Behavior of Reinforced Concrete Elements under Cyclic

Shear. II: Theoretical Model

Mohamad Mansour1 and Thomas T. C. Hsu2
Downloaded from ascelibrary.org by INDIAN INSTITUTE OF TECHNOLOGY on 06/08/14. Copyright ASCE. For personal use only; all rights reserved.

Abstract: In order to design reinforced concrete (RC) structures in earthquake regions, a cyclic softened membrane model (CSMM) is
presented to predict the load–deformation behavior of RC membrane elements subjected to reversed cyclic shear stresses. This model is
an extension of the softened membrane model for monotonic shear behavior. Both models are rational because they satisfy Navier’s
principles of mechanics of materials (stress equilibrium, strain compatibility, and constitutive laws of materials). Three new components
of material laws were required in developing CSMM: the stress–strain relationships for concrete and steel in the unloading and reloading
regions, the modifications of the envelope curve for compressive concrete, and the Hsu/Zhu ratios for cyclic loading. In order to verify the
CSMM as a valid theoretical model, this study focused on comparing the predictions of CSMM with actual test results reported in the
companion paper. We conclude from this comparison study that the CSMM can indeed predict the hysteretic loops and their pinched
shapes, and that the “pinching effect” was the result of steel bar direction deviating from that of the principal stresses. The “pinching
mechanism” and the “failure mechanism” are logically explained using Mohr’s circles of stresses and strains.
DOI: 10.1061/(ASCE)0733-9445(2005)131:1(54)
CE Database subject headings: Constitutive relations; Cyclic loads; Ductility; Energy dissipation; Hysteresis; Numerical models;
Concrete, reinforced; Shear stress.

Introduction crete elements in earthquake regions, a total of 15 panel elements

were subjected to reversed cyclic loading from 1997 to 2001. The
Over the past 30 years, an element-based approach was developed experimental cyclic behavior of such elements was presented in
to study reinforced concrete structures. The conceptual basis of the companion paper. From the observed behavior, a cyclic soft-
this approach is that the behavior of any whole structure can be ened membrane model (CSMM) was developed. This CSMM for
predicted by integrating what can be learned from experiments cyclic loading, which is presented in this paper, is an extension of
about its parts (elements). At the Univ. of Houston we constructed the model SMM for monotonic loading.
a Universal Panel Tester capable of testing full-size elements Three sources of information are required to extend SMM to
共1.398 m ⫻ 1.398 m ⫻ 0.178 m兲 as shown in Fig. 1. From the ex- CSMM: (1) New cyclic constitutive relationships need to be de-
perimental behavior of such elements we would further establish veloped for concrete and embedded steel bars in the unloading
analytical models. These analytical models can then be integrated and reloading branches. (2) The difference between the envelope
into finite element programs to compute and predict the responses curves of concrete and steel bars under cyclic loading as com-
of whole structures or subassemblages (Wang and Hsu 2001). pared to the stress–strain curves under monotonic loading need to
Following this element-based approach, over 80 panel ele- be clarified. Tests reported in the companion paper (Mansour and
ments were subjected to monotonic loading over the past 15 years Hsu 2005) indicate that a new damage coefficient of concrete
to study the shear response of reinforced concrete elements under needs to be taken into account. (3) New Hsu/Zhu ratios under
monotonic loading. As a result, three monotonic analytical mod- cyclic loading have to be defined.
els were developed: the rotating-angle softened truss model (Hsu The first of the three sources of information was established
1993; Belarbi and Hsu 1994, 1995; Pang and Hsu 1995), the from the test results of six panels with steel bar angle of 90°
fixed-angle softened truss model (Pang and Hsu 1996; Hsu and 共␣2 = 90° 兲 subjected to reversed cyclic loading, as reported by
Zhang 1997), and the softened membrane model (SMM) (Hsu Mansour et al. (2001). The second and third sources of informa-
and Zhu 2001, 2002; Zhu and Hsu 2002). tion will be presented in this paper. In addition, this paper will
In order to understand the shear behavior of reinforced con- concisely present the whole analytical model by summarizing the
equilibrium and compatibility equations, the concept of biaxial
1
Research Assistant, Dept. of Civil and Environmental Engineering, strains versus uniaxial strains, the constitutive relationships of
Univ. of Houston, Houston, TX 77204-4003. materials, the solution algorithm, and the comparison of the
2
Moores Professor, Dept. of Civil and Environmental Engineering, CSMM predictions with the test results of 12 panels reported in
Univ. of Houston, Houston, TX 77204-4003.
the companion paper (Mansour and Hsu 2005).
Note. Associate Editor: Dat Duthinh. Discussion open until June 1,
2005. Separate discussions must be submitted for individual papers. To
extend the closing date by one month, a written request must be filed with Cyclic Softened Membrane Model
the ASCE Managing Editor. The manuscript for this paper was submitted
for review and possible publication on June 6, 2002; approved on May Equilibrium and Compatibility Equations
25, 2004. This paper is part of the Journal of Structural Engineering,
Vol. 131, No. 1, January 1, 2005. ©ASCE, ISSN 0733-9445/2005/1-54– The in-plane element considered in this model and shown in Fig.
65/$25.00. 1 is reinforced by two layers of orthogonal steel grids oriented at

54 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JANUARY 2005

J. Struct. Eng. 2005.131:54-65.

 ␥ᐉt ␥VH
= 共− ␧V + ␧H兲sin ␣2 cos ␣2 + 共cos2 ␣2 − sin2 ␣2兲 共6兲
2 2
In CSMM the set of strains ␧H , ␧V , ␧ᐉ, and ␧t in Eqs. (4)–(6)
are biaxial strains, taking into account the Hsu/Zhu ratios of
cracked reinforced concrete (Hsu and Zhu 2002; Zhu and Hsu
2002).

