You are on page 1of 10

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/245305428

Simple Phenomenological Model for Reinforcing Steel under Arbitrary Load

Article  in  Journal of Structural Engineering · July 2006


DOI: 10.1061/(ASCE)0733-9445(2006)132:7(1061)

CITATIONS READS
42 1,352

2 authors:

Matthew Hoehler John Stanton


National Institute of Standards and Technology University of Washington Seattle
89 PUBLICATIONS   945 CITATIONS    84 PUBLICATIONS   1,898 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

National Fire Research Laboratory Metrology and Operations Project View project

Structural Fire Research Project View project

All content following this page was uploaded by Matthew Hoehler on 12 December 2014.

The user has requested enhancement of the downloaded file.


Simple Phenomenological Model for Reinforcing Steel
under Arbitrary Load
Matthew S. Hoehler1 and John F. Stanton2

Abstract: A new formulation for a robust, yet simple, uniaxial material model for reinforcing steel subjected to arbitrary loading is
presented. The model provides steel stress as an explicit function of the total steel strain and can account for nonlinear monotonic
envelope curves, i.e., with a yield plateau and nonlinear strain hardening, degradation of the yield limit as a function of the plastic strain
history 共Bauschinger effect兲 and strain hardening phenomena. Suggested model input parameters for typical reinforcing steels used in the
United States are provided and the characteristic behavior of the model is illustrated through comparison with experimental results.
DOI: 10.1061/共ASCE兲0733-9445共2006兲132:7共1061兲
CE Database subject headings: Concrete structures; Constitutive models; Cyclic loads; Finite element method; Hysteresis; Stress
strain relations; Steel.

Introduction Formulation of Material Model

Accurate modeling of reinforcing steel stress–strain behavior is The proposed model, which expresses stress as an explicit func-
important for predicting the response of reinforced concrete mem- tion of the total strain, consists of essentially three components: a
monotonic envelope curve, a branch curve on which the stress is
bers. For example, under monotonic loading, proper representa-
defined relative to the stress on the envelope curve for a given
tion of the steel envelope curve, i.e., to include a yield plateau and
strain, and a set of rules, implemented at each strain reversal, for
nonlinear strain hardening, is critical for detailed studies of steel– generating the parameter values that define the material behavior
concrete bond interaction and cracking phenomena. The steel on the upcoming branch curve. Those rules depend on the inelas-
model also influences computed deflections. Under cyclic load- tic strain history.
ing, such as that which occurs during an earthquake, accurate
modeling of the steel hysteretic behavior is necessary to properly
Envelope Curve
represent energy dissipation in a structure.
Numerous phenomenological models of the hysteretic stress- The equations describing the complete hysteretic behavior of re-
strain behavior of reinforcing steel are available in the literature inforcing steel presented in the next section require definition of a
共e.g., Ramberg and Osgood 1943; Park et al. 1972; Menegotto monotonic envelope curve. Although any realistic envelope curve
can be used, the envelope curve defined by the equations in
and Pinto 1973; Stanton and McNiven 1979; Monti and Nuti
Raynor et al. 共2002兲 is used in the proposed model due to its
1992; Dodd and Restrepo-Posada 1995; CEB 1996; Balan et al.
simplicity and versatility. The Raynor model for the monotonic
1998兲. This paper presents a new model that is versatile, compu- envelope curve is shown in Fig. 1.
tationally efficient, and easy to implement in a finite element The monotonic envelope curve is defined in the positive quad-
program. The primary advantage of the model is that it remains rant by the following equations:
conceptually simple while offering accuracy comparable with the
more sophisticated phenomenological steel models available ␴env = E · ␧ for ␧ 艋 ␧y 共1兲
today.
␴env = ␴y + 共␧ − ␧y兲 · Ey for ␧y ⬍ ␧ 艋 ␧sh 共2兲

1
Research Engineer, Institute of Construction Materials, Univ. of
Stuttgart, Pfaffenwaldring 4, D-70550 Stuttgart, Germany 共corresponding
␴env = ␴u − 共␴u − ␴sh兲 · 冉 ␧u − ␧
␧u − ␧sh
冊 C1
for ␧sh ⬍ ␧ 艋 ␧u 共3兲

author兲. E-mail: mhoehler@iwb.uni-stuttgart.de


2 where
Professor, Dept. of Civil Engineering, Box 352700, Univ. of
Washington, Seattle, WA 98195-2700. E-mail: stanton@ ␴y
u.washington.edu ␧y = 共4兲
Note. Associate Editor: Sashi K. Kunnath. Discussion open until E
December 1, 2006. Separate discussions must be submitted for individual
papers. To extend the closing date by one month, a written request must ␴sh = ␴y + 共␧sh − ␧y兲 · Ey 共5兲
be filed with the ASCE Managing Editor. The manuscript for this paper
was submitted for review and possible publication on January 26, 2005; ␴env = envelope stress; ␧ = total strain; E = elastic modulus; ␴y and
approved on June 2, 2005. This paper is part of the Journal of Structural ␧y = stress and strain at steel yield; Ey = slope of the yield plateau;
Engineering, Vol. 132, No. 7, July 1, 2006. ©ASCE, ISSN 0733-9445/ ␴sh and ␧sh = stress and strain at the onset of strain hardening; ␴u
2006/7-1061–1069/$25.00. and ␧u = ultimate 共peak兲 stress and strain; and C1 = parameter that

