HYSTERETIC MODEL OF ORDINARY AND HIGH-STRENGTH

REINFORCING STEEL

By Toader A. Balan; Member, ASCE, Filip C. Filippou,:z Associate Member, ASCE,

and Egor P. Popov,3 Honorary Member, ASCE

ABSTRACT: The hysteretic behavior of reinforced concrete structures depends to a large extent on the hysteretic
behavior of reinforcing steel. An accurate and computationally efficient numerical model of reinforcing steel is,
thus, very important in the analysis and evaluation of these structures under cyclic loads, including earthquake
loads. A new macroscopic hysteretic model of the short-term cyclic behavior or ordinary and high-strength
reinforcing steel is presented. The model is based on a uniaxial stress-strain relation that is expressed in terms
of natural stresses and strains, so that a single envelope curve governs the monotonic behavior in tension and
compression. The hysteretic model accounts for the degradation of strength properties with accumulation of
Downloaded from ascelibrary.org by Universita di Catania on 01/28/20. Copyright ASCE. For personal use only; all rights reserved.

plastic strains. The material parameters of the model are calibrated with monotonic tests of coupon specimens,
while strength degradation relations are derived from cyclic test data. Correlation studies of the model with
available experimental data for ordinary and high-strength reinforcing steel demonstrate the ability of the model
to simulate the hysteretic behavior of all types of reinforcing steel over a wide range of strain variations.

INTRODUCTION hysteretic behavior by explicit functions of stress or strain. In

The evaluation of the nonlinear hysteretic response of re-
inforced concrete structures under cyclic excitations and, in
particular, severe earthquake excitations necessitates the de-
velopment of accurate and computationally efficient models of
components and constituent materials. While component mod-
els might suffice for the determination of global parameters of
structural response, constitutive material models become nec-
essary in the local response evaluation and damage assessment
of existing and new structures. In well-designed new structures
and retrofitted old structures, steel dominates the response ei-
ther in the form of reinforcing steel or in the form of steel
jackets for columns and beam-column joints. The development
of an accurate and computationally efficient model for ordi-
nary and high-strength steel is, therefore, an important task in
the nonlinear response evaluation of these structures by finite-
element methods.
Several models of the cyclic stress-strain response of metals
have been proposed in the last two decades. Following Popov
and Ortiz (1979), these can be generally classified in two ma-
jor categories: (1) Macroscopic models, based on measured
stress-strain relations; and (2) microstructure or microscopic
models, based on dislocation theories. Discussions on existing
models can be found in "Response" (1983), Bate and Wilson
(1986), Chang and Mander (1994), Dodd and Restrepo-Posada
(1995). Microscopic models are typically derived from sound
theories, but are overly complex for use in the nonlinear anal-
ysis of large-scale structures. On the other hand, macroscopic
models are relatively simple and fail to represent important
features of the hysteretic behavior. In the latter case, accuracy
is increased by the introduction of many parameters, with no
clear physical significance and with significant penalty in com-
putational efficiency.
The most widely used reinforcing steel models describe the
'Prof., Dept. of Civ. Engrg., Tech. Univ. of Moldova, Kishinev, Mol-
dova, 277060.
2Assoc. Prof., Dept. of Civ. and Envir. Engrg., Univ. of California,
Berkeley, CA 94720-1710.
'Prof., Grad. School, Dept. of Civ. and Envir. Engrg., Univ. of Cali-
fornia, Berkeley, CA.
Note. Associate Editor: John B. Mander. Discussion open until August
1, 1998. To extend the closing date one month, a written request must
be filed with the ASCE Manager of Journals. The manuscript for this
paper was submitted for review and possible publication on June 10,
1996. This paper is part of the Journal of Structural Engineering, Vol.
124, No.3, March, 1998. ©ASCE, ISSN 0733-9445/98/0003-0288-
0297/$4.00 + $.50 per page. Paper No. 13404.
288/ JOURNAL OF STRUCTURAL ENGINEERING / MARCH 1998
this context there are two approaches: one using the consti-
tutive relation in the form £ = /(a) (Ramberg and Osgood
1943); the other, in the form a = /(E) (Menegotto and Pinto
1973; Chang and Mander 1994; Dodd and Restrepo-Posada
1995). The latter offers significant computational advantages
when the finite-element formulation is based on geometric
(kinematic) approximations, as is the case in displacement-
based finite elements and cross-section models with geometric
constraints, like the widely used Bernoulli hypothesis of beam
section deformations. The Ramberg-Osgood form of the ex-
plicit relation offers computational advantages when the finite-
element formulation is based on force approximations, as is
the case in flexibility-based finite elements.
A new, simple macroscopic constitutive model for ordinary
and high-strength reinforcing steels is presented here. The
model has the following characteristics: (1) It can be formu-
lated as either an explicit function of strain or an explicit func-
tion of stress; (b) it is based on material parameters that are
readily available from monotonic coupon tests; (c) it is com-
putationally efficient and displays excellent agreement with
available experimental data over a wide range of strain his-
tories and plastic strains. The model is based on an uniaxial
stress-strain relation that is expressed in terms of natural stress
and strain, so that a single envelope curve governs the mon-
otonic behavior in tension and compression. The same func-
tion governs the monotonic and hysteretic behavior of the
model by appropriate coordinate transformations. The hyster-
etic model accounts for the degradation of strength properties
with accumulation of plastic strains. The material parameters
of the model are calibrated with monotonic tests of coupon
specimens, while strength degradation relations are derived
from cyclic test data. The effects of local buckling and low
cycle fatigue behavior of reinforcing steel are outside the
scope of this paper. Future developments will extend the
model to include these effects.
MONOTONIC STRESS-STRAIN RESPONSE OF
REINFORCING STEEL
Tension and Compression Responses
The general shape of the uniaxial stress-strain diagram for
a mild reinforcing steel bar loaded in tension to failure is well
established from numerous experimental studies (Leonhardt
1980; Naaman 1982; Nawy 1995; Dodd and Restrepo-Posada
1995). As shown in Fig. I, it can be represented by four well-
pronounced regions:

J. Struct. Eng., 1998, 124(3): 288-297

 120
Ylild plltllu
defined in the natural, or true, coordinate system, the differ-
Slrlin-hordlnlng ,eglon POlt-ultlrnotl _ , Ion ences basically disappear. Data for monotonic loading of duc-
! 100 tile reinforcing steels (Aktan et al. 1973; Kent and Park 1973;

I 80
Ma et al. 1976; Panthaki 1991; Dodd and Cooke 1992;
Restrepo-Posada et al. 1994; Dodd and Restrepo-Posada 1995)
! assert that the tension and compression stress-strain curves
practically coincide up to the ultimate stress (point of plastic
60 instability when neckinglbarreling develops in the tension!
compression test).
40
Engineering and Natural Coordinates
20 Engineering stress and strain, a and E, respectively, are re-
lated to the original configuration of the specimen and are
0 defined as
0 0.02 0.04 0.06 0.08 0.1 0.12
f
Downloaded from ascelibrary.org by Universita di Catania on 01/28/20. Copyright ASCE. For personal use only; all rights reserved.

