Вы находитесь на странице: 1из 11

Engineering Structures 31 (2009) 1944–1954

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

A softened membrane model for torsion in reinforced concrete members


Chyuan-Hwan Jeng a,∗ , Thomas T.C. Hsu b
a
Department of Civil Engineering, National Chi Nan University, Nantou, Taiwan
b
Department of Civil and Environmental Engineering, University of Houston, Houston, TX, USA

article info abstract


Article history: The Softened Membrane Model (SMM), developed for predicting the behavior of reinforced concrete (RC)
Received 6 August 2008 membrane elements under shear, is extended to RC members subjected to torsion. This new analytical
Received in revised form method, referred to as the Softened Membrane Model for Torsion (SMMT), takes into account the strain
16 February 2009
gradient of concrete struts in the shear flow zone by making two modifications to the constitutive
Accepted 19 February 2009
Available online 15 May 2009
relationships of concrete. First, in the tensile stress–strain relationship of concrete, the pre-cracking
stiffness and the strain at peak stress should each be increased by 45%. Second, the Hsu/Zhu ratio for
Keywords:
torsion is taken as 80% of the Hsu/Zhu ratio for shear. Similar to the case of the SMM model for shear,
Cracking this new SMMT model can predict the entire torque–twist curve, including the ranges before and after
Shear cracking, as well as the ascending and descending branches. The theoretical predictions from the SMMT
Strain gradient compare very well with the test data on torsion available in the literature.
Stress–strain curve © 2009 Elsevier Ltd. All rights reserved.
Torque
Torsion
Twist
Ultimate state
Reinforced concrete

1. Introduction the SMM, to predict the hysteretic loops of two-dimensional (2-D)


membrane elements subjected to cyclic loading. The CSMM was
1.1. Shear theories for reinforced concrete implemented into a finite element program and was successfully
used to simulate the hysteretic loops of nine framed shear walls
Since the plane truss model concept was advanced by Ritter [1] subjected to vertical load and horizontal cyclic loading [14].
and Mörsch [2] for shear in reinforced concrete (RC), a series
of theoretical models for shear in membrane elements have 1.2. Softening coefficient in shear theories
been developed with increasing sophistication. Rational models
that satisfy equilibrium, compatibility and nonlinear constitutive An RC membrane element subjected to shear is a 2-D problem,
relationships include the Compression Field Theory (CFT) [3,4], because the shear stress can be resolved into a principal tensile
the Rotating-Angle Softened Truss Model (RA-STM) [5–7] and stress and a principal compressive stress in the 45◦ direction. For
the Fixed-Angle Softened Truss Model (FA-STM) [8,9]. The latest such a 2-D membrane element, Robinson and Demorieux [15]
theoretical development for treating RC membrane elements in found in 1968 that the principal compressive stress was reduced,
shear was the Softened Membrane Model (SMM) [10]. The SMM or ‘‘softened’’, by the principal tensile stress in the perpendicular
takes into account the Poisson effect of reinforced concrete, which direction.
is characterized by two Hsu/Zhu ratios [11]. As a result, the SMM Using a biaxial test facility called ‘‘shear rig’’, Vecchio and
can satisfactorily predict the entire monotonic response of the Collins [16] showed in 1981 that the softening coefficient of the
load–deformation curves, including both the ascending and the compressive stress–strain curve of concrete was a function of the
descending branches, as well as both the pre-cracking and post- principal tensile strain ε̄1 , rather than the principal tensile stress.
cracking responses. This is logical because the smeared strain ε̄1 is a direct measure of
In 2005, Mansour and Hsu [12,13] developed the Cyclic the severity of cracking. In 1995, Zhang and Hsu utilized the strain-
Softened Membrane Model (CSMM), which was an extension of controlled feature of the Universal Panel Tester [17] to study the
shear behavior of high-strength concrete RC panels up to 100 MPa.
They found that the softening coefficient was not only a function
∗ Corresponding author. Tel.: +886 492910960x4710; fax: +886 492918679. of the perpendicular tensile strain ε̄1 , but also a function of the
E-mail addresses: chjeng@ncnu.edu.tw (C.-H. Jeng), thsu@uh.edu (T.T.C. Hsu). compressive strength of concrete, fc0 [18].
0141-0296/$ – see front matter © 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engstruct.2009.02.038
C.-H. Jeng, T.T.C. Hsu / Engineering Structures 31 (2009) 1944–1954 1945