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JULY 2006 / 1061


Fig. 1. Raynor model for monotonic envelope curve for steel Fig. 2. Branch curve in proposed hysteretic steel model

defines the curvature of the strain hardening curve. The signs a1 · ␰


R = R0 − 共10兲
must be modified appropriately for excursions into the negative a2 + ␰
strain and stress quadrant.
The equations described by Raynor et al. 共2002兲 can be used to ␴ = stress for any given strain ␧; ␴env = envelope stress at strain ␧;
define bilinear, multilinear, or nonlinear envelope curves and are ␴rev and ␧rev = stress and strain at the last strain reversal; and
independent of units. For the model proposed in the next section the variable R, which controls the curvature of the branch
the initial linear elastic portion of the envelope curve 关Eq. 共1兲兴 is curve=function of the parameters R0 , a1 , a2, and the plastic strain
replaced by an extrapolation of either the yield plateau or the variable ␰. The starting point for any branch is determined by the
strain hardening branch, since linear elastic loading and unloading stress and strain at the last strain reversal. This is shown sche-
curves are automatically generated by the hysteretic branch curve matically in Fig. 2.
equations, and only the postyield part of the envelope is used. The function f共␧兲 in Eq. 共6兲 is a modified form of the equation
For the sake of simplicity, the shape of the tension and com- proposed by Menegotto and Pinto 共1973兲 to model the cyclic
pression envelope curves, which are expressed in terms of engi- behavior of steel. It is important to note that, while the
neering strain, are assumed to be identical. As shown by Dodd Menegotto–Pinto equation defines the transition between two in-
and Restrepo-Posada 共1995兲, a higher degree of similarity be- tersecting straight lines, the function f共␧兲 controls the shape of a
tween the tension and compression behavior of steel can be branch curve as it approaches an envelope curve of arbitrary
obtained using natural stress and natural strain, i.e., taking the shape.
reduction 共increase兲 of cross-sectional area during tension 共com- Eq. 共6兲 has three characteristics that make it well suited for
pression兲 loading into account. Although transformation of the modeling the hysteretic behavior of steel. First, the derivative of
compression envelope curve according to the equations provided Eq. 共6兲 with respect to ␧, evaluated at ␧rev, which gives the
by Dodd and Restrepo-Posada can be performed, this extra step tangent modulus immediately following a reversal, is equal to
was not observed to be necessary to obtain sufficiently accurate the elastic modulus E. Second, for large values of R 共R ⬎ 20兲 the
results with the proposed model. branch curve is a straight line, of slope E, that transitions sharply
to the envelope curve when the two meet. Third, for smaller val-
ues of R 共0 艋 R 艋 20兲, the slope of the branch curve decreases
Proposed Stress–Strain Relation from a value of E at the load reversal point to become tangent to
The new aspect of the proposed model is the way stress is ex- the envelope curve. The smaller the R value, the more quickly the
pressed relative to the envelope curve. The concept is reminiscent branch curve diverges from the elastic modulus and the more
of the bounding surface model proposed by Dafalias and Popov gradual is the transition to the envelope curve. Examples of
共1975兲. In the current model, the explicit stress–strain relation branch curves with various R values are shown in Fig. 3.
accounts for loading, unloading, and reloading. Stress on any
“branch” curve, i.e., a section of the stress–strain curve between
strain reversals, is described by the following equations:

␴ = ␴env − 共␴env − ␴rev兲 · f共␧兲 共6兲

where

␤ · ⌬␧
f共␧兲 = 1 − 共7兲
关1 + 共␤ · ⌬␧兲R兴1/R

E
␤= 共8兲
␴env − ␴rev

⌬␧ = ␧ − ␧rev 共9兲 Fig. 3. Branch curves with various R values

1062 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JULY 2006


As indicated in the previous section, initial linear elastic
stress–strain behavior is obtained simply by extrapolating the
yield plateau of the envelope curve to ␧ = 0, initializing ␴rev and
␧rev to zero, and selecting a sufficiently large initial value for R
共e.g., R0 = 25兲.
In the proposed implementation of the model, it is assumed
that the initial slope of any branch curve is equal to the elastic
modulus, E, regardless of the previous strain history. In reality the
unloading–reloading modulus Eu decreases and depends on
the plastic strain 共Dodd and Restrepo-Posada 1995兲, however, the
change in modulus for reinforcing steel is typically small
关Eu = 共0.82– 1.00兲 · E兴 and can be neglected for practical applica-
tions. If desired, the decrease in the unloading–reloading modulus
can easily be taken into account in the proposed model by reduc- Fig. 4. Shift of envelope curves and elimination of yield plateau
ing the elastic modulus dependent on a plastic strain variable 关see
for example Stanton and McNiven 共1979兲; Dodd and Restrepo- Thus the envelope curve, which in the proposed model is
Posada 共1995兲兴. based on Eqs. 共1兲–共3兲, can be reduced to the following two
The influence of strain rate is neglected in the proposed model. equations:
␴env = ␴y + 共兩␧ − ␧zm兩 − ␧y兲 · Ey
Model Parameter Changes Based on Inelastic Strain
for 兵兩␧ − ␧zm兩 艋 ␧sh and ⌬␧zm 艋 ␧sh其 共12兲
The proposed model as described up to this point can be used to

冉 冊
model monotonically loaded steel elements. Eqs. 共6兲–共10兲 simply
provide a more complex way of reproducing the monotonic en- ␧u − 兩␧ − ␧zm兩 C1
␴env = ␴u − 共␴u − ␴sh兲 ·
velope curve. To model the hysteretic behavior that occurs during ␧u − ␧sh
cycling, three groups of parameters that depend on the inelastic 共for 兵兩␧ − ␧zm兩 艋 ␧sh and ⌬␧zm ⬎ ␧sh其 or 兵兩␧ − ␧zm兩 ⬎ ␧sh其兲
strain history must be updated and stored at every strain reversal.
The parameters that must be updated and stored are the shape and 共13兲
position of the envelope curves, the branch curvature controlling where ␧zm = maximum positive or negative zero shift 共␧zmp , ␧zmn兲
parameter R and the critical values of stress and strain to ensure depending on the loading direction. The envelope stress ␴env must
proper return to the envelope curve for the case of reloading after be multiplied by −1 to obtain the compression envelope.
partial unloading. Two important remarks must be made about modeling the en-
While the equations in the previous sections are valid for ei- velope curves in this simplified manner. First, the decision to
ther tensile or compressive loading, separate values for the inelas- ignore the yield plateau after the difference in the maximum zero
tic strain based parameters must be stored for the positive and shifts becomes larger than ␧sh was based on observation of the
negative loading directions. The direction of loading is defined by available experimental data and not on theoretical grounds. Sec-
the sign of the strain increment. ond, although back projection of the strain hardening curve to
eliminate the yield plateau allows for the use of a simple set of
Modifications to Envelope Curves equations for the envelope curves and yields results that agree
A simple approach is applied to allow for hysteretic behavior and well with the investigated experimental data for reinforcing steel,
to simultaneously account for isotropic strain hardening of the this simplification could lead to a numerical problem for some
steel. At every strain reversal, the envelope curve is shifted along combinations of ␧sh and C1. When large values are given for both
the strain axis so that the branch curve can approach it from the ␧sh and C1 共a long yield plateau with steep strain hardening cur-
inside. This is achieved by defining the “zero shift” as the inter- vature兲, the back projection of the strain hardening curve could
section of the strain axis with the elastic unloading line from the fail to intersect the elastic unloading–reloading curve. In this case
reversal point the envelope curve would not be defined. A combination of input
␴rev parameters that would cause this numerical problem to occur was
␧z = ␧rev − 共11兲 not observed for typical reinforcing steels used in the United
E
States, Europe, or New Zealand, however, a check for this condi-
where ␧z = zero shift of the origin of the positive or negative en- tion should be implemented when programming the model.
velope curve. The maximum positive and negative zero shifts A further modification to the envelope curve that may be
共␧zmp , ␧zmn兲 that have occurred during the loading history are implemented in the model is the reduction of the onset of strain
stored. This is done in such a way that the positive envelope curve hardening ␧sh based on the previous plastic strain history. Reduc-
can only shift in the negative ␧ direction and the negative enve- tion of the onset of strain hardening further reduces the likelihood
lope curve can only shift in the positive ␧ direction. These maxi- of the potential numerical problem described above and allows
mum zero-shift values are used to modify the total strain in the for a slightly more accurate representation of the hysteretic be-
equations for the envelope curve. havior of reinforcing steel in some cases. Experimental data show
Additionally, in the proposed implementation, if the differ- that ␧sh decreases under cyclic conditions and that the magnitude
ence between the maximum negative and positive zero shifts of this reduction seems to be related to the accumulated comple-
共⌬␧zm = ␧zmn − ␧zmp兲 is greater than the strain at the onset of strain mentary strain energy during reversals 共Dodd and Restrepo-
hardening ␧sh, the strain hardening curve is extended toward the Posada 1995兲. In the proposed model, the initial value for the
shifted strain origin of the envelope curve, i.e., the yield plateau is onset of the strain hardening ␧sh 0
is assumed to reduce linearly to
ignored. This is shown schematically in Fig. 4. ␧y, i.e., no remaining plateau, as a function of a cumulative hys-