Biaxial Strains versus Uniaxial Strains

To solve the six equilibrium and compatibility equations, given
Downloaded from ascelibrary.org by INDIAN INSTITUTE OF TECHNOLOGY on 06/08/14. Copyright ASCE. For personal use only; all rights reserved.

by Eqs. (1)–(6), the stress–strain relationships of concrete and

steel have to be based on the biaxial strains ␧H , ␧V , ␧ᐉ, and ␧t.
Since laboratory experiments can provide only the uniaxial con-
stitutive relationships of concrete and steel (rather than the biaxial
constitutive relationships), the biaxial strains in Eqs. (4)–(6) must
be converted to the uniaxial strains before the uniaxial constitu-
tive relationships can be utilized. A set of four equations have
been derived (Hsu and Zhu 2002) to relate the set of biaxial
strains (␧H , ␧V , ␧ᐉ, and ␧t) to the set of uniaxial strains
¯ H , ¯␧V , ¯␧ᐉ, and ¯␧t) using the Hsu/Zhu ratios 共␯HV , ␯VH兲 as fol-
(␧
lows:
1 ␯HV
¯␧H = ␧H + ␧V 共7兲
1 − ␯HV␯VH 1 − ␯HV␯VH

␯VH 1
¯␧V = ␧H + ␧V 共8兲
Fig. 1. Typical reinforced concrete elements: (a) panels in CA series 1 − ␯HV␯VH 1 − ␯HV␯VH
and (b) panels in CE series
␥VH
¯␧ᐉ = ¯␧V cos2 ␣2 + ¯␧H sin2 ␣2 + 2 sin ␣2cos ␣2 共9兲
a given angle to the applied principal stress coordinate. Two ref- 2
erence Cartesian coordinates are used in the formulation. The first
reference Cartesian ᐉ – t coordinate system represents the longitu- ␥VH
¯␧t = ¯␧V sin2 ␣2 + ¯␧H cos2 ␣2 − 2 sin ␣2 cos ␣2 共10兲
dinal and transverse steel bar directions. The second reference 2
Cartesian H – V coordinate system represents the applied principal
Once the uniaxial strains, ¯␧H , ¯␧V , ¯␧ᐉ, and ¯␧t are determined, the
stresses in the horizontal and vertical directions. For computa-
tional purposes, it is assumed that the steel bars are uniformly stresses ␴Hc , ␴cV , ␶VH
c
, f ᐉ, and f t in Eqs. (1)–(3) can be calculated
distributed in the ᐉ and t directions, and the membrane element is using the uniaxial constitutive relationships under cyclic loading.
of uniform thickness. The principal applied stresses acting on the
four boundary edges of the element are assumed to be uniformly Cyclic Uniaxial Constitutive Relationships of Concrete
distributed.
The cyclic uniaxial constitutive relationships of cracked concrete
The three equilibrium equations that relate the applied stresses
in compression and tension are summarized in Fig. 2. The com-
(␴ᐉ , ␴t, and ␶lt) to the internal stresses of reinforcement (f ᐉ and f t)
pressive and tensile envelope curves for the cyclic stress–strain
and concrete (␴Hc , ␴cV, and ␶VH c
) in a membrane element are ex-
curves are taken from Mansour et al. (2001), except that a damage
pressed as (Pang and Hsu 1996)
coefficient D is incorporated in the compressive envelope of
␴ᐉ = ␴cV cos2 ␣2 + ␴Hc sin2 ␣2 + ␶VH
c
2 sin ␣2 cos ␣2 + ␳ᐉ f ᐉ 共1兲 Stages C1 and C2. This damage coefficient D takes into account
the effect of cyclic shear loading, where cyclic compression and
␴t = ␴cV sin2 ␣2 + ␴Hc cos2 ␣2 − ␶VH
c
2 sin ␣2 cos ␣2 + ␳t f t 共2兲 tension occur in both principal directions.
Tests conducted by Karsan and Jirsa (1969) showed that in the
␶ᐉt = 共− ␴cV + ␴Hc兲sin ␣2 cos ␣2 + ␶VH
c
共cos2 ␣2 − sin2 ␣2兲 共3兲 case of concrete cylinders under uniaxial cyclic compression, the
hysteretic loops of the compression stress–strain curves produced
The three compatibility equations, which define the compat- an envelope curve that was virtually identical to the curve for
ibility relationship between the steel strains (␧ᐉ , ␧t, and ␥ᐉt) in the monotonic compressive loading. In a similar vein, Mansour et al.
ᐉ – t coordinate of the reinforcement and the concrete strains (2001) showed that the compression envelope curves of the hys-
(␧H , ␧V, and ␥VH) in the V – H coordinate of the principal applied teretic loops for the three panels in the CVE3 series were the
stresses are expressed as follows (Pang and Hsu 1996): same as the monotonic compression curves proposed by Belarbi
and Hsu (1994, 1995).
␥VH
␧ᐉ = ␧V cos2 ␣2 + ␧H sin2 ␣2 + 2 sin ␣2 cos ␣2 共4兲 The difference between these two experiments is the fact that
2 the strain normal to the cyclic compression direction was zero in
Karsan and Jirsa’s tests, while a constant tensile strain was ap-
␥VH plied normal to the cyclic compression direction in the panels of
␧t = ␧V sin2 ␣2 + ␧H cos2 ␣2 − 2 sin ␣2 cos ␣2 共5兲
2 Mansour et al. (2001). This constant tensile strain in the orthogo-

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JANUARY 2005 / 55

J. Struct. Eng. 2005.131:54-65.

Downloaded from ascelibrary.org by INDIAN INSTITUTE OF TECHNOLOGY on 06/08/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 2. Cyclic smeared stress–strain curves of concrete

nal direction caused a “softening” of the concrete compressive caused by the history of tensile and compressive stress reversal
strength. Mansour et al. (2001) showed that the “softening coef- normal to the compression direction being considered.
ficient ␨” proposed by Zhang and Hsu (1998) was valid not only To be consistent with the concept of softening coefficient due
for the monotonic loading curves, but also for the envelope curves to tensile strain, the damage coefficient due to compression, de-
of cyclic loading. noted as D, is taken as a linear function of the compression strain
When a panel is subjected to cyclic shear loading, however, an ␧⬘c
additional phenomenon needs to be considered when modeling
the constitutive relationships of concrete in compression. Since ␧⬘c
the horizontal and vertical principal applied stresses are subjected D=1−␺ 艋 1.0 共11兲
␧0
to out-of-phase compression–tension stresses, a damage coeffi-
cient needs to be incorporated in the envelope compression The strain ␧⬘c (always negative) in Eq. (11) is the maximum com-
stress–strain curves of concrete to take into account the damage pression strain normal to the compression direction under consid-