Steel Strain (Inlin) a :0

= e =~ dL = L ~~ = ~ (1)
FIG. 1. Monotonic Curve for Mild Reinforcing Steel In Tension where Lo and L = initial and instantaneous length of the spec-
imen, respectively; aL = change in length of the specimen
1. The linear elastic region, defined for 0 S E. S Ey, where with respect to its initial length; F = axial force on the spec-
E. represents the generic steel strain and Ey is the yield imen; and A o = initial cross-sectional area.
strain. This region of the tensile stress-strain curve is According to the original definition by Ludwik (1909), the
defined by a straight line I. = E.E.. where I. represents natural strain, e, is the sum of the incremental strains Ae:
the steel stress and E. is the initial elastic modulus. The
yield strain Ey is defined as Ey = I,IE.. where fy is the yield (2)
stress.
2. The Liiders or yield plateau, defined for Ey < E. S E.h ,
where E.h marks the beginning of the strain hardening In the limit, as aL ~ 0, the natural strain corresponding to
region. The plateau, shown in the inset of Fig. 1, is typ- the instantaneous length L becomes
ically assumed to be horizontal, although small stress
fluctuations are observed. The yield stress, 1" is com-
puted as an average value of the observed fluctuations.
e= f~ = In (~) (3)

3. The strain-hardening region, defined for E.h < e. S e••, Natural and engineering strains are related by the simple equa-
where E•• is the strain corresponding to the ultimate steel tion
stress. The definition of the strain E.h is not a straight-
forward matter. A dip is generally observed at the end of £ =In(1 + E) (4)
the yield plateau before the stress begins increasing as In the plastic region of the stress-strain response of a spec-
the response stabilizes along a smooth strain hardening imen subjected to axial tension (compression), the cross-sec-
curve. This implies that the assumption of the yield stress tional area gets smaller (increases) as the specimen elongates
and of the strain hardening model affects the strain E.h at (shortens). These behaviors are considered in the definition of
initiation of strain hardening (see Fig. 1). the true stress 0':
4. The postultimate stress or strain-softening region, de-
fined for E. > E... In this region the stress-strain curve _ F
depends on the location and gauge length over which a=- (5)
A
experimental data are collected (Marin 1962). It is com-
monly assumed that beyond the ultimate point C(E••, I..) where A = instantaneous cross-sectional area of the specimen.
the stress-strain curve does not provide any usable in- Assuming that the plastic strain involves no volume change,
formation. it follows that AolA = U~ = (1 + E), which allows one to
relate true and engineering stresses as follows:
Even though measured data on reinforcing bars show some
difference in the responses in tension and compression, it is 0' = a(1 + e) (6)
commonly assumed that the monotonic stress-strain compres- The tangent modulus in natural coordinates £, is obtained upon
sion and tension curves are identical (Park et al. 1972; Aktan differentiation of 0' with respect to e:
et al' 1973; Geniev et al. 1974; Karpenko 1976; Ma et al.
1976; Filippou et aJ. 1983). More recently, some researchers
have suggested a modified tension envelope to model the com-
dO' = (da
-= -de + -a)
- (l + e) 2
(7)
de l+e
pression behavior of reinforcing steel. Mander et al. (1984),
for example, relate the compression parameters to the tension or
parameters using empirical equations. In order to include the
theoretical difference between the cross-sectional area of the
bar at equivalent compressive and tensile strains, Spurr and
£, = (E' + 1 : e) (1 + e)2 (8)

Paulay (1984) propose that the compressive stress at a given
strain be defined as the corresponding tension stress multiplied
by (1 - 2E.).
The difference in the tension and compression responses of
steel specimens is affected by the coordinate system used to
represent the measured data. Experimental data are typically
presented in engineering coordinates. If stresses and strains are
where E, = daltlE = tangent modulus of the reinforcing steel
in engineering coordinates.
The main advantage of the natural coordinate system is that
the compression and tension stress-strain curves practically co-
incide, whereas the two curves are quite different in engineer-
ing coordinates. This concept is illustrated in Fig. 2, which
shows in the same quadrant compression and tension mono-
JOURNAL OF STRUCTURAL ENGINEERING / MARCH 1998 /289

J. Struct. Eng., 1998, 124(3): 288-297

 160 120
! 140
~
c.,
., 100
M 120 ~
80
"i
! 100 iii fy
60
80
Engrg. Str....Str.ln 40
60 curv. In tension

40 20

20 0
o £, 0.01 0.02 0.03 0.04
0
Downloaded from ascelibrary.org by Universita di Catania on 01/28/20. Copyright ASCE. For personal use only; all rights reserved.