Nomenclature ε̄l , ε̄t smeared (average) uniaxial strain in the l-direction


and the t-direction of the steel bars, respectively
A0 area enclosed by the centerline of shear flow; A0 = ε̄n smeared (average) uniaxial yield strain of the steel
Ac − (0.5) pc td + td2 for a rectangular section bars
Al total cross-sectional area of longitudinal steel bars ε̄s smeared (average) uniaxial strain of the steel bars
At cross-sectional area of one transverse steel bar εsf smeared (average) strain of the steel bars that yield
Ac cross-sectional area bounded by the outer perimeter first, taking into account Hsu/Zhu ratios
of the concrete γ21 smear (average) shear strain in the 2–1 coordinate
B variable as defined in the constitutive relationship γlt smear (average) shear strain in the l–t coordinate of
of mild steel embedded in concrete in Fig. 3(c) the steel bars
c̄ distance from the concrete surface to the inner face σ1c , σ2c smeared (average) normal stresses of concrete in
of the transverse hoop bars the 1-direction and the 2-direction, respectively
Ec elastic modulus of concrete σl , σt applied normal stresses in the l-direction and the t-
Es elastic modulus of the steel bars direction of the steel bars, respectively
fc0 cylinder compressive strength of concrete τ21
c
smeared (average) shear stress of concrete in 2–1
fcr , εcr cracking stress and strain of concrete coordinate
fl , ft smeared (average) steel stress in the longitudinal τlt applied shear stresses in the l–t coordinate of the
and transverse directions, respectively steel bars
fs smeared (average) stress of steel bars ρ steel ratio
fy , εy yield stress and strain of bare steel bars ρl , ρt longitudinal and transverse steel ratios respec-
k1 ratio of the average compressive stress to the peak tively; ρl = Al /p0 td and ρt = At /s td
compressive stress in the concrete struts, neglecting ν12 , ν21 Hsu/Zhu ratios used in the SMM: ν12 = 0.2 + 850εsf
the tensile stress of concrete for εsf ≤ εy or ν12 = 1.9 for εsf > εy ; ν21 = 0
k1c ratio of the average compressive stress to the peak (ν12 )Shear same as ν12
compressive stress in the concrete struts, taking into (ν12 )Torsion modified Hsu/Zhu ratio used in the SMMT for
account the tensile stress of concrete torsion
k1t ratio of the average tensile stress to the peak tensile θ angle of twist per unit length
stress in the concrete struts θcr cracking angle of twist per unit length
p0 perimeter of the centerline of shear flow; p0 = pc − θu ultimate angle of twist per unit length
4td for rectangular sections φ curvature of the concrete struts along the 2-
pc perimeter of the outer concrete cross section direction
ph perimeter of the center line of stirrups ϕ curvature of the concrete struts along 1-direction
q shear flow ζ softened coefficient of concrete in compression
Q variable as defined by Eq. (9)
R reduction factor for the Hsu/Zhu ratio ν12 , Eq. (23)
and Fig. 6 In 2006, Wang [19] further improved the softening coefficient
s spacing of transverse hoop bars in the SMM by showing it to be a function of the deviation
s in-plane displacement as shown in Fig. 2(b) angle β . By treating the softening coefficient as a function of all
t in-plane displacement as shown in Fig. 2(b) three variables (ε1 , fc0 and β ), the SMM becomes very powerful,
T torque applicable to membrane elements with any ratios of longitudinal
Tcr cracking torque steel to transverse steel, any orientations of steel reinforcement,
td thickness of shear flow zone and any concrete strength up to 100 MPa.
Tu ultimate torque
[T (α2 )] transformation matrix as defined by Eq. (2) 1.3. Torsion theories for reinforced concrete
[V ] conversion matrix as defined by Eq. (11)
w out-of-plane displacement in the direction normal By extending the 2-D plane truss model of Ritter and Mörsch
to the membrane element as shown in Fig. 2 to a three-dimensional (3-D) space truss model, Rausch [20] in
α rotating angle, angle of principal compressive stress 1929 developed the first theory for torsion of RC members. Rausch
of concrete with respect to longitudinal steel bars utilized Bredt’s [21] tube theory to relate the torsional moment
(l-axis) T to the shear flow q in the tube wall by a simple relationship:
α2 fixed angle, angle of applied principal compressive T = (2A0 )q, where A0 is the lever arm area.
stress (2-axis) with respect to longitudinal steel bars Torsion is a more complicated problem than shear because it
(l-axis) is a 3-D problem involving not only the shear problem of 2-D
β deviation angle (α2 − α); 2β = tan−1 (γ21 / (ε2 − ε1 )) membrane elements in the tube wall, but also the equilibrium
ε0 concrete cylinder strain corresponding to peak and compatibility of the whole 3-D member and the warping of
cylinder strength fc0 tube walls that causes bending in the concrete struts. The effective
ε1 , ε2 smeared (average) biaxial strain in the 1-direction thickness of the tube wall was defined by the shear flow zone [22]
and the 2-direction, respectively in which the concrete strain varies from zero to a maximum at the
ε̄1 , ε̄2 smeared (average) uniaxial strain in the 1-direction edge, thus creating a strong strain gradient.
and the 2-direction, respectively By incorporating the two compatibility equations of a member,
ε̄1s , ε̄2s uniaxial surface strain in the 1-direction and the 2- the Compression Field Theory (CFT) for shear was applied to
direction, respectively predict the nonlinear, post-cracking torsional behavior of the
εl , εt smeared (average) biaxial strain in the l-direction member up to the peak point [3]. This torsion theory did not
and the t-direction of the steel bars, respectively utilize the softened stress–strain curve of concrete, but assumed
that the concrete cover to have spalled before reaching the
1946 C.-H. Jeng, T.T.C. Hsu / Engineering Structures 31 (2009) 1944–1954

maximum strength. In this ‘‘spalling model’’, the conservative


assumption of cover spalling appeared to roughly compensate
for the unconservative assumption of using a constitutive law of
concrete that was not ‘‘softened’’.
In 1985, Hsu and Mo [23] developed a ‘‘softened truss
model (STM)’’ to predict the post-cracking torsional behavior
of reinforced concrete members up to the peak point. The
softened stress–strain curve of concrete, in which the softening
coefficient was derived from shear without a strain gradient, was
incorporated. Hsu and Mo’s model predicted very well all the
available test results, and was able to explain why Rausch’s model
consistently overestimates the ultimate torque. In essence, the
softening of the concrete increases the effective thickness of the
shear flow zone and decreases the lever arm area. This, in turn,
reduces the torsional resistance of the cross section [22,24]. Since Fig. 1. RC section subjected to torsion: (a) shear flow of cross section; (b) in-plane
this STM utilized the same rotating angle concept as that of the RA- stress state of element A.
STM for shear, it is referred to as the RA-STM for torsion.
In the present paper, the powerful model for shear, SMM, is equation [21] is used to describe the equilibrium of the whole cross
applied to torsion to create a new theory ‘‘Softened Membrane section.
Model for Torsion (SMMT)’’ that is capable of predicting the
entire torque–twist curve, including the pre-cracking and the T = 2A0 q = 2A0 td τlt . (3)
post-cracking responses, as well as the pre-peak and the post-
peak branches. In the SMMT two modifications to the constitutive
relationships are required to take care of the strain gradient caused
2.2. Compatibility equations
by the bending of the concrete struts in the shear flow zone.
The strain gradient is expected to increase the tensile strength of
concrete and to decrease the Hsu/Zhu ratios. The in-plane compatibility of the shear element A (Fig. 1(b))
satisfies three equations [10], which can be expressed in matrix
form as follows:
1.4. Research significance
T  T
εl εt γlt /2 = T α2 ε2 ε1 γ21 /2 .
  
(4)
The shear model, the SMM, is extended to create a new
SMMT method for treating RC members subjected to torsion. This The fourth and the fifth compatibility equations are given as
extension required modifying the constitutive relationships of follows [24]:
concrete that relates to the strain gradient of concrete struts due
p0
to the bending. The new SMMT method is capable of predicting the θ= γlt (5)
torsional behavior of RC members before cracking, as well as after 2A0
the peak point. This represents a significant advancement beyond φ = θ sin 2α2 . (6)
previous torsion theories which were incapable of predicting the
torsional behavior in these two ranges. It should be mentioned that Eq. (6) was originally expressed in
terms of the rotating angle α in the rotating-angle theories [24].
2. The SMMT Since this equation is derived strictly from geometry, the equation
also holds when the rotating angle is replaced by the fixed angle α2 .
2.1. Equilibrium equations In Eq. (6), the curvature φ results in a nonuniform strain
distribution, or strain gradient, in the concrete struts. The strain
When an RC prismatic member is subjected to an external gradient is illustrated in Fig. 2(a) by the triangular strain diagrams
torque T as shown in Fig. 1(a), the external torque is resisted by in the 1- and 2-directions. As in the rotating angle theories, the
an internal torque formed by the circulatory shear flow q. This strain distribution is assumed to be linear, and the depth of the
shear flow q occupies a shear flow zone [22,24], with an effective compression zone of the concrete struts is assumed to be the
thickness td . The element A in the shear flow zone is subjected to a thickness of the shear flow zone td . Thus, we have
shear stress (Fig. 1(b)). The in-plane equilibrium of this element,
ε̄2s
according to the SMM [10], satisfies three algebraic equations, td = . (7)
which can be expressed in matrix form as follows: φ

