )
1 − h2
2h 2 − 1 h
B3 = e − hω n ∆t { (cosω d ∆t − sin ω d ∆t )
ω n ∆t
2
1 − h2
2h 1
− 3 (ω d sin ω d ∆t + hω n cos ω d ∆t )} + 2
ω n ∆t ω n ∆t
1 2h
B4 = e − hω n ∆t {( 2 + 3 )(ω d sin ω d ∆t + hω n cos ω d ∆t )
ω n ω n ∆t
h2h 2 − 1 h 1
− ( + 2 )(cosω d ∆t − sin ω d ∆t )} − 2
ω n ω n ∆t 1− h 2 ω n ∆t

A1 = E (H S + C)
4
 A2 = E S / ω d
A3 = E (1 / k)(H1 S + H2 C) + (1 - H2) / k
A4 = - E [(H + H1) S + (1 + H2) C] / k + H2 / k
B1 = - E ωn S / 1 − h2
B2 = E (C - H S)
B3 = E {H1( ω d C - h ω n S) - H2( ω d S + h ω n C)} / k + 1 / (k ∆t )
B4 = E [(1 + H2)( ω d S + h ω n C) - (H + H1)( ω d C - h ω n S)} / k - 1 / (k ∆t )
where
C = cos ω d ∆t
S = sin ω d ∆t
E = e - h ω n ∆t

H=h/ 1 − h2
H1 = (2h2 -1) / ( ω d ∆t )
H2 = 2h / ( ω n ∆t )

The coefficients Ai and Bi (i= 1, 2, 3, and 4) can be made constant if time increment ∆t of
numerical integration is fixed and system properties m, c, and k do not change with time. In such a
case, the computing time may be significantly reduced.

Not that this procedure is exact if the exciting function is given as a series of piece-wise linear
functions.

Therefore, this procedure is most desirable for the response analysis of a linearly elastic
single-degree-of-freedom system. However, in the case of a nonlinear problem, the coefficients As
and Bs must be evaluated whenever the tangent stiffness k and damping coefficient c are altered.

Reference:

Nigam, N.C., and P. C. Jennings, "Calculation of Response Spectra from Strong-Motion Earthquake
Records," Bulletin, Seismological Society of America, Vol. 59, No. 2, April 1969, pp. 909 - 922.

5
6.3 Linear Acceleration Method

The Taylor series expansion of a vector function {x} with respect to time t about time to is
expressed as
dx (t − t o ) 2 d 2 x (t − t o ) 3 d 3 x
{x(t )} = {x(t o )} + (t − t o ){ }t =to + { 2 }t = t o + { 3 }t = t o + L
dt 2! dt 3! dt
(t − t o ) d x
k k
=∑ { k }t =to
k =0 k! dt
The accuracy of the expression will be improved by considering more terms and by evaluating the
function near to.

The displacement and velocity are expanded by the Taylor series at time ti to evaluate the
functions at time ti+1;
∆t 2 ∆t 3
{x}i +1 = {x}i + ∆t{x&}i + {&x&}i + {&x&&}i + L
2 6
∆t 2
{x}i +1 = {x}i + ∆t{x}i +
& & && {&x&&}i + L
2
where ∆t = ti +1 − ti .

If the acceleration is assumed to vary linearly over a short time increment ∆t ;

x}i +1 − {&&
{&& x}i
x }i =
{&&&
∆t
The following relations are obtained for displacement and velocity increments;
∆t 2 ∆t 2
{∆x}i +1 = {x}i +1 − {x}i = ∆t{x&}i + {&x&}i + {&x&}i +1
3 6
∆t ∆t
{∆x&}i +1 = {x&}i +1 − {x&}i = {&x&}i + {&x&}i +1
2 2

The equation of motion at time ti+1 may be solved for {&x&}i +1 , {∆x&}i +1 and {∆x}i +1 ;
[ M ]{&x&}i +1 + [C ]{∆x&}i +1 + [ K ]{∆x}i +1 = −[ M ]{e} &y&i +1 − {D}i − {R}i

Using the integration relation and the equation of motion, acceleration {&&
x}i +1 at new time step ti+1
can be solved
∆t ∆t 2 ∆t ∆t 2
([m] + [c ] + x}i +1 = { p}i +1 − [k ]{x}i − ([c] + ∆t[k ]){x&}i − ( [c] +
[k ]){&& [k ]){&&
x}i
2 6 2 3
{&x&}i +1 = [ A1 ]{ p}i +1 + [ A2 ]{&x&}i + [ A3 ]{x&}i + [ A4 ]{x}i
where,
∆t ∆t 2
[ A1 ] = ([m] + [c ] + [k ]) −1
2 6
∆t ∆t 2
[ A2 ] = −[ A1 ]( [c] + [k ])
2 3
[ A3 ] = −[ A1 ]([c] + ∆t[k ])
[ A4 ] = −[ A1 ][k ]

The displacement and velocity are calculated using the linear acceleration procedure;

6
 1 1
{x}i +1 = {x}i + ∆t{x&}i + ∆t 2 {&x&}i + ∆t 2 {&x&}i +1
3 6
1 1
{x&}i +1 = {x&}i + ∆t{&x&}i + ∆t{&x&}i +1
2 2

The time increment ∆t for the response analysis must be chosen to satisfy the following
conditions;
(a) Excitation function, hysteresis relations, and response waveforms can be expressed
with a satisfactory accuracy,
(b) The accuracy of response results can be attained, and
(c) The numerical integration gives stable results.

The solution of the linear acceleration method diverges if the time increment is selected larger
than 1/3 of the shortest modal period of oscillation of the system.

7
6.4 Newmark Beta Method

For a set of differential equation of motion given below,

[ M ]{&x&}i +1 + [C ]{∆x&}i +1 + [ K ]{∆x}i +1 = −[ M ]{e} &y&i +1 − {D}i − {R}i
Newmark (1959) suggested the following relations for the numerical integration;
1
{∆x}i +1 = {x}i +1 − {x}i = ∆t{x&}i + ( − β ) ∆t 2 {&x&}i + β∆t 2 {&x&}i +1
2
{∆x&}i +1 = {x&}i +1 − {x&}i = (1 − γ ) ∆t{&x&}i + γ∆t{&x&}i +1
where β and γ are constants of the Newmark Beta method.

The value of γ must be 1/2, and β ≤ 1/ 4 . If the value of γ is selected to be greater than 1/2, the
response amplitude becomes greater than the true value; if the value is smaller than 1/2, then the
amplitude becomes smaller.

For the numerical integration to give stable results, the time increment ∆t must be less than
one-sixth of the shortest modal period of the system for β = 1/6. The scheme is known to be
unconditionally stable for β = 1/4.

The equation motion can be solved for the acceleration {&x&}i +1 at the new time step;
{[ M ] + γ∆t[C ] + β∆t 2 [ K ]){&x&}i +1
1
= { p}i +1 − {D}i − {R}i − (1 − γ ) ∆t[C ]{&x&}i − ∆t[ K ]{x&}i − ( − β ) ∆t 2 [ K ]{&x&}i
2
Newmark suggested to solve the set of equations by an iteration method;
*
(a) Assume acceleration vector {&x&}i +1 at time ti+1,
(b) Evaluate the displacement increment {∆x}i +1 and velocity increment {∆x&}i +1 by the
* *

following relations;
1
{∆x}*i +1 = ∆t{x&}i + ( − β ) ∆t 2 {&x&}i + β∆t 2 {&x&}*i +1
2
{∆x&}i +1 = (1 − γ ) ∆t{&x&}i + γ∆t{&x&}*i +1
*

{x}*i +1 = {x}i + {∆x}*i +1

{x&}*i +1 = {x&}i + {∆x&}*i +1
*
(c) Evaluate damping force {D}*i+1 and resistance {R}*i+1 for the calculated velocity {x&}i +1
*
and displacement {x}i +1 on the basis of damping and hysteresis models;
(d) Re-evaluate the acceleration {&x&}i +1 by the equation of motion at time ti+1;
[ M ]{&x&}i +1 = −[ M ]{e} &y&i +1 − {D}i +1 − {R}i +1
*
(e) If the re-evaluated acceleration {&x&}i +1 differs from the assumed acceleration {&x&}i +1 by
more than a specified tolerance {ε } , the assumed acceleration {&x&}i +1 is replaced by the
*

re-evaluated acceleration {&x&}i +1 in step (a), and the procedure is iterated until a satisfactory
conversion is achieved.

Sharpe and Carr (1974) studied a condition for numerical stability as follows;
derived acceleration − true acceleration
≤ 1.0
assumed acceleration − true acceleration
at the end of each time step using Newmark Beta scheme. They derived a convergence criteria as

8
follows;
∆t 1 h h 1
< [− + ( )2 + ]
T 2π 2 β 2β β

Newmark (1959) showed the convergence criteria for an undamped linear system as follows;
∆t 1 1
≤
T 2π β
1
For β = equivalent to the linear acceleration scheme,
6
∆t
≤ 0.39
T

Sharpe and Carr (1974) extended Newmark’s derivation of convergence for undamped system to
damped system. The criterion is shown below;
∆t 1 − h2
<
T π 1 − 4β

References

Newmark, N. M., "A Method of Computation for Structural Dynamics," Journal, Engineering
Mechanics Division, ASCE, July 1959, pp. 67 - 94.
Sharpe, R. D., and A. J. Carr, "The Seismic Response of Inelastic Structures," Research Report
74-13, Department of Civil Engineering, University of Canterbury, New Zealand, November
1974.

9
6.5 Wilson Theta Method

When the finite element method was applied in a

structural dynamic problem, a major problem was the
p(t)
computation time because the finite element analysis deals
a system having a large number of degrees of freedom. In
order to reduce the computation time, a stable numerical
method allowing the use of a large time increment ∆t is
desired. Wilson (Wilson, 1968, Bathe and Wilson, 1973)
proposed a stable numerical integration scheme. t

A unique feature of this method is to satisfy the ti ∆ti +1 ti +1 ti*+1

equation of motion at an imaginary time ti +1 = ti + θ ∆t
*
θ ∆t
( θ > 1.0) for an extrapolated imaginary force {p}*i+1
{ p}*i +1 = { p}i + [{ p}i +1 − { p}i ]θ
Using the Newmark Beta method, the equation of motion at time t=t*i+1 is given as
{[ M ] + γ θ ∆t[C ] + β (θ ∆t ) 2 [ K ]){&&
x}i +1
1
= { p}i +1 − {D}i − {R}i − (1 − γ ) θ ∆t[C ]{&&
x}i − θ ∆t[ K ]{x&}i − ( − β )(θ ∆t ) 2 [ K ]{&&
x}i
2
*
The relationship may be solved for {&x&}i +1 , and the acceleration at time ti+1 is determined by the
interpolation (assuming linear variation of acceleration over time increment);
1 1
{&x&}i +1 = {&x&}*i +1 + (1 − ){&x&}i
θ θ
Velocity {x&}i +1 and displacement {x}i +1 at time t + ∆t are evaluated by the Newmark Beta method;
1
{x}i +1 = {x}i + ∆t{x&}i + ( − β ) ∆t 2 {&x&}i + β∆t 2 {&x&}i +1
2
{x&}i +1 = {x&}i + (1 − γ ) ∆t{&x&}i + γ∆t{&x&}i +1

The value of θ was suggested to be greater than 1.37 for numerical stability; θ = 1.37 to 1.40
is often used.

The Wilson's Theta method is known to introduce numerical damping, a fictitious damping caused
by the numerical integration scheme especially for higher mode (short period) oscillation; the period
of oscillation is elongated and response amplitude decays with time (Bathe and Wilson, 1973). It is
believed that some error may be tolerated in higher mode oscillation because the higher mode
response may not govern the total response in a normal case.

Stability and Accuracy: It is important in selecting numerical integration scheme to examine the
stability of the procedure. Some scheme is stable when the time increment is selected less than a
certain fraction of the period of the highest mode; otherwise, error is continuously amplified during
the numerical integration and the response diverges with time. The stability of a numerical
integration scheme may be examined for a linearly elastic single-degree-of-freedom system without
any damping under free vibration;
x + ω n2 x = 0.0
&&
The error can be expressed in terms of period elongation and amplitude decay as a function of time
increment ∆t over natural period T .

The accuracy of numerical integration is another problem. Although a numerical integration

scheme may be unconditionally stable for the choice of time increment, the result may not be
accurate if the time increment is large with respect to the natural period. Such inaccuracy may be,

10
sometimes, tolerated for higher mode response because the higher mode response is often
negligibly small compared with the dominant response.

11
 It is often said that “numerical damping can be considered a good feature in numerical schemes
because it may be used to damp out and practically suppress the response of those modes for
which the response cannot be calculated accurately.” However, the range of errors should be
estimated.

A six-story two-bay linearly elastic frame was analyzed using the Wilson's Theta method and the
Newmark Beta method (Sharpe and Carr, 1974). Vertical and horizontal masses were considered at
each node. The stability criterion for an analysis using the linear acceleration technique requires a
time step of approximately 1/400 sec for this structure. Beta value of the Newmark Beta method was
varied from 1/12 ( ∆t = 1/400 sec) to 1/4 ( ∆t = 1/100 sec), and theta value of the Wilson's Theta
method was selected to be 1.5 and 2.0 with a time increment of ∆t =1/100 sec.

The top story displacement waveforms are compared. The response waveforms calculated by the
Newmark Beta method were almost identical, while the waveforms calculated by the Wilson Theta
method showed a difference from the response waveforms calculated by the Newmark Beta
methods. This particular response was dominated by the first mode component, but the appreciable
difference can be observed. If the oscillation is governed by higher frequency components such as
the acceleration waveform, then the effect of numerical (artificial) damping would appear more in the
calculated waveform.

It should be noted that the role of the analytical tool is to give results as close to the exact solution
as possible. In this standpoint, the Wilson Theta method does not satisfy the criteria.

References:

Bathe, K.-J., and E. L. Wilson, "Stability and Accuracy Analysis of Direct Integration Methods,"
International Journal of Earthquake Engineering and Structural Dynamics, Vol. 1, 1973, pp. 283
- 291.
Bathe, K.-J., and E. L. Wilson, "Linear and Nonlinear Earthquake Analysis of Complex Structures,"
Proceedings, World Conference on Earthquake Engineering, Rome, Italy, 1973, Paper No. 224.

12
Hilber, H.M., et al., “Improved Numerical Dissipation for Time Integration Algorithms in Structural
Dynamics,” Earthquake Engineering & Structural Dynamics, No. 5, 1977, pp. 283 - 292.
Sharpe, R. D., and A. J. Carr, "The Seismic Response of Inelastic Structures," Research Report
74-13, Department of Civil Engineering, University of Canterbury, New Zealand, November
1974.
Wilson, E. L., "A Computer Program for the Dynamic Stress Analysis of Underground Structures,"
SESM Report 68-1, Department of Civil Engineering, University of California, Berkeley, 1968.

13
6.6 Runge-Kutta-Gill Method (Fourth Order)

A procedure was developed by Gill for the numerical integration. For a differential equation of the
first order,
dx
= F ( x, t )
dt
the initial value problem is solved in a form
f 0 + 2 f1 + 2 f 2 + f 3
xi +1 = xi +
6
where,
f 0 = ∆tF ( xi , ti )
f0 ∆t
f1 = ∆tF ( xi + , ti + )
2 2
f1 ∆t
f 2 = ∆tF ( xi + , ti + )
2 2
f 3 = ∆tF ( xi + f 2 , ti + ∆t )

This integration scheme is extended for the ordinary differential equation of the second order;
{F ( x, x& , t )} = {x&}
{G ( x, x& , t )} = [ M ]−1 ({ p(t )} − [C ]{x&} − [ K ]{x})
then,
{ f 0 } + 2{ f1} + 2{ f 2 } + { f 3 }
{x}i +1 = {x}i +
6
{g } + 2{g1} + 2{g 2 } + {g 3 }
{x&}i +1 = {x&}i + 0
6
where,
{ f 0 } = ∆t{F ( xi , x& i , ti )} = ∆t{x&}i
f0 g ∆t {g }
{ f1} = ∆t{F ( xi + , x&i + 0 , ti + )} = ∆t ({x&}i + 0 )
2 2 2 2
f1 g1 ∆t { g1 }
{ f 2 } = ∆t{F ( xi + , x& i + , ti + )} = ∆t ({x&}i + )
2 2 2 2
{ f 3 } = ∆t{F ( xi + f 2 , x& i + g 2 , ti + ∆t )} = ∆t ({x&}i + {g 2 })

{g 0 } = ∆t{G ( xi , x& i , ti )}
= ∆t[ M ]−1 ({ p}i − [C ]{x&}i − [ K ]{x}i )
f0 g ∆t
{g1} = ∆t{G ( xi + , x& i + 0 , ti + )}
2 2 2
∆t g f0
= ∆t[ M ]−1 ({ p(ti + } − [C ]{x& + 0 }i − [ K ]{x + }i )
2 2 2
f1 g1 ∆t
{g 2 } = ∆t{G ( xi + , x& i + , ti + )}
2 2 2
∆t g f1
= ∆t[ M ]−1 ({ p(ti + } − [C ]{x& + 1 }i − [ K ]{x + }i )
2 2 2
{g 3 ) = ∆t{G ( xi + f 2 , x& i + g 2 , ti + ∆t )}
= ∆t[ M ]−1 ({ p}i +1 − [C ]{x& + g 2 }i − [ K ]{x + f 2 }i )

14
Reference:

Gill, S., "A Procedure for the Step-by-step Integration of Differential Equations in an Automatic
Computing Machine," Proceedings, Cambridge Philosophical Society, 49:96, 1951.

15
Appendix: Linearly elastic response of SDF System under earthquake motion (FORTRAN）

C
C RESPONSE OF A SINGLE-DEGREE-OF-FREEDOM LINEARLY ELASTIC SYSTEM
C TO EARTHQUAKE MOTION
C
C PROGRAMMED BY OTANI, S.
C ON OCTOBER 28, 1975
C AT UNIVERSITY OF TORONTO
C MODIFIED BY OTANI, S.
C ON APRIL 18, 1993
C AT UNIVERSITY OF TOKYO
C MODIFIED BY OTANI, S.
C ON MAY 26, 2000
C AT UNIVERSITY OF TOKYO
C
C THE EQUATION OF MOTION OF SDF SYSTEM UNDER GROUND MOTION
C
C DDX + 2.0 * BT * WN * DX + WN * WN * X = - DDY
C
C INPUT DATA
C 1. TITL
C TITL: TITLE OF STUDY
C 2. LUNT,TUNT
C LUNT: UNIT OF LENGTH
C TUNT: UNIT OF TIME
C 3. TN,BT,DT,ASCL,V0,D0
C TN: PERIOD
C BT: DAMPING RATIO
C DT: TIME INTERVAL OF NUMERICAL INTEGRATION
C ASC: SCALE FACTOR FOR GROUND ACCELERATION
C V0: INITIAL VELOCITY
C D0: INITIAL DISPLACEMENT
C 4. NSTT,NSTP,MTHD
C NSTT: STARTING POINT OF EARTHQUAKE DATA
C NSTP: TERMINATING POINT OF EARTHQUAKE DATA
C MTHD: POINTER FOR NUMBERICAL INTEGRATION METHOD
C 5. EQNM
C EQNM: NAME OF EARTHQUAKE RECORD
C 6. TIME,GACC
C TIME: TIME COORDINATE OF EARTHQUAKE DATA
C GACC: GROUND ACCELERATION OF EARTHQUAKE DATA
C
CHARACTER*4 TITLE(18)*4,EQNM(18)*4,LUNT*4,TUNT*4
DIMENSION TIME(2),GACC(2)
C
READ (5,500) TITL
READ (5,500) LUNT,TUNT
C
READ (5,*) TN,BT,DT,ASCL,V0,D0
C TN UNDAMPED NATURAL PERIOD
IF (TN.LE.0.0) GO TO 60
C BT DAMPING FACTOR
IF (BT.LT.0.0) BT=0.0
C DT TIME INCREMENT OF NUMERICAL INTEGRATION
IF (DT.LE.0.0) DT=0.05*TN
C ASCL SCALE FACTOR FOR GROUND ACCELERATION DATA
IF (ASCL.LE.0.0) ASCL=1.0
C FN UNDAMPED NATURAL FREQUENCY
FN=1.0/TN
WN=6.2830*FN
BW=2.0*BT*WN
W2=WN*WN
C PRINT INPUT INFORMATION
WRITE (6,600) TITL,LUNT,TUNT
WRITE (6,602) TN,TUNT,FN,TUNT,WN,TUNT,BT,DT,TUNT,ASCL,
1 V0,LUNT,TUNT,D0,LUNT

16
C
READ (5,*) NSTT,NSTP,MTHD
IF (NSTT.LE.0) NSTT=1
IF (NSTP.LE.NSTT) NSTP=NSTT
IF (MTHD.LE.0) MTHD=1
IF (MTHD.GT.4) MTHD=1
WRITE (6,604) NSTT,NSTP
C
C CALCULATION OF NUMERICAL INTEGRATION CONSTANTS
C
GO TO (1,2,3,4),MTHD
1 CALL JNNGS1 (DT,WN,BT,A1,A2,A3,A4,B1,B2,B3,B4,&9)
2 CALL LNACC1 (DT,WN,BT,A1,A2,A3,A4,B1,B2,B3,&9)
3 CALL NWMRK1 (DT,WN,BT,A1,A2,A3,A4,&9)
4 CALL RKGIL1 (DT,A1,&9)
C
C EARTHQUAKE RECORD
C
9 READ (5,500) EQNM
WRITE (6,606) EQNM
C
C INITIAL CONDITIONS
C
ISTP=0
IST=0
NST=0
GACC(2)=0.0
TIME(2)=0.0
TT=0.0
C
10 TIME(1)=TIME(2)
GACC(1)=GACC(2)
READ (5,*) TIME(2),GACC(2)
NST=NST+1
IF (NST.GT.NSTT) GO TO 20
IF (NST.LT.NSTT) GO TO 10
GA=GACC(2)*ASCL
A0=-(GA+BT*V0+W2*D0)
TIM=TIME(2)
CALL PRINT (IST,TIM,DT,A0,V0,D0,GA,LUNT,TUNT)
GO TO 10
C
20 IF (NST.GT.NSTP) GO TO 60
DLT=TIME(2)-TIME(1)
IF (DLT) 60,10,30
C
30 GR=GACC(1)*ASCL
SLP=(GACC(2)-GACC(1))*ASCL/DLT
TI=-TT
TT=TT+DLT
C
38 IF (DT.GT.TT) GO TO 10
TT=TT-DT
TI=TI+DT
G0=GA
GA=GR+SLP*TI
C
C NUMERICAL INTEGRATION
C PREVIOUS RESPONSE VALUE (A0,V0,D0) MUST BE REPLACED BY
C NEW RESPONSE VALUES (A0,V0,D0)
C A0: ACCELERATION RELATIVE TO GROUND
C V0: VELOCITY RELATIVE TO GROUND
C D0: DISPLACEMENT RELATIVE TO GROUND
C GA: GROUND ACCELERATION
C
40 GO TO (41,42,43,44),MTHD
41 CALL JNNGS2 (A0,V0,D0,G0,GA,W2,BW,A1,A2,A3,A4,B1,B2,B3,B4,&49)

17
 42 CALL LNACC2 (A0,V0,D0,GA,DT,A1,A2,A3,A4,B1,B2,B3,&49)
43 CALL NWMRK2 (A0,V0,D0,GA,DT,W2,BW,A1,A2,A3,A4,&49)
44 CALL RKGIL2 (A0,V0,D0,G0,GA,DT,W2,BW,A1,&49)
C
49 IST=IST+1
CALL PRINT (IST,TIM,DT,A0,V0,D0,GA,LUNT,TUNT)
GO TO 38
C END OF RESPONSE CALCULATION
60 WRITE (6,610)

STOP
500 FORMAT (18A4)
600 FORMAT (18A4,/,
1 "UNITS USED IN CALCULATION",/,
2 5X,"LENGTH = ",A4,/,
4 5X,"TIME = ",A4,/)
602 FORMAT ("SYSTEM PROPERTIES",/,
1 5X,"UNDAMPED PERIOD = ",F10.3," (",A4,")",/,
2 5X,"UNDAMPED FREQENCY = ",F10.3," (1/",A4,")",/,
3 5X,"UNDAMPED CIRCULAR FREQ. = ",F10.3," (rad/",A4,")",/,
4 5X,"DAMPING FACTOR = ",F10.3,/,
5 "CONSTANTS FOR NUMERICAL INTEGRATION",/,
6 5X,"TIME INTERVAL = ",F10.3," (",A4,")",/,
7 5X,"ACCELERATION SCALE FACTOR = ",F10.3,/,
8 "INITIAL CONDITIONS",/,
9 5X,"VELOCITY = ",1PE10.3," (",A4,"/",A4,")",/,
1 5X,"DISPLACEMENT = ",E10.3," (",A4,")",/)
604 FORMAT ("START COMPUTATION = ",I5," POINTS",/,
1 "END COMPUTATION = ",I5," POINTS",/)
606 FORMAT ("EARTHQUAKE RECORD ",/,
1 5X,"NAME: ",18A4,/)
610 FORMAT ("ALL DATA PROCESSED")
END
SUBROUTINE PRINT (IST,TIM,DT,A0,V0,D0,GA,LUNT,TUNT)
C
C PRINT RESPONSE VALUES
C
C IST: STEP NUMBER
C TIM: TIME AT FIRST RESPONSE CALCULATION
C DT: TIME STEP
C A0: RESPONSE ACCELERATION RELATIVE TO GROUND
C V0: RESPONSE VELOCITY RELATIVE TO GROUND
C D0: RESPONSE DISPLACEMENT RELATIVE TO GROUND
C GA: GROUND ACCELERATION
C
IF (IST.GT.0) GO TO 10
C FORMAT FOR PRINTING RESPONSE
WRITE (6,600) TUNT,LUNT,TUNT,LUNT,TUNT,
1 LUNT,TUNT,LUNT,LUNT,TUNT
10 T=TIM+DT*FLOAT(IST)
C ABSOLUTE ACCELERATION RESPONSE
A=A0+GA
WRITE (6,602) IST,T,GA,A0,V0,D0,A
600 FORMAT (" STEP"," TIME "," GR. ACC. "," REL. ACC. ",
1 " VELOCITY ","DISPLACEMENT"," ABS. ACC. ",/,
2 3X," (",A4,") ","(",A4,"/",A4,"**2)",
3 "(",A4,"/",A4,"**2)","(",A4,"/",A4,")",
4 " (",A4,") ","(",A4,"/",A4,"**2)",/)
602 FORMAT (I5,F8.3,5(2X,1PE10.3))
RETURN
END
SUBROUTINE JNNGS1 (DT,WN,BT,A1,A2,A3,A4,B1,B2,B3,B4,*)
C
C CALCULATION OF NUMERICAL CONSTANTS IN JENNINGS-NIGAM METHOD
C

C INPUT

18
C DT TIME INTERVAL OF NUMERICAL INTEGRATION
C WN UNDAMPED CIRCULAR FREQUENCY
C BT DAMPING FACTOR
C
C FUNCTIONS USED IN THE SUBROUTINE
C SQRT
C EXP
C SIN
C COS
C
DW=WN*DT
EX=EXP(-BT*DW)
SB=SQRT(1.0-BT*BT)
BS=BT/SB
WD=WN*SB
SW=SIN(WD*DT)
CW=COS(WD*DT)
B2=2.0*BT*BT-1.0
W2=1.0/WN**2
TB=B2/SB
WT=W2/DW
A1=EX*(BS*SW+CW)
A2=EX*SW/WD
A3=(EX*(TB*SW+2.0*BT*CW)-BT-BT)*WT+W2
A4=-EX*((BS+TB/DW)*SW+(1.0+2.0*BT/DW)*CW)*W2+2.0*BT*WT
B1=-EX*WN*SW/SB
B2= EX*(CW-BS*SW)
B3=(-EX*(CW+BS*SW)+1.0)*W2/DT
B4= EX*SW/WD+(EX*(BS*SW+CW)-1.0)*W2/DT
WRITE (6,600)
RETURN 1
600 FORMAT ("JENNINGS-NIGAM METHOD",/)
END
SUBROUTINE LNACC1 (DT,WN,BT,A1,A2,A3,A4,B1,B2,B3,*)
C
C CALCULATION OF NUMERICAL CONSTANTS FOR LINEAR ACCELERATION METHOD.
C
C INPUT
C DT TIME INTEVAL OF NUMERICAL INTEGRATION
C WN UNDAMPED CIRCULAR FREQUENCY
C BT DAMPING FACTOR
C
B1=0.5*DT
B2=DT*DT/3.0
B3=B2*0.5
A=1.0+BT*WN*DT+WN*WN*B3
A1=-1.0/A
A2=-WN*DT*(BT+WN*DT/3.0)/A
A3=-WN*(BT+BT+WN*DT)/A
A4=-WN*WN/A
WRITE (6,600)
RETURN 1
600 FORMAT ("LINEAR ACCELERATION METHOD",/)
END
SUBROUTINE NWMRK1 (DT,WN,BT,A1,A2,A3,A4,*)
C
C CALCULATION OF NUMERICAL CONSTANTS FOR NEWMARK BETA METHOD
C
C NIPUT
C DT TIME INTERVAL OF NUMERICAL INTEGRATION
C WN UNDAMPED CIRCULAR FREQUENCY
C BT DAMPING FACTOR
C
BETA=1.0/6.0
C
C BETA CONSTANT FOR NEWMARK BETA METHOD
C

19
 A1=0.5*DT
A2=(0.5-BETA)*DT**2
A3=BETA*DT**2
A4=0.0005
WRITE (6,600) BETA, A4
RETURN 1
600 FORMAT ("NEWMARK BETA METHOD", /,
1 5X,"BETA = ",F10.3, /,
2 5X,"ERROR LIMIT = ",1PE10.3,/)
END
SUBROUTINE RKGIL1 (DT,A1,*)
C
C CALCULATION OF A NUMERICAL CONSTANT FOR RUNGE-KUTTA-GILL METHOD.
C
C INPUT
C DT TIME INTERVAL OF NUMERICAL INTEGRATION
C
A1=0.5*DT
WRITE (6,600)
RETURN 1
600 FORMAT ("RUNGE-KUTTA-GILL MEHOTD",/)
END
SUBROUTINE JNNGS2 (A0,V0,D0,G0,GA,W2,BW,A1,A2,A3,A4,B1,B2,B3,B4,*)
C
C JENNNINGS-NIGAM METHOD OF NUMERICAL INTEGRATION
C
C INPUT
C A0 RESPONSE ACCLERATION AT PREVIOUS TIME STEP
C V0 RESPONSE VELOCITY AT PREVIOUS TIME STEP
C D0 RESPONSE DISPLACEMENT AT PREVIOUS TIME STEP
C G0 GROUND ACCELERATION AT PREVIOUS TIME STEP
C GA GROUND ACCELERATION AT CURRENT TIME STEP
C W2 = WN * WN
C BW = 2.0 * BT * WN
C A1 - A4 CONSTANTS FOR NUMERICAL INTEGRATION
C B1 - B4 CONSTANTS FOR NUMERICAL INTEGRATION
C
C OUTPUT
C A0 RESPONSE ACCELERATION AT CURRENT TIME STEP
C V0 RESPONSE VELOCITY AT CURRENT TIME STEP
C D0 RESPONSE DISPLACEMENT AT CURRENT TIME STEP
C
D=A1*D0+A2*V0-A3*GA-A4*G0
V0=B1*D0+B2*V0-B3*GA-B4*G0
A0=-GA-W2*D-BW*V0
D0=D
RETURN 1
END
SUBROUTINE LNACC2 (A0,V0,D0,GA,DT,A1,A2,A3,A4,B1,B2,B3,*)
C
C LINEAR ACCELERATION METHOD OF NUMERICAL INTEGRATION
C
C INPPUT
C A0 RESPONSE ACCLERATION AT PREVIOUS TIME STEP
C V0 RESPONSE VELOCITY AT PREVIOUS TIME STEP
C D0 RESPONSE DISPLACEMENT AT PREVIOUS TIME STEP
C GA GROUND ACCEERATION AT CURRENT TIME STEP
C DT TIME INTERVAL OF NUMERICAL INTEGRATION
C A1 - A4 CONSTANTS FOR NUMERICAL INTEGRATION
C B1 - B3 CONSTANTS FOR NUMERICAL INTEGRATION
C
C OUTPUT
C A0 RESPONSE ACCELERATION AT CURRENT TIME STEP
C V0 RESPONSE VELOCITY AT CURRENT TIME STEP
C D0 RESPONSE DISPLACEMENT AT CURRENT TIME STEP
C
A=A1*GA+A2*A0+A3*V0+A4*D0

20
 D0=D0+V0*DT+A0*B2+A*B3
V0=V0+(A0+A)*B1
A0=A
RETURN 1
END
SUBROUTINE NWMRK2 (A0,V0,D0,GA,DT,W2,BW,A1,A2,A3,A4,*)
C
C NEWMARK BETA METHOD OF NUMERICAL INTEGRATION
C
C INPPUT
C A0 RESPONSE ACCLERATION AT PREVIOUS TIME STEP
C V0 RESPONSE VELOCITY AT PREVIOUS TIME STEP
C D0 RESPONSE DISPLACEMENT AT PREVIOUS TIME STEP
C GA GROUND ACCELERATION AT CURRENT TIME STEP
C DT TIME INTERVAL OF NUMERICAL INTEGRATION
C W2 = WN * WN
C BW = 2.0 * BT * WN
C A1 = 0.5 * DT
C A2 = (0.5 - BETA) *DT **2
C A3 = BETA * DT **2
C
C OUTPUT
C A0 RESPONSE ACCELERATION AT CURRENT TIME STEP
C V0 RESPONSE VELOCITY AT CURRENT TIME STEP
C D0 RESPONSE DISPLACEMENT AT CURRENT TIME STEP
C
B=V0+A1*A0
C=D0+V0*DT+A0*A2
10 A=A0
V0=B+A1*A
D0=C+A3*A
A0=-GA-BW*V0-W2*D0
ER=ABS(A0-A)/(ABS(A0)+ABS(A))
IF (ER.GT.A4) GO TO 10
RETURN 1
END
SUBROUTINE RKGIL2 (A0,V0,D0,G0,GA,DT,W2,BW,A1,*)
C
C RUNGE-KUTTA-GILL METHOD OF NUMERICAL INTEGRATION
C
C INPPUT
C A0 RESPONSE ACCLERATION AT PREVIOUS TIME STEP
C V0 RESPONSE VELOCITY AT PREVIOUS TIME STEP
C D0 RESPONSE DISPLACEMENT AT PREVIOUS TIME STEP
C G0 GROUND ACCELERATION AT PREVIOUS TIME STEP
C GA GROUND ACCELERATION AT CURRENT TIME STEP
C DT TIME INTERVAL OF NUMERICAL INTEGRATION
C W2 = WN * WN
C BW = 2.0 * BT * WN
C A1 CONSTANT FOR NUMERICAL INTEGRATION
C
C OUTPUT
C A0 RESPONSE ACCELERATION AT CURRENT TIME STEP
C V0 RESPONSE VELOCITY AT CURRENT TIME STEP
C D0 RESPONSE DISPLACEMENT AT CURRENT TIME STEP
C
F0=DT*V0
G0=DT*A0
F1=F0+A1*G0
G1=G0-A1*(-G0+GA+BW*G0+W2*F0)
F2=F0+A1*G1
G2=G0-A1*(-G0+GA+BW*G1+W2*F1)
F3=F0+DT*G2
G3=G0-DT*(-G0+GA+BW*G2+W2*F2)
D0=D0+(F0+F1+F1+F2+F2+F3)/6.0
V0=V0+(G0+G1+G1+G2+G2+G3)/6.0
A0=-GA-BW*V0-W2*D0

21
RETURN 1
END

22
 Chapter 7. Matrix Analysis of Linearly Elastic Plane Frame

A structural analysis method for plane (two-dimensional) frames is presented in this chapter. The
efficiency of the analysis is not intended here, but the basic principle of the analysis should be
clearly understood. The procedure can be easily extended to a three-dimensional frame analysis or
a finite element analysis as long as the degree of freedom is increased for nodal displacements and
forces of each structural unit. The elastic behavior of members is considered, as an example, in the
formulation of member stiffness matrix. The procedure can be also used in the nonlinear earthquake
response analysis of a structure because the stiffness of members is assumed to be linearly elastic
over a short time increment in the analysis.

7.1 Assumptions

The following assumptions are made in the structural analysis of a moment-resisting plane frame
structure;
(a) All members and loads lie in the vertical plane of a frame (Plane Frame),
(b) Each member is prismatic and straight,
(c) Small displacement,
(d) Each element is rigidly connected at a joint. Member end displacement is equal to the
displacement at the connecting joint (Continuity Condition).
(e) Members behave within linearly elastic region,
(f) Axial and flexural deformation is considered for each member,
(g) External forces act at joints,
(h) External forces acting at a joint are equal to the sum of member end forces of members
connected to the joint (Equilibrium of Forces).

Each member is represented by a straight-line element passing through the geometrical centroid
of the section of a member. Axial and flexural elastic deformations are considered for each member.
Shear deformation is neglected for simplicity, but shear deformation can be included in the analysis
by introducing shear deformation in the member stiffness matrix. Flexible connection can be
considered by introducing rotational or shear springs at member ends.

Under a small displacement, the equilibrium of forces can be expressed using the coordinate
system for pre-deformed configuration.

Internal bending moment should be calculated at the geometrical centroid of the section. The
axial deformation is zero under bending at the geometrical centroid because the neutral axis under
bending passes through the geometrical centroid when the member is in the linearly elastic state.
Therefore, the axial deformation at the geometrical centroid is null under bending.
mz1 mz2
However, the neutral axis shifts from the
geometric centroid after the stress in section exceeds
the proportional limit of materials. Axial strain at the 2
1
geometric centroid develops elongation of the
member even under pure bending. The interaction of
axial and bending deformation should be considered Fixed End Forces
py1 py2
in the analysis.

The forces acting within a member should be

replaced by the equivalent member end actions
py1 py2
(forces equal to fixed end forces, but having opposite
sign). The effect of forces acting in the member will
be considered after the member end forces are
determined in the structural analysis. mz1 Equivalent Joint Loads mz2

1
7.2 Member Stiffness Matrix in Local Coordinates

A set of local (member) coordinates of a member are selected in the plane of a frame. The x-axis
is defined in the direction of the member, and y-axis perpendicular to the member following the
right-hand rule; i.e., rotation is positive in the counter-clockwise direction.

Coordinate System: Member direction is defined for each member from start end 1 to terminal end
2. Member end actions { p1}i and { p2 }i at start end 1 and terminal end 2 of member i are
 p x1   px 2 
   
{ p1}i =  p y1  { p2 }i =  p y 2 
m  m 
 z1  i  z 2 i
y

p y1 p y2
p x1 p x2 x
1 2
m z1 m z2

For each member, member end displacements {d1}i and {d 2 }i at start end 1 and terminal end
2 of member i, are
 d x1  d x 2 
   
{d1}i = d y1  {d 2 }i = d y 2 
θ  θ 
 z1  i  z 2 i

y θz2
2

dx1
py2
dy1 1 θz1
x

dx2

Member stiffness relation: The relation of member end actions and member end displacement
through member stiffness sub-matrices.
{ p1}i = [k11 ]i {d1}i + [k12 ]i {d 2 }i
{ p2 }i = [k 21 ]i {d1}i + [k 22 ]i {d 2 }i

For a linearly elastic member, stiffness sub-matrices are given below:

2
  EA  1 EA, EI 2
 L 0 0 
  L
12 EI 6 EI 
[ k11 ]i =  0 EA EA
 L3 L2  L d x1 = 1.0 −
L
 6 EI 4 EI 
 0
 L2 L  i
6 EI 6 EI
L2 L2
 EA 
− L 0 0  d y1 = 1.0

 12 EI 6 EI  12 EI
12EI
[k12 ]i = [k 21 ]i T =  0 − 3  −
L3
 L L2  L3
 0 6 EI 2 EI 
− 2
 L L  i
θ = 1.0
 EA  z1

 L 0 0  4 EI 6 EI 6 EI
− 2
2 EI
2
L L
 12 EI 6 EI 
L L
[k 22 ]i =  0 − 2 
 L3 L 
 0 − 6 EI 4 EI 
 L2 L  i
in which E : Young's modulus of material, A : cross sectional area, I : moment of inertia of section,
and L : member length.

3
7.3 Coordinate Transformation

Global coordinates ( X , Y , Z ) are defined in the plane of a frame. The X-axis is defined in the
horizontal direction, the Y-axis in the upward vertical direction, and the rotation Θ in the
counter-clockwise using the right-hand rule.

Suppose the member axis x of Y

member i is inclined from the horizontal y X = x cosα − y sin α
x
global axis X by angle α Y = x sinα + y cosα
counter-clockwise. Forces { p}i in a Θ =θ
member coordinate system may be
transformed into forces {P}i in the
global coordinate system considering the α X
equilibrium of forces in each direction.
PX = px cos α − p y sin α
PY = px sin α + p y cos α
M Z = mz
or
px = PX cos α + PY sin α
p y = − PX sin α + PY cos α
mz = M Z

These relations can be expressed in the matrix form,

 PX  cosα − sin α 0  p x 
    
 PY  =  sin α cosα 0  p y 

M   0 0 1 i m z  i
 Z i 
 px   cos α sin α 0   PX 
    
 p y  =  − sin α cos α 0   PY 
m   0 0 1  i  M Z i
 z i 

The relations of forces in the global and local coordinates can be expressed in a symbolic matrix
form using a transformation matrix [T ]i ;
{P}i = [T ]i { p}i
{ p}i = [T ]i−1{P}i

Note the following property of coordinate transformation matrix [T ]i of member i:

−1 T
[T ]i = [T ]i

Similarly, displacement {D}i in the global coordinate system may be transformed into
displacement {d}i in a member coordinate system;
 D X  cosα − sin α 0  d x 
    
 DY  =  sin α cosα 0 d y 
Θ   0 0 1 i θ z i
 Z i 

4
  d x   cosα sin α 0  D X 
    
d y  =  − sin α cosα 0  DY 

θ   0 0 1 i  Θ Z  i
 z i 
or
{D}i = [T ]i {d }i
T
{d}i = [T ]i {D}i
T
where, matrix [T ]i is the transpose of matrix [T ]i .

5
7.4 Member Stiffness Matrix in Global Coordinates

Member end actions {P1}i at start end of

Y x
member i in the global coordinate system are {P1}i
expressed by those in the local coordinate system {p1}i
by the coordinate transformation; y
{P1}i = [T ]i { p1}i 2

Using the member stiffness relation,

{ p1}i = [k11 ]i {d1}i + [k12 ]i {d 2 }i 1
The member end actions {P}i in the global
coordinate system can be expressed by member X
end displacements {d }i in the member
coordinate system:
{P1}i = [T ]i { p1}i = [T ]i ([k11 ]i {d1}i + [k12 ]i {d 2 }i )
{P2 }i = [T ]i { p2 }i = [T ]i ([k 21 ]i {d1}i + [k 22 ]i {d 2 }i )

Transformation of displacement coordinates from the global coordinate system to the local
coordinate system;
T
{d1}i = [T ]i {D1}i
T
{d 2 }i = [T ]i {D2 }i

The member end actions {P1}i at the starting end of member i are expressed in terms of
member end displacements {D1}i and {D2 }i both in the global coordinate system:
{P1 }i = [T ]i ([k11 ]i {d 1 }i + [k12 ]i {d 2 }i ) Y Θ z2
2
= [T ]i ([k11 ]i [T ]Ti {D1 }i + [k12 ]i [T ]Ti {D2 }i )
= [T ]i [k11 ]i [T ]Ti {D1 }i + [T ]i [k12 ]i [T ]Ti {D2 }i Dx1
Dy2
= [ K11 ]i {D1 }i + [ K12 ]i {D2 }i 1 Θ z1
where, Dy1
Dx2 X
[ K11 ]i = [T ]i [k11 ]i [T ] T
i

[ K12 ]i = [T ]i [k12 ]i [T ]Ti Member End Displacements in Global Coordinates

Similarly, the member end action {P2}i at the terminal end can be expressed in the global
coordinate system;
{P2 }i = [T ]i ([k 21 ]i {d 1 }i + [k 22 ]i {d 2 }i )
= [T ]i ([k 21 ]i [T ]Ti {D1 }i + [k 22 ]i [T ]Ti {D2 }i )
= [T ]i [k 21 ]i [T ]Ti {D1 }i + [T ]i [k 22 ]i [T ]Ti {D2 }i
= [ K 21 ]i {D1 }i + [ K 22 ]i {D2 }i
where,
[ K 21 ]i = [T ]i [k 21 ]i [T ]Ti
[ K 22 ]i = [T ]i [k 22 ]i [T ]Ti Joint J Starting End 1 Joint K

Term inal End 2

Therefore, member end actions {P}i , in
the global coordinates, of member i can be M em ber i
expressed by member end displacements

6
{D}i using member stiffness matrix [ K ]i in the global coordinates;
{P}i = [ K ]i {D}i
 P1   K11 K12   D1 
  =  
 P2  i  K 21 K 22 i  D2 i
where,
[ K jk ] = [T ]i [k jk ][T ]Ti for j, k = 1,2

7
7.5 Continuity of Displacement at Joint

Suppose member i is rigidly

connected to joints J and K of the
structure; starting end 1 of member i is Joint J Starting End 1 Joint K
connected to joint J and terminal end 2
of member i to joint K. Term inal End 2

By the continuity conditions of Mem ber i

displacement at joints, member end
displacements and joint displacements
are equal as long as they are expressed
in the common coordinate system:
{D1}i = {D}J
{D2 }i = {D}K
where {D1 }i : displacement at start end 1 of member i, connected to joint J of the structure,
expressed in the global coordinate system, {D2 }i : displacement at terminal end 2 of member i,
connected to joint K of the structure, expressed in the global coordinate system, {D}J :
displacement at joint J of the structure expressed in the global coordinate system, {D}K :
displacement at joint K of the structure expressed in the global coordinate system.

8
7.6 Equilibrium of Forces at Joint

The sum of member end forces of members connected at a joint is equal to the external forces
acting at the joint, both expressed in the global coordinate system.
n
{P}J = ∑{PJ }i
i =1

where, {P}J : external forces acting at joint J of the structure expressed in the global coordinate
system, {PJ }i : member end forces of member i at a member end connected to joint J of the
structure expressed in the global coordinate system, n: number of members connected to joint J of
the structure. If start end 1 of member i is connected to joint J of the structure, {PJ }i = {P1 }i ; if
terminal end 2 is connected to joint J, {PJ }i = {P2 }i .

Member n

Jn
Member 1 Joint J
Member i
J1 Ji
J2

Ji: End of member i connected to

Joint J of the structure.
Member 2

9
7.7 Formulation of Structural Stiffness Matrix

Suppose start end 1 of member i is connected to joint J of the structure and terminal end 2 to joint
Ki of the structure, member end forces {PJ }i of member i at joint J are expressed in the global
coordinate system, as
{PJ }i = {P1 }i
= [ K11 ]i {D1 }i + [ K12 ]i {D2 }i

Y
{PJ}i={P1}i

2
Joint J Joint Ki
1 Member i
X

By the continuity condition of member end displacements and structural joint displacements at a
joint in the global coordinate system,
{D1 }i = {D}J
{D2 }i = {D}Ki

Therefore,
{PJ }i = [ K11 ]i {D}J + [ K12 ]i {D}Ki

Similarly, if terminal end 2 of member i is connected to joint J and starting end 1 to joint Ki,
member end forces {PJ }i of member i at Joint J are expressed as
{PJ }i = {P2 }i
= [ K 21 ]i {D1}i + [ K 22 ]i {D2 }i
= [ K 21 ]i {D}Ki + [ K 22 ]i {D}J

{PJ}i={P2}i

2
Joint Ki
1 Member i Joint J
X

By the equilibrium condition of member end forces and external forces at joint J, where n
members are connected,
n
{P}J = ∑{PJ }i
i =1

Suppose m members of start end 1 are connected to joint J of the structure and (n-m) members of
terminal end 2 are connected to the same joint,
10
 m n
{P}J = ∑ [ K11 ]i {D}J + [ K12 ]i {D}Ki + ∑[ K ] {D}Ki + [ K 22 ]i {D}J
21 i
i =1 i = m +1
m m n n
= ∑ [ K11 ]i {D}J + ∑ [ K12 ]i {D}Ki + ∑[ K 21 ]i {D}Ki + ∑[ K ] {D}J
22 i
i =1 i =1 i = m +1 i = m +1

If the equilibrium of forces is written for every joint including the support joint, the external forces
{P} at joints are expressed as a linear function of joint displacement {D} ;
{P} = [ K ]{D}
in which [K] is called a structural stiffness matrix.

Normally, the structural stiffness matrix [K] is formulated member by member rather than joint by
joint. Suppose starting end of member i is connected to joint J and terminal end to joint K, then
member stiffness sub-matrix [ K11 ]i is added at (J, J) location of the structural stiffness matrix [K],
[ K12 ]i at (J, K) location, [ K 21 ]i at (K, J) location, and [ K 22 ]i at (K, K) location. This process of
adding member sub-matrices to the structural stiffness matrix is repeated for all members.

J K
l l
l l
l l
J -------------(J,J)------------------------------(J,K)-----
l l
l l
l l
l l
[K] = l l
l l
l l
l l
l l
l l
K ------------(K,J)-----------------------------(K,K)-----
l l
l l
l l

[ K ( J , J )] = [ K ( J , J )] + [ K11 ]i
[ K ( J , K )] = [ K ( J , K )] + [ K12 ]i
[ K ( K , J )] = [ K ( K , J )] + [ K12 ]Ti
[ K ( K , K )] = [ K ( K , K )] + [ K 22 ]i

If the symmetric properties of the structural stiffness matrix are recognized, then only upper
triangle part of the structural stiffness matrix need be formulated.

 M  O M M L  M 
 P  L K K IJ L  DI 
 I 
 =
II
 
P
 J L K JI K JJ L  DJ 
 M   M M M

O  M 

11
7.8 Free Joint Displacements and Support Reactions

Joints are classified into support joints and free joints. Rigidly supported joints cannot displace in
any direction, hence the displacement {Ds } at the support joint is zero; however, force (reaction)
{Ps } at a support joint is not known. On the other hand, at free end, displacement {D f } is not
known, but force (external load) {Pf } is given.
known: {Pf } (given as joint loads)and {Ds } (=0 for fixed supports)
unknown: {D f } and {Ps }

Therefore, it is desirable to reorganize the structural stiffness relation and separate displacement
and force at free joints and support joints;
 Pf   K ff K fs   D f 
 =  
 Ps   K sf K ss   Ds 
where {Ds } = {0} . The free joint displacement {D f } can be solved from the first equation:
{D f } = [ K ff ]−1{Pf }
and then, the support reaction {Ps} is solved.
{Ps } = [ K sf ]{D f }

Free Joint

Support Joint

12
7.9 Member End Actions

Once joint displacements are all calculated, then member end actions in the local coordinate
system can be calculated using joint displacements and member stiffness matrix.

The member end actions and displacements

relation for member i is
Joint J Starting End 1 Joint K
{ p1}i = [k11 ]i {d1}i + [k12 ]i {d 2 }i
{ p2 }i = [k 21 ]i {d1}i + [k 22 ]i {d 2 }i Term inal End 2

Mem ber i
If starting end 1 of the member is connected
to joint J and terminal end 2 to joint K, using the
continuity condition of displacement at joint an transformation of coordinates, joint end
displacements in local coordinates are expressed
{d1}i = [T ]Ti {D}J
{d 2 }i = [T ]Ti {D}K

Combining the above two relations;

{ p1}i = [k11 ]i [T ]Ti {D}J + [k12 ]i [T ]Ti {D}K
{ p 2 }i = [k 21 ]i [T ]Ti {D}J + [k 22 ]i [T ]Ti {D}K

The bending moment distributes linearly along the member because no intermediate loads act in
the member.

If intermediate loads act in the member, the above member end actions should be added to the
member actions of fixed-fixed member under the given intermediate loads.

py1 py2
mz2 FEM2
mz1 FEM1

px1 px2
Member i

Structural Analysis Stress due to member loading

Final Member Stresses

13
7.10 Example: One-story One-bay Frame

Structure: A one-story one-bay frame is analyzed. The stiffness matrix of the structure is formulated
in this example.

The story height is h m and the span is l Y 1 2

m. The base of the columns is fixed. Only 1
flexural deformation is considered in the beam EIb
and axial and flexural deformations are
considered in the columns. The flexural rigidity 3
2 EIc,EAc EIc,EAc h
of the beam and the columns are EI b and
EI c , and the axial rigidity of the columns is X
EAc .
3 l 4
Joints are numbered from 1 to 4 starting
from the left beam-column joint to the right
column base. Members are numbered from 1 to 3 starting from the beam.

Global Coordinate System: The global coordinate ΘZ1 ΘZ 2

DY 1 DY 2
system is taken as X-axis in the horizontal direction DX 2
DX 1
and Y-axis in the vertical direction. Displacement
1 2
{DX , DY , Θ Z } and force {PX , PY , M Z } are
defined at each joint in the direction of the global
coordinates. Rotation and moment are positive ΘZ 3 ΘZ 4
counterclockwise. DY 3 DY 4
DX 3 DX 4
3 4
Member Coordinate System: The x-axis of a member is taken in the direction of the member, and
the y-axis normal to the member axis. The member direction of a beam is from left to right, and that
of a column is downward. Displacement {d x , d y ,θ z } and force { px , p y , mz } are defined at starting
and terminal ends in the direction of the member coordinates. Rotation and moment are positive
counterclockwise.

d x1 , p x1 θ z1 , m z1
y
1
d y1 , p y1
y

θ z1 , m z1 θ z2 , mz2
d y1 , p y1 d y2 , py2

d x1 , p x1 1 2 d x2 , px2 x d y2 , py2
2

d x2 , px2
θ z2 , mz2

The transformation matrices for a column ( α =-90.0 deg) and a beam ( α =0.0 deg) are given as
follows;

14
  0.0 1.0 0.0  1.0 0.0 0.0 
[T ]column =  −1.0 0.0 0.0  [T ]beam =  0.0 1.0 0.0 
 0.0 0.0 1.0   0.0 0.0 1.0 

Beam Stiffness Matrix in Member Coordinates: For a beam 1-2, no axial deformation is
considered. Therefore, the stiffness matrix of a beam in the member coordinates is given as below;

 EAb   EAb 
 l 0 0  − l 0 0 
 px1     d x1    d x2 
   12 EI b 6 EI b     12 EI b 6 EI b   
 p y1  = 0  d y1  + 0 − 3  d y2 
 l 3
l 
2  l l 
2
m     
 z1 beam  6 EI 4 EI  θ z1 beam  6 EI 2 EI  θ z 2 beam
 0 b b   0 − 2 b b 
 l2 l  beam  l l  beam
 EAb   EAb 
− l 0 0   l 0 0 
 px 2     d x1    d x2 
   12 EI b 6 EI b     12 EI b 6 EI b   
 py 2  = 0 − 3 − 2  d y1  + 0 − 2 d y 2 
 l l   l 3
l 
m     
 z 2 beam  6 EI 2 EI  θ z1 beam  6 EI 4 EI  θ z 2 beam
 0 b b   0 − 2b b 

 l2 l  beam  l l  beam
but d x1 = d x 2 and px1 = − px 2 .
θ z2

θ z1
d y2
p y1
mz1 d y1
py2 mz 2

px1 1 2 px 2

Beam Stiffness Matrix in Global Coordinates: The member coordinates of the beam coincide with
the global coordinates; hence the member stiffness matrix of the beam is expressed in the same
form;
 EAb   EAb 
 l 0 0  − l 0 0 
 PX 1     DX 1     DX 2 
   12 EI b 6 EI b     12 EI b 6 EI b   
 PY 1  = 0  DY 1  + 0 − 3  DY 2 
 l 3
l 2   l l 2 
M  Θ  Θ 
 Z 1 beam  6 EI 4 EI   Z 1 beam  6 EI 2 EI   Z 2 beam
 0 b b   0 − 2 b b 
 l 2
l  beam  l l  beam
 EAb   EAb 
− l 0 0   l 0 0 
 X2 
P    DX 1     DX 2 
   0 12 EI 6 EI   12 EI b 6 EI b   
P
 Y2  = − 3 b − 2b D
 Y1  +  0 − 2  DY 2 
 l l   l3 l 
M  Θ  Θ 
 Z 2 beam  6 EI b 
2 EI b   Z 1 beam  6 EI 4 EI b   Z 2 beam
 0  0 − 2b
 l2 l  beam  l l  beam
but DX 1 = DX 2 and PX1=-PX2.