0 0.02 0.04 0.06 0.08 0.1 0.12 Steel Strain (lnlin)

Steel Strain (lnlin)
FIG. 2. Tension and Compression Monotonic Curves
tonic stress-strain envelopes in both natural and engineering
coordinate systems. The figure is only qualitative and data are
not represented to scale. Natural compression and tension
stress-strain curves are very similar until buckling of the bar
becomes noticeable at a compression strain of approximately
6%. In tension the natural stress is larger than the engineering
stress (as the specimen elongates, the cross-sectional area de-
creases), while in compression the true stress is smaller than
the corresponding engineering stress (as the specimen short-
ens, the cross-sectional area increases).
FIG. 3. Stress-8traln Curve Defined by Eq. (9)
vertex point (shadowed region in the inset of Fig. 3). When
80 = 0, the hyperbola defined by (9) degenerates into a bilinear
curve. The definition of asymptotes to the stress-strain
monotonic curve has already been proposed by Dafalias and
Popov (1975) and Stanton and McNiven (1976) for an arbi-
trary envelope curve.
Eq. (9) can be inverted and expressed as a strain-stress re-
lation in the form

Proposed Monotonic Stress-Strain Curve where

The formulation of the proposed steel model is based on
the assumption that in the natural coordinate system the
monotonic curve in compression is equal and opposite to the
tension curve (Nadai 1950; Marin 1962; Dodd and Restrepo-
Posada 1995). The response of a reinforcing steel bar in ten-
sion is defined with respect to the following parameters: the
initial elastic modulus, E,; the slope of the symptote in the
strain hardening region, or strain hardening modulus, Eh ; the
yield stress and strain, J;, and €y, respectively; E'h' strain at the
onset of strain hardening; fsu, ultimate stress; and E,u' ultimate
strain. These parameters are all defined in the engineering co-
ordinate system with respect to the monotonic curve in ten-
sion. The corresponding curve in tension and compression in
a=l-p. r.I.=l+p
2 ' I-' 1 _ p (12)
In the preceding expression, the strain hardening ratio p in the
Ltiders plateau should be set equal to a very small positive
number rather than zero in order to avoid numerical problems.
Eq. (11) may be useful in flexibility-based elements, where the
strains are computed from known stresses. In the remainder
of the paper the proposed hysteretic model will be based on
the stress-strain relation defined by (9).
The relation of (9) can be extended to generate the idealized
tension and compression monotonic curves. In a single equa-
tion the following expression defines the linear elastic region,

r
the natural coordinate system is obtained from (4) and (6), the yield plateau and the strain-hardening region:
which are then used to find the monotonic curve in compres-
sion in the engineering coordinate system.
The proposed monotonic stress-strain relation for reinforc-
Is =1' (1 - p') [ 1 + (1 + p') E, - Eo -
y 2 (1 - p') E;
~ (E' -E Eo - 1 + 8']
y
ing steel in tension is based on the following function, for- (13)
mulated in the engineering coordinate system:
where

[1 + (1(1 -+ p)p) ~ - VI(~Ey - 1)2 + 8] I'={ E,e;h

Is =J; (1 - p) (9) for E, ::s; E,h
(14)
y 2 Ey y for E, > E'h

where p = Eh/E, = hardening ratio; and 8 = model parameter

defined later in this section. Eq. (9) represents a hyperbola p,~{~, for E, :5 E'h
for E, >E'h
(15)
with two asymptotes, one with slope E, and the other with E,
slope Eh , as shown in Fig. 3. This equation describes a family

{~(~-1)80
of "parallel" hyperbolas with two asymptotes that intersect at
point A. They are parallel in the sense that the slopes of the for E, ::s; E'h
8' _ 2 (16)
two asymptotes, E, and E h , are the same for all curves. These for E, > Esh
parallel curves depend on parameter 8, defined by -

8=~

~~{("-~J
(10)
1- p for E, ::s; E'h
(17)
where 80 can be interpreted as the area of the triangle bounded for E, > E'h
by the two asymptotes and the tangent to the hyperbola at its
290 I JOURNAL OF STRUCTURAL ENGINEERING I MARCH 1998

J. Struct. Eng., 1998, 124(3): 288-297

 I, f.
E.
E

1 E.
E.
S1raIn hardening
branch I. Tenelon
envelope cuve
I,

E,
I,
E, Compl'88llon I;
E.
envelope cuve I.
Downloaded from ascelibrary.org by Universita di Catania on 01/28/20. Copyright ASCE. For personal use only; all rights reserved.

FIG. 4. Monotonic Stress-5traln Curve for Reinforcing Steel

Fixed asymptote
E

e; =E,(1 ~ p') [(f,u - ,{y) - (E,u - E'h)E,p'] (18) FIG. 5. Tension and Compression Envelopes

The curve defined by (13) is illustrated in Fig. 4, which shows ing into account the shift/stretching transformation, (13) be-
all the parameters used in the definition. The tension curve is comes
first defined in the engineering coordinate system, and then
converted to the natural coordinate system using (4) and (6),
which are applied once again to obtain the monotonic curve ~ =1' (1 -
J' y 2
pi) [1 + (1 + pi)
(I - p')
E, -e; eo _ ~(E' e;- eo _ 1)2 + 8 1]

in compression. Eq. (13) will be used as "backbone" for the

general hysteretic material model for reinforcing steel pre- (19)
sented hereafter. where
CYCLIC BEHAVIOR OF REINFORCING STEEL
Several uniaxial models have been proposed to simulate the
I'-{hE,e;
Y -
for
for
E, - Eo :S Esh
E, - £0 > Esh
(20)

cyclic behavior of reinforcing steel (Park et al. 1972; Aktan et

al. 1973; Ma et al. 1976; Filippou et al. 1983; Chang and
Mander 1994). These models are typically based on either the (21)
Ramberg-Osgood law, or on the Menegotto-Pinto curve. Both
approaches, however, present some difficulties in describing
the cyclic behavior of different types of reinforcing steels un-
der generalized loading histories.
(22)
This section introduces a general purpose cyclic material
model for ordinary and high-strength reinforcing steel bars.
The proposed model is a rule-based model with the following
basic components. Eo for E, - Eo S E'h
• Envelope curves, which are assumed to be identical to the
eo-- { Eo + (E'h - ~,)
f,
for E, - Eo > E'h
(23)

monotonic stress-strain curves in the engineering stress

coordinate system. Starting from the tension envelope
curve in engineering coordinates given by (13), (4) and e; =E,O ~ pi) [(Ius - f,) - (Eus - E'h)E,p'] (24)
(6) are used first to obtain the two identical tension and
compression curves in the natural system, and then to Eq. (19) provides a single expression for the cyclic envelopes
compute the compression envelope in engineering coor- in tension and compression defined in terms of engineering
dinates. These curves are the "backbones" of the model. coordinates. Differentiation of (19) with respect to E, yields
Shifting is used to simulate cyclic strength degradation. the tangent modulus E, on the envelope branches in terms of
Shifting implies that the point of return to an envelope the engineering coordinates
curve is different from the point where the last reversal
started.
• Reversal branches, which are used to model unloading
and reloading within the envelope curves.
• Loading function, which is used to establish whether the
material is loading or unloading.
E,=E'(I _ p)
2
[0 0 -
+ p) _
p) ~(E
(¥- 1) ]
_ )2
£0
(25)