56 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JANUARY 2005

J. Struct. Eng. 2005.131:54-65.

Downloaded from ascelibrary.org by INDIAN INSTITUTE OF TECHNOLOGY on 06/08/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 3. Cyclic smeared stress–strain curves of mild steel bars embedded in concrete

eration and occurred in the previous loading cycles. The compres- Constitutive Relationships of Concrete in Shear
sive strain ␧0 (always negative) is the concrete cylinder
The rational equation relating the shear stress of concrete 共␶VHc
兲
compressive strain at the peak cylinder stress f ⬘c . The symbol ␺ is
and the shear strain 共␥VH兲 in the H – V plane is given by Zhu et al.
a constant taken as 0.4. The value of ␺ = 0.4 was chosen to best fit
(2001)
the test results of the cyclic shear stress-strain curves of the test
panels (Mansour 2001).
The damage coefficient D is incorporated into the backbone ␴Hc − ␴cV
␶VH
c
= ␥VH 共12兲
envelope curves of concrete in compression together with the 2共␧H − ␧V兲
softening coefficient ␨, as shown by the concrete compression
equations (Stages C1 and C2) in Fig. 2. Because the damaging
effect of the perpendicular tensile strain ␧T⬘ (or the uniaxial ¯␧T⬘ ) is Cyclic Constitutive Relationships of Embedded Mild
taken care of by the softening coefficient ␨, the damage coeffi- Steel Bars
cient D in Eq. (11) cannot be greater than unity, and the strain ␧c⬘
cannot be positive. When the damage coefficient D becomes The cyclic constitutive relationships of reinforcing steel bars em-
unity, the envelope compression stress–strain curve of concrete bedded in concrete and subjected to uniaxial strains (Mansour et
becomes identical to the monotonic compression stress–strain al. 2001) are summarized in Fig. 3. These smeared stress versus
curve of concrete. smeared strain curves (solid curves) of embedded steel bars under

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JANUARY 2005 / 57

J. Struct. Eng. 2005.131:54-65.

 cyclic loading are compared to the monotonic stress–strain rela-
tionship of a bare bar (dotted lines). It can be seen that the cyclic
stress–strain curves of embedded steel bars have the following
two characteristics. First, the tensile envelope curves for the cy-
clic stress–strain curves of embedded steel bars closely resemble
the monotonic stress–strain curves of embedded steel bars devel-
oped by Belarbi and Hsu (1994, 1995) and verified by Hsu and
Zhang (1996). Second, the new unloading and reloading stress–
strain curves of embedded steel bars take into account the
Bauschinger effect.
Downloaded from ascelibrary.org by INDIAN INSTITUTE OF TECHNOLOGY on 06/08/14. Copyright ASCE. For personal use only; all rights reserved.

Hsu/Zhu Ratios for Cyclic Loading

The SMM was developed to predict the entire behavior (before
and after peak point) of reinforced concrete panel elements under
monotonic shear stresses taking into account the Hsu/Zhu ratios.
Zhu and Hsu (2002) measured the two Hsu/Zhu ratios of panels
under monotonic shear stresses. The first ratio ␯12 (representing
the effect of the compression strain in the 2 direction on the
tensile strain in the 1 direction) was found to increase with the
increase of steel strain and became a constant 1.9 after yielding.
The second ratio ␯21 (representing the effect of the tensile strain
in the 1 direction on the compression strain in the 2 direction) was
found to be essentially zero throughout the loading history.
Under cyclic loading, however, repeated cycles of crack clos-
ing and opening are expected to reduce the Hsu/Zhu ratio ␯12 (or
␯TC as it will be denoted under cyclic loading). In CSMM, it is
assumed that the Hsu/Zhu ratio ␯TC (the effect of the compression
strain on the tensile strain) will be 1.0 under cyclic loading based
on the comparative studies of shear strains of panels (Mansour Fig. 4. Flow chart of solution procedure
2001). The Hsu/Zhu ratio ␯CT (the effect of the tensile strain on
the compression strain) under cyclic loading will be taken as zero,
identical to the Hsu/Zhu ratio ␯21 under monotonic loading case. Comparison of Cyclic Softened Membrane Model
with Test Results

Solution Algorithm Cyclic Shear Stress–Strain Curves (Hysteretic Loops)

The CSMM was used to predict the behavior of 12 panels tested
The solution procedure for the proposed CSMM model is given by Mansour and Hsu (2005). Figs. 5(a–l) compare the predicted
by the flow chart in Fig. 4. Two equilibrium equations, which cyclic shear-stress versus shear-strain curves of panels (dotted
make the solution procedure more efficient, are derived from the curves) to the experimental curves (solid curves). The planning of
basic equilibrium Eqs. (1) and (2) the 12 panels in Fig. 5 includes two variables: (1) angle of steel
bar orientation with respect to the applied principal vertical stress
(␣2 = 45, 68.2, 79.8, and 90°), and (2) percentages of steel 共␳t , ␳ᐉ兲
␳ᐉ f ᐉ + ␳t f t = 共␴ᐉ + ␴t兲 − 共␴cV + ␴Hc兲 共13兲 in the two directions of the panels. ␳t and ␳ᐉ could each vary from
0.54 to 2.7%. It can be seen that the CSMM is able to predict the
pinched shapes of the hysteretic loops of the shear stress–strain
␳ᐉ f ᐉ − ␳t f t = 共␴ᐉ − ␴t兲 − 共␴cV − ␴Hc兲cos 2␣2 − 2␶VH
c
sin 2␣2 curves. It is clear that the pinched shape of hysteretic loops is the
result of the steel bar direction deviating from the principal stress
共14兲
directions and can be rationally predicted. The postpeak shear
stress–strain curves of all the test panels are also reasonably pre-
dicted, testifying to the correctness of the proposed damage coef-
Eqs. (13) and (14) replace Eqs. (1) and (2) in the solution algo- ficient and the Hsu/Zhu ratios for cyclic loading.
rithm as shown in Fig. 4.
In this model the nonlinear solution is controlled by means of
a stepwise increment in the value of the horizontal biaxial strain Horizontal-Strain versus Vertical-Strain Curves
␧H according to a prescribed biaxial horizontal strain history. An Figs. 6(a–d) compare the predicted horizontal-strain versus
iteration procedure is performed for each value of ␧H until all vertical-strain curves to the experimental curves of four test pan-
equilibrium, compatibility, and constitutive equations are satis- els (CA3, CD3, CF2, and CE3). These four panels are chosen to
fied. Fig. 4 shows that this solution procedure of CSMM evolves cover the four angles of steel bar orientation (␣2 = 45, 68.2, 79.8,
from that of SMM. Even for the complicated cyclic loading, the and 90°). The mechanical and geometrical characteristics of these
clarity and the generality of this solution procedure are evident. panels are given in the companion paper (Mansour and Hsu