σl σt τlt
T
= T α2
 c T
σ2 σ1c τ21
  c The smeared (or average) uniaxial strain ε̄2 is assumed to be
T related to the maximum uniaxial strain ε̄2s at the surface by ε̄2 =
+ ρl fl ρt ft 0 . ε̄2s /2 [24].


(1) Substitution and manipulation of Eqs. (5)–(7) and the equations


for computing p0 and A0 of rectangular sections listed in Nomen-
The transformation matrix T α2
 
is given by clature yields the following explicit expression for calculating the
effective thickness td of the shear flow zone:
c2 s2
 
2sc
[T (α2 )] =  s2 c2 −2sc  (2) 1
td =
−sc sc c 2 − s2 2 (Q + 4)
where c = cos (α2 ), s = sin (α2 ). In pure torsion, the element A is
 s 
   2
Q Q
subjected to pure shear, with the normal stresses σl = σt = 0 × pc 1 + − 1+ p2c − 4Q (Q + 4) Ac  (8)
and α2 = 45◦ . By employing the thin tube concept, a fourth 2 2
C.-H. Jeng, T.T.C. Hsu / Engineering Structures 31 (2009) 1944–1954 1947

Fig. 2. Bending of concrete struts: (a) in-plane biaxial stress state with out-of-plane
bending; (b) hyperbolic paraboloid surface for the out-of-plane bending; (c) 3-D
Fig. 3. Constitutive relationships of materials: (a) concrete in compression;
view of a hyperbolic paraboloid surface.
(b) concrete in tension; (c) mild steel embedded in concrete.

where The Hsu/Zhu ratios ν12 and ν21 proposed by Zhu and Hsu [11]
2ε̄2s 4ε̄2 for RC membrane elements in uniform shear are as listed in
Q = = . (9) Nomenclature.
γlt sin 2α2 γlt sin 2α2
Eqs. (8) and (9) are derived to simplify the solution algorithm by 2.4. Constitutive relationships of concrete in compression
avoiding the nested loop for the iterative calculation of td .
As in the SMM, a compressive stress–strain relationship of
2.3. Relationships between biaxial strains and uniaxial strains softened concrete [19] is employed and is illustrated in Fig. 3(a).
The softening coefficient ζ in Fig. 3(a), which represents the
The relationships between biaxial strains and uniaxial strains reduction of compressive strength of concrete, is expressed as a
were given by Hsu and Zhu [10] in algebraic form. These function of three variables: the tensile strain ε̄1 in the principal
relationships in matrix expression are direction, the concrete strength fc0 and the deviation angle β .
As in the RA-STM for torsion [24], an average stress factor, k1 ,
{ε̄2 , ε̄1 , γ21 /2}T = [V ] {ε2 , ε1 , γ21 /2}T (10) is used to compute the average compressive stress of the concrete
struts.
where [V ] is the conversion matrix:
σ2c = k1c ζ fc0 . (13)
1 ν21
 
0 The average stress factor k1c is obtained by integration of the
 1 − ν12 ν21 1 − ν12 ν21 
ν12 1 stress–strain equations in Fig. 3(a), as follows:
[V ] =  (11)
 
0


1 − ν12 ν21 1 − ν12 ν21 ε̄2s (ε̄2s )2 ε̄2s
0 0 1 k1c = − for ≤1
ζ ε0 3 (ζ ε0 )2 ζ ε0
and
ζ ε0 (ε̄2s − ζ ε0 )3 ε̄2s
= 1− − for > 1. (14)
{ε̄` , ε̄t , γ`t /2} = [T (α2 )] {ε̄2 , ε̄1 , γ21 /2} .
T T
(12) 3ε̄2s 3ε̄2s (4ε0 − ζ ε0 )2 ζ ε0
1948 C.-H. Jeng, T.T.C. Hsu / Engineering Structures 31 (2009) 1944–1954

2.5. Constitutive relationships of concrete in tension

In the various theories for torsion [3,23,25–28], the tensile


stress of concrete is neglected. None of these models can predict
the cracking torque, Tcr , or the pre-cracking and the pre-peak
stiffness of the torque–twist response. These handicaps are
overcome by the new SMMT, which has successfully incorporated
this effect.
The element A of Fig. 1 is subjected to an in-plane biaxial
stress state with out-of-plane bending. The concrete struts in
such an element is schematically illustrated in Fig. 2(a) with the
assumption of linear distribution of normal strain. The out-of-
plane bending creates a hyperbolic paraboloid surface as shown
in Fig. 2(b) [24]. In the 2-direction, the curvature φ of the curve OD
is the second derivative of w with respect to the length in the 2-
direction, and has been found to be related to the angle of twist θ
and the fixed angle α2 by Eq. (6). Similarly, in the 1-direction, we
can derive the curvature ϕ of the curve AC as the derivative of w
with respect to the length in the 1-direction. Fig. 4. Effect of constitutive models of concrete in tension.

d2 w Also, it can be seen from Fig. 4 that increasing the values


ϕ= = −θ sin 2α2 = −φ. (15) of fcr and Ec affects primarily the SMMT-predicted cracking
dt 2
torque and twist, but increase only slightly the post-cracking
This curvature ϕ produces the distribution of tensile strains and
torque. Therefore, by conducting a systematic and comprehensive
stresses of concrete in the shear flow zone as shown in Fig. 2(a).
parametric study primarily with all the experimental cracking
Assuming that the curvature ϕ results in a linear tensile strain
torque and twist in Table 1 [30], a new tensile constitutive
distribution within an effective thickness td , and assuming the relationship is proposed for torsion (solid curve in Fig. 3(b)). In
same depth td for the tension zone as that for the compression Fig. 3(b), both Ec and εcr of the proposed model for torsion are
zone, the maximum tensile strain ε̄1s can be calculated as 1.45 times those of the Belarbi and Hsu model for shear. The pre-
ε̄1s = ϕ td . (16) cracking stiffness is 45% higher, and the concrete cracking strength
fcr is 110% stronger.
The average uniaxial tensile strain ε̄1 is related to the maximum Substituting the equations of tensile stress of Fig. 3(b) into
uniaxial tensile strain ε̄1s by ε̄1 = ε̄1s /2. Eq. (18) yields the following expression for the average stress factor
The constitutive relationship for concrete in tension can now be k1t for tension:
established. Similarly to the compressive relationship, an average
ε̄1s ε̄1s
stress factor is devised to deal with the tensile constitutive k1t = for ≤1
relationship of concrete. To distinguish the compressive from 2εcr εcr
the tensile constitutive relationship, however, a subscript ‘c’ is εcr (εcr )0.4  ε̄1s
(ε̄1s )0.6 − (εcr )0.6 > 1.