Cyclic Envelope Curves
The compression and tension envelopes of the cyclic stress-
strain curve are defined by the monotonic stress-strain relation
of (13), whose origin is shifted to point (Eo, 0) in order to
describe cyclic response (Fig. 5). In the following develop-
ments the strain £0 will represent the generic origin shift. Tak-
-'---1 +8 '
e;
It is important to note that in Fig. 5 the tension and com-
pression envelopes are represented with respect to the engi-
neering coordinates. Up to the end of the Liiders plateau the
two envelopes are identical. In the strain hardening regions,
the engineering stresses corresponding to identical engineering
strains are larger on the compression envelope.
JOURNAL OF STRUCTURAL ENGINEERING / MARCH 1998 /291

J. Struct. Eng., 1998, 124(3): 288-297

 Reversal Branches
When a load reversal takes place, the stress-strain behavior
is described by an unloading-reloading branch, or reversal
branch. The reversal branch is approximated by a hyperbola
that begins at the reversal strain E, (Fig. 6). The hyperbola has
two asymptotes. The first one, with slope Eu , describes the
initial unloading phase and its position depends on the reversal
point. The second asymptote, with slope Eh , has a fixed po-
sition and defines the envelope curve in the opposite loading
direction. In engineering coordinates, the reversal branch is
obtained by rewriting (9) in the following recurrence form:
applicable to both normal and high strength steel bars, is used
for k8:
k O= I ~ p [ 1 + a (:~r]
k (k == 0, 1, ... ,n) (29)
where k Ep == k E, - k-I E, = plastic strain amplitude defined as
the strain difference between two consecutive reversal points
(Fig. 6); 80 = initial value of 8; and ka = amplification factor
that will be defined later. Calibration of 80 was based on the
statistical analysis of available experimental tests on reinforc-
ing steel bars (Aktan et al. 1973; Kent and Park 1973; Ma et

is = %(l - p)
2
[I + (l + p) Es :- 'E,
(l - p) Ey
al. 1976; Panthaki 1991). Based on this analysis, the following
values of Eo were selected:
o _ {0.005 for ordinary steel (grades 40 and 60)
Downloaded from ascelibrary.org by Universita di Catania on 01/28/20. Copyright ASCE. For personal use only; all rights reserved.

0- 0.010 for high-strength steel (higher grade) (30)

- ~ (Es ~y'e, _ I Y+ O] k (k = 0, I, ... ,n)
(26) The tangent modulus kE, is obtained upon differentiation of
(26) with respect to lOs:
where superscript k indicates the unloading-reloading cycles:
p = EhlEu = instantaneous hardening; kEu = unloading mod-
ulus; and % and k ev = instantaneous yield stress and strain, (
E. - k
E
, _ I) ]
respectively. All the parameters carrying superscript k are up- kE, = 'Eu (l _ p) (I + p) _ 'Ey (31)
dated after each load reversal. The instantaneous yield stress
% is defined by the following expression:
2 [ (l - p) ~(~
_' )2 +kO
key - 1
%='E}Ey (k=O, I, .. . ,n) (27)
As reported by Bauschinger (1887), the steel modulus re-
where duces when steel is strained beyond the elastic limit. Available
experimental data indicate that the unloading modulus of re-
key =Ey _P_ k Eo an d k
Eo == ' E, - kY'(kO
=, 1, ... , n) (28)
inforcing steel E u also decreases, at a rate that is especially
I - p Eu high after first yielding, and then stabilizes at larger strains.
The plastic strain amplitude 'Ep and the unloading modulus
where kEo == strain at the intersection of the instantaneous un- appear, therefore, to be closely related. In this study it is as-
loading asymptote with the strain axis after the k-reversal; Ey sumed that the unloading modulus kEu varies according to the
== initial engineering yield strain; and % and k E, == stress and following expression given by Dodd and Restrepo-Posada
strain, respectively, at k-reversal point. The yield stress %and (1995):
the yield strain kEy are the coordinates of the point where un-
loading and strain hardening asymptotes intersect. ,
E u ==[
Es 0.83 + 0.001]
, (32)
In (26), k8 is the parameter that shapes the transition be- 0.006 + ( Ep - Ey )
tween the two asymptotes and allows for a good representation
of the Bauschinger effect. The parameter k8 is also called the Even though the above equation has been implemented in the
degradation parameter because it takes into account the cyclic proposed constitutive model, for all practical purposes the un-
strength degradation of the material. Degradation of the loading modulus kEu can be set equal to the initial stiffness Es •
strength properties is directly associated with plastic defor- Three types of reversals are defined in the proposed model:
mation, and the value of k8 is assumed to depend on the ratio reversals from the yield plateau of the envelope curves, re-
between the plastic strain amplitude and the instantaneous versals from the strain hardening region of the envelope
yield strain kEy. The following expression, which proved to be curves, and reversals from the reversal curves. Each reversal
curve type is described separately hereafter.
Is
Reversals from Yield Plateau
Fixed asymptote
When reversal takes place from the yield plateau of the
1 E, envelope branch, the reversal curve is described by a softened
envelope shifted in the opposite direction, as shown in Fig. 7.
The reversal branch is described by (19), starting at reversal
point ('En fy). The current asymptote of the shifted envelope
passes through point ('Eo, 0), where 'Eo is defined as:

£, 'Eo = 'E, - fy (k = 0, 1, ... , n) (33)

Es
As shown in Fig. 7, the plastic strain amplitude 'Ep in this case
is

Current aSYlT1Jtote
'ep = 'E, (k =0, 1, ... , n) (34)
Available experimental data from cyclic tests indicate that
the total strain of the Liiders plateau decreases under cyclic
loading conditions. The data also point to a relation between
the reduction of the Liiders strain and the complementary
FIG. 6. General Load Reversal Curve strain energy accumulated during load reversals. This relation
292/ JOURNAL OF STRUCTURAL ENGINEERING / MARCH 1998

J. Struct. Eng., 1998, 124(3): 288-297

 f, /, 1:-1£,
Fixed asyrT4llote ........ 1 E•
" .. "' ....
.. ' 1 E•

f,

E,

Current asyIllltole E,
Downloaded from ascelibrary.org by Universita di Catania on 01/28/20. Copyright ASCE. For personal use only; all rights reserved.