58 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JANUARY 2005

J. Struct. Eng. 2005.131:54-65.

 2005). Fig. 6 shows a good agreement between the predicted
curves (dotted curves) and the experimental curves (solid curves),
and thus illustrates the success of CSMM in predicting the defor-
mation history.
For panels CA3 共␣2 = 45° 兲 and CD3 共␣2 = 68.2° 兲, shown in
Figs. 6(a and b), the horizontal and vertical strains reach the nega-
tive region in every cycle, meaning that the cracks are fully
closed and the concrete struts are formed to resist compression. In
contrast, Figs. 6(c and d) show that the horizontal and vertical
strains in Panels CF2 共␣2 = 79.8° 兲 and CE3 共␣2 = 90° 兲 move away
Downloaded from ascelibrary.org by INDIAN INSTITUTE OF TECHNOLOGY on 06/08/14. Copyright ASCE. For personal use only; all rights reserved.

from the negative region with every cycle of loading, meaning

that the cracks in these panels were never completely closed after
yielding, and the principal compressive stresses are resisted by
steel bars in compression, rather than concrete.

Preyield Shear Stiffness „K␥…

The predicted and experimental preyield shear stiffnesses
共K␥+ , K␥−兲 of the 12 test panels subjected to positive and negative
shear stresses are recorded in Table 1. These panels are designed
to study the two main variables: the steel grid orientation 共␣2兲 and
the steel ratios in the longitudinal and transverse directions
共␳ᐉ , ␳t兲. Table 1 clearly shows that the CSMM can accurately
predict the preyield shear stiffnesses of the panels.

Envelope Shear Ductility Factors „␮E␥…

The CSMM predicted and the experimental values of the enve-
lope shear ductility factors ␮E␥
+
and ␮E␥−
in both the positive and
the negative loading directions are also listed in Table 1. It can be
seen that the predicted and the experimental values agree very
well.

Shear Energy Dissipation Factors „␮D…

Table 1 also gives the CSMM predicted and the experimental
values of shear energy dissipation factors ␮D for the 12 test pan-
els. Here again, the CSMM is shown to be capable of accurately
predicting the ␮D factors of all test panels.

Pinching Mechanism

To better understand the pinching mechanism in the hysteretic

loops of the shear stress–strain curves, the predicted results of two
typical panels (CA3 and CE3) with two different steel bar orien-
tations will be analyzed by means of Mohr circles of strains and
stresses.

Pinching Mechanism in Panel CA3 (␣2 = 45°)

The CSMM predicted results of panel CA3 are analyzed to illus-
trate the severe pinching effect, [Fig. 5(b)], when the steel bars
Fig. 5. Predicted and experimental hysteretic loops of shear-stress are oriented at 45° to the direction of the principal applied
versus shear-strain curves of test panels (solid curves show experi- stresses, [Fig. 1(a)]. Fig. 7 shows the first cycle of the hysteretic
mental results, dotted curves show predicted results): (a) CA2 共␣2 loops beyond yielding in terms of shear stress ␶45° versus shear
= 45° ; ␳ᐉ = ␳t = 0.77% 兲; (b) CA3 共␣2 = 45° ; ␳ᐉ = ␳t = 1.7% 兲: (c) CA4 strain ␥45° in the 45° direction. Four points A , B , C, and D are
共␣2 = 45° ; ␳ᐉ = ␳t = 2.7% 兲; (d) CB3 共␣2 = 45° ; ␳ᐉ = 1.7% ; ␳t chosen in Fig. 7 to illustrate the presence of the pinched shape.
= 0.77% 兲; (e) CB4 共␣2 = 45° ; ␳ᐉ = 2.7% ; ␳t = 0.67% 兲; (g) CD3 共␣2
Point A is at the maximum positive shear strain of the first cycle
= 68.2° ; ␳ᐉ = ␳t = 1.3% 兲; (h) CD4 共␣2 = 68.2° ; ␳ᐉ = ␳t = 2.3% 兲; (i)
beyond yielding. Point B is at the stage where the shear stress is
CF2 共␣2 = 79.8° ; ␳ᐉ = ␳t = 0.56% 兲; (j) CE2 共␣2 = 90° ; ␳ᐉ = ␳t
almost zero after unloading. Point C is taken in the negative shear
= 0.54% 兲; (k) CE3 共␣2 = 90° ; ␳ᐉ = ␳t = 1.2% 兲; and (l) CE4 共␣2
strain region at the end of the low-stress pinching zone just before
= 90° ; ␳ᐉ = ␳t = 1.9% 兲
the sudden increase in stiffness. Point D is at the maximum nega-