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JULY 2006 / 1063


Bauschinger Effect
The parameter R in Eq. 共7兲 controls the shape of the hysteretic
loops and thus the degree to which the Bauschinger effect is
present for any branch curve. In the proposed model, the relation
for R suggested by Menegotto and Pinto 共1973兲 is used. In this
equation, R0 is the value of the parameter R during the initial
loading and a1 and a2 determine the extent and rate of the degra-
dation of the yield limit as a function of the previous plastic
strain. All three parameters carry no units and must be determined
by fitting of experimental data. R tends towards the value
共R0 − a1兲 as ␰ approaches infinity, so a1 must be less than R0 in
order to avoid a negative value of R. Moreover, for values of R
less than about 1.5, the branch curve might not rejoin the enve-
lope curve in the practical strain range for reinforcing steel. In a
Fig. 5. Reduction of onset of strain hardening study of experimental results from typical reinforcement bars
used in the United States, Europe, and New Zealand, default val-
ues of R0 = 25, a1 ⬇ 22.5, and a2 = 0.10– 0.50 were found to give
teretic strain parameter, ␧h. This is shown schematically in Fig. 5. good results for the proposed model.
The superscripts are used to designate the strain reversals. The In the proposed model the plastic strain variable ␰ is defined as
components of the cumulative strain parameter 共␧1h, ␧2h, etc.兲 are
based on the widths of the closed strain loops. 共␧zmn − ␧zmp兲
␰= 共16兲
The instantaneous value for the onset of strain hardening ␧sh is ␧y
defined as
where ␧zmp, ␧zmn, and ␧y are as defined previously. Fig. 6 illus-

␧sh = ␧sh
0

· 1−
共␧sh
0
− ␧y兲
8 · 共␧sh兲
0 2 册
· ␧h 艌 ␧ y 共14兲
trates the definition of the curvature parameter.

Proper Return to Envelope Curve


where Fig. 7 illustrates a problem that can arise in phenomenological
models for steel cycling. This problem has been documented by
␧h = 兺
n=0
␧nh 共15兲 other authors in the context of the Menegotto-Pinto model 共CEB
1996兲, but it has the potential to arise in any model in which a
branch curve approaches an envelope. For reloading following
and ␧y and ␧sh0
are as defined previously. Although the reduction
partial unloading on a curve that has experienced reduction of the
of the onset of strain hardening has been observed experimentally
yield limit due to the Bauschinger effect, the curve 共a兲 shown in
共e.g., Ma et al. 1976兲, the writers are not aware of any detailed
Fig. 7 should be followed. If the model does not store the infor-
studies describing the parameters influencing this behavior. The
mation that defines the branch curve 共a兲, it will return to the
assumption of a linear reduction of the onset of strain hardening,
monotonic envelope along curve 共b兲.
as well as the assumption that the yield plateau completely dis-
To avoid this behavior, the model must store the information
appears when the above-defined cumulative strain parameter
that defines every branch curve, i.e., ␴rev, ␧rev, and the R value.
reaches a value of eight times the initial value for the onset of
For general load cases with many branches this is not practicable
strain hardening, is based on observation of a limited amount of
from a memory standpoint. In the proposed model this problem
experimental data and thus should be viewed critically. Eq. 共14兲
was dealt with in a pragmatic fashion. Critical maximum stresses
was found to yield relatively good performance for all of the
and strains that have occurred at positive and negative strain re-
experimental results investigated, however, for some reinforcing
versals are stored 共␴rmp , ␧rmp , ␴rmn , ␧rmn兲, i.e., the memory of the
steels the reduction of the onset of strain hardening appeared to
model consists of one positive and one negative branch. At a
occur faster and in some cases no reduction was observed. The
given strain reversal, the reversal stress and strain 共␴rev , ␧rev兲 are
reduction of the onset of hardening can be ignored 共␧sh = ␧sh
0
兲 for
practical applications.

Fig. 7. Reloading curve after partial unloading: 共a兲 correct behavior;


Fig. 6. Definition of curvature parameter 共b兲 behavior of model with limited memory

1064 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JULY 2006


between constant strain levels, this leads to relaxation of the
“mean stress” as described by Morrow and Sinclear 共1959兲. This
feature of the model is sometimes undesirable. Numerical inves-
tigations with the model showed, however, that the amount of
relaxation per cycle decreases with the number of cycles, i.e., it
converges.
The rules for cycling described above were found to provide
good results for practical load histories and are efficient in terms
of memory use.