15
Column Stiffness Matrix in Member Coordinates: The stiffness matrix of a column is given in the
member coordinates as follows;

 EAc   EAc 
 h 0 0  − h 0 0 
 px1     d x1    d x 2 
  12 EI c 6 EI c    12 EI c 6 EI c   
 p y1  = 0 d
2   y1 
+ 0 − d y 2 
 h3 h  h3 h2   
m    θ z1 column   θ
 z1 column 6 EI c 4 EI c  6 EI 2 EI c   z 2 column
 0  0 − 2c
 h2 h   h h 
 EAc   EAc 
− h 0 0   h 0 0 
 px 2     d x1    d x2 
  12 EI c 6 EI c    12 EI c 6 EI c   
 py 2  = 0 − − 2  d y1  + 0 − 2 d y 2 
 h3 h   h3 h  
m    θ z1 column   θ
 z 2 column 6 EI c 2 EI c  6 EI 4 EI c   z 2 column
 0  0 − 2c
 h2 h   h h 

u y1 y
1
A ux1

θ z1

θz2
2
ux 2

x uy 2

Column Stiffness Matrix in Global Coordinates: The member coordinates of a column rotate 90
degrees clockwise. The member stiffness matrix of the beam is expressed as follows;
12 EI c 6 EI c   12 EI c 6 EI c 
 h3 0
h 2   − h3 0
h2   D 
 PX 1     DX 1    X2
  EAc   EAc  
 PY 1  = 0 
0  DY 1  + 0 − 0   DY 2 
 h   h 
M     Θ Z 1 column   Θ 
 Z 1 column 2 EI c   Z 2 column
 6 EI c 0
4 EI c   − 6 EI c 0
 h 2 h   h 2 h 
 12 EI c 6 EI c   12 EI c 6 EI c 
 − h3 0 −
h 2   h3 0 −
h2   D 
 PX 2     DX 1    X2
  EAc   EAc  
 PY 2  = 0 − 0  D + 0 0   DY 2 
 h   Y1   h 
M     Θ Z 1 column   Θ 
 Z 2 column 4 EI c   Z 2 column
 6 EI c 0
2 EI c   − 6 EI c 0
 h 2 h   h 2 h 

16
Continuity of Displacement at Joint: The continuity of displacement is considered at each joint
using the global coordinates;

Joint 1:
1 1 1 2 2
 DX   DX 1   DX 1 
     
 DY  =  DY 1  =  DY 1  1 1
Θ     
 Z  Jo int 1  Θ Z 1 Member 1  Θ Z 1 Member 2
Joint 2:
2 3
 DX   DX 2   DX 1  2 2
     
 DY  =  DY 2  =  DY 1  3 4
Θ     
 Z  Jo int 2  Θ Z 2 Member 1  Θ Z 1 Member 3
Joint 3:
 DX   DX 2 
   
 DY  =  DY 2 
Θ   
 Z  Jo int 3  Θ Z 2 Member 2
Joint 4:
 DX   DX 2 
   
 DY  =  DY 2 
Θ   
 Z  Jo int 4  Θ Z 2 Member 3

Equilibrium of Forces at Joint: The equilibrium of forces is considered at each joint;

1 1 1 2 2
Joint 1:
 PX   PX 1   PX 1  1 1
     
 PY  =  PY 1  +  PY 1 
M      2 3
 Z  Jo int 1  M Z 1 Member 1  M Z 1 Member 2
Joint 2: 2 2
 PX   PX 2   PX 1  3 4
     
 PY  =  PY 2  +  PY 1 
M     
 Z  Jo int 2  M Z 2 Member 1  M Z 1 Member 3

Joint 3:
 PX   PX 2 
   
 PY  =  PY 2 
M   
 Z  Jo int 3  M Z 2 Member 2
Joint 4:
 PX   PX 2 
   
 PY  =  PY 2 
M   
 Z  Jo int 3  M Z 2 Member 2

17
Structural Stiffness: The stiffness of the structure is formulated by using the equilibrium and
continuity conditions at each joint;

Joint 1:
 PX   PX 1   PX 1 
     
 PY  =  PY 1  +  PY 1 
M     
 Z  Jo int 1  M Z 1 Member 1  M Z 1 Member 2
 EAb   EAb 
 l 0 0  − l 0 0 
   DX     DX 
12 EI b 6 EI b     12 EI b 6 EI b   
= 0  DY  + 0 −  DY 
 l3 l2     l3 l2   
 Θ Θ
 0 6 EI b 4 EI b   Z 1  6 EI 2 EI b   Z 2
0 − 2b
 l2 l 1  l l 1
12 EI c 6 EI c   12 EI c 6 EI c 
 h3 0
h 2   − h3 0
h2   D 
   DX    X
EAc   EAc  
+ 0 0   DY  +  0 − 0  D
 h   h   Y
   Θ Z 1   Θ 
 6 EI c 4 EI c   − 6 EI c 2 EI c   Z 3
0 0
 h h  2  h h  2
2 2

Joint 2:
 PX   PX 2   PX 1 
     
 PY  =  PY 2  +  PY 1 
M     
 Z  Jo int 2  M Z 2 Member 1  M Z 1 Member 3
 EAb   EAb 
− l 0 0   l 0 0 
   DX     DX 
12 EI b 6 EI   12 EI b 6 EI b   
= 0 − − 2 b   DY  +  0 − 2  DY 
 l3 l     l3 l   
 Θ Θ
 0 6 EI b 2 EI b   Z 1  6 EI 4 EI b   Z 2
0 − 2b
 l2 l 1  l l 1
12 EI c 6 EI c   12 EI c 6 EI c 
 h3 0
h 2   − h3 0
h2   D 
   X 
D  X
EAc   EAc  
+ 0 0   DY  +  0 − 0  D
 h   h   Y
   Θ Z 2   Θ 

 6 EI c 4 EI c   − 6 EI c 2 EI c   Z 4
0 0
 h 2 h  3  h 2 h  3

Joint 3:

18
  PX   PX 2 
   
 PY  =  PY 2 
M   
 Z  Jo int 3  M Z 2 Member 2
 12 EI c 6 EI c   12 EI c 6 EI c 
 − h3 0 −
h 2   h3 0 −
h2   D 
   DX 1    X2
EAc   EAc  
= 0 − 0   DY 1  +  0 0   DY 2 
 h   h 
   Θ Z 1 2   Θ 
 6 EI c 2 EI c   − 6 EI c 4 EI c   Z 2 2
0 0
 h 2 h   h 2 h 
 12 EI c 6 EI c   12 EI c 6 EI c 
 − h3 0 −
h 2   h3 0 −
h2   D 
   DX    X
EAc   EAc  
= 0 − 0   DY  +  0 0   DY 
 h   h 
   Θ Z 1   Θ 
 6 EI c 2 EI c   − 6 EI c 4 EI c   Z 3
0 0
 h 2 h   h 2 h 

Joint 4:
 PX   PX 2 
   
 PY  =  PY 2 
M   
 Z  Jo int 4  M Z 2 Member 3
 12 EI c 6 EI c   12 EI c 6 EI c 
 − h3 0 −
h 2   h3 0 −
h2   D 
   DX 1    X2
EAc   EAc  
= 0 − 0   DY 1  +  0 0   DY 2 
 h   h 
   Θ Z 1 3   Θ 
 6 EI c 2 EI c   − 6 EI c 4 EI c   Z 2 3
0 0
 h h   h h 
2 2

 12 EI c 6 EI c   12 EI c 6 EI c 
 − h3 0 −
h 2   h3 0 −
h2   D 
 
 X D    X
EAc    EAc  
= 0 − 0  D + 0 0   DY 
 h  Y   h 
 
 Θ    Θ 
 6 EI c 2 EI c   Z 2  6 EI c 4 EI c   Z 4
0 − 2 0
 h h   h h 
2

From the boundary condition of the support;

 DX  0.0 
   
 DY  = 0.0 
 Θ  0.0 
 Z 3  
 DX  0.0 
   
 DY  = 0.0 
 Θ  0.0 
 Z 4  

Therefore, the stiffness matrix of the structure is written as

19
  EAb  12 EI c 6 EI c 
0 0   h3 0
 PX 
 l h2   D 
   DX    X
  12 EI b 6 EI b     EAc  
 PY  = 0  DY  + 0 0   DY 
M 
 l3 l2     h 
 Z  Jo int 1  Θ   Θ 
 0 6 EI b 4 EI b   Z 1  6 EI c 4 EI c   Z 1
0
 l2 l 1  h h  2
2

Free joint 1
 EAb 
− l 0 0 
   DX 
 12 EI b 6 EI b   
+ 0 − 3  DY 
 l l2   
 Θ
 0 6 EI b 2 EI b   Z 2
− 2
 l l 1
 EAb   EAb 
− l 0 0   l 0 0 
 PX     DX     DX 
  12 EI 6 EI   12 EI b 6 EI b   
 PY  = 0 − 3 b − 2 b   DY  +  0 − 2  DY 
 l l     l3 l   
M   Θ Θ
 Z  Jo int 2
 0 6 EI b 2 EI b   Z 1  6 EI 4 EI b   Z 2
0 − 2b
 l2 l 1  l l 1
Free joint 2
12 EI c 6 EI c 
 h3 0
h2   D 
  X
EAc  
+ 0 0   DY 
 h 
  Θ 
 6 EI c 4 EI c   Z 2
0
 h h  3
2

 12 EI c 6 EI c 
 − h3 0 −
h2   D 
 PX    X
  EAc  
 PY  = 0 − 0   DY  Support joint 3
 h 
M    Θ 
 Z  Jo int 3 2 EI c   Z 1
 6 EI c 0
 h h 
2

 12 EI c 6 EI c 
 − h3 0 −
h2   D 
 PX    X
  EAc  
 PY  = 0 − 0   DY  Support joint 4
 h 
    Θ 
 M Z  Jo int 4 2 EI c   Z 2
 6 EI c 0
 h h 
2

but, DX @ jo int1 = DX @ jo int 2 and PX 1 = − PX 2 .

The horizontal joint displacement at joints 1 and 2 are the same because no axial deformation is
considered in the beam; D X atjnt1 = D X atjnt 2 = D X 1 . As we consider the horizontal displacement at

20
joints 1 and 2 simultaneously, PX 1 = PX at jnt1 + PX at jnt 2 .

PX 1 24 EI 0 6 EI c 0 6 EI c DX 1
h3 h2 h2
PY at jnt1 0 12 EI b EAc 6 EI b 12 EI b 6 EI b DY at jnt 1
+ −
l3 h l2 l3 l2
M Z at jnt1 6 EI c 6 EI b 4 EI b 4 EI c 6 EI 2 EI b M Z at jnt 1
= + − 2b
h2 l2 l h l l
PY at jnt 2 0 12 EI 6 EI b 12 EI b EAc 6 EI b DY at jnt 2
− 3 b − 2 + − 2
l l l3 h l
M Z at jnt 2 6 EI c 6 EI b 2 EI b 6 EI 4 EI b 4 EI c M Z at jnt 2
− 2b +
h2 l2 l l l h

21
 Home Assignment No. 3

2002-03-04
S. Otani

Analyze a two-member structure. Define the member stiffness matrices in local coordinate
system, coordinate transformation matrices, member stiffness matrix in global coordinate system,
and stiffness matrix of the structure. Determine bending moment, shear and axial forces of the two
members. The length, cross sectional area and moment of inertia are common in the two members.

l = 3.00 m
l, E , A, I C
E = 200 × 10 3 N / mm 2 M= 100 kN m
A = 10 5
mm 2

θ = 30o
I = 1010 mm 4 B

l, E , A, I

22
Joint displacement (unit: m)
 0.0124 
 
{DB } =  0.0214  ×10−4
−0.1937 
 

Joint reactions (unit: m, N)

 0.2473 
 
{PC } = −0.1428 × 105
−0.2418
 
 −0.2473
 
{PA } =  0.1428  ×105
−0.2418
 

23
 Chapter 8. Numerical Solution of Linear Equations
The equilibrium of a structure at time t is expressed as the equilibrium of internal force (structural
resistance) {R (t )} and external force {F (t )} at structural nodes;
{F (t )} = {R (t )}
In a dynamic problem, the external force may be the sum of externally applied dynamic forces, and
negative values of inertia and damping forces. The internal force is the restoring force of the
structure.
{F (t )} = {P (t )} − [ M ]{&&
z (t )} − {D(t )}
where {P(t )} : external force at time t, [ M ] : mass matrix, normally defined as lumped mass at
each floor corresponding to the horizontal degree of freedom, {&&
z (t )} : absolute acceleration at mass
point, {D (t )} : damping resistance, normally defined at each floor for the horizontal degree of
freedom. Damping force may be assumed to be proportional to velocity {x& (t )} relative to the base
of the structure, but the damping matrix may be made proportional to instantaneous stiffness in the
horizontal degree of freedom.

8.1 Incremental Formulation

A nonlinear problem is normally solved by reducing the problem into a linear problem over a short
time increment in an incremental approach.

For example, the equilibrium of dynamic forces at time ti +1 can be written as

[M ]{&z&(ti+1 )}+ {D(ti+1 )}+ {R(ti +1 )} = {0}
The structural resistance and damping resistance may be written in incremental form;
[M ]({&x&(ti+1 )}+ &y&(ti+1 ){e}) + {D(ti )}+ {∆D(ti+1 )}+ {R(ti )}+ {∆R(ti+1 )} = {0}
y (ti +1 ) : ground acceleration (given value as an input earthquake motion) at ti +1 , and {e} :
where &&
vector having unit value for all elements.

Separating the known quantities and unknown quantities, the equilibrium of dynamic forces at
time t i +1 can be written as
[M ]{&x&(ti+1 )}+ {∆D(ti +1 )}+ {∆R(ti+1 )} = − &y&(ti+1 )[M ]&y&(ti+1 ){e}− {D(ti )}− {R(ti )}
The increment of internal force {∆R (ti +1 )} may be approximated by product of tangent stiffness
[ K ]ti and incremental displacement {∆x(t )} relative to the base of the structure, and the
increment of damping force {∆D (ti +1 )} by the product of tangent damping matrix [C ]ti and
incremental velocity {∆x& (t )} .
[M ]{&x&(ti+1 )}+ [C ]t {∆x& (ti+1 )}+ [K ]t {∆x(ti+1 )} = − &y&(ti+1 )[M ]&y&(ti+1 ){e}− {D(ti )}− {R(ti )}
i i

The incremental acceleration, velocity and displacement are related through a numerical
integration method assumed in the analysis. For example, using the Newmark β method scheme,
1
{∆x(ti +1 )} = ∆t{x& (ti )} + ( − β )∆t 2 {&&
x(ti )} + β∆t 2 {&&
x(ti +1 )}
2
1 1
{∆x& (ti +1 )} = ∆t{&&x(ti )} + ∆t{&&x(ti +1 )}
2 2
where β is a constant of the Newmark Beta method, {x& (ti )} and {&& x(ti )} are known velocity and
acceleration vectors at previous time step ti.

1
 The three linear equations may be solved to determine incremental displacement and velocity
{∆x(ti +1 )} , {∆x& (ti +1 )} and acceleration {&&
x(ti +1 )} at the new time step.

The stiffness of constituent members is nonlinear and the incremental internal force {∆R (ti +1 )}
may not be expressed as the product of tangent stiffness [ K ]ti and incremental displacement
{∆x(ti +1 )} . Nor may the incremental damping force {∆D(ti +1 )} at new time step ti +1 not be equal
to the product of tangent damping matrix [C ]ti and incremental velocity {∆x& (ti +1 )} . Therefore, the
equilibrium force may not be satisfied at new time step ti +1 . In other words, the internal force
{R(ti + ∆ti+1 )} may not be calculated as the sum of internal force {R(ti )} at previous time step t i
and the product of tangent stiffness matrix [ K ]t and incremental displacement {∆x(ti+1 )}.

Fi +1 Di +1
Ki ∆xi +1 Ci ∆x&i +1

∆Fi +1
Ki ∆Di +1
Fi Ci

x Di x&

xi ∆xi +1 xi +1 x&i ∆x&i +1 x&i +1

Structural resistance {R (ti +1 )} and damping resistance {D (ti +1 )} at new time step must be
re-evaluated for calculated new displacement {x(ti +1 )} and velocity {x& (ti +1 )} to satisfy the
hysteresis and damping relations. It should be noted, therefore, that the equilibrium of dynamic force
at new time increment is not satisfied;
{R (ti +1 )} + [ M ]{&&
z (ti +1 )} + {D (ti +1 )} − {P (ti +1 )} = {Error (ti +1 )}

It is generally time consuming to correct this error within the current time step because stiffness
and damping may be nonlinear in the dynamic problem. The unbalanced force {Error (ti +1 )} must
be corrected in the equilibrium of dynamic force in the next time interval.

If the correction of error is desired within the same time step, Newton-Raphson iteration method
may be used. The incremental linearization of the equilibrium equation leads to;
[ K ]{∆u} = {S }
(i )
The right hand side of the equation {S } may be updated for residue vector {Error (ti +1 )} , and
new solution may be sought;
[ K ]{∆u (i ) } = {S ( i ) }
The solution may be updated at iteration i;
{u ( i +1) } = {u (i ) } + {∆u (i ) }
In this solution process, the stiffness matrix should be reformulated. However, the reformulation of
stiffness matrix and its factorization is computationally expensive. Therefore, the initial stiffness
[ K 0 ] is normally maintained during iteration steps. The convergence may be accelerated by the use
of a scale factor δ in updating the solution;
{u ( i +1)
} = {u (i ) } + δ {∆u (i ) }
2
8.2 Modified Cholesky Matrix Decomposition

A set of n simultaneous linear algebraic equations are expressed in a matrix form:

[ A]{ X } = {b}
where, [ A] : coefficient matrix of size n x n, {X } : column vector of n unknowns, {b} : column
vector of n constants. The coefficient matrix [ A] is symmetric and positive definite in a normal
structural analysis.

A symmetric positive definite matrix [ A] can be decomposed into the product of three matrices as
follows:
[ A] = [ L][ D][U ]
= [U ]T [ D][U ]
in which [L] : lower unit triangular matrix, [D] : diagonal matrix, and [U ] : upper unit triangular
T
matrix (= [ L] ). The diagonal elements of lower and upper unit triangular matrices are equal to
unity.
1 0 0 L 0 d 11 0 0 L 0  1 u12 u13 L u1n 
l 1 0 L 0  0 d 22 0 L 0  0 1 u 23 L u 2 n 
 21 
[ A] = l 31 l 32 1 L 0  0 0 d 33 L 0  0 0 1 L u 3n 
   
 M M M O M  M M M O M M M M O M 
l n1 l n2 l n3 L 1  0 0 0 L d nn  0 0 0 L 1 
 1 0 0 L 0 d 11 0 0 L 0  1 u12 u13 L u1n 
u 1 0 L 0  0 d 22 0 L 0  0 1 u 23 L u 2 n 
 12
=  u13 u 23 1 L 0  0 0 d 33 L 0  0 0 1 L u 3n 
   
 M M M O M  M M M O M M M M O M 
ul 1n ul 2 n u 3n L 1  0 0 0 L d nn  0 0 0 L 1 

where, l ij = u ji .
 d 11 0 0 L 0  1 u12 u13 L u1n 
d u d 22 0 L 0  0 1 u 23 L u 2 n 
 11 12  
[ A] =  d 11u13 d 22 u 23 d 33 L 0  0 0 1 L u 3n 
  
 M M M O M M M M O M 
d 11u1n d 22 u 2 n d 33 u 3n L d nn  0 0 0 L 1 
 d 11 d 11u12 d 11u13 L d 11u1n 
d u d 11u12 + d 22
2
d 11u12 u13 + d 22 u 23 L d 11u12 u1n + d 22 u 2 n 
 11 12 
=  d 11u13 d 11u12 u13 + d 22 u 23 d 11u132 + d 22 u 23
2
+ d 33 L d 11u13 u1n + d 22 u 23 u 2 n + d 33 u 3n 
 
 M M M O M 
d 11u1n d 11u12 u1n + d 22 u 2 n d 11u13 + d 22 u 23 u 2 n + + d 33 u 3n L d 11u12n + d 22 u 22n + d 33 u 32n + L + d nn 

Expanding the product,

For i = 1; j = i: a11 = d 11
j > i: a1 j = d 11u1 j

3
 i −1

∑d
2 2 2 2
For i > 1; j = i: aii = d11u1i + d 22 u 2i + d 33 u 3i + L + d ii = d ii + kk u ki
k =1
i −1
j > i: aij = d11u1i u1 j + d 22 u 2i u 2 j + d 33 u 3i u 3 j + L + d ii u ij = d ii u ij + ∑d
k =1
kk u ki u kj

The relations may be solved for u ij (for i<j) and d ii . Note that [u ij ] is an upper triangular
matrix with null lower triangular elements ( u ij = 0 for i > j) and with unit diagonal element ( u ii = 1 ).
Therefore, only upper triangular elements need be stored; unit diagonal elements may not have to
be stored. Instead diagonal element d ii of the diagonal matrix [D] can be stored at the diagonals
u ii of upper triangle matrix [U ] .

For i = 1; j = i: u11 ( = d11 ) = a11

j > i: u1 j = a1 j / d11
i −1
For i > 1; j = i: uii ( = d jj ) = aii − ∑d
k =1
kk uki
2

i −1
j > i: uij = ( aij − ∑d
k =1
kk uki ukj ) / d ii

It should be noted that matrix elements aij and u ij appear in the same expression; in other words,
aij may be replaced by u ij in the operation to save computer memory in programming.

Banded Coefficient Matrix: The coefficient matrix [ A] of linear algebraic equations is normally
banded. Remember the formulation of structural stiffness matrix;
{P} = [ K ]{D}
The structural stiffness matrix is formulated member by member rather than joint by joint. For
member i with start end connected to joint J and terminal end to joint K, member stiffness sub-matrix
[K11]i is added at (J, J) location of the structural stiffness matrix [K ] , [ K 12 ]i at (J, K) location,
[ K 21 ]i at (K, J) location, and [ K 22 ]i at (K, K) location.

J K
l l
l l
l l
J --------(J,J)-----------------------------(J,K)-----
l l
l l
l l
l l
[K] = l l
l l J K
l l Member i
l l
l l
l l
K --------(K,J)-----------------------------(K,K)-----
l l
l l
l l

4
 K K12 
[ K ]i =  11
 K 21 K 22  i

Unless a member connects joints L and M, a structural stiffness sub-matrix at (L,M) is null; in
other words, (L,M) sub-matrix is not null if joints L and M are connected by a member. By proper
choice of joint numbering, the structural stiffness can be made narrowly banded.

To conserve computer time and storage, only the upper band of the stiffness matrix [K] is
normally constructed in a rectangular array of size N x NB, where N: the number of degrees of
freedom and NB: semi-band-width. The diagonal elements are stored in the first column of the
rectangular array.

The band width of the upper unit triangular matrix [U] is the same as that of the coefficient matrix
[A]. An element at ID(J) line and ID(K) column of the original square matrix can be stored at IR line
and IC column in a compact storage format.
IR=ID(J)
IC=ID(K)-(IR-1)

5
 SUBROUTINE BANFAC (N, NB, A, *)

A flow chart to decompose a banded rectangular coefficient matrix [ A] to the product of lower
triangular matrix, diagonal matrix and upper triangular matrix [U ] (Weaver and Gere, 1980). The
decomposed banded upper triangular matrix [U ] is stored in matrix [ A] . The diagonal elements of
the diagonal matrix [D ] are stored in the first column of the array [ A] .

N: number of unknowns in coefficient matrix [A] ,

NB: semi-band width of rectangular coefficient matrix [ A] ,
[ A] : rectangular coefficient matrix of array size N x NB,
*: statement number in a calling program to which a nonstandard RETURN is directed.

References:

L. Fox: An Introduction to Numerical Linear Algebra, Oxford University Press, New York, 1965.
W. Weaver, Jr. and J. M. Gere: Matrix Structural Analysis of Framed Structures, Second Edition,
Van Nostrand Reinhold, New York, 1980.

6
8.3 Solution of Linear Algebraic Equations

A set of linear algebraic equations are given in a matrix form

[ A]{ X } = {B}

The coefficient matrix [ A] can be decomposed into the products of lower triangular matrix,
diagonal matrix and upper triangular matrix:
[U ]T [ D][U ]{ X } = {B}

Let
[U ]{ X } = {Y }
and
[ D]{Y } = {Z }
then
[U ]T {Z } = {B}

The third relation can be expressed in a matrix form:

1 0 0 L 0  z1   b1 
u 1 0 L 0  z 2  b2 
 12    
u13 u 23 1 L 0  z 3  = b3 
 
 M M M O M  M   M 
   
u1n u 2n u 3n L 1  z n  bn 

This relation can be solved for zi's by "forward substitution procedure",

z1 = b1
z 2 = b2 − u12 z1
z 3 = b3 − u13 z1 − u 23 z 2
M
i −1
z i = bi − ∑ u ki z k
k =1

M
n −1
z n = bn − ∑ u kn z k
k =1
The z’s are stored in vector {b}.

The equation [ D ]{Y } = {Z } can be solved for yi's as follows;

d 11 0 0 L 0   y1   z1 
0 d 22 0 L 0   y 2   z 2 
    
0 0 d 33 L 0  y3  =  z3 
 
 M M M O M  M   M 
   
 0 0 0 L d nn   y n   z n 

y i = z i / d ii

The y’s are stored in vector {z}.

7
The equation [U ]{ X } = {Y } is expressed as follows:
1 u12 u13 L u1n   x1   y1 
0 1
 u 23 L u 2 n   x 2   y 2 
   
0 0 1 L u 3n   x 3  =  y 3 
 
M M M O M  M   M 
   
0 0 0 L 1   x n   y n 

The equation can be solved for {X } by "backward substitution procedure":

xn = y n
x n −1 = y n −1 − u n −1,n x n
x n − 2 = y n − 2 − u n − 2,n −1 x n −1 − u n − 2,n x n
M
n
xi = y i − ∑u
k =i +1
ik xk

M
n
x1 = y1 − ∑ u1k x k
k =2
The final solution x’s are stored in vector {y}.

8
 SUBROUTINE BANSOL (N, NB, U, B, X)

The solution of a set of linear equation in which the coefficient matrix [ A] has been decomposed
T
into the product of unit lower triangular matrix [U ] , diagonal matrix [D ] and unit upper triangular
matrix [U ] .

N: number of unknowns (number of columns of coefficient matrix [ A] ),

NB: semi-band width of coefficient matrix [ A] ,
[U ] : decomposed unit upper triangular matrix, stored in banded rectangular array, with
diagonal elements of diagonal matrix [D ] stored in the first column, array size of N x NB,
{B} : column vector of n constants,
{X } : column vector of n unknowns, used as working area for {Y } and {Z } .

9
8.4 Static Condensation

Frame structures are often analyzed by assuming floor slabs to be rigid in their own plane,
leading the same horizontal displacement at each floor. The mass at a floor is assumed to
concentrate at a floor level; the mass associated with vertical and rotational inertia is ignored.

The displacement degrees of freedom are selected to be

29 31 33
one horizontal displacement for a floor level, and vertical and
30 32 34
rotational displacements at each node. 35
22 24
Floor horizontal displacements 26
23 25 27
 X1  28
X  15 17 19
 
{X } =  2  16 18 20
21
 M  10
 X n  8
9 11
12
13
14
Node displacements
 Y1  1
2
3
4
5
6
Θ  7
 
{Y } =  1 
 Ym 
Θ m 
where Xi: horizontal displacement at floor i, Yj: vertical Degrees of freedom
displacement at node j, and θ j : rotation at node j.

If lumped translational masses are assumed at each floor level, the mass matrix becomes
diagonal with non-zero elements associated with the floor horizontal degrees of freedom and zero
elements associated with the nodal displacements.
[M ] = diagonal [M1,M2,...,Mn,0,0,....,0]
 M XX 0
=
 0 0
The stiffness matrix is normally banded, but can be expressed in a partitioned form;
K K XY 
[ K ] =  XX
 K YX K YY 

d2y
The equation of motion of an undamped system under horizontal base motion = &y& may be
dt 2
expressed as
 M XX 0  X&&   K XX K XY  ∆X   M XX 0 e  R X 
 0   &&  +     = −    &y&i +1 −  
 0  Y  i +1  K YX K YY   ∆Y  i +1  0 0 0  RY  i
where, {X } : horizontal floor displacement, {Y } : nodal vertical displacement and rotation, {e} :
vector of unit elements.

From the second equation;

[ K YX ]{∆X }i +1 + [ K YY ]{∆Y }i +1 = −{RY }i
and
{∆Y }i +1 = −[ K YY ] −1 ([ K YX ]{∆X }i +1 + {RY }i )

Therefore, the first equation can be written as

10
 [ M XX ]{ X&&}i +1 + ([ K XX ] − [ K XY ][ K YY ] −1 [ K YX ]){∆X }i +1
or
= −[ M XX ]{e}&y& + [ K KY ][ K YY ] −1{RY }i − {R X }i
[ M * ]{ X&&} + [ K * ]{∆X }
i +1 i +1

= −[ M ]{e}&y& + [ K XY ][ K YY ] −1 {RY }i − {R X }i
*

−1
where, [ M ] : story mass matrix, [ K ] : story stiffness matrix (= [ K XX ] − [ K XY ][ K YY ] [ K YX ] ).
* *

The equation of motion can be solved for incremental story displacement {∆X }i +1 and then
incremental nodal displacement and rotation {∆Y }i +1 can be determined by

{∆Y }i +1 = −[ K YY ] −1 ([ K YX ]{∆X }i +1 + {RY }i )

It should be noted that the inverse matrix of [ K YY ] is not formed in formulating the story matrix;
but a set of linear algebraic equations of the following form are solved;

[ K YY ][a] = [ K YX ]
[ K YY ]{b} = {RY }
or
[ K YY ][a M b] = [ K YX M RY ]

by decomposing the matrix [ K YY ] into [ L][ D][U ] .

{∆Y }i +1 = −[a]{∆X }i +1 − {b}

A series of plane frames were analyzed by (a) direct [ L][ D][U ] decomposition of the entire
matrix and (b) the static decomposition. The number of stories was varied from 3 to 30. The number
of bays in a frame was 2, 4 or 9. The computation time required to solve a set of equations was
compared.

11
 9-bay frame

4-bay frame

Computation time, ms
2-bay frame

LTDL Decomposition
Static condensation

Number of Stories

The computation time is reduced to one-half to one-third by the direct [L][D][U] decomposition
method compared to the static condensation method for a frame of more than 10-story high.

However, the number of operation is much less for the static condensation if the stiffness does
not change during a time increment. The computation time by the static condensation may not be
reduced even if the number of parallel frames increased.

12
 Home Assignment No. 4

2002-03-04
Otani, S.

For a four-story one-bay with rigid beams, formulate the stiffness matrix of a structure. The
material properties are constant throughout the structure. The moment of inertia of columns is 3I
in the first and second stories and 2I in the third and fourth stories.

2EI h

2EI h

3EI h

3EI h

If shear and axial deformation are ignored, each floor joint can move only in the horizontal
direction. Neither vertical nor rotation can take place at the joint.

Slope deflection equation of column AB with flexural rigidity EI ;

2 EI
M AB = (2θ A + θ B − 3RAB )
h
2 EI
M BA = (θ A + 2θ B − 3RAB )
h
Because the beam is rigid;
θ A = θ B = 0.0
Story shear and member rotation angle relation
M AB + M BA 24 EI
Vi = −2 = 2 RAB
h h
24 EI 24 EI
= 3 ∆ AB = 3 ( xi − xi −1 )
h h

The stiffness matrix of the structure is

 6 −3 0 0
− 3 5 − 2 0 
[K ] = 24 EI  
h 3  0 − 2 4 − 2
 
0 0 −2 2 

Decompose the stiffness matrix in [ K ] = [ L][ D ][U ] form.

13
[Solution]
The stiffness matrix of the structure is
 6 −3 0 0 
 
24 EI −3 5 −2 0 
[K ] = 3 
h  0 −2 4 −2 
 
 0 0 −2 2 
 1 0 0 0  1 
 1  144 0 0 0  1 − 0 0 
  2
− 0 
 2
1 0
 EI  0 84 0 0  4
 480  0 1 − 0 
= 4 
 0 0 0  7 
 0 − 1 0  h3 7  7
 7   
 0 72  0 0 1 − 
 7  0 0  10 
 0 0 − 1  5   0 0
 10   0 1 

14
 Chapter 9. Formulation of Member Stiffness Matrix

9.1 Introduction

Nonlinear analysis of a building considers the nonlinear behavior of ∆

materials. The nonlinear geometrical properties associated with large
P
deformation are normally ignored in the analysis. However, the P − ∆
effect, which is defined as the overturning effect of horizontally
displaced mass, may be included in the analysis using an approximate
formulation. The P − ∆ effect is known to increase the structural
displacement response when a story drift angle (inter-story
deformation divided by story height) exceeds approximately 0.01.

The equilibrium of external forces {P}i at time step ti and

internal resistances {R}i at all joints are expressed as; M = P×∆
{R}i = {P}i
P
Assuming that structural members behave linearly elastic between time steps ti and ti +1 ,
incremental resistance {∆R}i +1 between the two adjacent time steps may be written as the product
of instantaneous (tangent) stiffness [ K ]i +1 and incremental displacement {∆D}i +1 at joints. Hence,
[ K ]i +1{∆D}i +1 = {∆R}i +1 = {∆P}i +1
in which {∆P}i +1 : incremental external forces at joint during the time increment.

Thus, the linearly analysis of a structure reflecting the damage state is necessary in the nonlinear
response analysis.

The structural stiffness matrix of a frame structure can be formulated for free joints from
(a) member stiffness matrices in the local coordinate system,
(b) transformation of coordinates,
(c) continuity conditions of displacement at joint,
(d) equilibrium condition of forces at joint,
(e) separating the free joint and the support joint displacements and forces,
(f) solution of a set of linear equations for free joint displacement, and then
(g) determination of member end actions using the member stiffness matrix.

Steps (b) to (f) can be processed automatically by considering the geometrical conditions and the
coordinate transformation. Therefore, it is more important to formulate stiffness matrices of structural
members considering their damaged state and the formulation of a member stiffness matrix needs a
special attention.

1
9.2 Formulation of Member Stiffness Matrix

A general procedure to formulate a member stiffness matrix is outlined in this section. The
stiffness matrix of a member can be formulated from a stiffness relation of a statically determinate
member. First, a method is introduced to formulate the member stiffness from a stiffness relation of
a cantilever member. The equilibrium matrix is used in the derivation.

Flexibility of Arbitrary Cantilever: For a member of given stiffness distribution, fixed at the start
end and free at the terminal end, free end displacement {d 2 }i can be calculated for any free end
actions { p2 }i , for example, by Castigliano's theorem or the unit load method.

Displacement f jk 2 at the free end in j-direction due to a unit load in k-direction applied at the
free end is calculated by the unit load method, by
mz 2 ,θ z 2
evaluating axial forces n( x) j and n( x) k , bending
py2 , d y 2
member i
moments m( x) j and m( x) k , and shear forces
v( x) j and v( x) k due to unit load in j- and 1 2 px 2 , d x 2
EA( x), EI ( x), GAs ( x)
k-directions, separately;
L n j ( x) ⋅ nk (k ) v j ( x) ⋅ vk (k ) m j ( x) ⋅ mk (k )
f jk 2 = ∫ { + + }dx
0 EA( x) GAs ( x) EI ( x)

where, E : Young’s modulus, A( x) : cross sectional area, As ( x) : effective shear area, and I ( x) :
moment of inertia of section at coordinate x.

Castigliano's Theorem is expressed in the following form to calculate displacement di :

∂U
di =
∂pi
∂ L m( x ) 2 κ v ( x ) 2 n( x ) 2
=( )∫ ( + + )dx
∂pi 0 2 EI ( x) 2GA( x) 2 EA( x)
where U: strain energy stored in the structure expressed in terms of external forces { px 2 , p y 2 , mz 2 }
at the free end, m( x) : bending moment, v( x) : shear, n( x) : axial force, κ : shape factor for shear
deformation, E : Young’s modulus of material, I ( x) : moment of inertia of section, A( x) : cross
sectional area.

The flexibility matrix of a cantilever can be expressed as

{d 2 }i = [ f 22 ]i { p2 }i
where, each element f ij of flexibility matrix [ f 22 ]i represents the displacement in j-direction at the
free end due to unit load applied in the k-direction at the free
end.
{p2}i

Stiffness of Elastic Prismatic Cantilever: For a linearly member i

elastic prismatic cantilever, neglecting shear deformation, 1 2 {d2}i
the flexibility relation is expressed as follows;
EI, EA
L

2
  L 
 EA 0 0 
d x 2    px 2 
  L3 L2   
d y 2  = 0   py2 
θ   3EI 2EI   
 z 2 i  L2 L  mz 2  i
 0 EI  i
 2 EI

Solving this relationship for free end action { p 2 }i , we obtain the stiffness relation for the
cantilever;
−1
{ p2 }i = [ f 22 ]i {d 2 }i
= [k 22 ]i {d 2 }i
−1
[k 22 ]i = [ f 22 ]i
in which [k22 ]i is the stiffness matrix of the cantilever member i .

For a linearly elastic prismatic member; the stiffness matrix is expressed

 EA 
 L 0 0 
 px 2   d x 2 
  12EI 6EI   
 p y2  = 0 − 2  d y 2 
m   L3 L  
 z 2 i  0 − 6 EI 4EI  θ z 2  i
 L2 L  i

Stiffness Matrix of Cantilever with Shear Deformation: The shear deformation increases lateral
deformation d y 2 at the free end due to lateral force p y 2 . The flexibility relation is written as
 L 
 EA 0 0 
d x2     px 2 
  κL L3 L2   
d y 2  =  0 +  py 2 
θ  GA 3EI 2 EI   
 z 2 i   m
 0 L2 L   z 2 i
 2 EI EI  i

By inverting the flexibility relation, the stiffness matrix of a prismatic cantilever can be expressed as
 
 EA 
 0 0 
 px 2   L  d 
   1 12 EI 1 6 EI   
x2

 py 2  = 0 −  d y 2 
m   1 + 2γ L3 1 + 2γ L2   
 z 2 i θ
 γ   z 2 i
 1+ 
1 6 EI 2 4 EI 
 0 −
 1 + 2γ L2 1 + 2γ L  i
6EIκ
where γ= , κ : shape factor for shear deformation.
GAL2

3
Equilibrium Matrix [ H jk ]i : For a straight member i of
py1 py2
length L, without any intermediate loads, member end mz1
actions { p1}i and { p 2 }i in the member coordinates mz2
must satisfy the equilibrium conditions: 1 member i 2
∑p x = 0 : px1 + px 2 = 0
px1 px2
∑p y = 0 : p y1 + p y 2 = 0 L
∑m z @1 = 0 : mz1 + mz 2 + Lp y 2 = 0
in which, x-axis of the member coordinate system is taken in the direction of the straight member.

Writing in a matrix form; the member end forces { p1}i and { p2 }i of member i are related
through an equilibrium matrix [ H12 ]i
{ p1}i + [ H12 ]i { p2 }i = {0}
where,
 p x1 
 
{p1 }i =  p y1 
m 
 z1  i
 px2 
 
{p2 }i =  py2 
m 
 z 2 i
1 0 0
[ H12 ]i = 0 1 0
0 L 1 i
Force [ H12 ]i { p2 }i is a force developed at end 1 due to force { p2 }i acting at end 2.

By the same token,

[ H 21 ]i { p1}i + { p2 }i = {0}
where the member length L in matrix [ H 21 ]i is measured from the terminal point.
1 0 0
[ H 21 ]i = 0 1 0
0 − L 1 i

Note that
{ p1}i = −[ H12 ]i { p2 }i
= [ H12 ]i [ H 21 ]i { p1}i
hence,
[ H12 ] = [ H 21 ]−1

The inverse matrix of an equilibrium matrix is obtained by changing the sign of off-diagonal terms.

4
Rigid Body Displacement: For small displacement
( sin θ ≈ θ , cos θ ≈ 1.0 ), member end displacements θ2z
{d1}i and {d 2 }i through rigid body displacement are dx1
θ1z
also related by the equilibrium matrix; dy2
d x 2 = d x1 dy1 1 2
d y 2 =d y1 + Lθ z1
θ z 2 = θ z1 dx

This relation can be written in a matrix form:

 d x 2  1 0 0   d x1 
    
 d y 2  =  0 1 L   d y1 
θ  0 0 1  θ 
 z 2 i   i  z1  i

or symbolically,
T
{d 2 }i = [ H12 ]i {d1}i

Note that {d 2 }i is a displacement at the member end 2 caused by a rigid body displacement at
member end 1. By the same token, we obtain
T
{d1}i = [ H 21 ]i {d 2 }i

Member Stiffness Matrix: For a free body of member i, the member end displacement and force
relation is formulated. If a starting end 1 of member i is allowed to displace, additional displacement
takes place at the terminal end 2 by a rigid
body movement; i.e., the deformation {e}i
2”
of the member is the difference of terminal
end displacement {d 2 }i and rigid body
T
displacement [ H12 ]i {d1}i at the terminal {e}i
end 2 caused by displacement {d1}i at the
starting end 1: {d2}i
2’
T
{e}i = {d 2 }i − [ H12 ]i {d1}i
1’
{d1}i [H12]Ti {d1}i
The deformation {e}i is related to the
deformation at the free end of a cantilever
1 2
member due to forces { p2 }i applied at the member i
free end. Therefore,
{ p2 }i = [k22 ]i {e}i

Expressing the deformation by the member end displacements {d1}i and {d 2 }i , the member
end force { p2 }i is related to the member end displacements;
{ p2 }i = [k 22 ]i {e}i
T
= [k 22 ]i ({d 2 }i − [ H12 ]i {d1}i )
= −[k 22 ]i [ H12 ]i T {d1}i + [k 22 ]i {d 2 }i
= [k 21 ]i {d1}i + [k 22 ]i {d 2 }i
where

5
 T
[k 21 ]i = −[k 22 ]i [ H12 ]i

For a prismatic member,

 EA 
 L 0 0 
 
 12 EI 6 EI 
[k22 ]i = 0 − 2
 L3 L 
 4 EI 
 0 − 6 EI
 L2 L  i
 EA 
− L 0 0 
 
 12 EI 6 EI 
[k21 ]i = 0 − 3 − 2
 L L 
 6 EI 2 EI 
 0
 L2 L  i

From the equilibrium of member end forces,

{ p1 }i = −[ H 12 ]i { p 2 }i
= −[ H 12 ]i [k 22 ]( −[ H 12 ]Ti {d 1 }i + {d 2 }i )
= [ H 12 ]i [k 22 ][ H 12 ]Ti {d 1 }i − [ H 12 ]i [k 22 ]{d 2 }i
= [k11 ]i {d 1 }i + [k12 ]i {d 2 }i
where,
[k11 ]i = [ H 12 ]i [k 22 ]i [ H 12 ]Ti
[k12 ]i = −[ H 12 ]i [k 22 ]i
= [k 21 ]Ti

For a prismatic member,

 EA 
 L 0 0 
 
 12 EI 6 EI 
[k11 ]i = 0
 L3 L2 
 6 EI 4 EI 
 0
 L2 L  i
 EA 
− L 0 0 
 
 12 EI 6 EI 
[k12 ]i = 0 − 3
 L L2 
 6 EI 2 EI 
 0 − 2
 L L  i

 p1   k11 k12   d1 
  =  
 p2 i  k21 k22  i d 2 i

6
------------------------------
Note:

([ A][ B ]) T = [ B ]T [ A]T

7
9.3 Member with Rigid Ends

It is often necessary to consider rigid zones at member ends. For example, a beam-column
connection may be assumed to be rigid or a structural wall is represented by a column with rigid
beams at each floor. Let us consider the treatment of a member having rigid zones at the ends.

Beam-column Connection Structural Wall

Equilibrium Matrix: The equilibrium matrix [ H AB ] for member AB, not lying in x-direction, can be
defined in a more general form if the lengths of the member are x AB and y AB in x- and y-direction,
respectively;
∑p x = 0 : pxA + pxB = 0
∑p y = 0 : p yA + p yB = 0 y
B

∑m z@ A = 0 : mzA + mzB − pxB y AB + p yB x AB = 0

{ p}A + [ H AB ]{ p}B = {0} yAB
A
The equilibrium matrix for this member is given as
 1 0 0 xAB x

[ H AB ] =  0 1 0
 
 − y AB x AB 1

It should be noted that force { p A } at starting end A is necessary to satisfy the equilibrium when
force { pB } acts at the terminal end B.

Product of Equilibrium Matrices: Consider regions AB and BC connected at B. No external force

is assumed to act at joint B.

If force { pC } acts at member end C of region BC, force { pB '} at B end is necessary to satisfy
the equilibrium in the region;
{ p B ' } + [ H BC ]{ pC } = {0}
{ p B ' } = −[ H BC ]{ pC }

The equilibrium of forces at end B is

{ p B } + { p B ' } = {0}
{ p B } = −{ p B '} = [ H BC ]{ pC }

8
In other words, force { pB } is a force developed at
joint B due to force { pC } acting at joint C.
{ pC }
C
By the same token, force { p A '} is developed by
member end force { pB } acting at member end B to
B
satisfy the equilibrium of forces in member AB.
{ p A ' } = −[ H AB ]{ p B } A { pC }

= −[ H AB ][ H BC ]{ pC } {pA '} C

= −[ H AC ]{ pC }
B
Therefore, { pB }
{ pB ' }
[ H AC ] = [ H AB ][ H BC ]
A
 1 0 0  1 0 0 B

=  0 1 0   0 1 0  {pA '}

 − y AB x AB 1   − yBC xBC 1 
 1 0 0

= 0 1 0 
 −( y AB + yBC ) ( x AB + xBC ) 1 
y C

y BC
 1 0 0
B
y AC

= 0 1 0  A y AB
 − y AC x AC 1 
x
xAB xBC

The equilibrium matrix [ H AC ] of a region ABC x AC

combing regions AB and BC can be expressed as the
product of corresponding equilibrium matrices [ H AB ] and [ H BC ] . The product matrix can be
formulated by adding the corresponding off-diagonal elements of the element matrices.

Rigid Zones and Equilibrium Matrices: Consider a member consisting of an elastic middle part BC
and rigid zones AB and CD at the two ends. The two rigid zones may not be in line with the elastic
part. The local coordinate system is defined for the middle elastic part. No external loads acts within
compound region of ABCD.

y D
A

C yCD
yAB B
yBC

x
xAB xBC xCD

The equilibrium matrix [ H AB ] for rigid zone AB can be defined if the lengths of the rigid zones
are x AB and y AB in x- and y-directions:

9
  1 0 0
[ H AB ] =  0 1 0
 
 − y AB x AB 1

Similarly, for the middle elastic part BC and rigid zone CD, equilibrium matrices [ H BC ] and
[ H CD ] can be formulated:
 1 0 0
[ H BC ] =  0 1 0
 
 − y BC x BC 1
 1 0 0
[ H CD ] =  0 1 0
 
 − y CD xCD 1

Member Stiffness Matrix of Elastic Part: Consider a {p2}BC

cantilever member; fixed at start end B and free at
terminal end C. A flexibility matrix [ f 22 ]BC of the middle
elastic part can be formulated by calculating deformation 1 2
{d 2 }BC at free end 2 caused by free end force { p2 }BC . BC
{d2}BC
{d 2 }BC = [ f 22 ] BC { p2 }BC

The relation can be inverted to express force { p2 }BC

required to deform the terminal end by {d 2 }BC ;
{ p2 }BC = [k 22 ] BC [d 2 }BC
For a prismatic member,
 EA 
 0 0 
 x BC 
12 EI 6 EI 
[k 22 ] BC = 0 − 2
 3
x BC x BC 
 6 EI 4 EI 
 0 − 2
 x BC x BC 

As discussed in Section 6.2 "Formulation of Member Stiffness Matrix, member stiffness

sub-matrices of the middle elastic part BC are formulated;
[k11 ] BC = [ H 12 ] BC [k 22 ] BC [ H 12 ]TBC
py1 py2
= [ H BC ][k 22 ] BC [ H BC ]T
[k12 ] BC = [k 21 ]TBC mz1 mz2
1 BC 2
= −[ H 12 ] BC [k 22 ] BC
= −[ H BC ][k 22 ] BC px1 px2
L
 p1   k11 k12   d1 
  =  
 p 2  BC k 21 k 22  BC d 2  BC

10
Member Stiffness Matrix of Compound Member: Consider a cantilever member ABCD (rigid zone
AB, elastic zone BC and rigid zone CD) with fixed end A and free end D. Let force { pD } be
member end force at end D of rigid zone CD,
{pD}
and { p2 }BC be member end force at end C of {p2}BC={pC}
{pB} D {dD}
elastic part BC. These two forces are related by A
the equilibrium matrix [ H CD ] as follows; B C

{ p 2 }BC = [ H CD ]{ p D }
-{pB}
-{pC}
The rigid body displacement {d D } at D
caused by displacement {dC } at end C (or terminal end of elastic part BC) is expressed using
equilibrium matrix [ H CD ] ;
{d D } = [ H CD ]T {d C }
= [ H CD ]T {d 2 }BC

The region AB is rigid and does not deform. Therefore, terminal member end displacement
{d 2 }BC of part BC is expressed by the flexibility [ f 22 ]BC of part BC and terminal end force { p2 }BC :
{d 2 }BC = [ f 22 ] BC { p2 }BC

Using the relation of displacements at nodes C and D, the flexibility [ f DD ] of ABCD as a

cantilever fixed at A can be expressed as follows;
{d D } = [ H CD ]T {d 2 }BC
= [ H CD ]T [ f 22 ] BC { p 2 }BC
= [ H CD ]T [ f 22 ] BC [ H CD ]{ p D }
= [ f DD ]{ p D }
where
[ f DD ] = [ H CD ]T [ f 22 ] BC [ H CD ]

The stiffness sub-matrix [ k DD ] of a cantilever ABCD is obtained by inverting the flexibility [ f DD ]

[k DD ] = [ f DD ]−1
= ([ H CD ]T [ f 22 ] BC [ H CD ]) −1
= [ H CD ]−1 [ f 22 ]−BC1 ([ H CD ]T ) −1
= [ H CD ]−1 [k 22 ] BC ([ H CD ]T ) −1

Note that the inverse of an equilibrium matrix can be obtained by changing the sign of off-diagonal
elements; i.e.,
 1 0 0
[ H CD ] =  0
−1
1 0
 
 y CD − xCD 1
and also the inverse of a product of matrices is the product of inverse matrices reversing the order of
product;
([ A][ B]) −1 = [ B]−1[ A]−1
provided each element matrix has an inverse matrix.

Other member stiffness sub-matrices are obtained as follows by noting:

11
 [ H AD ] = [ H AB ][ H BC ][ H CD ]
[k DD ] = [ H CD ]−1 [k 22 ] BC ([ H CD ]T ) −1
[ H AD ]T = ([ H AB ][ H BC ][ H CD ]) T = [ H CD ]T [ H BC ]T [ H AB ]T
and
[k11 ]BC = [ H BC ][k 22 ] BC [ H BC ]T
then
[k AA ] = [ H AD ][k DD ][ H AD ]T
= [ H AB ][ H BC ][[ H CD ][ H CD ]−1 [k 22 ] BC ([ H CD ]T ) −1 [ H CD ]T [ H BC ]T [ H AB ]T
= [ H AB ][ H BC ][k 22 ] BC [ H BC ]T [ H AB ]T
= [ H AB ][k11 ] BC [ H AB ]T
[k AD ] = [k DA ]T
= −[ H AD ][k DD ]
= −[ H AB ][ H BC ][k 22 ] BC ([ H CD ]T ) −1
= [ H AB ][k12 ] BC ([ H CD ]T ) −1
or compound member AD consisting of
rigid zones AB and CD at ends and {PA } {PD }
elastic part BC, member stiffness matrix
D
relations in local coordinate system are A
expressed as follows; {d D }
{d A } B C
{ p A } = [k AA ]{d A } + [k AD ]{d D }
{ p D } = [k DA ]{d A } + [k DD ]{d D }

It should be noted that the sub-matrices [k IJ ] of the compound member could be easily derived
from the member stiffness sub-matrices [k ij ]b of the elastic part by transformation operation.

12
9.4 Member with Flexible Ends

A member AB has springs AA’ and BB’ at both ends. y kθ kx

Each spring has stiffness for axial deformation, lateral x
deformation and rotation with spring constants
k x , k y and kθ , respectively. The length of the spring is A A’
none.
ky
The flexibility matrix of spring at A’ or B’ is expressed as

1  Spring at AA’
 0 0
 kx 
  A’ B’
1
[ f AA ' ] =  0 0 A B
 ky  EI , EA
 1
0  L
0
 kθ  AA '
1 
 0 0
 kx 
 1 
[ fB'B ] =  0 0
 ky 
 1
0 0 
 kθ  B ' B

The equilibrium matrices of parts AB and A’B are equal because no length is given at the springs.
[ H AB ] = [ H A ' B ]
1 0 0
=  0 1 0 
 0 LAB 1 
{ pB }
and the equilibrium matrix of part B’B is unit A
matrix; A’
{d B }
B’
1 0 0  B
[ H B ' B ] =  0 1 0 
 0 0 1  L

For a cantilever member AB fixed at end A, internal forces at end A’ induced by external force
{ pB } is given as
{ p A ' } = [ H A ' B ]{ pB }

The deformation of spring {d A ' } at AA’ is given as

{d A ' } = [ f AA ' ]{ p A ' }
= [ f AA ' ][ H A ' B ]{ pB }
The rigid body displacement at end B due to the deformation of the spring at A end is
[ H A ' B ]T {d A ' } = [ H A ' B ]T [ f AA ' ][ H A ' B ]{ pB }

13
 The displacement at end B’ due to the deformation of middle region A’B’ caused by external force
{ pB } is given as
[ H B ' B ]T [ f 22 ] A ' B ' [ H B ' B ]{ pB } = [ f 22 ] A ' B '{ pB }

The displacement at B end due to the deformation of spring B’B caused by external force { pB }
is given by
[ H B ' B ]T [ f B ' B ][ H B ' B ]{ pB } = [ f B ' B ]{ pB }

The total displacement at end B caused by the external force { pB } is the sum of the
contributions of springs AA’ and B’B and middle elastic region A’B’;
{d B } = [ H A ' B ]T [ f AA ' ][ H A ' B ]{ pB } + [ f 22 ] A ' B '{ pB } + [ f B ' B ]{ pB }
= ([ H A ' B ]T [ f AA ' ][ H A ' B ] + [ f 22 ]A ' B ' + [ f B ' B ]){ pB }

The flexibility matrix [ f BB ] is given as

[ f BB ] = [ H A ' B ]T [ f AA ' ][ H A ' B ] + [ f A ' B ' ] + [ f B ' B ]

The stiffness matrix [k BB ] of the cantilever can be obtained by inverting the flexibility matrix
[ f BB ] ;
[k BB ] = [ f BB ]−1

For the member AB, the stiffness sub-matrices

are obtained as { pA} { pB }
[k AA ] = [ H AB ][k BB ][ H AB ] T
{d B }
[k AB ] = −[ H AB ][k BB ]
{d A}
{ p A } = [k AA ]{d A } + [k AB ]{d B } A A’ B’ B
{ pB } = [k BA ]{d A } + [k BB ]{d B }

For a special case where no resistance is given in the spring, the spring constant can be set to
be zero in the member stiffness relation.

14
9.5 Member Stiffness based on Stiffness of Simply Supported Member

A simply supported member is often used as a basic statically determinate system in formulating
the member stiffness matrix. Deformation of a member can be expressed by the axial deformation e,
member end rotations θ A and θ B , and member deformation angle RAB .

Member end displacements of a simply supported

member AB are extension e and member end y
rotations θ A and θ B . These displacements are θ z2
related with displacements {d1} and {d 2 } at the RAB
member ends expressed in the member coordinates: θA
eAB = − d x1 + d x 2 d x1
θ z1 θB d y2
θ A = θ z 1 + ( d y1 − d y 2 ) / L
d y1
θ B = θ z 2 + ( d y1 − d y 2 ) / L 1 2 dx2 x
member i
In a matrix form;
   
 −1 0 0 1 0 0
eAB     d x1    d x2 
  1   1  
θ A  =0 1   d y1  +  0 − 0 d y 2 
θ   L     L   
 B i  1  θ z1   1  θ z 2 i
0 0 0 − 1
 L  i  L  i

The relation can be written in a matrix form:

{eAB }i = [ B1 ]i {d1}i + [ B2 ]i {d 2 }i

Member end forces in the member coordinates

can be expressed by member end forces of the mB
simply supported member AB;
p x1 = − p AB mA
pAB
p y1 = ( m A + m B ) / L
m z1 = m A mz1 mz2
p x 2 = p AB px1
p y 2 = −( m A + m B ) / L px2
py1 py2
mz 2 = mB
or in a matrix form:
 p x1  − 1 0 0   p AB 
     
 p y1  =  0 1 / L 1 / L   m A 
m   0 1 0  i  m B  i
 z1  i 
 p x 2  1 0 0   p AB 
     
 p y 2  = 0 − 1 / L − 1 / L   m A 
m   0 0 1  i  m B  i
 z 2 i 
or

15
 {p1 }i = [B1 ]i T {p AB }i
{p 2 }i = [B2 ]i T {p AB }i

Member Stiffness Matrix in Local Coordinates: The incremental member end force-deformation
of a simply supported member may be expressed in a stiffness matrix form;
{p AB } = [k AB ]{e AB }
 p AB 
 
{ p AB } =  m A 
m 
 B
e AB 
 
{e AB } =  θ A 
θ 
 B

For a linearly elastic prismatic member,

 EA 
 L 0 0 
 4 EI 2 EI 
[k AB ] =  0 
 L L 
 0 2 EI 4 EI 
 L L 

The member stiffness matrix in the local coordinate system is obtained by the transformation of
forces and displacements;
 p1   d1 
  = [B1 , B2 ]i [k AB ]i [B1 , B2 ]i  
T

 p2 i d 2  i
k k12   d1 
=  11   
k 21 k 22  i d 2 i

in which
[k11 ]i = [B1 ]i T [k AB ]i [B1 ]i
[k12 ]i = [k 21 ]i T = [B1 ]i T [k AB ]i [B2 ]i
[k 22 ]i = [B2 ]i T [k AB ]i [B2 ]i

16
Member with Rigid Zones: If rigid zones of length
λ A LAB and λ B LAB are attached at member ends, the
transformation of deformation and force must be A' B'
considered.

C MB’
For a simply supported flexible part A'B', member
end moments mA' and mB' and member end rotations MA’
θ A' and θ B' are defined. For a simply supported total
member AB, member end moments mA and mB and
member end rotation θ A and θ B are defined. Let the
θB’ B'
length of the total member be LAB , rigid zones λ A LAB A RA’B’ B
and λ B LAB . The inflection point is denoted by C. θB
θA D
θA’
Looking at a triangle AA'C of the deformed shape, A'
we have the following relation among angles,
MB’
θ A' = R A' B ' + θ A MB
in which C
θ A λ A LAB + θ B λ B LAB
R A' B ' =
(1 − λ A − λ B )LAB
MA MA’
Therefore, λ A LAB (1 − λ A − λB ) LAB λB LAB
(1 − λB )θ A + λBθ B
θ A' =
1 − λ A − λB

Similarly,
λ Aθ A + (1 − λ A )θ B
θ B' =
1 − λ A − λB

In a matrix form,
θ A'  1  (1 − λB ) λB  θ A 
 ' =   
θ B  1 − λ A − λB  λ A (1 − λ A )  θ B 
or
{θ '} = [A] {θ }
For the moments, member end moments mA and mB are expressed by moments m A ' and
mB ' at the ends of the middle elastic region, by using constant shear acting throughout the member;
(mA ' + mB ' )
mA = mA ' + (λ A LAB )
(1 − λ A − λB ) LAB
(1 − λB )mA ' λ A mB '
= +
(1 − λ A − λB ) (1 − λ A − λB )
(mA ' + mB ' )
mB = mB ' + (λB LAB )
(1 − λ A − λB ) LAB
λB m A ' (1 − λ A )mB '
= +
(1 − λ A − λB ) (1 − λ A − λB )
In a matrix form,

17
 mA  1  (1 − λB ) λ A  m ' A 
 =   
mB  1 − λ A − λB  λB (1 − λ A )  m 'B 
{m} = [ A]T {m'}
The member end rotation and moment stiffness relation of a simple member is expressed as
 4 EI 2 EI 
mA '   L L  θ A ' 
  =   
mB ' i  2 EI 4 EI  θ B ' 
 L L  i
{m'} = [k ']{θ '}
The stiffness matrix of a simple member having rigid ends can be assembled by using the
following relations:
θ A'  1  (1 − λB ) λB  θ A 
 ' =   
θ B  1 − λ A − λB  λ A (1 − λ A )  θ B 
mA  1  (1 − λB ) λ A  m ' A 
 =   
mB  1 − λ A − λB  λB (1 − λ A )  m 'B 
 4 EI 2 EI 
mA '   L L  θ A ' 
  =   
mB ' i  2 EI 4 EI  θ B ' 
 L L  i

In a symbolic format,
{m} = [A]T {m'}
= [ A] [k ']{θ '}
T

= [ A] [k '][ A]{θ }
T

18
 Home Assignment No. 5
2002-03-06
Otani, S.

Formulate a member stiffness sub-matrices [k11], [k12], [k21] and [k22] for a beam having rigid ends
and rotational springs as shown below. Ignore the axial deformation and shear deformation, and
consider only flexural deformation of the elastic member and rotation of rotational springs. Use
member end forces and displacements in the local coordinate system.

Use the following symbols; E: Young's modulus, I: moment of inertia of prismatic section, L: total
length of member, and λ1 L : length of rigid zone at starting end, and λ2 L : length of rigid zone at
terminal end, kθ 1 : spring constant of the rotational spring at the starting end 1, kθ 2 : spring constant
of the rotational spring at the terminal end 2.

p 
{ p1 } =  1 y 
m1z  y

p 
{ p2 } =  2 y 
m 2 z 
k1θ EI k 2θ x
d  2
{d1 } =  1 y  1
θ 1z  A B C D
d 
{d 2 } =  2 y 
θ 2 z  λ1 L (1 − λ1 − λ2 ) L λ2 L
 p1   k11 k12   d 1 
 =  
 p 2  k 21 k 22  d 2 

Each matrix must be defined.

(a) Calculate [k 22 ] BC without the rotational springs at the two ends.
(b) Calculate [k 22 ] BC including the rotational springs at the two ends,
(c) Express [k 22 ] AD and define all related matrices necessary to calculate [k 22 ] AD . Matrix
operation is not required.
(d) Express [k11 ] AD and [k12 ] AD . Define all related matrices necessary to calculate [k11 ] AD
and [k12 ] AD . Matrix operation is not required.

19
(1) Stiffness [k22 ]BC without any rotational spring
 12 EI 6 EI 
 L3 BC −
L2BC 
[k22 ]BC = 
 6 EI 4 EI 
 − L2 LBC 
 BC

(2) Calculation of stiffness matrix of the part consisting of two rotational spring and mid elastic
element.
The flexibility matrix [ F22 ]BC of the mid elastic member BC
 L3BC L2BC 
 
3EI 2 EI 
[ f 22 ]BC = [k22 ]−BC1 = 2
 LBC LBC 
 2 EI EI 
The flexibility matrix [ f 22 ]B 'C ' of the part consisting of the two springs and mid elastic element;
[ f 22 ]B 'C ' = [ H BC ' ]T [ f 22 ]B ' B [ H BC ' ] + [ H CC ' ]T [ f 22 ]BC [ H CC ' ] + [ f 22 ]CC '
 L3BC L2BC 
0 0  0 0 
1 LBC   0   3EI 2 EI  
 1
= 1   + + 1 

0 1  0 1   L2BC LBC  0
L
 kθ 1   BC  2 EI  kθ 2 
EI 
 L3BC L2BC L2BC LBC 
 3EI k+ +
 θ1 2 EI kθ 1 
= 2
 LBC LBC LBC 1 1 
 + + + 
 2 EI kθ 1 EI kθ 1 kθ 2 
Inverting the flexibility matrix [ f 22 ]B 'C ' , the stiffness matrix of the part consisting of the two springs
and mid elastic element;
 LBC 1 1 L2BC LBC 
 + + − − 
[k22 ]BC =
1  EI kθ 1 kθ 2 2 EI kθ 1 
L 4
L L3 3
L2  L2BC LBC L3BC L2BC 
+
BC
2 2
+ BC + BC
BC
 − − + 
12 E I 3EIkθ 1 3EIkθ 2 kθ 1kθ 2  2 EI kθ 1 EI kθ 1 
(3) Calculation of stiffness matrix [k22 ] AD of the entire member with rigid zones at the ends.
[k22 ]AD = [ H CD ]−1[k22 ]B 'C ' ([ H CD ]T ) −1
and
 1 0
[ H CD ]−1 =  
 − LCD 1 
1 − LCD 
([ H CD ]T ) −1 = 
0 1 
Therefore,

20
 1
[k22 ]AD = 4 3
L L L3 L2
+
BC
2 2
+ BC + BC
BC
12 E I 3EIkθ 1 3EIkθ 2 kθ 1kθ 2
 LBC 1 1 (2 LCD + LBC ) LBC LCD + LBC LCD 
 + + − − − 
EI kθ 1 kθ 2 2 EI kθ 1 kθ 2
× 
 (2 LCD + LBC ) LBC LCD + LBC LCD {3LCD ( LCD + LBC ) + L2BC }LBC ( LCD + LBC ) 2 L2CD 
− − − + + 
 2 EI kθ 1 kθ 2 3EI kθ 1 kθ 2 

(3) Other stiffness sub-matrices [ k11 ] AD and [ k12 ] AD

[k11 ]AD = [ H AD ][k22 ] AD [ H AD ]T

1
=
L4BC L3BC L3 L2
2 2
+ + BC + BC
12 E I 3EIkθ 1 3EIkθ 2 kθ 1kθ 2
 LBC 1 1 (2 LCD + LBC − L) LBC LCD + LBC − L LCD − L 
 + + − − − 
EI kθ 1 kθ 2 2 EI kθ 1 kθ 2
× 
 (2 LCD + LBC − L) LBC LCD + LBC − L LCD − L {3L( L − 2 LCD ) + 6( LCD − L ) LBC + 2 L2BC }LBC ( LCD + LBC − L) 2 ( L − LCD ) 2 
− − − + + 
 2 EI kθ 1 kθ 2 6 EI kθ 1 kθ 2 

[k12 ] AD = −[ H AD ][k22 ] AD
1
= 4 3
L L L3 L2
BC
+
2 2
+ BC + BC
BC
12 E I 3EIkθ 1 3EIkθ 2 kθ 1kθ 2
 LBC 1 1 (2 LCD + LBC ) LBC LCD + LBC LCD − L 
 + + − − − 
EI kθ 1 kθ 2 2 EI kθ 1 kθ 2
× 
 (2 LCD + LBC ) LBC LCD + LBC LCD − L {6 LCD ( LCD − L) + 3(2 LCD − L) LBC + 2 L2BC }LBC ( LCD + LBC )( LCD + LBC − L) LCD ( LCD − L) 
− − − + + 
 2 EI kθ 1 kθ 2 6 EI kθ 1 kθ 2 

21
 Chapter 10. Member Stiffness Models

10.1 Member Stiffness Model

The modeling of a reinforced concrete structure at the material level, such as the finite element
method analysis, allows the representation of details of the structural geometry and material
properties. In the past decade, a great progress has been made in the field of constitutive modeling
of plain and reinforced concrete under multiaxial loading. Despite this progress, the computational
and memory requirements of such a modeling have restricted its application to the analysis of
individual members or their sub-assemblage. The efforts to extend the application to the dynamic
analysis of a small reinforced concrete structure have not been successful. Therefore, the
member-by-member modeling is normally used in the nonlinear response analysis of a reinforced
concrete structure.