Current asyIlllIole

FIG. 7. Load Reversal from Yield Plateau .-1 eo

f,
Fixed asyrrptote FIG. 9. Load Reversal from Reversal Curve
1 E.
Reversals from Reversal Curves
A complete reversal is defined as any reversal from the en-
velope branches or a reversal from another complete reversal
curve in which the stress difference between the two consec-
E,
utive reversal points (Fig. 9) satisfies the condition
't - '-1. ~ 2f, (38)
If the above condition is not satisfied, the reversal is assumed
to be incomplete. Both complete and incomplete reversals fol-
low the same rules as the reversals from the strain hardening
-+--+0--1 ....- Current asyrT4llole region of the envelope. Thus, each one of these reversal curves
is described by the hyperbola of (9), starting at the reversal
point (te" %), as shown in Fig. 9. The curve has two asymp-
'e p totes, an initial one with slope tE., defined by (32), and the
strain hardening asymptote with fixed slope Eh • The two as-
FIG. 8. Load Reversal from Strain Hardening Region ymptotes intersect at point (leo + lEy, %). The instantaneous
asymptote passes through the shifted strain point (leo, 0),
is empirically accounted for in the model proposed by Ma et where the shift strain leo after the k-th reversal is defined by
al. (1976). In the present study, the Liiders strain is assumed (33). The current plastic strain amplitude is
to be a constant, independently of the number of performed
reversals (Thompson and Park 1978; Mander et al. 1984; Dodd (39)
and Restrepo-Posada 1995). A delay in the initiation of strain The only difference between complete and incomplete re-
hardening relative to the measured material response is there- versals is in the definition of the amplification parameter la,
fore to be expected with the proposed model. which is computed as:
Reversals from Strain Hardening Region 6 for y, - HI, ~ 2
When reversal from the strain-hardening region of the en- 'a = f, (40)
velope curve occurs, the reversal is described by a softened
hyperbolic curve, defined by (9), in which the hardening ratio
{ 3 for y, ~Hf, < 2
p is
The above empirical relation was obtained using statistical
(35) analysis on available experimental tests, and applies to both
normal and high-strength reinforcing steel bars. It should be
This hyperbola begins at the reversal point (le" %) and has an noted, however, that additional statistical data are required to
initial asymptote with slope lEu and a strain hardening asymp- ensure accuracy in the evaluation of both 80 and t a . A database
tote with slope E h • As one can see in Fig. 8, these two as- of these factors is needed in order to establish a sound theo-
ymptotes intersect at point (lEy, Yy ). The initial asymptote retical foundation for their definitions.
passes through point (\:0' 0), where
Loading, Unloading, and Reloading Criteria
,Eo = ,e, - 'E't
. (k =0, 1, ... ,n) (36) During loading cycles, the state of loading or unloading
affects the yield stress %, yield strain lEy, and the degradation
and the current plastic strain amplitude is parameter 18. The loading or unloading conditions are defined
by a loading function formulated in terms of engineering
'ep='e, (k=O, 1, ... ,n) (37) strains as follows:
JOURNAL OF STRUCTURAL ENGINEERING / MARCH 1998/293

J. Struct. Eng., 1998, 124(3): 288-297

 'l= Ie, ~kerl (41)
50
:! 40

where e, = current engineering strain; k er =strain at the current

reversal point; and key = current yield strain. The conditions
j : - Anolyllo

of material loading/unloading are I 10

-- Elq:JorImont

ojt-----------+----j
'l> 'lrnu loading; 'ls'lrnu unloading (42a,b) ·10
·20
where Ymax = maximum value of the loading function during
the loading history of the material. Under conditions of ma- ·30
terial loading, the current stress!s and the tangent modulus E, ·40 _.---- ----
corresponding to the current steel strain e, are determined ac- ·50 0L-~-0....J.0-02-----0. ..L004---0....LOO-:8-""'-0-...J.OO8-~---JO'Ol
cording to (19) and (25), respectively. Under conditions of
Steel Strain (irVin)
unloading and subsequent reloading, !s and E, are obtained
from (26) and (31), respectively, where the degradation param- a) Experimental and Analytical Response for Specimen 6
Downloaded from ascelibrary.org by Universita di Catania on 01/28/20. Copyright ASCE. For personal use only; all rights reserved.

eter k8 and the amplification factor ka are updated after every

load reversal.
50
MATERIAL MODEL VERIFICATION :! 40
30
A set of correlation studies between experimental data and j 20 - Anolyllo
-- e-tmont
analytical results on ordinary and high-strength reinforcing
steel bars under cyclic loading was performed to assess the I 10
0
accuracy of the proposed model for reinforcing steel. The anal- ·10
yses were conducted using a computer program that runs under ·20
imposed strain histories. Input material data include all the ·30
parameters that describe the tensile envelope curve in engi- ·40
neering coordinates. The reported studies are all based on the ·50
following assumptions: -80
-0.005 0 0.005 0.01 0.015 0.02 0.025
Steel Strain (IrVin)
• The number of cycles, including those of large strain am-
plitude, is too small for low-cycle fatigue to have an im- b) Experimental and Analytical Response for Specimen 8
portant influence on the ultimate state of the tested
specimens
• Buckling of the specimen has no effect on the test results 50
• Strain aging effects do not develop in any of the tests :!
I
~
The first set of analyses refers to a series of tests on normal- (J)

strength reinforcing steel coupons performed by Kent and Park

(1973). The tests are designated as specimens 6, 8, 15, and 17 I - Anolyoil
-- e-tmem
in the original work of Kent and Park. For these analyses, the
tensile steel material properties are as follows: h = 45.0 ksi; 0
!s. = 50.0 ksi; E, = 29,000.0 ksi; Eh = 290.0 ksi; 80 = 0.005;
e,h = 0.0155; and e,u = 0.14. Experimental and analytical re-
sults from the cyclic tests are illustrated in Fig. 10. The tests
on specimens 6 and 8 consider a single half-cycle with a large 0.008 0.01 0.012 0.014 0.018
maximum strain. While a simple half-cycle was applied to
Steel Strain (irVin)
specimen 6, specimen 8 was subjected to some unloading be-
fore being stressed to a maximum strain of 2%. Specimens 15 c) Experimental and Analytical Response for Specimen 15
and 17 were also loaded to a large maximum strain, but in-
complete stress reversals (one for specimen 15 and four for
50
specimen 17) were applied before loading to the maximum
strain. For all four specimens, the stress-strain responses pre- :! 40
dicted by the proposed model are in good agreement with the I
~
30
(J)
experimental results. 20
A second set of analyses was performed based on experi- :I
(J)
10
- Anolyol.
mental results from cyclic tests on machined ordinary rein- - - Elope".....