appended to the k1 factor in Eqs. (13) and (14) for concrete in = + for (19)
2ε̄1s (0.6) ε̄1s εcr
compression and a subscript ‘t’ is appended to the k1 factor for
concrete in tension as follows:
2.6. Constitutive relationships of mild steel bars embedded in concrete
σ1c = k1t fcr (17)
Now that the smeared tensile stress of concrete is incorporated
where k1t is defined as the ratio of the average tensile stress to the
into the SMMT, the stress–strain relationship of bare steel bars
peak tensile stress fcr . Similar to k1c , the factor k1t can be obtained
needs to be replaced by the smeared stress–strain relationship of
by integration:
steel bars embedded in concrete [5,6] as listed in Fig. 3(c).
Z ε̄1s
1
k1t = σ1 (ε̄1 ) dε̄1 , (18) 2.7. Constitutive relationships of concrete in shear
ε̄1s fcr 0

where σ1 (ε̄1 ) is a uniaxial tensile stress–strain relationship of As in the SMM, a rational shear modulus [31] is incorporated in
concrete. the SMMT to relate the concrete shear stress to the shear strain, as
In this study, the tensile constitutive relationship of concrete follows:
proposed by Belarbi and Hsu [6] (dashed curve in Fig. 3(b)) for shear σ1c − σ2c
elements was first tried. It was found, however, that the cracking τ21
c
= γ21 . (20)
2 (ε1 − ε2 )
stress (fcr ) and the pre-cracking stiffness (Ec ) of the Belarbi and
Hsu model were too small when applied to torsion specimens,
as illustrated in Fig. 4. Fig. 4 shows that the original fcr and Ec of 2.8. Solution algorithm
the Belarbi and Hsu model yield an analytical cracking torque and
The solution algorithm of the proposed SMMT is an extension
twist that are about half of the experimental values for specimen
of that of the SMM. As in the SMM, the first two basic equilibrium
M3 [29].
equations in Eq. (1) are summed and subtracted to obtain the
The reason for the lower fcr and Ec can be attributed to the
following two equations, which are used as the convergence
strain gradient effect. The Belarbi and Hsu model was based on
criteria for the solution procedure [10]:
tests of shear panels (uniform strain), so its fcr was analogous to
ρl fl + ρt ft = (σl + σt ) − σ2c + σ1c

the direct tensile strength of plain concrete. To reflect the strain (21)
gradient effect, the fcr for torsion should be much higher because
ρl fl − ρt ft = (σl − σt ) − σ − σ c c
cos 2α2 − 2τ c
sin 2α2 . (22)

it is analogous to the modulus of rupture of very shallow plain 2 1 21
concrete beams. The solution algorithm of the SMMT with the modification of the
C.-H. Jeng, T.T.C. Hsu / Engineering Structures 31 (2009) 1944–1954 1949

Table 1
Comparison of the SMMT with tests at cracking.
θcr ,test Tcr ,test
Ref. Item no. Specimen fc0 (MPa) θcr ,test (rad/m × 10−3 ) θcr ,calc (rad/m × 10−3 ) θcr ,calc
Tcr ,test (kN m) Tcr ,calc (kN m) Tcr ,calc

1 N-06-06 35.5 1.4 1.570 0.892 43.2 55.02 0.785


2 N-06-12 35.5 2.0 1.604 1.247 51.8 55.49 0.933
3 N-12-12 35.5 1.6 1.595 1.003 49.3 56.10 0.879
4 N-12-16 35.5 2.1 1.629 1.289 57.1 56.48 1.011
5 N-20-20 35.5 2.2 1.653 1.331 55.0 57.38 0.959
6 N-07-10 33.5 1.9 1.601 1.187 41.6 55.39 0.751
7 N-07-16 33.5 1.9 1.619 1.174 40.0 55.99 0.714
[35] 8 N-14-10 33.5 2.1 1.618 1.298 41.8 55.98 0.747
9 H-06-12 78.5 1.5 1.154 1.300 75.0 78.18 0.959
10 H-12-12 78.5 1.3 1.312 0.991 77.1 77.46 0.995
11 H-12-16 78.5 1.6 1.043 1.533 79.3 78.32 1.013
12 H-20-20 78.5 1.6 1.353 1.183 76.0 79.93 0.951
13 H-07-10 68.4 1.6 1.178 1.358 70.5 73.80 0.955
14 H-07-16 68.4 1.5 1.370 1.095 65.3 73.37 0.890
15 H-14-10 68.4 1.5 1.067 1.406 61.8 73.21 0.844

16 A2 38.2 – 2.520 – 11.29 13.13 0.860


17 A3 39.3 – 2.523 – 12.19 13.45 0.906
[34] 18 A4 39.1 – 2.564 – 12.53 13.64 0.919
19 B3 38.6 – 2.821 – 8.805 11.93 0.738
20 B4 38.4 – 2.862 – 10.16 12.11 0.839