In the response of a structure in the nonlinear range, piece-wise linear response is assumed and
the member stiffness changes with the development of damage along the member. Therefore,
member stiffness matrix needs to be re-evaluated and re-formulated with a development of new
damage.

The damage distribution within a member

is affected by the distribution of stress within Bending Moment
the member; namely, the stress caused by under Gravity
gravity loads affects the damage distribution Loads
along the member. This gravity load effect is
important in a building designed with
relatively small earthquake forces; the
yielding may take place in the middle region
of a member. Unfortunately, the effect of Bending Moment
gravity loading cannot be treated in a simple under
Earthquake
model at the moment. The effect of gravity
loading is normally ignored in the laboratory
testing of structural members as well as in
the nonlinear response analysis.
Consequently, moment is normally assumed
to distribute linearly along a member. Bending Moment
under combined
Although the moment is assumed to Gravity and
distribute linearly within a member, where Earthquake Loads
the largest moment occurs at member ends
under earthquake-induced forces, the
damage (inelastic deformation) does not concentrate at the critical section, but rather spreads along
the member. Many cracks develop in the middle part of a member. Therefore, the distribution of
stiffness within a member needs to be
modeled in the nonlinear analysis. Many
member models have been proposed in the
past to represent the distribution of stiffness
along the member reflecting the
development of damage.

For the formulation of a member stiffness

matrix, let us consider a simply supported
member AB of any stiffness distribution within the member. The relation between incremental
member end forces and deformations needs to be formulated; i.e., the relation between incremental
member end moments, ∆mA and ∆mB , axial force, ∆p AB , and incremental member end rotations,
∆θ A and ∆θ B , axial deformation, ∆eAB , at member ends. In a matrix form, the elements of the
instantaneous member stiffness matrix must be evaluated:
1
 ∆mA ∆mB
∆p AB   k11 k12 k13  ∆e AB 
    
 ∆m A  = k 21 k 22 k 23   ∆θ A 
 ∆m   k ∆p AB
 B  i  31 k 32 k 33  i  ∆θ B  i
∆mz1 ∆mz 2
In a linearly elastic structural analysis, a
∆p x1
prismatic member, such as a beam or column, is
∆p x 2
represented by a straight element passing ∆p y1 ∆p y 2
through "the geometrical centroid" of the section
because the longitudinal strain under bending is zero at the geometrical centroid. Therefore, a
member does not develop axial deformation under bending moment acting along the member, nor
an axial force causes any member end rotation. In other words, there is no interaction between axial
and rotational response in a linearly elastic stage.

With flexural cracks forming along a reinforced concrete member due to bending, the neutral axis
shifts from the geometrical centroid into the compression side. The tensile strain is developed at the
geometric centroid of previously uncracked section, causing an elongation of the member measured
at the centroid of the section. Such elongation of a member is also measured in member tests under
bending in the laboratory. This phenomenon should be recognized. Very few model recognizes the
interaction of axial and bending response.

Strain before Strain after

Section
Cracking Cracking

In a nonlinear analysis of a frame structure, however, this axial and rotational interaction is
normally ignored. Therefore, member end moment-rotation relation and member end axial
force-elongation relation are treated separately; i.e., it is assumed that k12 = k 21 = 0 and
k13 = k31 = 0 .

Once the moment-axial force interaction is ignored, stiffness element k11 represents the axial
force-deformation relation of the reinforced concrete Yielding
member under uniaxial loading and is determined Stress
from the axial force-axial deformation relationship
observed in uni-axial tests. The relationship in tension
Ccompresson
is linear up to initial tensile cracking, and then the
stiffness gradually deteriorates with additional
cracking, followed by yielding of all longitudinal
reinforcement. The relationship in compression is Strain
similar to that of concrete in compression. The axial Tension
stress-strain relation is normally considered to be
linear for a practical range of analysis, especially in Yielding
the analysis of low-rise buildings, where axial stress
due to the overturning effect of earthquake forces is relatively small.

If the interaction of bending moment and axial force interaction, the member end moment and

2
rotation relation should be defined; i.e., stiffness elements k 22 , k23 , k 32 and k33 must be defined.
The symmetry of a stiffness matrix (the reciprocal theorem) gives k 23 = k 32 , and three independent
stiffness elements k 22 , k23 and k33 must be determined. The test of reinforced concrete members
is carried out under a prescribed loading history. It is not possible to vary the ratio of member end
moments because the combination is infinite. Therefore, the member end moments are chosen to be
same in the test, and the anti-symmetric bending moment distribution is developed in a specimen
with the inflection point at the mid-span. The member end moment and rotation relation may be
determined to define diagonal elements k 22 and k33 , but the off-diagonal element k23 may not be
defined. A member model is necessary to define the member stiffness matrix.

Yielding
e AB
Moment

mA
pAB
θB
θA
Cracking
mB=mA

Rotation
Member end moment-rotation relation from tests

3
10.2 Fiber Model

A member may be divided into short segments, and boundary section between adjacent short
segments may be divided into small element, where the Bernoulli’s assumption of plane section
remaining plane after deformation may be used as kinematic constraints. Rotations about two axis
and elongation at the centroid may express the degrees of freedom at the section. Such a model is
generally called a fiber model and useful to represent flexural behavior of a member.

y
y

z
x

z
Coordinate System of Fiber Model

The fiber model may be considered as a simple finite element method applied to the
one-dimensional continuum. A member coordinate system may consist of x-axis in the direction of
the member, y-axis in the vertical direction, and z-axis in the horizontal direction. A section is divided
into small elements (filaments or fibers) by lines parallel to the z-axis and y-axis. Each fiber
represents either concrete or steel reinforcement. Nonlinear uniaxial stress-strain relation of the
material is assigned to the center of each fiber. Shear deformation is normally ignored in this
formulation.

Stress and Strain in Section: Plane section before bending y

is assumed to remain plane after bending. The normal tensile
strain ε ( y, z ) at ( y, z ) in the section is expressed by the
ε 0 at the geometric centroid ( y = 0, z = 0)
tensile strain y

and curvatures φ y and φ z (positive counter-clockwise

rotation about the coordinate axis) about the centroidal axes y z
and z;
ε x ( y, z ) = ε 0 ( x) − y φ z ( x) + z φ y ( x) z 0

The incremental tensile stress ∆σ x ( y, z ) in the fiber is

t
evaluated by using tangent modulus of elasticity Ex ( y, z ) of
the material at coordinate ( y, z ) at section x ;
∆σ x ( y, z ) = Ext ( y, z ) ⋅ ∆ε x ( y, z ) Coordinate System

Stiffness of Section: The incremental tensile force

σ
∆n( x) of section at x is evaluated by summing the ∆σ E’
incremental tensile stress ∆σ ( y, z ) over the section;

∆ε ε
4
 ∆n( x) = ∫
Section
∆σ x ( y, z )dA

= ∫
Section
Ext ( y, z ) ⋅ ∆ε x ( y, z )dA

= ∫
Section
Ext ( y, z ) ⋅ {∆ε 0 ( x) − y ∆φ z ( x) + z ∆φ y ( x)}dA

= ∫
Section
Ext ( y, z ) ⋅ ∆ε 0 ( x)dA − ∫
Section
Ext ( y, z ) y ∆φ z ( x)dA + ∫
Section
Ext ( y, z ) z ∆φ y ( x)dA

= ∆ε 0 ( x) ∫
Section
Ext ( y, z )dA − ∆φ z ( x) ∫
Section
Ext ( y, z ) ydA + ∆φ y ( x) ∫
Section
Ext ( y, z ) z dA

Similarly, the incremental bending moment about y-axis is calculated by summing up the
contribution of fiber stresses to the moment;
∆m y ( x) = ∫
Section
∆σ x ( y, z ) zdA

= ∫
Section
Ext ( y, z ) ⋅ ∆ε ( y, z ) ⋅ zdA

= ∫
Section
Ext ( y, z ) ⋅ {∆ε 0 ( x) − y ∆φ z ( x) + z ∆φ y ( x)} ⋅ zdA

= ∫
Section
Ext ( y, z ) ⋅ z ∆ε 0 ( x)dA −
Section
∫ Ext ( y, z ) ⋅ yz ∆φ z ( x)dA + ∫
Section
Ext ( y, z ) z 2 ∆φ y ( x)dA

= ∆ε 0 ( x ) ∫ E ( y, z ) ⋅ zdA − ∆φ z ( x) ∫ E ( y, z ) ⋅ yzdA + ∆φ y ( x) ∫
t t
x x Ext ( y, z ) z 2 dA
Section Section Section

The incremental moment about z-axis is evaluated by summing up the moment contributions of
stresses in fiber elements;
∆mz ( x) = − ∫
Section
∆σ x ( y, z ) ydA

=− ∫
Section
Ext ( y, z ) ⋅ ∆ε ( y, z ) ⋅ ydA

=− ∫
Section
Ext ( y, z ) ⋅{∆ε 0 ( x) − y ∆φ z ( x) + z ∆φ y ( x)} ⋅ ydA

=− ∫
Section
Ext ( y, z ) ⋅ y∆ε 0 ( x)dA + ∫
Section
Ext ( y, z ) ⋅ y 2 ∆φ z ( x)dA − ∫
Section
Ext ( y, z ) ⋅ yz ∆φ y ( x)dA

= −∆ε 0 ( x) ∫ E ( y, z ) ⋅ ydA + ∆φ z ( x) ∫ E ( y, z ) ⋅ y dA − ∆φ y ( x) ∫
t t 2
x x Ext ( y, z ) yz dA
Section Section Section

The instantaneous stiffness matrix [k s ( x)] relates the incremental internal forces {∆ss ( x)} and
incremental strains {∆ε ( x)} ;
{∆ss ( x)} = {∆m y ( x), ∆mz ( x), ∆n( x)}T
{∆ε s ( x)} = {∆φ y ( x), ∆φ z ( x), ∆ε 0 ( x)}T ;
{∆ss ( x)} = [k s ( x)]{∆ε ( x)}

5
  Ext ⋅ z 2 dA − Ext ⋅ yzdA
∫ ∫ ∫ Ex ⋅ zdA 
t

[k s ( x)] =  − ∫ Ext ⋅ yzdA ∫ Ext ⋅ y 2 dA − ∫ Ext ⋅ ydA

 
 E t ⋅ zdA
 ∫ x − ∫ Ex ⋅ ydA ∫ Ex dA 
t t

For uniaxial bending with m y ( x) = 0 , the stiffness relation is reduced to 2x2.

∆mz ( x)   ∫ Ex ⋅ y dA − ∫ Ex ⋅ ydA  ∆φ z ( x) 
t 2 t

 =  
 ∆n( x)   − ∫ Ext ⋅ ydA   ∆ε 0 ( x) 
∫
t
E x dA 
Actual evaluation of section stiffness is carried out not by integration but by summing up the
contribution from small fiber segments. Zeris and Mahin (1988) pointed out that this formulation
sometimes causes a numerical problem once the maximum section capacity is reached. An iterative
approach is suggested.

The section stiffness may be defined by a hysteresis model once skeleton moment-curvature
relation is estimated under monotonically increasing curvature.

Member Stiffness Matrix: The

principle can be discussed, y
without loosing generality, using ∆mA , ∆θ A ∆mB , ∆θ B
a two-dimensional plane frame
member under uniaxial bending A B x
and deformation. A simple
beam is considered as a basic ∆p AB , ∆eAB
statically determinate system.

The tangent stiffness matrix of the member relates the member end forces and displacements
without intermediate loading.
 ∆m A   ∆θ A 
  t  
 ∆mB  = [k m ] ∆θ B 
∆p  ∆e 
 AB   AB 

(1) Stiffness Approach

t
Member stiffness matrix [km ] is evaluated by integrating the tangent section stiffness matrix
[k st ] along the length of the member;
L
[kmt ] = ∫ [ B ( x)]T [k st ][ B( x)]dx
0

where [ B( x)] : matrix relating the section generalized strain increment vector to the member
deformation increment vector;
{∆ε ( x)} = [ B( x)]{∆θ }

For a linearly elastic prismatic member, the transverse incremental displacement is cubic
polynomials of the distance along the member axis, and matrix [ B ( x)] may be evaluated as
 x x 
1  2(3 − 2) 2(3 − 1) 0 
[ B( x)] = l l
l 0
 0 0 
For non-uniform distribution of stiffness, the evaluation of matrix [ B( x)] is the critical problem.

6
 The main shortcoming of the stiffness-based elements is their inability to represent member
behavior near the peak resistance since there exists numerical instability problem. Therefore, the
member may be sub-divided into short segments, and simple flexibility distribution should be
assumed for each short segments. Mahasuverachai and Powell (1982) suggested the use of
flexibility-dependent shape functions that are continuously updated during the analysis.

(2) Flexibility Approach

For a statically determinate member such as a cantilever or a simple beam, section forces (e.g.,
axial force and bending moment) {s ( x)} along a member are precisely defined as a function of
member end forces {Q} .
{s ( x)} = [ N Q ( x)]{Q}
where [ N Q ( x)] : force interpolation functions which define section forces as a function of member
end forces. For a simply supported plane frame member, matrix [ N Q ( x)] is expressed as

x x 
−1 0
[ NQ ] =  l l
 1
 0 0 

A section constitutive law is written in the incremental form at each section;

{∆ε ( x)} = [ f ( x)]{∆s ( x)}
where [ f ( x)] : section flexibility matrix and {ε ( x)} : section deformations.

The compatibility equation is expressed in the following form;

[ F ]{∆Q} = {∆U }
where {U } : nodal displacement of the statically determinate member, and flexibility matrix [ F ] is
defined as
L
[ F ] = ∫ [ N ( x)]T { f ( x)}[ N ( x)]dx
0

The flexibility matrix of a member is evaluated in discretized form. Therefore, the compatibility
between section {e( x)} and member end displacement {U } is maintained in an integral sense
(Coleman and Spacone, 2001);
L
{U } = ∫ [ N Q ( x)]{e( x)}dx
0

The draw back of the flexibility-based formulation is the implementation the flexibility relation in
existing stiffness-based analysis procedure.

The drawback of the fiber model is the difficulty in evaluating member tangent stiffness matrix.
The memory requirement is also significant to keep track of stress and strain levels in each fiber in
the analysis.

References:

Menegotto, M., and P. E. Pinto, “Method of Analysis for Cyclically Loaded RC Plane Frames
Including Changes in Geometry and Non-elastic Behaviour of Elements under Combined Normal
Force and Bending,” Preliminary Report, IABSSE, 1973, Vol. 13, pp. 15 - 22.
Aktan, A. E., et al., “R/C Column Earthquake Response in Two Dimensions,” Journal, Structures
Division, ASCE, Vol. 100, No. ST10, October 1974, 1999 - 2015.

7
Aziz, T. S., “Inelastic Dynamic Analysis of Building Frames,” Research Report R76-37, Department
of Civil Engineering, Massachusetts Institute of Technology, Cambridge, MA, 1976.
Coleman, J., and E. Spacone, “Localization Issues in Force-based Frame Elements,” Journal,
Structural Engineering, ASCE, Vol. 127, No. 11, November 2001, pp. 1257 - 1265.
Kaba, S. and S. A. Mahin, “Refined Modeling of Reinforced Concrete Columns for Seismic Analysis,”
EERC Report 84/03, Earthquake Engineering Research Center, University of California, Berkeley,
1984.
Mahasuverachai, M., and G. H. Powell, “Inelastic Analysis of Piping and Tubular Structures,” EERC
Report 82-27, Earthquake Engineering Research Center, University of California, Berkeley, 1982.
Mark, K., “Nonlinear Dynamic Response of Reinforced Concrete Frames,” Research Report R76-38,
Department of Civil Engineering, Massachusetts Institute of Technology, Cambridge, MA, 1976.
Spacone, E., F. C. Filippou and F. F Taucer, “Fiber Beam-column Model for Non-linear Analysis of
R/C Frames: Part I. Formulation,” Earthquake Engineering and Structural Dynamics, Vol. 25,
1996, pp. 711 - 725.
Spacone, E., F. C. Filippou and F. F Taucer, “Fiber Beam-column Model for Non-linear Analysis of
R/C Frames: Part II. Applications,” Earthquake Engineering and Structural Dynamics, Vol. 25,
1996, pp. 727 - 742.
Zeris, C. and S. A. Mahin, “Analysis of Reinforced Concrete Beam-columns under uniaxial excitation,
Journal, Structures Division, ASCE, Vol. 114, No. ST4, April 1988, pp. 804 - 820.
Zeris, C., and S. A. Mahin, “Behavior of Reinforced Concrete Structures subjected to Biaxial
Excitation,” Journal, Structures Division, ASCE, Vol. 117, No. ST9, September 1991, pp. 2657 -
2673.

8
10.3 Discrete Element Models

In order to overcome difficult problems of variable stiffness along a member, the member can be
sub-divided into short line segments along the length, with each short segment assigned nonlinear
hysteretic characteristics. The nonlinear stiffness can be assigned within a segment, or to flexible
springs at the connection of two adjacent segments.

Rotational Spring
Rigid Element

Flexural Rigidity EI

Discrete spring model Discrete segment model

Discrete Segment Model: A member is divided into short segments, each segment i with uniform
flexural rigidity EI i that varies with a stress history of the segment. Variation of stiffness along the
member can be easily handled by this model if the flexural rigidity can be estimated for each short
segment. Structural walls are often idealized by this model although the interaction of axial
deformation and flexure cannot be considered in this model; i.e., the elongation due to the shift of
neutral axis in the section cannot be considered. More computation effort is required in this model
due to the increased number of degrees of freedom.

A member is sub-divided into (n+1) A 1 2 i n B

elements; element i has constant flexural
rigidity EI i and length ∆xi . A common EIi
local coordinate system may be chosen for
the entire member. The stiffness matrix for Discrete Segment Model
each element using the common local
coordinate system is expressed as
 ∆p1   k11 k12   ∆d 1  y
  =  
∆p2  i k 21 k 22  i ∆d 2  i ∆p y 2
where
1 EIi, EAi ∆p x 2 x
 EAi  ∆px1
 ∆x 0 0 
 i  i 2 ∆m z 2
12 EI i 6 EI i  ∆p y1
[k11 ]i = 0

∆m z 1
( ∆xi ) 3 ( ∆x i ) 2  ∆x i
 6 EI i 4 EI i 
 0 
 ( ∆xi ) 2 ∆xi 
 EAi 
− ∆xi
0 0 
 
12 EI i 6 EI i 
[k12 ]i = [k 21 ]Ti =

0 −
( ∆xi ) 3 ( ∆xi ) 2 
 6 EI i 2 EI i 
 0 − 
 ( ∆x i ) 2 ∆xi 

9
  EAi 
 ∆x 0 0 
 i 
12 EI i 6 EI i 
[k 22 ]i = 0 −
 ( ∆x i ) 3 ( ∆xi ) 2 
 6 EI i 4 EI i 
 0 − 
 ( ∆xi ) 2 ∆xi 

At each node, two elements are connected. Using the {∆p1}i +1 ,{∆d1}i +1
{∆p2 }i ,{∆d 2 }i
continuity condition of element end displacement at the
connection and the equilibrium condition of member end
forces at each node, a stiffness matrix of the member can be
formulated;
Member i Member i+1
{∆p} = [k ]{∆d }

∆p A  k11A1 k12A1 0 ⋅ 0 ⋅ 0  ∆d A 

 ∆p   A1  
 1  k 21 k A1
22+k 12
11 k 12
12 0 0 ⋅ 0   ∆d1 
 ⋅   0 12
k 21 ⋅ ⋅ 0 ⋅ 0  ⋅ 
    
 ∆pi  =  0 0   ∆d i 
i −1,i i ,i +1
0 ⋅ k 22 +k 11 ⋅ ⋅ ⋅
 ⋅   ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ 0  ⋅ 
    
 ∆p n   0 0 ⋅ 0 ⋅ k n −1, n
22 + k11nB k12nB   ∆d n 
    
∆p B   0 0 0 0 0 k 21nB k 22nB  ∆d B 

The linear equation can be re-arranged in the form

∆p e  k ee k ei  ∆d e 
 =  
 ∆pi   k ie k ii   ∆d i 
in which, {∆p e } and {∆d e } are forces and displacements at member ends A and B, {∆pi } and
{∆d i } are forces and displacements at intermediate nodes from 1 to n.

∆p  ∆d 
{∆p e } =  A  {∆d e } =  A 
∆p B  ∆d B 
 ∆p1   ∆d1 
   
{∆pi } =  M  {∆d i } =  M 
∆p  ∆d 
 n  n

If intermediate loads are not considered along a member, the sum of internal forces should be
zero at intermediate joints;
{∆pi } = {0}
hence,
{∆d i } = −[k ii ]−1 [k ie ]{∆d e }
and
{∆pe } = ([k ee ] − [k ei ][k ii ]−1 [k ie ]){∆d e }
= [k ]{∆d e }

10
 This procedure is called "static condensation." A reduced stiffness matrix of 3 x 3 is obtained.
Incremental displacement {∆d i } at internal joints is calculated after the end incremental
displacement {∆d e } is obtained:
{∆d i } = −[k ii ]−1 [k ie ]{∆d e }

Correction of Unbalanced Forces: The F

structural analysis is based on (a) constitutive
relation of members, (b) equilibrium of forces F*
j+1
at joints, and (c) compatibility of displacement
∆p *j +1 Correction of overshooting
at joints. It is generally assumed that the F
j+1
stiffness does not change during a small load
(displacement) increment. However, this ∆F j +1 kj ∆F j +1 = k j ∆D j − ∆p *j +1
assumption is often violated and the Fj
force-deformation relationship deviates from
the linear instantaneous stiffness. If the
constitutive relation is to be satisfied, either
equilibrium or compatibility must be violated ∆D j +1 D
at a joint at the next load (displacement)
increment. Dj Dj+1

It is normally advisable to correct the

element force in accordance with the Correction of Overshooting in Hysteresis Relation
constitutive relation because the violation of
equilibrium can be corrected easily by applying an imaginary external force at the joint.

Due to the unbalanced force ∆pi { } at intermediate joints within a member and {∆p } at the
* *
e
ends, the static condensation must be performed in a more complicated manner;
∆p e  k ee k ei  ∆d e  ∆p e* 
 =   −  *
 ∆pi   k ie k ii   ∆d i  ∆pi 
and {∆pi } = {0} because there acts no external load at the internal nodes. Hence,
{∆p } = [k ]{∆d } + [k ]{∆d }
*
i ie e ii i

{∆d } = [k ] ({∆p }− [k ]{∆d })

i ii
−1 *
i ie e
and,
{∆pe } = [k ee ]{∆d e } + [k ei ]{∆d i } − {∆pe* }
= ([k ee ] − [k ei ][k ii ] [k ie ]){∆d e } − {∆p e* }+ [k ei ][k ii ] {∆pi* }
−1 −1

Application of Unit Load Method: If there are no external loads acting at internal nodes, the
stiffness matrix of a member can be formulated in a simpler manner using unit load method of
calculating displacement under a given loading set.

Suppose a member AB of length L is simply supported at the two ends subjected to incremental
member end moments ∆m A and ∆m B . The member is divided into n segments, not necessarily of
equal length. Length of segment i is ∆xi , flexural rigidity EI i , and distance from A end to the
center of segment i is xi.

Incremental bending moment ∆m x (positive for tension at bottom) at distance x is expressed as

11
 x x
∆m x = − ∆m A (1 − ) + ∆m B ( )
L L
and incremental curvature ∆φ x at the
center of segment i is
∆m x
∆φ x =
EI i
∆xi ∆x Flexural Rigidity EI
for ( xi − ) ≤ x ≤ ( xi + i )
2 2

For a unit moment applied at A end, x

bending moment mux at distance x is
∆mA ∆m x ∆mB
given as
x
mux = −1 +
L
L
Incremental member end rotation ∆θ A at
A end is calculated by unit load method:
L
∆θ A = ∫ ∆φ x mux dx
0
xi + ∆xi / 2

=∑ ∫ ∆φ x mux dx
i xi − ∆xi / 2

∆m x
∆φ x =
EI i

i
A B
xi
∆xi
mux
mA = 1

Calculation of member end rotation

xi + ∆xi / 2 xi + ∆xi / 2 x x x
∫
xi − ∆xi / 2
∆φ x mux dx = ∫ (−1 + ){∆m A (1 − ) + ∆m B ( )} / EI i dx
xi − ∆xi / 2 L L L
∆m A xi + ∆xi / 2 x ∆m B xi + ∆xi / 2 x x
= ∫
EI i i i
x − ∆x / 2
(1 − ) 2 dx −
L ∫
EI i i i L
x − ∆x / 2
( )(1 − )dx
L
= f AAi ∆m A + f ABi ∆m B

12
 1 xi + ∆xi / 2 x ∆x x 1 ∆x
f AAi =
EI i ∫ xi − ∆xi / 2
(1 − ) 2 dx = i {(1 − i ) 2 + ( i ) 2 }
L EI i L 12 L
1 x
xi + ∆xi / 2 x ∆x x
f ABi = −
EI i ∫ ( )(1 − )dx = f BBi + i i
xi − ∆xi / 2 L L EI i L
1 x
xi + ∆xi / 2 ∆x x 1 ∆x
f BBi =
EI i ∫ ( ) 2 dx = i {( i ) 2 + ( i ) 2 }
xi − ∆xi / 2 L EI i L 12 L
∆θ A = (∑ f AAi ) ∆m A + (∑ f ABi ) ∆m B
i i

= f AA ∆m A + f AB ∆m B

Similarly, applying unit moment at end B, bending moment mux at distance x from end A;
x
mux =
L
and incremental rotation ∆θ B at member end B is calculated using the unit load method:
L
∆θ B = ∫ ∆φ x mux dx
o
xi + ∆xi / 2

=∑ ∫ ∆φ x mux dx
i xi − ∆xi / 2

= (∑ f BAi ) m A + (∑ f BBi ) m B
i i

= f BA ∆m A + f BB ∆m B
where,
f BAi = f ABi

∆θ A   f AA f AB  ∆m A 
 =  
∆θ B   f BA f BB  ∆m B 

The stiffness relation at the member end can be expressed by inverting the flexibility relation:
∆m A  k AA k AB  ∆θ A 
 =  
∆m B   k BA k BB  ∆θ B 
in which
f BB
k AA =
f AA f BB − f AB
2

f AB
k AB = k BA = −
f AA f BB − f AB
2

f AA
k BB =
f AA f BB − f AB
2

The number of degrees of freedom of a member is normally reduce to three (two member end
rotations and an extension) using the static condensation technique so that the size of a structural
stiffness matrix is kept small.

Discrete Spring Model: Wen and Janssen (1965) presented a method of analysis for a plane frame

13
consisting of elasto-plastic segments. The mass and flexibility of a member were lumped at the
connecting points on a tributary basis. Powell (1975) suggested the use of a degrading stiffness
hysteresis model for rigid inelastic connecting springs. Shorter segments were recommended in a
region of high bending moment, and longer segments in a low bending moment region.

Flexibility relation of a simply supported

member consisting of short rigid segments
and rotational springs at internal joints is y
expressed as; Rotational Spring
Rigid Element
∆θ A   f AA f AB  ∆m A 
 =   spring i x
∆θ B   f BA f BB  ∆m B 
in which
xi 2 xi
f AA = ∑ (1 − ) fi
i L L
x x
f AB = f BA = ∑ ( i )(1 − i ) f i mB
i L L mA
m(xi)
x
f BB = ∑ ( i )2 fi
i L
where, xi : distance of spring i from A end,
L: length of member, f i : rotational flexibility of spring i. The stiffness relation can be obtained by
inverting the flexibility relation.

References:

Kaba, S. and S. A. Mahin, “Refined modeling of reinforced concrete columns for seismic analysis,”
Earthquake Engineering Research Center, University of California at Berkeley, EERC Report
84-3, 1984.
Powell, G. H., "Supplement to Computer Program DRAIN-2D, Supplement to Report, DRAIN-2D
User's Guide, University of California, Berkeley, 1975.
Wen, R. K., and J. G. Janssen, "Dynamic Analysis of Elasto-Inelastic Frames," Proceedings, Third
World Conference on Earthquake Engineering, Wellington, New Zealand, January 1965, Vol. II,
pp. 713-729.
Zeris, C. and S.A. Mahin, “Analysis of Reinforced Concrete Beam-columns under Uniaxial
Excitation,” Journal of Structural Engineering, ASCE, Vol. 114, No. 4, April 1988, pp. 804 - 820.
Zeris, C., and S.A. Mahin, “Behavior of Reinforced Concrete Structures subjected to Biaxial
Excitation,” Journal of Structural Engineering, ASCE, Vol. 117, No. 9, September 1991, pp.
2657 - 2673.

14
10.4 One-component Model

An elasto-plastic frame structure was often analyzed by introducing a plastic hinge at the location
where the bending moment reached the plastic moment (Berg and DaDeppo, 1960). All plastic
deformation was assumed to occur in the hinge and no strain hardening was considered.

B C P B C P

A D A D
Elastic Response Inelastic Response

This model, called one-component model, was extended for the general use in a nonlinear frame
analysis under earthquake excitation by Giberson (1967); he used nonlinear rotational springs at two
member ends instead of rigid-plastic hinges. All the inelastic deformation of a member was assumed
to concentrate at the member ends, and the middle part was assumed to remain elastic.

mA Elastic Element Elastic

kA kB m Deformation
A
θB

θA L B mB

One Component Model Spring Rotation

If the stiffness properties of each spring can be

evaluated from the material properties and θ
geometry of the member, the flexibility matrix of a
simply supported member can be formulated by
considering the member end rotation as the sum of
an elastic element rotation and spring rotation at Member End Moment-Rotation Relation
each end:
∆θ A = ∆θ Ae + ∆θ Ap
∆θ B = ∆θ Be + ∆θ Bp

For a simply supported elastic element:

 L L 
∆θ Ae   3EI −
6 EI  ∆m A 
 = L  ∆m B 
 ∆θ Be   − L 
 6 EI 3EI 
in which EI: flexural rigidity of section in the elastic region, L: member length. For each inelastic
rotational spring:

15
 ∆m A
∆θ Ap =
kA
∆m B
∆θ Bp =
kB
where kA and kB: instantaneous (tangent) spring constant of rotational springs at A end and B end,
respectively. Therefore, the flexibility relation is written in the form;
 L 1 L 
 + −
∆θ A  3EI k A 6 EI  ∆m A 
 = 
1  
 ∆θ B   − L L
+  ∆m B 
 6 EI 3EI k B 

6 EI 6 EI
Let s A = / k A , and s B = / kB ,
L L
∆θ A  L 2 + s A − 1  ∆m A 
 =   
∆θ B  6 EI  − 1 2 + s B   ∆mb 

The flexibility relation can be inverted to the stiffness relation; i.e.,

∆m A  (6 EI / L) 2 + s B 1  ∆θ A 
 =   
∆m B  ( 2 + s A )( 2 + s B ) − 1  1 2 + s A  ∆θ B 

The spring properties may be determined by assuming the length of a yield hinge zone and a
uniform curvature distribution over the hinge region.
L
In many cases, however, it becomes more desirable to
∆mA
6EI
faithfully simulate a member end moment-rotation relation
of a member observed in the laboratory under a
prescribed loading condition. Such a relation is idealized
and given by a hysteresis model. Therefore, it is more
' '
convenient to use an instantaneous stiffness k A and k B
of the member end moment-rotation relation, in which ∆mA
member end rotation includes elastic rotation (=
∆m A 6 EI / L ) at the member end under anti-symmetric
moment distribution; consider m A = m B in the flexibility ∆mA
relation above, ∆θ A =
kA'
L 1 L
∆θ A = ( + ) ∆m A − ∆mB (= ∆mA )
3EI k A 6 EI
L 1 Member End Moment-Rotation Relation
=( + ) ∆m A
6 EI k A

This relation is now expressed as

1
∆θ A = ( )∆m A
k A'
Therefore,
1 L 1
'
= +
k A 6 EI k A

16
 6 EI 6 EI
or s ' A = 1 + s A , where, s ' A = / k ' A and s ' A = / k'A .
L L

This relation can be substituted in the stiffness relation above:

6 EI
∆m A  L 1 + s B' 1   ∆θ A 
  =   
∆m B  (1 + s A )(1 + s B ) − 1  1
' '
1 + s A'  ∆θ B 

A major advantage of the model is that inelastic member-end deformation depends solely on the
moment acting at the end so that any moment-rotation hysteresis relation can be assigned to the
spring. The stiffness of an inelastic spring is normally defined by assuming an anti-symmetric
moment distribution along a member with the inflection point at mid-span.

This fact is also a weakness of the model because the member-end rotation should be dependent
on the curvature distribution along the member, hence dependent on moments at both member ends.
Consider two cases of moment distribution along a member AB with corresponding to a curvature
distribution shown below;

The inelastic rotations at the A end are given by

shaded areas. For the same moments at A end,
Case II causes larger inelastic rotation at A end.
Consequently, this simple model does not simulate
actual member behavior if the member moment
distribution changes significantly during an
earthquake. Furthermore, it is not rational to lump all
inelastic deformation of a reinforced concrete
member at member ends.

Therefore, the moment-rotation relation at a

member end using this member model tends to
deviate from the actual relation if the stress
distribution becomes different from that assumed in
determining the spring properties. Inelastic member-end rotation
for different moment distributions
Suko and Adams (1971) suggested the use of the
initial location of the inflection point in evaluating spring properties, assuming the inflection point
during an earthquake does not shift much from the initial elastic location. However, once yielding is
developed at one member end, the moment at the other end must increase to resist a higher stress,
moving the inflection point toward the member center. At the same time, a large concentrated
rotation starts to occur near the critical section.

Despite rational criticisms against this simple model, the performance of the one-component
model is expected to be reasonably good for a relatively low-rise frame structure, in which the
inflection point of a column locates reasonable close to mid-height.

A finite size of the plastic regions may be considered in the analysis (Roufaiel and Meyer, 1987).

References:

Berg, G. V., and D. A. DaDeppo, "Dynamic Analysis of Elasto-Plastic Structures," Proceedings,

Engineering Mechanics Division, ASCE, Vol. 86, EM2, April 1960, pp. 35-58.
Giberson, M. F., "The Response of Nonlinear Multi-story Structures subjected to Earthquake
Excitation," EERL Report, Earthquake Engineering Research Laboratory, California Institute of
Technology, Pasadena, 1967.
Roufaiel, M.S.L., and C. Meyer, “Analytical Modeling of Hysteretic Behavior of R/C Frames,” Journal

17
 of Structural Engineering, ASCE, Vol. 113, No. 3, March 1987, pp. 429-444.
Suko, M., and P. F. Adams, "Dynamic Analysis of Multi-bay Multi-story Frames," Journal, Structural
Division, ASCE, Vol. 97, No. ST10, October 1971, pp. 2519-2533.

18
10.5 Multi-component Model

In an effort to analyze frame structures well

into the inelastic range under earthquake
excitation, Clough et al. (1965) proposed a
two-component model: a frame member was
divided into two imaginary parallel elements: an (a) Member
elasto-plastic element to represent a yielding
phenomenon, and a fully elastic element to
representing strain-hardening behavior. The sum
of the flexural rigidity of the two elements was
equal to the initial elastic flexural rigidity, EI. The
two elements were rigidly connected at the two
(b) Member division
member ends, and the member end-moment
was equal to the sum of the moments at the
element ends. When a member-end moment
reached the yield level, a plastic hinge was
inserted at the end of the elasto-plastic element.
Upon unloading from a peak, the plastic hinge
was removed at the yielding end. A member-end
(c) Left end yielding
rotation depended on both member-end
moments.

This model was unique at the time when a frame was Elements I and II
analyzed only for an elastic-perfectly plastic condition. Bilinear
force-deformation relation was made possible to use at Member end moment
member ends by this model.
Element II
Aoyama and Sugano (1968) adapted the two component
model, creating the multi-component model using four parallel Element I
elements to account for flexural cracking, yield levels different
at two member ends, and strain-hardening. The deformation
compatibility of the imaginary components is satisfied only at Member Rotation
their ends. Member rotation

The stiffness matrix of a simply supported

member having four parallel elements can be
formulated by recognizing (a) the rotation at each
member end is common among four elements, and
(b) the moment at each end is the sum of moment
resisted by four elements. In other words, the
(a) Member
member stiffness matrix is the sum of stiffness
matrices of four parallel elements.

For a simply supported elastic member with

flexural rigidity p1 EI ,
∆m A1  p1 EI 4 2 ∆θ A 
 =   (b) Stiffness of divided members
∆m B1  L 2 4 ∆θ B 

For an element with flexural rigidity p2 EI and a plastic hinge at A end, the moment at A end is
known to be zero;
∆m A 2  p 2 EI 0 0 ∆θ A 
 =  
∆m B 2  L 0 3 ∆θ B 

19
Similarly, for an element with flexural rigidity p3 EI and a plastic hinge at B end;
∆m A3  p3 EI 3 0 ∆θ A 
 =  
∆m B 3  L 0 0 ∆θ B 

For an element with plastic hinges at the two ends, the stiffness matrix is zero;
∆m A4  0 0 ∆θ A 
 =  
∆m B 4  0 0 ∆θ B 

From the equilibrium of forces;

∆m A = ∆m A1 + ∆m A2 + ∆m A3 + ∆m A4
∆m B = ∆m B1 + ∆m B 2 + ∆m B 3 + ∆m B 4

Therefore, the stiffness matrix of a simply supported member is written as follows;

∆m A  EI ( 4 p1 + 3 p3 ) 2 p1  ∆θ A 
 =   
 ∆m B  L  2 p1 ( 4 p1 + 3 p2 ) ∆θ B 

Note that p 2 ⋅ p3 = 0 ; an element having a plastic hinge at A end and another element having a
plastic hinge at B end cannot exist simultaneously; i.e., either p 2 or p3 or both should be zero at
a stage.

The multi-component model appears to have a merit; rotation at one end of a member depends
on both member-end moments. In other words, the moment distribution along a member can be
approximately reflected in the analysis. However, the stiffness of the multi-parallel components must
be evaluated under a certain assumed moment distribution. Therefore, the stiffness parameters are
valid only under such a moment distribution, and are bound to be approximate when the moment
distribution becomes drastically different in the analysis.

Giberson (1967) discussed the advantage and disadvantage of the one-component and the
two-component models, and concluded that the one-component model was more versatile than the
two-component model because the two-component model was restricted to the bilinear hysteresis
characteristics.

The ratios p's of element stiffness may be varied with damage if a more general hysteresis
relation is desired for a member. Takizawa (1976) suggested the ratios p's be determined as a
function of member end stiffness under an anti-symmetric moment distribution. The flexibility matrix
including shear deformation is given below:
∆θ A   f11 f 12  ∆m A 
 =  
∆θ B   f 21 f 22  ∆m B 
where, for f B < f A ( p2 = 0) :
3 (1 − γ ) 2
f11 = fA + fB
2+γ (1 + 2γ )( 2 + γ )
(1 − γ )
f12 = f 21 = − fB
1 + 2γ
2+γ
f 22 = fB
1 + 2γ
and for f B > f A ( p3 = 0) ,

20
 2+γ
f11 = fA
1 + 2γ
(1 − γ )
f12 = f 21 = − fA
1 + 2γ
(1 − γ ) 2 3
f 22 = fA + fB
(1 + 2γ )( 2 + γ ) 2+γ

γ : parameter describing the relative contribution of shear deformation (= 6κ EI / GAL2 ). Flexibility

parameters f A and f B are given by a hysteresis model at member ends A and B;
∆θ A
fA =
∆m A
∆θ B
fB =
∆m B
under anti-symmetric bending moment distribution.

The axial load-bending moment interaction may be approximated by changing the yield moment
level of elements as a function of existing axial load.

References:

Aoyama, H., and T. Sugano, "A Generalized Inelastic Analysis of Reinforced Concrete Structures
based on the Tests of Members," Recent Researches of Structural Mechanics, Contribution in
Honor of the 60-th Birthday of Professor Y. Tsuboi, Uno Shoten, Tokyo, 1968, pp. 15-30.
Clough, R. W., K. L. Benuska and E. L. Wilson, "Inelastic Earthquake response of tall buildings,"
Proceedings, Third World Conference on Earthquake Engineering, New Zealand, Vol. II,
Session II, 1965, pp. 68-89.
Giberson, M. F., "The Response of Nonlinear Multi-story Structures subjected to Earthquake
Excitation," EERL Report, Earthquake Engineering Research Laboratory, California Institute of
Technology, 1967.
Takizawa, T., "Notes on Some Basic Problems in Inelastic Analysis of Planar R/C Structures (Part
1)," Transactions, Architectural Institute of Japan, No. 240, February 1976, pp. 51-62.

21
10.6 Distributed Flexibility Model

Once cracks develop in a member, the stiffness becomes non-uniform along the member length.
Instead of dividing a member into short segments, Takizawa (1973,1976) developed a model that
assumed a prescribed distribution pattern of cross-sectional flexural flexibility (reciprocal of flexural
rigidity EI) along the member length. The flexibility matrix of a simply supported member can be
formulated by considering a member with varying section properties EI(x) and (GA/ κ )(x) with
distance x from A end.
∆θ A   f AA f AB  ∆m A 
 =  
∆θ B   f BA f BB  ∆m B 
in which,
x
L (1 − ) 2 L
=∫ L dx + 1 1
f AA
0
EI ( x ) L2 ∫ GA / κ ( x )dx
0

x x
L− ( )(1 − ) L
=∫ L L dx + 1 1
f AB = f BA
0
EI ( x ) L2 ∫ GA / κ ( x )dx
0

x
L ( )2 L
f BB = ∫ L dx + 1 1
0
EI ( x ) L2 ∫ GA / κ ( x )dx
0
EI(x): tangent flexural rigidity, GA/ κ (x): tangent shear rigidity.

A parabolic distribution of flexural flexibility (1/EI(x)) was assumed; the function may be defined if
two end values (1/EI(0) and 1/EI(L)) and the minimum value (min {1/EI(x)}) are given; an inflection
point can stay within a member or outside the member. The smallest flexural flexibility may be
assumed to be the same as the initial elastic value. For member end moments, the member end
flexural flexibility may be evaluated on the basis of a moment-curvature relation.

Takizawa (1973, 1976) assumed a moment-rotation relation at the member end under the
anti-symmetric moment distribution to determine the flexibility coefficients f AA , f AB and f BB ,
rather than flexural flexibility.

Flexibility relation of a member for incremental member end rotations ∆θ A and ∆θ B and
moments ∆m A and ∆m B is given in the following form by ignoring shear deformation;

22
  f B − fo ( f A + f B ) 2 f AB 
∆θ A  2 f A + 3
− f AB −
2
+
3  ∆m A 
 = f − fo  
∆θ B   − ( f A + f B ) + 2 f AB 2 fB + A − f AB  ∆m B 
 2 3 3 
in which f o : member end flexibility at the initial elastic stage (= L / 6 EI).

Member end flexibility f A or f B was determined as an incremental member end rotation to

moment ratio under the anti-symmetric bending moment distribution with the same member end
moments of amplitude m A or m B acting at the two ends. The curvature was assumed to distribute
along the member length in a parabolic form with the initial elastic value at the mid-span. Normally,
this flexibility is given by a hysteresis model.

The interacting element f AB is defined as

f AB = ( f A − f o )( f B − f o ) sgn(m A m B )

This is an interesting concept in analyzing an inelastic member. However, the parabolic flexibility
distribution may not describe the actual concentration of deformation at critical section (normally at
member ends) due to flexural yielding and deformation attributable to slippage of longitudinal
reinforcement within a beam-column connection. The usage of inelastic springs at locations of
concentrated deformation in conjunction with this model may be a useful solution.

References:

Takizawa, H., "Analysis of Reinforced Concrete Buildings under Strong Earthquake Motion (in
Japanese)," Concrete Journal, Japan Concrete Institute, Vol. 11, No. 2, February 1973, pp.
10-21.
Takizawa, H., "Notes on Some Basic Problems in Inelastic Analysis of Planar R/C Structures (Part
1)," Transactions, Architectural Institute of Japan, No. 240, February 1976, pp. 51-62.

23
 10.7 Multi-spring Model

The moment-curvature analysis of a reinforced concrete section under load reversal is normally
based on (a) an assumption, known as Bernoulli's hypothesis, that a plane section remains plane
after bending, (b) uni-axial stress-strain relations of materials and (c) equilibrium of forces. The
"lamina (fiber) model" may be used for a sectional moment-curvature analysis. However, the
deformation of a member calculated by the integration of curvature is known to underestimate the
deformation observed in a test. Therefore, a member model, based on member end moment-rotation
relation, is favored in a nonlinear analysis of a reinforced concrete structure.

The flexural behavior of a reinforced concrete column section is influenced by existing axial load;
the phenomenon normally called as "the interaction of axial force and bending moment." The
interaction of axial deformation and curvature also exists. For example, in a reinforced concrete
member under pure bending, the neutral axis of section shifts to compression side after flexural
cracking, which accompanies the elongation at the centroid of the section although no tensile force
is applied; i. e., pure bending causes the elongation of a member after cracking. Member stiffness
models such as one-component and multi-component models cannot model such interaction.

Furthermore, a column during a real earthquake motion is subjected to bi-directional shear and
bending. The bending capacity in a principal direction is influenced by an existing bending moment
in the orthogonal direction, the phenomenon called "the bi-directional bending interaction." Proper
analytical model is necessary to analyze a three-dimensional structure under bi-directional horizontal
and vertical ground motions.

Lai Model: Lai et al. (1984) proposed a multi-spring (MS) model to simulate the flexural behavior of
reinforced concrete columns under varying axial load and bi-directional lateral load reversals. A
column member was idealized to be a linear element with its length equal to the column clear height
and two multi-spring elements with zero length at the top and bottom. The multi-spring element
model was similar to the "lamina (fiber) model," but the section was not divided into many elements;
cross sectional properties at each member end were represented by four steel springs and five
concrete springs. All inelastic flexural deformation was assumed to concentrate in the multi-spring
elements at the member ends (one-component model).

A column was idealized by an elastic line element b with two multi-spring elements a and c;
member end nodes are numbered A and B. The node between the top multi-spring element and the
elastic element is C and the node between the bottom multi-spring element and the elastic element
is D. For each element, the start and terminal ends 1 and 2 are assigned in the direction of A to B.

z
A y
a
C z

D
c
B
x

24
 The following assumptions were
used in formulating stiffness of a
multi-spring element:
(1) The plane section remains plane
after bending in the multi-spring
element;
(2) The force-deformation relation of
each spring under monotonically
increasing load is elastic-perfectly
plastic;
(3) Concrete springs resist only
compressive stresses;
(4) Deformation of steel springs is
caused by the pullout deformation of the
longitudinal reinforcement from
anchorage zone and the yielding of the
reinforcement;
(5) Concrete and steel springs yield
at the same deformation.

The elastic element in the middle

part is already cracked and deforms in
flexure only. The moment of inertia of
section is evaluated for fully cracked
section.

The steel spring simulates the elasto-plastic behavior and the Bauschinger effect under load
reversals. The area of a steel spring is calculated by summing the areas of reinforcing bars in the
corner quadrant of the section. The yield force Psy of a steel spring is determined as the steel area
(= As / 4) in the tributary area and the yield
strength σ sy of the steel.
Ag σ sy
Psy =
4

The deformation due to the pull-out of

longitudinal reinforcement from the
anchorage zone is assigned to the steel
spring. The steel stress was assumed to
distribute linearly with constant bond stress
u along the development length l d
within the joint;
Ab f y
ld =
π db u
where d b : diameter of bar, Ab : cross sectional area of bar, f y : yield stress of bar. The pullout
deformation d sy of the longitudinal reinforcement at yielding is estimated by assuming a linear
distribution of steel strain over the development length l d ;
1
d sy = ld ε y
2
where, ε y : yield strain of steel.

The elastic stiffness k se of steel spring is the ratio of yield force to the yield displacement;

25
 As f y
kse =
d sy
Post-yield stiffness of a steel spring
was assumed to be zero.

The concrete spring resisted only

compression stress. The initial elastic
deformation of the spring simulates the
initial elastic depression at the joint
core. The plastic deformation of a
concrete spring represents the
accumulated crushing behavior of the
concrete over the plastic hinge length.
The yielding force level of the effective
concrete spring is expressed by
Pcy = 0.85σ B Aci
where Pcy : yielding force of the
concrete spring, Aci : tributary area for a Pb
concrete spring.
Mb
The tributary area Aci of a corner
concrete spring was evaluated by
considering the equilibrium of axial force
at the balanced point, when the tensile
steel springs yield simultaneously with
the crushing (compression yielding) of Deformation and resistance in springs
the concrete springs. As the yield Cross Section
at the balanced point
deformation of the steel and concrete
springs was assumed to be the same,
the steel spring elements in tension and
compression both yield at the balanced
point. The neutral axis lies at the center
of the section and the central concrete
element does not carry any force.
Therefore, the axial force at the
balanced point must be resisted by the
two concrete spring elements;
Pb
Aci =
2(0.85σ B )
The area Ac 0 of the central concrete
spring was calculated as the remaining
area after steel area and tributary areas
of the four corner concrete springs were
removed from the gross sectional area.

The distance d between the four

corner springs was determined by the
equilibrium of bending moment M b at
the balanced point. As the yield
deformation of the steel and concrete
springs was assumed to be identical, the
neutral axis is at the centroid of the
26
section. Therefore, the distance between the springs is determined as

2M b
d=
(∑ Asi f y ) + ∑ 0.85σ B Aci

The axial force-bending moment interaction diagram may be represented by four zones. The
equations for the four zonse may be expressed as follows;

(a) Zone 1: Tension failure zone (TAB)

2M y
P = 2( Ps1 y + Ps 3 y ) −
d

(b) Zone 2: Compression failure zone (BD)

Pc 0 y
P = −[2 + ]( Pc 2 y + Ps 2 y + Pc 4 y + Ps 4 y ) )
2( Ps1 y + Ps 3 y )
Pc 0 y Pc 0 y 2M y
− + [1 + ]
2 2( Ps1 y + Ps 3 y ) d

(c) Zone 3 (CD):

Pc 0 y
P = −[2 + ]( Pc 2 y + Ps 2 y + Pc 4 y + Ps 4 y )
2( Ps1 y + Pc1 y + Ps 3 y + Pc 3 y )
Pc 0 y Pc 0 y 2M y
− + [1 + ]
2 2( Ps1 y + Pc1 y + Ps 3 y + Pc 3 y ) d

The interaction curve of the model deviated from the interaction curve due to the use of fewer
number of springs. With an increase in the number of springs in a multi-spring element, the
simulation of the interaction behavior is improved, and the determination of the stiffness properties of
each spring is simplified.

Jiang and Saiidi (1990) proposed to combine the F

hysteresis properties of the concrete and steel
springs in each corner to simplify the model . Fsy+Fcy
because the yield deformations of the concrete
and steel springs located in the same quadrant is
β s k2
identical.

-dsy k1 d
k2 dsy

β s k2 -Fsy

Force-deformation relation
(Jiang and Saiidi, 1990)

27
Li Model: Li et al. (1988, 1990) simplified the method to determine spring properties and modified
the hysteretic properties of the concrete and steel springs; they demonstrated the reliability of the
model with respect to column test results using simple five spring models.

Li (1991) suggested the use of fiber models at the pz

member end. A member is represented by plastic
zones (multi-axial spring elements) at the member Multi-spring element
ends and an elastic zone in the middle part.
Elastic element
A multi-axial spring element consists of uni-axial
springs in the direction of member axis. The number
of springs may be chosen considering the material
Lo
properties, section size and reinforcement
arrangement. Each single steel bar may be represented by a steel spring at the bar center. Concrete
section may be divided into several sub-areas; a concrete spring is placed at the geometrical
centroid of a concrete sub-area. The multi-spring element is rigid against shearing force. The length
of a multi-axial spring element is given as pz, which is suggested as the smaller of one-half of total
depth of end section or one-tenth of the clear length of the member.

28
 The axial force fi and deformation di of each axial spring are given as follows;
f i = σ i Ai
di = ε i pz
where σ i and ε i : stress and strain at spring point i, and Ai . area of sub-area i. Multi-linear
stress-strain relation may be used for concrete and steel springs. Some adjustment is required for
the stress-strain relation of a steel spring to take into account the stiffness degradation, for example,
due to the bond slip along the longitudinal reinforcement or the pullout deformation of the
longitudinal reinforcement from the anchorage zone. The yield deformation may be increased by a
factor κ ;
ho
− 1.0
ho
κ = 1.0 + D for > 1.0
ho D
D
ho
κ = 1.0 for ≤ 1.0
D
where ho: shear span, and D: overall depth of section.

29
 The flexural and axial deformations may be considered in the middle elastic part.

A common member coordinate system is used to define forces and displacements, with x-axis in
the direction of the member and y- and z-axes in the principal directions of the section. Forces and
displacements at node J are denoted by {P} J and {D} J , whereas the start and terminal end
forces and displacements of an element i are denoted by { p1 }i and { p 2 }i , {d1 }i and {d 2 }i .

The incremental stiffness relation of a multi-spring model i (i = a or c) may be expressed;

 ∆p1x   ∆p 2 x 
   
∆m1 y  = −∆m2 y 
 ∆m   ∆m  A
 1z  i  2 z i
a
 k xx k xy k xz   ∆d1x   ∆d 2 x 
      C
= k yx k yy k yz  (∆θ 1 y  − ∆θ 2 y  )
 k zx
 k zy k zz   ∆θ 1z  i  ∆θ 2 z  i
i
b
in which,
k xx = ∑ k i
i

k xy = ∑ k i zi
i D
k xz = ∑ k i y i c
i

k yy = ∑ k i zi2 B
i

k yz = ∑ k i yi zi
i

k zz = ∑ k i y i2
i

and k i : tangent stiffness of spring i, (yi, zi): coordinates of spring i with respect to the centroid of
section, p x : axial force, m y : bending moment about y-axis, m z : bending moment about z-axis,
d x : axial deformation at the centroid, θ y : rotation about y-axis, θ z : rotation about z-axis.

It should be noted that the following relations hold because no length is considered in multi-spring
elements a and c;
∆p1 y   ∆p 2 y  ∆p1 y  ∆p 2 y 
  = −    = − 
 ∆p1z  a  ∆p2 z  a  ∆p1z  c  ∆p 2 z  c
and
∆d 1 y   ∆d 2 y  ∆d1 y  ∆d 2 y 
  = −    = − 
 ∆d 1z  a  ∆d 2 z  a  ∆d1z  c  ∆d 2 z  c

In a symbolic expression, for multi-spring a;

{∆p1}a = [k ]a {∆d1}a − [k ]a {∆d 2 }a
{∆p2 }a = −[k ]a {∆d1}a + [k ]a {∆d 2 }a
and for multi-spring element c;
{∆p1}c = [k ]c {∆d1}c − [k ]c {∆d 2 }c
{∆p2 }c = −[k ]c {∆d1}c + [k ]c {∆d 2 }c

30
 For an elastic element of length L, flexural rigidity EI, axial rigidity EA, and shear rigidity GA/κ, a
stiffness relation can be formulated in the form:
{∆p1}b = [k11 ]b {∆d 1}b + [k12 ]b {∆d 2 }b
{∆p2 }b = [k 21 ]b {∆d1}b + [k 22 ]b {∆d 2 }b
and {d xk , d yk ,θ yk , d zk ,θ zk } .
T T
where, { p k }b and {d k }b at k-end are { p xk , p yk , m yk , p zk , m zk }

The compatibility of displacements at nodes:

{∆D} A = {∆d1}a
{∆D}B = {∆d 2 }c
{∆D}C = {∆d 2 }a = {∆d1}b
{∆D}D = {∆d 2 }b = {∆d1}c

Equilibrium of external forces and the sum of internal element end forces at a node;
{∆P} A = {∆p1}a
{∆P}B = {∆p2 }c
{∆P}C = {∆p2 }a + {∆p1}b
{∆P}D = {∆p2 }b + {∆p1}c

Special care must be exercised in formulating a member stiffness matrix with four nodes (A, B, C
and D) to include the following conditions;

(a) for displacements:

∆D y  ∆D y  ∆d1 y  ∆d 2 y 
  =  =  = 
 ∆D z  A  ∆D z  C  ∆d1z  a  ∆d 2 z  a
∆D y  ∆D y  ∆d1 y  ∆d 2 y 
  =  =  = 
 ∆D z  D  ∆D z  B  ∆d1z  c  ∆d 2 z  c
and (b) for forces:
∆p1 y   ∆p 2 y  ∆p1 y  ∆p 2 y 
  = −    = − 
 ∆p1z  a  ∆p2 z  a  ∆p1z  c  ∆p 2 z  c

References:

Lai, S.-S., G. T. Will and S. Otani, "Model for Inelastic Bi-axial Bending of Concrete Member,"
Journal, Structural Division, ASCE, Vol. 110, No. ST11, November 1984, pp. 2563-2584.
Jiang, Y., and S. M. Saiidi, "Four-Spring Element for Cyclic Response of R/C Columns," Journal,
Structural Engineering Division, ASCE, Vol. 116, No. ST4, April 1990, pp. 1018-1029.
Li, K.-N., S. Otani and H. Aoyama, "Reinforced Concrete Columns under Varying Axial Load and
B-directional Lateral Load Reversals," Proceedings, Ninth World Conference on Earthquake
Engineering, Tokyo-Kyoto, August 1988, Vol. VIII, pp. 537-542.
Li, K.-N., S. Otani and H. Aoyama, "Study on Reinforced Concrete Columns subjected to Varying
Axial Load and Bi-directional Horizontal Earthquake Loads (in Japanese)," Report, Aoyama
Laboratory, Department of Architecture, Faculty of Engineering, University of Tokyo, March
1990.
Li, Kang-Ning, S. Otani and H. Aoyama, “R/C Columns under Axial and Bi-directional Lateral Loads,”
Proceedings, Mechanics Computing in 1990’s and Beyond, ASCE, Vol. 2, Structural and
Material Mechanics, May 1991, pp. 681 - 685.

31
10.10 Wall Models

Column Model: A structural wall is often represented by a

column (lineal) model at the center of the wall section. The kA
boundary girders at the top and bottom of a wall panel are
normally assumed to be rigid. The bending and shear Axial Spring
deformations are considered for the wall.
Shear Spring
One-component model is often used in an inelastic analysis,
in which a yield rotation is estimated for uniform or kB
anti-symmetric moment distribution along the wall height. The
uniform bending moment distribution may be realistic for a
structural wall at the lower level. The flexibility relation of a wall
including a shear spring is
Flexibility of springs
 h 
∆e AB   0 0  ∆p AB 
   EA  
 ∆θ A  = 0 2 f + f A + γ − f + γ   ∆m A 
 ∆θ   0 − f +γ

2 f + f B + γ   ∆m B 
 B 
 
h 1 1 κ
in which f = , fA = , fB = , γ = , EI : flexural rigidity of the middle elastic
6 EI kA kB GAw h
GAw
region, h : clear height of the wall panel, : shear rigidity of wall section including shape factor
κ
κ for shear. Shear rigidity may be reduced with inelastic deformation.

A rigid zone, length equal to the one-half width of a wall, must be considered at the end of a
girder connected to the structural wall.

Distributed flexibility model may be used to represent a distribution of damage for a wall. Shear
deformation of a wall needs be considered. A wall may be sub-divided into a short segment (the
discrete element model) to reflect the distribution of the damage.