forcing steel coupons tested by Ma et al. (1976). Numerical 0

and experimental results for specimens I, 3, 4, and 5 are il- -10
lustrated in Fig. 11. For these specimens the analytical predic- ·20
tion is based on the following model parameters: h = 66.0 ·30
ksi; !s. = 95.0 ksi; E, = 29,000.0 ksi; Eh = 580.0 ksi; 80 =
·40
0.005; e,h = 0.014; and e,u = 0.123. In specimens 1 and 4 0 0.005 0.01 0.015 0.02 0.025
the strain hardening envelope is clearly visible after the yield S1Be1 Strain (IrVin)
plateau. Specimens 3 and 5 were subjected to several load
cycles of strain magnitude up to approximately 4.5 and 4%, d) Experimental and Analytical Response for Specimen 17
respectively. In spite of the complexity of the imposed strain FIG. 10. Measured and Modeled Stress-Strain Responses for
histories, the results obtained with the proposed model match Tests by Kent and Park (1973)
294/ JOURNAL OF STRUCTURAL ENGINEERING / MARCH 1998

J. Struct. Eng., 1998, 124(3): 288-297

 100 . - - - - - - - - - - - - - - - - - - - - , the measured data well. Strain hardening, Bauschinger effect,
I 80
and unloading stiffness are all closely traced.
The results of an extensive test program on the behavior of
~ ordinary reinforcing steel bars subjected to cyclic loading are
reported by Panthaki (1991). The comparison between experi-
J -
--
AnoJytio
EJlPerime'" mental and analytical results from cyclic tests on ordinary re-
inforcing steel coupons, denoted as specimens Rl and R5, is
ol---------/,;------:lff----j
illustrated in Fig. 12. For these specimens the steel material
-20 properties are defined by the following model parameters;
-40 h = 48.0 ksi; /.. = 82.0 ksi; E, = 31,200.0 ksi; Eh = 580.0 ksi;
-60 80 = 0.005; £'h = 0.0091; and £,. = 0.144. Both tests consist
of two complete cycles between fixed strain amplitudes, :!:3%
-80 0 0.005 0.Q1 0.Q15 0.02 0.025 0.03 0.035 0.04 0.045 0.05 for specimen Rl and :!:2% for specimen R5. Predicted and
Sleel Strain (irVin) measured experimental stress-strain responses are in good
agreement for both specimens. Panthaki (1991) also reports on
a) Experimental and Analytical Response for Specimen 1
Downloaded from ascelibrary.org by Universita di Catania on 01/28/20. Copyright ASCE. For personal use only; all rights reserved.

a series of experiments on high-strength reinforcing steel bars

under cyclic loading. Experimental and analytical results for
100
specimens P2 and P16 are illustrated in Fig. 13. For these
'i" 80
o!!. specimens, the steel material properties are defined by the fol-
lowing tensile model parameters: h = 126.0 ksi; f.. = 163.9
!
rJ)
60
40 ksi; E, = 32,100.0 ksi; Eh =700.0 ksi; 80 := 0.01; £'h =0.0039;
:I and £,. = 0.063. While specimen P2 was subjected to a single
rJ) 20
full cycle in tension, compression, and reloading in tension,
0 specimen P16 was cycled in tension only. For these compar-
-20 isons on high-strength steel bars, simulated and measured re-
-40 sults are in good agreement.
- - Eloporimonl The results of an extensive experimental investigation on
-60
the behavior of ordinary reinforcing steel bar coupons sub-
-80 jected to large strain reversals are discussed by Aktan et al.
-<l.02 -<l.01 o 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08
(1973). The comparison between experimental and analytical
Sleel Strain (inlin) results for reinforcing steel bar coupons, labeled as test 5 and
b) Experimental and Analytical Response for Specimen 4 6 in the original report by Aktan et al. (1973), are illustrated
in Fig. 14. For both specimens the analytical prediction is
100 , . . - - - - - - - , . - - - - - - - - - - - - - - - - - - - ,
100 r----------,,-------------,
80
80
60
60
40
40
20
20
o
-20
oI - - , - - - f - - - - - + - - - - - - I ' f - - - - I
-20
-40
-40
-60 - Ana/iOlS
ElCPOrimont -60
-80
-1 00 L-...........--J_~__J._~__'_ _ _........_~....L...___'"---' -80
-<l.01 0 0.01 0.02 0.03 0.04 0.05 -100 L--.-L-....o..--::-l--.......J.-~.............L--'-_L._~..L-o.-J
-<l.04 -<l.03 -<l.02 -<l.01 0 0.01 0.02 0.03 0.04
Steel Slrain (irVin)
Steet Strain (IrVin)
c) Experimental and Analytical Response for Specimen 3
a) Experimental and Analytical Response for Specimen Rl
100
! 75 t='
100
80
j
rJ)
50
!.,
60
:I 25 !
rJ) 40
-
--
Analpis
E......m...
rJ)
0 I
rJ)
20
·25 0
-20
-50
-Anal)'lis
-40
-75
---- E""orimont ·60
-100 -80
·0.01 0 0.01 0.02 0.03 0.04 0.05
-100
Steel Strain (In/in) -0.03 -<l.02 -0.01 0 0.01 0.02 C>.03
d) Experimental and Analytical Response for Specimen 5 Steel Strain (IrVin)

b) Experimental and Analytical Response for Specimen R5

FIG. 11. Measured and Modeled Stress-Straln Responses for FIG. 12. Measured and Modeled Stress-Strain Response for
Tests by Ma et al. (1976) Teata on Ordinary Reinforcing Steel Bars by Panthakl (1991)