21 B3 28.0 2.164 2.341 0.925 20.09 20.61 0.975


22 B4 30.5 2.075 2.312 0.897 21.90 21.69 1.010
23 B5 29.0 2.405 2.371 1.014 22.58 21.44 1.053
24 B6 28.8 2.845 2.403 1.184 24.95 21.62 1.154
25 B7 23.2 1.890 2.450 0.771 20.21 18.70 1.080
26 B8 26.7 2.199 2.390 0.920 21.79 20.28 1.075
27 B9 28.8 1.855 2.306 0.805 19.64 20.67 0.950
28 B10 26.5 1.546 2.404 0.643 17.61 20.25 0.870
29 M1 29.8 1.855 2.267 0.818 19.19 20.88 0.919
30 M2 30.5 1.931 2.271 0.850 20.55 21.30 0.965
31 M3 26.7 2.110 2.383 0.885 20.66 20.25 1.020
32 M4 26.5 2.343 2.415 0.970 20.66 20.39 1.013
33 M5 28.0 2.116 2.404 0.880 21.67 21.14 1.025
34 M6 29.3 2.549 2.391 1.066 22.69 21.81 1.040
35 I3 44.7 1.855 2.058 0.901 25.51 25.30 1.008
36 I4 44.9 2.151 2.080 1.034 28.00 25.73 1.088
37 I5 45.0 2.659 2.107 1.262 28.11 26.17 1.074
38 I6 45.7 2.226 2.125 1.048 27.54 26.78 1.029
39 J4 16.7 1.917 2.727 0.703 17.95 16.15 1.112
[29] 40 J3 16.9 1.677 2.682 0.625 16.93 16.12 1.050
41 J2 14.5 2.164 2.778 0.779 17.05 14.87 1.147
42 J1 14.3 1.491 2.736 0.545 14.00 14.72 0.951
43 K2 30.6 3.099 3.141 0.987 12.19 12.15 1.003
44 K3 29.0 3.367 3.262 1.032 12.42 12.02 1.033
45 K4 28.6 3.594 3.366 1.068 13.10 12.16 1.077
46 N1 29.5 2.233 3.429 0.651 7.586 6.544 1.159
47 N1a 28.7 1.890 3.456 0.547 7.022 6.461 1.087
48 N2 30.4 3.793 3.444 1.101 7.439 6.727 1.106
49 N2a 28.4 3.669 3.573 1.027 7.496 6.515 1.150
50 N3 27.3 3.717 3.594 1.034 7.406 6.365 1.163
51 N4 27.3 3.697 3.648 1.013 7.597 6.453 1.177
52 C4 27.1 2.350 2.838 0.828 11.85 11.56 1.025
53 C5 27.2 2.900 2.876 1.008 14.00 11.74 1.192
54 C6 27.6 2.996 2.906 1.031 13.89 12.00 1.157
55 G3 26.8 1.677 2.126 0.788 27.09 29.23 0.927
56 G4 28.2 2.041 2.154 0.947 28.67 30.19 0.950
57 G5 26.9 1.986 2.206 0.900 29.46 29.79 0.989
58 G7 30.9 2.350 2.047 1.148 33.64 31.24 1.077
59 G8 28.3 2.151 2.158 0.997 33.64 30.32 1.110
(continued on next page)

Hsu/Zhu ratio (Eq. (24)) is illustrated in Fig. 5. Note that one of the concrete. The element A of a torsional beam (Fig. 1) is subjected
differences between the algorithms of the SMMT and the SMM is to out-of-plane bending, resulting in a strain gradient in the
the variable td (and so are the reinforcement ratios ρl and ρt ) in shear flow zone (Fig. 2). Compared to the uniform strain in
the SMMT. The SMM algorithm is simpler because the membrane membrane elements subjected to shear, the strain gradient in the
thickness and the reinforcement ratios in the SMM are constant. concrete struts of torsional beams should mitigate the concrete
cracking, and thus lower the Hsu/Zhu ratio ν12 for torsion. Denoting
3. Modification of Hsu/Zhu ratio for torsion (ν12 )Torsion to be the Hsu/Zhu ratio for torsion, we can assume that

The Hsu/Zhu ratios proposed by Zhu and Hsu [11] were derived
(ν12 )Torsion = R · (ν12 )Shear (23)
from tests of RC membrane elements under uniform shear. The where (ν12 )Shear is the Hsu/Zhu ratio proposed by Zhu and Hsu [11]
large value (1.9) of ν12 is mainly due to extensive cracking of for membrane elements under uniform shear and R is a reduction
1950 C.-H. Jeng, T.T.C. Hsu / Engineering Structures 31 (2009) 1944–1954

Table 1 (continued)
θcr ,test Tcr ,test
Ref. Item no. Specimen fc0 (MPa) θcr ,test (rad/m × 10−3 ) θcr ,calc (rad/m × 10−3 ) θcr ,calc
Tcr ,test (kN m) Tcr ,calc (kN m) Tcr ,calc

60 VM3 40.0 – 1.554 – – 61.88 –


61 VM2 36.1 – 2.125 – – 24.96 –
62 VM1 39.1 – 3.128 – – 7.668 –
63 VU1 19.5 – 2.544 – – 18.82 –
64 VU2 19.5 – 2.493 – – 18.88 –
65 VU3 18.5 – 2.597 – – 18.43 –
66 VU4 18.5 – 2.588 – – 18.40 –
67 VS2,VQ2 19.0 – 2.498 – – 18.55 –
68 VS3 19.0 – 2.510 – – 18.64 –
[33]
69 VS4,VQ5 19.0 – 2.521 – – 18.71 –
70 VS9 17.6 – 2.557 – – 17.85 –
71 VS10,VB1 19.0 – 2.519 – – 18.70 –
72 VB2 26.4 – 2.297 – – 21.79 –
73 VB3 39.1 – 2.068 – – 25.88 –
74 VB4 49.8 – 1.949 – – 28.61 –
75 VQ1 19.0 – 2.314 – – 19.60 –
76 VQ3 17.6 – 2.994 – – 15.97 –
77 VQ9 19.5 – 3.643 – – 14.32 –

Average 0.9986 Average 0.9939


Standard deviation 0.21781 Standard deviation 0.11652
1. The experimental data are taken from the references: (Fang and Siau 2004) [35], (McMullen and Rangan 1978) [34], (Hsu 1968) [29], and (Leonhardt and Schelling
1974) [33].
2. ‘–’ indicates that the test data are not available in the respective reference.