The problem of representing a wall by a single line member at the center is that the “three
dimensional effect” of a flexural wall cannot be represented; i.e., the axial elongation of a wall at the
centroid due to the shift of the neutral axis after flexural cracking cannot be modeled. The boundary
girders connected on both sides of a wall displace the same amount in the vertical direction at the
wall faces.
py
mz

Brace Model: A structural wall is sometimes idealized by a px

braced frame, in which shear deformation is represented by
the deformation of diagonal braces and flexural deformation by Rigid Beam
the deformation of vertical elements. This model is useful
when the shear deformation is dominant in a structural wall.
EA1
The axial stiffness of tensile bracing and vertical elements EA2 EA2 h
may be reduced to take into account the degradation of
stiffness due to cracking.

The flexibility of the wall element is given for the coordinate Rigid Beam
system shown in the figure,
L

32
  1 L '3 h3 1 h2 
 2 L2 ( EA + EA ) 0 −
EA2 L2 
 1 2

 L '3 
dx     px 
   h EA1   p 
d y  =  0 0
 y
θ   2 EA2 L '3 h3
 z +   mz 
 EA1 EA2 
 1 h2 2h 
 − 0 
 EA2 L2 EA2 L2 
1
where L ' = ( L + h ) 2 : length of diagonal braces.
2 2

Boundary Column Model: A structural wall, especially behaving dominantly in flexure, is modeled
by three springs at the boundary columns and at the wall center (Otani et al., 1985). The two outside
springs are provided with the axial stiffness of the boundary columns. The central element
represents the vertical uni-axial, lateral shear and flexural rotational characteristics of the wall panel;
a rotational spring is placed only at the bottom of the central element. The girder within a wall is
considered to be rigid.

The axial stiffness in compression is assumed to be

elastic, and the tensile stiffness is reduced in comparison
with the compression stiffness ignoring "the tension Rigid Girder
stiffening effect," and the stiffness is reduced to a small
value after tensile yielding of the longitudinal
reinforcement.
aw
The flexibility relation of the wall panel as a simply
supported member is a1 γ a2
∆e AB  a 0 0  ∆p AB 
    
 ∆θ A  =  0 2 f + γ − f + γ   ∆m A  kB

 ∆θ   0 − f + γ 2 f + f B + γ   ∆m B 
 B 
h h
in which α = a1 + a 2 + a w , aw = , f = , fB :
EAw 6 EI w Boundary Column Model
κ GAw (Kabeyasawa Model)
flexibility of rotational spring, γ = , : shear
GAw h κ
rigidity of the wall section, taking stiffness degradation with shear cracking into account, EI w :
elastic flexural rigidity of wall panel section. Forces and displacements are defined at the mid-point
of the rigid girders.

The stiffness of the wall model is formulated by including the stiffness contribution from the
boundary columns.

Reference:

Kabeyasawa, Toshimi, Chapter 7 Earthquake Response Analysis, Design of Modern Highrise

Reinforced Concrete Structures, Series on Innovation in Structures and Construction vol. 3,
Imperial College Press, 2001, pp. 315 - 344.
Otani, S., T. Kabeyasawa, H. Shiohara and H. Aoyama, "Analysis of the Full-scale Seven-story
Reinforced Concrete Test Structure," ACI SP-84, Earthquake Effects on Reinforced Concrete

33
 Structure, US-Japan Research, American Concrete Institute, Detroit, 1985, pp. 203-239.
Vulcano, A., and V. V. Bertero, “Analytical Models for Predicting the Lateral Response of RC Shear
Walls,” University of California at Berkeley, Earthquake Engineering Research Center, Report
No. EERC 87-19, 1987.

34
 Home Assignment No. 6
(Formulation of Member Stiffness Matrix)
2002-03-08
Otani, S.

Consider a multi-component model, consisting Plastic hinge

of two parallel elements I and II, supported at A p1 EI
end by a pin and at B end by a roller. Element I is A B
connected at A end by a plastic hinge and is
rigidly connected at B end. Element II is rigidly
connected at the two ends. The elastic flexural
rigidity of elements I and II is p1 EI and p2 EI. The p2 EI
stiffness relation is expressed as follows:
∆m A  EI 4 p2 2 p2   ∆θ A 
 =   
∆m B  L 2 p2 4 p2 + 3 p1  ∆θ B  mA

(1) Express member end rotations ∆θ A and

mB
∆θ B in terms of member end moments ∆m A
and ∆m B in a matrix form.

(2) The incremental moment-and rotation relations are given for anti-symmetric bending moment
distribution below. Namely, the relations were obtained by applying equal incremental moments
∆m A at the two ends of a simply supported beam and measuring resultant incremental member
end rotations ∆θ A at A end. Similarly, equal incremental member end moments ∆m B were
applied to a simple beam and incremental member end rotation ∆θ B was measured at B end. The
flexibility at the member ends is defined from the incremental relation as follows;
1 ∆θ A
fA = =
k A ∆mA
1 ∆θ B
fB = =
k B ∆mB

mA mB

∆mB
kB
∆mA
kA
θA θB

∆θ A ∆θ B

Determine the stiffness ratios p1 and p 2 which satisfy the flexibility f A and f B at the two
member ends as defined above.

35
[Solution]
For a two-component model with a plastic hinge at left end, the tangent stiffness relation is given as
∆m A  EI 4 p2 2 p 2   ∆θ A 
 =   
∆m B  L 2 p2 4 p2 + 3 p1  ∆θ B 

(1) The stiffness relation is solved for the flexibility relation,

∆θ A  L 1  4 p2 + 3 p1 −2 p2   ∆mA 
 = 2   
∆θ B  EI (4 p2 + 3 p1 )4 p2 − 4 p2  −2 p2 4 p2   ∆mB 

(2) Under anti-symmetric bending, ∆mA = ∆mB ;

L 1
∆θ A = (4 p2 + 3 p1 − 2 p2 )∆mA
EI (4 p2 + 3 p1 )4 p2 − 4 p22
L 1
= (2 p2 + 3 p1 )∆mA
EI (4 p2 + 3 p1 )4 p2 − 4 p22
L 2 p2 + 3 p1
= ∆mA
EI 12( p1 + p2 ) p2
∆mA
=
kA
L 1
∆θ B = (−2 p2 + 4 p2 )∆mB
EI (4 p2 + 3 p1 )4 p2 − 4 p22
L 1
= (2 p2 )∆mB
EI (4 p2 + 3 p1 )4 p2 − 4 p22
L 1
= ∆mB
EI 6( p1 + p2 )
∆mB
=
kB
Solving for p1 and p1 ,
L k AkB
p2 =
2 EI k A + 2k B
L k − kA
p1 = kB B
3EI k A + 2k B

36
 Home Assignment No. 6
2001-12-05
Otani, S.

Consider a system consisting of diagonal braces with py

mz
axial rigidity EA1 and vertical elements with axial rigidity
EA2. The braces and vertical elements are connected to
rigid girders (girder depth is zero) at the top and bottom by px
hinges. The distance between the vertical elements is L
and clear height between the top and bottom girder is h. Rigid Beam

Formulate a flexibility relation of the system at the

center of the rigid girder at the top, considering the bottom EA1
girder to be fixed. In other words, find horizontal, vertical EA2 h
EA2
and rotational displacements under each of unit horizontal
force, unit vertical force and unit moment applied at the
mid-span of the top girder.

Rigid Beam

37
 [Solution]
This structure is statically indeterminate by one degree. py
mz
The reaction at the support at lower right is selected as
an indeterminate force and is released to make a
statically determinate structure. px
B C Rigid Beam

EA1
EA2 EA2 h

α
A D

Rigid Beam
Basic Determinate Structure
L
Axial forces in the members are calculated for unit force
applied separately in the horizontal, vertical and
rotational directions at the center of the top rigid beam, and also for a unit applied in the horizontal
direction at the released support. Axial force is positive when in tension.

Px=1.0 Mz=1.0
Py=1.0
B C B B
C C
0 0 0

h 1 0 1 0
0 − 1 1
−
L 2 2 L L
1
A cos α D A D A D

NM0
NH0 NV0

B C
1
cos α

h h
− −
L L
1
A cos α N=1
D
Nu

Horizontal displacement is calculated in the direction of the indeterminate force under the unit force
using the unit load method.
Member L NH 0 NV 0 N M 0 Nu L L L L 2
N H 0 Nu NV 0 N u N M 0 Nu Nu
EA EA EA EA EA
AB h 0 1 1 h 0 h2 1 h2 1 h3 1
− − − −
EA2 2 L L EA2 2 L EA2 L2 EA2 L2
AC L' 1 0 0 1 L' 1 0 0 L' 1
EA1 cos α cos α EA1 cos 2 α EA1 cos 2 α

38
BD L' 0 0 0 1 0 0 0 L' 1
EA1 cos α EA1 cos 2 α
CD h h 1 1 h h3 1 h2 1 h2 1 h3 1
− − −
EA2 L 2 L L EA2 L2 EA2 2 L EA2 L2 EA2 L2
Total - - - - - 1 L '3 h3 h2 1 0 2 L '3 h3
( + ) − ( + )
L2 EA1 EA2 EA2 L L2 EA1 EA2
where L ' = L2 + h 2

From the displacement boundary condition at support D, the horizontal reaction N1 at the support
due to unit horizontal load acting at the top beam is calculated;
1 L '3 h3 2 L '3 h3
( + ) + N1 2 ( + ) = 0.0
L2 EA1 EA2 L EA1 EA2
1
N1 = −
2
The horizontal reaction N1 due to unit vertical force is calculated;
h 2 1 2 L '3 h3
− + 2( + ) N1 = 0.0
EA2 L L EA1 EA2
h2
L EA2
N1 =
2 L '3 h3
+
EA1 EA2
The horizontal reaction N1 due to unit moment is calculated;
2 L '3 h3
0+ ( + ) N1 = 0.0
L2 EA1 EA2
N1 = 0.0
Therefore, the axial forces in he brace members due to unit load applied at the top beam are
expressed as;
Px=1.0 Mz=1.0
Py=1.0
B C B B
C C
0
h 0
1 h 1 h 1 h 1 1
2L − − − N1 − N1 −
2 cos α 2L 2 L 2 L L L
1
N1 N1
2 cos α
A D A cos α cos α D A D

NM
NH 0.5 NV 0.0
h2
L EA2
N1 =
2 L '3 h3
+
EA1 EA2
Calculation of displacement due to unit force applied at the center of top beam. Note that a unit force
in the direction of desired displacement may be applied to the original structure, but unit forces in the
direction of applied forces can be applied to the statically determinate structure.

39
 Px=1.0 Mz=1.0
Py=1.0
B C B B
C C
0
h 0
1 h 1 h 1 h 1 1
2L − − − N1 − N1 −
2 cos α 2L 2 L 2 L L L
1
N1 N1
2 cos α
A D A cos α cos α D A D

NM0
NH0 0.5 NV0 0.0
h2
L EA2
N1 =
2 L '3 h3
+
EA1 EA2

Px=1.0 Mz=1.0
Py=1.0
B C B B
C C
0 0 0

h 1 0 1 0
0 − 1 1
−
L 2 2 L L
1
A cos α D A D A D

NMu
NHu NVu

(1) Horizontal displacement d x at the top beam due to horizontal force px , vertical force p y and
moment mz acting at the top beam.

Member L NH 0 NV 0 NM 0 Nu L L L
N H 0 Nu NV 0 N u N M 0 Nu
EA EA EA EA
AB h 0 1 1 h 0 h2 1 h2 1
− −
EA2 2 L 2L EA2 4 L EA2 2 L2
AC L' 1 0 0 1 L' 1 0 0
EA1 cos α 2 cos α EA1 2 cos 2 α
BD L' 0 0 0 1 0 0 0
−
EA1 2 cos α
CD h h 1 1 h h3 1 h2 1 h2 1
− − − −
EA2 L 2 L 2L EA2 2 L2 EA2 4 L EA2 2 L2
Total - - - - - 1 L '3 h3 0 h2 1
( + ) −
2 L2 EA1 EA2 EA2 L2

1 L '3 h3 h2 1
dx = ( + ) p x − mz
2 L2 EA1 EA2 EA2 L2

40
 (2) Vertical displacement d y at the top beam due to horizontal force px , vertical force p y and
moment mz acting at the top beam.
Member L NH 0 NV 0 N Nu L L L
M0 N H 0 Nu NV 0 N u N M 0 Nu
EA EA EA EA
AB h 0 1 1 1 h 0 h 1 h h 1 h
− − N1 ( − N1 ) − ( − N1 )
2 L 2 EA2 2 L EA2 2 L L2
EA2 2 L
AC L' 1 0 0 N1 L ' N1 0 0
EA1 cos α cos α EA1 cos 2 α
BD L' 0 0 0 N1 0 0 0
EA1 cos α
CD h h 1 1 1 h h h2 h h 1 h
( − N1 )
h 1 h
− − N1 ( 2 N1 − ) ( − N1 )
2 L EA2 L 2L 2 EA2 2 L EA2 2 L L2
EA2 L 2 L
Total - - - - - L ' N1 h 1 h 0
+ ( − N1 )
EA1 cos 2 α EA2 2 L

h h2 h
( 2 N1 − )
EA2 L 2L

h2
L EA2
N1 =
2 L '3 h3
+
EA1 EA2
L ' N1 h h2 h h 1 h
dy = { + ( 2 N1 − )} px + ( − N1 ) p y
EA1 cos α EA2 L
2
2L EA2 2 L
L ' N1 h h2 h
+ ( 2 N1 − )
EA1 cos α EA2 L
2
2L
N1 L '3 h3 h h
= 2( + )−
L EA1 EA2 EA2 2 L
h2
L EA2 1 L '3 h3 1 h2
= ( + ) −
2 L '3 h3 L2 EA1 EA2 EA2 2 L
+
EA1 EA2
L h2 1 h2
= − =0
2 EA2 EA2 2 L
Therefore,

41
 h 1 h
dy = ( − N1 ) p y
EA2 2 L
h2
h 1 hL EA2
= ( − ) py
EA2 2 L 2 L '3 h3
+
EA1 EA2
h3
h EA2
= (1 − 3 ) py
2 EA2 L' h3
+
EA1 EA2
L '3
h EA1
= py
2 EA2 L '3 h3
+
EA1 EA2

(3) Rotation θ z at the top beam due to horizontal force px , vertical force p y and moment mz
acting at the top beam.

Member L NH 0 NV 0 NM 0 Nu L L L
N H 0 Nu NV 0 N u N M 0 Nu
EA EA EA EA
AB h 0 1 1 1 0 h h
− − −
2 EA2 L EA2 L2
EA2 2 L L
AC L' 1 0 0 0 0 0 0
EA1 cos α
BD L' 0 0 0 0 0 0 0
EA1
CD h h 1 1 1 h2 h h
− −
EA2 L2 2 EA2 L EA2 L2
EA2 L 2 L L
Total - - - - - 1 h2 0 2h
− EA2 L2
EA2 L2

1 h2 2h
θz = − 2
px + mz
EA2 L EA2 L2

Collecting the information,

42
  1 L '3 h3 1 h2 
 2 L2 ( EA + EA ) 0 −
EA2 L2 
 1 2

 L '3 
dx     px 
   h EA1   p 
d y  =  0 0
 y
θ   2 EA2 L '3 h3
 z +   mz 
 EA1 EA2 
 1 h2 2h 
 − 0 
 EA2 L2 EA2 L2 
1
where L ' = ( L + h ) 2
2 2

43
10.11 Modeling of Foundation

A building structure is often analyzed with rigid foundation. A structure stands on flexible ground,
and it sometimes becomes necessary to consider the effect of soil flexibility. The finite element
model is sometimes used for the ground. A simple model is to consider a sway spring and a rocking
spring under a rigid foundation.

Free surface

Sway Rocking

Pile foundation supporting a structural wall is

idealized by an elastic member supported on a fixed
base or by a line member supported by vertical springs
at various levels. Vertical elastic stiffness kv of a pile
may be evaluated by using a reduced length taking into
account friction along the depth of a pile (Japan Road
Association, 1990);
Free surface
Ap E p
kp =α
l

where, α: effective length factor, A p : cross sectional No mass

area at the bottom end of a pile, E p : Young's modulus

of pile material, l : length of the pile.

The value of a was studied for loading test data (L/D > 10, where L/D: length to diameter ratio of
a pile). The vertical stiffness was determined as the secant stiffness at yielding point on log P-log S
relation, where P: load applied at the pile top, and S: vertical deformation at the pile top. The
following expressions are suggested (Japan Road Association, 1990);

(a) driving steel tube pile:

l
α = 0.014 + 0.78
D
(b) driving PC (precast prestressed concrete) pile or HPC (high strength precast prestressed
concrete) pile:
l
α = 0.013 + 0.61
D
(c) cast-in-situ reinforced concrete pile:
l
α = 0.031 − 0.15
D
(d) drilled steel tube pile:

44
 l
α = 0.009 + 0.39
D
(e) drilled PC and HPC pile:
l
α = 0.011 + 0.36
D

The above expressions tend to give small vertical stiffness of a pile foundation.

The vertical stiffness of a pile may be calculated by considering a pile being modeled by a single
line member with a series of vertical friction springs attached along the depth and a vertical spring at
the bottom of the pile. The friction spring properties may be determined for the displacement.

Sway-rocking Model: A sway-rocking model of a large structure may be formulated by the dynamic
ground compliance proposed by T. Kobori for square foundation.

The effective shear modulus G0 is estimated from shear wave velocity Vs and unit weight per
volume ρ;
ρ
G0 = Vs2
g
where, g : gravity acceleration.

The dynamic shear modulus may be estimated to be one-half of the static shear modulus;
Ge = G0 / 2

The first mode un-dimensional frequency a0 is defined as

ρ
a0 = ω d
Ge
where, ω : circular frequency of structure on rigid foundation, d : dimension of the square
foundation.
Equivalent spring constants KeS and

Equivalent spring constant KeR and

Equivalent damping coefficient CeR
Equivalent damping coefficient CeS

Non-dimensional frequency a0 Non-dimensional frequency a0

Equivalent constants for sway Equivalent constants for rocking

The equivalent spring constants K eS and K eR for sway and rocking are evaluated from charts

45
for longitudinal dimension 2c and transverse dimension 2b. Spring constants K S and K R for
rocking and sway springs and associated damping factors hS and hR are evaluated as;
K S = K eS d Ge
Ge
K R = K eR d 3
3
1 a0 CeS
hS =
2 K eS
1 a0 CeR
hR =
2 K eR

Reference:

Japan Road Association, "Standard Specifications for Design of Road Bridge and Commentary, Part
4: Underground Structures (in Japanese)," revised in February 1990.

46
 Chapter 11. Member Hysteresis Models

11.1 Introduction

An inelastic earthquake response analysis of structures requires realistic hysteresis models,

which can represent resistance-deformation relationship of a structural member model.

The resistance-deformation relations are different for constitutive materials of a section, for a
section, for a member, for a story and for an entire structure. The resistance-deformation relation of
a structural analysis unit observed in a laboratory test must be idealized into a
resistance-deformation hysteresis model. Different levels of resistance-deformation models must be
used for structural elements considered in an analysis; e.g., a constitutive model of materials in a
finite element method analysis, a hysteresis model for a rotational spring in a one-component
member model, a story shear-drift hysteresis model for a mass-spring model.

A hysteresis model is derived by extracting common features of resistance-deformation relations

observed in laboratory tests of members of similar properties. The hysteresis model of a member
must be able to express resistance-deformation relations under any loading history, including load
reversals.

Resistance-deformation relationship under monotonically increasing loading is called the primary

curve, skeleton curve or backbone curve. The skeleton curve provides an envelope of the hysteresis
resistance-deformation relationship if the behavior is governed by stable flexure. The skeleton curve
for reinforced concrete member is normally represented by a trilinear relation with stiffness changes
at flexural cracking and tensile yielding of longitudinal reinforcement. The skeleton curve of a
member must be defined on the basis of mechanical properties of constitutive materials and
geometry of the member. Some researchers suggest the use of a bilinear relation with a stiffness
change at yielding, ignoring the initial uncracked stage, because a reinforced concrete member
subjected to light axial force can be easily cracked by shrinkage or accidental and gravity loading.

The state-of-the-art does not provide a reliable method to estimate the initial stiffness, yield
deformation and ultimate deformation. The stiffness degrades from the initial elastic stiffness with
increased inelastic deformation and the number of cycles under reversed loading. The elastic
modulus of concrete varies significantly with concrete strength and mix; initial cracks cause decay in
the stiffness. The estimate of yield deformation is more complicated by the interaction of bending
and shear deformation and additional deformation due to pullout of longitudinal reinforcement from
the anchorage zone and due to bar slip of longitudinal reinforcement along the longitudinal
reinforcement within the member. Empirical expressions are necessary for the estimate of yield and
ultimate deformation.

The coordinates of a response point on a deformation-resistance plane are given by (D, F), in
which, D: deformation, F: resistance. The skeleton curve is represented by either "bilinear" or
"tri-linear" lines for a reinforced concrete member, with stiffness changes at "cracking (C)" and
"yielding (Y)" points.
F Loading
The following terms are defined to clarify the hysteresis
description;
Loading: a case where the absolute value of resistance (or
deformation) increases on the skeleton curve; Unloading
Unloading: a case where the absolute value of resistance
(or deformation) decreases after loading or reloading; and
Reloading; a case where the absolute value of resistance D
(or deformation) increases after unloading before the
response point reaching the skeleton curve. Reloading

The hysteresis model is formulated on the basis of resistance-deformation relations observed in

the laboratory tests. The loading program for a test should include the followings;
(1) At least two cycles of load reversals at an amplitude to study the decay in resistance at the
1
amplitude,
(2) Small deflection amplitude
excursion must be placed after a large
amplitude excursion to study the
slip-type behavior

A lateral load-deflection relation of a

reinforced concrete member was
obtained from the test of a slender
column (Otani and Cheung, 1981). The
behavior was dominantly by flexure
although flexural cracks started to
incline due to the presence of high
shear stresses before flexural yielding.
The yielding of the longitudinal
reinforcement was observed in cycle 3.

The general hysteretic characteristics

can be summarized as follows:
(a) Stiffness changed due to the flexural cracking of concrete and the tensile yielding of the
longitudinal reinforcement (cycle 1);
(b) When a deflection reversal was repeated at the same newly attained maximum deformation
amplitude, the loading stiffness in the second cycle was noticeably lower than that in the first cycle,
although the resistance at the peak displacement was almost identical (cycles 3 and 4). This
reduction in stiffness is attributable to the formation of new cracks during loading cycle 3, and also to
a reduced stiffness of the longitudinal reinforcement in cycle 4 due to the Bauschinger effect.
(c) Average peak-to-peak stiffness of a complete cycle decreases with previous maximum
displacement. Note that the peak-to-peak stiffness of cycle 5 is significantly smaller than that of cycle
2, although the displacement amplitudes of the two cycles are comparable. The peak-to-peak
stiffness of cycle 5 is closer to that of cycles 3 and 4;
(d) The hysteresis characteristics of reinforced concrete are dependent on the loading history,
and
(e) The resistance at the peak deflection is almost the same for the two successive cycles in the
member dominated by flexural behavior.

A hysteresis model of a reinforced concrete "flexural" member must be able to represent the
above characteristics. The skeleton curve is similar to an "envelope curve" of a force-deformation
relation under load reversals. The state of the art is not sufficient to determine the ultimate point, at
the deformation of which the resistance of a member starts to decay. The force-deformation relation
after the onset of strength decay is normally not modeled because the behavior is strongly
dependent on a particular local deterioration of materials.

If the reinforced concrete is subjected to

high shear stress reversals, or if the
slippage of the reinforcement from concrete
within the anchorage area occurs, the
force-deflection curve exhibits a pronounced
"pinching". The pinching behavior is also
observed;
(a) in a "flexural" member when the
amount of longitudinal reinforcement differs
significantly for the tension and compression
sides at the critical sections, typically in a
girder with monolithically cast slabs,
(b) at a member end where additional
deformation may be caused by anchorage
slip of longitudinal reinforcement within the
adjacent member or connection, and
Hysteresis of slip type (Bertero and Popov, 1977)
2
 (c) in a member where bond splitting cracks develop along the longitudinal reinforcement.

Because such hysteresis relationship is highly dependent on loading history and structural
properties of the member, a general hysteresis model is difficult to formulate; or the parameters of
hysteresis models cannot be analytically determined by the properties of the member. In the design
of earthquake resistant structures, the pinching type behavior is generally thought to be undesirable
because small hysteresis energy can be dissipated by the behavior. Therefore, a proper design care
must be exercised to reduce such pinching behavior due to shear and bond deterioration.

Many hysteresis models have been developed in the past. Some hysteresis models are elaborate,
and include many hysteresis rules; others are simple. The complicatedness of a hysteresis model
indicates a large memory to store the hysteresis rule program in a computer. It does not lead to a
longer computation time because the complicatedness of a hysteresis model requires simply many
branches in a computer program, and only a few branches are referred to for a step of response
computation.

A class of hysteresis models, in which the unloading and reloading relation is defined by
enlarging the skeleton curve by a factor of two, are called "Masing type." Some examples of Masing
type models are shown below:

A hysteresis energy dissipation index (Eh) is

used to express the amount of hysteresis energy
dissipation ∆W per cycle during displacement
reversals of equal amplitudes in the positive and
negative directions;
∆W
Eh =
2π Fm Dm
in which Fm: resistance at peak displacement Dm.
The value of the index was derived by equating
the area of hysteresis and the energy ∆W
dissipated by an equivalent viscous damper of a
linearly elastic system in one cycle under the
"resonant" "steady-state" oscillation.

The steady state response amplitude Dm Hysteresis energy dissipation index

under sinusoidal excitation with amplitude p o
and circular frequency ω , is given by

3
 po 1
Dm =
k ω 2 ω
{1 − ( ) } + 4h 2 ( ) 2
ωn ωn
x(t ) = Dm sin(ω t + φ )

The energy dissipated ∆W by viscous damper per cycle is

Tn T
dx dx n

∆W = ∫ (c )( )dt = ∫ c Dm ω 2 cos 2 (ω t + φ )dt

2

0
dt dt 0

= π c ω Dm
2

= 2π h mk ω Dm
2

where m, c, k ,: mass, damping coefficient and stiffness of an SDF system, h : damping factor
c m
(= ), Tn : natural period of the system ( = 2π ), ω n : circular frequency of the system
2 mk k
k
(= ).
m

At the resonant condition ( ω = ω n ), the energy dissipated per cycle can be expressed
∆W = 2π h k Dm
2

Therefore, the damping factor corresponding to the hysteresis energy dissipation ∆W is

∆W ∆W
h= =
2π k Dm
2
2π Fm Dm
Fm
k =
Dm

The equivalent damping factor should not be confused with a damping factor of a viscously
damped system because the equivalent damping factor is not relevant in random oscillation.

References:

Bertero, V. V., and E. P. Popov, "Seismic Behavior of Ductile Moment Resisting Reinforced
Concrete Frames," ACI SP-53, American Concrete Institute, Detroit, 1977, pp. 247-291.
Comite Euro-International du Beton: RC Frames under Earthquake Loading, State of the Art Report,
Thomas Telford, 1996.
Otani, S, "Hysteresis Models of Reinforced Concrete for Earthquake Response Analysis," Journal,
Faculty of Engineering, University of Tokyo, Vol. XXXVI, No. 2, 1981, pp. 125-156.
Otani, S., and V. W.-T. Cheung, "Behavior of Reinforced Concrete Columns Under Bi-axial Lateral
Load Reversals - (II) Test Without Axial Load," Publication 81-02, Department of Civil
Engineering, University of Toronto, 1981.
Saatcioglu, M., "Modeling Hysteretic Force-Deformation Relationships for Reinforced Concrete
Elements," ACI-SP127, American Concrete Institute, Detroit, 1991, pp. 153-198.

4
11.2 Bilinear Model

At the initial development stage of nonlinear dynamic analysis, the elastic-perfectly plastic
hysteretic model ("elasto-plastic model") was used by many investigators. The response point
moves on the elastic stiffness line before the yield stress is reached. After yielding, the response
point moves on the perfectly plastic line until unloading takes place. Upon unloading, the response
point moves on the line parallel to the initial elastic line.

This model does not consider degradation of stiffness under cyclic loading. Energy dissipation
during a small excursion is not included.

A finite positive slope was assigned to

the stiffness after yielding to simulate the
strain hardening characteristics of the Bilinear Model
steel and the reinforced concrete Specimen SP-5
("bilinear model"). Unloading stiffness
after yielding is equal to the initial elastic
stiffness. The stiffness degradation with
inelastic deformation and energy
Column top force, kN

dissipation during small amplitude

oscillation are not considered in the
model.

Neither the elasto-plastic model nor

the bilinear model represents the
behavior of reinforced concrete and steel
members. The steel member softens
during reloading after plastic deformation
by the "Bauschinger effect." The
Column top displacement, cm
response of the elasto-plastic model is
compared with a test result of a Response of Bilinear model and RC column
reinforced concrete column above.

When the degradation in stiffness was recognized in the behavior of the reinforced concrete, the
loading and unloading stiffness Kr was proposed to degrade with the previous maximum
displacement (Nielsen and Imbeault, 1970) in a form:
Dm −α
Kr = K y ( )
Dy
in which, α : unloading stiffness degradation parameters (0 < α <1); Ky: initial elastic stiffness, and
Dm: previously attained maximum displacement in any direction. The unloading stiffness remains
5
constant until the response displacement amplitude exceeds the previous maximum displacement in
either direction. The model is called a "degrading" bilinear hysteresis model." If the value of a is
chosen to be zero, the unloading stiffness does not degrade with yielding. A smaller value of a tends
to yield a larger residual displacement. The degrading bilinear model does not dissipate hysteretic
energy until the yield is developed. For a reinforced concrete member, the value of α is normally
selected to be around 0.4.

The hysteretic energy dissipation index Eh of

the degrading bilinear model is given by
2(1 − β ){µ − µ α (1 − β + µβ )}
Eh =
π µ (1 − β + µβ )(1 − βµ α )
in which β : ratio of the post-yielding stiffness
to the initial elastic stiffness; and µ : "ductility
factor" (ratio of the maximum displacement to
the initial yield displacement).

The equation is valid for a ductility factor

greater than 1.0. The hysteresis energy index of
a regular bilinear model ( α = 0) reaches as
high as 0.33 at a ductility factor of 4.0. However,
such large amplitude oscillations do not
continue during an earthquake; no hysteresis
energy is dissipated by the model during small
amplitude oscillations. The total energy
dissipation of the bilinear model over the
duration of an earthquake is much smaller than
that expected from the hysteretic energy
dissipation index.

Reference:

Nielsen, N. N., and F. A. Imbeault, "Validity of

Various Hysteretic Systems," Proceedings,
Third Japan National Conference on
Earthquake Engineering, 1971, pp. 707-714.

6
 Appendix FORTRAN PROGRAM LISTING OF BILINEAR HYSTERESIS MODEL

SUBROUTINE HYSTR1 (LL,SS,DD,DS,FF,FS)

C
C BILINEAR HYSTERESIS RULES FOR A GIVEN DISPLACEMENT INCREMENT.
C
C PROGRAMMED BY OTANI, S.
C ON NOVEMBER 11,1978
C AT THE UNIVERSITY OF TORONTO.
C
C INPUT DATA
C FY INITIAL YIELD FORCE LEVEL.
C DY INITIAL YIELD DISPLACEMENT.
C SY INITIAL ELASTIC STIFFNESS.
C SU POST YIELDING STIFFNESS.
C B0 STIFFNESS DEGRADATION FACTOR.
C LL HYSTERESIS RULE POINTER AT PREVIOUS STEP.
C DD DISPLACEMENT AT CURRENT STEP.
C DS DISPLACEMENT AT PREVIOUS STEP.
C
C OUTPUT DATA
C LL HYSTERESIS RULE POINTER AT CURRENT STEP.
C SS STIFFNESS AT CURRENT STEP.
C FF FORCE AT CURRENT STEP.
C
C VARIABLES
C ES VARIABLE ELASTIC STIFFNESS AFTER FIRST YIELDING
C DU UPPER LIMIT DISPLACEMENT FOR POST-YIELDING ELASTIC STAGE.
C DL LOWER LIMIT DISPLACEMENT FOR POST-YIELDING ELASTIC STAGE.
C FL LOWER LIMIT FORCE FOR POST-YIELDING ELASTIC STAGE.
C DMX ABSOLUTE MAXIMUM DISPLACEMENT.
C
C
C BILINEAR HYSTERESIS RULES
C
C RULE 1 INITIAL ELASTIC STAGE.
C FF=SY*DD
C RULE 2 POST-YIELDING STAGE IN POSITIVE DIRECTION.
C FF= FY+(DD-DY)*SU
C RULE 3 POST-YIELDING STAGE IN NEGATIVE DIRECTION.
C FF=-FY+(DD+DY)*SU
C RULE 4 POST-YIELDING ELASTIC STAGE.
C FF=FL+(DD-DL)*SS
C
COMMON /STFF/MD,DC,DY,FC,FY,SC,SY,SU,B0,B1
GO TO (1,2,3,4),LL
C INITIAL ELASTIC STAGE.
1 IF (DY-ABS(DD)) 110,110,100
100 FF=SY*DD
GO TO 1000
110 DMX=DY
IF (DD) 300,200,200
C POST-YIELDING STAGE IN POSITIVE DIRECTION.
2 IF (DD-DS) 220,220,210
200 LL=2
SS=SU
210 FF= FY+(DD-DY)*SU
GO TO 1000
220 IF (DMX.LT.ABS(DS)) DMX=ABS(DS)
ES=SY*(DY/DMX)**B0
DU=DS
DL=(FS+FY-DY*SU-DU*ES)/(SU-ES)
FL=-FY+(DL+DY)*SU
IF (DD-DL) 300,300,400
C POST-YIELDING STAGE IN NEGATIVE DIRECTION.
3 IF (DD-DS) 310,320,320
300 LL=3
SS=SU
7
 310 FF=-FY+(DD+DY)*SU
GO TO 1000
320 IF (DMX.LT.ABS(DS)) DMX=ABS(DS)
ES=SY*(DY/DMX)**B0
DL=DS
FL=-FY+(DL+DY)*SU
DU=(FY-FL+DL*ES-DY*SU)/(ES-SU)
IF (DD-DU) 400,200,200
C POST-YIELDING ELASTIC STAGE.
4 IF (DD-DU) 420,200,200
400 LL=4
SS=ES
410 FF=FL+(DD-DL)*ES
GO TO 1000
420 IF (DD-DL) 300,300,410
C
1000 RETURN
END

8
11.3 Ramberg-Osgood Model

γ by
A stress-strain relation of the metal was expressed using three parameters D y , Fy and
Ramberg and Osgood (1943), where D y : yield displacement, Fy : yield resistance and γ : a
parameter of the model. Jennings (1963) introduced the fourth parameter η to the model. The
initial loading curve of the model under monotonically increasing deformation, as modified by
Jennings, is expressed by
γ −1
D F F
= (1 + η )
D y Fy Fy
in which, γ : exponent of the Ramberg-Osgood model; and η : parameter introduced by Jennings
(1963).

The initial tangent modulus is equal to (Fy/Dy), and the initial loading curve passes a point (Fy,
(1+ η )Dy) for any value of γ . The shape of the primary curve can be controlled by the exponent γ
from linearly elastic ( γ = 1.0) to elasto-plastic ( γ = infinity). For a larger value of γ , the behavior
becomes similar to that of the bilinear model.

Upon unloading from a peak response point (Do, Fo), the unloading, load reversal and reloading
branches of the relationship is given by
γ −1
D − Do F − Fo F − Fo
= (1 + η )
2Dy 2 Fy 2 Fy
until the response point reaches the peak point of one outer hysteresis loop.

The resistance F is not explicitly expressed by a given displacement D in this model. The
resistance F at a given displacement D must be computed numerically, for example, using the
Newton-Rapson's iterative procedure.

The Ramberg-Osgood model is often used for stress-strain relation of the steel in the finite
element analysis or in the lamina model, and for resistance-deformation relation of steel members in
a frame analysis.

The hysteresis energy dissipation index of the Ramberg-Osgood model is expressed as

2 2η D F
Eh = (1 − )(1 − y m )
π 1+γ Fy Dm
The model can dissipate some hysteresis energy even if the ductility factor is less than unity. The
9
index is sensitive to the exponent γ
of the model, and the hysteresis
energy dissipation capacity increases
with increasing value of the exponent.

References:

Jennings, P. C., "Response of Simple

Yielding Structures to Earthquake
Excitation," Ph.D. Thesis,
California Institute of Technology,
Pasadena, 1963.
Ramberg, W., and W. R. Osgood,
"Description of Stress-Strain
Curves by Three Parameters,"
National Advisory Committee on
Aeronautics, Technical Note 902,
1943.

10
 SUBROUTINE HYSTR2 (LL,SS,DD,DS,FF,FS)
C
C RAMBERG-OSGOOD HYSTERESIS MODEL
C
C PROGRAMMED BY OTANI, S.
C ON FEBRUARY 9, 1979
C AT UNIVERSITY OF TORONTO
C
C
C (DD/DY)=(FF/FY)*(1.0+ABS(FF/FY)**(B0-1.0))
C
C THIS PROGRAM FINDS THE RESISTANCE AT A GIVEN DISPLACEMENT.
C THE NEWTON-RAPSON'S METHOD IS USED TO SOLVE THE NONLINEAR PROBLEM.
C
C
C INPUT DATA
C LL HYSTERESIS RULE POINTER
C DD CURRENT DISPLACEMENT
C DS PREVIOUS DISPLACEMENT
C FS PREVIOUS RESISTANCE
C FY YIELD RESISTANCE
C DY YIELD DISPLACEMENT
C B0 RAMBERG-OSGOOD PARAMETER
C B1 CONVERGENCE LIMIT
C
C OUTPUT DATA
C FF CURRENT RESISTANCE
C SS CURRENT TANGENT STIFFNESS
C
COMMON /STFF/MD,DC,DY,FC,FY,SC,SY,SU,B0,B1
C
GO TO (1,2,3,4,5,6,7,8,9,10), LL
C RULE 1 LOADING ON PRIMARY CURVE
1 CONTINUE
IF ((DD-DS)*DS) 120,110,110
100 CONTINUE
LL=1
110 CONTINUE
XX=DD/DY
QQ=FS/FY
CALL RAMOSG (XX,QQ,B0,B1)
FF=QQ*FY
SS=FY/DY/XX
GO TO 10000
120 CONTINUE
D0=DS
F0=FS
130 CONTINUE
IF (ABS(DD)-ABS(D0)) 200,200,100
C RULE 2 UNLOADING FROM POINT (D0,F0) ON PRIMARY CURVE
2 CONTINUE
IF ((DS-D0)*(DD-DS)) 230,220,220
200 CONTINUE
LL=2
210 CONTINUE
XX=(DD-D0)/(DY+DY)
QQ=(FS-F0)/(FY+FY)
CALL RAMOSG (XX,QQ,B0,B1)
FF=QQ*(FY+FY)+F0
SS=FY/DY/XX
GO TO 10000
220 CONTINUE
IF (ABS(DD)-ABS(D0)) 210,210,100
230 CONTINUE
D1=DS
F1=FS
240 IF (ABS(DD)-ABS(D0)) 300,300,100
C RULE 3 UNLOADING FROM POINT (D1,F1) ON FIRST INNER LOOP
11
 3 CONTINUE
IF ((DS-D1)*(DD-DS)) 330,320,320
300 CONTINUE
LL=3
310 CONTINUE
XX=(DD-D1)/(DY+DY)
QQ=(FS-F1)/(FY+FY)
CALL RAMOSG (XX,QQ,B0,B1)
FF=QQ*(FY+FY)+F1
SS=FY/DY/XX
GO TO 10000
320 CONTINUE
IF (ABS(DD)-ABS(D0)) 310,310,100
330 CONTINUE
D2=DS
F2=FS
340 CONTINUE
IF ((D1-D2)*(DD-D1)) 400,130,130
C RULE 4 UNLOADING FROM POINT (D2,F2) ON FIRST INNER LOOP
4 CONTINUE
IF ((D1-D2)*(DD-DS)) 430,420,420
400 CONTINUE
LL=4
410 CONTINUE
XX=(DD-D2)/(DY+DY)
QQ=(FS-F2)/(FY+FY)
CALL RAMOSG (XX,QQ,B0,B1)
FF=QQ*(FY+FY)+F2
SS=FY/DY/XX
GO TO 10000
420 CONTINUE
IF ((D1-D2)*(DD-D1)) 410,130,130
430 CONTINUE
D3=DS
F3=FS
440 CONTINUE
IF ((D1-D2)*(DD-D2)) 240,240,500
C RULE 5 UNLOADING FROM POINT (D3,F3) ON SECOND INNER LOOP
5 CONTINUE
IF ((D2-D3)*(DD-DS)) 530,520,520
500 CONTINUE
LL=5
510 CONTINUE
XX=(DD-D3)/(DY+DY)
QQ=(FS-F3)/(FY+FY)
CALL RAMOSG (XX,QQ,B0,B1)
FF=QQ*(FY+FY)+F3
SS=FY/DY/XX
GO TO 10000
520 CONTINUE
IF ((D2-D3)*(DD-D2)) 510,240,240
530 CONTINUE
D4=DS
F4=FS
540 CONTINUE
IF ((D2-D3)*(DD-D3)) 340,340,600
C RULE 6 UNLOADING FROM POINT (D4,F4) ON SECOND INNER LOOP
6 CONTINUE
IF ((D3-D4)*(DD-DS)) 630,620,620
600 CONTINUE
LL=6
610 CONTINUE
XX=(DD-D4)/(DY+DY)
QQ=(FS-F4)/(FY+FY)
CALL RAMOSG (XX,QQ,B0,B1)
FF=QQ*(FY+FY)+F4
SS=FY/DY/XX
GO TO 10000
12
 620 CONTINUE
IF ((D3-D4)*(DD-D3)) 610,340,340
630 CONTINUE
D5=DS
F5=FS
640 CONTINUE
IF ((D3-D4)*(DD-D4)) 440,440,700
C RULE 7 UNLOADING FROM POINT (D5,F5) ON THIRD INNER LOOP
7 CONTINUE
IF ((D4-D5)*(DD-DS)) 730,720,720
700 CONTINUE
LL=7
710 CONTINUE
XX=(DD-D5)/(DY+DY)
QQ=(FS-F5)/(FY+FY)
CALL RAMOSG (XX,QQ,B0,B1)
FF=QQ*(FY+FY)+F5
SS=FY/DY/XX
GO TO 10000
720 CONTINUE
IF ((D4-D5)*(DD-D4)) 710,440,440
730 CONTINUE
D6=DS
F6=FS
740 CONTINUE
IF ((D4-D5)*(DD-D5)) 540,540,800
C RULE 8 UNLOADING FROM POINT (D6,F6) ON THIRD INNER LOOP
8 CONTINUE
IF ((D5-D6)*(DD-DS)) 830,820,820
800 CONTINUE
LL=8
810 CONTINUE
XX=(DD-D6)/(DY+DY)
QQ=(FS-F6)/(FY+FY)
CALL RAMOSG (XX,QQ,B0,B1)
FF=QQ*(FY+FY)+F6
SS=FY/DY/XX
GO TO 10000
820 CONTINUE
IF ((D5-D6)*(DD-D5)) 810,540,540
830 CONTINUE
D7=DS
F7=FS
840 CONTINUE
IF ((D5-D6)*(DD-D6)) 640,640,900
C RULE 9 UNLOADING FROM POINT (D7,F7) ON FOURTH INNER LOOP
9 CONTINUE
IF ((D6-D7)*(DD-DS)) 930,920,920
900 CONTINUE
LL=9
910 CONTINUE
XX=(DD-D7)/(DY+DY)
QQ=(FS-F7)/(FY+FY)
CALL RAMOSG (XX,QQ,B0,B1)
FF=QQ*(FY+FY)+F7
SS=FY/DY/XX
GO TO 10000
920 CONTINUE
IF ((D6-D7)*(DD-D6)) 910,640,640
930 CONTINUE
D8=DS
F8=FS
940 CONTINUE
IF ((D6-D7)*(DD-D7)) 740,740,1000
C RULE 10 LINEAR RELATION BETWEEN POINTS (D7,F7) AND (D8,F8)
10 CONTINUE
IF ((D7-D8)*(DD-D8)) 840,1020,1020
1000 CONTINUE
13
 LL=10
1010 CONTINUE
SS=(F7-F8)/(D7-D8)
FF=F8+(DD-D8)*SS
GO TO 10000
1020 CONTINUE
IF ((D7-D8)*(DD-D7)) 1010,740,740
C
10000 CONTINUE
RETURN
END
SUBROUTINE RAMOSG (XX,QQ,B0,ERR)
C
C DETERMINATION OF FORCE LEVEL QQ AT A GIVEN DISPLACEMENT XX OF A
C RAMBERG-OSGOOD HYSTERESIS MODEL BY THE NEWTON-RAPSON'S ITERATIVE
C PROCEDURE.
C
C XX=QQ*(1.0+ABS(QQ)**(B0-1.0))
C
C
C INPUT DATA
C XX CURRENT DISPLACEMENT
C QQ INITIAL VALUE OF RESISTANCE
C B0 RAMBERG-OSGOOD PARAMETER
C ERR CONVERGENCE LIMIT
C
C OUTPUT
C QQ CURRENT RESISTANCE OF THE HYSTERESIS MODEL
C XX CURRENT STIFFNESS
C =1.0+B0*ABS(QQ)**(B0-1.0)
C
10 CONTINUE
Q0=QQ
EX=ABS(Q0)**(B0-1.0)
GQ=Q0*(1.0+EX)-XX
TQ=1.0+B0*EX
QQ=Q0-GQ/TQ
IF (ABS(QQ-Q0)/(ABS(Q0)+ABS(QQ)).LT.ERR) GO TO 20
GO TO 10
20 CONTINUE
XX=1.0+B0*ABS(QQ)**(B0-1.0)
RETURN
END

14
11.4 Degrading Tri-linear Model

A model that simulates dominantly flexural stiffness characteristics of the reinforced concrete was
used extensively in Japan (Fukada, 1969). The primary curve is of tri-linear shape with stiffness
changes at flexural cracking and yielding. Up to yielding, the model behaves in a manner the same
as the bilinear model. When the response exceeds a yield point, response point follows the
strain-hardening part of the
primary curve. Once
unloading takes place from
a point on the primary curve,
the unloading point is
considered to be a new
"yield point" in the direction.
The model behaves in a
bilinear manner between
the positive and negative
"yield points" with stiffness
degraded proportional to
the ratio of the slopes
connecting "current yield
points" and "the initial yield Degrading tri-linear model
points."

The ratio of the first and second stiffness is kept constant even after yielding.

This model has the following properties:

(a) the stiffness continuously degrades
with increasing maximum amplitude beyond
yielding,
(b) the hysteretic energy dissipation is
large in the first load reversal cycle after
yielding, and becomes steady in the following
cycles, and
(c) the steady hysteretic energy dissipation
is proportional to the displacement amplitude.

The hysteretic energy dissipation index of

the degrading tri-linear model is expressed as
2 K y Fc
Eh = (1 − )
π K c Fy
in which Ky: secant stiffness at yielding (=
Fy / D y ), and K c : initial elastic stiffness (=
Fc / Dc ). The index is independent of the
displacement amplitude, but dependent on
the stiffness and resistance ratios at cracking
and yielding. Cracking point of this model
controls the fatness of a hysteresis loop.
Therefore, it is important to choose the
cracking point taking into account the degree
of a hysteresis loop.

Nomura (1976) used an arbitrary skeleton

curve; when the response point reached the previous maximum response point, it moves on the
skeleton curve. Upon unloading, the newly attained maximum response point was considered as the
yield point in the direction, similar to the degrading tri-linear model.

15
References:

Fukada, Y., "Study on the Restoring Force

Characteristics of Reinforced Concrete Buildings
(in Japanese)," Proceedings, Kanto Branch
Symposium, Architectural Institute of Japan, No.
40, 1969, pp. 121-124.
Nomura, S., "Restoring Characteristics and their
Modeling," Data for Earthquake Resistant
Design for Buildings, No. 65, Magazine of
Architectural Institute of Japan, June 1976.
Nomura model (1976)

16
 SUBROUTINE HYST6 (LL,SS,DD,DS,FF,FS)
C
C DEGRADING TRILINEAR MODEL
C
C PROGRAMMED BY OTANI, S.
C ON FEBRUARY 9, 1979
C AT THE UNIVERSITY OF TORONTO.
C
C THE MODEL BEHAVES AS A BILINEAR MODEL BETWEEN POSITIVE (DX,FX) AND
C NEGATIVE (DN,FN) YIELD POINTS WITH A CHANGE OF STIFFNESS AT A
C CRACKING POINT (DU,FU) OR (DL,FL). ONCE THE RESPONSE EXCEEDS A
C YIELD DISPLACEMENT (DX OR DN), THE RESPONSE FOLLOWS THE PRIMARY
C CURVE. THE MAXIMUM RESPONSE POINT (DX,FX) OR (DN,FN) IS NOW
C TREATED AS A NEW YIELD POINT IN THAT DIRECTION. THE STIFFNESSES
C ARE MODIFIED ACCORDINGLY.
C
C REFERENCE
C FUKADA, Y.,'STUDY ON THE RESTORING FORCE CHARACTERISTICS OF
C REINFORCED CONCRETE BUILDINGS (IN JAPANESE)', PROCEEDINGS,KANTO
C DISTRICT SYMPOSIUM, AIJ, NO.40, NOV. 1969, PP.121-4.
C
C THE PROGRAM COMPUTES THE RESISTANCE OF THE MODEL AT A GIVEN
C DISPLACEMENT.
C
C
C INPUT DATA
C FC INITIAL CRACKING FORCE LEVEL
C DC INITIAL CRACKING DISPLACEMENT
C FY INITIAL YIELD FORCE LEVEL.
C DY INITIAL YIELD DISPLACEMENT.
C SC INITIAL ELASTIC STIFFNESS
C SY STIFFNESS AFTER INITIAL CRACKING, BUT BEFORE
C INITIAL YIELDING.
C SU POST YIELDING STIFFNESS.
C LL HYSTERESIS RULE POINTER AT PREVIOUS STEP.
C SS STIFFNESS AT PREVIOUS STEP.
C DD DISPLACEMENT AT CURRENT STEP.
C DS DISPLACEMENT AT PREVIOUS STEP.
C FS RESISTANCE AT PREVIOUS STEP
C
C OUTPUT DATA
C LL HYSTERESIS RULE POINTER AT CURRENT STEP.
C SS STIFFNESS AT CURRENT STEP.
C FF FORCE AT CURRENT STEP.
C
C VARIABLES
C DU UPPER LIMIT DISPLACEMENT FOR POST-CRACKING ELASTIC
C STAGE.
C DL LOWER LIMIT DISPLACEMENT FOR POST-CRACKING ELASTIC
C STAGE.
C FL LOWER LIMIT FORCE FOR POST-CRACKING ELASTIC STAGE.
C FU UPPER LIMIT FORCE FOR POST-CRACKING ELASTIC STAGE.
C DX POSITIVE MAXIMUM DISPLACEMENT
C DN NEGATIVE MAXIMUM DISPLACEMENT
C FX POSITIVE MAXIMUM RESISTANCE
C FN NEGATIVE MAXIMUM RESISTANCE
C B0 STIFFNESS DEGRADATION FACTOR.
C =(DY*(FX-FN))/(FY*(DX-DN))
C
C
C DEGRADING TRILINEAR HYSTERESIS RULES
C
C RULE 1 INITIAL ELASTIC STAGE.
C FF=SC*DD
C RULE 2 LOADING IN POST-CRACKING STAGE IN POSITIVE DIRECTION
C S1=SY*B0
C FF=FU+(DD-DU)*S1
C RULE 3 LOADING IN POST-CRACKING STAGE IN NEGATIVE DIRECTION
17
C S1=SY*B0
C FF=FL+(DD-DL)*S1
C RULE 4 POST-CRACKING ELASTIC STAGE
C S0=SC*B0
C FF=FL+(DD-DL)*S0
C RULE 5 LOADING IN POST-YIELDING STAGE IN POSITIVE DIRECTION
C FF= FY+(DD-DY)*SU
C RULE 6 LOADING IN POST-YIELDING STAGE IN NEGATIVE DIRECTION
C FF=-FY+(DD+DY)*SU
C
COMMON /STFF/MD,DC,DY,FC,FY,SC,SY,SU,B0,B1
C
GO TO (1,2,3,4,5,6), LL
C RULE 1 INITIAL ELASTIC STGE BEFORE CRACKING AT (DC,FC)
1 IF (DC-ABS(DD)) 110,110,100
100 FF=SC*DD
GO TO 1000
110 DX= DY
DN=-DY
FX= FY
FN=-FY
S0=SC
S1=SY
IF (DD.GT.DY) GO TO 500
IF (DD.LT.-DY) GO TO 600
IF (DD) 300,200,200
C RULE 2 LOADING IN POST-CRACKING STAGE IN POSITIVE DIRECTION
2 IF (DD-DS) 230,230,220
200 LL=2
SS=S1
210 FF=FX+(DD-DX)*S1
GO TO 1000
220 IF (DX-DD) 500,500,210
230 DU=DS
FU=FS
DL=(FU-FN+DN*S1-DU*S0)/(S1-S0)
FL=FU+(DL-DU)*S0
IF (DL-DD) 400,240,240
240 IF (DN-DD) 300,600,600
C RULE 3 LOADING IN POST-CRACKING STAGE IN NEGATIVE DIRECTION
3 IF (DD-DS) 320,330,330
300 LL=3
SS=S1
310 FF=FN+(DD-DN)*S1
GO TO 1000
320 IF (DN-DD) 310,600,600
330 DL=DS
FL=FS
DU=(FX-FL-DX*S1+DL*S0)/(S0-S1)
FU=FL+(DU-DL)*S0
IF (DU-DD) 340,340,400
340 IF (DX-DD) 500,500,200
C RULE 4 POST-CRACKING ELASTIC STAGE
4 IF (DU-DD) 340,340,420
400 LL=4
SS=S0
410 FF=FL+(DD-DL)*S0
GO TO 1000
420 IF (DL-DD) 410,240,240
C RULE 5 LOADING IN POST-YIELDING STAGE IN POSITIVE DIRECTION
5 IF (DD-DS) 520,520,510
500 LL=5
SS=SU
510 FF= FY+(DD-DY)*SU
GO TO 1000
520 DX=DS
FX=FS
B0=DY*(FX-FN)/(FY*(DX-DN))
18
 S0=SC*B0
S1=SY*B0
GO TO 230
C RULE 6 LOADING IN POST-YIELDING STAGE IN NEGATIVE DIRECTION
6 IF (DD-DS) 610,620,620
600 LL=6
SS=SU
610 FF=-FY+(DD+DY)*SU
GO TO 1000
620 DN=DS
FN=FS
B0=DY*(FX-FN)/(FY*(DX-DN))
S0=SC*B0
S1=SY*B0
GO TO 330
C
1000 RETURN
END

19
11.5 Clough Degrading Model

A hysteretic model with an elasto-plastic C

skeleton curve was proposed by Clough F Y
and Johnston (1966) to represent the A
hysteretic behavior of a reinforced concrete
beam-column sub-assemblage. Kr=Ky
Ky
During loading, the response point B

follows the elasto-plastic skeleton curve.

D
The unloading stiffness after yielding was
kept equal to the initial elastic stiffness. The
response point during reloading moves
toward the previous maximum response
point in the direction of reloading, simulating Y
the stiffness degradation. If yielding has not Clough Model
taken place in the direction of reloading, the
response point moves toward the yield point C
in the reloading direction. F
Y
A minor deficiency of the Clough model Fy A

was pointed out by Mahin and Bertero

(1976). After unloading from point A, Kr
consider a situation in which reloading takes Ky
B D
place from point B. The original Clough
model assumed that the response point Dy Dm
should move toward the previous maximum
response point C. This is not realistic.
Therefore, a minor modification was added
so that the response point should move
toward an immediately preceding unloading Y
point A during reloading. When the Modified Clough Model
response point reaches the point A, the
response point moves toward the previous maximum point C.

The model was made more versatile by incorporating the reduction in unloading stiffness Kr with
a maximum displacement in a form:
Dm −α
Kr = K y ( )
Dy
200
in which, α : unloading stiffness
Clough Model
degradation parameter; K y : initial elastic
100 RC Column
stiffness; and Dm : previous maximum
Column Resistance, kN

displacement. The different unloading

stiffness may be assigned taking Dm to be
0
a maximum deformation in the direction
unloading takes place.
-100
If the value of a is chosen to be zero, the
unloading stiffness of the model remains
equal to the initial elastic stiffness.
-200
The response of the Clough model is -100 -50 0 50 100
shown to compare well with the response of Column Top Displacement, mm
a reinforced concrete column tested in the
structures laboratory.

20
 Saiidi and Sozen (1979) and Riddell and Newmark (1979) used models similar to the modified
Clough model.

Wang and Shah (1987) introduced the strength and stiffness degradation effect of cumulative
damage. The strength and stiffness degrade in proportion to (1-Dws), where Dws is the Wang and
Shah damage index. The ordinates of the bilinear skeleton curve in monotonic loading is multiplied
by the current value of (1-Dws). Unloading and reloading stiffness is reduced by the same amount, as
they are defined on the basis of the location of the point of reversal and of the maximum previous
deformation in the direction of loading, on the degraded skeleton curve. The Wang and Shah
damage index is defined separately for each direction of loading as
enδ − 1
Dws = n
e −1
where the damage prameter δ is expressed in terms of chord rotation,
∑θ i
δ =c i

θu

The hysteretic energy dissipation index

of the modified Clough model is expressed
as
1 (1 − β + µβ ) µ α
Eh = {1 − }
π µ
where β : ratio of post-yielding stiffness to
the initial elastic stiffness, and µ : ductility
factor.

The equation is valid for ductility factor

greater than unity. The Clough model can
continuously dissipate hysteretic energy
even at a small amplitude oscillation after
yielding.

References:

Clough, R. W., and S. B. Johnston, "Effect

of Stiffness Degradation on
Earthquake Ductility Requirements,"
Proceedings, Second Japan National
Conference on Earthquake
Engineering, 1966, pp. 227-232.
Mahin, S. A., and V. V. Bertero, "Rate of
Loading Effect on Uncracked and Repaired Reinforced Concrete Members," EERC No. 73-6,
Earthquake Engineering Research Center, University of California, Berkeley, 1972.
Riddell, R., and N. M. Newmark, "Statistical Analysis of the Response of Nonlinear Systems
subjected to Earthquakes," Structural Research Series No. 468, Civil Engineering Studies,
University of Illinois at Urbana-Champaign, Illinois, 1979.
Saiidi, M., and M. A. Sozen, "Simple and Complex Models for Nonlinear Seismic Response of
Reinforced Concrete Structures," Structural Research Series No. 465, Civil Engineering Studies,
University of Illinois at Urbana-Champaign, Illinois, 1979.
Wang, M.-L., and S. P. Shah, “Reinforced Concrete Hysteresis Model based on the Damage
Concept,” Earthquake Engineering and Structural Dynamics, John Wiley & Sons, Chichester,
Sussex, Vol. 15, 1987, pp. 993 -1003.