Model Input Parameters

This section provides suggested input parameters for the proposed


Fig. 8. Reloading curve after partial unloading in proposed model model for typical reinforcement bar steels currently used in the
United States. These values are provided only as a guide for users
of the model and are intended for analysis, i.e., not for design,
compared with the critical values to establish whether a previous when the user does not have experimental data available. It is
branch curve should be followed or if an envelope curve should important to note that the expected 共mean兲 strengths 共␴ye , ␴ue兲
be followed and the critical maximum values updated. Further- provided are averages based on tests with many different rein-
more, the envelope curve is automatically followed upon reload- forcement bar diameters produced by various manufacturing pro-
ing if the previous branch curve in the loading direction was cesses and thus could vary significantly from the actual values for
“sufficiently close” to the monotonic envelope curve. This proce- a specific reinforcement bar. Furthermore, the strain at the onset
dure is illustrated in Fig. 8 for the case of reloading in the of strain hardening ␧sh and at the ultimate 共peak兲 strength, ␧u, as
negative direction after unloading from a branch curve with well as the strain hardening curvature, C1, are strongly dependent
exaggerated softening. on the steel composition, the reinforcement bar diameter and the
Reloading from Point A in Fig. 8 will occur to the previous process by which the bar was manufactured, and thus only a
negative branch curve and the critical stress and strain values range of typical values can be provided.
共␴rmn , ␧rmn兲 will remain unchanged since the strain at reversal In the proposed model it was assumed that the slope of the
Point A is less than the critical maximum negative reversal strain yield plateau Ey was constant for all steel types and was defined
共␧revគA ⬍ ␧rmn兲 and the stress at reversal Point A 共␴revគA兲 is greater as
than the stress on the previous negative branch curve at the rever-
sal strain. Reloading from Point B in Fig. 8 will occur to the
previous envelope curve and the critical stress and strain values 0.001 · ␴y
Ey = 共17兲
共␴rmn , ␧rmn兲 will be set equal to the reversal stress and strain val- 共␧sh − ␧y兲
ues 共␴rev , ␧rev兲 since the strain at reversal Point B is less than the
critical maximum negative reversal strain 共␧revគB ⬍ ␧rmn兲, however, and that ␧y and ␴sh were determined as given in Eqs. 共4兲 and 共5兲.
the stress at reversal Point B 共␴revគB兲 is less than the stress on the Thus a total of nine input parameters are required: E, ␴y, ␴u, ␧sh,
previous negative branch curve. Reloading from Point C in Fig. 8 ␧u, C1, and R0, a1, a2.
will occur to a new envelope curve and the critical stress and Table 1 provides suggested input parameters for the proposed
strain values 共␴rmn , ␧rmn兲 will be set equal to the reversal stress model for typical Grade 60 reinforcement bar steels currently
and strain values 共␴rev , ␧rev兲 since the strain at reversal Point C is used in the United States. The values were compiled from Malvar
greater than the critical maximum negative reversal strain and Crawford 共1998兲, ASTM 共1999兲, Caltrans 共2004兲, and the
共␧revគC ⬎ ␧rmn兲. writers’ investigations of cyclic test data for reinforcement steels
Regardless of the relation between the actual and critical strain used in the United States, Europe, and New Zealand. For analysis
reversal values, if the previous negative branch curve had been purposes, the expected 共mean兲 values for the yield and ultimate
“sufficiently close” to the envelope curve, reloading would occur strength 共␴ye , ␴ue兲 are typically more appropriate than the mini-
to an envelope curve and the critical stress and strain values mum values 共␴y , ␴u兲.
共␴rmn , ␧rmn兲 would be updated. In the proposed model, the toler-
ance for “sufficiently close” was defined by a stress difference
between the envelope and branch curves of E · 10−5, i.e., a stress Comparison of Proposed Model with Experimental
jump of around 2 MPa 共0.3 ksi兲. Data
It may be noted that for all three Cases 共A, B, and C兲 that if a
further reversal was to occur as loading progressed toward The performance of the proposed model was evaluated using
the negative branch curve 共or envelope兲, the previous positive numerous sets of experimental results for steel reinforcing bars
branch curve would be rejoined. This is consistent with the afore- commonly used in the United States, Europe, and New Zealand.
mentioned criteria for reloading from Point A. Thus, the proposed Selected results illustrating characteristic behavior of the pro-
criteria allow for closing of small loops within larger strain posed model are presented below.
reversals. The capability of the proposed model to represent the stress–
Due to the form of the proposed stress–strain relation in strain behavior of low ductility 共Type A兲 and ductile 共Type B兲
Eq. 共6兲, a strain loop typically closes below the point where the reinforcing steel under monotonic loading is shown in Fig. 9
loop initiated, e.g., the stress at Point B in Fig. 8 is below ␴rmn. using experimental results for European grade BSt 500 S re-
For small values of R 共strong Bauschinger effect兲 and cycling inforcing bar coupons tested by Eligehausen et al. 共2003兲. The