Fig. 6. Effect of the Hsu/Zhu ratio ν12 .

factor. Fig. 6 illustrates the effect of the variation of the Hsu/Zhu


ratio on the SMMT-predicted torque–twist curves of specimen
M3 [29] by varying the R value from 0.0, 0.5, 0.8 to 1.0. The other
Hsu/Zhu ratio ν21 remains zero, as shown in Nomenclature. It
can be seen that as the factor R decreases, the analytical post-
cracking torque becomes larger and the analytical post-cracking
twist smaller. The pre-cracking torque–twist curves are practically
not influenced by the factor R as expected, and are not shown
in Fig. 6.
A systematic and comprehensive parametric study of the
reduction factor R with all the experimental ultimate torque and
twist in Table 2 were conducted [32], from which an optimum
value of R = 0.8 is determined. Hence, the Hsu/Zhu ratio for the
SMMT is proposed as
Fig. 5. Solution algorithm of the SMMT.
(ν12 )Torsion = 0.8 (ν12 )Shear . (24)
C.-H. Jeng, T.T.C. Hsu / Engineering Structures 31 (2009) 1944–1954 1951

Table 2
Comparison of the SMMT with tests at ultimate.
θu,test Tu,test
Ref. Item no. Specimen fc0 (MPa) θu,test (rad/m × 10−3 ) θu,calc (rad/m × 10−3 ) θu,calc
Tu,test (kN m) Tu,calc (kN m) Tu,calc

1 N-06-06 35.5 56.6 48.75 1.161 79.7 84.07 0.948


2 N-06-12 35.5 40.4 48.98 0.823 95.2 98.58 0.966
3 N-12-12 35.5 47.8 59.61 0.795 116.8 135.27 0.863
4 N-12-16 35.5 40.2 55.25 0.728 138.0 137.1 1.006
5 N-20-20 35.5 30.5 50.09 0.862 158.0 152.2 1.038
6 N-07-10 33.5 50.8 50.69 1.073 111.7 105.5 1.059
7 N-07-16 33.5 55.2 42.71 1.246 117.3 110.3 1.063
[35] 8 N-14-10 33.5 58.8 49.91 1.178 125.0 126.5 0.988
9 H-06-12 78.5 25.4 54.80 0.463 115.1 114.9 1.001
10 H-12-12 78.5 37.8 41.60 0.909 155.3 156.0 0.995
11 H-12-16 78.5 39.3 45.47 0.864 196.0 175.5 1.117
12 H-20-20 78.5 46.5 55.11 0.844 239.0 220.3 1.085
13 H-07-10 68.4 33.2 58.31 0.552 126.7 120.1 1.055
14 H-07-16 68.4 37.5 44.83 0.836 144.5 126.4 1.143
15 H-14-10 68.4 38.1 45.57 0.836 135.2 143.2 0.944

16 A2 38.2 69.40 51.91 1.337 22.58 21.96 1.028


17 A3 39.3 56.35 59.09 0.954 27.77 30.20 0.920
[34] 18 A4 39.1 69.40 86.03 0.807 34.43 36.28 0.949
19 B3 38.6 82.46 71.91 1.147 25.29 27.74 0.912
20 B4 38.4 79.02 93.14 0.848 31.72 32.45 0.977

21 B3 28.0 51.54 60.60 0.850 37.48 45.01 0.833


22 B4 30.5 57.03 68.67 0.831 47.30 53.39 0.886
23 B5 29.0 61.84 63.03 0.981 56.11 54.41 1.031
24 B6 28.8 65.28 59.23 1.102 61.64 56.20 1.097
25 B7 23.2 48.10 61.82 0.778 26.87 27.47 0.978
26 B8 26.7 – 43.83 – 32.51 30.68 1.060
27 B9 28.8 50.85 59.04 0.861 29.80 30.91 0.964
28 B10 26.5 54.97 41.66 1.320 34.32 31.99 1.073
29 M1 29.8 43.98 46.50 0.946 30.37 30.23 1.005
30 M2 30.5 51.54 62.77 0.821 40.53 39.43 1.028
31 M3 26.7 53.60 54.20 0.989 43.80 43.49 1.007
32 M4 26.5 57.72 63.66 0.907 49.56 47.24 1.049
33 M5 28.0 62.53 62.24 1.005 55.65 51.61 1.078
34 M6 29.3 – 61.43 – 60.06 55.29 1.086
35 I3 44.7 47.41 41.14 1.153 45.61 50.69 0.900
36 I4 44.9 – 48.74 – 58.02 62.74 0.925
37 I5 45.0 53.60 69.33 0.773 70.67 73.62 0.960
38 I6 45.7 54.97 65.58 0.838 76.65 77.58 0.988
39 J4 16.7 58.41 47.68 1.225 40.64 34.26 1.186
[29] 40 J3 16.9 54.97 61.84 0.889 35.22 32.16 1.095
41 J2 14.5 51.54 67.05 0.769 29.13 26.84 1.085
42 J1 14.3 46.04 57.92 0.795 21.45 22.74 0.943
43 K2 30.6 – 66.64 – 23.71 24.96 0.950
44 K3 29.0 75.59 93.38 0.809 28.45 29.64 0.960
45 K4 28.6 85.89 83.69 1.026 35.00 31.54 1.109
46 N1 29.5 75.59 54.53 1.386 9.088 9.053 1.004
47 N1a 28.7 75.59 53.37 1.416 8.986 9.005 0.998
48 N2 30.4 85.89 78.93 1.088 14.45 14.26 1.013
49 N2a 28.4 84.52 94.44 0.895 13.21 14.09 0.937
50 N3 27.3 81.77 79.35 1.030 12.19 12.49 0.976
51 N4 27.3 89.33 103.0 0.867 15.69 15.08 1.041
52 C4 27.1 77.65 76.39 1.016 25.29 27.98 0.904
53 C5 27.2 84.52 54.03 1.564 29.69 29.71 0.999
54 C6 27.6 91.39 53.29 1.715 34.21 32.12 1.065
55 G3 26.8 50.85 43.40 1.172 49.56 54.05 0.917
56 G4 28.2 53.60 52.18 1.027 64.80 66.91 0.968
57 G5 26.9 54.28 60.11 0.903 71.91 70.85 1.015
58 G7 30.9 44.66 35.43 1.261 52.61 55.87 0.942
59 G8 28.3 49.47 63.66 0.777 73.38 70.49 1.041
(continued on next page)

4. Comparison of the SMMT with tests A total of 109 torsional specimens were initially collected from
literature. Some of them were excluded for comparison based
on four criteria, namely, (1) insufficient reinforcement such that
The computer program based on the proposed SMMT solution
Tu ≤ Tcr ; (2) excessive stirrup spacing exceeding ph /8 or 12
algorithm (Fig. 5) was used to predict the response of a test inches, where ph is the perimeter of the center line of stirrups;
specimen throughout its entire loading history. Fig. 7 compares (3) excessive or insufficient concrete cover c̄, where c̄ /td is either
the analytical torque–twist curve with the experimental curve of less than 0.25 or greater than 0.75; (4) hollow sections are exclu-
a typical test specimen M3 [29]. It can be seen that the SMMT ded, because their cracking torques and cracking angles of twist
predicts very well the entire torque–twist curve of specimen M3. are somewhat lower than those of solid sections. As a result, the
The physical meanings of four points on the curve are also given. remaining 80 specimens were eligible for comparison [29,33–35].
1952 C.-H. Jeng, T.T.C. Hsu / Engineering Structures 31 (2009) 1944–1954