21
 SUBROUTINE HYSTR3 (LL,SS,DD,DS,FF,FS)
C
C CLOUGH'S DEGRADING HYSTERESIS MODEL WITH TWO MODIFICATIONS.
C (1) UNLOADING STIFFNESS DEGRADES WITH MAXIMUM DISPLACEMENT ON
C THE SIDE OF UNLOADING POINT,
C (2) RELOADING STIFFNESS, AFTER UNLOADING IN THE SAME STRESS
C REGION, IS THE SAME AS THE UNLOADING STIFFNESS UP TO THE
C INNER UNLOADING POINT.
C
C PROGRAMMED BY OTANI, S.
C ON FEBRUARY 9, 1979
C AT UNIVERSITY OF TORONTO
C
C
C THE CLOUGH MODEL ASSUMES THE YIELD POINT TO BE THE PREVIOUS MAXIMUM
C DISPLACEMENT POINT IN THE DIRECTION OF LOADING.
C THE RESISTANCE INCREASES TOWARD THIS VARIABLE YIELD POINT IN THE
C DIRECTION OF LOADING. THE UNLOADING STIFFNESS IS MODIFIED IN THIS
C PROGRAM TO DEGRADE WITH THE VALUE OF MAXIMUM DISPLACEMENT ON THE
C SIDE OF UNLOADING POINT. THIS PROGRAM RETURNS TO THE CALLING
C PROGRAM WITH NEW HYSTERESIS POINT VALUES.
C
C
C INPUT INFORMATION
C DY INITIAL YIELD DISPLACEMENT
C FY INITIAL YIELD RESISTANCE
C SY ELASTIC STIFFNESS
C SU POST-YIELDING STIFFNESS
C B0 CONSTANT DEFINING DEGRADATION OF STIFFNESS
C LL HYSTERESIS CASE NUMBER AT PREVIOUS STEP
C DD DISPLACEMENT AT CURRENT STEP
C FS RESISTANCE AT PREVIOUS STEP
C DS DISPLACEMENT AT PREVIOUS STEP
C
C OUTPUT INFORMATION
C LL HYSTERESIS CASE NUMBER AT CURRENT STEP
C SS STIFFNESS AT CURRENT STEP
C FF RESISTANCE AT CURRENT STEP
C
C VARIABLES
C FX(1) NEGATIVE MAXIMUM RESISTANCE REACHED
C FX(2) POSITIVE MAXIMUM RESISTANCE REACHED
C DX(1) NEGATIVE MAXIMUM DISPLACEMENT REACHED
C DX(2) POSITIVE MAXIMUM DISPLACEMENT REACHED
C ES(1) UNLOADING STIFFNESS AFTER YIELDING IN NEGATIVE SIDE.
C ES(2) UNLOADING STIFFNESS AFTER YIELDING IN POSITIVE SIDE.
C XD DISPLACEMENT AT CROSSING OF ZERO-RESISTANCE AXIS IN
C HYSTERESIS RULE 3.
C F1 RESISTANCE AT INNER UNLOADING POINT
C D1 DISPLACEMENT AT INNER UNLOADING POINT
C X1 DISPLACEMENT AT CROSSING OF ZERO-RESISTANCE AXIS IN
C HYSTERESIS RULE 5.
C DC DUMMY VARIABLE
C FC DUMMY VARIABLE
C SC DUMMY VARIABLE
C B1 DUMMY VARIABLE
C
C
C CLOUGH HYSTERESIS RULES
C
C RULE 1 INITIAL ELASTIC STAGE
C FF=SY*DD
C RULE 2 POST-YIELDING STAGE
C FF=SN*FY+(DD-SN*DY)*SU
C RULE 3 UNLOADING FROM POST-YIELDING POINT (DMX(IS),FMX(IS))
C FF=FX(IS)+(DD-DX(IS))*ES(IS)
C RULE 4 LOADING TOWARD PREVIOUS MAXIMUM POINT (DMX(IS),FMX(IS))
C FF=FX(IS)*(DD-XD)/(DX(IS)-XD)
22
C RULE 5 UNLOADING FROM INNER PEAK POINT (D1,F1)
C FF=F1+(DD-D1)*ES(IS)
C
COMMON /STFF/MD,DC,DY,FC,FY,SC,SY,SU,B0,B1
DIMENSION DX(2),FX(2),ES(2)
C
C IS=1 NEGATIVE RESISTANCE AT PREVIOUS STEP
C IS=2 POSITIVE RESISTANCE AT PREVIOUS STEP
IS=2
IF (FS.LT.0.0) IS=1
SN=FLOAT(IS+IS-3)
C GO TO PREVIOUS HYSTERESIS RULE
GO TO (1,2,3,4,5), LL
C RULE 1 = INITIAL ELASTIC STAGE
1 IF (DY-ABS(DD)) 110,110,100
100 FF=SY*DD
GO TO 1000
110 DX(1)=-DY
DX(2)= DY
FX(1)=-FY
FX(2)= FY
ES(1)=SY
ES(2)=SY
GO TO 200
C RULE 2 = LOADING BEYOND YIELD POINT (DX,FX)
2 IF ((DD-DS)*SN) 220,220,210
200 LL=2
SS=SU
210 FF=SN*FY+(DD-SN*DY)*SU
GO TO 1000
220 DX(IS)=DS
FX(IS)=FS
ES(IS)=SY*(DY/ABS(DS))**B0
XD=DX(IS)-FX(IS)/ES(IS)
IF ((XD-DD)*SN) 300,230,230
230 IS=3-IS
SN=FLOAT(IS+IS-3)
240 IF ((DX(IS)-DD)*SN) 200,200,400
C RULE 3 = UNLOADING FROM YIELD POINT (DMX,FMX)
3 IF ((XD-DD)*SN) 320,230,230
300 LL=3
SS=ES(IS)
310 FF=FX(IS)+(DD-DX(IS))*ES(IS)
GO TO 1000
320 IF ((DX(IS)-DD)*SN) 200,200,310
C RULE 4 = LOADING TOWARD YIELD POINT (DMX,FMX)
4 IF ((DD-DS)*SN) 430,430,420
400 LL=4
SS=FX(IS)/(DX(IS)-XD)
410 FF=SS*(DD-XD)
GO TO 1000
420 IF ((DX(IS)-DD)*SN) 200,200,410
430 D1=DS
F1=FS
X1=D1-F1/ES(IS)
IF ((X1-DD)*SN) 500,440,440
440 XD=X1
GO TO 230
C RULE 5 = UNLOADING FROM INNER PEAK (D1,F1)
5 IF ((X1-DD)*SN) 520,440,440
500 LL=5
SS=ES(IS)
510 FF=F1+(DD-D1)*ES(IS)
GO TO 1000
520 IF ((D1-DD)*SN) 240,240,510
C
1000 RETURN
END
23
11.6 Takeda Degrading Model

Based on the experimental observation on the behavior of a number of medium-size reinforced

concrete members tested under lateral load reversals with light to medium amount of axial load, a
hysteresis model was developed by Takeda, Sozen and Nielsen (1970). The model has been widely
used in the nonlinear earthquake response analysis of reinforced concrete structures.

Takeda Hysteresis Model:

1. Condition: The cracking load, Pcr, has not been exceeded in one direction. The load is reversed
from a load P in the other direction. The load P is smaller than the yield load Py.
Rule: Unloading follows a straight line from the position at load P to the point representing the
cracking load in the other direction.

2. Condition: A load P1 is reached in one direction on the primary curve such that P1 is larger than
Pcr but smaller the yield load Py. The load is then reversed to -P2 such that P2<P1.
Rule: Unload parallel to loading curve for that half cycle.

3. Condition: A load P1 is reached in one direction such that P1 is larger than Pcr but not larger
than the yield load Py. The load is then reversed to -P3 such that P3>P1.
Rule: Unloading follows a straight line joining the point of return and the point representing
cracking in the other direction.

4. Condition: One or more loading cycles have occurred. The load is zero.
Rule: To construct the loading curve, connect the point at zero load to the point reached in the
previous cycle, if that point lies on the primary curve or on a line aimed at a point on the primary
curve. If the previous loading cycle contains no such point, go to the preceding cycle and continue
the process until such a point is found. Then connect that point to the point at zero load.
Exception: If the yield point has not been exceeded and if the point at zero load is not located
within the horizontal projection of the primary curve for that direction of loading, connect the point at
zero load to the yield point to obtain the loading slope.

5. Condition: The yield load Py is exceeded in one direction.

Rule: Unloading curve follows the slope given by the following equation:
Dy
kr = k y ( ) 0.4
D
in which k r : slope of unloading curve, k y : slope of a line joining the yield point in one direction to
the cracking point in the other direction, D : maximum deflection attained in the direction of the
loading, and D y : deflection at yield.

6. Condition: The yield load is exceeded in one direction but the cracking load is not exceeded in
the opposite direction.
Rule: Unloading follows Rule 5. Loading in the other direction continues as an extension of the
unloading line up to the cracking load. Then, the loading curve is aimed at the yield point.

7. Condition: One or more loading cycles have occurred.

Rule: If the immediately preceding quarter-cycle remained on one side of the zero-load axis,
unload at the rate based on rule 2, 3 and 5 whichever governed in the previous loading history. If the
immediately preceding quarter-cycle crossed the zero-load axis, unload at 70% of the rate based on
rule 2, 3, or 5, whichever governed in the previous loading history, but not at a slope flatter than the
immediately preceding loading slope.

Takeda model included (a) stiffness changes at flexural cracking and yielding, (b) hysteresis rules
for inner hysteresis loops inside the outer loop, and (c) unloading stiffness degradation with
deformation. The response point moves toward a peak of the one outer hysteresis loop. The
unloading stiffness Kr after yielding is given by

24
 −α
Fc + Fy Dm
Kr =
Dc + D y D y
in which, α : unloading stiffness degradation parameter; and Dm : previous maximum displacement
beyond yielding in the direction concerned. The hysteresis rules are extensive and comprehensive.

The hysteretic energy dissipation index of

the Takeda model is expressed as
Dc
1+
1 D y µ α (1 − β + µβ )
E h = {1 − }
π 1+
Fc µ
Fy
The expression is valid for a ductility factor
greater than unity.

It should be noted that the Takeda

hysteresis rule was originally developed to
simulate the behavior of reinforced concrete
members. If this model is used to simulate the
behavior of a story or a simplified structure,
some rules need to be simplified.

For example, hysteresis rules prior to

yielding may be simplified such that unloading
takes place toward the origin of the relation
(Muto Model). This model is often used in a
story-based (mass-spring) earthquake
response analysis.

25
Bilinear Takeda Model: The primary curve of the Takeda model can be made bilinear simply
choosing the cracking point to be the origin of the hysteretic plane. Such a model is called the
"bilinear Takeda" model, similar to the Clough model except that the bilinear Takeda model has
more hysteresis rules for inner hysteresis loops (Otani and Sozen, 1972); i.e., the response point
moves toward an unloading point on the immediately outer hysteresis loop.

The behavior before yielding is

F
sometimes made simple by letting the
response point moves toward the (D2,F2) (D0,F0)
origin during unloading, and toward
the maximum response point in the
opposite side upon reloading. The D’m X0 D
X1
Takeda hysteresis rules are applied
after the yielding. X3 Dm
(D3,F3)
This model is similar to Clough
(D1,F1)
Degrading Model, but is more
complicated having rules for inner
hysteresis loops.

Additional modifications of the Takeda model with bilinear backbone curve may be found in
literature (Powell, 1975, Riddle and Newmark, 1979, Saiidi and Sozen, 1979, Saiidi, 1982). Riddle
and Newmark (1979) used a bilinear skeleton curve and unloading stiffness equal to the initial elastic
stiffness; loading occurs either on the strain hardening branch or towards the furthest point attained
in the previous cycle. Saiidi and Sozen (1979) claimed to simplify the Takeda model using a bilinear
skeleton curve; the model, however, is identical to the modified Clough model with reduced
unloading stiffness with maximum deformation, and reloading to the immediate prior unloading point
if reloading occurs during unloading and then to the unloading point on the skeleton curve.

References:

Takeda, T., M. A. Sozen and N. N. Nielsen, "Reinforced Concrete Response to Simulated

Earthquakes," Journal, Structural Division, ASCE, Vol. 96, No. ST12, 1970, pp. 2557-2573.
Otani, S., and M. A. Sozen, "Behavior of Multistory Reinforced Concrete Frames During
Earthquakes," Structural Research Series No. 392, Civil Engineering Studies, University of
Illinois, Urbana, 1972.
Powell, G. H., “Supplement to Computer Program DRAIN-2D,” Supplement to Report, DRAIN-2D
User’s Guide, University of California, Berkeley, August 1975.
Riddle, R., and N. M. Newmark, “Statistical Analysis of the Response of Nonlinear Systems
subjected to Earthquakes,” Structural Research Series No. 468, Civil Engineering Studies,
University of Illinois, Urbana, 1979.
Saiidi, M., “Hysteresis Models for Reinforced Concrete,” Journal, Structural Division, ASCE, Vol. 108,
No. ST5, May 1982, pp. 1077 - 1087.
Saiidi, M., and M. A. Sozen, “Simple and Complex Models for Nonlinear Seismic Response of
Reinforced Concrete Structures,” Structural Research Series No. 465, Civil Engineering Studies,
University of Illinois, Urbana, 1979.

26
 SUBROUTINE HYSTR4 (LL,SS,DD,DS,FF,FS)
C
C BILINEAR TAKEDA MODEL
C
C PROGRAMMED BY OTANI, S.
C ON FEBRUARY 9, 1979
C AT UNIVERSITY OF TORONTO
C
C
C THE MODEL WAS SLIGHTLY MODIFIED FROM THE ORIGINAL TAKEDA MODEL.
C THE PRIMARY CURVE IS CHANGED TO A BILINEAR TYPE.
C
C MODIFICATION IS MADE SUCH THAT UNLOADING STIFFNESS IS DEGRADED
C WITH THE MAXIMUM DISPLACEMENT AMPLITUDE IN THE DIRECTION, RATHER
C THAN WITH THE ABSOLUTE MAXIMUM DISPLACEMENT AMPLITUDE.
C
C THIS PROGRAM FINDS A RESISTANCE OF THE TAKEDA HYSTERESIS MODEL AT
C A GIVEN DISPLACEMENT.
C
C
C INPUT DATA
C LL HYSTERESIS RULE POINTER AT PREVIOUS STEP.
C SS STIFFNESS AT PREVIOUS TIME STEP
C DD DISPLACEMENT AT PRESENT STEP
C DS DISPLACEMENT PREVIOUS TIME STEP
C FS FORCE AT PREVIOUS TIME STEP
C FY YIELDING FORCE
C DY YIELDING DISPLACEMENT
C SY DISPLACEMENT STIFFNESS BEFORE YIELDING
C SU DISPLACEMENT STIFFNESS AFTER YIELDING
C B0 STIFFNESS DEGRADATION FACTOR
C B1 STIFFNESS DEGRADATION FACTOR
C
C OUTPUT DATA
C LL HYSTERESIS RULE POINTER AT PRESENT STEP.
C SS STIFFNESS AT PRESENT STEP.
C FF FORCE AT PRESENT TIME STEP
C
C DESCRIPTION OF VARIABLES
C S1(IS) UNLOADING STIFFNESS IN OUTER HYSTERESIS LOOP
C S1(IS)=(DY/DM(IS))**B0*(FY/DY)
C S2(IS) UNLOADING STIFFNESS IN INNER HYSTERESIS LOOPS
C S2(IS)=S1(IS)*B1
C F0 UNLOADING FORCE LEVEL FROM STEP 6
C F1 UNLOADING FORCE LEVEL FROM STEP 8
C F2 UNLOADING FORCE LEVEL FROM STEP 10
C F3 UNLOADING FORCE LEVEL FROM STEP 12
C D0 UNLOADING DISPLACEMENT FROM STEP 6
C D1 UNLOADING DISPLACEMENT FROM STEP 8
C D2 UNLOADING DISPLACEMENT FROM STEP 10
C D3 UNLOADING DISPLACEMENT FROM STEP 12
C X0 INTERCEPT OF DISPLACEMENT AXIS WITH STEP 6
C X1 INTERCEPT OF DISPLACEMENT AXIS WITH STEP 8
C X2 INTERCEPT OF DISPLACEMENT AXIS WITH STEP 10
C X3 INTERCEPT OF DISPLACEMENT AXIS WITH STEP 12
C FM(*) UNLOADING FORCE LEVEL ON PRIMARY CURVE
C DM(*) UNLOADING DISPLACEMENT ON PRIMARY CURVE
C
COMMON /STFF/MD,DC,DY,FC,FY,SC,SY,SU,B0,B1
DIMENSION FM(2),DM(2),S1(2),S2(2)
C
C
IS=2
IF (FS.LT.0.0) IS=1
SN=FLOAT(IS+IS-3)
C
GO TO (1,2,3,4,5,6,7,8,9,10,11), LL
C RULE 1 LINEARLY ELASTIC STAGE BEFORE YIELDING AT (DY,F!)
27
 1 IF (DY-ABS(DD)) 110,110,100
100 FF=SY*DD
GO TO 10000
110 FM(1)=-FY
FM(2)= FY
DM(1)=-DY
DM(2)= DY
S1(1)= SY
S1(2)= SY
GO TO 200
C RULE 2 LOADING ON PRIMARY CURVE AFTER YIELDING AT (DY,FY),
C OR BEYOND PREVIOUS MAXIMUM POINT (DM,FM).
2 IF ((DD-DS)*SN) 220,220,210
200 LL=2
SS=SU
210 FF=(FY+(ABS(DD)-DY)*SU)*SN
GO TO 10000
220 FM(IS)=FS
DM(IS)=DS
S1(IS)=SY*(DY/ABS(DM(IS)))**B0
X0=DM(IS)-FM(IS)/S1(IS)
IF ((X0-DD)*SN) 300,230,230
230 IS=3-IS
SN=FLOAT(IS+IS-3)
240 IF ((DM(IS)-DD)*SN) 200,200,400
C RULE 3 UNLOADING WITH STIFFNESS S1 FROM PREVIOUS MAXIMUM
C POINT (DM,FM) AFTER YIELDING AT (DY,FY).
3 IF ((X0-DD)*SN) 320,230,230
300 LL=3
SS=S1(IS)
310 FF=FM(IS)+(DD-DM(IS))*S1(IS)
GO TO 10000
320 IF ((DM(IS)-DD)*SN) 200,200,310
C RULE 4 LOADING TOWARD PREVIOUS MAXIMUM POINT (DM,FM) ON
C PRIMARY CURVE FROM ZERO CROSSING POINT (X0,0.0).
4 IF ((DD-DS)*SN) 430,430,420
400 LL=4
SS=FM(IS)/(DM(IS)-X0)
410 FF=(DD-X0)*SS
GO TO 10000
420 IF ((DM(IS)-DD)*SN) 200,200,410
430 F0=FS
D0=DS
S2(IS)=S1(IS)*B1
IF (SS.LT.S2(IS)) SS=S2(IS)
X1=D0-F0/SS
IF ((X1-DD)*SN) 500,440,440
440 IS=3-IS
SN=FLOAT(IS+IS-3)
450 IF ((DM(IS)-DD)*SN) 200,200,600
C RULE 5 UNLOADING FROM INNER PEAK POINT (D0,F0) WITH
C STIFFNESS S2.
5 IF ((X1-DD)*SN) 520,440,440
500 LL=5
IF (SS.LT.S2(IS)) SS=S2(IS)
510 FF=F0+(DD-D0)*SS
GO TO 10000
520 IF ((D0-DD)*SN) 240,240,510
C RULE 6 LOADING TOWARD PREVIOUS MAXIMUM POINT (DM,FM) ON
C PRIMARY CURVE FROM ZERO CROSSING POINT (X1,0.0).
6 IF ((DD-DS)*SN) 630,630,620
600 LL=6
SS=FM(IS)/(DM(IS)-X1)
610 FF=(DD-X1)*SS
GO TO 10000
620 IF ((DM(IS)-DD)*SN) 200,200,610
630 F1=FS
D1=DS
28
 S2(IS)=S1(IS)*B1
IF (SS.LT.S2(IS)) SS=S2(IS)
X2=D1-F1/SS
IF ((X2-DD)*SN) 700,640,640
640 IS=3-IS
SN=FLOAT(IS+IS-3)
650 IF ((D0-DD)*SN) 240,240,800
C RULE 7 UNLOADING FROM INNER PEAK POINT (D1,F1) WITH
C STIFFNESS S2.
7 IF ((X2-DD)*SN) 720,640,640
700 LL=7
IF (SS.LT.S2(IS)) SS=S2(IS)
710 FF=F1+(DD-D1)*SS
GO TO 10000
720 IF ((D1-DD)*SN) 450,450,710
C RULE 8 LOADING TOWARD INNER PEAK POINT (D0,F0) FROM ZERO
C CROSSING POINT (X2,0.0).
8 IF ((DD-DS)*SN) 830,830,820
800 LL=8
SS=F0/(D0-X2)
810 FF=(DD-X2)*SS
GO TO 10000
820 IF ((D0-DD)*SN) 240,240,810
830 F2=FS
D2=DS
IF (SS.LT.S2(IS)) SS=S2(IS)
X3=D2-F2/SS
IF ((X3-DD)*SN) 900,840,840
840 IS=3-IS
SN=FLOAT(IS+IS-3)
850 IF ((D1-DD)*SN) 450,450,1000
C RULE 9 UNLOADING FROM INNER PEAK POINT (D2,F2) WITH
C STIFFNESS S2.
9 IF ((X3-DD)*SN) 920,840,840
900 LL=9
IF (SS.LT.S2(IS)) SS=S2(IS)
910 FF=F2+(DD-D2)*SS
GO TO 10000
920 IF ((D2-DD)*SN) 650,650,910
C RULE 10 LOADING TOWARD INNER PEAK POINT (D1,F1) FROM ZERO
C CROSSING POINT (X3,0.0).
10 IF ((DD-DS)*SN) 1030,1030,1020
1000 LL=10
SS=F1/(D1-X3)
1010 FF=(DD-X3)*SS
GO TO 10000
1020 IF ((D1-DD)*SN) 450,450,1010
1030 F3=FS
D3=DS
IF (SS.LT.S2(IS)) SS=S2(IS)
X2=D3-F3/SS
IF ((X2-DD)*SN) 1100,640,640
C RULE 11 UNLOADING FROM INNER PEAK POINT (D3,F3) WITH
C STIFFNESS S2.
11 IF ((X2-DD)*SN) 1120,640,640
1100 LL=11
IF (SS.LT.S2(IS)) SS=S2(IS)
1110 FF=F3+(DD-D3)*SS
GO TO 10000
1120 IF ((D3-DD)*SN) 850,850,1110
10000 RETURN
END

29
 SUBROUTINE HYST5 (LL,SS,DD,DS,FF,FS)
C
C TAKEDA HYSTERESIS MODEL
C
C PROGRAMMED BY OTANI, S.
C ON FEBRUARY 9, 1979
C AT UNIVERSITY OF TORONTO
C
C
C INPUT DATA
C LL HYSTERESIS RULE POINTER AT PREVIOUS TIME STEP
C DS DISPLACEMENT AT PREVIOUS TIME STEP
C DD DISPLACEMENT AT PRESENT TIME STEP
C FS FORCE AT PREVIOUS TIME STEP
C SS STIFFNESS AT PREVIOUS TIME STEP
C FC CRACKING FORCE
C FY YIELDING FORCE
C DC CRACKING DISPLACEMENT
C DY YIELDING DISPLACEMENT
C SC DISPLACEMENT STIFFNESS BEFORE CRACKING
C SY DISPLACEMENT STIFFNESS BEFORE YIELDING
C SU DISPLACEMENT STIFFNESS AFTER YIELDING
C B0 STIFFNESS DEGRADATION FACTOR
C B1 STIFFNESS DEGRADATION FACTOR
C OUTPUT DATA
C LL HYSTERESIS RULE POINTER AT PRESENT TIME STEP
C SS STIFFNESS AT PRESENT TIME STEP
C FF FORCE AT PRESENT TIME STEP
C DESCRIPTION OF VARIABLES
C S1(*) UNLOADING STIFFNESS FROM A PEAK ON PRIMARY CURVE
C S2(*) UNLOADING STIFFNESS ON INNER HYSTERESIS LOOPS
C F0 UNLOADING FORCE LEVEL FROM STEP 6
C F1 UNLOADING FORCE LEVEL FROM STEP 8
C F2 UNLOADING FORCE LEVEL FROM STEP 10
C F3 UNLOADING FORCE LEVEL FROM STEP 12
C D0 UNLOADING DISPLACEMENT FROM STEP 6
C D1 UNLOADING DISPLACEMENT FROM STEP 8
C D2 UNLOADING DISPLACEMENT FROM STEP 10
C D3 UNLOADING DISPLACEMENT FROM STEP 12
C X0 ZERO-CROSSING POINT FROM STEP 4 TO STEP 5
C ZERO-CROSSING POINT FROM STEP 4 TO STEP 6
C X1 ZERO-CROSSING POINT FROM STEP 7 TO STEP 8
C X2 ZERO-CROSSING POINT FROM STEP 9 TO STEP 10
C ZERO-CROSSING POINT FROM STEP 13 TO STEP 10
C X3 ZERO-CROSSING POINT FROM STEP 11 TO STEP 12
C FM(*) UNLOADING FORCE LEVEL ON PRIMARY CURVE
C DM(*) UNLOADING DISPLACEMENT ON PRIMARY CURVE
C
C
COMMON /STFF/MD,DC,DY,FC,FY,SC,SY,SU,B0,B1
DIMENSION FM(2),DM(2),S1(2),S2(2)
C
IS=2
IF (FS.LT.0.0) IS=1
SN=FLOAT(IS+IS-3)
GO TO (1,2,3,4,5,6,7,8,9,10,11,12,13), LL
C RULE 1 ELASTIC STAGE UP TO CRACKING
1 IF (DC-ABS(DD)) 110,110,100
100 FF=SC*DD
GO TO 10000
110 FM(1)=-FC
FM(2)= FC
DM(1)=-DC
DM(2)= DC
S1(1)= SC
S1(2)= SC
120 IF (DY-ABS(DD)) 300,300,200
C RULE 2 LOADING ON THE POST-CRACKING FRIMARY CURVE UP TO
30
C YIELDING.
2 IF ((DD-DS)*SN) 230,230,220
200 LL=2
SS=SY
210 FF=(FC+(ABS(DD)-DC)*SY)*SN
GO TO 10000
220 IF (DY-ABS(DD)) 300,300,210
230 S1(IS)=(ABS(FS)+FC)/(ABS(DS)+DC)
GO TO 330
C RULE 3 LOADING ON THE POST-YIELDING PRIMARY CURVE
3 IF ((DD-DS)*SN) 320,320,310
300 LL=3
SS=SU
310 FF=(FY+(ABS(DD)-DY)*SU)*SN
GO TO 10000
320 S1(IS)=(DY/ABS(DS))**B0*(FC+FY)/(DC+DY)
330 FM(IS)=FS
DM(IS)=DS
X0=DS-FS/S1(IS)
IF ((DD-X0)*SN) 430,430,400
C RULE 4 UNLOADING FROM A PEAK (DM,FM) ON THE PRIMARY CURVE.
4 IF ((DM(IS)-DD)*SN) 120,120,420
400 LL=4
SS=S1(IS)
410 FF=FM(IS)+(DD-DM(IS))*S1(IS)
GO TO 10000
420 IF ((DD-X0)*SN) 430,430,410
430 IS=3-IS
SN=FLOAT(IS+IS-3)
IF (ABS(DM(IS)).GE.DY) GO TO 740
X=FM(IS)/(DM(IS)-X0)
Y=FY*SN/(DY*SN-X0)
IF (X-Y) 440,440,450
440 FM(IS)=FY*SN
DM(IS)=DY*SN
450 IF (FC-ABS(FM(IS))) 740,460,460
460 X=FM(IS)/(DM(IS)-X0)
IF (X-SS) 470,470,740
470 D0=X0+FC/SS*SN
F0=FC*SN
IF ((D0-DD)*SN) 530,530,500
C RULE 5 LOAD REVERSED AT A ZERO-CROSSING POINT (X0,0) WITHOUT
C PREVIOUS CRACKING IN NEW LOADING DIRECTION.
5 IF ((DD-X0)*SN) 540,540,520
500 LL=5
510 FF=(DD-X0)*SS
GO TO 10000
520 IF ((D0-DD)*SN) 530,530,510
530 FM(IS)=FY*SN
DM(IS)=DY*SN
X0=SN*(DY-FY*(DY*SN-D0)/(FY*SN-F0))
GO TO 740
540 IS=3-IS
SN=FLOAT(IS+IS-3)
IF ((DM(IS)-DD)*SN) 120,120,400
C RULE 6 LOAD REVERSED AT ZERO-CROSSING POINT (X0,0), AND
C LOADING TOWARD A PEAK (DM,FM) ON THE PRIMARY CURVE.
6 IF ((DD-DS)*SN) 630,630,620
600 LL=6
SS=FM(IS)/(DM(IS)-X0)
610 FF=(DD-X0)*SS
GO TO 10000
620 IF ((DM(IS)-DD)*SN) 120,120,610
630 F0=FS
D0=DS
S2(IS)=S1(IS)*B1
IF (SS.LT.S2(IS)) SS=S2(IS)
X1=D0-F0/SS
31
 640 IF ((DD-X1)*SN) 730,730,700
C RULE 7 UNLOADING FROM A PEAK (D0,F0) TOWARD A ZERO-CROSSING
C POINT (X1,0).
7 IF ((D0-DD)*SN) 740,740,720
700 LL=7
IF (SS.LT.S2(IS)) SS=S2(IS)
710 FF=F0+(DD-D0)*SS
GO TO 10000
720 IF ((DD-X1)*SN) 730,730,710
730 IS=3-IS
SN=FLOAT(IS+IS-3)
GO TO 940
740 IF ((DM(IS)-DD)*SN) 120,120,600
C RULE 8 LOAD REVERSED AT ZERO CROSSING POINT (X1,0), AND
C LOADING TOWARD A PEAK (DM,FM) ON THE PRIMARY CURVE.
8 IF ((DD-DS)*SN) 830,830,820
800 LL=8
SS=FM(IS)/(DM(IS)-X1)
810 FF=(DD-X1)*SS
GO TO 10000
820 IF ((DM(IS)-DD)*SN) 120,120,810
830 F1=FS
D1=DS
S2(IS)=S1(IS)*B1
IF (SS.LT.S2(IS)) SS=S2(IS)
X2=D1-F1/SS
IF ((DD-X2)*SN) 930,930,900
C RULE 9 UNLOADING FROM A PEAK (D1,F1) TOWARD A ZERO-CROSSING
C POINT (X2,0).
9 IF ((D1-DD)*SN) 940,940,920
900 LL=9
IF (SS.LT.S2(IS)) SS=S2(IS)
910 FF=F1+(DD-D1)*SS
GO TO 10000
920 IF ((DD-X2)*SN) 930,930,910
930 IS=3-IS
SN=FLOAT(IS+IS-3)
GO TO 1140
940 IF ((DM(IS)-DD)*SN) 120,120,800
C RULE 10 LOAD REVERSED AT A ZERO-CROSSING POINT (X2,0), AND
C LOADING TOWARD A PEAK (D0,F0).
10 IF ((DD-DS)*SN) 1030,1030,1020
1000 LL=10
SS=F0/(D0-X2)
1010 FF=(DD-X2)*SS
GO TO 10000
1020 IF ((D0-DD)*SN) 740,740,1010
1030 F2=FS
D2=DS
IF (SS.LT.S2(IS)) SS=S2(IS)
X3=D2-F2/SS
IF ((DD-X3)*SN) 1130,1130,1100
C RULE 11 UNLOADING FROM A PEAK (D2,F2) TOWARD A ZERO-CROSSING
C POINT (X3,0).
11 IF ((D2-DD)*SN) 1140,1140,1120
1100 LL=11
IF (SS.LT.S2(IS)) SS=S2(IS)
1110 FF=F2+(DD-D2)*SS
GO TO 10000
1120 IF ((DD-X3)*SN) 1130,1130,1110
1130 IS=3-IS
SN=FLOAT (IS+IS-3)
GO TO 1330
1140 IF ((D0-DD)*SN) 740,740,1000
C RULE 12 LOAD REVERSED AT A ZERO-CROSSING POINT (X3,0), AND
C LOADING TOWARD A PEAK (D1,F1).
12 IF ((DD-DS)*SN) 1230,1230,1220
1200 LL=12
32
 SS=F1/(D1-X3)
1210 FF=(DD-X3)*SS
GO TO 10000
1220 IF ((D1-DD)*SN) 940,940,1210
1230 F3=FS
D3=DS
IF (SS.LT.S2(IS)) SS=S2(IS)
X2=D3-F3/SS
IF ((DD-X2)*SN) 930,930,1300
C RULE 13 UNLOADING FROM A PEAK (D3,F3) TOWARD A ZERO-CROSSING
C POINT (X2,0).
13 IF ((D3-DD)*SN) 1330,1330,1320
1300 LL=13
IF (SS.LT.S2(IS)) SS=S2(IS)
1310 FF=F3+(DD-D3)*SS
GO TO 10000
1320 IF ((DD-X2)*SN) 930,930,1310
1330 IF ((D1-DD)*SN) 940,940,1200
10000 RETURN
END

33
11.7 Pivot Model

Major features of the force-deflection hysteresis results of large-scale reinforced concrete

members are;
(1) Unloading stiffness decreases as displacement
ductility increases, P4 (α 2 Fy 2 )
(2) Following a nonlinear excursion in one direction, Q4
upon load reversal, the force-deflection path crosses the Y1(Dy1,Fy1)
idealized initial stiffness line prior to reaching the idealized
yield force, and PP4 ( β1* Dy1 , β1* Fy1 ) Q1
(3) The effect of pre-cracked stiffness may be ignored.
The use of the pivot point in defining degraded unloading
stiffness was first proposed by Kunnath et al. (1990).
Q3 PP ( β * D , β * F )
2 2 y2 2 y2
Four quadrants are defined by the horizontal axis and
Y2(Dy2,Fy2)
the elastic loading lines (positive and negative). Primary
Q2
Pivot points P1 through P4 on the elastic loading lines
control the amount of softening in each quadrant. Pinching P2 (α1 Fy1 )
Pivot points PP2 and PP4 fix the degree of pinching
following load reversal in each quadrant.

(1) The response follows the strength envelope so F P4 P3

long as no displacement reversal occurs. The initial α 2 Fy 2
elastic stiffness, yield resistance and maximum
resistance can be different in each direction.
Ft1
(2) Once the yield deformation has been exceeded Fy1 Y1 S1(D1max,F1max)
in either direction, a subsequent strength envelope
PP4
is developed requiring the introduction of points S1
Q4 Q1
and S2 which move along the strength envelope and
defined by the previous maximum displacements.
The initial points of S1 and S2 are yield point Y1 and df2 dd2 dt2 dy2 D
Y2 in each direction. dt1 dd1 df1
dy
PP2
(3) The modified strength envelope (acting as the Q3
upper bound for future cyclic loading) is defined by Fy2
lines joining the pinching pivot point PP4 (PP2) to S2(D2max,F2max) Y2 Q2
Ft2
maximum response point S1 (S2) until the response
point reaches the strength envelope.
P2 P1 α1 Fy1
(4) The pinching pivot points PP4 and PP2 are
initially fixed, but they move toward the
force-deflection origin with the strength degradation. The F P4
resistance at a pinching pivot point is given by β i Fyi
* P4* Q1

where
PP4 S1
β i* = β i (d iMAX ≤ d ti )
PP4*
F Q4 D
β = β i iMAX
i
*
(d iMAX > d ti )
Fti
d1max
where β i defines the degree of pinching for a ductile
flexural response prior to strength degradation. Q3
d iMAX , d ti : maximum displacement and strength α1 Fy1 Q2
degradation displacement (displacement at the highest P1
P1* (1 + η )α1 Fy1
resistance) in the i-th direction of loading (i=1 or 2).

(5) The unlading stiffness of the maximum displacement

34
excursion in Quadrant Q1 is guided toward point P1 until point P1* is reached at force ( 1 + η ) times
larger than the force at point P1. A line extending from point P1* through origin defines the new
softened elastic loading line K*. Point P4* is on the new elastic loading line at the same force level
as point P4. Point PP4* is also on the new elastic loading line but at a force defined by the
intersection of the modified strength envelope (line between points PP4 and S1) and K*.

Hysteresis Rules:
(1) Loading and unloading in Quadrants Qn is directed away from or toward point Pn, respectively.
(2) Loading in Quadrant Qn is directed toward point PPn, then to maximum response point Si, followed
by the strength envelope.
(3) Unloading in Quadrant Qn is directed away from point Pn.

F P4 P3

Q1
Y1
S1
PP4

Q4 D

Q3 PP2
Q2
Y2

S2
P2 P1

Reference:

Dowell, R. K., F. Seible and E. L. Wilson, “Pivot

Hysteresis Model for Reinforced Concrete
Members,” ACI Structural Journal, Title No.
95-S55, Vol. 95, No. 5, September-October
1998, pp. 607 - 617.
Kunnath, S. K., A. M. Reinhorn, and Y. J. Park,
“Analytical Modeling of Inelastic Seismic
Response of RC Structures,” Journal,
Structural Engineering Division, ASCE, Vol.
116, No. 4, April 1990, pp. 996 - 1017,”

35
11.8 Stable Hysteresis Models with Pinching

The force-deformation relation of a reinforced concrete member is highly dependent on a loading

history, characterized by strength decay with load reversals and pinching behavior at a low stress
level during reloading, when the behavior of the member is dominated by sliding along inclined shear
cracks or slippage of longitudinal reinforcement.

A flexure-dominated reinforced concrete

girder sometimes exhibits a pinching
characteristic when the amount of
longitudinal reinforcement (or bending
resistance) is significantly different at the top
and bottom of section. This is attributable to
the fact that a wide crack in weak side
cannot close due to large residual strain in
tensile reinforcement after load reversal; the
compressive stress must be resisted by the
longitudinal reinforcement before concrete
faces make contact at cracks.

Many hysteresis models have been Hysteresis Relation of Beams with

developed on the basis of test results of a Unbalanced Amount of Reinforcement
particular set of specimens under a specific
loading history. However, the parameters of most models may not be analytically defined by the
member properties (material properties and member geometry).

Takeda-slip Model: Eto and Takeda (1973) modified the Takeda model to incorporate a slip-type
behavior at low stress level due to pull-out of longitudinal reinforcement from the anchorage zone.

The skeleton curve is tri-linear with stiffness changes at cracking and yielding where the cracking
and yielding levels can be different in positive and negative directions. The performance of the
model is identical to the Takeda model before yielding.

Pinching takes place only when the

yielding has occurred in the direction of (Dm,Fm)
reloading. The reloading (pinching)
F Y
stiffness Ks is defined as
−γ
Fm Dm
Ks = C
Dm − Do Dy
where Do : displacement at the end of
Ks Kd
unloading (resistance equal to zero), Do D 'o
Dm and Fm : maximum deformation and
resistance in the direction of reloading, Ks’ D
D y : yield deformation in the direction of
C
reloading, γ : slip stiffness degradation
index (slip stiffness degradation index γ
is suggested to be 0.5). The pinching Y
(Dm’,Fm’)
stiffness is revised only when the
maximum response point is exceeded in
the direction of reloading. Takeda-slip model

When the response point crosses a line connecting the origin and the maximum response point in
the direction of reloading, the response point moved toward the previous maximum response point
and then on the skeleton curve. The unloading stiffness is defined in the same manner as the
36
Takeda model.

The same pinching and unloading stiffness is used during reloading and unloading in an inner
loop.
−α
F 'c + Fy Dm
Kd =
D 'c + Dy Dy
where, F 'c and D 'c : resistance and deformation at cracking on the opposite side, Fy and Dy :
resistance and deformation at yielding on the unloading side, Dm : maximum deformation on the
unloading side, α : unloading degradation index.

Kabeyasawa-Shiohara Model: Kabeyasawa et al. (1983) modified the Takeda-Eto slip model to
represent the behavior of a girder with the amount of longitudinal reinforcement significantly different
at the top and bottom;

37
 (1) the pinching occurs only in one direction where the yield resistance is higher than the other
direction,
(2) the pinching occurs only after the initial yielding in the direction of reloading, and
(3) the stiffness Ks during slipping is a function of the maximum response point (Dm, Fm) and the
point of load reversal (Do, Fo=0.0) in the resistance-deformation plane.

The reloading (slip) stiffness Ks, after unloading in the direction of the smaller yield resistance,
was determined as
γ
Fm Dm
Ks =
Dm − Do Dm − Do
where ( Dm , Fm ): deformation and resistance at the previous maximum response point, Do :
displacement at the end of unloading on the zero-load axis, γ : slip stiffness degradation index. No
slip behavior will be generated for γ = 0; the degree of slip behavior increases with γ > 1.0. γ =
1.2 was suggested.

The slip stiffness is used until the response point crosses a line with slope Kp through the
previous maximum response point (Dm, Fm); the stiffness is reduced from the slope connecting the
origin and the maximum response point by reloading stiffness index η ,
Fm
K p = η( )
Dm

The values of unloading stiffness degradation index α of Takeda model, slipping stiffness
degradation index γ , and reloading stiffness index η were chosen to be 0.4, 1.0 and 1.0,
respectively by Kabeyasawa et al. (1983).

Costa and Costa model: Costa and Costa (1987) proposed a Fy

trilinear model for the force-displacement response of a
single-degree-of-freedom oscillator, including pinching and
strength degradation. Fc

Unloading-reloading loops prior to yielding in either direction D’y D’c

are bilinear, with slopes equal to those of the pre-cracking and
Dc Dy
post-cracking branches in the virgin loading. After the initial
yielding, the reloading stiffness K s is reduced from the stiffness
γ
toward the previous extreme point by factor ( Dy / Dm ) ; i.e., F’c

Fm Dy γ
Ks = ( ) F’y
Dm − Do Dm
where, Fm and Dm : resistance and deformation at the F
Y
previous maximum response point, and Do : deformation at
load reversal point. Once the response point crosses the line C
connecting the origin and the maximum response point, then
response point moves toward the maximum response point.
Ks
The unloading stiffness after yielding is reduced from the
α O D
elastic stiffness by factor ( Dy / Dr ) .

Post-yield strength and stiffness degradation with cycling is

modeled by directing the reloading branch, after modification for
pinching, toward a point at a displacement equal to (1 + λ ) Dm

38
and at a moment (1 − λ ) Fm , where Fm is the resistance at the extreme point if the previous
excursion. After reaching this terminal point of the reloading branch, further loading takes place
parallel to the post-yielding stiffness of the virgin loading curve.

Fy

Fc

D’y Dy
D’c Dc

F’c

F’y

References:

Costa, A. C., and A. G. Costa, “Hysteretic Model of Force-Displacement Relationships for Seismic
Analysis of Structures,” National Laboratory for Civil Engineering, Lisbon, 1987.
Eto, H, and T. Takeda, "Elasto Plastic Earthquake Response Analysis of Reinforced Concrete
Frame Structure (in Japanese)," Report, Annual Meeting, Architectural Institute of Japan, 1973,
pp. 1261-1262.
Kabeyasawa, T., H. Shiohara, S. Otani and H. Aoyama, "Analysis of the Full-scale Seven-story
Reinforced Concrete Test Structure," Journal of the Faculty of Engineering, the University of
Tokyo, (B), Vol. XXXVII, No. 2, 1983, pp. 431-478.

39
 SUBROUTINE TAKEDSLP (FC,RC,FY,RY,SU,B0,B1,B2,
+ FM,RM,F0,R0,F1,R1,X0,X1,XM,S1,
+ FF,RR,DR,SS,IL)
C
C TAKEDA-SLIP HYSTERESIS MODEL
C THE PRIMARY CURVE IS TRILINEAR WITH STIFFNESS CHANGES AT
C CRACKING AND YIELDING. THE PINCHING BEHAVIOUR DUE TO THE
C DIFFERENT AMOUNT OF TOP AND BOTTOM LONGITUDINAL REINFORCEMENT
C WAS MODELED BY MODIFYING THE TAKEDA MODEL.
C
C REFERENCE:
C OTANI, S., T. KABEYASAWA, H. SHIOHARA, AND H. AOYAMA,
C "ANALYSIS OF THE FULL-SCALE SEVEN-STORY REINFORCED CONCRETE
C TEST STRUCTURE", EARTHQUAKE EFFECTS ON REINFORCED CONCRETE
C STRUCTURES, U.S.-JAPAN RESEARCH, PUBLICATION SP-84, ACI, 1985,
C PP. 203-239.
C
C PROGRAMMED BY SHIOHARA, H.
C ON DECEMBER 15, 1982
C AT THE UNIVERSITY OF TOKYO
C MODIFIED BY OTANI, S.
C ON APRIL 21, 1994
C AT UNIVERSITY OF CANTERBURY
C
C THE ORIGINAL PROGRAM BY SHIOHARA WAS WRITTEN TO DEFINE THE FORCE
C AND DEFORMATION FOR A GIVEN FORCE INCREMENT.
C THE PROGRAM WAS MODIFIED BY OTANI TO DEFINE THE HYSTERESIS RELATION
C FOR A GIVEN DISPLACEMENT INCREMENT. MANY COMMENTS WERE ADDED TO
C DESCRIBE THE HYSTERESIS RULES IN DETAILS.
C
C TAKEDA HYSTERESIS MODEL WAS MODIFIED AS FOLLOWS :
C (1) THE YIELD RESISTANCES IN POSITIVE AND NEGATIVE DIRECTIONS
C ARE MADE DIFFERENT ALTHOUGH CRACKING FORCE AND DEFORMATION
C ARE KEPT THE SAME. POST-YIELDING STIFFNESS IS ALSO THE SAME
C IN POSITIVE AND NEGATIVE DIRECTIONS.
C (2) THE PINCHING TAKES PLACE ONLY IN THE DIRECTION OF THE
C HIGHER YIELD RESISTANCE, AND ONLY AFTER THE INITIAL YIELDING.
C (3) THE HYSTERESIS RULES WERE SIMPLIFIED FROM THE ORIGINAL
C TAKEDA MODEL.
C (4) STIFFNESS DURING UNLOADING IS MADE EQUAL FOR UNLOADING
C FROM MAXIMUM RESPONSE POINT AND UNLADING FROM INNER PEAK.
C
C INPUT DATA:
C IL HYSTERESIS RULE POINTER AT PREVIOUS STEP.
C INITIAL VALUE MUST BE 1.
C RR DISPLACEMENT AT PREVIOUS STEP.
C INITIAL VALUE MUST BE LESS THAN RC.
C FF FORCE AT PREVIOUS STEP.
C INITIAL VALUE MUST BE LESS THAN FC.
C SS STIFFNESS AT PREVIOUS STEP.
C INITIAL VALUE MUST BE SC.
C DR DISPLACEMENT INCREMENT.
C PROPERTIES OF HYSTERESIS SYSTEM (INPUT DATA):
C FC CRACKING FORCE.
C RC CRACKING DISPLACEMENT.
C FY(*) YIELDING FORCE (POSITIVE AND NEGATIVE).
C RY(*) YIELDING DISPLACEMENT (POSITIVE AND NEGATIVE).
C SU STIFFNESS AFTER YIELDING.
C B0 UNLOADING STIFFNESS DEGRADATION INDEX.
C STIFFNESS S1(*) DURING UNLOADING IS DEGRADED AFTER
C YIELDING BY FACTOR (RY(IS)/RM(IS))**B(1) FROM THE
C STIFFNESS CONNECTING PEAK POINT (DM(*),FM(*)) AND
C CRACKING POINT IN THE OTHER DIRECTION.
C B0=0.0 TO 0.5 IS COMMONLY USED FOR REINFORCED CONCRETE.
C B1 SLIP STIFFNESS DEGRADATION PARAMETER.
C SLIP STIFFNESS ET IS DEFINED BY
C ET=ABS(FM(IS)/(X-RM(IS)))*ABS(RM(IS)/(X-RM(IS)))**B1
C B2 RELOADING STIFFNESS PARAMETER.
40
C STIFFNESS EU AFTER SLIPPING IS DEFINED BY
C EU=B2*FM(IS)/RM(IS)
C
C OUTPUT DATA:
C IL HYSTERESIS RULE POINTER AT PRESENT STEP
C RR DISPLACEMENT AT PRESENT STEP
C FF FORCE AT PRESENT STEP
C SS STIFFNESS AT PRESENT STEP
C
C DESCRIPTION OF VARIABLES
C RZ DISPLACEMENT AT PREVIOUS STEPS (TEMPORARY)
C FZ FORCE AT PREVIOUS STEP (TEMPORARY)
C FM(*) FORCE AT UNLOADING POINT FROM PRIMARY CURVE
C RM(*) DISPLACEMENT AT UNLOADING POINT FROM PRIMARY CURVE
C F0 FORCE AT UNLOADING POINT TO RULE 5
C R0 DISPLACEMENT AT UNLOADING POINT TO RULE 5
C F1 FORCE AT UNLOADING POINT TO RULE 8
C R1 DISPLACEMENT AT UNLOADING POINT TO RULE 8
C XM DISPLACEMENT AT STIFFNESS HARDENING POINT AFTER SLIPPING
C TOWARD PREVIOUS MAXIMUM RESPONSE POINT (RM(*),FM(*)),
C (FROM RULE 6 TO RULE 7).
C X0 DISPLACEMENT AT ZERO-CROSSING POINT AFTER UNLOADING FROM
C PEAK (RM(*),FM(*)) ON PRIMARY CURVE.
C X1 DISPLACEMENT AT ZERO-CROSSING POINT AFTER UNLOADING FROM
C PEAK.
C S1(*) STIFFNESS DURING UNLOADING; THE STIFFNESS IS COMMON FOR
C UNLOADING FROM PEAK (RM(*),FM(*)) ON PRIMARY AND FROM
C PEAK OF INNER LOOP.
C EU STIFFNESS AFTER SLIPPING TOWARD PREVIOUS MAXIMUM POINT.
C EU=B2*FM(IS)/RM(IS)
C ET STIFFNESS DURING SLIPPING AFTER RELOADING IN THE DIRECTION
C OF HIGHER YIELD RESISTANCE.
C ET=ABS(FM(IS)/(X-RM(IS)))*ABS(RM(IS)/(X-RM(IS)))**B1
C IS SIGN POINTER FOR FORCE AT PREVIOUS STEP.
C (=1 FOR NEGATIVE FORCE, AND =2 FOR POISTIVE FORCE).
C SN SIGN OF FORCE AT PREVIOUS STEP (=-1.0 OR 1.0).
C ERR SMALL VALUE TO CHECK THE DIFFERENCE IN STIFFNESS.
C
PARAMETER (ERR=0.0001)
C
DIMENSION RY(2),FY(2),S1(2),FM(2),RM(2)
C
C DR INCREMENTAL DISPLACEMENT
C RR DISPLACEMENT AT PREVIIOUS STEP, REPLACED BY
C DISPLACEMENT AT PRESENT STEP
C SS STIFFNESS AT PREVIOUS STEP
C FF FORCE AT PREVIOUS STEP, REPLACED BY FORCE AT PRESENT
C STEP
C RZ DISPLACEMENT AT PREVIOUS STEP (TEMPORARY USE)
C FZ FORCE AT PREVIOUS STEP (TEMPORARY USE)
C
RZ=RR
FZ=FF
RR=RR+DR
FF=FF+SS*DR
C IS: SIGN POINTER FOR FORCE FZ AT PREVIOUS STEP
C (=1 FOR POISTIVE FORCE, AND =2 FOR NEGATIVE FORCE)
C SN: SIGN OF FORCE FZ AT PREVIOUS STEP (=-1.0 OR 1.0)
IS=1
IF (FZ.LT.0.0) IS=2
SN=FLOAT(3-IS-IS)
C IL: HYSTERESIS RULE POINTER AT PREVIOUS TIME STEP
GO TO (1, 2, 3, 4, 5, 6, 7, 8, 9), IL
C
C RULE 1: ELASTIC STAGE UP TO CRACKING
C
C TEST IF DISPLACEMENT RR EXCEEDS CRACKING DISPLACEMENT RC.
C
41
 1 IF (RC-ABS(RR)) 110, 110, 2000
C THE FOLLOWING PARAMETERS ARE SET
C FM(*) FORCE AT PREVIOUS MAXIMUM RESPONSE ON PRIMARY CURVE
C RM(*) DISPLACEMENT AT PREVIOUS MAXIMUM RESPONSE ON PRIMARY
C CURVE
C S1(*) STIFFNESS DURING UNLOADING FROM (RM,FM) ON PRIMARY
C CURVE
C WHERE *=1 FOR POSITIVE AND *=2 FOR NEGATIVE DIRECTION.
110 FM(1)= FC
FM(2)=-FC
RM(1)= RC
RM(2)=-RC
S1(1)= FC/RC
S1(2)= FC/RC
C CHECK IF DISPLACEMENT RR EXCEEDS YIELDING DISPLACEMENT AFTER CRACKING
120 IF (ABS(RY(IS))-ABS(RR)) 300, 300, 200
C
C RULE 2: LOADING ON POST-CRACKING PRIMARY CURVE TO YIELD POINT
C
C TEST IF LOADING CONTINUES OR UNLOADING TAKES PLACE
C DR*SN>0: LOADING CONTINUES
C DR*SN<0: LOADING DIRECTION CHANGES, AND UNLOADING TAKES PLACE
C
2 IF (DR*SN) 220, 220, 210
C RESPONSE POINT MOVES ON PRIMARY CURVE AFTER CRACKING TOWARD YIELD POINT
C STIFFNESS SS AND FORCE FF ARE REVISED WITH STIFFNESS CHANGE
200 IL=2
SS=(FY(IS)-SN*FC)/(RY(IS)-SN*RC)
FF=SN*FC+(RR-RC*SN)*SS
GO TO 2000
C TEST IF YIELDING TAKES PLACE DURING LOADING ON POST-CRACKING PRIMARY CURVE
210 IF ((RY(IS)-RR)*SN) 300, 300, 2000
C UNLOADING TAKES PLACE FROM PEAK (RM(*),FM(*)) ON PRIMARY CURVE
C BETWEEN CRACKING AND YIELDING POINTS TOWARD CRACKING POINT
C IN THE OTHER DIRECTION.
C FORCE (FZ) AND DISPLACEMENT (RZ) AT PREVIOUS STEP ARE ASSIGNED
C TO FORCE FM(*) AND DISPLACEMENT RM(*) AT MAXIMUM RESPONSE POINT.
C STIFFNESS S1(*) DURING UNLOADING IS DEFINED AS SLOPE CONNECTING
C PREVIOUS MAXIMUM RESPONSE POINT (FM(*),RM(*)) AND CRACKING POINT
C IN THE OTHER DIRECTION.
220 RM(IS)=RZ
FM(IS)=FZ
S1(IS)=(FM(IS)+SN*FC)/(RM(IS)+SN*RC)
GO TO 320
C
C RULE 3: LOADING ON THE POST-YIELDING PRIMARY CURVE
C
C TEST IF LOADING CONTINUES OR UNLOADING TAKES PLACE
C DR*SN>0: LOADING CONTINUES
C DR*SN<0: LOADING DIRECTION CHANGES, AND UNLOADING TAKES PLACE
C
3 IF (DR*SN) 310, 310, 2000
C RESPONSE POINT MOVES ON POST-YIEDING PRIMARY CURVE.
C FORCE FF AND STIFFNESS SS ARE REVISED FOR STIFFNESS CHANGE.
300 IL=3
FF=FY(IS)+(RR-RY(IS))*SU
SS=SU
GO TO 2000
C UNLOADING TAKES PLACE FROM PEAK (RM(*),FM(*)) ON PRIMARY
C CURVE AFTER YIELDING USING UNLOADING STIFFNESS S1(*).
C FORCE (FZ) AND DISPLACEMENT (RZ) AT PREVIOUS STEP ARE ASSIGNED
C TO FORCE FM(*) AND DISPLACEMENT RM(*) AT MAXIMUM RESPONSE POINT.
C STIFFNESS S1(*) DURING UNLOADING IS DEGRADED BY FACTOR
C (RY(IS)/RM(IS))**B0 FROM THE STIFFNESS CONNECTING THE PREVIOUS
C MAXIMUM RESPONSE POINT (DM(*),FM(*)) AND THE CRACKING POINT IN
C THE OTHER DIRECTION.
310 RM(IS)=RZ
FM(IS)=FZ
42
 S1(IS)=(SN*FC+FY(IS))/(SN*RC+RY(IS))*(RY(IS)/RM(IS))**B0
C X0: DISPLACEMENT AT ZERO CROSSING POINT DURING UNLOADING
C FROM PEAK (RM(*),FM(*)) ON PRIMARY CURVE.
320 X0=RM(IS)-FM(IS)/S1(IS)
C TEST IF THE LOADING DIRECTION CHANGES AFTER UNLOADING.
C (RR-X0)*SN>0 UNLOADING CONTINUES.
C (RR-X0)*SN<0 UNLOADING FINISHES, AND RELOADING TAKES PLACE IN
C THE OPPOSITE DIRECTION.
IF ((RR-X0)*SN) 420, 400, 400
C
C RULE 4: UNLOADING FROM PEAK (FM,RM) ON THE PRIMARY CURVE
C
C TEST IF PREVIOUS RESPPONSE DISPLACEMENT IS EXCEEDED DURING
C REOADING AFTER UNLOADING FROM PEAK (RM(*),FM(*)).
C
4 IF ((RM(IS)-RR)*SN) 120, 120, 410
C UNLOADING FROM POINT (RM,FM) ON PRIMARY CURVE.
C FORCE AND STIFFNESS ARE REVISED WITH STIFFNESS CHANGE.
400 IL=4
FF=FM(IS)+(RR-RM(IS))*S1(IS)
SS=S1(IS)
GO TO 2000
C TEST IF UNLOADING IS COMPLETED, AND RELOADING TAKES PLACE IN
C THE OTHER DIRECTION
C (RR-X0)*SN>0 UNLOADING CONTINUES.
C (RR-X0)*SN<0 UNLOADING IS COMPLETED AND RELOADING TAKES PLACE IN
C THE OPPOSITE DIRECTION.
410 IF ((RR-X0)*SN) 420, 420, 2000
C UNLOADING IS COMPLETED AND RELOADING TAKES PLACE IN THE OPPOSITE
C DIRECTION PASSING ZERO CROSSING POINT (X0,0).
C SIGN SN OF FORCE FF IS CHANGED FROM FORCE FZ OF PREVIOUS STEP.
420 IS=3-IS
SN=FLOAT(3-IS-IS)
C IF CRACKING HAS NOT OCCURRED IN THE RELOADING DIRECTION, RELAODING
C ASSUMES THE SAME STIFFNESS AS THE UNLOADING STIFFNESS UNTIL FORCE
C EXCEEDS THE CRACKING FORCE AT POINT (R0,F0).
IF (ABS(RM(IS)).LE.RC) THEN R0=X0+SN*FC/S1(3-IS)
F0=FC*SN
IF ((R0-RR)*SN) 520, 520, 500
C IF YIELDING HAS NOT TAKEN PLACE IN THE RELOADING DIRECTION, AND IF
C STIFFNESS TOWARD PREVIOUS MAXIMUM RESPONSE POINT (RM(*),FM(*))
C IS SMALLER THAN STIFFNESS TOWARD YIELDING POINT, PREVIOUS MAXIMUM
C RESPONSE POINT (RM(*),FM(*)) IS REPLACED BY THE YIELD POINT.
ELSE IF (ABS(RM(IS)).LE.ABS(RY(IS))) THEN
X=FM(IS)/(RM(IS)-X0)
Y=FY(IS)/(RY(IS)-X0)
IF (X.LT.Y) THEN RM(IS)=RY(IS)
FM(IS)=FY(IS)
END IF
GO TO 530
C TEST IF YIELD FORCE IN RELOADING DIRECTION IS LARGER THAN YIELD
C FORCE IN THE OPPOSITE DIRECTION
ELSE IF (ABS(FY(IS)).GT.ABS(FY(3-IS))) THEN GO TO 430
ELSE GO TO 530
END IF
C AFTER COMPLETION OF UNLOADING FROM MAXIMUM RESPONSE POINT
C (RM(*),FM(*)) AT ZERO CROSSING POINT (X,0), SLIP TAKES
C PLACE DURING RELOADING IN OPPOSITE DIRECTION AT SLIP STIFFNESS ET.
430 X=RM(3-IS)-FM(3-IS)/S1(3-IS)
ET=FM(IS)/(RM(IS)-SN*X)*(RM(IS)/(RM(IS)-SN*X))**B1
EU=FM(IS)/RM(IS)*B2
C DISPLACEMENT X AT START OF STRAIN HARDENING AFTER SLIP IS
C SOLVED AS THE INTERSECTION OF LINE PASSING ZERO CROSSING POINT
C (X0,0) WITH SLOPE ET AND LINE PASSING PREVIOUS MAXIMUM POINT
C (RM(*),FM(*)) WITH SLOP EU.
IF (ABS(EU-ET).LT.ERR) THEN XM=X0
ELSE XM=(EU*RM(IS)-ET*X0-FM(IS))/(EU-ET)
END IF
43
C TEST IF RESPONSE PASSES STRAINHARDENING POINT
IF ((RR-XM)*SN.GT.0.0) THEN
IF ((RR-RM(IS))*SN) 700, 120, 120
ELSE GO TO 600
END IF
C
C RULE 5: LOAD REVERSED AT ZERO CROSSING PONT (X0,0) WITHOUT
C PREVIOUS CRACKING IN RELOADING DIRECTION.
C
5 IF ((RR-X0)*SN) 540, 540, 510
C RELAODING TOWARD POINT (R0,F0) AFTER UNLODING FROM PEAK
C (RM(*),FM(*)) ON PRIMARY CURVE USING THE UNLOADING STIFFNESS
C S1(*). NO CRACKING HAS TAKEN PLACE IN THE RELOADING DIRECTION.
500 IL=5
SS=S1(3-IS)
FF=(RR-X0)*SS
GO TO 2000
C RELOADING TOWARD POINT (R0,F0).
C TEST IF RESPONSE POINT EXCEEDS POINT (R0,F0).
510 IF ((R0-RR)*SN) 520, 520, 2000
C RELOADING BEYOND POINT (R0,F0) AND RELOADING TOWARD
C PREVIOUS MAXIMUM RESPONSE POINT (RM(*),FM(*)).
520 IF (ABS(RM(IS)).LE.RC
+ .AND.ABS(RM(3-IS)).LE.ABS(RY(3-IS))) GO TO 120
C IF YIELDED IN THE OPPOSITE DIRECTION, MAXIMUM RESPONSE POINT
C (RM(*),FM(*)) IS REPLACED BY THE YIELD POINT ALTHOUGH CRACKING
C HAS NOT TAKEN PLACE IN RELOADING DIRECTION.
FM(IS)=FY(IS)
RM(IS)=RY(IS)
X0=RY(IS)-FY(IS)*(RY(IS)-R0)/(FY(IS)-F0)
530 IF ((RR-RM(IS))*SN) 900, 120, 120
C DURING RELOADING, UNLOADING TAKES PLACE INTO OPPOSITE
C DIRECTION PASSING ZERO CROSSING POINT (X0,0), AND SIGN
C OF FORCE SN HAS BEEN CHANGED.
540 IS=3-IS
SN=FLOAT(3-IS-IS)
C TEST IF RESPONSE POINT PASSES PREVIOUS MAXIMUM RESPONSE POINT
C (RM(*),FM(*)) DURING RELOADING TOWARD PREVIOUS MAXIMUM RESPONSE
C POINT (RM(*),FM(*))
IF ((RM(IS)-RR)*SN) 120, 120, 400
C
C RULE 6: RELOADING WITH SLIP STIFFNESS ET TOWARD STRAIN HARDENING
C POINT AT DISPLACEMENT XM
C
C TEST IF RELOADING CONTINUES
C
6 IF (DR*SN) 630, 630, 610
C STIFFNESS SS AND FORCE FF ARE REVISED FOR STIFFNESS CHANGE
600 IL=6
X=RM(3-IS)-FM(3-IS)/S1(3-IS)
ET=FM(IS)/(RM(IS)-SN*X)*(RM(IS)/(RM(IS)-SN*X))**B1
SS=ET
FF=ET*(RR-X0)
GO TO 2000
C RELOADING CONTINUES WITH SLIP STIFFNESS ET
C TEST IF RESPONSE POINT PASSES STRAIN HARDENING POINT AT
C DISPLACEMENT AT XM.
610 IF ((XM-RR)*SN) 620, 620, 2000
C RESPONSE POINT PASSES THE STRAIN HARDENING POINT AT
C DISPLACEMENT XM. TEST IF RESPONSE PINT PASSES PREVIOUS
C MAXIMUM RESPONSE POINT (RM(*),FM(*)).
620 IF ((RR-RM(IS))*SN) 700, 700, 300
C UNLOADING TAKES PLACE FROM THE PREVIOUS RESPONSE POINT
C AT PEAK (R1,F1) OF INNER LOOP.
630 R1=RZ
F1=FZ
C X1: DISPLACEMENT AT ZERO CROSSING POINT DURING UNLOADING
C FROM PEAK (R1,F1) OF INNER LOOP WITH UNLOADING STIFFNESS S1(*).
44
 X1=R1-F1/S1(IS)
GO TO 8
C
C RULE 7: LOADING WITH STRAIN HARDENING AFTER SLIPPING TOWARD
C PREVIOUS PEAK (RM(*),FM(*))
C
C TEST IF LOADING CONTINUES.
C
7 IF (DR*SN) 630, 630, 710
700 IL=7
SS=FM(IS)/RM(IS)*B2
FF=FM(IS)-SS*(RM(IS)-RR)
GO TO 2000
C LOADING CONTINUES TOWARD PREVIOUS MAXIMUM RESPONSE POINT
C (RM(*),FM(*)).
C TEST IF RESPONSE POINT EXCEEDS THE PREVIOUS MAXIMUM POINT
C (RM(*),FM(*)).
710 IF ((RR-RM(IS))*SN) 2000, 300, 300
C
C RULE 8: UNLOADING FROM INNER PEAK POINT (R1,F1).
C
C TEST IF UNLOADING IS COMPLETED AND RELOADING STARTS IN THE
C OPPOSITE DIRECTION.
C
8 IF ((X1-RR)*SN) 810, 840, 840
C UNLOADING FROM INNER PEAK (R1,F1) WITH STIFFNESS CHANGE.
800 IL=8
SS=S1(IS)
FF=F1-(R1-RR)*SS
GO TO 2000
C UNLOADING OR RELOADING IN THE SAME DIRECTION CONTINUES.
C TEST IF RELOADING RESPONSE POINT EXCEEDS THE INNER PEAK
C POINT (R1,F1).
810 IF ((R1-RR)*SN) 820, 820, 2000
C RESPONSE POINT DURING RELOADING EXCEEDS THE INNER PEAK
C POINT (R1,F1), AND RESPONSE POINT MOVES TOWARD PREVIOUS
C MAXIMUM POINT (RM(*),FM(*)).
820 IF (ABS(RM(IS)).GT.ABS(RY(IS))
: .AND.ABS(FY(IS)).GT.ABS(FY(3-IS))) THEN
C YIELDING HAS TAKEN PLACE IN BOTH DIRECTIONS.
IF ((RR-XM)*SN.LE.0.0) GO TO 600
C TEST IF THE PREVIOUS MAXIMUM POINT (RM(*),FM(*)) IS EXCEEDED.
830 IF ((RR-RM(IS))*SN) 900, 120, 120
C YIELDING HAS NOT TAKEN PLACE IN ONE DIRECTION.
ELSE GO TO 830
840 X0=X1
IS=3-IS
SN=FLOAT(3-IS-IS)
IF (ABS(RM(IS)).GT.ABS(RY(IS))
+ .AND.ABS(FY(IS)).GT.ABS(FY(3-IS))) THEN
EU=FM(IS)/RM(IS)*B2
XF=RM(IS)-FM(IS)/EU
XM=X0
IF ((XF-X0)*SN.LE.0.0) THEN
IF ((RR-RM(IS))*SN) 900, 120, 120
ELSE
GO TO 430
END IF
ELSE
IF ((RR-RM(IS))*SN) 900, 120, 120
END IF
C
C RULE 9: RELOADING TOWARD PEAK (RM(*),FM(*)) ON PRIMARY CURVE
C WITHOUT PINCHING
C
C TEST IF LOADING CONTINUES.
C
9 IF (DR*SN) 920, 920, 910
45
C RELOADING TOWARD PREVIOUS MAXIMUM POINT (RM(*),FM(*)) FROM
C ZERO CROSSING POINT (X0,0).
900 IL=9
SS=FM(IS)/(RM(IS)-X0)
FF=SS*(RR-X0)
GO TO 2000
C TEST IF RESPONSE POINT EXCEEDS PREVIOUS MAXIMUM RESPONSE
C POINT (RM(*),FM(*)) ON PRIMARY CURVE DURING RELOADING.
910 IF ((RR-RM(IS))*SN) 2000, 2000, 120
C UNLOADING TAKES PLACE. PEAK (R1,F1) IS DEFINED AS RESPONSE
C POINT AT PREVIOUS STEP.
920 R1=RZ
F1=FZ
C X1 ZERO CROSSIG POINT DURING UNLOADING FROM INNER PEAK
C (R1,F1) USING COMMON UNLOADING STIFFNESS.
X1=R1-F1/S1(IS)
C TEST IF UNLOADING IS COMPLETED.
IF ((X1-RR)*SN) 800, 840, 840
C RETURN TO CALLING PROGRAM
2000 RETURN
END

46
11.9 Shear-type Hysteresis Models

Reinforced concrete members exhibit progressive loss of strength under reversed cycles of
inelastic deformation due to lack of shear capacity of member or bond resistance along longitudinal
reinforcement; the monotonic strength of such members cannot be attained.