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JULY 2006 / 1065


Table 1. Suggested Model Input Parameters
Grade 60

Parameter Units A706a A615b


E MPa 200,000 200,000
␴y MPa 420 420c
␴ye MPa 475 475
␴u MPa 550 620c
␴ue MPa 655 750
␧sh — 0.003–0.030d
#8 共25 mm兲 bars — 0.0150 —
#9 共28 mm兲 bars — 0.0125 —
#10 共32 mm兲 bars — 0.0115 —
#14 共43 mm兲 bars — 0.0075 —
#18 共57 mm兲 bars — 0.0050 —
␧u — 0.07–0.18
#10 共32 mm兲 bars — 0.12 —
and smaller
#11 共36 mm兲 bars — 0.09 —
and larger
C1 — 2–6 2–6
R0 — 25 25
a1 — 21.0–23.5 21.0–23.5
a2 — 0.10–0.50 0.10–0.50
a
Envelope values from Caltrans 共2004兲.
b
Envelope values from Malvar and Crawford 共1998兲, with permission.
c
Extracted, with permission, from A615/A 615M-96a “Standard
specification for deformed and plain billet-steel bars for concrete
reinforcements,” copyright ASTM International.
d
Upper limit increased from Malvar and Crawford value.
Fig. 10. Experimental and predicted stress–strain behavior for tests
by Kent and Park 共1973兲, experimental data reprinted with
permission from the British Society for Strain Measurement: 共a兲
model parameters for the Type A steel were: E = 205,000 MPa, experimental and numerical results for Specimen 8; 共b兲 experimental
␴y = 580 MPa, ␴u = 620 MPa, ␧sh = 0.0028, ␧u = 0.04, C1 = 6, and numerical results for Specimen 17
R0 = 25, a1 = 22.0, and a2 = 0.3. The model parameters for the Type
B steel were: E = 205,000 MPa, ␴y = 564 MPa, ␴u = 650 MPa,
␧sh = 0.0260, ␧u = 0.130, C1 = 4, R0 = 25, a1 = 22.0, and a2 = 0.3. 共1973兲. The tests shown are designated as Specimens 8 and 17 in
The response of the proposed model to simple strain cycling the original work by Kent and Park. The model parameters were
histories is illustrated in Fig. 10 using experimental results for nominally: E = 200,000 MPa 共29,000 ksi兲, ␴y = 310 MPa 共45 ksi兲,
normal strength reinforcing steel coupons tested by Kent and Park ␴u = 380 MPa 共55 ksi兲, ␧sh = 0.0100, ␧u = 0.140, C1 = 4, R0 = 25,
a1 = 23.0, and a2 = 0.3. The effect of neglecting softening of the
elastic modulus can be observed in Fig. 10共a兲.
The response of the proposed model to more complicated
strain cycling histories is illustrated in Fig. 11 using experimental
results for steel coupons machined from normal strength 共ASTM
A 615兲 reinforcing steel bars tested by Ma et al. 共1976兲. The
tests shown are designated as Specimens 1, 2, 3, and 4
in the original work by Ma et al. The model parameters were
nominally: E = 200,000 MPa 共29,000 ksi兲, ␴y = 455 MPa 共66 ksi兲,
␴u = 655 MPa 共95 ksi兲, ␧sh = 0.0110, ␧u = 0.090, C1 = 3, R0 = 25,
a1 = 23.0, and a2 = 0.1. The specimens tested by Ma et al. showed
a large asymmetry between the tension and compression envelope
curves. This can be observed in the overprediction of the peak
compressive stresses by the proposed model in Figs. 11共b and c兲.
Although this asymmetry can be accounted for by introducing
more complicated relations to describe the positive and negative
branch and envelope curves in the proposed model, i.e., by intro-
ducing a load direction dependent plastic strain variable, these
Fig. 9. Experimental and predicted stress–strain behavior for tests modifications are not presented here as the writers’ intention is to
by Eligehausen et al. 共2003兲, experimental data reprinted with present a simple and easy to implement model that yields suffi-
permission ciently accurate behavior for practical applications. Furthermore,