Table 2 (continued)
θu,test Tu,test
Ref. Item no. Specimen fc0 (MPa) θu,test (rad/m × 10−3 ) θu,calc (rad/m × 10−3 ) θu,calc
Tu,test (kN m) Tu,calc (kN m) Tu,calc

60 VM3 40.0 41.92 38.21 1.097 100.8 98.71 1.021


61 VM2 36.1 63.90 43.62 1.465 39.17 38.37 1.021
62 VM1 39.1 98.26 115.4 0.852 13.89 13.07 1.062
63 VU1 19.5 51.54 73.60 0.700 23.93 24.08 0.994
64 VU2 19.5 71.46 65.08 1.098 30.37 31.21 0.973
65 VU3 18.5 72.15 61.28 1.177 31.04 30.17 1.029
66 VU4 18.5 56.35 62.44 0.902 25.96 24.45 1.062
67 VS2,VQ2 19.0 46.73 44.76 1.044 19.53 22.71 0.860
68 VS3 19.0 76.96 65.30 1.178 28.56 31.14 0.917
[33]
69 VS4,VQ5 19.0 70.78 74.08 0.955 34.32 34.49 0.995
70 VS9 17.6 63.22 84.38 0.749 21.56 26.58 0.811
71 VS10,VB1 19.0 68.71 75.64 0.908 33.30 34.02 0.979
72 VB2 26.4 73.52 85.33 0.862 42.11 42.49 0.991
73 VB3 39.1 76.96 63.53 1.211 46.40 50.25 0.923
74 VB4 49.8 74.90 55.16 1.358 48.54 51.40 0.944
75 VQ1 19.0 57.03 45.19 1.262 21.11 22.09 0.956
76 VQ3 17.6 52.22 57.56 0.907 19.98 22.15 0.902
77 VQ9 19.5 81.08 72.70 1.115 21.90 22.26 0.984

Average 0.9985 Average 0.9945


Standard deviation 0.22905 Standard deviation 0.07148
Same as those of Table 1.

5. Comparison of the SMMT with previous torsion theory

62 specimens in Tables 1 and 2 (Item No. 16–77) have been


analyzed by the RA-STM for torsion in [23]. The average and
standard deviation of the ultimate state ratios for the 62 specimens
are shown in Table 3. The SMMT-predicted results compared with
tests for the 62 specimens are also given in Table 3. It can be seen
that the SMMT predictions achieve better agreement with tests
than the RA-STM predictions in terms of both ultimate torque and
ultimate angle of twist.
Fig. 7 shows the experimental torque–twist curve of specimen
M3 and the two analytical torque–twist curves predicted by the
SMMT and the RA-STM for torsion. It can be seen that the main
superiority of the SMMT over the RA-STM is the ability of the SMMT
to accurately predict the pre-cracked branch and the post-cracked
ascending branch of the torque–twist curves. Table 3 also lists the
average and standard deviation of the cracking state ratios for the
62 specimens predicted by the SMMT. In contrast, the average
and standard deviation of the cracking state ratios could not be
Fig. 7. Comparison of the SMMT analytical torque–twist curve with experiment; a predicted by the RA-STM [23].
typical instance. The SMMT provides three insights into the constitutive
relationships of concrete and steel bars for future study of
members subjected to combined forces, such as torsion–shear,
However, since there are three identical pairs in the 80 specimens,
torsion–bending and torsion–axial forces [37]. First, the softening
the total count of specimens taken for comparison is 77.
coefficient in the compressive stress–strain curve of concrete
Tables 1 and 2 compare the analytical values with the 80
remains the same when the SMM for shear was extended to the
available test data for both ultimate and cracking states. The SMMT for torsion. In other words, the softening coefficient is a
averages of the ultimate state ratios (Tu,test /Tu,calc , θu,test /θu,calc ) are very fundamental and crucial property of concrete, and should be
(0.9945, 0.9985), and the standard deviations of the two ratios included in all the shear and torsion theories. Second, the smeared
are (7.148%, 22.905%) (Table 2). The averages of the cracking state stress–strain curve of mild steel bars embedded in concrete
ratios (Tcr ,test /Tcr ,calc , θcr ,test /θcr ,calc ) are (0.9939, 0.9986), and the remains the same when applied to shear as well as torsion. Third,
standard deviations are (11.652%, 21.781%) (Table 1). Fig. 8(a) and only two properties of concrete—the tensile stress–strain curve
(b) show that the calculated torque values are very close to the and the Hsu/Zhu ratios—need to be changed when applied to the
experimental torques for both the ultimate and the cracking states. wall elements. For the case of torsion, where the wall elements
Fig. 8(c) and (d) show that the calculated angles of twist agree in the shear flow zone are warped into a parabolic paraboloid
reasonably well with the test data, considering the difficulties in shape, these two properties could be quantified through a physical
measuring small deformations. Also, it should be noted that the understanding of the strain gradient effect.
fc0 of the specimens ranges between 14.3 MPa and 78.5 MPa. In
short, the SMMT and the incorporated constitutive relationships 6. Conclusions
can be applied to high strength concrete as well as normal strength
concrete. A figure showing the comparison between the analytical 1. A new model, the SMMT, is proposed which is capable of
and experimental torque–twist curves for all of the 80 specimens predicting the entire torque–twist response, including the
can be accessed on the website [36] or from [30]. ultimate torque and twist and the cracking torque and twist.
C.-H. Jeng, T.T.C. Hsu / Engineering Structures 31 (2009) 1944–1954 1953

Table 3
Comparison of the SMMT with previous torsion theory.
Tu,test θu,test Tcr ,test θcr ,test
Tu,calc
(No. of data = 62) θu,calc
(No. of data = 58a ) Tcr ,calc
(No. of data = 44a ) θcr ,calc
(No. of data = 39a )
SMMT RA-STM SMMT RA-STM SMMT RA-STM SMMT RA-STM
b
Average 0.989 1.059 1.023 1.126 1.029 N/A 0.914 N/Ab
Standard deviation 7.06% 7.81% 21.26% 21.36% 10.07% N/Ab 16.97% N/Ab
a
The total number of specimens taken for comparison is 62. However, the experimental data of certain specimens are not available in the respective references, so the
number of available test data is less than 62.
b
These data are not available from [23] because the RA-STM for torsion was unable to predict the pre-cracking branch of torque–twist curves.

for concrete proposed by Belarbi and Hsu [6], the stiffness and
the strain at peak stress should each be increased by 45%. The
Hsu/Zhu ratio ν12 proposed by Zhu and Hsu [11] needs to be
reduced by 20%.
3. The two modifications to the properties of concrete are
necessary because the out-of-plane bending of concrete struts
in the shear flow zone of a torsional member causes a large
strain gradient.