The response of a reinforced concrete member, exhibiting early strength decay, is difficult to
model because such behavior is sensitive to loading history. General features can be summarized
as the decay in resistance with cyclic loading and pinching response during reloading followed by
hardening.

The undesirable features can be avoided or reduced by following design requirements and
detailing of reinforcement. Therefore, hysteresis models for shear-failing performance may not be
necessary for the response analysis of new construction, but may be necessary for the seismic
evaluation of existing construction.

Takayanagi-Schnobrich Model: Takayanagi and Schnobrich (1976) modified the Takeda model to
incorporate pinching and strength decay features caused by high shear acting in short coupling
beams connecting parallel structural walls. The skeleton curve is trilinear.

The reloading (loading in the opposite direction after unloading) is made smaller than the stiffness
toward the previous maximum response point in the direction of reloading; the response point moves
toward the previous maximum response point after the response deformation changes its sign.

The resistance at a target point for reloading in the hardening range is reduced from the
resistance at the previous maximum response point; e.g., the resistance at the target point is
selected on a strength decay guideline which descends from the yield point. After the response
reaches the target point, the response point moves along a line parallel to the post yielding line.

The pinching stiffness is based on the reinforcement resistance for bending. The rate of strength
decay is assumed to proportionally increase with the rotation.

M Y
My
Decay Guideline

Mc

Dm’
Dm
Pinching θ
Mc

My
Y’
Takayanagi-Schnobrich Model of
Pinching and Strength Decay

Roufaiel-Meyer Model: Roufaiel and Meyer (1987) used a hysteresis model that includes strength
decay, stiffness degradation and pinching effect.

47
 The moment resistance of a bilinear F A
Y
moment-curvature relation was assumed to decay
when a given strain is reached at the extreme
compression fiber. The curvature at the
commencement of strength decay is called the
critical curvature. The degradation in resistance was
assumed to be proportional to the amount by which
the critical curvature was exceeded. B

C
An auxiliary unloading branch AB is drawn O Ks Do
D
parallel to the elastic branch of the bilinear skeleton
curve until it intersects a line OB through the origin O
parallel to the strain-hardening branch YA of the
skeleton curve. The line connecting this latter point B
of intersection to the point of previous extreme
deformation in the opposite direction defines the end
C of the unloading branch on the horizontal axis. If (Dm,Fm)
yielding has not taken place in the direction of Roufaiel and Meyer Model
loading, the yield point is used as the previous
maximum response point.

From that point on reloading is not always directed straight to the point of the previous extreme
post-yield excursion in the direction of reloading, but it may include pinching, depending on the shear
ration, M/Vh. Pinching is accomplished by directing the reloading branch first towards a point on the
elastic branch of the skeleton curve at an ordinate equal to that of the intersection of this branch with
the line of straight reloading to the previous extreme deformation point, times m<1. The second part
of the reloading branch heads towards this latter extreme deformation point. Parameter m assumes
the following values;
for M/Vh<1.5 m=0
for 1.5<M/Vh<4 m=0.4(M/Vh)-0.6
for 4<M/Vh m=1
The slope of slipping stiffness is
F 'm
Ks = m
D 'm − Do

Chung et al. (1987) extended the

Roufaiel and Meyer model to include
strength and stiffness degradation at
constant amplitude cycling. The
degradation model requires two
additional parameters: the value of
curvature φ f and the moment m f at
failure in monotonic loading. The failure
is defined as rupture or buckling of
longitudinal reinforcement, concrete
crushing, or the reduction of resistance
to 75 %. If the bilinear approximation to
the moment-curvature curve under
monotonic loading is denoted by m p (φ ) , Modification of Roufaiel and Meyer model
by Chung (1987)
a single half-cycle of loading up to
monotonic failure causes a drop in the moment at failure is m p (φ f ) − m f . By extension, during a
half-cycle to a curvature φ the drop in moment relative to the bilinear monotonic envelope m p (φ )
is given by

48
 3
 φ −φy 2
∆m(half cycle at φ ) = {m p (φ f ) − m f }  
 φ −φ
 f y 
Accordingly, a branch of reloading in the direction where the previous maximum curvature is equal to
φ , moves toward a point at ( m p (φ ) − ∆m, φ ), rather than at ( m p (φ ), φ ) as in the original Roufaiel
and Meyer model.

Banon-Biggs-Irvine Model: Banon, Biggs and Max F Y K2

Irvine (1981) modified Takeda hysteresis model by (a)
using a bilinear skeleton curve, (b) incorporating
pinching and stiffness degradation. The pinching K1
hysteresis was adopted to simulate the propagation of
inclined cracks due to high shear and slippage of
longitudinal reinforcement. D

Hysteresis rules are summarized below; D’m

(a) Moment-rotation relationship is elastic up to the
K s Dm
yield point,
(b) Once the yield point is exceeded, loading
proceeds on the second slope of the bilinear envelope,
(c) Unloading is parallel to the elastic stiffness,
Y’
(d) The stiffness during reloading immediately after
unloading is reduced to 50 % of the second slope of the Banon-Biggs-Irvine Model (1981)
bilinear envelope,
K
Ks = 2
2
(e) When the direction of loading changes during unloading and resistance (or deformation)
starts to increase again, the reloading stiffness is parallel to the elastic stiffness before the response
point reaches a point where the last unloading started,
(f) When the sign of deformation changes during reloading, the response point moves toward
previous maximum response point in the direction of reloading.

If the strength-degrading feature is introduced, the response point after the pinching does not
move toward the previous maximum point, but a point on the skeleton curve at deformation greater
than the previous maximum deformation.
Dm
D *m =
α
and α = 0.8 is suggested in the study.
The skeleton curve may be different in positive and negative directions.

Kato Shear Model: Kato et al. (1983) used a

hysteresis model to represent the behavior of a F B
reinforced concrete member failing in shear, in
which strength decay and stiffness reduction due to
load reversals were incorporated. A trilinear A
skeleton curve was used with stiffness changes at A
and B. By choosing the skeleton curve without
descending stiffness, the stable flexural behavior
may be represented by this model. The trilinear
skeleton curve may include descending slopes. The
following example shows a skeleton curve with two D
descending slopes.
Skeleton Curves of Kato Model (1983)
49
 The response is linearly elastic before the response point reaches point A. The response point
follows the skeleton curve if the slope of the skeleton curve is positive; if the slope of the skeleton
curve is negative, the response point increases its deformation without the change in resistance
(plastic behavior).

If a response point crosses the descending branches during loading or reloading, the deformation
increases without change in resistance (perfectly plastic stiffness). Upon unloading from a maximum
response point on the perfectly plastic branch, the response point moves on a line parallel to the
initial elastic stiffness K e until the response point crosses the descending skeleton curve; the point
is termed as the maximum response point (Dmax, Fmax). Then the response point follows a line with
reduced stiffness K u ;
Dmax −α
Ku = K e ( )
Dy
where α : unloading stiffness degradation index, D y : yield deformation.

Upon reloading after crossing zero resistance line, the response point moves on a line with
reloading (slip) stiffness K s ;
Fmin D
Ks = ( max )− β
Dmin − Do Dy
where ( Dmin , Fmin ): previous maximum response point on the skeleton curve in the direction of
reloading, Do : deformation at the completion of unloading, D y : yield deformation in the opposite
direction.

Ke
D xo
Dmin D yp K s Ku

Dmax
ls
Fmin
l

This slip stiffness is used for deformation ls (= γ l), where l: length from the unloading point to
the intersection of slip line and the line connecting the origin and the negative maximum response
point ( Dmin , Fmin ). The response point during strain softening moves toward the previous maximum
point ( Dmin , Fmin ) or the yield point if no yielding was experienced in the reloading direction.

If unloading takes place during reloading toward previous maximum response point, the
unloading stiffness from the previous maximum response point is used. If the response point crosses

50
the zero resistance axis,
the response point
follows the same slip
stiffness previously
defined in the reloading
direction. The length of
slip deformation is
defined for l: length from
the new unloading point
to the intersection of slip
line and the line
connecting the origin and
the maximum response
point ( Dmax , Fmax ).

Values for the

parameters of this model
recommended for shear
failing reinforced
concrete members are
α =0.4, β =0.6 and
γ =0.95.

Values for flexure

dominated members are
α =0.2, β = γ = 0.0.

Park et al. model: The

model developed by Park
et al. (1987) includes (a)
stiffness degradation, (b)
pinching and (c) strength
degradation with cycling.
The skeleton curve is a Hysteresis model by Park et al. (1987)
trilinear relation. The
extension of unloading from the post-cracking branch of the virgin loading curve intersects the
pre-cracking branch of the trilinear virgin loading in the direction of unloading at an ordinate equal to
approximately two times the corresponding yield moment. The reloading branch is initially directed
towards a point on the previous extreme unloading branch, at a moment ordinate equal to a
user-specified percentage γ (approximately 0.5) of the yield moment. Before reaching this point
and upon exceedance of the previous maximum permanent deformation (curvature at the
intersection of the previous extreme unloading branch and the horizontal axis), the reloading branch
stiffens and moves toward the point of maximum deformation in the direction of reloading. The
strength degrades in proportion to the amount of energy dissipated up to the current point. The
proportionality constant depends on the amount of longitudinal reinforcement and confining
reinforcement.
F
Origin Oriented Model: Shiga (1976) suggested Y
a simple hysteresis model, in which the response
moves on the line connecting the previous
absolute maximum response point and the origin. C
If the response point reaches the maximum
response point, it moves on the skeleton curve. D
When unloading takes place from a point on the
skeleton curve, the response point moves on the
line connecting the newly attained maximum
51 C

Y
response point and the origin.

The model was obtained from the observation on steady-state response of reinforced concrete
structural model which oscillated about the origin of the force-deformation relation. No hysteresis
energy is dissipated during the oscillation within the
previous maximum response amplitude. Therefore,
viscous damping proportional to the initial stiffness is F Y
suggested as a mechanism to dissipate energy with
degradation of stiffness in a system.
C
Any shape may be used for the skeleton curve of
this model. This model is sometimes used to represent
a feature of shear-dominated member, which
dissipates small hysteresis energy and degrades its D
stiffness with plastic deformation. The model, however,
does not give residual displacement when the load
was removed. Therefore, the model may not be suited C
for the simulation analysis of response waveform.
Y
Similar to the origin oriented model, the response
point may directed toward the previous maximum
response point on the opposite direction. Such model may be called a peak oriented model.

Matsushima Strength Reduction Model: Short reinforced F

concrete columns, failing in shear, exhibit strength decay Fy 0
with load reversals and associated stiffness degradation.
Matsushima (1969) used a model to explain the damage of a
structure after shear failure in columns. The characteristics
of the model are basically of bilinear type, but the elastic k
stiffness Kn and the yield resistance Fn were degraded k0
D
whenever unloading takes place from a point on the
post-yielding line in a form;
Fy = a N Fy 0
Kn = α nk y k = b N k0
Fn = β Fyn

where Ky: initial elastic stiffness, Fy: initial yield resistance, n:

number of unloading from the post-yield stiffness line, α
and β are constants to decay rate.

References:

Banon, H., J. M. Biggs and H. Max Irvine, "Seismic Damage in Reinforced Concrete Frames,"
Journal of Structural Division, ASCE, Vol. 107, No. ST9, September 1981, pp. 1713-1729.
Chung, Y. S., et al., “Seismic Damage Assessment of Reinforced Concrete Members,” National
Center for Earthquake Engineering Research, State University of New York, Buffalo, Technical
Report NCEER-87-0022, 1987.
Kato, D., S. Otani, H. Katsumata and H. Aoyama, "Effect of Wall Base Rotation Behavior of
Reinforced Concrete Frame-Wall Building," Proceedings, Third South Pacific Regional
Conference on Earthquake Engineering, Victoria University of Wellington, New Zealand, May
1983.
Matsushima, Y., "Discussion of Restoring Force Characteristics of Buildings, the Damage from
Tokachi-oki Earthquake (in Japanese)," Report, Annual Meeting, Architectural Institute of Japan,
August 1969, pp. 587-588.
Park, Y. J., et al., “IDARC: Inelastic Damage Analysis of Reinforced Concrete Frame-Shear Wall
Structures,” National Center for Earthquake Engineering Research, State University of New

52
 York at Buffalo, Technical Report NCEER-87-0008, 1987.
Roufaiel, M. S. L., and C. Meyer, "Analytical Modeling of Hysteretic Behavior of R/C Frames,"
Journal of Structural Division, ASCE, Vol. 113, No. 3, March 1987, pp. 429-444.
Shiga, T., Vibration of Structures (in Japanese), Structural Series, Vol. 2, Kyoritsu Shuppan, 1976.
Takayanagi, T., and W. C. Schnobrich, "Computed Behavior of Reinforced concrete Coupled Shear
Walls," Structural Research Series No. 434, Civil Engineering Studies, University of Illinois at
Urbana-Champaign, 1976.

53
11.10 Axial Force-Bending Moment Interaction

It is known that bending resistance varies with

existing axial force in a reinforced concrete section.
The effect of axial load on flexural yield level was
considered by Mahin and Bertero (1976), in which
the yield moment of the multi-component model
was varied with the amount of axial load.

Takayanagi and Schnobrich (1976) modified

the Takeda model to include the effect of axial
force-bending resistance interaction in the analysis
of a coupled structural wall. The skeleton curve is
trilinear. A set of trilinear skeleton curves were
prepared for different level of axial force, and the
change in bending resistance with unit axial load
was evaluated. The moment m is assumed to
vary with curvature φ and axial force n , while
the axial force n is assumed to vary with
Takayanagi-Schnobrich model for
curvature φ and axial strain ε ;
axial load-moment interaction
m = m(φ , n)
n = n(φ , ε )

The assumption leads to an un-symmetric relation in an incremental form;

∂m ∂m ∂m ∂m ∂n ∂m ∂n
∆m = ∆φ + ∆n = ( + ) ∆φ + ∆ε
∂φ ∂n ∂φ ∂n ∂φ ∂n ∂ε
∂n ∂n
∆n = ∆φ + ∆ε
∂φ ∂ε
The above relation for incremental curvature ∆φ and strain ∆ε , and then modification factor was
developed to regain the symmetry;
∂m 1
∆m = ( )∆φ = EI * ∆φ
∂φ 1 − ∂m ∆n
∂n ∆m
∂n 1
∆n = { }∆ε = EA * ∆ε
∂
∂ε 1 − ( /
n ∂m ∆m ∂m
)( − )
∂φ ∂φ ∆n ∂n
∆n
where EI * : instantaneous flexural rigidity, and EA * : instantaneous axial rigidity. The ratio
∆m
is assumed to remain constant during a small load increment.

The stiffness is updated for the subsequent load increment considering the existing axial force
level. For an increase in axial force, the moment-rotation hysteresis relation is directed to the
corresponding loop with increased yield moment.

The axial force-moment interaction effect can be easily handled by "fiber" model. Curvature may
be assumed to distribute uniformly over a specified hinge region, for which a moment-rotation
relation can be evaluated on the basis of the moment-curvature relation at the critical section.

References:

Mahin, S. A., and V. V. Bertero, "Nonlinear Seismic Response of a Coupled Wall System," Journal
54
 of Structural Division, ASCE, Vol. 102, 1976, pp. 1759-1780.
Takayanagi, T., and W. C. Schnobrich, "Computed Behavior of Reinforced Concrete Coupled Shear
Walls," Structural Research Series No. 434, Civil Engineering Studies, University of Illinois at
Urbana-Champaign, 1976.

55
11.11 Special Purpose Models

Hysteresis Model for MS Model: A steel spring and a

concrete spring in the corner of section are located in the
same point, and are subjected to identical displacement
history. Therefore, the two springs may be combined into
a single composite spring. The skeleton curve is
expressed by a bilinear relation; the compressive yield
resistance is determined as the sum of the compressive
strengths of the concrete and the steel springs, and the Tension Compression
tensile yield resistance is equal to the yield resistance
of the steel spring.

Hysteresis relation is of the Takeda model type with

the bilinear skeleton curve; unloading stiffness in a
compression zone and in a tension zone was made
different:

In a compression zone:
K1 = S1 ( K se + K ce ) for Dm ≤ Dsy
−λ Composite spring hysteresis model
D for MS model
= S1 ( K SE + K ce ) m for Dm > Dsy
Dy

In a tensile zone:
K 2 = S 2 K se for D ' m ≥ − Dsy
−λ
D'
= S 2 K se m for D' m < − Dsy
Dsy
where Kse and Kce: initial elastic stiffness of the steel spring and the concrete spring, Dsy: yield
deformation of the concrete and steel springs, Dm: previous maximum response deformation in
compression, Dm': previous maximum deformation in tension, S1 = S2 =2.0 and k = 0.4.

Post yielding stiffness was chosen to be 0.02 times the initial elastic stiffness of the direction of
loading. Upon reloading in compression, the response point moves on the slip stiffness line toward a
point (Dm, Fm"), where F " m = θFm and θ = 0.4. When the sign of deformation changes, the
response point moves toward the previous maximum point in compression. Similar to the Takeda
model, the response point moves toward a peak of immediately outer loop.

Axial Force-deformation Model: Kabeyasawa and Shiohara et al. (1983) used a hysteresis model
for an axial force-deformation relation of a boundary column in the analysis of a structural wall. The
model was developed on the basis of the observed axial deformation behavior of the boundary
column in the test of the full-scale seven-story structure tested as a part of U.S.-Japan Cooperative
Program (Yoshimura and Kurose, 1985).

The tension stiffening was ignored; concrete was assumed to resist no tensile stress. The axial
stiffness in tension was made equal to the stiffness of the reinforcing steel in the boundary column,
and the stiffness in compression was assumed to be linearly elastic including the stiffness of the
concrete. The stiffness in tension changed at the tensile yielding of the longitudinal reinforcement.

The gravity loads was considered as the initial stress.

56
 A response point followed bilinear
hysteresis rules between the maximum Tension
response point ( Dmax , Fmax ) in the tension
side after yielding and a reference point Y'
( D yc ,− Fy ) on the skeleton curve in the
compression zone. The resistance − Fy at
the reference point was determined at the
compressive yielding of the longitudinal Elongation
reinforcement. Initial Load

The unloading stiffness Kr was degraded

with plastic deformation;
Dmax −α
Kr = Kc ( )
D yt
where, D yt : tensile yielding deformation,
Dmax : maximum deformation greater than Compression

D y , α : unloading stiffness degradation Axial force-deformation model for

parameter (= 0.9). wall boundary element (Kabeyasawa et al., 1983)

When the response point reached the previous maximum point ( Dmax , Fmax ) in tension, the
response point moved on the second slope of the skeleton curve, renewing the maximum response
point.

When the response point approached the compressive characteristic point Y' (Dyc, -Fy) in
compression, the response point was directed to move toward a point Y" (2Dyc, -2Fy) from a point P
(Dp, Fp) on the bilinear relation:
D p = D yc + β ( D x − D yc )
where, β : parameter for stiffness hardening point (=0.2), and D x : deformation at unloading
stiffness changing point. This rule was introduced to reduce an unbalanced force at the compressive
characteristic point Y' due to a large stiffness change. The compressive characteristic point Y' did
not change under any loading history.

This axial-stiffness hysteresis model was used for the axial deformation of an independent
column as well as boundary columns of a wall.

Slip Model: Reinforced concrete

members exhibit slip-type (pinching)
behavior before a wide crack closes or
when longitudinal reinforcing bars slip
after bond deterioration. The slip-type
behavior is characterized by a small
stiffness during reloading at low
resistance level after a large amplitude
deformation in the opposite direction
and by the gradual increase in
stiffness with deformation.

Tanabashi and Kaneta (1962) used

a slip model with elasto-plastic
skeleton curve and zero slip stiffness in their nonlinear response analysis. No hysteresis energy was
dissipated until the response point exceeded the previous maximum response point.

57
 A finite stiffness may be assigned to the slip stiffness and a stress hardening may start to occur
before the initiation of slip at preceding unloading.

Bond Slip Model: Morita and Kaku

(1984) proposed a hysteresis model to Bond Stress
represent the bond stress-bar slip relation
on the basis of their observation of the
test results. The model is prepared for
assuming various loading situations and
may be useful in a finite element analysis
of a reinforced concrete member.

Slip

(a) Reloading relation (S > SA)

Bond Stress
Bond Stress

Slip Slip

References:

Fillipou, F. C., E. P. Popov and V. V. Bertero, “Effect of Bond Deterioration on Hysteretic Behavior of
Reinforced Concrete Joints,” Report No. EERC 83-19, University of California, Berkeley, August
1983, 184 pp.
Fillipou, F. C., E. P. Popov and V. V. Bertero, “Modeling of Reinforced Concrete Joints under Cyclic
Excitations,” Journal, Structural Engineering, ASCE, Vol. 109, No. 11, November 1983, pp.
2666 - 2684.
Fillipou, F. C., “A Simplified Model for Reinforcing Bar Anchorages under Cyclic Excitations,” Report
No. EERC 85-05, University of California, Berkeley, March 1985, 61 pp.
Kabeyasawa, T., H. Shiohara, S. Otani and H. Aoyama, "Analysis of the Full-scale Seven-story
Reinforced Concrete Test Structure," Journal, Faculty of Engineering, University of Tokyo (B),
Vol. XXXVII, No. 2, 1983, pp. 432-478.
Li, K.-N., S. Otani and H. Aoyama, "Study on the Elastic-plastic Behavior of Reinforced Concrete
Columns subjected to Bi-directional Horizontal Earthquake Forces and Varying Axial Load (in
Japanese)," Report, Aoyama Laboratory, Department of Architecture, Faculty of Engineering,
58
 University of Tokyo, March 1990.
Morita, S., and T. Kaku, "Slippage of Reinforcement in Beam-column Joint of Reinforced Concrete
Frames," Proceedings, Eighth World Conference on Earthquake Engineering, San Francisco, U.
S. A., Vol. 6, 1984, pp. 477-484.
Tanabashi, R., and K. Kaneta, "On the Relation between the Restoring Force Characteristics of
Structures and the Pattern of Earthquake Ground Motion," Proceedings, Japan National
Conference on Earthquake Engineering, November 1962, pp. 57-62.
Yoshimura, M., and Y. Kurose, "Inelastic Behavior of the Building," ACI SP-84, Earthquake Effects
on Reinforced Concrete Structures, U.S.-Japan Research, American Concrete Institute, Detroit,
1985, pp. 163-202.

59
11.12 Hysteresis Model for Prestressed Concrete Members

This note summarizes a hysteresis model for prestressed concrete members. The model was
proposed by M. Hayashi et al. (1995).
mA
Member end moment M and rotation
θ of a prestressed concrete member under
anti-symmetric bending moment distribution
with the inflection point at mid span is mB = mA
considered. The skeleton curve of the
moment-curvature ( θ , M ) relation is
represented by a trilinear relation with
δ
stiffness changes at flexural cracking of
concrete and tensile yielding of longitudinal
reinforcement.
θA
θB = θ A

Trilinear Skeleton Relation: The two points to define a trilinear skeleton curve may be estimated as
follows.
(1) Initial elastic stiffness K1 is calculated for a prismatic line member considering flexural and
shear deformation:
L
K1 =
L 2
κ
+
3Ec I c Gc Ac
where L : member length from the face of the orthogonal member to the inflection point (0ne-half of
clear span or height), Ec and Gc : elastic and shear moduli of concrete, I e : moment of inertia of
the transformed concrete section, Ac : cross sectional area of the transformed concrete section, κ:
shape factor for shear deformation (=1.2 for a rectangular section).

(2) Cracking moment M c and rotation θ c are calculated for a condition that the tensile stress at
the extreme tensile fiber reaches the tensile strength σ t of concrete:
Pe
M c = (σ t + ) Ze
Ac
Mc
θc =
K1
where, Pe : axial force acting on the section including effective prestressing force, Ac : cross
sectional area of concrete, Z e : section modulus of the transformed section. Tensile strength σ t of
concrete may be assumed to be equal to 1.8 σ B in kgf/cm2 where compressive strength σ B of
concrete is expressed in kgf/cm2.

(3) Yielding moment M y should be calculated for a given axial force and effective prestressing
force assuming (a) plain section to remain plain after bending, (b) nonlinear axial stress-strain
relation of concrete and reinforcement, and (c) equilibrium of internal and external forces. A parabola
and straight descending line may be used to represent stress-strain relation of concrete in
compression; an elasto-plastic stress-strain relation may be used for steel reinforcement ignoring
strain hardening.

Rotation θ y at yielding may be evaluated by integrating the curvature along the member, but
this often underestimates the deformation. Sugano (1970) proposed an empirical expression for the

60
ratio of secant stiffness at yielding to the initial elastic stiffness as follows;
My
θy a N d
αy = = {0.43 + 1.64n pt + 0.043
+ 0.33 }( )2
K1 D b D σB D
where n : modular ratio of steel to concrete, pt :
tensile reinforcement ratio including prestressing Y
reinforcement area as increased by the yield stress My
ratio of prestressing reinforcement to ordinary
reinforcement, a : shear span, b and D : width and
depth of member section, N : axial force of section

Moment
including effective prestressing force, d : effective
depth of section, σ B : compressive strength of
α y K1
concrete. C
Mc
MA A
Characteristic Points on Hysteresis Relations:
The following points and stiffness are used in this
model. K1 Rotation
(1) Characteristic point A ( θ A , M A ) is defined on the
initial elastic stiffness line with stiffness K1 . This θ A θc θy
point is used for the hysteresis relation of a prestressed concrete member. Moment resistance M A
of the characteristic point is defined as decompression moment; i.e., for the effective prestressing
force Pe,
Pe
MA = Ze
Ac
The moment is zero for a reinforced concrete member without prestressing force. Rotation θ A is
calculated for moment M A and initial stiffness K1 ;
MA
θA =
K1

(2) Characteristic point B ( θ B , M B ) is defined for Takeda hysteresis model (Takeda, Sozen and
Nielsen, 1970) as the terminal point (zero moment resistance) of unloading from the maximum
response point M ( θ m , M m ). The unloading stiffness K B is defined as follows;
(a) unloading before yielding:
M m − M 'c
KB =
θ m − θ 'c
(b) unloading after yielding:
M y − M 'c θ m −γ
KB = ( )
θ y − θ 'c θ y
where M y and θ y : yield moment and rotation on the side of the unloading point, M 'c and θ 'c :
cracking moment and rotation on the opposite side, and γ : unloading stiffness degradation index of
the Takeda model (=0.5 for normal reinforced concrete members). The rotation θ B is calculated as
Mm
θB = θm −
KB

61
 Y M

Moment
Y
My

Moment
Mm

C C
KB
A A KB
Rotation Rotation
θ 'c B θ 'c B

θB θm θB θy θm

M 'c M 'c
C’ C’

(3) Unloading stiffness K A of fully prestressed concrete members from the maximum response
point M ( θ m , M m ) on the skeleton curve is defined as follows;
Mm M
Y Y
Moment

My

M
Mm
Moment

KA

C C
KA
MA MA A
A
Rotation Rotation

θA θm θA θy θm

(a) unloading before yielding:

Mm − M A
KA =
θm −θ A
(b) unloading after yielding:
M y − M A θ m −γ
KA = ( )
θ y −θ A θ y
(4) Unloading stiffness K D of this model from the maximum response point M ( θ m , M m ) on the
skeleton curve is defined by index λ ' taking into consideration the characteristics of both reinforced
concrete and fully prestressed concrete members.
K D = λ ' K A + (1 − λ ') K B

(5) Characteristic point D ( θ D , M D ) is defined as an intersection of line AB and the unloading line
MD of this model with unloading stiffness K D from the maximum response point M( θ m , M m ) on the

62
skeleton curve.

Moment M
Moment
Y
Mm Y

M
Mm

KD
C C
KD
MA MA A
A
D D

B Rotation Rotation

θ A θB θm θA B θy θm

Coordinates of the intersection D ( θ D , M D ) must satisfy the two equations;

MA MD
= : line AB
θ A −θ B θ D −θ B
Mm − MD
= KD : line MD
θm −θ D
The moment resistance M D is thus solved from the two simultaneous equations as
M
θm −θ B − m
KD
MD = MA
M
θ A −θ B − A
KD

(6) Unloading stiffness K E after reaching characteristic point D

(6-1) No yielding has taken place on the unloading side:
K E = K1 (initial elastic stiffness)
(a-1) The terminal point of this unloading stiffness is point E’ at moment level equal to moment
M ' A of characteristic point A’ on the opposite side if no cracking has taken place on the
opposite side.
(a-2) The terminal point of this unloading stiffness is point E’ at moment level equal to moment
M 'D of characteristic point D’, which was defined during unloading after cracking on the
other side.
(6-2) Yielding has already occurred on the unloading side,
θ m −γ
K '1 = K1 ( )
θy
M y − M 'c θ m −γ
KB = ( )
θ y − θ 'c θ y
K E = λ ' K '1 + (1 − λ ') K B
where rotation θ m is the maximum response rotation where the unloading initiated. The terminal
point of this unloading stiffness is point E’ at moment level equal to moment M 'D of
characteristic point D’ in the opposite direction.
63
Characteristics of Prestressed Concrete:

(1) Unlading stiffness degradation index γ

Index γ was initially used in
the Takeda hysteresis model
1.0

Unloading stiffness degradation index

(Takeda, Sozen and Nielsen,
1970) to control the unloading 0.8
stiffness from the maximum
deformation on the post-yielding 0.6
skeleton curve and also to control
the hysteresis area per cycle. The
0.4
unloading stiffness was degraded
from the reference stiffness,
defined as a slope connecting the 0.2
yield point of unloading side and
the cracking point on the opposite 0.0
side. 0.0 0.2 0.4 0.6 0.8 1.0
M y − M 'c θ m −γ Ultimate moment ratio λ of prestressing reinforcement
KB = ( )
θ y − θ 'c θ y

The index value from 0.4 to 0.5 is normally used for reinforced concrete members. The hysteresis
energy dissipation decreases with increasing value of the index.

The unloading stiffness in moment-rotation relations of reinforced concrete and prestressed

concrete member tests were examined. The following graph was suggested by Hayashi et al. (1995)
to express the unloading stiffness degradation index γ as a function of the ratio λ of ultimate
moment resistance attributable to prestressing reinforcement to the total ultimate moment. The ratio
λ is zero for a reinforced concrete member, unity for a fully prestressed concrete member, and
between zero and unity for a partially prestressed concrete member. The unloading stiffness
degradation index γ for partially prestressed to fully prestressed concrete is 0.7 to 0.8. The
hysteresis energy dissipation per cycle is less compared with a reinforced concrete member.

(2) Index λ ' for unloading stiffness

Index λ ' controls unloading
1.0
stiffness of reinforced concrete,
Index for unloading stiffness

partially prestressed reinforced

concrete and fully prestressed 0.8
concrete members. The unloading
stiffness of the moment-rotation 0.6
relations of member tests was
examined. Hayashi et al. (1995) 0.4
suggested expressing index λ '
for unloading stiffness as a function
0.2
of the ratio λ of ultimate moment
resistance attributable to
prestressing reinforcement to the 0.0
0.0 0.2 0.4 0.6 0.8 1.0
total ultimate moment. The index
λ ' increases with increasing Ultimate moment ratio λ of prestressing reinforcement
ultimate moment ratio λ .

The effect of prestressing on unloading stiffness is negligible when the ultimate moment ratio was
less than 0.3.

64
Hysteresis Rules:

Rule 1: Before flexural cracking at C ( θ c , M c ), the Y

My
relation is linearly elastic with stiffness K1 . Rule 3

Moment
If the response point reaches flexural cracking point C Rule 2
K2
( θ c , M c ), the response point follows Rule 2.

C
Rule 2: The response point ( θ , M ) moves on the Mc
second skeleton line CY toward yield point Y ( θ y , M y ) Rule 1
with stiffness K 2 .
K1 Rotation
If the response point reaches the yielding point Y
θc θy
( θ y , M y ), the response point follows Rule 3.
Moment-rotation relation during loading
If the unloading takes place from the maximum
response point M ( θ m , M m ), the response point follows Sub-rule 2-1.

Sub-rule 2-1: The response point moves on line MD

with unloading stiffness K D , where M ( θ m , M m ) is the
Moment
Y
previous maximum point on the second skeleton line
CY. The unloading stiffness K D is defined as M
Mm − M A Mm
KA =
θm −θ A
M m − M 'c C
KB = KD
θ m − θ 'c MA A

K D = λ ' K A + (1 − λ ') K B D

Characteristic point D is defined as the intersection of B Rotation

the unloading line MD and line AB connecting two θ A θB θm
characteristic points A and B. The moment M D at
point D is given by
Mm
θm −θ B −
KD
MD = MA
M
θ A −θ B − A
KD
The response point moves on line MD during unloading and reloading.

If the response point reaches the previous maximum point M ( θ m , M m ) during reloading, the
response point follows Rule 2.

If the response point reaches point D during unloading, the response point follows Sub-rule
2-2.

Sub-rule 2-2: The response point follows Sub-rule 2-2-1 if no cracking has taken place in the
opposite direction, or Sub-rule 2-2-2 if cracking has taken place in the opposite direction.

Sub-rule 2-2-1: The response point moves elastically from the characteristic point D
( θ D , M D ) of Sub-rule 2-1 to point E’ whose moment level is equal to moment M ' A of

65
 characteristic point A’ in the direction of reloading.
M
The slope K E of line DE’ is equal to the initial Mm
stiffness K1 . Between characteristic point D and
cracking point E’, the response point moves on line
DE’ during reloading and unloading.
C KD
The previous response point M’ on the side of
point E’ is defined as the cracking point C’. The K1
Sub-rule 2-1
unloading stiffness K 'D from the previous A

maximum M’ (cracking point C’) is defined as K1 .

Characteristic point D’ is defined as the
D
characteristic point A’, and moment M 'D of the
characteristic point D’ is equal to moment M ' A . B

Sub-rule 2-2-1
θm
If the response point reaches point E’, the
response point follows Rule 4. K E = K1

A’ No cracking in
Sub-rule 2-2-2: The response point moves E’
reloading
elastically from the characteristic point D ( θ D , M D ) Rule 4 direction
of Sub-rule 2-1 to point E’ whose moment level is C’
equal to moment level M 'D of characteristic point
D’ in the direction of reloading. The unloading
stiffness K E is equal to the initial elastic stiffness K1 . Point D’ and its moment M 'D have
been defined by Sub-rule 2-1 or
M
Sub-rule 3-1 upon previous unloading
Mm
from point M’ ( θ 'm , M 'm ) on the second
skeleton line.

Between characteristic points D and

E’, the response point moves on the
C Sub-rule 2-1
same line DE’ during reloading and
unloading. K1
A KD
If the response point reaches
characteristic point D during reloading,
the response point moves toward
previous maximum response point M MD D
( θ m , M m ) in the direction of reloading
following Sub-rule 2-1. B’ B
Sub-rule 2-2-2 θm
If the response point reaches point E’
after crossing zero moment axis D’ E’
Cracked in
(moment reversal), the response point Rule 4 reloading
follows Sub-rule 4.
direction
A’
Rule 3: The response point follows the third
skeleton line with stiffness K 3 .
C’

If the unloading takes place at maximum

response point M ( θ m , M m ), the response point M’
follows Sub-rule 3-1.

66
Sub-rule 3-1: The response point follows the unloading stiffness K D ,
M y − M A θ m −γ
KA = ( ) Mu
θ y −θ A θ y Y
Rule 3 M

M y − M 'c θ m −γ My
KB = ( )
θ y − θ 'c θ y
K D = λ ' K A + (1 − λ ') K B Sub-rule 3-1

where yielding point Y ( θ y , M y ) is on the side of C

KD
maximum response point M ( θ m , M m ) and
cracking point C’ ( θ 'c , M 'c ) is on the other side. A

Characteristic point D is defined as the D

intersection of the unloading line MD and line AB
connecting characteristic points A and B. B
Moment M D at characteristic point D is
determined by θy θm
M
θm −θ B − m
KD
MD = MA
M
θ A −θ B − A
KD
The response is elastic between unloading point M and characteristic point D.

If the response point reaches the unloading point M, the response point follows Rule 3 for
loading on the third skeleton line.

If the response point reaches the

characteristic point D ( θ D , M D ) of unloading, M
the response point follows Sub-rule 3-2. Y

Sub-rule 3-2: The response point follows

Sub-rule 3-2-1 if no cracking has taken place in
the direction of reloading, Sub-rule 3-2-2 if
cracking has taken place in the direction of C
reloading.
KD
A Sub-rule 3-1
Sub-rule 3-2-1: The response point
moves elastically on line DE’ with
MD D
unloading stiffness K E , where unloading
stiffness is defined as
θ m −γ KE B
K '1 = K1 ( )
θy Sub-rule 3-2-1

M 'A
M y − M 'c θ m −γ A’ E’
KB = ( )
θ y − θ 'c θ y C’
No cracking on
K E = λ ' K '1 + (1 − λ ') K B opposite side
Point E’ is defined on the unloading line at
Rule 4
moment level equal to moment M ' A of
characteristic point A’ on the initial
stiffness. Y’

67
 The previous response point M’ on the side of point E’ is defined as the yielding point Y’.
The unloading stiffness K 'D from the yield point is defined as
M ' y − M 'A
KA =
θ 'y − θ 'A
M 'y − M c
KB =
θ 'y −θc
K D = λ ' K A + (1 − λ ') K B
Characteristic point D is defined as the intersection of the unloading line MD and line AB
connecting two characteristic points A and B. The moment M D at point D is given by
My
θ y −θ B −
KD
MD = MA
M
θ A −θ B − A
KD

If the response point reaches point D, the response point follows Sub-rule 3-1.

If the response point reaches point E’, the response point follows Rule 4.

Sub-rule 3-2-2: The response M

point moves elastically on line DE’ Y
with unloading stiffness K E . The
unloading stiffness K E is defined
by
θ m −γ C
K '1 = K1 ( )
θy A
Sub-rule 3-1

M y − M 'c θ m −γ
KB = ( )
θ y − θ 'c θ y D
B’
K E = λ ' K '1 + (1 − λ ') K B
KE B
Point E’ is defined on the
unloading line at moment level D’ Sub-rule 3-2-2

equal to moment level M 'D of E’

A’
the characteristic point D’. The Rule 4

characteristic point D’ and its Cracking in reloading

C’ direction
moment M 'D were defined
during previous unloading from
point M’ on the skeleton curve in
the opposite direction under
Sub-rule 2-1 or 3-1.
M’ Y’
If the response point reaches point D, the response point follows Sub-rule 3-1.

If the response point reaches point E’, then the response point follows Rule 4.

Rule 4: The response point moves on line E’M’ toward the previous maximum response point M’
( θ 'm , M 'm ) in the direction of reloading. The characteristic point E’ is defined either in Sub-rule 2-2
or 3-2.

68
 When the response point reaches the previous maximum response point M’ in the direction of
reloading, the response point follows either Rule 2 if no yielding has taken place or Rule 3 if yielding
has taken place in the direction of reloading.

If unloading takes place at point N’ before reaching the previous maximum response point, the
response point follows Sub-rule 4-1.
C
M
Y

A KD N

G C
MD D
A

θ 'm B’ B
E
D
F’ K E = K1 B’
G F
D’ E’ B
F’
D’ G’
K 'D
A’
A’
N’ N’ C’

C’

M 'm
M’ Y’
M’

Sub-rule 4-1: The response point moves on line N’F’ with unloading stiffness K 'D where
point N’ is an unloading point before the response point reaches the previous maximum
response point M’ in Sub-rule 2-2 or M
3-2. Point F’ is defined on the
unloading line N’F’ at moment level Y
equal to moment M 'D of
characteristic point D’. The
unloading stiffness K 'D ,
characteristic point D’ and its C KD
moment M 'D were defined during
KD
previous unloading from point M’ A
( θ 'm , M 'm ) on the skeleton curve in G
Sub-rule 2-1 or 3-1. MD D
B’
The response point follows line KE B
N’F’ during reloading and
D’ F’ M 'D
unloading.
M 'A
If the response point reaches K 'D A’ E’
point N’ during reloading, it follows
C’
Rule 4.

If the response point reached

point F’ during unloading, the
response point follows Sub-rule 4-2. N’
Y’

69
 Sub-rule 4-2: The response point follows line F’G with unloading (reloading) stiffness K 'E
where characteristic point F’ is defined in Sub-rule 4-1. The moment level of point G is equal
to moment M D of characteristic point D. The characteristic point D’ and its moment M 'D
were defined during previous unloading from point M’ ( θ 'm , M 'm ) on the skeleton curve in
Sub-rule 2-1 or 3-1. The unloading stiffness K 'E depends on the previous maximum
response on the unloading side;
(a) If no yielding has taken place on the unloading side, the unloading stiffness K 'E is
equal to the initial elastic stiffness K1 .
(b) If yielding has taken place on the unloading side, the unloading stiffness is given
below;
θ m −γ
K '1 = K1 ( )
θy
M ' y − M c θ 'm − γ
KB = ( )
θ 'y −θ c θ 'y
K 'E = λ ' K '1 + (1 − λ ') K B

The response point follows line F’G during unloading and reloading.

If the response point reaches point G, the response point follows Rule 4.

If the response point reaches F’ during reloading, the response point moves toward point
N’ following Sub-rule 4-1.

References:

Hayashi, M., S. Okamoto, S. Otani, H. Kato, and J. Fu, “Hysteresis Model for Prestressed Concrete
Members and its Effect on Earthquake Response (in Japanese),” Journal, Prestressed Concrete,
Japan Prestressed Concrete Engineering Association, Vol. 37, No. 4, July 1995, pp. 57-67.
Sugano, S, "Experimental Study on Restoring Force Characteristics of Reinforced Concrete
Members (in Japanese)," Thesis submitted to fulfill the requirements of Doctor of Philosophy,
University of Tokyo, March 1970.
Takeda, T., M. A. Sozen and N. N. Nielsen, "Reinforced Concrete Response to Simulated
Earthquakes," Journal, Structural Division, ASCE, Vol. 96, No. ST12, 1970, pp. 2557-2573.

70
 Home Assignment No. 6

2002-03-11
Otani, S.

Displacement history is given below;

(1) D= 0.0
(2) D= 2.0 Dy
(3) D=-2.0 Dy
(4) D= Dy
(5) D=-0.5 Dy
(6) D= 4.0 Dy
(7) D= -2.0 Dy
(8) D= 4.0 Dy
where Fy=1.0, Dy=1.0. The skeleton curve is elasto-plastic.

Draw resistance-deformation (F-D) relation for the two hysteresis models;

(1) Clough Model (no unloading stiffness degradation)
(2) Pivot hysteresis model ( α = 2.0 , β = 0.5 )

71
 Chapter 12 Response of Different Models

12.1 Effect of Member Modeling

Various member models have been proposed to represent the distribution of stiffness within a
reinforced concrete member. The effect of member models on the nonlinear response is studied by
Shiohara et al. (1983).

Member Models: Four member models are studied; (a) one-component model (Giberson, 1967), (b)
multi-component model (Clough et al., 1965), (c) distributed flexibility model (Takizawa, 1976) and (d)
discrete spring model.

One-component model consists of a linearly mA Elastic Element

kA kB
elastic element and two rotational springs at the A
θB
ends of the elastic element. All inelastic
deformation is assumed to concentrate at the
member ends. θA L B mB

The multi-component model assumes that a One Component Model

member is divided to four imaginary parallel
elements; an element rigidly connected at both
ends, two elements with a hinge at one end and
rigidly connected at the other end and a truss
element with hinges at both ends. The last
element does not contribute to the resistance
and stiffness of the member. The dividing ratios are varied with a strain history to accommodate an
arbitrary member end hysteresis relation as proposed by Takizawa (1976).

The discrete spring model utilized 20

equal-length rigid segments and 19 rotational Rotational Spring
springs in-between; the number of segments was Rigid Element
increased in a preliminary study, but the
response was not influenced by the number
greater than 20.

The distributed flexibility model assumes a

parabolic distribution of flexural flexibility
(reciprocal of flexural rigidity EI) along the member
with an elastic flexibility at the lowest flexibility
point. The flexural flexibility at member ends was
given by a hysteresis model. The model is useful
to represent spread of damage (cracks) along the
member, but not suitable to represent the
concentration of inelastic deformation at the
member end.

Stiffness Properties: Moment-curvature relation of section F

was assumed to be bilinear with a post-yielding stiffness of 2
percent of the initial elastic stiffness. The response of the discrete
element model was believed to give the most realistic response of
the four models if the behavior of the member was dominated by Fy
kp=0.02ky
flexure; however, the yielding was found to occur at slightly later
stage than the other models because rotational springs were not
placed at the member ends. The Clough hysteresis model ku
(Clough, 1966) was used for the moment-curvature relation with a ky D
bilinear skeleton relation and a degrading unloading stiffness ku:
Dy Dm
1
 k p = 0.02k y
k u = k y ( Dm / D y )−α
in which, k p : post-yielding stiffness, k u : unloading stiffness, k y : initial elastic stiffness, Dm :
maximum deformation beyond the yield displacement D y , α : unloading stiffness degradation
coefficient (= 0.4).

A member end moment-rotation relation was calculated for the anti-symmetric moment distribution
using the discrete element model. The calculated member end moment-rotation relation was
idealized into a bilinear relationship connecting the yield point and a point corresponding to rotational
ductility factor (a ratio of a rotation divided by the yield rotation) of 9.0. The Clough hysteresis model
was also used for a member end moment-rotation relation on the basis of the skeleton relation
calculated by the discrete spring model. The unloading degradation factor for a member end
moment-rotation relation was estimated to be 0.43 for the discrete spring model under load reversal
at a rotational ductility factor of 9.0.

The stiffness properties of the one-component mode, multi-component model and distributed
flexibility modes are determined on the basis of a hysteresis model provided for the member end
moment-rotation relation under an imaginary anti-symmetric moment distribution along the member.
Note that the actual member end rotation is not necessarily identical to the rotation given by the
hysteresis model, or the member stiffness identical to the stiffness given by the corresponding
hysteresis model.

Response under Uniform Bending: A hysteresis mA mB

model is derived for a member end moment-rotation
relation under the anti-symmetric moment distribution. θA θB
The use of such a hysteresis relation for a loading
A B
situation drastically different from the anti-symmetric
moment distribution is expected to clarify the
difference in the performance of the member models. Member under uniform bending
Therefore, the member models are subjected to a
stress history of uniform bending moment distribution along the member, although such a uniform
bending moment distribution is not expected to develop in a member during an earthquake.

A simply supported member was

Stiffness reduction factor after yielding, %
subjected to external moments, m A and
0 5 10 15 20
m B , of equal amplitude but of the opposite
sign at the two ends. The amplitude of the
moments was monotonically increased, and One component model
the response of the member was calculated
for the four member models. Naturally, the Multi-component model
four models exhibited the identical elastic
stiffness; the yield point was also identical
for the three models except for the discrete Distributed flexibility model
spring model which showed a slightly higher
yield resistance and deformation because
Discrete element model
rotational springs were not placed at the
ends.

The difference among the four models was observed in the post yield stiffness. The post yield
stiffness of a member end moment-rotation relation must be identical to the that of the
moment-curvature relation because a uniform bending moment was applied; the member end
rotation should be equal to φ ( L / 4) ; i.e. the post yielding stiffness should be 0.02 of the initial
stiffness.

2
 The one component model exhibited the highest post-yield stiffness, more than 13 percent of the
initial stiffness because the model does not consider the distribution of curvature along the member.
The multi-component model and the distributed flexibility model developed 5 percent of the initial
stiffness.

Response under Predetermined

Displacement History: A simply sup-ported
member is subjected to member end rotations θ A
and θ B of a predetermined history. A member
A end
end rotation θ B was delayed from the other end B end
rotation θ A by a quarter of a cycle. Large

Ductility factor
inelastic deformation is intended at the member
ends; the rotation amplitudes in the first and
second loading cycles are 6 and 12 times the yield
rotation.

A similar tendency is observed in the first cycle

of loading. The three models exhibit comparable
hysteresis shapes. The discrete spring model dissipates the largest hysteresis energy per cycle,
followed by the one component model, and the multi-component model. The multi-component model
dissipates less hysteresis energy at A end and more at B end.

One component model

Discrete spring model
Multi-component model

(a) A end (b) B end

The moment-rotation relation in the second cycle (load stages 5 to 8 in the loading history) after
yielding in the first cycle is compared for the four models. The hysteresis area is much thinner at A
end than at B end; the hysteresis loops were generally thin at A end. Lager hysteresis energy is
dissipated at A-end by the one-component model compared to the other two models. The post
yielding stiffness is highest for the one-component model as observed in the member end
moment-rotation relation at B end. The response of the multi-component model is closer to that of the
discrete element model.

Earthquake Response of Two-story One-bay Frame: A two-story one-bay frame is analyzed under
El Centro (NS) 1940 earthquake motion, the acceleration amplitude is scaled to the maximum
acceleration of 3.90 m/sec2. The one component model and the multi-component model are used in
the analysis.

3
 Span width is 6.0 m; the first-story height is
3.75 m, and the second-story height 3.00 m. Floor
weights at the two levels are the same and 50 tonf.
The base of the first-story columns is fixed to the
rigid ground. Young's modulus of the concrete is
2.37 x 105 kgf/cm2. No damping is assumed in the
analysis.

The beams have the following stiffness

properties; initial elastic moment of inertia Ib = 5.9
x 10-3 m4, cracking moment Mc = 5.2 tonf-m, My+ =
21.7 tonf-m for positive bending, and My- = 8.5
tonf-m for negative bending. The ratio β of
tangent stiffness after cracking to the initial
stiffness is 0.273 for positive bending and 0.080
for negative bending; post yield stiffness is
arbitrarily assumed to be 0.02 of the initial stiffness.
The Takeda-slip model (Kabeyasawa, 1982) is
used for the hysteresis relation.

The columns have the following stiffness

properties both in positive and negative directions;
initial elastic moment of inertia Ic = 5.0 x 10-3 m4,
cracking moment Mc = 5.2 tonf-m, and yield
moment My = 19.3 tonf-m. The ratio β of tangent
stiffness after cracking to the initial stiffness is
0.154; post yield stiffness is arbitrarily assumed to
be 0.02 of the initial stiffness. The Takeda model
(Takeda et al., 1970) is used for the hysteresis
relation.

The response waveform of the second floor

level (roof level) displacement, base shear,
member end rotations is compared for the two
models.

Although the second-floor displacement

waveform is similar for the two model, the
multi-component model exhibited a slightly larger
maximum response and also residual
displacement. The residual displacement must be
associated with the large plastic deformation at beam ends of the multi-component model.

Member end deformation of the second floor beam, using the multi-component model, shows the
drift corresponding to the deformation under uniform bending. The multi-component model developed
a maximum deformation larger than the one-component model.

The deformation waveform at the ends of the first story column is similar, but the one-component
model calculates larger amplitudes Some residual deformation was calculated by the
multi-component model at the top of the first story column.

Response of Four-story and Seven-story Buildings: In the nonlinear earthquake response

analysis of a four-story five-bay frame and a seven-story three-bay frame, the response of three
member models was compared (Shiohara et al., 1983); i.e., (a) the one-component model, (b)
multi-component model and (c) distributed flexibility model. No damping was assumed. The 1940 El
Centro (NS) motion was scaled to yield the maximum acceleration of 4.37 m/sec2. Displacement

4
response at the roof level and base shear response were compared.

One-component model
Multi-component model
4.0 Distributed flexibility model

-4.0
Ground motion acceleration, m/sec2
15

-15
Roof-level Displacement, cm
2.0

-2.0 Base shear, x 103 tonf

Reponse of four-story building

15

-15 Roof level displacement, cm

8.0

-8.0 Base shear, x103 tonf

Response of seven-story building

The response waveforms were comparable for the three models in the analysis of the four-story
frame. The multi-component model and distributed flexibility model developed slightly larger residual
displacement. Multi-component model developed slightly smaller base shear response. Beam end
rotation response of the three models was comparable, but the distributed flexibility model showed
accumulation of residual displacement with a number of oscillations.

The three models responded almost the same way up to 6 sec in the analysis of the seven-story
frame, but the multi-spring model and the distributed flexibility model lengthened the period of
oscillation after a large amplitude response at 6 sec; the distributed flexibility model behaved quite
different manner than the other two models. The difference may be attributable to the contribution of
higher modes in the response; i.e., the inflection point of a column may locate near the mid-height
under the oscillation in the fundamental mode. Therefore, the difference in the model characteristics
may not appear in the response of a low-rise structure. However, the shift of an inflection point due to
the higher mode oscillation tends to reveal the difference in model characteristics.

References:

Clough, R. W., K. L. Benuska and E. L. Wilson, "Inelastic Earth-quake response of tall buildings,"
Proceedings, Third World Conference on Earthquake Engineering, New Zealand, Vol. II, Session
II, 1965, pp. 68 - 89.
5
Clough, R. W., and S. B. Johnston, "Effect of Stiffness Degradation on Earthquake Ductility
Requirements," Proceedings, Second Japan Conference of Earthquake Engineering, October
1966, pp. 227 - 232.
Giberson, M. F., "The Response of Nonlinear Multi-story Structures subjected to Earthquake
Excitation," EERL Report, Earthquake Engineering Research Laboratory, California Institute of
Technology, Pasadena, 1967.
Kabeyasawa, T., "U.S.-Japan Cooperative Research, - Study on the Earthquake Resistance of
Seven-story Reinforced Concrete Structure (Part 3: Pseudo-Dynamic Analysis) (in Japanese),"
Proceedings, Sixth Japan Conference of Earthquake Engineering, December 1982, pp. 1161 -
1168.
Shiohara, H., S. Otani and H. Aoyama, "Comparison of Various Member Models for Reinforced
Concrete Earthquake Response Analysis," Transactions, Japan Concrete Institute, Vol. 5, 1983,
pp. 269 - 276.
Shiohara, H., S. Otani and H. Aoyama, "Comparison of Various Member Models for Reinforced
Concrete Earthquake Response Analysis," Transactions, Japan Concrete Institute, Vol. 5, 1983,
pp. 269 - 276.
Takeda, T., M. A. Sozen and N. N. Nielsen, "Reinforced Concrete Response to Simulated
Earthquakes," Journal, Structural Division, ASCE, Vol. 96, No. ST12. December 1970, pp. 2557 -
2573.
Takizawa, H., "Notes on Some Basic Problems in Inelastic Analysis of Planar R/C Structures (Part
1)," Transactions, Architectural Institute of Japan, No. 240, February 1976, pp. 51 - 62.

6
12.2 Effect of Damping Modeling

The effect of viscous damping on the earthquake response of a single-degree-of-freedom system

is studied using an instantaneous damping coefficient c * proportional to constant mass m and
*
varying instantaneous stiffness k :
c * = cm m + ck k *
*
An instantaneous damping factor h may be defined as
*
c
h* =
2 mk *
c m ck k*
= m +
2 k* 2 m
Note that, with the degradation of stiffness, the instantaneous damping factor h * associated with
the mass increases and that associated with the instantaneous stiffness tends to decrease.

A hysteresis model includes the hysteresis energy dissipation, and it may not be reasonable to
expect additional energy dissipation by the damping during inelastic oscillation. A constant
mass-proportional damping tends to exaggerate the damping effect.