JOURNAL OF STRUCTURAL ENGINEERING / MARCH 1998/295

J. Struct. Eng., 1998, 124(3): 288-297

 150 125 r-------,-----------,
'i" 100
O!. - Analysis
100

~
- - E,.,orimonl 75
50
SO
I 0
25
Or---jf-f---t-f-r--~-__j~--I-'---I

·25
·50
-so
-75
-100 -AnIIVI.
-100 .- .... - E""lrimllll
.125 '--......._~.......:_'~-'-_--'-~--'-~--'--J
'1~.0L..1-2---o..J.0-08---o--l..0-04~-0.L-~-0-.004'--~-0..J.00-8--0--'.012 ·0.04 -0.02 0 0.02 0.04 0.08
Steel Strain (1I1I1n) Steet Strain (1I1I1n)

a) Experimental and Analytical Response lor Specimen P2 a) Experimental and Analytical Response lor Test 5
Downloaded from ascelibrary.org by Universita di Catania on 01/28/20. Copyright ASCE. For personal use only; all rights reserved.

2OOr--r---------------, 125 r-------....,----------,

'i" 100
O!. 150

~
75
100 50

I so
25
Ot-t-t---H--tf"tt-H'-+-f-h/--f-t--+--f
ol--+---ft:......----------A~___j -2S
-50
-so
-75
-100 -100
-1 SO '__--'-_~--l._~___' '__~_.L_~___'
-125 --l._ ___'_.......L...-~_'_~..J...-'-......L.~--l___'
L..J;,.........

-0.08 -0.04 -0.02 0 0.02 0.04 0.08 0.08

o 0.01 0.02 0.03 0.04 0.05
Steel Strain (1I1I1n) Steel Strain (InIIn)

b) Experimental and Analytical Response for Specimen P16
FIG. 13. Measured and Modeled Stress-Strain Response for
Tests on High Strength Reinforcing Steel Bars by Panthakl
(1991)
based on the following model parameters: /.; = 70.0 ksi; Is. =
100.0 ksi; E s = 29000.0 ksi; E h = 580.0 ksi; 80 = 0.005; £sh =
0.011; and £s. = 0.29. A large number of plastic cycles were
applied in both tests, and in both cases the yield plateau and
the strain hardening regions are clearly visible in the results.
Predicted and experimental stress-strain responses are once
again in close agreement. However, as the cycles progress,
some discrepancies appear, probably related to the fact that
low-cycle fatigue is ignored in the proposed model.
CONCLUSIONS
A general-purpose hysteretic material model is proposed for
the simulation of the cyclic response of different types of or-
dinary and high-strength reinforcing steels. The model is sim-
ple and is formulated in terms of natural stresses and strains.
The formulation is based on the assumption that the cyclic
stress-strain behavior of reinforcing steel in the natural coor-
dinate system is identical in tension and compression up to
the point of plastic instability when necking (barreling) de-
velops in the tensile (compression) test.
A simple hyperbolic curve describes the entire envelope
curve, including both the yield plateau and the strain hardening
branch. This curve is simply redefined and shifted during the
load reversal cycles. Formulation of the model requires steel
material properties in tension only, thus making the model par-
ticularly easy to program and use for any study that requires
an uniaxial constitutive law for reinforcing steel. The accuracy
of the proposed model has been verified through comparisons
with several experimental tests on steel coupons.
Further studies are necessary to calibrate the model pa-
rameters governing material strength degradation. Limited ex-
b) Experimental and Analytical Response lor Test 6
FIG. 14. Measured and Modeled Stress-Strain Response for
Tests by Aktan et al. (1973)
perimental data are presently available and a larger database
is needed for a precise definition of the degradation parame-
ters. Low-cycle fatigue (Aktan et aI. 1973; Panthaki 1991;
Chang and Mander 1994) and buckling of the reinforcing bars
(Monti and Nuti 1992) will also be important enhancements
of the proposed model that will allow for the realistic analysis
of reinforced concrete structures subjected to seismic loading.
These extensions of the basic model proposed in this paper
are presently under investigation.
ACKNOWLEDGMENTS
The research reported in this paper was carried out while the first
writer was a visiting scholar at the Civil Engineering Department of the
University of California, Berkeley. This research was supported by the
Technical University of Moldova (UTM). The writers would like to grate-
fully acknowledge Ion A. Bostan, president of UTM, for his support. The
suggestions of E. Spacone from the Department of Civil, Architectural,
and Environmental Engineering at the University of Colorado, Boulder,
and his valuable comments about the model description are much appre-
ciated.
APPENDIX I. REFERENCES
Aktan, A. E., Karlsson, B. I., and Sozen, M. A. (1973). "Stress-strain
relationships of reinforcing bars subjected to large strain reversals."
Res. Rep. NSFRG G/ 29934, Dept. of Civ. Engrg., Univ. of Illinois,
Urbana, III.
Bate, P. S., and Wilson, D. V. (1986). "Analysis of Bauschinger effect."
Acta Metallurgica, 34(6), 1097-1105.
Bauschinger, J. (1887). "Variations in the elastic limit of iron and steel
[summarized translation]." J. Iron and Steel/nst., I, 442-444.
Chang, G. A., and Mander, J. B. (1994). "Seismic energy based fatigue
damage analysis of bridge columns. Part I: Evaluation of seismic ca-
pacity." Res. Rep. NCEER-94-0006, Dept. of Civ. Engrg., State Univ.
of New York at Buffalo, N.Y.