1066 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JULY 2006


Fig. 11. Experimental and predicted stress–strain behavior for tests by Ma et al. 共1976兲, experimental data reprinted with permission: 共a兲
experimental and numerical results for Specimen 1; 共b兲 experimental and numerical results for Specimen 2; 共c兲 experimental and numerical results
for Specimen 3; and 共d兲 experimental and numerical results for Specimen 4

the high degree of asymmetry between tension and compression tension and compression strain cycling is illustrated in Fig. 13
behavior was only significant for a limited number of the data sets using experimental results for reinforcing steel coupons tested
investigated. It should also be mentioned that the stress history by Leslie 共1974兲. The model parameters were: E = 200,000 MPa,
predicted by the proposed model resembled the measured values ␴y = 400 MPa, ␴u = 700 MPa, ␧sh = 0.0020, ␧u = 0.150, C1 = 3,
from Specimen 2 even more closely when the reduction of R0 = 25, a1 = 23.0, and a2 = 0.1.
the onset of strain hardening was neglected. This suggests
that Eq. 共14兲 may be an oversimplification of the true physical
behavior.
Summary and Conclusions
Experimental results for machined reinforcing steel coupons
subjected to strain histories representative of those that a reinforc-
ing bar might undergo during an earthquake, i.e., small compres- The formulation of a general uniaxial material model for reinforc-
sive strains combined with large tension strains, are presented in ing steel under arbitrary cyclic loading was presented. The pri-
Dodd and Restrepo-Posada 共1995兲. The specimens shown were mary advantage of the model is that it remains conceptually
New Zealand manufactured steel Grades 300 and 430. The re- simple while offering accuracy comparable with the more sophis-
sponse of the proposed model to the experimental data described ticated phenomenological steel models developed to date.
in Dodd and Restrepo-Posada 共1995兲 is illustrated in Fig. 12. The following conclusions may be drawn from the study:
The model parameters for the Grade 300 steel were: 1. The model is suitable for incorporation into a nonlinear finite
E = 200,000 MPa, ␴y = 300 MPa, ␴u = 465 MPa, ␧sh = 0.0190, element formulation, such as a fiber model, for reinforced
␧u = 0.180, C1 = 6, R0 = 25, a1 = 22.0, and a2 = 0.1. The model pa- concrete members.
rameters for the Grade 430 steel were: E = 200,000 MPa, 2. The model accounts for a wide variety of nonlinear mono-
␴y = 465 MPa, ␴u = 650 MPa, ␧sh = 0.0100, ␧u = 0.150, C1 = 5, tonic envelope curves, for reduction with cycling of the
R0 = 25, a1 = 23.0, and a2 = 0.2. The experimental and predicted strain at the onset of strain hardening, for the Bauschinger
results are in quite good agreement. No significant asymmetry effect, and for strain hardening phenomena.
between the tension and compression envelopes was observed. 3. By expressing stress as an explicit function of strain, the
The response of the proposed model to large, symmetric, model is computationally efficient.

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JULY 2006 / 1067


Notation

The following symbols are used in this paper:


a1 ⫽ branch curvature control parameter 1;
a2 ⫽ branch curvature control parameter 2;
C1 ⫽ parameter defining curvature of strain hardening
curve;
E ⫽ elastic modulus;
Eu ⫽ unloading–reloading modulus;
Ey ⫽ slope of yield plateau;
R ⫽ variable that controls curvature of branch curve
共Bauschinger effect兲;
R0 ⫽ initial curvature of branch curve;
⌬␧zm ⫽ difference between the maximum negative and
positive “zero-shift” 共⌬␧zm = ␧zmn − ␧zmp兲;
␧ ⫽ total strain;
␧h ⫽ cumulative hysteretic strain parameter;
␧rev ⫽ strain at last load reversal;
␧rmn ⫽ critical maximum negative strain reversal value;
␧rmp ⫽ critical maximum positive strain reversal value;
␧sh ⫽ strain at onset of strain hardening;
␧sh
0
⫽ initial value for strain at onset of strain hardening;
␧u ⫽ ultimate 共peak兲 strain;
␧y ⫽ strain at steel yield;
␧z ⫽ “zero-shift” of positive or negative envelope curve;
␧zm ⫽ maximum positive or negative “zero-shift” depending
on loading direction;
␧zmn ⫽ maximum negative “zero-shift”;
␧zmp ⫽ maximum positive “zero-shift”;
␰ ⫽ plastic strain variable;
␴ ⫽ stress for any given strain;
Fig. 12. Experimental and predicted stress–strain behavior for tests ␴env ⫽ envelope stress;
by Dodd and Restrepo-Posada 共1995兲, ASCE: 共a兲 experimental and ␴rev ⫽ stress at last load reversal;
numerical results for Grade 300 reinforcing steel; 共b兲 experimental ␴rmn ⫽ critical maximum negative stress reversal value;
and numerical results for Grade 430 reinforcing steel ␴rmp ⫽ critical maximum positive stress reversal value;
␴sh ⫽ stress at onset of strain hardening;
␴u ⫽ ultimate 共peak兲 stress;
4. Simple rules are suggested for addressing the problem of ␴ue ⫽ expected ultimate 共peak兲 stress;
nested branch curves. ␴y ⫽ stress at steel yield; and
5. Values for model input parameters for typical reinforcing ␴ye ⫽ expected stress at steel yield.
steels used in the United States are suggested.
6. The proposed model was successfully validated against ex-
perimental data from a number of different sources in the References
United States, Europe, and New Zealand.
ASTM. 共1999兲. “A 615/A 615M-96a: Standard specification for deformed
and plain billet-steel bars for concrete reinforcement.” Annual book of
ASTM standards, ASTM International, Philadelphia.
Balan, T. A., Filippou, F. C., and Popov, E. P. 共1998兲. “Hysteretic model
of ordinary and high-strength reinforcing steel.” J. Struct. Eng.,
124共3兲, 288–297.
California Department of Transportation 共Caltrans兲. 共2004兲. Seismic de-
sign criteria, Version 1.3, Caltrans, Sacramento, Calif. 具http://
www.dot.ca.gov典 共Jan. 18, 2005兲.
Comité Euro-International du Béton 共CEB兲. 共1996兲. “RC elements under
cyclic loading: State-of-the-art report.” Bulletin d’information no.
230, Thomas Telford Service Ltd., London.
Dafalias, Y. F., and Popov, E. P. 共1975兲. “A model of nonlinear hardening
materials for complex loading.” Acta Mech., 21, 173–192.
Dodd, L. L., and Restrepo-Posada, J. I. 共1995兲. “Model for predicting
cyclic behavior of reinforcing steel.” J. Struct. Eng., 121共3兲,
433–445.
Eligehausen, R., Mayer, U., and Lettow, S. 共2003兲. “Mitwirkung des
betons zwischen den rissen in stahlbetonbauteilen.” Final Rep. DFG-
Fig. 13. Experimental and predicted stress–strain behavior for tests
Research Project EL 72/8-1, Institut für Werkstoffe im Bauwesen,
by Leslie 共1974兲 Univ. of Stuttgart, Stuttgart, Germany 共in German兲.