For a given earthquake motion, the degree of damping effect on the response may depend on (a)
type of damping, (b) period of vibration, (c) capacity of hysteresis energy dissipation, and (d) level of
ductility demand. The effect of damping is studied from these view points.

Type of Damping: Mass-proportional

damping is expected to be more effective
reducing the response amplitude where
many cycles of oscillation occurs with
highly degraded stiffness. On the contrary,
instantaneous stiffness-proportional
damping is effective during oscillation in a
small ductility range.

Attained ductility of the Takeda model

with an unloading stiffness degradation
parameter α of 0.0 is compared. The
"yielding period" of
single-degree-of-freedom systems was
varied from 0.14 sec to 1.13 sec. Taft
(N21E) record was used as an excitation
function.

The system with mass-proportional

damping produced small displacement
response than that with
stiffness-proportional damping having the
same initial damping factor. With an Effect of types of damping
increase in the value of initial damping
factor, the mass proportional damping is
more effective in reducing the response
amplitude. This tendency is larger for a shorter period system. The response amplitudes of systems
with stiffness-proportional damping are not so sensitive to the increase in the value of initial damping
factor partially attributable to the fact that the hysteretic energy dissipation is appreciable when the
unloading stiffness degradation parameter α was 0.0. When the initial damping factor is made of
equal contributions for the mass-proportional damping and the stiffness-proportional damping, the
mass-proportional damping tends to have a dominant influence on the maximum response.

7
 Displacement response waveforms are compared for two types of damping.
Single-degree-of-freedom systems with yielding period of 0.4 sec were subjected to the El Centro
(NS) 1940 earthquake motion. The Takeda hysteresis model was used with unloading stiffness
degradation parameters α of 0.0 and 0.5.

The waveforms are generally similar. The effect of damping is large when an unloading stiffness
degradation parameter α is 0.5; i.e., the hysteresis energy dissipation is small. The system with the
stiffness-proportional damping produces a larger response.

Period and Ductility Range: The effect of mass-proportional damping on maximum response is
pronounced, and the mass-proportional damping was found useful to exaggerate the damping effect
and to clarify a general trend of the damping effect of maximum response. Therefore, the
mass-proportional damping is used to study the variation of maximum response with the amount of
damping.

Single-degree-of-freedom (SDF)
systems with the Takeda hysteresis
model (unloading stiffness degrading
parameter α = 0.0) were subjected
to Taft (N21E) 1952 earthquake
motion. The yielding period was
selected to be 0.14 sec and 1.13 sec.
The yield resistance level was varied
to control the maximum ductility
response and to study the effect of
damping at different ductility ranges.
The initial damping factor was varied
from zero to 20 percent of the
critical.

The maximum response is Effect of damping with yield level and period
reduced significantly with increasing
damping amplitude in short-period
systems (Ty = 0.14 sec), but is not so
much affected in long-period systems
(Ty = 1.13 sec). The general trend of
decreasing response amplitude with logarithmically increasing damping amplitude is observed for
both high and low ductility ranges.

Hysteretic Energy Dissipation Capacity: Some models have large hysteretic energy dissipation

8
capacity, and others have small capacity. The latter model can dissipate kinetic energy only through
viscous damping, hence its response amplitude is likely affected by the amount of viscous damping.

The effect of damping on the response amplitude of single-degree-of-freedom systems with a

Takeda hysteresis model (unloading stiffness degrading parameter α = 0.0) and a Peak-oriented
hysteresis model is studied under the Taft (N21E) 1952 earthquake motion. The yield period of the
systems was selected to be 0.14 sec and 1.13 sec.

The response point of the Peak-oriented hysteresis model moves toward a previous maximum
response point in the loading direction, and the model behavior is linearly elastic between the positive
and negative maximum response points without any hysteretic energy dissipation. Once the
response point reaches the previous maximum response point, it moves on the primary curve.

The maximum response of the Peak-oriented model at a yielding period of 0.14 sec was too large,
and the yield resistance was increased by 50 percent from the standard value to reduce the response
amplitude.

Note that the amount of damping has a larger influence on the response ductility of the
Peak-oriented models, especially in the short-period systems. The difference in ductility of the two
models was relatively small in the long-period systems.

It should be noted that the viscous damping dissipated energy even at a small amplitude
oscillation as long as there exists velocity. Hysteresis energy is dissipated at a large amplitude
oscillation beyond yielding. Therefore, the viscous damper is more effective in reducing response
amplitudes.

9
 Chapter 13. Response of Different Hysteresis Models

The effect of different stiffness parameters on earthquake response of single-degree-of-freedom

systems is studied. By choosing stiffness properties and hysteretic energy dissipation capacity as
similar as possible, the effect of different hysteresis models on earthquake response waveforms and
amplitudes was investigated. Hysteresis models simulating flexural behavior of the reinforced
concrete were used.

13.1 Analysis Method

Common force-deflection properties were specified to all the models so that the influence of
different model hysteretic behavior would be clarified. However, some models have a tri-linear
skeleton curve under monotonically increasing load, and others have a bilinear skeleton curve.
Consequently, the yield point and the post-yielding stiffness were chosen common among the
models, and the cracking point was added to the models with a tri-linear skeleton curve. Hysteresis
models used are Degrading bilinear model, Ramberg-Osgood model, Clough model, Bilinear Takeda
model, Takeda model, and Degrading trilinear model.

The mass of an SDF system was arbitrarily chosen to be 1.0 ton since the overturning effect
( P − ∆ effect) due to the mass' side sway was not included in the analysis. A series of hysteretic
models were designed with "yield period" (periods related to the secant stiffness at the yield point) of
0.10, 0.14, 0.20, 0.28, 0.40, 0.57, 0.80, 1.13 and 1.60 sec.

Earthquake response amplitudes are known to

vary with the system's period and the yield level.
To make the comparative study easy, SDF
systems with different periods were desired to
produce comparable ductility ratios (attained
maximum displacement divided by the yield
value). Hence, the Newmark's design criteria
(Veletsos and Newmark, 1960) were adopted.
Namely, the yield resistance of an SDF system
was determined by:
(a) dividing the maximum elastic inertia force
by the allowable ductility factor, µ , of the system
if the system's period is greater than 0.5 sec; and
(b) dividing the maximum elastic inertia force
by a factor 2 µ − 1 if the system's period less
than 0.5 sec.

The stiffness of an elastic system was made

equal to the "yielding stiffness," Ky (the secant
Determination of stiffness properties
stiffness at the yield point). The allowable ductility
factor of a nonlinear system was arbitrarily
assumed to be 4.0. In this manner, the yield resistance was different for different periods and
different earthquake motions.

The post-yielding stiffness, Ku, was assumed to be 10 percent of the yielding stiffness; the
uncracked stiffness, Kc, to be 2.0 times the yield stiffness; and the cracking resistance to be
one-third the yield resistance. These assumptions were used to approximate the stiffness properties
of a reinforced concrete structure.

The parameters of the Ramberg-Osgood model were chosen so that the resistance at the
allowable ductility should be the same as the other models; i.e., η =1.0 and γ =3.79.

Hysteresis energy dissipation indices of different models are calculated at a ductility factor of 4.0
1
and compared below:
Clough Model ( α = 0.5): Eh = 0.11
Degrading Tri-linear Model: Eh = 0.11
Takeda Model ( α = 0.5): Eh = 0.14
Degrading Bilinear Model ( α = 0.5): Eh = 0.19

Clough Model ( α = 0.0): Eh = 0.21

Takeda Model ( α = 0.0): Eh = 0.23
Ramberg-Osgood Model( η =1.0; γ = 3.79): Eh = 0.28
Bilinear Model ( α = 0.0): Eh = 0.33
where, α : unloading stiffness degradation index. Note that a large discrepancy exists among the
models in the capacity to dissipate hysteretic energy under a steady-state condition. the unloading
stiffness degradation prameter α has an appreciable effect on the value of hysteretic energy
dissipation index.

Damping: Viscous damping was assumed; the damping coefficient was assumed to be proportional
to varying instantaneous stiffness. The damping factor was 5 percent of the critical at the initial
elastic stage.

The damping is assumed to be proportional to instantaneous stiffness, where the maximum

response of an SDF system is not so sensitive to the amplitude of initial damping factor. Therefore,
an initial elastic damping factor of 0.05 is used for a system with bilinear primary curve, and 0.0707
for a system with tri-linear primary curve. In this manner, the damping factor of all the system is
made identical at the yielding period because the pre-cracking stiffness of a tri-linear primary curve
is chosen twice the yielding stiffness.

Earthquake Motions: Four

earthquake accelerograms from two
California earthquakes were used in
this study: the NS and EW
components of the 1940 El Centro
record and the N21E and S69E
components of the 1952 Taft record,
digitized at the University of Illinois
at Urbana-Champaign (Amin and
Ang, 1966).

Linearly elastic response spectra

of these four records were studied
using the entire duration and the
first 15 sec part of the records. The
damped spectra were almost
identical for a period range less
than 2.0 sec using either the entire
duration or the first 15 sec part
except for the El Centro (EW)
record. Consequently, the response
computation was terminated
approximately at 15 sec when the
Taft (N21E and S69E) and El
Centro (NS) records were used. On
the other hand, the maximum Earthquake accelerograms
response of some linearly elastic
systems under the El Centro (EW)
motion occurred after 15 sec. Therefore, the entire 30 sec record of the El Centro (EW) motion was

2
used.

The response spectra of the four motions are shown below:

Numerical Method: The equation of motion was solved numerically using the Newmark- β method
(Newmark, 1959) with β =1/6 and γ =1/2. Both the equation of motion and the displacement -
velocity - acceleration relations were satisfied only at the discrete time step using an iterative
procedure. In other words, the "overshooting" of the hysteresis curve was adjusted within the time
step.

A constant time increment of the numerical integration was taken either as one-twentieth the
initial elastic period or 0.02 sec, whichever was shorter. The former was necessary to faithfully trace
the hysteresis curve rather than numerical stability requirements. The latter criterion became
necessary because the earthquake accelerograms were given at a 0.02 sec interval.

3
13.2 Effect of Initial Stiffness (Takeda Model)

The initial stiffness was arbitrarily chosen in this

paper to be 2.0 times the yielding stiffness. For a
normal reinforced concrete member, the ratio of initial
to the yielding stiffness may vary from 1.5 to 4.0.

Fy

ky

Fc

Maximum response of the Takeda models is compared by varying the stiffness ratio, keeping a
cracking-to-yielding resistance ratio to be one-third and secant yield stiffness the same. When an
attained ductility is greater than 4.0, the effect of initial stiffness is minimal for both short-period and
intermediate-period systems. It is expected, however, that the initial stiffness should influence
maximum response amplitude if an attained ductility is less than or around unity.

13.3 Effect of Cracking Force Level (Takeda Model)

The effect of cracking force levels on

maximum response amplitudes of the Takeda
model is studied, keeping the initial stiffness to
be 2.0 times the yielding stiffness. Little effect is
observed when an attained ductility is greater
than 4.0 even for a short-period system.

Fy

ky

Fc

13.4 Effect of Yield Resistance Level (Takeda Model)

The level of yield resistance is expected to be one major factor to influence maximum response
amplitudes. The yield resistance was varied from the standard value. As yield strength increases, an
attained ductility factor is significantly reduced, especially for a system with a short yielding period;
the required ductility was reduced to one half due to a 30 percent increase in the yield strength.

4
 It is important to note that the value of
yield displacement increased proportional to
the level of yield resistance. Consequently,
the maximum response amplitude did not
decrease with the level of yield resistance so
much as the attained ductility did, although
the 0.14 sec period system showed a rapid
increase in the displacement amplitude with
the reduction in yield strength. The 1.13 sec
system also showed an increase in the
maximum displacement with decreasing
yield strength, but reached a peak at the
standard resistance, and then gradually
decreased its maximum displacement
amplitude.

The level of yield resistance has a

significant effect on maximum response
amplitude, especially in a short-period
range.

Response waveforms of Takeda models

( α =0.0) with different yield resistances are F

compared. The yield resistances are 0.70, 0.60,

Fy
0.47 times the standard yield strength at a 0.4
sec yield period under El Centro (NS) 1940
motion. The maximum amplitude is largest for
the weakest system. However, the peak
amplitudes in the positive direction were largest Fc
for the strongest system since the lower
D
strength system produced a large residual
displacement in the negative direction. This Dy Dm
observation is not necessarily true for a general Maximum displacement and ductility factor
weak-strength system, but it is related to the
unloading stiffness degradation parameter. A low value of the parameter, for example α =0.0, tends
to cause a large residual displacement with little elastic recovery during unloading.

5
13.5 Effect of Post-Yielding Stiffness (Takeda Model)

The strain hardening of reinforcing

bars will give a finite positive stiffness
after the flexural yielding. Very small
post-yielding stiffness has been routinely
used in Japan. The standard model in
this paper assumes a 10 percent of the
yielding stiffness as the post-yielding
stiffness.

Maximum response of Takeda models

is compared varying post-yield stiffness.
Maximum response decreases with an
increasing post-yielding stiffness,
remarkably in a short-period system, and
insignificantly in a long-period system.
The response amplitude changes more
with post-yielding stiffness when the
post-yielding stiffness is 0.05 to 0.20
times the yield stiffness. For a high
post-yielding stiffness, less inelastic
displacement is required to store a given
magnitude of strain energy. Effect of post-yielding stiffness

6
13.6 Effect of Unloading Stiffness Degradation Parameters (Takeda Model)

Some models use an unloading stiffness degradation parameter, which controls the fatness of a
hysteresis loop and also the plastic residual deformation. It is not possible to determine the value of
this parameter from the material and geometrical properties of a reinforced concrete structure.
Normal range of this parameter is 0.0 to 0.5, and a value of 0.4 has been often used for the
reinforced concrete.

The effect of the value of the

unloading stiffness degradation
parameter on maximum response
of Takeda models is studied.
Maximum response increases with
an increasing value of the
parameter, and this tendency is
remarkable for shorter period
systems. The system's capacity
(either through damping or through
hysteresis) to dissipate kinetic
energy has a conspicuous
influence on the maximum
response of a short period
structure. The same tendency is
observed when the yielding period
of systems was varied from 0.1 to
1.6 sec; the effect becomes small
for a system of yielding period
greater than 0.4 sec.

Response waveforms of Takeda

models under El Centro (NS) 1940
motion are compared for the
yielding period of 0.4 sec. The yield level was chosen to be 0.6 times that of the standard model. For
7
a large value of the parameter, peak amplitudes are larger both in the positive and negative
directions, having comparable amplitudes in the two directions. For smaller values of the parameter,
the system tends to produce large amplitude only in one direction. This is clearly observed in the
hysteretic curve. Peak-to-peak stiffness in a low amplitude oscillation is lower for a system with a
larger parameter, causing a long period of oscillation from approximately 6.0 sec.

The effect of the unloading stiffness degradation parameter is significant on response amplitude,
response waveform, residual displacement and hysteresis shape.

8
13.7 Effect of Hysteresis Energy Dissipation

The hysteretic energy dissipation capacity of a Degrading

Trilinear model is known to be sensitive to the choice of a
cracking point relative to the yielding point. The effect of
cracking force level on maximum response amplitude of
Degrading Trilinear models is studied. An attained ductility
factor decreases with an increasing cracking force level,
especially in a short-period system. This is another example to
show that maximum response amplitude of a short-period
system is significantly influenced by the capacity to dissipate
kinetic energy through either damping or hysteresis.

Effect of energy dissipation

The response of two degrading tri-linear models is shown above. The yielding period is 0.4 sec. The
two systems were subjected to the 1940 El Centro (NS) motion. A significant difference in hysteresis
shape can be observed. The response up to the first large oscillation is similar, but difference started
to be apparent in subsequent response cycles in the response waveform.

9
13.8 Effect of Parameter of Ramberg-Osgood Model

The parameter γ of the Ramberg-Osgood model influences the hysteresis shape. When the
parameter is small, the hysteresis area becomes small, but
post-yielding stiffness is high. When the parameter increases,
the hysteresis shape becomes similar to that of the
elasto-plastic model.

The response of short- and long-period systems is compared

with the parameter γ . The short-period system increased the
ductility demand for increasing value of the parameter.

The response waveforms and hysteresis relations of systems

are compared using different parameters γ . The yielding
period of the systems was 0.4 sec, and the 1940 El Centro (NS)
motion was used. The response waveforms are similar up to 2.0
sec from the beginning of the motion. The system with using the
smallest parameter exhibited the largest amplitude response.
The residual displacement increased with the value of the
parameter.

10
13.9 Response to Different Earthquake Motions

The each model was subjected to four different earthquake motions. The maximum response
ductility demand is compared with respect to yielding periods.

(a) Degrading Bilinear Model ( α = 0.0, 0.5 )

11
12
13.10 Response of Different Models

Four earthquake records are used in this study; i.e., El Centro 1940 (NS) and (EW), and Taft 1952
(N21E) and (S69E). Maximum response of six different hysteresis models is compared using the
standard stiffness properties. Hysteresis models are (a) Ramberg-Osgood model ( γ =3.79), (b)
Degrading Bilinear model ( α = 0.0 and 0.5), (c) Clough model ( α = 0.0 and 0.5), (d) Bilinear Takeda
model ( α = 0.0 and 0.5), (e) Takeda model ( α = 0.0 and 0.5), and (f) Degrading Trilinear model
(Fc/Fy = 1/3).

An attained ductility factor is defined as the ratio of the maximum displacement to the yield
displacement. For a design procedure to be conservative, the attained ductility factor should be less
than the allowable ductility factor of 4.0. Note that the Newmark's design criteria give a reasonable
ductility demand from all six hysteresis models for a wide range of yielding periods in the case of El
Centro (NS) 1940 motion. The undamped yield period is a period associated with secant stiffness at
the yield point. The initial uncracked period of an SDF system with a trilinear primary curve is
approximately 70 percent of the yield period.

Although the Newmark's design criteria appear to be acceptable for the El Centro (NS) 1940
motion, the other three earthquake motions caused attained ductility factors much greater than the
allowable value at the various periods. In general, the design criteria are not satisfactory in a very
short-period range, for example less than 0.15 sec. Distribution of maximum response with periods
is different from one earthquake motion to another, showing an irregular shape, although each
hysteresis models was designed on the basis of elastic response of individual earthquake motion.
On the other hand, distribution of maximum response with periods is similar from one hysteresis
model to another for a given earthquake motion, implying that maximum response amplitudes of
different hysteresis models can be made comparable if hysteresis parameters of each model are
properly adjusted.

13
 For an unloading stiffness degradation parameter of 0.5, the Takeda, Clough and Degrading
Bilinear models developed comparable ductility factors. The Degrading Trilinear model also
developed ductility factors similar in magnitude to those three models at corresponding periods.

Therefore, maximum response amplitudes are not as sensitive to detail difference in hysteretic
rules of these models, but rather are influenced by more basic characteristics of hysteresis loops,
such as stiffness properties to define a primary curve and the fatness (hysteretic energy dissipating
capacity) of a hysteresis loop.

14
13.11 Response Waveforms and Hysteresis Relations

Resistance response normally oscillates about its neutral axis, and its amplitude is limited by the
yield resistance. On the other hand, displacement response does not necessarily oscillate about the
neutral axis, but the residual displacement amplitude is easily shifted by the properties of a
hysteresis model. Therefore, it is easy to study the effect of different hysteretic properties in a
displacement response waveform.

The El Centro (NS) 1940 motion was used for response computation. Five hysteretic models
were used for comparison; i.e., (a) Degrading Bilinear model ( α = 0.0 and 0.5), (b)
Ramberg-Osgood model ( γ = 3.79), (c) Clough model ( α = 0.0 and 0.5), (d) Takeda model ( α = 0.0
and 0.5), and (e) Degrading Trilinear model (Fc/Fy= 1/3). The yielding period of these models was
arbitrarily chosen to be 0.4 sec, and the yield resistance level was taken to be 60 percent of the
standard model to allow a larger inelastic action.

The response amplitude was shown to be influenced by the fatness of a hysteresis loop even if
the stiffness properties of the primary curve are identical. Consequently, the response waveforms
are compared among systems having a relatively fat hysteresis loop and among those having a
relatively thin hysteresis loop.

In all fat-hysteresis systems, maximum displacement at 2.0 sec, and the second largest
amplitude at around 5.3 sec. The Takeda model shows a short-period oscillation at 1.0 sec, since
only the Takeda model has a trilinear primary curve among the models shown in this figure. The
Bilinear model oscillates in a period shorter than the other models, e.g., between 2.5 to 4.5 sec,
attributable to the non-degradation of stiffness with displacement amplitude. The Bilinear and
Ramberg-Osgood models developed residual displacement in the negative direction at 7.0 sec,
whereas the Clough and Takeda models developed positive residual displacement. The former two
models behaved in a manner different from the Clough and Takeda models. The Ramberg-Osgood,
Clough and Takeda models show similar hysteresis relations.

Thin hysteresis models show displacement response waveforms distinctly different from those of
fat-hysteresis models, oscillating regularly in larger amplitudes and in longer periods. The Degrading
Bilinear model exhibited a behavior different from the other models, especially in a waveform
between 6.5 and 8.0 sec. The Clough, Takeda and Degrading Trilinear models produced similar
displacement waveforms.

The Clough and Takeda models developed similar hysteretic relations although the Takeda
model had a trilinear primary curve. This may be attributable to the fact that a large-amplitude
oscillation occurred at an early stage of the earthquake motion. In other words, the behavior of
Takeda and Clough models can be different if a small oscillation continues for a long duration, or if
the yielding does not occur during an earthquake.

Therefore, the Takeda model is more preferable to the Clough model, although the former model
requires a larger memory in a computer to store the complicated hysteresis rules.

A hysteresis loop of the Degrading Trilinear model appears to be thinner than the Takeda model,
but the Degrading Trilinear model can dissipate larger hysteretic energy during medium-amplitude
oscillation.

The comparison of response waveforms of different hysteresis models points out the less
sensitive nature of response waveforms to a minor difference in hysteresis rules, as long as the
same primary curve is used in conjunction with a comparable capacity to dissipate hysteretic energy.
If maximum response amplitude is known, before analysis, to be much larger than the yield
displacement, the Clough model can produce a response waveform similar to that of the complicated
Takeda model. However, if that premise is not guaranteed, it is more conservative to use a
hysteresis model with a trilinear primary curve in the analysis of the reinforced concrete, recognizing
the stiffness changes at cracking and yielding; i.e., the Takeda model.

15
References:

Amin, M., and A. H.-S. Ang, "A Nonstationary Model for Strong Motion Earthquakes," Structural
Research Series No. 306, Civil Engineering Study, University of Illinois, Urbana, 1966.
Otani, S, "Hysteresis Models of Reinforced Concrete for Earthquake Response Analysis," Journal,
Faculty of Engineering, University of Tokyo, Vol. XXXVI, No. 2, 1981, pp. 125-156.
Newmark, N. M., "A Method of Computation for Structural Dynamics," Journal, Engineering
Mechanics Division, ASCE, Vol. 85, No. EM3, 1959, pp. 67-94.
Veletsos, A. S., and N.M. Newmark, "Effect of Inelastic Behavior on the Response of Simple
Systems to Earthquake Motions," Proceedings, Second World Conference on Earthquake
Engineering, 1960, Vol. II, pp. 895-912.

16
17
18
13.12 Effect of Hysteresis Shape on Frame Response

The bond deterioration along the beam longitudinal reinforcement within a beam-column joint is
normally thought to be undesirable because the energy dissipation at beam ends is reduced by
pinching in the hysteresis shape; the decay in energy dissipation capacity might increase the
response of a structure during an earthquake. However, it is not practical to expect perfect bond
along the beam longitudinal reinforcement.

The acceptable level of bond deterioration cannot be determined by the laboratory test of a
beam-column sub-assemblage if the problem is to evaluate how much response may be increased
by the decay in the energy dissipation characteristics. The influence of the energy dissipation on the
earthquake response well into an inelastic range is studied to re-examine the acceptable level of
bond deterioration (Kitayama, 1993).

Four-story, seven-story and sixteen-story reinforced concrete moment resisting frames were
analyzed with uniform 6.0 m bays and uniform story height of 3.5 m. In the analysis, a single
continuous column with connecting girders on the both sides was removed by cutting the connecting
girders at the mid-span, and the girder ends were supported by pin-horizontal rollers. The yield was
allowed at the girder ends, but the columns were assumed to remain elastic during an earthquake.
The mass of each floor was estimated on the basis of member dimensions and floor slab thickness
(= 130 mm). The fundamental periods of oscillation were 0.36 sec, 0.62 sec and 0.98 sec for 4-, 7-
and 16-story buildings.

Each member was represented by the one-component model, in which inelastic deformation
(rotation) was assumed to concentrate at member ends; a beam-column connection was assumed
to be rigid.

The Takeda-slip hysteresis model (Kabeyasawa et al., 1983) at beam ends was selected to
simulate the pinching behavior caused by the bond deterioration along the beam reinforcement. As a
reference, the Takeda model (Takeda, Sozen and Nielsen, 1970) was used to simulate a good bond
situation with a spindle-shape hysteresis. The primary curves of the both models were made
identical. Additional deformation caused by the pull-out of reinforcement from the connection was not
considered. The force-deformation relation of the hysteresis models are compared with the
hysteresis relations of interior beam-column sub-assemblages observed in the laboratory test.

Instantaneous viscous damping matrix was assumed to be proportional to instantaneous stiffness

matrix, and the initial elastic damping factor for the first mode was chosen to be 0.05 of the critical.

The 1940 El Centro (NS) and the 1952 Taft (S69E) records were used in the analysis. The
intensity of ground motion was adjusted to develop maximum member ductility factors of
approximately 4.0 at the beam ends using the Takeda hysteresis model.
19
 Time histories of displacement response at the roof-level are compared. Although the
displacement response waveforms of the Takeda-slip model deviated from those of the Takeda
model, attained maximum response amplitudes are comparable. Large drift was noted more
frequently in the response waveforms of the Takeda-slip models.

The attained story drift angles are compared. The maximum beam ductility factor of 4.0 was
adopted to determine the intensity of ground motion. The maximum drift angle was smaller than 1/50
rad in the three structures with the Takeda model. The story drift angle increased slightly with the
use of the Takeda-slip model, by dissipating less kinetic energy; the story drift angle exceeded 1/50
rad in the 16-story building using the Takeda-slip model with small hysteresis energy dissipation
capacity.

The attained ductility factors at beam ends are compared. The distribution of beam-end ductility
factors of a structure with the Takeda model is similar to that with the Takeda-slip model (equivalent
damping factor he = 0.15). The change in the energy dissipating capacity in terms of equivalent
viscous damping factors did not affect the ductility demand appreciably.

20
21
22
References:

Giberson, M. F., "The Response of Nonlinear Multi-story Structures subjected to Earthquake

Excitation," EERL Report, Earthquake Engineering Research Laboratory, California Institute of
Technology, Pasadena, 1967.
Kabeyasawa, T., H. Shiohara, S. Otani and H. Aoyama, "Analysis of the Full-scale Seven-story
Reinforced Concrete Test Structure," Journal, Faculty of Engineering, University of Tokyo (B), Vol.
XXXVII, No. 2, 1983, pp. 432-478.
Kitayama, K., "Limitation of Beam Bar Bond Deterioration within Beam-column Joint," Earthquake
Resistance of Reinforced Concrete Structures, A Volume Honoring Hiroyuki Aoyama, November
1993, pp. 297 - 306.
Takeda, T., M. A. Sozen and N. N. Nielsen, "Reinforced Concrete Response to Simulated
Earthquakes," Journal, Structural Division, ASCE, Vol. 96, No. ST12, 1970, pp. 2557-2573.

23
 Chapter 14. Reliability of Analysis Methods
14.1 Introduction

Various dynamic tests have been carried out to understand the dynamic behavior of a structure
and to test the reliability of analysis methods. The methods of dynamic testing may be classified by
(a) tests on a real structure or a model structure, (b) behavior within an elastic range or in an inelastic
range, (c) excitation by harmonic or random function, and (d) excitation at specific points within a
model or at the base.

A real structure is normally tested in an elastic range to avoid any damage to structural as well as
non-structural elements; the structure is excited by a harmonic exciter to generate steady state
oscillation. The vibration characteristics such as natural periods, damping factors and mode shapes
are normally studied. The amplitudes of oscillation were measured at different point during the
steady-state response at resonance frequency. The state-of-the-art in testing technique can define
the soil-structure interaction behavior or in-plane and out-of-plane vibration mode shape of floor slabs
(Foutch, 1976).

An old building preceding demolition may be tested in an inelastic range under harmonic excitation.
The behavior of the structure to failure in such a test is significantly different from the behavior during
an earthquake because the structure may fail in low-cycle fatigue mode under steady-state excitation.

Small-scale model specimens are normally used to study the failing behavior due to the limitation
in the capacity of testing facilities or in the research funds. To study the behavior during an
earthquake, specimens may be tested on an earthquake simulator.

In testing a small scale test, the similitude laws must be carefully studied. The basic dimensions
for physical problem are force (mass), length and time. The relations between a proto-type and model
may be expressed as
Fm = a1 F p
Lm = a 2 L p
Tm = a 3T p
Once the three scale factors a1, a2 and a3 are selected, then other properties must be determined;
e.g.,
Fm a1
Stress: σm = = σp
L2m a 22
F a
Young’s modulus: E m = 2m = 12 E p
Lm a2
1
 Lm a 2
Velocity: v m = = vp
Tm a 3
L a
Acceleration: a m = m2 = 22 a p
Tm a3
Material properties (strength and stiffness) must satisfy the similitude laws although every property
cannot be satisfied by a scaled model. Note that the gravity acceleration must be the same in the
prototype and model environments.

When a small-scale test specimen is used in a dynamic test, the scale effect should be carefully
studied. The behavior of a specimen is sometimes influenced by its size; especially in shear failure
and bond failure in the reinforced concrete.

The pseudo-dynamic (on-line) testing method is also used in the laboratory, in which the response
of a specimen under given earthquake motion is calculated by a computer on the basis of the
observed resistance of the specimen; explicit numerical integration technique is used to determine
the displacement response at the next time step.

The test on an earthquake simulator may be classified into two types; (a) a proof test of a
particular structure against design earthquake excitation, in which the test specimen must represent
necessary stiffness and dynamic characteristics of the proto-type structure and the excitation motion
must be carefully selected considering the soil properties of the construction site, and (b) a behavioral
test of a general structure, the test results of which may be used to understand the general behavior
and to test the reliability of an analytical method. The specimen for the behavioral test should
represent a mathematical model of a general structure rather than an actual structure. The difference
in the two types of testing lies in the design of a specimen and selection of loading function, but the
testing technique is the same.

Reference:

1. Foutch, D. A., "A Study of the Vibrational Characteristics of Two Multistory Buildings," EERL 76-03,
Earthquake Engineering Research Laboratory, California Institute of Technology, September
1976.

2
14.2 Reinforced Concrete Column

As a simple model representing the

characteristics of the reinforced concrete
structure, a series of reinforced concrete
columns were tested under uni-directional base
excitation on the University of Illinois Earthquake
Simulator (Takeda, Sozen and Nielsen, 1970).

The dimensions of a specimen were 152 x

152 mm in section, reinforced by 4-No.4 bars
(129 mm2/bar, yield stress of 351 MPa). Shear Wire
reinforcement was diameter 4.8 mm plain wire Section
(yield stress of 276 MPa). Concrete strength was
30.3 MPa in compression and 2.52 MPa in
tension. A heavy steel weight (8.97 kN) was
attached at the top of the column through a
mechanical hinge so that rotational inertia of the Unit
mass would not affect the response of the
specimen.

The lateral force resistance-deformation at the top of the specimen was calculated to define the
primary curve. Cracking moment was calculated for modulus of rupture concrete using the flexural
theory. The cracking deformation was calculated for the elastic stiffness and cracking moment. Yield
moment was calculated using the parabolic stress-strain relation of concrete and the elasto-plastic
stress-strain of reinforcement. The yield deflection was calculated as the sum of (a) deflection caused
by curvature based on cracked section, (b) deflection caused by slip of the reinforcement (assuming
uniform anchorage bond over 20 bar diameter) and depression of semi-infinite plate under flexural
compression stress at the beam-column interface, (c) deflection caused by deformation of the test
platform (observed in a static test), and (d) the shearing deflection. The deflection was dominated by
parts (a) and (b) above.

Cycle 1 Cycle 4
Load, kips

Load, kips

Displ., inch Displ., inch

kips

Cycle 8
Cycle 6
Load
Load, kips

Displ., inch

Displ., inch Observed

Calculated

3
 The force-deflection relation obtained by a static test is compared with the hysteresis relation of
the Takeda model with the calculated primary curve. The model was subjected to the observed
deflection at the top. The overall behavior of the specimen and the model is generally similar.

In the earthquake simulator test, the time axis of earthquake motion was compressed to
one-eighth to satisfy the similitude relations (equal velocity, equal stress and scaled length) and the
capacity of the earthquake simulator. El Centro (NS) 1940 motion was simulated on the table. The
observed motion on the earthquake simulator table was used in the simulation analysis. Damping
factor was assumed to be zero and 2 percent of the critical; the damping coefficient was assumed to
remain constant (mass proportional damping).

The correlation was found more favorable with damping to 3.0 sec from the beginning of the
motion; however, the correlation was better without damping at around 3.5 sec probably because the
constant damping coefficient tends to dissipate more energy after the deterioration of stiffness with
damage.

The study showed that the analysis could simulate a complex behavior of the reinforced concrete
member in an inelastic range if the proper hysteresis model were to be used which was capable of
simulating static behavior under load reversals.

(a) Base acceleration

(b) Observed acceleration

Acceleration, G

(c) Calculated acceleration (h=0.00)

(d) Calculated acceleration (h=0.02)

Time, sec

Reference

Takeda, T., M. A. Sozen, and N. N. Nielsen, "Reinforced Concrete Response to Simulated

Earthquakes," Journal, Structural Division, ASCE, Vol. 96, No. ST12, December 1970, pp. 2557 -
2573.

4
14.3 Frame Structures

Three-story Shear Model: A simple multi-story

building model is a frame with rigid girders, in
which all the deformation takes place only in the
columns. The stiffness of each story can be Weight
defined without coupling with stiffness of the other Weight
stories.

Eto, Takeda and Omote (1973) tested a

one-fifth scale three-story one-bay frame model Weight
Weight
with rigid girders on the Ohbayashi-gumi
earthquake simulator. Another specimen, designed
and constructed using identical specifications, was
tested under static lateral force reversals.
Weight
Weight
The resistance-deformation relation of each
column under monotonically increasing force was
evaluated taking into account flexural and shear
deformation and deformation due to pull-out of
longitudinal reinforcement at column ends.

The Takeda model (Takeda, Sozen and Nielsen, 1970) was used to simulate the hysteresis
relation under lateral force reversals. The unloading stiffness degradation index of the model was
determined on the basis of the static test. The story shear-inter-story deformation relations calculated
and observed in the static test agreed reasonably well.

Static test result

Calculated force-deformation relation

The damping matrix was assumed to be proportional to initial stiffness, and the damping factor at
the initial elastic stage was determined to be 0.03 for the first mode on the basis of the free vibration
test conducted prior to the dynamic test.

The calculated and observed acceleration response waveforms are compared for test runs R2 and
R3. The columns in the three stories yielded in test run R3, but the analysis indicated no yielding in
the top story columns. The analysis could favorably simulate large amplitude oscillations.

5
 (a) Third story acceleration (a) Third story acceleration
Acceleration, G

Acceleration, G
(b) Second story accleration (b) Second story acceleration

(c) First story accleration (c) First story acceleration

(d) Ground acceleration (d) Ground acceleration

Time, sec Time, sec

(a) Third story accelration

(a) Third story acceleration
Acceleration, G

Acceleration, G

(b) Second story accelration

(b) Second story acceleration

(c) First story acceleration (c) First story acceleration

(d) Ground acceleration (d) Ground acceleration

Time, sec Time, sec

6
Three-story One-bay Frame: A series of three-story one-bay plane frame structures were tested
uni-directional base motion on the University of Illinois earthquake simulator (Otani and Sozen, 1972).

Specimens were approximately one-sixth scale (=1/2.52) of an imaginary prototype structure; two
parallel frames were tested to attain the stability in the orthogonal direction. Rigid steel weight was
attached at beam end through mechanical hinges, modeling rigid floor diaphragm.

The dimensions and reinforcement are shown below;

Wire

Wire

Wire

Beam Section

Wire

Column section Unit

As the similitude relations, the followings were chosen as the basic relations;
(a) Linear scale: Lm = (1/6.25) Lp
(b) Stress: σ m = σ p
(c) Acceleration: am = ap
where subscript m stands for model and p for prototype, and L: length, σ : stress, a: acceleration.
Linear scale was necessary to test a scaled model. Equal stress became necessary because the
materials used in the specimen were the same or similar to those used in a prototype structures;
strengths and elastic moduli of concrete and steel should be modeled. Equal acceleration was
adopted because the gravity acceleration could not be changed.

Three independent similitude relations can be selected in a dynamic model test. Some
researchers used equal velocity, instead of equal acceleration, as a criterion

The relations for forces and times can be derived from the chosen set of similitude relations; i.e.,
Fm F p
= : equal stress
L2m L2p
Lm L p
= : equal acceleration
Tm2 T p2

7
 and
2
L 
Fm =  m  Fp
L 
 p 
Lp
Tm = Tp
L
Therefore, the time axis of an earthquake record is compressed by 2.5 in the test.

In an earthquake simulator test, a specimen is normally subjected to a series of excitations of

increasing magnitude. The excitation at a preceding test run may affect the response of the specimen
in the following test run.

The response waveforms of two specimens are compared under the same base motion; one
specimen (F1) was previously subjected to three test runs, doubling the magnitude after each test run,
and the other specimen (F2) was virgin at the test run.

The acceleration and displacement response waveforms were observed to be similar in the two
specimens. From this observation, the each test run of a reinforced concrete specimen may be
considered independent as long as (a) the behavior is governed by flexure without decay in
resistance and (b) large amplitude response occurs at the beginning of the motion.
Acceleration, G
Displacement, inch

(a) Observed in Test F1-3

(a) Observed in Test F1-3

(b) Observed in Test F2-1

(b) Observed in Test F2-1

Time, sec
Time, sec

The one-component model (Giberson, 1967) was used to represent the distribution of stiffness in
each member; the beam-column connection was assumed to be rigid. The moment-rotation relation
under monotonically increasing force at a member end was calculated for the anti-symmetric bending
moment distribution on the basis of observed material properties of the specimen. The Takeda
hysteresis model (Takeda, Sozen and Nielsen, 1970) was used to represent the force-deformation
relation under load reversals.

The deformation at a member end due to the bar slip of longitudinal reinforcement within a
beam-column connection was considered in the analysis; i.e., the elastic deformation of a tensile
longitudinal bar under uniform bond stress was calculated as the pull-out deformation, and the center
of rotation was assumed at the compressive reinforcement at the member end. The primary curve of
the moment-bar slip rotation relation was simplified to a bilinear relation and the Bilinear Takeda
model (Otani and Sozen, 1972) was used for the hysteresis relationship. However, the slip type
behavior of bar slip was not represented by the model.

The damping properties could not be determined by the material and geometrical properties of a
specimen. Therefore, two types of damping matrix were studied (Otani and Sozen, 1972); i.e., (a) a

8
damping matrix proportional to the constant mass matrix, and (b) a damping matrix proportional to the
instantaneous stiffness. The damping factor for the first mode at the initial elastic stage was selected
to be 0.02 or 0.10.

The test specimen was subjected to an intense base motion causing a first-story drift angle of 1/20
rad. Two response waveforms were studied; (a) the top-floor (relative) displacement response
governed by the fundamental mode of oscillation, and (b) the second-floor (absolute) acceleration
response with participation of higher modes.

(a) Observed displacement

(b) Calculated displacement

(stiffness proportional damping, h1=0.02)
Third story displacement, inch

(c) Calculated displacement,

(stiffness proportional damping, h1=0.10)

(d) Calculated displacement,

(mass proportional damping, h1=0.02)

(e) Calculated displacement

(mass proportional damping, h1=0.10)

Time, sec

The displacement response of the model using the instantaneous stiffness proportional damping
was not affected by the choice of the initial damping factor, but the higher frequency components in
the acceleration response was suppressed by the use of a larger initial damping factor; this
observation is consistent with the properties of elastic response using the stiffness proportional
damping.
9
 The displacement amplitude was reduced by the use of large initial damping factor in the mass
proportional damping, but the value of initial damping factor did not influence the acceleration
response waveforms. The mass proportional damping has significant influence on the fundamental
mode response.

(a) Observed acceleration (Test F2-1)

(b) Calculated acceleration

(stiffness proportional damping, h1=0.02)
First story acceleration, G

(c) Calculated acceleration

(stiffness proportional damping, h1=0.10)

(d) Calculated acceleration

(mass proportional damping, h1=0.02)

(e) Calculated acceleration

(mass proportional damping, h1=0.10)

Time, sec

From the comparison of observed and calculated response waveforms, the mass proportional
damping must be small to better simulate the displacement response waveform, and must be large to
simulate the acceleration response waveform. The mass proportional damping may not be suited for
the simulation of the response of the particular specimen.
10
 If damping proportional to instantaneous stiffness was assumed, the calculated displacement
response waveforms for the two values of initial damping factors agreed reasonably well, while the
calculated acceleration amplitude was too large for the initial damping factor of 0.02, and too small for
the initial damping factor of 0.10. Therefore, the initial damping factor of approximately 0.05 might be
suitable for this particular specimen using the damping proportional to instantaneous stiffness.

If the damping proportional to instantaneous stiffness was used, the large amplitude oscillation of
displacement response was reasonably well simulated, but medium amplitude oscillations at 5 and 11
seconds from the beginning were not favorably reproduced. The calculated response showed
oscillation in a shorter period; the stiffness was estimated to be too large compared to the specimen.
This may be because the hysteresis model for bar slip deformation does not consider the pinching
characteristics.

The yielding was calculated at the ends of second-floor beam, at top of the second-story columns
and at the base of the first-story columns, where the yielding was observed in the test.

Pseudo-dynamic Test: In the nonlinear response of a building structure, the damping proportional to
the instantaneous stiffness gives better correlation with the observed response; the effect of the
damping proportional to the instantaneous stiffness is relatively small on the calculated response in
an inelastic range. Therefore, the equilibrium of inertia force and resistance should be considered.

Using an explicit integration scheme, for example the central difference method, it is possible to
determine the displacement response at the next time step:
{x}i +1 − {x}i
{x&}i +1 =
∆t
{x&}i +1 − {x&}i
{&x&}i =
∆t
{x}i +1 − {x}i {x}i − {x}i −1
−
= ∆ t ∆t
∆t
{x}i +1 − 2{x}i + {x}i −1
=
∆t 2

ti+1 ti ti-1
---------------------------------------------
xi+1 xi xi-1
x& i +1 x& i
&x&i

Solving for the displacement {x}i+1 at time ti+1,

{x}i +1 = 2{x}i − {x}i −1 + ∆t 2 {&x&}i

The resistance vector {R}i+1 at time ti+1 can be determined in an experiment by applying statically
the forced displacement {x}i+1 to the specimen in the laboratory.

The equation of motion without damping at time ti+1 gives

[ M ]{&x&}i +1 + {R}i +1 = −[ M ]{e}&y&i +1

The observed resistance {R}i+1 and calculated displacement {x}i+1 may be input to the above
relation to obtain the acceleration {&x&}i +1 .

11
 This testing method was called "On-line test" or "Pseudo-dynamic test," originally developed by Dr.
M. Hakuno (1969) using an analog computer.

The method has been extensively used in various earthquake response tests (e.g., Okada and
Seki, 1979). The major advantages of the test method are that (a) the development of damage in the
specimen can be observed and (b) the hysteresis relations need not be assumed in the analysis. The
difficulty in the method is to evaluate the numerical stability associated with the accuracy in the
instrumentation and the procedure and sequence to apply the forced deformation to a
multi-degree-of-freedom system.

Teshigawara (1980) tested two-story one-bay frames. One specimen (specimen FDR-1) was
designed to form the beam yielding mechanism and the other specimen (specimen FDR-2) was
designed to form the story side-sway mechanism. The beam section was 150x180 mm, and the
column section 150x150 mm. The horizontal forces were applied at the mid-span of the beams.

Loading Device Loading device

Steel Steel

Loading device Loading device

Steel Steel

Unit

(a) Specimen FDR-1 (b) Specimen FDR-2

In the analysis, one component model (Giberson, 1967) was used for beams and columns. The
moment-rotation relation was determined by first analyzing the moment-curvature relation of section
by the lamina model, and then member end rotation was assumed proportional to the member end
curvature for the inflection point at the mid-point of the member. The degrading trilinear model
(Fukada, 1969) was used to define the hysteresis relation.

The observed (solid line) and calculated (dashed line) second-floor displacement and first-story
shear waveforms are compared. The large amplitude oscillation was successfully reproduced by the
model, but calculated small to medium amplitude waveforms shows a shorter response period,
indicating the model stiffness was higher than the observed at a low stress level.

12
 Calculated
Observed

Time, sec

Free vibration

(a) First story displacement (Test FDR-1)

Free vibration

Time, sec

(b) First story shear (Test FDR-1)

Free vibration

Time, sec

(c) First story displacement (Test FDR-2)

Free vibration

Time, sec

(d) First story shear (Test FDR-2)

13
References:
Clough, R. W., and J. Gidwani, “Reinforced Concrete Frame 2; Seismic Testing and Analytical
Correlation,” EERC Report 76-15, Earthquake Engineering Research Center, University of
California at Berkeley, 1976.
Eto, H., T. Takeda and Y. Omote, "Dynamic Destruction Test of Three-story One-span Reinforced
Concrete Frames (Part 1: Test Results) (in Japanese)," Report, AIJ Annual Meeting, October 1972,
pp. 1119 - 1120, and "ditto, (Part 2: Discussion on Test Results) (in Japanese)," Report, AIJ Kanto
District Meeting, 1973, pp. 45 - 48.
Fukada, Y., "Study on the Restoring Force Characteristics of Reinforced Concrete Buildings (in
Japanese)," Proceedings, Kanto Branch Symposium, Architectural Institute of Japan, No. 40,
1969, pp. 121-124.
Giberson, M. F., "The Response of Nonlinear Multi-story Structures subjected to Earthquake
Excitation," EERL Report, Earthquake Engineering Research Laboratory, California Institute of
Technology, Pasadena, 1967.
Hakuno, M., Shidawara, and Hara, "Dynamic Failure Test of Beams Controlled by Computer,"
Transactions, Japan Society of Civil Engineers, No. 171, November 1969.
Healey, T. J. and M. A. Sozen, “Experimental Study of the Dynamic Response of a Ten-story
Reinforced Concrete Frame with a Tall First-story,” Structural Research Series No. 450,
Department of Civil Engineering, University of Illinois at Urbana-Champaign, 1978.
Hidalgo, P., and R. W. Clough, “Earthquake Simulator Study of a Reinforced Concrete Frame,” EERC
Report 74-13, Earthquake Engineering Research Center, University of California at Berkeley,
1974.
Kabeyasawa, T., H. Shiohara, S. Otani and H. Aoyama, "Analysis of the Full-scale Seven-story
Reinforced Concrete Test Structure," Journal, Faculty of Engineering, University of Tokyo (B), Vol.
XXXVII, No. 2, 1983, pp. 432-478.
Kunanath, S. K., and A. M. Reinhorn, “Inelastic Three-dimensional Response Analysis of Reinforced
Concrete Building Structure (IDARC-3D), Part 1- Modeling,” Technical Report NCEER-89-0011,
State University of New York at Buffalo, New York, 1989.
Kunnath, S. K., A. M. Reinhorn, and Y. J. Park, “Analytical Modeling of Inelastic Seismic Response of
R/C Structures,“ Proceedings, Journal, Structural Engineering, ASCE, Vol. 116, No. 4, April 1990,
pp. 996-10117.
Okada, T., and M. Seki, "Earthquake Response Testing of Reinforced Concrete Frames using
Computer-Actuator On-line System, Part 1: Objectives and Methodology (in Japanese),"
Transactions, Architectural Institute of Japan, No. 275, January 1979, pp. 25 - 31, "ditto, Part 2:
On-line Test 1 (in Japanese)," Transactions, AIJ, No. 279, May 1979, pp. 77 - 84, "ditto, Part 3:
On-line Test 2 (in Japanese)," Transactions, AIJ, No. 280, June 1979, pp. 79 - 89, "ditto, Part 4:
Discussion of Earthquake Response Characteristics (in Japanese)," Transactions, AIJ, No. 282,
August 1979, pp. 57 - 64.
Otani, S., "Earthquake Tests of Shear Wall-Frame Structures to Failure," Proceedings, ASCE/EMD
Specialty Conference, University of California at Los Angeles, March 1976, pp. 298 - 307.
Otani, S., and M. A. Sozen, "Behavior of Multistory Reinforced Concrete Frames during
Earthquakes," Civil Engineering Studies, SRS No. 392, University of Illinois at Urbana, November
1972.
Otani, S., "Failure Test of Reinforced Concrete Structure, Data for Dynamic Analysis (in Japanese),"
Report, AIJ Annual Meeting, September 1980, pp. 1555 - 1556.
Roufail, M. S. L., and C. Meyer, “Analytical Modeling of Hysteretic Behavior of R/C Frames,
“Proceedings, ASCE, Journal, Structural Engineering, Vol. 113, No. 3, March 1987, pp. 429-444.
Takeda, T., M. A. Sozen and N. N. Nielsen, "Reinforced Concrete Response to Simulated
Earthquakes," Journal, Structural Division, ASCE, Vol. 96, No. ST12, 1970, pp. 2557-2573.
Teshigawara, M., "Simulation of Nonlinear Earthquake Response of Two-story Reinforced Concrete
Frames by Computer-Actuator On-line System," M. Eng. Thesis submitted to the University of
Tokyo, February 1980.

14
13.4 Frame-Wall Structures
Column
Seven-story Three-bay Frame: A Beam
seven-story three-bay frame with a
structural wall in the center bay was
tested on the Ohbayashi-gumi Weight
earthquake simulator (Koike, Omote and
Takeda, 1980). The test specimen was
approximately one-tenth scale of an
imaginary prototype structure.

Story height was uniform and 320 mm, Wall

and bay width was 540 mm. Steel weights Reinf.
were attached at each beam-column joint
through mechanical hinges. Beam

The column section was 60 x 60 mm,

and beam section was 40 x 70 mm; wall
thickness was 20 mm. Four 4-mm
diameter indented wires were used as Column
longitudinal reinforcement, 1.4-mm
diameter plane wires as lateral
reinforcement in beams and columns,
2-mm diameter wires as wall
reinforcement. Tensile reinforcement ratio
pt of the wall and beams was 0.52 % and
0.90 %, respectively. Gross reinforcement
ratio pg of columns was 1.40 %. The shear Indented Wire Unit
reinforcement ratio pw was 0.21 % in
beams and columns.

In the analysis, the wall and columns

were divided into 13 short segments; the moment-curvature relation was varied in each segment. The
shear and axial deformations were considered in the column and wall. One component model was
used for a beam.

The Takeda hysteresis model (Takeda, Sozen and Nielsen, 1979) was used at beam ends
connected to the wall, and the Takeda-slip hysteresis model, especially developed for this study, was
used at the beam ends connected to the exterior columns. The Takeda model was used for the
moment-curvature relation of the wall and column segments. The stiffness of the shear deformation
was reduced to one-tenth after shear cracking, but the Clough model (Clough and Johnston, 1966)
was used as the hysteresis relation. The parameters of the hysteresis models were determined on
the basis of the observation in the static tests on individual components.

The specimen was tested using the Hachinohe Harbor (NS) record recorded during the 1968
Tokachi-oki earthquake. The observed and calculated response waveforms are compared for the
top-floor acceleration and displacement.

Reference:

Koike, Y. Omote and T. Takeda, "Reinforced Concrete Wall-frame Structures subjected to Dynamic
Loading - Model Tests and the Simulations," Proceedings, Seventh World Conference on
Earthquake Engineering, Istanbul, Vol. 6, 1980, pp. 419 - 426.
Takeda, T., M. A. Sozen and N. N. Nielsen, "Reinforced Concrete Response to Simulated
Earthquakes," Journal, Structural Division, ASCE, Vol. 96, No. ST12, 1970, pp. 2557-2573.

15
 Beam-end -1

Wall Beam end -2 Beam end -2

16
Beam-column Beam end -1
joint

Sub-element
Beam
Column, wall
element

Column

Test
Analysis
Member model Wall shear
Hysteresis
models
Acceleration, cm/sec2

Seventh floor acceleration (observed)

Seventh floor acceleration (calculated)

Displacement, mm

Seventh floor displacement (observed)

Acceleration, cm/ssec2

Seventh floor displacement (calculated)

Base motion

Time, sec

17
13.5 Wall Structures

Ten-storey Coupled Shear Walls: Ten-storey coupled shear walls were tested on the University of
Illinois earthquake simulator (Aristizaba1-Ochoa and Sozen 1976). Takayanagi and Schnobrich (l976)
divided a wall into short segments of uniform stiffness, and represented connecting beams by the
one-component model. The Takeda-Takayanagi model with changing axial force was assigned to a
wall element, and the Takeda-Takayanagi model with pinching action and strength decay was used in
a beam. It was judged that the usage of two-dimensional plane stress elements for the walls was less
desirable because such an approach might cost more computational effort without any compensating
increase in accuracy.

The amplitude of the exterior column axial load varies greatly due to the earthquake overturning
moment, and changes its moment-carrying capacity. Takayanagi and Schnobrich (1976) incorporated
the effect of axial force variation in the Takeda model by preparing various backbone curves at
different axial load levels.

(a) axial force variation (b) pinching and strength decay

Takeda-Takayanagi models (Takayanagi and Schnobrich 1976):

A pinching action and strength decay are inevitable in a short and deep member due to bar slip
and deterioration in shear resistance. Takayanagi and Schnobrich (1976) introduced a pinching
action and strength decay in the Takeda model. Whenever a response point was located in the
positive rotation-negative moment range or the negative rotation-positive moment range, the pinching
was introduced. After the moment exceeded the yield level, strength decay was incorporated. The
values of guideline for strength decay and pinching stiffness were not related to the member
geometry and material properties.

18
 (a) displacement at level 10, in inches

(b)Acceleration at level 10, g.

Analysis of ten-story coupled shear wall (Takayanagi and Schnobrich 1976)

The comparison of the measured and

calculated displacement and acceleration is
excellent. It is necessary to include the
effects of inelastic axial rigidity of the wall
section and pinching action and strength
decay of the connecting beams to reproduce
the maximum displacement response and
the elongation of the period. Some stiffness
parameters for the walls and connecting
beams were defined on the basis of static
tests of connecting beam-wall assemblies.

Chimney Structure: A model of a tall

reinforced concrete chimney was tested on
the Ohbayashi-gumi earthquake simulator
with four lumped masses (Omote and
Takeda, 1974). The time axis in the test was
not compressed.

In the analysis of the model, the chimney Displ.

was divided into short segments, and the Acc.
flexural stiffness was assumed to be uniform
within a segment over a short time increment.
The primary curve of moment-curvature
relationship (flexural rigidity) was evaluated
for the sectional geometry and material
properties, and then idealized to a trilinear
relation. The Takeda hysteresis model was Unit
used to represent the moment-curvature
relation under load reversals.
19
 The resistance-displacement relation at the top observed in the static test was favorably simulated
by the analysis.

The observed and calculated displacement response waveforms at the top are compared. The
Rayleigh-type damping, proportional to mass and initial stiffness, was assumed; the initial damping
factors for the first two modes were assumed to be 0.02. Large amplitude oscillation was reasonably
reproduced by the model, but small and medium amplitude response was not.

Reference:

Aristizabal-Ochoa, J. D., and M. A. Sozen, “Behavior of Ten-story Reinforced Concrete Walls

subjected to Earthquake Motion,” Structural Research Series No. 431, Department of Civil
Engineering, University of Illinois at Urbana-Champaign, 1976.
Omote, Y., and T. Takeda, "Study on Elasto-plastic Response of Reinforced Concrete Chimney, Part
1: Model Test (in Japanese)," Transactions, Architectural Institute of Japan, No. 215, January
1974, pp. 21 - 32.
Takayanagi, T., and Schnobrich, W. C, “Computed Behaviour of Reinforced Concrete Coupled Shear
Walls,” Structural Research Series, No. 434, Department of Civil Engineering, University of Illinois at
Urbana-Champaign, 1976.
Load, kgf

Displacement,

Drift

Test
Analysis

20
 Observed in Test R1

Calculated with h1=h2=0.02

Top floor displacement, cm

Observed in Test R2

Calculated with h1=h2=0.02

Time, sec

21
 Chapter 15. U.S.-Japan Full-scale Test

This paper describes an analysis of the full-scale seven-story reinforced concrete building tested
as a part of the U.S.-Japan Cooperative Research Program Utilizing Large Scale Testing Facilities. A
general purpose computer program was developed to simulate the inelastic behavior of a structure
during an earthquake. On the basis of given structural geometry and material properties, this paper
places an emphasis to describe (a) methods to model member behavior, and (b) methods to
determine member stiffness properties. At the time of analysis, the results of small-scale
sub-assemblage tests and the full-scale test were made available to the authors, hence, the
information was reflected in the development of the analytical models and evaluation methods. A
good correlation of the observed and computed responses of the test structure is reported in this
paper.

15.1 Test Program of Full-scale Seven-story RC Building

A test of a full-scale reinforced concrete building was conducted, as a part of U.S.-Japan

Cooperative Research Program Utilizing Large Scale Testing Facilities, at the Large Size Structures
Laboratory of Building Research Institute, Ministry of Construction, Tsukuba, Japan (1).

A full-scale seven-story reinforced concrete building was designed and constructed in the Large
Size Structures Laboratory, Building Research Institute, in accordance with normal construction
specifications and practices (Fig. 1). The building had three bays in the longitudinal direction, and two
spans in the transverse direction. A shear wall was placed, parallel to the direction of loading, in the
middle bay of the center frame.

The structure was subjected to lateral load of an inverted triangular distribution at each level by
eight actuators; two actuators used at the roof level and one actuator each at the other six levels. The
pseudo-dynamic test method was used to control the roof-level displacement; i.e., the response
displacement under an imaginary earthquake motion was computed, in parallel with the test, for a
system having the observed restoring force characteristics. The computed response displacement
was applied to the roof level of the test structure while the load amplitude at the first to the sixth floor
levels were made proportional to the load measured in the roof level actuators; in this fashion, the
number of degrees of freedom of the test structure was reduced to one. A total of 716 channel strains,
displacements, rotations and loads were measured during the test.

The intensity of imaginary earthquake motions was varied in four test runs to yield expected
maximum roof-level displacements of approximately 1/7000, 1/400, 3/400, and 1/75 of the total height
(Table 1). The earthquake record used in the test was modified from the original record so that the
first mode response should dominate in the response of the test structure; i.e., higher frequency
components were removed from the original records. Free and forced vibration tests were carried out
between pseudo-dynamic earthquake test runs to study the change in period and damping.

After the fourth test run, epoxy resin was injected into major cracks in structural members, and
non-structural partitions were installed to the original bare structure. The repaired structure was
tested in three runs using the pseudo-dynamic test method, and finally tested statically under
reversals of uniform load distribution to a roof-level displacement of 1/50 the total height.

The second, third and fourth pseudo-dynamic earthquake tests of the bare structure were
simulated by analytical models in this paper.

1
 Table 1: Test Program

Test No. Brief Description

VT-1 Free and forced vibration tests
PSD-1 Pseudo-dynamic earthquake test
Modified Miyagi-ken Oki Earthquake (1978)
Tohoku University Record (NS), amax*= 23.5 Gal
Rmax**= 2.52 mm, Smax***= 31.5 tonf
PSD-2 Pseudo-dynamic earthquake test
Modified Miyagi-ken Oki Earthquake (1978)
Tohoku University Record (NS), amax*= 105 Gal
Rmax**= 32.5 mm, Smax***= 226 tonf
PSD-3 Pseudo-dynamic earthquake test
Modified Tehachapi Shock (1952)
Taft Record (EW), amax*= 320 Gal
Rmax**= 238 mm, Smax***= 411 tonf
PSD-4 Pseudo-dynamic earthquake test
Modified Tokachi-oki Earthquake (1968)
Hachinohe Harbor Record (EW), amax*= 350 Gal
Rmax**= 342 mm, Smax***= 439 tonf
VT-2 Free and forced vibration tests
Repair of test structure by epoxy injection
VT-3 Free and forced vibration tests
Placement of non-structural elements in test structure
PSD-5 Pseudo-dynamic earthquake test
Modified Miyagi-ken Oki Earthquake (1978)
Tohoku University Record (NS), amax*= 23.5 Gal
Rmax**= 3.03 mm, Smax***= 26.7 tonf
PSD-6 Pseudo-dynamic earthquake test
Modified Miyagi-ken Oki Earthquake (1978)
Tohoku University Record (NS), amax*= 105 Gal
Rmax**= 65.3 mm, Smax***= 234 tonf
PSD-7 Pseudo-dynamic earthquake test
Modified Tehachapi Shock (1952)
Taft Record (EW), amax*= 320 Gal
Rmax**= 244 mm, Smax***= 452 tonf
SL Static test under uniform load distribution
Rmax**= 326 mm, Smax***= 597 tonf
*amax: maximum acceleration of input ground motion
**Rmax: maximum roof level displacement
***Smax: maximum base shear

2
(a) Typical Floor Plan

(b) Typical Elevation

Fig. 1: Test Structure

3
15.2 Description of Test

The construction of the full-scale seven-story reinforced concrete test structure is described in
detail in a paper by J. K. Wight and S. Nakata (1). The method of testing is described by S. Okamoto
et a1. (2). The information relevant to the structural analysis is summarized herein from the two
papers.