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JANUARY 2005 / 59

J. Struct. Eng. 2005.131:54-65.

Downloaded from ascelibrary.org by INDIAN INSTITUTE OF TECHNOLOGY on 06/08/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 6. Predicted and experimental horizontal-strain versus vertical-strain curves of four test panels (solid curves show experimental results,
dotted curves show predicted results): (a) CA3; ␣2 = 45° , ␳ᐉ = ␳t = 1.7%; (b) CD3; ␣2 = 68.2° , ␳ᐉ = ␳t = 1.3%; (c) CF2; ␣2 = 79.8° , ␳ᐉ = ␳t
= 0.56%; and (d) CE3; ␣2 = 90° , ␳ᐉ = ␳t = 1.2%

tive shear strain of the first negative cycle after yielding. The the maximum applied shear stress ␶45° of 6.84 MPa produces a
three segments of curves from Points A to D in Fig. 7 clearly large shear strain ␥45° of 0.00828 (twice the number shown in the
define the pinched shape of the hysteretic loops. It should be Mohr circle because the vertical axis represents ␥ / 2). To resist
recalled that a state of positive shear stress is created in the panel this applied shear stress, the concrete is subjected to a maximum
by applying a principal tensile stress in the horizontal direction vertical compressive stress ␴cV of 13.64 MPa, and a maximum
and an equal but opposite compressive stress in the vertical direc-
smeared steel stresses ␳f of 6.80 MPa in both the longitudinal and
tion, while a state of negative shear stress is created by applying
transverse directions.
a principal tensile stress in the vertical direction and an equal but
opposite compressive stress in the horizontal direction. When the panel is unloaded from Points A to B, the applied
The smeared biaxial strains, the applied stresses, the smeared shear stress of 6.84 MPa is reduced to virtually zero (Mohr circle
concrete stresses, and the smeared steel stresses at the four points of applied stresses shrinks to almost a point). From equilibrium,
(A , B , C, and D) chosen in Fig. 7 are represented by Mohr the compressive stress in the concrete and the tensile stress in the
circles as shown in Fig. 8. At Point A in the first postyield cycle, steel also approach zero (Mohr circle of concrete stresses ap-

60 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JANUARY 2005

J. Struct. Eng. 2005.131:54-65.

 Table 1. Comparison of Cyclic Softened Membrane Model (CSMM) Predicted and Experimental Preyield Shear Stiffnesses, Envelope Shear Ductility
Factors, and Shear Energy Dissipation Factors
K␥+ K␥− K␥+ K␥−
(MPa) (MPa) (MPa) (MPa) ␮E␥
+
␮E␥
−
␮E␥
+
␮E␥
−
␮D ␮D
Panel (Experimental) (Experimental) (CSMM) (CSMM) (Experimental) (Experimental) (CSMM) (CSMM) (Experimental) (CSMM)
CA2 901 884 920 907 7.54+ 8.15+ 7.84 7.67 26.3+ 27.0
CA3 1,489 1,377 1,495 1,480 4.18 3.66 4.34 4.23 11.3 12.3
CA4 1,790 1,804 1,820 1,835 1.99 1.66 1.76 1.74 5.60 5.89
CB3 1,147 1,041 960 965 4.83 5.24 5.81 4.54 20.9 22.1
Downloaded from ascelibrary.org by INDIAN INSTITUTE OF TECHNOLOGY on 06/08/14. Copyright ASCE. For personal use only; all rights reserved.

CB4 1,124 1,030 975 980 2.92 2.95 2.75 2.86 19.6 21.0
CD2 1,167 1,049 1,070 1,056 7.77+ 7.68+ 7.67 7.65 46.5+ 44.1
CD3 2,273 1,981 2,334 2,340 5.38 5.82 6.12 5.94 34.4 36.1
CD4 2,880 2,771 2,960 2,950 4.80 4.39 4.54 4.47 30.2 34.2
CF2 1,633 1,470 1,670 1,540 6.66+ 6.39+ 6.68 6.51 54.8+ 56.9
CE2 1,155 1,426 1,310 1,360 10.45+ 10.53+ 10.3 9.76 224.3+ 212.4
CE3 2,414 2,576 2,440 2,430 9.54+ 9.10+ 9.71 9.33 217.5+ 204.6
CE4 4,268 3,304 4,303 4,290 6.42+ 4.96+ 6.42 5.25 50.0+ 46.7
Note: Definitions of preyield shear stiffnesses 共K␥兲, envelope shear ductility factors 共␮E␥兲, and shear energy dissipation factors 共␮D兲 are given in the
companion paper (Mansour and Hsu 2005). Values followed by the positive signs 共+兲 correspond to the maximum stroke of jacks or maximum limits of
LVDTs. (More details are given in Table 4 of companion paper).

proaches a point). Correspondingly, the shear strain is reduced the concrete stress and the steel stress remain small at Point C.
from 0.00828 at Point A to a value of 0.00288 at Point B. This The corresponding Mohr circles of applied stresses and concrete
unloading process produces an almost linear shear stiffness due to stresses continue to look like two points.
the normal relaxation of steel and concrete. These nearly propor- When the negative shear strain reaches −0.00862 at Point D,
tional reductions of stresses in the concrete and steel are also the vertical cracks are fully closed and the concrete compressive
related to the closing of vertical cracks. The corresponding hori- struts are fully formed. The concrete struts can now resist a com-
zontal biaxial strain ␧H decreases from 0.00762 at Point A to pressive stress of 13.50 MPa. In order to satisfy equilibrium con-
0.00278 at Point B, but remains in tension with crack widths of ditions, the smeared steel in both the longitudinal and transverse
significant size. The compressive biaxial strain in the vertical di- directions have to resist a tensile stress of 6.59 MPa (the tensile
rection ␧V decrease from −0.00067 at Point A to −0.00011 at resistance of concrete can be ignored). Correspondingly, the ele-
Point B. ment is resisting an applied shear stress of 6.91 MPa. In other
When the positive shear strain of 0.00288 at Point B is re- words, in the reloading CD range the shear stiffness is restored to
versed to become a negative shear strain of −0.00269 at Point C, its normal magnitude.
the vertical biaxial strain increases to a tensile strain of 0.00273, The pinched shape of the hysteretic loop of Panel CA3 in Fig.
while the horizontal biaxial strain further decreases to a small 7 is formed by a small shear stiffness in the BC region, sand-
tensile value of 0.000039 (not in compression). This large change wiched between two large shear stiffnesses in the AB and the CD
of shear strain through the origin, however, is not accompanied by regions.
a corresponding change in the applied shear stress, meaning that
the shear stiffness in the BC region is very small. This is because
Absence of Pinching Mechanism in Panel CE3
at Point C the vertical cracks did not fully close, and the horizon-
(␣2 = 90°)
tal compressive stress of concrete could not be developed. With-
out forming an effective set of concrete compressive struts, the The predicted results of Panel CE3 are analyzed to illustrate the
stresses in the steel bars also could not be developed. Hence, both absence of pinching mechanism [Fig. 5(k)] when the steel bars
are oriented in the direction of the principal applied stresses [Fig.
1(b)]. In this case the hysteretic loops of the shear stress–strain
curves are robust, well rounded, and did not show any trace of
pinching. Fig. 9 shows the first cycle of hysteretic loop beyond
yielding in terms of shear stress ␶45° versus shear strain ␥45° in the
45° direction. Four points A , B , C, and D are chosen in Fig. 9 to
illustrate the absence of the pinched shape. The definitions of
Points A , B, and D are the same as those given for Panel CA3,
except that Point C is now taken in the negative strain region
when the steel reaches yielding. The three segments of curves
from Points A to D in Fig. 9 clearly define the absence of the
pinched shape in the hysteretic loops of Panel CE3.
Fig. 10 shows the Mohr circles at the four points, A , B , C,
and D, that are chosen in Fig. 9. At Point A in the first postyield
cycle, the maximum applied shear stress ␶45° of 4.79 MPa pro-
Fig. 7. Shear-stress versus shear-strain curves of panel CA3 after first duces a large shear strain ␥45° of 0.00374. To resist this applied
yield shear stress, the concrete is subjected to a maximum vertical com-