1068 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JULY 2006


Kent, D. C., and Park, R. 共1973兲. “Cyclic load behavior of reinforcing Monti, G., and Nuti, C. 共1992兲. “Nonlinear cyclic behavior of reinforcing
steel.” J. British Society Strain Measurement, 9共3兲, 98–103. bars including buckling.” J. Struct. Eng., 118共12兲, Dec., 3268–3284.
Leslie, P. D. 共1974兲. “Ductility of reinforced concrete bridge piers.” Morrow, J., and Sinclear, G. M. 共1959兲. “Cyclic dependent stress relax-
Master of Engineering Rep., Dept. of Civil Engineering, Univ. of ation.” Proc. Symp on Basic Mechanics of Fatigue, ASTM, STP–237.
Canterbury, Christchurch, New Zealand. Park, R., Kent, D. C., and Sampson, R. A. 共1972兲. “Reinforced concrete
Ma, S.-Y. M., Bertero, V. V., and Popov, E. P. 共1976兲. “Experimental and members with cyclic loading.” J. Struct. Div. ASCE, 98共7兲,
analytical studies on the hysteretic behavior of reinforced concrete 1341–1360.
rectangular and T-beams.” Earthquake Engineering Research Center Ramberg, W., and Osgood, W. R. 共1943兲. “Description of stress-strain
(EERC), Rep. UBC/EERC 76-2, Univ. of California, Berkeley, Calif. curves by three parameters.” Technical Note 902, National Advisory
Malvar, L. J., and Crawford, J. E. 共1998兲. “Dynamic increase factors for Committee for Aeronautics, Washington, D.C.
steel reinforcing bars.” Proc., 28th DDESB Seminar, Orlando, Fla, Raynor, D. J., Lehman, D. L., and Stanton, J. F. 共2002兲. “Bond-slip re-
具http://www.nfesc.navy.mil典 共Jan. 18, 2005兲. sponse of reinforcing bars grouted in ducts.” ACI Struct. J., 99共5兲,
Menegotto, M., and Pinto, P. E. 共1973兲. “Method of analysis for cycli- 568–576.
cally loaded RC plane fames including changes in geometry and Stanton, J. F., and McNiven, H. D. 共1979兲. “The development of a math-
non-elastic behavior of elements under combined normal force and ematical model to predict the flexural response of reinforced concrete
bending.” Proc., IABSE Symp. on the Resistance and Ultimate De- beams to cyclic loads, using system identification.” Earthquake Engi-
formability of Structures Acted on by Well Defined Repeated Loads, neering Research Center (EERC), Rep. UBC/EERC 79-02, Univ. of
Preliminary Rep., Lisbon, Portugal, 15–22. California, Berkeley, Calif.

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JULY 2006 / 1069

View publication stats

You might also like