296/ JOURNAL OF STRUCTURAL ENGINEERING / MARCH 1998

J. Struct. Eng., 1998, 124(3): 288-297

 Dafalias. Y. F.. and Popov. E. P. (1975). "A model of nonlinearly hard-
ening materials for complex lading." Acta Mechanica. 21(3). 173-
192.
Dodd. L. L.. and Cooke. N. (1992). "The dynamic behavior of reinforced
concrete bridge piers subjected to New Zealand seismicity." Res. Rep.
92-04. Dept. of Civ. Engrg., Univ. of Canterbury, Christchurch, New
Zealand.
Dodd, L. L.. and Restrepo-Posada, J. 1. (1995). "Model for predicting
cyclic behavior of reinforcing steel." J. Struct. Engrg., ASCE, 121(3),
433-445.
Filippou, F. C., Popov. E. P., and Bertero, V. V. (1983). "Effects of bond
deterioration on hysteretic behavior of reinforced concrete joints."
EERC Rep. 83-19, Earthquake Engrg. Res. Ctr., University of Califor-
nia, Berkeley, Calif.
Geniev, G. A., Kissiuk, V. N.• and Tiupin, G. A. (1974). Theory of plas-
ticity of concrete and reinforced concrete. Stroiizdat. Moscow, Russia
(in Russian).
Karpenko, N.1. (1976). Theory ofdeformation ofreinforced concrete with
Downloaded from ascelibrary.org by Universita di Catania on 01/28/20. Copyright ASCE. For personal use only; all rights reserved.
"Response of RC critical regions under large amplitude reversed ac-
tions." (1983). Bull. D'lnformation No. 161, Comit6 Euro-International
du Beton, Lausanne, Switzerland.
Restrepo-Posada, J. I., Dodd, L. L., Park, R., and Cooke, N. (1994).
"Variables affecting cyclic behavior of reinforcing steel." J. Struet.
Engrg., ASCE, 120(11), 3178-3196.
Spurr, D. D., and Paulay, T. (1984). "Post-elastic behavior of reinforced
concrete frame-wall components and assemblages subjected to simu-
lated seismic loading." Res. Rep. 84-19, Dept. of Civ. Engrg., Univ.
of Canterbury, Christchurch. New Zealand.
Stanton, J. P., and McNiven, H. D. (1979). "The development of a math-
ematical model to predict the flexural response of reinforced concrete
beams to cyclic loads, using system identification." Rep. No. EERC
79-02. Univ. of California, Berkeley, Calif.
Thompson, K. J.• and Park. R. (1978). "Stress-strain model for grade 275
reinforcing steel with cyclic loading." Bull. New Zealand Nat. Soc. for
Earthquake Engrg., Waikanae, New Zealand, 11(2), 101-109.

cracks. Stroiizdat, Moscow. Russia (in Russian). APPENDIX II. NOTATION

Kent, D. C., and Park, R. (1973). "Cyclic load behavior of reinforcing
steel." Stain, J. British Soc. for Strain Measurement. 9(3), 98-103. The following symbols are used in this paper:
Leonhardt, F. (1980). Vorlesungen aber Massivbau. Vol. 1-6. Springer,
Berlin, Germany. A = instantaneous cross-sectional area;
Ludwik, P. (1909). Elemente der technologischen Mechanik. Verlag Julius Ao = initial cross-sectional area;
Springer. Berlin, Germany. a = amplification factor in the definition of the degradation pa-
Ma. S.-Y. M.• Bertero. V. V.• and Popov. E. P. (1976). "Experimental and rameter 8;
analytical studies on the hysteretic behavior of reinforced concrete rec- Eh = slope of the asymptote to the strain hardening region;
tangular and T-beams." EERC Rep. 76-02. Earthquake Engrg. Res. E. = initial modulus of elasticity;
Ctr., University of California. Berkeley, Calif. E, = tangent modulus of elasticity in engineering coordinates;
Mander. J. B.• Priestly. M. J. N., and Park. R. (1984). "Seismic design E. = unloading modulus of elasticity;
of bridge piers." Res. Rep. 84-02, Dept. of Civ. Engrg., Univ. of Can- E t = tangent modulus of elasticity in natural coordinates;
terbury. Christchurch. New Zealand.
Marin, J. (1962). Mechanical behaviour of engineering materials. Pren-
F = axial force;
tice-Hall Inc., Englewood Cliffs, N. J. f = model loading function;
Menegotto, M., and Pinto, P. E. (1973). "Method of analysis for cycli- ftnJ4X = maximum value of the loading functionj;
cally loaded reinforced concrete plane frames including changes in ge- f,. = engineering stress at k-reversal point;
ometry and non-elastic behavior of elements under combined normal h = steel stress;
force and bending." Proc., IABSE Symp. of Resistance and Ultimate h. = steel ultimate stress;
Deformability of Structures Acted on by Well-Defined Repeated Loads. h = steel yield stress;
International Association of Bridge and Structural Engineering, Lisbon,
Portugal, 15-22.
k = left superscript of model parameters indicating the number
of load reversals;
Monti, G., and Nuti. C. (1992). "Nonlinear cyclic behavior of reinforcing
bars including buckling." J. Struct. Engrg.• ASCE. 118(12), 3268-
L = instantaneous specimen length;
3284.
La = initial specimen length;
Naaman. A. E. (1982). Prestressed concrete, analysis and design. Mc- AL = change in length of the specimen;
Graw-Hill Book Co. Inc., New York. N.Y. l1e = incremental natural strain;
Nadai, A. (1950). Theory offlow and fracture of solids, 2nd Ed., Vol. I, 8 = cyclic strength degradation parameter;
McGraw-Hill Book Co. Inc.• New York. N.Y. 80 = initial value of 8;
Nawy, E. G. (1995). Prestressed concrete: fundamental approach, 2nd £ = engineering strain;
Ed.• Prentice-Hall, Inc.• Upper Saddle River. N.J. £0 = generic strain shift of the stress-strain envelope;
Panthaki. F. D. (1991). "Low cycle fatigue behavior of high strength and
ordinary reinforcing steels," MS thesis, Dept. of Civ. Engrg., State
£p = amplitude of engineering plastic strain;
£, = strain at reversal point;
Univ. of New York at Buffalo. N.Y.
Park. R.. Kent, D., and Sampson. R. (1972). "Reinforced concrete mem-
£. = steel bar engineering strain;
bers with cyclic loading." J. Struct. Div., ASCE, 98(7), 1341-1360. £.h = engineering strain at the beginning of the strain hardening
Popov, E. P., and Ortiz. M. (1979). "Macroscopic and microscopic cyclic region;
plasticity." Proc. 3rd Engrg. Mech. Div. Specialty Conf., ASCE, 303- £y yield strain;
330. e = natural strain;
Ramberg, w.. and Osgood. W. R. (1943). "Description of stress-strain p = hardening ratio;
curves by three parameters." NACA Rep. TN-902. National Advisory cr = engineering stress; and
Committee on Aeronautics, Washington. D.C. i1 = natural stress.

JOURNAL OF STRUCTURAL ENGINEERING / MARCH 1998/297

J. Struct. Eng., 1998, 124(3): 288-297