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JANUARY 2005 / 61

J. Struct. Eng. 2005.131:54-65.

Downloaded from ascelibrary.org by INDIAN INSTITUTE OF TECHNOLOGY on 06/08/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 8. Mohr circles of smeared biaxial strains and stresses at points A , B , C, and D for panel CA3 共␣2 = 45° 兲

pressive stress ␴cV of 4.58 MPa, and a maximum smeared tensile When the positive shear strain of 0.00182 at Point B is re-
steel stresses ␳ᐉ f ᐉ of 4.47 MPa in the longitudinal directions. The versed to become a negative shear strain of −0.00178 at Point C,
steel stress of −0.21 MPa in the transverse direction is negligible, the vertical strain increases to a tensile strain of 0.00192, while
since most of the compressive stresses are being carried by con- the horizontal strain further decreases to a small tensile value of
crete in the vertical directions rather than the vertical steel. 0.00014 (not in compression). This large change of shear strain
When the panel is unloaded from Points A to B, the applied through the origin is accompanied by a large increase of the ap-
shear stress of 4.79 MPa is reduced to virtually zero (Mohr’s
circle of applied stresses becomes virtually a point as shown in
Fig. 10). From equilibrium, the compressive stress in the concrete
struts and the tensile stress in the steel bars also approach zero.
Correspondingly, the shear strain is reduced from 0.00374 at
Point A to a value of 0.00182 at Point B. This unloading process
produces an almost linear shear stiffness due to the normal relax-
ation of steel and concrete. These nearly proportional reductions
of stresses in the concrete and steel are also related to the closing
of the vertical cracks. The corresponding horizontal strain ␧H de-
creases from 0.00362 at Point A to 0.00181 at Point B, but re-
mains in tension with crack width of significant size. The com-
pressive strain in the vertical direction ␧V also decreases from Fig. 9. Shear-stress versus shear-strain curves of panel CE3 after first
−0.00012 at Point A to nearly zero at Point B. yield

62 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JANUARY 2005

J. Struct. Eng. 2005.131:54-65.

Downloaded from ascelibrary.org by INDIAN INSTITUTE OF TECHNOLOGY on 06/08/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 10. Mohr circles of smeared biaxial strains and stresses at points A , B , C, and D for panel CE3 共␣2=90° 兲

plied shear stress, meaning that the shear stiffness in the BC re- stant up to Point D. The smeared strain, however, continues to
gion is large. This large shear stress is supplied by a simple in- increase. At Point D the shear strain is −0.00386, and the vertical
ternal mechanism as follows: although the vertical cracks at Point strain reaches a value of 0.00400.
C did not fully close to form compression struts and the Mohr
circle for concrete is very small, the principal compressive stress
can be resisted by the longitudinal steel. It is clear that the large Failure Mechanism under Cyclic Loading
shear stress is resisted primarily by the transverse steel with a
smeared stress of 4.24 MPa in the principal tension direction, and When a reinforced concrete structure is subjected to static load-
by the longitudinal steel with a smeared compressive stress of ing, the principal compression stresses in the structure can be
−3.43 MPa in the principal compression direction. resisted by concrete struts while the principal tensile stresses are
After yielding of steel at Point C, the applied shear stress and resisted by the reinforcing bars. This strut-and-tie concept can be
the stresses in the concrete and steel all remain essentially con- used to design all reinforced concrete structures under static

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JANUARY 2005 / 63

J. Struct. Eng. 2005.131:54-65.

 loads. However, comparison of the cyclic behavior of Panel CE3 cracked reinforced concrete under reversed cyclic shear load-
共␣2 = 90° 兲 and Panel CA3 共␣2 = 45° 兲 in the previous sections ing, and to design structures that are capable of dissipating
clearly shows that the strut-and-tie model is no longer valid for earthquake energy.
application to reversed cyclic loading. When reinforced concrete
elements (panels) are resisting cyclic loading beyond the yielding
of the steel, crack widths increase in both directions with each
cycle of loading because panels are expanding. The crack will not Acknowledgments
close after unloading and the concrete struts cannot be formed to
resist the reversed loading, unless significant pinching is intro- This research was supported by Grant No. CMS-9711084 from
duced to close the cracks. Therefore, the steel bars must be de- the National Science Foundation. The steel bars were specially
made and donated by Chaparral Steel Co., Midlothian, Tex.
Downloaded from ascelibrary.org by INDIAN INSTITUTE OF TECHNOLOGY on 06/08/14. Copyright ASCE. For personal use only; all rights reserved.