Geometry of Test Structure: A general plan view is shown in Fig. 2. The test structure consisted of
three three-bay frames, (frames A, B, and C) parallel to the loading direction, and four two-bay frames,
(frames 1 , 2, 3 and 4) perpendicular to the loading direction. The span widths were 6.0, 5.0, and 6.0
m in the longitudinal direction, and 6.0 m each in the transverse direction. Frame B had a shear wall
in the central bay continuous from the first to the seventh story.

An elevation of Frame B is shown in Fig. 3. Floor level and story notations are defined in the figure.
The inter-story height was 3.75 m in the first story, and 3.0 m from the second to the seventh story.
Note that the girders of spans 1-2 and 3-4 were not continued through the shear wall. Frames A and
C were three-span open frames, having inter-story heights and bay widths identical to Frame B.

A general elevation of Frame 4 is shown in Fig. 4. Two walls were installed in the frame
perpendicular to the loading direction so as to reduce the torsional and transverse displacements of
the test structure. A 1.0-m gap was provided between the face of a column and the edge of the wall to
eliminate the contribution of the wall in the stiffness of the structure in the loading direction. Frame 1
was identical to Frame 4 except for pairs of openings in the walls for the loading beams. Frames 2
and 3 were open frames without walls.

A plan view of the foundation is given in Fig. 5, a floor plan for the second floor through the
seventh floor levels in Fig. 6, and the roof plan in Fig. 7. Notations for beams, columns and walls are
shown in the figures. The foundation was post tensioned to the test floor with 33-mm diameter
high-strength (10,000 kgf/cm2) rods at a stress level of 5,900 kgf/cm2.

The dimensions of a column section were 50 x 50 cm2 throughout the test structure. The size of
the girders parallel to the loading direction was 30 x 50 cm2 from the second to the roof level. The
dimensions of a transverse beam were 30 x 45 cm2. The wall (W1) parallel to the loading direction had
a thickness of 20 cm, and the transverse walls a thickness of 15 cm, both from the first to the seventh
story. The floor slab was 12 cm thick throughout the structure. Two 120 x 120 x 80 cm3 loading points
were placed at each floor from the second to the seventh floor in the floor slab at the mid-span of
beams B2 (Fig. 6). At the roof level, the loading point dimensions were 70 x 530 x 64 cm3 (Fig. 7).

The as-built dimensions were reported to be very close to the nominal dimensions. In the first
story, areas of poorly compacted concrete were found near the base of the first story columns. The
voids did not penetrate into the column core even in the worst case. However, longitudinal and
transverse reinforcing bars were reported to be visible in some locations.

Reinforcement Details: Cross section reinforcement details for the foundation beams and floor
beams are shown in Figs. 8 and 9. D19 and D25 deformed bars were used as flexural reinforcement,
and D10 and D19 deformed bars as stirrup reinforcement. Two digits after alphabet D denote
approximate bar diameter in mm. In the foundation beams, D16 bars were used in the web to hold the
stirrups in position. All of the beam bars terminating at an exterior column or in the wall boundary
columns were anchored with a 90-degree hook. Within a region extending one-quarter of the clear
span from a column face, floor beam stirrups were spaced at an approximately one-fourth of the
effective beam depth. The spacing was increased to approximately one-half of the effective depth in
the middle region.

A typical column cross section is given in Fig. 10. The columns were all 50 x 50 cm2, and
reinforced with 8-D22 deformed bars. All of the column bars terminated at the roof level with a 180
degree hook. Perimeter hoops were spaced at 10 cm over the total height of the columns, including
4
the beam-to-column joint regions. Cross ties in the first-story independent columns were provided at
a 10-cm spacing over the first 60 cm above the foundation, and at a 60-cm spacing elsewhere. Cross
ties in the boundary columns of the shear wall were provided at a 10-cm spacing over the full height
of the first three stories except in the beam-to-column joint regions, and at a 60-cm spacing
elsewhere.

The shear wall, parallel to the direction of loading, was reinforced with 2-D10 bars at a spacing of
20 cm in the horizontal and vertical directions. The horizontal wall reinforcement was anchored into
the boundary columns, and the vertical wall reinforcement into the foundation.

Reinforcement details for the floor slabs at the second through the roof levels are shown in Fig. 11.
Different spacing was used in the column strips, middle strips and in the cantilevered portion of the
floor slabs.

Materials: Deformed bars were used in the constructions of the test structure. The grade of
reinforcing steel was SD35. Geometrical and mechanical properties are listed in Table 2 taken from
Reference 1. All bars showed a clear yield plateau after yielding up to a strain of 0.012 to 0.022,
depending on the bar size.

The ready-mixed concrete was used in the test structure. Following the Japanese construction
practices, the concrete was placed in columns of a story and into beams and slabs immediately
above the columns in a single job. The mechanical properties of the concrete are listed in Table 3.
The values were obtained from the tests of 15 x 30-cm2 standard cylinders cured in the field. The
sixth and the seventh story concrete strengths were found to be significantly weaker than the
specified strength of 270 kgf/cm2. However, the compression tests on standard cured cylinders did
not show such a change in concrete strength (1). The tensile strength was determined by the splitting
test of cylinders.

Method of Testing: Test of the full-scale seven-story structure was carried out using "SDF Pseudo
Dynamic Earthquake Response Test Procedure". The theoretical background is outlined by Okamoto
et al. (2). The summary is given below.

The equation of motion of a multi-degree-of-freedom system without damping can be written in a

matrix form;

x} + { f } = −[m]{1}&&
[m]{&& y (1)

in which [ m] : mass matrix, { f } : restoring force vector (resistance of structure), {&x&} : structural
response acceleration vector relative to the base, {1} : vector consisting of unit elements, and &y& :
ground acceleration.

In order to reduce the number of degrees of freedom to one, the structure was assumed to
oscillate in a single "governing" mode. The restoring force distribution pattern was assumed to remain
unchanged during an earthquake. In other words,

{ f } = {v} f R (2)

in which {v} : constant vector, each element of which represents the lateral resistance amplitude
normalized to the roof-level resistance amplitude f R .

Under the specified distribution of the lateral resistance (or loads), the structure would deform in a
certain shape, reflecting the stiffness distribution of the structure. Namely, the "mode shape" {u}
and its "amplitude" q .

5
 {x} = {u}q (3)

If the mode shape is normalized to the roof-level amplitude, the value of q represents a
roof-level lateral displacement xR . Displacement distribution vector {u} normally varies with
stiffness deterioration associated with structural damage. However, the deflected shape pattern did
not change appreciably regardless of load amplitudes in a preliminary analysis of the test structure
under an inverted triangular distribution of lateral loads. Therefore, the structure was assumed to
respond in the fixed mode shape {u} during an earthquake, and the equation of motion was
expressed as

xR + {v} f R = −[m]{1}&&
[m]{u}&& y (4)

T
Pre-multiplying {u} to Eq. 4,

xR + {u}T {v} f R = −{u}T [m]{1}&&

{u}T [m]{u}&& y (5)

Or,

mx&&R + f = −m( β &&

y) (6)

in which m = {u}T [m]{u} : effective mass, f = {u}T {v} f R : effective restoring force, and
β = {u}T [m]{1}/ m : effective participation factor. The seven-story structure was forced to reduce to
an "equivalent" single-degree-of-freedom system in this manner.

An inverted triangular shape was used to represent the lateral resistance distribution {v} . The
corresponding deflected shape {u} was obtained as an average of deflected shapes at different
load amplitudes. Hence, the properties of the equivalent single-degree-of-freedom system are

{u}T = [1.000, 0.850, 0.696, 0.540, 0.384, 0.234, 0.102]

{v}T = [1.000, 0.862, 0.724, 0.586, 0.448, 0.310, 0.172]
β = 1.442
m = 0.643 tonf sec2/cm

The pseudo-dynamic earthquake response test was carried out on the "equivalent" single
degree-of-freedom system.

The central difference method was used for a numerical integration procedure in the
pseudo-dynamic earthquake response test. The central difference method is given

q&&i = ( qi +1 − 2qi + qi −1 ) / ∆t 2 (7)

&&i : acceleration at time step i, and ∆t : time increment for

in which, qi : displacement at time step i, q
numerical integration.

Equation 7 can be rewritten in a form,

qi +1 = 2qi − qi −1 + ∆t 2 q&&i (8)

In other words, from displacements and acceleration at old time steps i-1 and i, the displacement at
new time step i+1 can be evaluated, hence the roof-level displacement xR (= qi +1 ).
6
 Although displacement amplitudes at other levels can also be determined by Eq. 3, only the
roof-level displacement was controlled in the test. Eight actuators, maintaining the fixed load
distribution {v} , applied load to the structure until the roof-level displacement reached the specified
displacement.

When the roof-level displacement attained the calculated amplitude xR ( = qi +1 ), the resistance
f R at the roof level was measured. The acceleration amplitude &x&R (= q&&i +1 ) was evaluated by Eq.
yi +1 . With a new acceleration value at time step i+1, Eq.
6 with given ground acceleration amplitude &&
8 was used to calculate the displacement at further time step i+2.

Repeating the procedure outlined above, the test structure was subjected to an imaginary
earthquake motion.

Equations 6 and 8 may be combined to yield a single-step procedure,

qi +1 = 2qi − qi −1 − (∆t 2 / m) f i − ( β∆t 2 ) &&

yi (9)

or in an incremental form,

∆qi +1 = ∆qi − (∆t 2 / m) fi − ( β∆t 2 ) &&

yi (10)

in which, ∆qi = qi − qi −1 . Equation 10 was used in the analysis.

7
 Table 2: Properties of Reinforcing Bars

Nominal Nominal Nominal Yield Strain Tensile

Bar Diameter Perimeter Area Strength Hardening Strength Fracture
size mm mm mm2 Kgf/cm2 Strain Kgf/cm2 Strain
D10 9.5 30.0 71 3870 0.018 5670 0.17
D16 15.9 50.0 199 3850 0.019 5720 0.18
D19 19.1 60.0 287 3650 0.017 5730 0.20
D22 22.2 70.0 387 3530 0.012 5750 0.21
D25 25.4 80.0 507 3780 0.022 5660 0.20

Table 3: Properties of Field Cured Concrete

Story Test Compressive Strain at Tensile

Age Strength Compressive Strength
(days) (kgf/cm2) Strength (kgf/cm2)
7 67 189 0.0019 13.2
6 87 144 0.0019 13.3
5 98 295 0.0019 23.6
4 111 290 0.0023 23.3
3 119 274 0.0023 22.8
2 132 292 0.0024 24.6
1 145 289 0.0022 24.2

8
Fig. 2: General Plan View and Frame Notations

Fig. 3: Elevation of Frame B

9
 Fig. 4: Elevation of Frame 4

Fig. 5: Plan View of Foundation

10
Fig. 6: Plan View of Second through Seventh Floor Levels

Fig. 7: Plan View at Roof Level

11
 Fig. 8: Reinforcement in Foundation Beams

Fig. 9: Reinforcement in Beams at Second through Roof Levels

12
Fig. 10: Typical Column Cross Section

Fig. 11: Slab Reinforcement

13
15.3 Modeling of Structural Members

It is not feasible to analyze an entire structure using microscopic material models. Therefore, it is
necessary to develop a simple analytical model of structural members.

Nonlinear dynamic analysis of a reinforced concrete structure requires two types of mathematical
modeling: (a) modeling for the distribution of stiffness along a member; and (b) modeling for the
force-deformation relationship under stress reversals. The former models are called "member
models", and the latter "hysteresis models".

Inelastic deformation of a reinforced concrete member does not concentrate in a critical location,
but rather spreads along the member. Various member models have been proposed to represent the
distribution of stiffness within a reinforced concrete member (3, 4).

The member models used to represent the stiffness behavior of beams, columns, and walls are
presented in this part.

Beam and Column Model: Many member models have been proposed for the beam and column
members; for example, (a) One-component model (5), (b) Multi-component model (6), (c) Connected
two-cantilever model (7), (d) Distributed flexibility model (8).

The One-component model was
column member was idealized as
rotational springs at the two ends.
springs as shown in Fig. 12. Axial
member.
used for beams and columns in this paper. Namely, beam or
a perfectly elastic massless line element with two nonlinear
The model could have two rigid zones outside the rotational
deformation is considered in the elastic element of a column

The stiffness properties of a rotational spring are evaluated for an imaginary anti-symmetric
loading conditions with the inflection point at the center of the flexible portion of a member. The
rotation at a flexible end less the elastic rotation is assigned to the rotational spring. The shear
deformation within a member and the member end rotation due to bar slip within the beam-to-column
connection should be considered in the evaluation of the deformation.

The shear deformation of a beam-to-column connection panel is not considered in the analysis.

Wall Model: A shear wall is normally idealized as (a) an equivalent column taking flexural and shear
deformation into account, (b) a braced frame, in which the shear deformation is represented by the
deformation of diagonal elements, whereas the flexural deformation by the deformation of vertical
elements, and (c) short line segments along the height with each short segment with hysteretic
characteristics (9, 10). These models have advantages and disadvantages. In most cases, the
horizontal boundary beams (or slabs) are assumed to be rigid.

The Japanese support tests on three-story walls with connecting beams (11) indicated a large
elongation of a tension-side column due to cracking, and a small compression of the
compression-side column, with the neutral axis of wall section close to the compression-side column.
In other words, the bending deformation of a wall was caused primarily by the extension of the
tension-side boundary column. The resistance of a wall came from the resistances of the boundary
columns and that of the central wall section.

The wall member of a story was, therefore, idealized as three vertical line elements with infinitely
rigid beams at the top and bottom floor levels (Fig. 13). Two outside truss elements represented the
axial stiffness of boundary columns. The axial stiffness varied with the sign and level of axial stress,
and degraded with tensile stress history. The central vertical element was a one-component model in
which vertical, horizontal and rotational springs were concentrated at the base. A finite rigid zone
could be placed between the spring assembly and the lower rigid chord.

14
 The model was intended to simulate the wall deformation under uniform bending, the resistance of
wall section being lumped at the locations of the outer truss elements and the central vertical spring.
The effect of strain gradient across the wall section was represented by the rotational spring in the
central element, and the shear deformation expressed by the deformation of the horizontal spring.

The stiffness matrix of a wall element was formulated as the sum of the stiffness of the three
vertical elements evaluated at the top and bottom of the two boundary columns.

Transverse Beam Model: The tensile boundary column of a wall tends to elongate extensively under
bending deformation, yielding a significant vertical displacement at a beam-to-wall-joint node,
whereas the vertical displacement of a beam-to-column-joint node of an open frame is relatively small.
Consequently, the transverse beam connecting the boundary column of a shear wall and an adjacent
parallel open frame is subjected to vertical differential displacement at the two ends, and resists the
upward movement of a wall boundary column.

Vertical spring elements, therefore, were introduced to reflect the effect of such transverse beams
to restrain the elongation of a tensile boundary column (Fig. 14). A spring was placed between the
joints of the wall and an open frame connected by a transverse beam.

15
Fig. 12: One-component Model for Beams and Columns

Fig. 13: Wall Model

Fig. 14: Transverse Beam Model

16
15.4 Stiffness of Member Models

Force-deformation relationship of member models under monotonically increasing load (called

skeleton force-deformation relationship) was evaluated on the basis of idealized stress-strain
relations of the concrete and the reinforcing steel.

Force-Deformation Relation: The force-deformation relationship is described for each member

model. As the analysis reported herein was of preliminary nature, approximate methods were used in
evaluating member deformations and resistances. Nominal member dimensions and material
properties obtained from coupon tests were used.

(1) Beam Stiffness: The beams were analyzed as a T-shaped beam, taking the contribution of slab
into account. The effective width of slab for the elastic stiffness of a beam was taken in accordance
with the Architectural Institute of Japan Standard (AIJ Standard for R/C) for Structural Calculation of
Reinforced Concrete Structure (12); i.e., the cooperating flange width ba in a T-shape member (one
side) is

ba = (0.5 − 0.6a / l)a when a < 0.5l (11.a)

ba = 0.1l when a ≥ 0.5l (11.b)

where a : distance from the side of a beam to the side of the adjacent parallel T-beam (Fig. 15), and
l : span length of the beam.
Equation (11.b) governed in all beams, and the total effective width B of beams parallel to the
loading direction was 150 cm in spans 1-2 and 3-4, and 130 cm in span 2-3 (Fig. 2).

The moment of inertia of a T-shaped beam section was computed about the geometrical centroid
ignoring the contribution of reinforcing steel. The elastic modulus of concrete was assumed to be 2.37
x 105 kgf/cm2, ignoring the fact that the field cured cylinders from the sixth and seventh story concrete
showed lower strength. The elastic stiffness properties were given to the perfectly elastic massless
line element of a one-component model.

Cracking moment M c of a beam at the face of the supporting column was computed on the
basis of the flexural theory and an assumed concrete tensile strength of 20 kgf/cm2 (Table 3); i.e.,

∗
M c = cσ t Ze (12)

where c σ t : tensile strength of concrete (=20 kgf/cm2), and Z e : section modulus without reinforcing
steel.

The value of cracking moment was different for the positive and negative bending because the
geometrical center does not locate at the mid-height of the section. The average value of positive and
negative cracking moments was used in the analysis.

Yield moment and curvature of a T-shaped beam section were calculated based on the flexural
theory. A linear strain variation across the section was assumed and the stress-strain relationships for
the longitudinal steel and concrete were considered as input factors.

Bi-linear model was used for the stress-strain relationship of steel as shown in Fig.16.a. Yield
stress (=3,650 kgf/cm2), and elastic modulus (=1.710 x 106 kgf/cm2), for D19 deformed bars were
determined according to the results of the material tests. The stiffness after yielding was assumed to
be zero.

The stress-strain relation model by Aoyama (17) was used for concrete as shown in Fig. 16.b,
17
which defined the primary curve according to the following equation; i.e.,

α
 σ B −σ   εB − ε  Ecε B
 =  where α = (13)
 σB   εB  σB

where σ , ε : compressive stress and strain, σ B : stress at compressive strength (=290 kgf/cm2), ε B :
strain at compressive strength (=0.0021), and Ec : initial tangent modulus (=2.37 x 105 kgf/cm2).

The slab can contribute to the resistance of a beam. The region, in which slab reinforcement
parallel to the loading direction yielded under beam negative moment, progressively spread with
increasing beam rotation. The strains measured in the slab reinforcing bars during the full-scale test
indicated that the effective slab width B (Fig. 15) was 350 cm in Frames A and C and 510 cm in Frame
B at maximum structural deformation (18). Therefore, the slab effective width B of 430 cm was used
in computation. Consequently, the yield moments for the positive and negative bending were
significantly different.

The inelastic beam deformation was assumed to concentrate at the locations of two nonlinear
rotational springs. The beam-end rotations at cracking and yielding were computed on the basis of
corresponding curvature distribution of the beam with an inflection point assumed to locate at the
mid-span of the flexible portion of the beam. The shear deformation was assumed to be proportional
to the flexural deformation. The calculated beam-end rotation less the elastic deformation was
assigned to the rotational spring at the end.

The skeleton moment-rotation curve was represented by a trilinear relation in each direction of
loading. The stiffness after yielding was arbitrarily assumed to be 3 % of the initial elastic stiffness.
The calculated stiffness properties of a beam model are listed in Table 4. The elastic deformation is
included in the calculated rotation.

(2) Column Stiffness: The dimensions of a column section and the amount of longitudinal
reinforcement were identical in all the column. The elastic stiffness properties (moment of inertia,
cross sectional area, and area effective for shear deformation) were calculated for gross concrete
section, ignoring the contribution of the steel reinforcement.

The existing axial force of a column due to the gravity loading was not the same for a column at
different story levels, and for columns of a story at different locations. The weight of slab, beams, and
girders within the tributary area of a column (Fig. 17) was used to calculate the axial load. The
calculated values (Table 3) were generally in reasonable agreement with the values obtained from
strain gauge measurement on column longitudinal bars. Columns C1 and C3 carried the weight of
actuators and loading beams. For columns C1 and C1 ' , or C3 and C3 ' , the average axial load of
the two columns was used in the analysis. The variation of axial load due to the overturning effect of
earthquake forces was not considered in evaluating flexural resisting capacity.

Simple approximate expressions (12) were used to evaluate cracking moment M c and yield
moment M y ; i.e.,

M c = cσ t Ze + N D / 6 (14)
M y = 0.8 at σ y D + 0.5 N D(1 − N / b D Fc ) (15)

where N: axial force in column section (Table 5.2), b: width of column section (=50 cm), D: overall
depth of column section (=50 cm), and Fc : compressive strength of concrete (=290 kgf/cm2).

The area at of tensile reinforcement was 3-D22 (=11.61 cm2). The yield strength of D22

18
reinforcing bars was taken from the coupon test to be 3,530 kgf/cm2. The tensile strength of concrete
was assumed to be 20 kgf/cm2.

The rotations of a column were evaluated by a simple empirical formula by Sugano (19). The
formula was prepared for reinforced concrete beams and columns subjected to anti-symmetric
bending. The secant stiffness ( M y / θ y ) at the yield point was proposed:

M N d 6 EI
M y / θ y = (0.043 + 1.64n pt + 0.043 + 0.33 )( ) 2 ( ) (16)
QD bDFc D l

in which, M y : yield moment applied at two member ends, θ y : member end rotation at yielding, n:
Young's modulus ratio ( = Es / Ec ), pt : tensile reinforcement ratio ( = at / bD ) , M/QD: shear
span-to-depth ratio, and l : total length of member.

Ninety percent of test data studied fell within 30 per cent range of the value predicted by Eq. 16.
The Sugano's formula was used to estimate the yield rotation of a column. The column-end rotation
less the elastic deformation was assigned to the rotational spring. The skeleton moment-rotation
curve was represented by a trilinear relation with stiffness changes at cracking and yielding. The
skeleton curves were the same for positive and negative directions. The calculated stiffness
properties are listed in Table 6.

The axial rigidity (= EA / l ) of a column in compression was defined by the gross sectional area,
elastic modulus (2.37 x 105 kgf/cm2) and height of the column (Fig. 3). When the axial force due to the
gravity effect was overcome by the overturning effect of earthquake forces, the axial rigidity was
reduced to 90% of the initial elastic stiffness. The column was assumed to yield in tension when the
net tensile load reached a tensile force equal to the sum of yield forces carried by all the column
longitudinal reinforcement (= 109.3 tonf). After tensile yielding, the stiffness was arbitrarily reduced to
0.1 % of the initial axial stiffness.

(3) Wall Stiffness: The boundary columns and a wall were analyzed as a unit. The wall model
consists of three sub-elements; i.e., (a) two vertical truss elements for the boundary columns, and (b)
vertical one-component element for the wall panel.

The axial rigidity (= EA / l ) of a truss element (Table 7) was determined in the same way as that of
an independent column. The axial rigidity in compression remained linearly elastic. When a net axial
load changed its sign from compression to tension, the stiffness was reduced to 90% of the initial
elastic stiffness. The initial axial forces due to the gravity loads are listed in Table 5.2.b. Tensile
yielding occurred when a net tensile force reached a force level (=109.3 tonf) at which all column
longitudinal reinforcement yielded. Then the stiffness was reduced to 0.1% of the initial elastic
stiffness.

The shear resistance of a shear wall was provided by the lateral spring in the central vertical
element. The initial elastic shear rigidity K s was defined as
GAw
Ks = (17)
κh
in which, G: elastic shear modulus (=0.98 x 105 kgf/cm2), Aw : area of shear wall section (Fig. 18),
κ :shape factor for shear deformation (= 3(1 + u )[1 − u 2 (1 − v)] / 4[1 − u 3 (1 − v)] , h: inter-story height,
and u, v : geometrical parameters defined in Fig. 18.

Shear cracking was assumed to occur at a shear force s Qc (in kgf),

s Qc = 1.4 Fc Aw (18)
2 2
in which Fc : compressive strength of concrete in kgf/cm (=290 kgf/cm ).
19
 Hirosawa's empirical equation (13) was used to evaluate the ultimate shear resisting capacity
s Qu (kgf);

 0.0679 pt 0.23 ( Fc + 180 ) 

Q
s u =  + 2.7 σ p
wh wh + 0.1σ 0  be j (19)
 M / QL + 0.12 

D
where pt : effective tensile reinforcement ratio (%), = 100at / be ( L − ) , at : area of longitudinal
2
reinforcement in tension-side boundary column, M/QL: shear span-to-depth ratio, σ wh : yield strength
2
of horizontal reinforcement in the wall (kgf/cm ), pwh : effective horizontal wall reinforcement ratio =
awh / be x , x : spacing of horizontal wall reinforcement (= 20 cm), σ 0 : average axial stress over
7 D
entire wall cross sectional area (Table 5.2.a), be : average width of wall section, j = ( L − ) , L,D:
8 2
geometrical parameters defined in Fig. 18.

Ratio β s of the secant stiffness at shear yield point to the elastic stiffness was determined
empirically by

β s = 0.46 pwhσ wh / Fc + 0.14 (20)

The shear stiffness reduction factor β s was approximately 0.16 for the shear wall analyzed.

The stiffness after shear yielding was taken to be 0.1 % of the initial elastic shear rigidity.
Calculated stiffness properties are listed in Table 8.a.

Axial stiffness properties of the central vertical element (Table 8.b) were determined in the same
way as the truss element. Area of a shear wall bounded by the inner faces of two boundary columns
was used for the cross sectional area of the central vertical element.

Rotational stiffness properties of the central vertical element (Table 8.c) were defined for wall area
bounded by the inner faces of two boundary columns. Wall rotation was computed as the product of
the curvature at base and the inter-story height. In other words, for the purpose of computing wall
rotation, moment was assumed to distribute uniformly along the story height with an amplitude equal
to the moment at wall critical section. Cracking was to occur when the extreme tensile fiber strain
became zero under the gravity load and overturning moment;

N ( uL )
Mc = (21)
6

Yielding moment M y was taken to be the full plastic moment; moment about the centroid of wall
section caused by the yielding of all vertical wall reinforcement. The gravity load was ignored in
computing the full plastic moment. The stiffness after yielding was taken to be 0.1 % of the initial
elastic stiffness.

(4) Transverse Beam Stiffness: The effect of transverse beams to restrain the upward movement of
a tensile wall boundary column was represented by a vertical spring. The initial elastic stiffness K t
was calculated for a fixed-fixed beam as

20
 12 EI
Kt = (22)
l3

where EI : flexural rigidity of transverse beam, and l : span length of transverse beam.

Cracking and yielding forces were determined as shear force acting in the transverse beam when
both ends cracked and yielded simultaneously in flexure. Cracking moment, yielding moment and
curvature of T-shaped transverse beam section were evaluated based on the flexural theory in the
same way used for beam stiffness evaluation. These values calculated for positive and negative
bending moments were averaged. The effective width B (Fig. 15) of 190 cm was determined referring
to the results of the full-scale test (18). The stiffness after yielding was reduced to 3 % of the initial
stiffness.

The numerical values of the stiffness properties of the vertical spring are listed in Table 9.

Hysteresis Models: A hysteresis model must be able to provide the stiffness and resistance relation
under any displacement history. Four different hysteresis models were used in the analysis; i.e., (a)
Takeda hysteresis model (14), (b) Takeda-slip hysteresis model (16), (c) Axial-stiffness hysteresis
model, and (d) Origin-oriented hysteresis model. The characteristics of each model are briefly
described in this section.

(1) Takeda Hysteresis Model: Based on the experimental observation on the behavior of a number
of medium-size reinforced concrete members tested under lateral load reversals with light to medium
amount of axial load, a comprehensive hysteresis model was developed by Takeda, Sozen and
Nielsen (l4). The model included (a) stiffness changes at flexural cracking and yielding, utilizing a
trilinear skeleton force-deformation relationship, (b) hysteresis rules for inner hysteresis loops inside
the outer loop; i.e., the response point during loading moves toward a peak of the immediately outer
hysteresis loop, and (c) unloading stiffness degradation with a maximum deformation amplitude. The
unloading stiffness K r is given by

Fc + Fy Dm − a
Kr = ⋅| | (23)
Dc + Dy Dy

in which ( Dc , Fc ) : cracking point deformation and resistance, ( Dy , Fy ) : yielding point deformation

and resistance, Dm : maximum deformation amplitude greater than Dy , α : unloading stiffness
degradation parameter (normally between 0.0 and 0.6).

The general hysteresis rules are outlined in Fig. 19. The detail description of the model can be
found in References 7 and 14.

The Takeda hysteresis model was used in inelastic rotational springs of the independent column
one-component model, and in a vertical spring of the transverse beam model.

The unloading stiffness degradation parameter α for an independent column and a transverse
beam model was arbitrarily chosen to be 0.4.

(2) Takeda-Slip Hysteresis Model: Half-scale beam-to-column joint assemblies with slab were
tested (15) to obtain preliminary information about possible behavior of the full-scale seven-story
building. Force-deformation relation of a beam with slab showed obvious pinching characteristics in
negative moment region (loading under which the beam top was in tension) as shown in Fig, 20. This
pinching behavior was not associated with that often observed in a member failing in shear, but rather
associated with a wide crack opening at the bottom of the beam during positive-moment loading; i.e.,
after a load reversal from positive-moment loading, the stiffness did not recover until the crack closed
at the beam bottom.
21
 Eto and Takeda (16) introduced pinching characteristics into a hysteresis model in simulating
member-end rotation behavior due to bar slip within a beam-column connection. The Takeda and
Eto's model was modified in this paper for use in a rotational spring of a beam one-component model.

The Takeda hysteresis model was modified as follows :

(a) The pinching occurs only in one direction where the yield resistance is higher than that in the
other direction, and the pinching occurs only after the initial yielding in the direction concerned.
(b) The stiffness K s during slipping is a function of the maximum response point ( Dm , Em ) and
the point of load reversal ( D0 , F0 ) in the force-deformation plane (Fig. 21.d)

γ
Fm  Dm 
Ks =   (24)
Dm − D0  Dm − D0 

where γ : reloading stiffness parameter.

(c) After pinching, the response point moves towards the previous maximum response point with
stiffness K p ;

Fm
K p = η( ) (25)
Dm

where η : reloading stiffness parameter. In other words, the stiffness change occurs at an
intersection of the two straight lines having slopes K s and K p .

The Takeda-slip hysteresis model was used in the inelastic rotational spring of a beam
one-component model. The values of unloading stiffness degradation parameter α , slipping
stiffness degradation parameter γ , and reloading stiffness parameter η were 0.4, 1.0, and 1.0,
respectively.

(3) Axial-Stiffness Hysteresis Model: The behavior of a column under axial load reversals is not
clearly understood. The following hysteresis model was developed and tentatively used for the axial
force-deformation relation of a column.

Referring to Fig. 22, a point Y' is defined on the elastic slope in compression at a force level equal
to the tensile yield strength Fy . The response point follows the regular bilinear hysteresis rules
between the two points Y and Y' (Fig. 22.a). Once tensile yielding occurs, then a response point
moves following the regular bi-linear hysteresis rules between point Y’, and previous maximum
tensile response point M with a force level of Fy using unloading stiffness K r (Fig. 22.b) :

−a
D 
K r = K c  max  (26)
 Dyt
 

where, Dyt : tensile yielding point deformation, Dmax : maximum deformation amplitude greater than
Dyt , α : unloading stiffness degradation parameter (=0.9). When the response point reaches the
previous maximum tensile point M, then the response point moves on the second slope of the
skeleton curve, renewing the maximum response point M.

When the response point approaches the compressive characteristic point Y' and moves on the
elastic slope in compression, the response moves toward a point Y" from a point P of deformation

22
Dp :

D p = Dyc + β ( Dx − Dyc ) (27)

where, β: parameter for stiffness hardening point (=0.2), Dx : deformation at unloading stiffness
changing point. This rule is introduced only to reduce an unbalanced force by a sudden stiffness
change at compressive characteristic point Y’. The compressive characteristic point Y’ will be
maintained under any loading history.

This axial-stiffness hysteresis model was used for the axial deformation of an independent column
as well as a boundary column of a wall. The initial response point located in the compression zone
because a column carried gravity loads.

(4) Origin-Oriented Hysteresis Model: A hysteresis model which dissipates small hysteretic energy
was used for the rotational and horizontal springs at the base of the central vertical element of a wall
model.

The response point moves along a line connecting the origin and the previous maximum response
point in the direction of reloading (Fig. 23). Once the response point reaches the previous maximum
point, the response point follows the skeleton force-deformation relation renewing the maximum
response point. In this model, no residual deformation occurs, and the stiffness changes when the
sign of resistance changes. No hysteretic energy is dissipated when the response point oscillates
within a region defined by the positive and negative maximum response points. The skeleton curve of
this model can be of any shape.

Table 4: Skeleton Moment-rotation Relations of Beams

Stiffness Properties Top in Tension Bottom in Tension

Cracking Moment (tonf-m) 9.1 4.2 (6.6)*
Cracking Rotation (x10-3x l rad) 1.08 0.50 (0.79)*
Yield Moment (tonf-m) 43.9 9.8
-2
Yield Rotation (x10 x l rad) 1.08 0.52
Note: Elastic deformation included in rotation.
l : span length of beam
*: average values used in the analysis

23
 Table 5: Initial Axial Loads in Vertical Members
(a) Independent Columns, tonf
Story C1* C1’ C2 C3* C3’
1 87.9 70.1 96.8 127.9 92.3
2 74.2 59.4 82.8 107.7 78.1
3 61.1 49.2 69.1 88.3 64.6
4 47.9 39.0 55.3 68.9 51.2
5 34.8 28.8 41.6 49.5 37.7
6 21.6 18.6 27.8 30.2 24.2
7 8.4 8.4 14.0 10.8 10.8
Note: Column notation given in Fig. 3.5
* Loading-side column carried additional weight of actuators and loading beam

(b) Shear Wall and Boundary Columns, tonf

Story Boundary Wall
Column C4 Panel W1
7 97.5 87.2
6 83.6 73.8
5 69.9 61.2
4 56.2 48.6
3 42.5 36.0
2 28.8 23.4
1 15.1 10.8

Table 6: Calculated Stiffness Properties of Columns

Type C1 C2 C3
Story N, Mc, My θy N Mc, My θy N Mc, My θy
tf tf-m tf-m tf tf-m tf-m tf tf-m tf-m
7 8.4 4.9 18.5 2.50 14.0 5.4 19.8 2.67 10.8 5.1 19.0 2.57
6 20.1 5.9 21.3 2.88 27.8 6.5 23.1 3.12 27.2 6.4 22.9 3.09
5 31.8 6.8 24.0 3.24 41.6 7.7 26.2 3.34 43.6 7.8 26.6 3.59
4 43.5 7.8 26.6 3.59 55.3 8.8 29.1 3.93 60.6 9.2 30.0 4.05
3 55.2 8.8 29.1 3.93 69.1 9.9 31.9 4.31 76.5 10.5 33.0 4.46
2 66.8 9.8 31.4 4.24 82.8 11.1 34.5 4.66 92.9 11.9 36.3 4.90
1 79.0 10.8 33.8 4.56 96.8 12.3 37.0 5.00 110.1 13.3 39.3 5.31
Yield rotation θ y in 10-3 rad for a unit length column.

Table 7: Axial Stiffness Properties for Shear Wall (Outside Truss Element)

Story Elastic Stiffness Tension Yield

Compression (tonf/m) Tension (tonf/m) Load (tonf)
First Story 158,000 142,000 109.3
Second through 198,000 178,000 109.3
Seventh stories

24
 Table 8: Stiffness Properties of Shear Wall (Central Element)

(a) Shear Stiffness Properties

Elastic Shear Cracking Ultimate Yield Story
Rigidity Ks Shear Shear Displacement
Story (tonf/cm) (tonf) (tonf) (mm)
First story 2,770 238 381 4.17
Second story 3,400 238 381 3.33
and above

(b) Axial Stiffness Properties

Elastic Stiffness Tensile Yield
Compression Tension Load
Story (tonf/cm) (tonf/cm) (tonf)
First story 5,690 5,120 121.4
Second story 7,110 6,400 121.4
and above

(c) Rotational Stiffness Properties

Story Elastic Rotation Cracking Yielding Yielding
Stiffness Moment Moment, Rotation
(tonf-m/rad) (tonf-m) (tonf-m) X10-5 l rad
1 1920000 8.1 144.7 2.75
2 17.6 154.2 2.84
3 27.0 163.6 2.93
4 2400000 36.5 173.1 3.01
5 45.9 182.5 3.10
6 55.4 192.0 3.19
7 65.4 202.0 3.23
l : span length

Table 9: Stiffness Properties of Vertical Spring for Transverse Beams

Elastic Spring Cracking Force Yield Force Yield Displacement

(tonf/cm) (tonf) (tonf) (cm)
14.88 2.0 5.6 1.16

25
 Fig. 15: Notation for Effective Width Evaluation

Fig. 16: Stress-strain Relationships assumed in the Flexural Theory

Fig. 17: Tributary Area for Gravity Axial Load Computation of Columns and Wall
26
Fig. 18: Notation for Shear Wall Section

Fig. 19: Takeda Hysteresis Model

27
Fig. 20: Observed Behavior of Beam-to-column Connection with Slab

Fig. 21: Takeda-slip Hysteresis Model

28
Fig. 22: Axial Stiffness Hysteresis Model

Fig. 23: Origin-oriented Hysteresis Model

29
15.5 Method of Response Analysis

The seven-story test structure was idealized as three parallel plane frames with beams, columns
and walls represented by corresponding member models. The transverse beams connecting the
shear wall boundary columns and adjacent parallel frames were idealized by vertical springs. A
routine stiffness method was used in the analysis.

Floor slab was assumed to be rigid in its own plane, causing identical horizontal displacements of
all the joints in a floor level. The mass of the structure was assumed to be concentrated at each floor
level.

Vertical displacement and rotation were two degrees of freedom at each joint. The frames and a
shear wall were assumed to be fixed at the base of the structure.

A numerical procedure was developed to simulate the "equivalent" single-degree-of-freedom

pseudo-dynamic earthquake response procedure. The mode shape {u} , participation factor β and
resistance distribution {v} were taken from the test as outlined in Section 3.4.

The test structure changed its stiffness continuously with applied displacement even in a short
time increment, whereas the analytical model assumed a constant tangent stiffness during the time
increment. Therefore, the unbalanced forces, caused by overshooting at a break point of hysteresis
rules, must be released at the next time step. The analytical procedure from time step i to i+1 is
briefly outlined below.

Step 1: Determine displacement increment ∆qi+1 at the top floor using Eq. 10;

∆qi +1 = ∆qi − (∆t 2 / m) f i − ( β∆t 2 ) &&

yi (28)

where ∆qi = qi − qi −1 : incremental displacement at roof level from time step i-1 to i, ∆t : time
increment, m : effective mass,( = {u} [m]{u} ), f : effective resistance at time step i, ( = {u} {v} f R ),
T T

β : effective participation factor,( = {u}T [m]{1}/ m ), &y&i : ground acceleration at time step i, {u} :
assumed lateral deflection mode shape, {v} : assumed lateral resistance distribution, [ m] : mass
matrix, and {1} : vector consisting of unit element.

Step 2: Unbalance force correction. This step is necessary only when the stiffness changed
between time steps i-1 and i.
0
(a) Calculate displacement vector {x } due to unit load {v}

{x 0 } = [ K ]i−1{v} (29)

(b) Calculate displacement vector {x '} due to unbalanced force {F '} , at time step i,

{x '}i +1 = [ K ]i−1{F '}i (30)

where [ K ]i : tangent stiffness matrix evaluated at time step i.

Step 3: Determine incremental lateral resistance ∆f R i +1 at the roof level,

∆f R i +1 = ( ∆qi +1 − x 'R i +1 ) / xR0 (31)

30
 0 0
in which xR ' and xR are the values of vectors {x '} and {x } evaluated at the roof level.

Step 4: Calculate displacement increment {∆x}i due to incremental load and unbalanced forces
{F '}i ,

{∆x}i = {x 0 }∆f R i +1 + {x '}i +1 (32)

Step 5: Calculate incremental member forces from incremental joint displacement and
tangent member stiffness. Check if member stiffness changed during time step i and i+1.

Steps 1 through 5 were repeated for each time step. Computed response was temporarily stored
on computer files, and plots of response waveforms and force-deformation curves were made if
necessary.

31
15.6 Results of Analysis

The analytical models and procedure described above was applied to the full-scale seven-story
test structure. Equivalent single-degree-of-freedom pseudo-dynamic tests PSD-2 through PSD-4
were simulated continuously so that the structural damage in the preceding test runs could be
reflected in the analysis of the following test runs. In other words, the analytical model was given the
calculated residual displacement and structural damage from the immediately preceding test run.
Initial velocity of the analytical model was set to be null at the beginning of each test run. The actual
test was conducted in the same manner.

Pseudo-dynamic test PSD-1 was not analyzed herein because the test run was carried out to
examine the reliability of the testing technique and procedure. The maximum roof-level displacement
was as small as 1/8600 of the total structural height. The structure was observed to remain in the
elastic range during the test run. Therefore, the test run was not included for study here.

The analytical response of test PSD-3 was studied in detail to examine the reliability of the
analytical method. The structure was subjected to a displacement beyond the formation of collapse
mechanism during the test run. The roof-level displacement was observed to reach 1/91 the total
story height, a displacement which may be expected from this type of a structure during a "strong"
earthquake motion. This test included a wide range of response prior to and after the yielding of
various members and the shear wall. The structure did not have non-structural elements. These were
major reasons to choose this particular test run for careful inspection. Studied are the analytical
results of (a) response waveforms at the roof level displacement and base shear, (b) hysteresis
relation between base shear and roof-level displacement, (c) base shear-local deformation relations,
and (d) force and deformation distribution at maximum computed response.

The calculated response waveforms and the roof-level displacement vs. base shear relations are
briefly compared with the observed for tests PSD-2 and PSD4.

Response Waveforms (Test PSD-3): The artificial earthquake accelerogram based on EW

component of the Taft record (1952) was used in test PSD-3. The higher frequency components were
removed from the original record so that the first mode should govern the response of the test
structure. The maximum ground acceleration was 320 Gal (cm/sec2).

No damping was assumed in the test structure in the pseudo-dynamic response computation
during the test. Observed and calculated response waveforms are compared for the roof-level
displacement and base shear as shown in Fig. 24. The input base motion is shown in the same figure.
Note that the base motion oscillates in relatively low frequency compared with the original earthquake
record.

Response waveforms observed in test PSD-3 are shown in broken lines. Analytical responses are
in good agreement with the observed response over the entire duration of earthquake excitation.

Maximum displacement at roof-level was 238 mm from the test attained at 4.48 sec, while the
calculated maximum amplitude was slightly larger (= 248 mm) than the observed. Both maxima
occurred at the same time step. The period of oscillation elongated significantly after this time step in
the test and analysis.

At 10.16 second, the ground motion input was terminated in the test, and pseudo-dynamic
free-vibration test was started with existing residual displacement and no velocity. In the free vibration
range, the period of the analytical model appeared slightly longer than that of the test structure.

Maximum base shear of 414 tonf was attained at 4.48 sec in the test. The computed value was
425 tonf, slightly higher than the observed. Maximum base shear amplitude of an analytical model
can easily be controlled by choosing yield resistance level and post-yielding stiffness of constituent
members, especially of beam members in this analysis.

32
 Parametric studies by varying beam yield resistance and post-yielding stiffness indicated that the
combination of the values described in Chapter 5 was most suited to the test structure; i.e., the beam
yield resistance to be computed with the contribution of slab reinforcement within an effective width of
430 cm and post-yielding stiffness to be 3 % of the initial elastic stiffness.

SDF Hysteresis Relation (Test PSD-3): Roof-level displacement and base shear were the
corresponding force and displacement in the "equivalent" single-degree-of-freedom pseudo-dynamic
test, and their relation may be called "SDF hysteresis" as shown in Fig. 25.

As can be expected from a good correlation of observed and computed response waveforms, the
observed and computed hysteresis relations as an equivalent single-degree-of-freedom system are
in fair agreement, especially at the peaks of hysteresis loops. General shapes of the two curves are
slightly different; the stiffness of the test structure changed gradually during unloading, whereas the
stiffness of the analytical model changed when the sign of resistance changed. The latter stiffness
change was associated with that of member hysteresis models such as Takeda, Takeda-slip, and
Origin-oriented hysteresis models. The analytical model showed some pinching behavior, which was
also appreciable in the observed hysteresis relation. The pinching behavior of an analytical model
was caused by Takeda-slip hysteresis model used with beam one-component models and
axial-stiffness hysteresis model used with vertical line elements in the shear wall.

Local Deformations (Test PSD-3): During the pseudo-dynamic tests, local deformations of
members were measured at various locations of the test structure; (a) flexural rotation at beam ends,
(b) flexural rotation at column base, (c) elongation of boundary columns of the wall, and (d) shear
deformation of the wall panel. Computed local deformations of typical members were compared with
the observed deformations so as to examine the reliability of the analysis method.

(1) Beam End Rotation: Rotations at beam ends were determined from the axial elongation and
compression measurements by two displacement gauges, one placed above the slab face and the
other placed below the beam, parallel to the beam member axis (Fig. 26.c). The gauge length was
one half the effective beam depth from the column face.

The observed base shear-beam end rotation relation of a sixth floor beam at the wall connection is
shown in Fig. 26.a. The calculated relation is shown in Fig. 26.b. The calculated and the observed
relations do not necessarily agree because the beam end rotation was measured for a given gauge
length, whereas the rotation was calculated for an entire beam under imaginary anti-symmetric
loading condition. In other words, the calculated deformation corresponds to the deformation over
one-half span length of the beam. Therefore, the measured deformations were generally smaller, and
approximately 60 to 70% of the calculated amplitudes. General shapes of the base shear-beam end
rotation relation curves of the two were similar.

The beam was subjected to larger deformation in the negative loading direction when the
connecting tension-side boundary column moved upward. The upward displacement of a boundary
column joint (node) was significantly larger than the downward displacement because the bending
deformation of a wall was mainly attributable to the elongation of a tension-side boundary column.
Both observed and calculated beam-end response show this behavior.

Negative maximum deformations were larger than the positive deformation, although positive and
negative amplitudes of overall structural displacement were comparable. Negative deformation
amplitudes at the two ends of the beam were comparable, whereas the positive deformation at the
wall end was approximately 1.3 times larger than that of the behavior observed at the further end (left
end); the behavior was observed both in the measured and calculated beam-end rotations. At the
exterior column-beam joint, beam negative moment capacity was large due to the participation of slab
reinforcement. Hence, the exterior column was subjected to higher bending moment under the
positive loading (load applied from right to left), and experienced a larger rotation at column ends.
Therefore, nodal rotation at the exterior beam-column joint under the positive loading was smaller,
resulting in a smaller beam-end rotation at the exterior column end.

33
 Figure 27 shows the beam-end rotations at a sixth floor beam in Frame A. The observed
beam-end rotation amplitudes were smaller than the calculated amplitudes.

(2) Column Axial Deformation: A large vertical displacement was observed at the top (roof level) of
the tension-side boundary column of the wall during the test. Large axial elongations were measured
in the tensile region of the wall, especially at lower stories. Compressive axial deformations in the
corresponding region were small under opposite direction loading. Larger deformation was observed
in a transverse beam connected at the tensile edge (boundary column) of the wall. The boundary
columns were measured to elongate as much as 44 mm in the first story as shown in Fig. 28.a,
whereas the maximum compressive deformation reached only 5 mm.

Computed axial deformations of the boundary column, as expressed as the deformation of outer
truss elements, are shown in Fig. 28.b. General deformation amplitudes and hysteresis shapes of the
analytical model agree reasonably well with those of the test structure. The computed axial
deformation was larger.

Response at Maximum Displacement (Test PSD-3): It is important from design point of view to
estimate possible force amplitudes and deformation ductility factors at various critical sections of the
test structure at maximum deformation. However, member forces could not be measured in the test.
The frame analysis method may be applied to estimate these quantities. The maximum deformation
of the test structure was observed as well as calculated to occur at 4.48 see of the earthquake time.

(1) Member Forces: Member forces in wall-frame B calculated at maximum structural deformation by
the analytical model are shown in Fig. 29. The wall carried smaller shear forces in the first story than
in the second story.

Vertical forces transferred by transverse beams to the wall boundary columns are also shown in
the figure. Yield force level was reached by the transverse beam connected to the boundary column
in tension.

(2) Member Ductility: Ductility factors are defined in the analysis as a ratio of maximum deformation
amplitude to the calculated yield amplitude. Figure 30 shows the distribution of ductility factors at the
maximum structural deformation for frames A and C and frame B.

In open frames A and C (Fig. 30.a), almost all beam ends yielded at the maximum displacement
except those at the roof level. Under this deformed configuration, the top chord was in tension at the
left end of a beam, and the bottom chord was in tension at the right end. Ductility factors, ratios of
beam end rotations to the yield rotation, ranged from 0.8 to 1.5 at the left end of the beams, and from
2.3 to 4.7 at the right end. The rotation amplitudes at the left and right ends of the beams are
comparable. The difference in ductility factors at the two ends of a beam was caused by the
difference in the yield rotations at the two ends (see Table 4); the yield: rotation amplitude under
negative moment (top chord in tension) is approximately twice as much as that under positive
moment (bottom chord in tension).

Ductility factors at the same end (left or right) of the beams varied with the level of the beam; the
ductility factor decreased with the beam level. A beam end rotation appeared to be inversely related
to the column end rotation of the joint. Ductility factors of beam ends were smaller at the upper floor
levels where the columns yielded, and they were larger at the right exterior joints where the column
rotations were smaller.

In wall-frame B (Fig. 30.b), all beams yielded. Under the deformed configuration, the top chord
was in tension at the left end of a beam, and the bottom chord in tension at right end. The distribution
of beam end ductility factors was relatively uniform along the height; 1.4 to 1.7 at left exterior beam
ends, 4.1 to 4.7 at beam ends immediately left of the wall, 3.2 to 3.4 at beam ends immediately right
of the wall, and 6.5 to 7.9 at right exterior beam ends. The beam end rotation was generally larger in
the right exterior span than that at the corresponding end in the left exterior beams, which was

34
caused by the large vertical displacement along the tensile boundary column.

(3) Deflected Shape: Observed and calculated deflected shapes at maximum structural deformation
are compared in Fig. 31. A good agreement can be noted at every floor level. The deflection mode
shape used for the equivalent single-degree-of-freedom pseudo-dynamic earthquake test slightly
deviated at lower floor levels.

Analysis of Test PSD-2: The maximum roof-level displacement during the second test run (PSD-2)
reached 1/660 of the total structural height, or 33 mm. An artificial earthquake motion, modified from
the NS component of the Tohoku University record measured during the 1978 Miyagi-Oki Earthquake,
was used with the maximum acceleration amplitude of 105 Gal (cm/sec2). The calculated response
waveforms and equivalent SDF hysteresis relation are examined below.

(1) Response Waveforms: Observed and calculated roof-level displacement and base shear
waveforms are compared in Fig. 32. The analysis indicated that the test structure responded
elastically up to 1.5 sec, and then started to suffer damages. The calculated response waveforms
(solid lines) are in good agreement with the observed (broken lines) in the first 2.5 sec, and then
significantly deviates from the observed.

The maximum roof-level displacement of 32.9 mm was observed at 2.03 sec, while the maximum
amplitude of 36.5 mm was calculated at 2.06 sec. The maximum base shear of 224 tonf was attained
at 2.01 sec in the test. The maximum displacement and base shear did not occur at the same time in
the test. The maximum base shear of 219 tonf, slightly smaller than the observed, was calculated at
the same time as the calculated maximum displacement.

The calculated residual displacement at the termination of the base motion was so small that the
free vibration response was not excited in the analysis.

(2) Equivalent SDF Hysteresis Relation: Observed and calculated roof-level displacement vs. base
shear relation is compared in Fig. 33. Note that the two curves are similar. However, a careful
inspection reveals that calculated stiffness and resistance (solid lines) were generally lower than the
observed (broken lines). The calculated stiffness in a small amplitude oscillation following a large
amplitude excursion was lower, which may be a major cause to create the discrepancy in the two
waveforms (Fig. 32) after 2.5 sec.

Analysis of Test PSD-4: After test PSD-3, the roof-level displacement during test PSD-4 reached as
large as 1/64 of the total story height, or 342 mm. The EW component of the Hachinohe Harbor
record measured during the 1968 Tokachi-Oki Earthquake was used in the test with the maximum
acceleration of 350 Gal. The analysis was carried out continuously using the PSD-2, PSD-3 , and
PSD-4 input motions. Calculated and observed response waveforms and equivalent SDF hysteresis
relations are compared below.

(1) Response Waveforms: Observed and calculated roof-level displacement and base shear
waveforms are compared in Fig. 34. Note the good agreement of the two waveforms over the entire
duration of the test.

Maximum roof-level displacement reached 342 mm at 4.36 sec during the test, while the maximum
amplitude of 391 mm was calculated at 4.33 sec. Observed maximum base shear of 439 tonf was
attained at 2.52 sec, much before the maximum displacement was attained. The base shear at the
maximum displacement amplitude was observed to be 433 tonf, almost of the same amplitude as the
observed maximum base shear. The maximum base shear of 463 tonf was calculated at 4.33 sec,
slightly larger than the observed. The calculated and observed waveforms oscillated in the same
phase with a common dominant period of 1 .36 sec.

(2) Equivalent SDF Hysteresis Relations: Observed and calculated roof-level displacement vs.
base shear relations are compared in Fig. 35. The two hysteresis curves show a pinching behavior at

35
low stress levels. As expected from the good agreement in the response waveforms, the two
hysteresis curves agreed well. The observed base shear in the positive direction was slightly lower
than that in the negative direction. Such degradation in resistance was not reproduced by the
analytical model.

36
 Fig. 24: Observed and Calculated Response Waveforms of Test PSD-3

Fig. 25: Observed and Calculated SDF Hysteresis Relation of Test PSD-3

37
Fig. 26: Observed and Calculated Beam-end Rotation at Wall Connection (Test PSDD-3)

38
Fig. 27: Observed and Calculated Beam End Rotation in Frame A (Test PSD-3)

39
Fig. 28: Observed and Calculated Axial Deformation of Wall Boundary Columns (Test PSD-3)

40
Fig. 29: Calculated Member Forces at Maximum Structural Deformation in Test PSD-3

(a) Frames A and C (b) Frame B

Fig. 30: Calculated Ductility Factors at Maximum Structural Deformation in Test PSD-3

41
Fig. 31: Deflected Shape at Maximum Structural Deformation in Test PSD-3

Fig. 32 Observed and Calculated Response Waveforms in Test PSD-2

42
Fig. 33 Observed and Calculated SDF Hysteresis Relations in Test PSD-2

Fig. 34: Observed and Calculated Response Waveforms in Test PSD-4

43
Fig. 35: Observed and Calculated SDF Hysteresis Relations in Test PSD-4

44
15.7 Concluding Remarks

A full-scale seven-story reinforced concrete structure was tested using "equivalent"

single-degree-of-freedom pseudo-dynamic earthquake response test procedure at Building Research
Institute, Tsukuba, as a part of U.S.-Japan Cooperative Research Program Utilizing Large Scale
Testing Facilities.

A nonlinear dynamic analysis method was used to simulate the observed behavior. The method
utilized three different member models for (a) beams and columns, (b) shear walls, and (c) transverse
beams, and four hysteresis models for elements of member models: (a) Takeda hysteresis model, (b)
Takeda-slip hysteresis model, (c) Axial-stiffness hysteresis model, and (d) Origin-oriented hysteresis
model.

A procedure was outlined as to the method to determine stiffness properties used for the analysis
on the basis of material properties and structural geometry.

The response of the test structure was computed by a numerical procedure specially developed to
simulate the "equivalent" single-degree-of-freedom pseudo-dynamic earthquake response test
procedure.

A good correlation was reported between the observed and calculated response when the
structure responded well in an inelastic range. However, it was felt more difficult to attain a good
correlation when the structural response reached barely yielding.

The method of nonlinear dynamic analysis of reinforced concrete buildings can be made
significantly reliable not only to outline the overall structural behavior, but also to describe the local
behavior.

References:

1) Wight, J.K. and S. Nakata, “Construction of the Full-scale Seven-story Reinforced Concrete Test
Structure,” Report Presented during The Second Joint Technical Coordinating Committee,
U.S.-Japan Cooperative Earthquake Research Program Utilizing Large-Scale Testing Facilities,
Tsukuba, Japan, 1982.

2) Okamoto, S., S. Nakata, Y. Kitagawa, M. Yoshimura and T. Kaminosono, “A Progress Report on the
Full-scale Seismic Experiment of a Seven-story Reinforced Concrete Building - Part of the
U.S.-Japan Cooperative Program,” BRI Research Paper No. 94, Building Research Institute,
Ministry of Construction, 1982.

3) Otani, S., “Nonlinear Dynamic Analysis of Reinforced Concrete Building Structures,” Canadian
Journal of Civil Engineering, Vol. 7, No. 2, 1980, pp. 333 - 344.

4) Umemura, H. and H. Takizawa, “A State-Of-the-Art Report On the Dynamic Response of

Reinforced Concrete Buildings,” Structural Engineering Documents 2, IABSE, 1981.

5) Giberson, M. F., “The Response of Nonlinear Multi-Story Structures Subjected to Earthquake

Excitation,” Earthquake Engineering Research Laboratory, EERL Report, California Institute of
Technology, Pasadena, California, 1967.

6) Clough, R. W., K. L. Wilson, “Inelastic Earthquake Response of Tall Building,” Proceedings, Third
World Conference on Earthquake Engineering, New Zealand, Vol. II Section II, 1965, pp. 68 - 89.

7) Otani, S. and M. A. Sozen, “Behavior of Multistory Reinforced Concrete Frames during

Earthquake,” Structural Research Series No. 392, Department of Civil Engineering, University of
Illinois, Urbana, 1972.

45
8) Takizawa, H., “Strong Motion Response Analysis of Reinforced Concrete Buildings (in Japanese),”
Concrete Journal, Japan National Council on Concrete, Vol. 11, No. 2, 1973, pp. 10 - 21.

9) Omote, Y. and T. Takeda, “Nonlinear Earthquake Response Study on the Reinforced Concrete
Chimney - Part 1 Model Tests and Analysis (in Japanese),” Transactions, Architectural Institute of
Japan, No. 215, 1974, pp. 21-32.

10) Takayanagi, T. and W. C. Schnobrich, “Computed Behavior of Reinforced Concrete Coupled

Shear Walls,” Structural Research Series No. 434, University of Illinois, Urbana, 1976.

11) Hiraishi, H., M. Yoshimura, H. Isoishi and S. Nakata, “Planer Tests on Reinforced Concrete Shear
Wall Assemblies - U.S.-Japan Cooperative Research Program -, Report submitted at Joint
Technical Coordinating Committee, U.S.-Japan Cooperative Research Program, Building
Research Institute of Japan, 1981.

12) Architectural Institute of Japan: AIJ Standard for Structural Calculation of Reinforced Concrete
Structures (Revised in 1982), 1950.

13) Hirosawa, M., “Past Experimental Results on Reinforced Concrete Shear Walls and Analysis on
them (in Japanese),” Kenchiku Kenkyu Shiryo No. 6, Building Research Institute, Ministry of
Construction, 1975.

14) Takeda, T., M. A. Sozen and N. N. Nielsen, “Reinforced Concrete Response to Simulated
Earthquakes,” ASCE, Journal of the Structural Division, Vol. 96, No. ST12, 1970, pp. 2557 - 2573.

15) Nakata, S., S. Otani; T. Kabeyasawa, Y. Kai and S. Kimura, “Tests of Reinforced Concrete
Beam-Column Assemblages, - U.S.-Japan Cooperative Research Program -,” Report submitted
to Joint Technical Coordinating Committee, U.S.-Japan Cooperative Research Program, Building
Research Institute and University of Tokyo, 1980.

16) Eto. H. and T. Takeda, “Elasto Plastic Earthquake Response Analysis of Reinforced Concrete
Frame Structure (in Japanese),” Proceedings, Architectural Institute of Japan Annual Meeting,
1977, pp. 1877 - l878.

17) Fujii, S, H. Aoyama and H. Umemura, “Moment-Curvature Relations of Reinforced Concrete

Sections Obtained from Material Characteristics (in Japanese),” Proceedings, Architectural
Institute of Japan Annual Meeting, 1973, pp. 1261 - 1262.

18) Kaminosono. T., S. Okamoto, Y. Kitagawa, S. Nakata, M. Yoshimura, S. Kurose and H. Tsubosaki,
“The Full-Scale Seismic Experiment of a Seven-story Reinforced Concrete Building, - Part 1, 2 -
(in Japanese” Proceedings, Sixth Japan Earthquake Engineering Symposium, 1982, pp. 865-880.

19) Sugano, S., “Experimental Study on Restoring Force Characteristics of Reinforced Concrete
Members (in Japanese),” Doctor of Engineering Thesis, Department of Architecture, University of
Tokyo, 1970.

46