Professional Documents
Culture Documents
1998 - Filippou - Hysteretic Model of Ordinary and High-Strength Reinforcing Steel PDF
1998 - Filippou - Hysteretic Model of Ordinary and High-Strength Reinforcing Steel PDF
REINFORCING STEEL
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.
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.
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 where E, = daltlE = tangent modulus of the reinforcing steel
strain be defined as the corresponding tension stress multiplied in engineering coordinates.
by (1 - 2E.). The main advantage of the natural coordinate system is that
The difference in the tension and compression responses of the compression and tension stress-strain curves practically co-
steel specimens is affected by the coordinate system used to incide, whereas the two curves are quite different in engineer-
represent the measured data. Experimental data are typically ing coordinates. This concept is illustrated in Fig. 2, which
presented in engineering coordinates. If stresses and strains are shows in the same quadrant compression and tension mono-
JOURNAL OF STRUCTURAL ENGINEERING / MARCH 1998 /289
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.
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)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
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.
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]
-'---1 +8 '
e;
Cyclic Envelope Curves
It is important to note that in Fig. 5 the tension and com-
The compression and tension envelopes of the cyclic stress- pression envelopes are represented with respect to the engi-
strain curve are defined by the monotonic stress-strain relation neering coordinates. Up to the end of the Liiders plateau the
of (13), whose origin is shifted to point (Eo, 0) in order to two envelopes are identical. In the strain hardening regions,
describe cyclic response (Fig. 5). In the following develop- the engineering stresses corresponding to identical engineering
ments the strain £0 will represent the generic origin shift. Tak- strains are larger on the compression envelope.
JOURNAL OF STRUCTURAL ENGINEERING / MARCH 1998 /291
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.
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
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
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.
~
- - 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.
~
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;,.........
b) Experimental and Analytical Response for Specimen P16 b) Experimental and Analytical Response lor Test 6
FIG. 13. Measured and Modeled Stress-Strain Response for FIG. 14. Measured and Modeled Stress-Strain Response for
Tests on High Strength Reinforcing Steel Bars by Panthakl Tests by Aktan et al. (1973)
(1991)
perimental data are presently available and a larger database
based on the following model parameters: /.; = 70.0 ksi; Is. = is needed for a precise definition of the degradation parame-
100.0 ksi; E s = 29000.0 ksi; E h = 580.0 ksi; 80 = 0.005; £sh = ters. Low-cycle fatigue (Aktan et aI. 1973; Panthaki 1991;
0.011; and £s. = 0.29. A large number of plastic cycles were Chang and Mander 1994) and buckling of the reinforcing bars
applied in both tests, and in both cases the yield plateau and (Monti and Nuti 1992) will also be important enhancements
the strain hardening regions are clearly visible in the results. of the proposed model that will allow for the realistic analysis
Predicted and experimental stress-strain responses are once of reinforced concrete structures subjected to seismic loading.
again in close agreement. However, as the cycles progress, These extensions of the basic model proposed in this paper
some discrepancies appear, probably related to the fact that are presently under investigation.
low-cycle fatigue is ignored in the proposed model.
ACKNOWLEDGMENTS
CONCLUSIONS
The research reported in this paper was carried out while the first
A general-purpose hysteretic material model is proposed for writer was a visiting scholar at the Civil Engineering Department of the
University of California, Berkeley. This research was supported by the
the simulation of the cyclic response of different types of or- Technical University of Moldova (UTM). The writers would like to grate-
dinary and high-strength reinforcing steels. The model is sim- fully acknowledge Ion A. Bostan, president of UTM, for his support. The
ple and is formulated in terms of natural stresses and strains. suggestions of E. Spacone from the Department of Civil, Architectural,
The formulation is based on the assumption that the cyclic and Environmental Engineering at the University of Colorado, Boulder,
stress-strain behavior of reinforcing steel in the natural coor- and his valuable comments about the model description are much appre-
dinate system is identical in tension and compression up to ciated.
the point of plastic instability when necking (barreling) de-
velops in the tensile (compression) test. APPENDIX I. REFERENCES
A simple hyperbolic curve describes the entire envelope
Aktan, A. E., Karlsson, B. I., and Sozen, M. A. (1973). "Stress-strain
curve, including both the yield plateau and the strain hardening relationships of reinforcing bars subjected to large strain reversals."
branch. This curve is simply redefined and shifted during the Res. Rep. NSFRG G/ 29934, Dept. of Civ. Engrg., Univ. of Illinois,
load reversal cycles. Formulation of the model requires steel Urbana, III.
material properties in tension only, thus making the model par- Bate, P. S., and Wilson, D. V. (1986). "Analysis of Bauschinger effect."
ticularly easy to program and use for any study that requires Acta Metallurgica, 34(6), 1097-1105.
an uniaxial constitutive law for reinforcing steel. The accuracy Bauschinger, J. (1887). "Variations in the elastic limit of iron and steel
[summarized translation]." J. Iron and Steel/nst., I, 442-444.
of the proposed model has been verified through comparisons Chang, G. A., and Mander, J. B. (1994). "Seismic energy based fatigue
with several experimental tests on steel coupons. damage analysis of bridge columns. Part I: Evaluation of seismic ca-
Further studies are necessary to calibrate the model pa- pacity." Res. Rep. NCEER-94-0006, Dept. of Civ. Engrg., State Univ.
rameters governing material strength degradation. Limited ex- of New York at Buffalo, N.Y.