signed to resist principal compression stresses as well as principal

tension stresses. Panel CE3 共␣2 = 90° 兲, which is designed based
on this concept, performs very well. It does not have the pinching
problem, and exhibits ductile behavior and high energy dissipa-
tion capacity. Notation
However, when the steel bars are oriented at an angle of 45° to
the principal applied stress coordinate as in Panel CA3 共␣2 The following symbols are used in this paper:
= 45° 兲, steel bars in both longitudinal and transverse directions A ⫽ coefficient in unloading and reloading equation of
are always in tension. Then the cracks are forced to close in order steel bar, given as 1.9k−0.1p ;
to form the concrete compression struts required in establishing B ⫽ 共f cr / f y兲1.5 / ␳, parameter defining smeared yielding
an equilibrium truss with the steel tension ties. This forced clos- stress of mild steel bars embedded in concrete;
ing of cracks and the subsequent reopening of cracks in each D ⫽ damage factor of compressive concrete under cyclic
cycle of loading represent a very destructive failure mechanism shear;
(Hsu and Mansour 2002), leading to the rapid deterioration of Ec ⫽ modulus of elasticity of concrete;
bond and the ultimate disintegration of concrete. This undesirable Ecc ⫽ slope of linear unloading and reloading expression
failure mechanism is responsible for the early arrival of the de- of concrete;
scending branch, the low ductility, and the low capacity in energy Es ⫽ modulus of elasticity of bare steel bar;
dissipations. f c⬘ ⫽ cylinder compressive strength of concrete;
f cr ⫽ cracking tensile strength of concrete;
f ᐉ ⫽ smeared steel stress in longitudinal direction
(ᐉ axis);
Conclusions f n ⫽ smeared yield stress of embedded steel bar in
concrete;
1. The CSMM, a new rational smeared model, is presented to f s ⫽ stress in mild steel, f s becomes f ᐉ or f t, when
predict the behavior of reinforced concrete elements under applied to the longitudinal and transverse steel,
reversed cyclic shear stresses. The CSMM is an extension of respectively;
the SMM for monotonic loading (Hsu and Zhu 2002) and f t ⫽ smeared steel stress in transversal direction (t axis);
includes the following three new ingredients: (1) new consti- f y ⫽ yielding strength in bare steel bars;
tutive relationships are established for concrete and steel in H ⫽ direction of applied principal horizontal stress;
the unloading and reloading stages under cyclic loading K␥ ⫽ preyield shear stiffness;
(Mansour et al. 2001); (2) a damage factor is introduced into k p ⫽ plastic strain ratio given as ␧ p / ␧n;
the equations of the envelope curves of concrete in compres- ᐉ ⫽ direction of longitudinal steel bars;
sion; and (3) Hsu/Zhu ratios are modified for application to R ⫽ coefficient in unloading and reloading equation of
cyclic loading. steel bar, given as 10k−0.2p ;
2. The CSMM is capable of predicting the hysteretic loops of t ⫽ direction of transverse steel bars;
the shear-stress versus shear-strain curves of concrete ele- V ⫽ direction of applied principal vertical stress;
ments reinforced with various percentages of mild steel bars ␣2 ⫽ angle between longitudinal steel bars (ᐉ axis) and
that are oriented in various angles with respect to the princi- applied principal vertical stress (V axis);
pal applied stresses. The CSMM can correctly predict the ␥VH ⫽ smeared shear strain in H – V coordinate;
behavior of shear elements required for earthquake design: ␥ᐉt ⫽ smeared shear strain in ᐉ – t coordinate;
the stresses and strains at yield and at failure, the shear stiff- ␥45° ⫽ smeared shear strain at 45° to horizontal direction;
ness, the shear ductility, and the energy dissipation capacity. ¯␧ ⫽ smeared uniaxial concrete strain, could be
3. The CSMM can explain the presence or the absence of compression or tension;
pinched shape in the hysteretic loops, which are related to the ␧⬘c ⫽ maximum compression strain normal to direction
process of how the cracks open and close and how the steel being considered and occurring in previous loading
bars are oriented to resist such opening and closing. The best cycles;
way to control the opening and closing of cracks is to orient ␧cr ⫽ cracking strain of concrete taken as 0.00008;
the steel bars in the two directions of the applied principal ␧H ⫽ smeared strain in horizontal direction;
stresses, so that the steel bars can resist both the principal ¯␧H ⫽ uniaxial smeared strain in horizontal direction;
tensile stress and the principal compressive stress. The steel ¯␧i ⫽ uniaxial concrete strain at load reversal point “i”;
bars in compression can counter the destructive failure ¯␧i+1 ⫽ uniaxial concrete strain at end of stage;
mechanism of forced closing and reopening of cracks. The ␧ᐉ ⫽ smeared strain of longitudinal steel bars (ᐉ axis);
CSMM allows us to understand the failure mechanism of ¯␧ᐉ ⫽ uniaxial smeared strain of longitudinal steel bars;

64 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JANUARY 2005

J. Struct. Eng. 2005.131:54-65.

 ¯␧n ⫽ smeared tensile strain of mild steel bars embedded References
in concrete at first yield;
¯␧ p ⫽ plastic strain of steel bars embedded in concrete; Belarbi, A., and Hsu, T. T. C. (1994). “Constitutive laws of concrete in
␧s ⫽ smeared strain of steel bars embedded in concrete; tension and reinforcing bars stiffened by concrete.” Struct. J. Am.
¯␧T⬘ ⫽ uniaxial tensile strain normal to compression Concrete Institute, 91(4), 465–474.
direction being considered; Belarbi, A., and Hsu, T. T. C. (1995). “Constitutive laws of softened
␧t ⫽ smeared strain of transverse steel bars (t axis); concrete in biaxial tension–compression.” Struct. J. Am. Concrete In-
¯␧t ⫽ uniaxial smeared strain of transverse steel bars; stitute, 92(5), 562–573.
␧V ⫽ smeared strain in vertical direction; Hsu, T. T. C. (1993). Unified theory of reinforced concrete, CRC, Boca
¯␧V ⫽ uniaxial smeared strain in vertical direction; Raton, Fla.
␧y ⫽ yield strain of bare steel bar;
Downloaded from ascelibrary.org by INDIAN INSTITUTE OF TECHNOLOGY on 06/08/14. Copyright ASCE. For personal use only; all rights reserved.

Hsu, T. T. C., and Mansour, M. Y. (2002). “Failure mechanism of rein-

␧0 ⫽ concrete cylinder strain at maximum (peak) forced concrete elements under cyclic loading.” Concrete: Material
compressive stress f ⬘c ;
science to applications—A tribute of surendra shah, American Con-
␨ ⫽ softening coefficient of concrete in compression;
crete Institute, Farmington Hill, Mich., 1–24.
␩ ⫽ parameter defined as 共␳t f ty − ␴t兲 / 共␳ᐉ f ᐉy − ␴ᐉ兲;
Hsu, T. T. C., and Zhang, L. X. (1996). “Tension stiffening in reinforced
␩⬘ ⫽ ␩ or its reciprocal whichever is less than unity;
concrete membrane elements.” Struct. J. Am. Concrete Institute,
␮D ⫽ shear energy dissipation factor;
93(1), 108–115.
␮E␥ ⫽ envelope shear ductility factor;
Hsu, T. T. C., and Zhang, L. X. (1997). “Nonlinear analysis of membrane
␯CT ⫽ Hsu/Zhu ratio of cracked reinforced concrete
elements by fixed-angle softened-truss model.” Struct. J. Am. Con-
(increment of compression strain due to increment of
crete Institute, 94(5), 483–492.
tensile strain);
Hsu, T. T. C., and Zhu, R. R. H. (2001). “Post-yield behavior of rein-
␯HV ⫽ Hsu/Zhu ratio of cracked reinforced concrete
forced concrete membrane elements - The Hsu/Zhu ratios.” Modeling
(increment in horizontal strain due to increment in
of inelastic behavior of RC structures under seismic loads, American
vertical strain);
Society of Civil Engineers, Reston, Va., 139–157.
␯TC ⫽ Hsu/Zhu ratio of cracked reinforced concrete
Hsu, T. T. C., and Zhu, R. R. H. (2002). “Softened membrane model for
(increment of tensile strain due to increment of
reinforced concrete elements in shear.” Struct. J. Am. Concrete Insti-
compression strain);
tute, 99(4), 460–469.
␯VH ⫽ Hsu/Zhu ratio of cracked reinforced concrete
Karsan, I. D., and Jirsa, J. O. (1969). “Behavior of concrete under com-
(increment in vertical strain due to increment
pressive loadings.” J. Struct. Div. ASCE, 95(12), 2543–2563.
in horizontal strain);
Mansour, M. (2001). “Behavior of reinforced concrete membrane ele-
␯12 ⫽ Hsu/Zhu ratio of cracked reinforced concrete
ments under cyclic shear: Experiments to theory.” PhD dissertation,
(increment of strain in 1 direction due to increment of
Univ. of Houston, Houston.
strain in 2 direction);
Mansour, M., and Hsu, T. T. C. (2005). “Behavior of reinforced concrete
␯21 ⫽ Hsu/Zhu ratio of cracked reinforced concrete
elements under cyclic shear. I: Experiments.” J. Struct. Eng., 131(1),
(increment of strain in 2 direction due to increment of
44-53.
strain in 1 direction);
Mansour, M., Lee, J. Y., and Hsu, T. T. C. (2001). “Cyclic stress–strain
␳ᐉ ⫽ longitudinal steel ratio;
curves of concrete and steel bars in membrane elements.” J. Struct.
␳t ⫽ transversal steel ratio;
Eng., 127(12), 1402–1411.
␴c ⫽ concrete stress, could be compression or tension;
Pang, X. B., and Hsu, T. T. C. (1995). “Behavior of reinforced concrete
␴H ⫽ applied principal stress in the horizontal direction;
membrane elements in shear.” Struct. J. Am. Concrete Institute,
␴Hc ⫽ smeared concrete stress in the horizontal direction;
92(6), 665–679.
␴ci ⫽ concrete stress at load reversal point “i”;
Pang, X. B., and Hsu, T. T. C. (1996). “Fixed-angle softened-truss model
␴i+1 ⫽ concrete stress at end of stage;
c
for reinforced concrete.” Struct. J. Am. Concrete Institute, 93(2),
␴l ⫽ applied normal stress in ᐉ-axis;
197–207.
␴cᐉ ⫽ smeared stress of concrete in ᐉ-axis;
Wang, T. J., and Hsu, T. T. C. (2001). “Nonlinear finite element analysis
␴t ⫽ applied normal stress in t-axis;
of concrete structures using new constitutive models.” Comput.
␴V ⫽ applied principal stress in vertical direction;
Struct. 79, 2781–2791.
␴cV ⫽ smeared concrete stress in vertical direction;
Zhang, L. X., and Hsu, T. T. C. (1998). “Behavior and analysis of 100
␶ᐉt ⫽ applied shear stress in ᐉ – t coordinate;
MPa concrete membrane elements.” J. Struct. Eng., 124(1), 24–34.
␶cᐉt ⫽ smeared shear stress of cracked concrete in ᐉ – t
Zhu, R. R. H., and Hsu, T. T. C. (2002). “Poisson effect of reinforced
coordinate;
concrete membrane elements.” Struct. J. Am. Concrete Institute,
␶VHc
⫽ smeared shear stress of concrete in V – H coordinate; 99(5), 631–640.
␶45° ⫽ shear stress at 45° to horizontal direction; Zhu, R. H., Hsu, T. T. C., and Lee, J. Y. (2001). “Rational shear modulus
␶45° ⫽ smeared shear stress at 45° direction to horizontal for smeared crack analysis of reinforced concrete.” Struct. J. Am.
direction; and Concrete Institute, 98(4), 443–450.
␺ ⫽ constant taken as 0.4 in damage coefficient D.

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JANUARY 2005 / 65

J. Struct. Eng. 2005.131:54-65.