Acknowledgments

This research was funded by National Science Council, Taiwan,


through Grant NSC94-2211-E-260-007. The authors wish to thank
Professor I.-K. Fang and Dr. J.-K. Shiau at National Cheng Kung
University, Taiwan, for providing valuable experimental data. The
authors also gratefully acknowledge the assistance by Mr. J.-R. Shih
and Mr. C.-C. Hsu, former graduate students at National Chi Nan
University, Taiwan.

References

[1] Ritter W. Die bauweise hennebique. Schweiz Bauzeitung 1899;33(7):59–61.


[2] Mörsch E. Der Eisenbetonbau, seine Anwendung und Theorie. 1st ed. Wayss
and Fretag, A. G., Im Selbstverlag der Firma, Neustadt a d. Haardt; 1902. p. 118.
[3] Collins MP, Mitchell D. Shear and torsion design of prestressed and non-
prestressed concrete beams. PCI J 1980;25(5):32–100.
[4] Vecchio FJ, Collins MP. Modified compression-field theory for reinforced
concrete element subjected to shear. ACI Struct J 1986;83(2):219–31.
[5] Pang XB, Hsu TTC. Behavior of reinforced concrete membrane elements in
shear. ACI Struct J 1995;92(6):665–79.
[6] Belarbi A, Hsu TTC. Constitutive laws of concrete in tension and reinforcing
bars stiffened by concrete. ACI Struct J 1994;91(4):465–74.
[7] Belarbi A, Hsu TTC. Constitutive laws of softened concrete in biaxial
tension–compression. ACI Struct J 1995;92(5):562–73.
[8] Pang XB, Hsu TTC. Fixed-angle softened-truss model for reinforced concrete.
ACI Struct J 1996;93(2):197–207.
[9] Hsu TTC, Zhang LX. Nonlinear analysis of membrane elements by fixed-angle
softened-truss model. ACI Struct J 1997;94(5):483–92.
[10] Hsu TTC, Zhu RRH. Softened membrane model for reinforced concrete
elements in shear. ACI Struct J 2002;99(4):460–9.
[11] Zhu RRH, Hsu TTC. Poisson effect in reinforced concrete membrane elements.
ACI Struct J 2002;99(5):631–40.
[12] Mansour M, Hsu TTC. Behavior of reinforced concrete elements under cyclic
shear: Part I—Experiments. J Struct Eng ASCE 2005;131(1):44–53.
[13] Mansour M, Hsu TTC. Behavior of reinforced concrete elements under cyclic
shear: Part II—Theoretical model. J Struct Eng ASCE 2005;131(1):54–65.
[14] Mo YL, Zhong J, Hsu TTC. Seismic simulation of RC wall-type structures. Eng
Struct 2008;30(11):3167–75.
[15] Robinson JR, Demorieux JM. Essai de Traction-Compression sur Modeles
d’Ames de Poutres en Beton Arme. Compte Rendu Partiel I, U.T.I., Institut de
Recherches Appliquees du Beton Arme. Paris; 1968. p. 43.
[16] Vecchio FJ, Collins MP. Stress–strain characteristics of reinforced concrete in
pure shear. IABSE colloquium on advanced mechanics of reinforced concrete
Fig. 8. Summary of ratios of test data to calculated data: (a) ultimate torque Tu ; final report 1981. Delft; p. 32–100.
(b) cracking torque Tcr ; (c) angle of twist at ultimate torque θu ; (d) cracking angle [17] Hsu TTC, Zhang LX, Gomez T. A servo-control system for universal panel tester.
of twist θcr . J Test Eval ASTM 1995;23(6):424–30.
[18] Zhang LX, Hsu TTC. Behavior and analysis of 100 MPa concrete membrane
elements. J Struct Eng ASCE 1998;124(1):24–34.
The analytical results using the SMMT compare very well with [19] Wang J. Constitutive relationships of prestressed concrete membrane
elements. Ph.D. dissertation. Houston (TX, USA): University of Houston; 2006.
the experimental data available in the literature. [20] Rausch, E. Design of reinforced concrete in torsion (Berechnung des
2. Two modifications are required to extend the SMM for 2-D RC Eisenbetons gegen Verdrehung). Ph.D. thesis. Berlin: Technische Hochschule;
1929. p. 53.
membrane elements under shear to the new SMMT for 3-D [21] Bredt R. Kritische BemerKungen zür Drehungselastizitat. Z des Vereines
members under torsion. In the tensile constitutive relationship Deutscher Ingenieure 1896;40(28):785–90; (29):813–817 (in German).
1954 C.-H. Jeng, T.T.C. Hsu / Engineering Structures 31 (2009) 1944–1954

[22] Hsu TTC. Shear flow zone in torsion of reinforced concrete. J Struct Eng ASCE [30] Shih JR. Softened membrane model for torsion. Master thesis. Nantou
1990;116(1):3206–26. (Taiwan): National Chi Nan University; 2007 (in Chinese).
[23] Hsu TTC, Mo YL. Softening of concrete in torsional members – Theory and tests. [31] Zhu RRH, Hsu TTC, Lee JY. Rational shear modulus for smeared crack analysis
ACI J Proc 1985;82(3):290–303. of reinforced concrete. ACI Struct J 2001;98(4):343–50.
[24] Hsu TTC. Unified theory of reinforced concrete. Boca Raton (FL): CRC Press; [32] Jeng CH, Chang CM. Quantifying the reduction factor for Hsu/Zhu ratio for
1993. RC members under torsion. In: Proceedings of structural engineers world
[25] Hsu TTC, Mo YL. Softening of concrete in torsional members – Prestressed congress. 2007.
concrete. ACI J Proc 1985;82(5):603–15. [33] Leonhardt F, Schelling G. Torsionsversuche an Stahlbetonbalken. Heft 239.
[26] Ashour SA, Samman TA, Radain TA. Torsional behavior of reinforced high- Deutscher Ausschuss für Stahlbeton. Berlin; 1974. p. 122.
strength concrete deep beams. ACI Struct J 1999;96(6):1049–58. [34] McMullen AE, Rangan BV. Pure torsion in rectangular section – A re-
[27] Mo YL, Jeng CH, Chang YS. Torsional behavior of prestressed concrete box- examination. ACI Struct J 1978;75(10):511–9.
girder bridges with corrugated steel webs. ACI Struct J 2000;97(6):849–59. [35] Fang IK, Shiau JK. Torsional behavior of normal- and high- strength concrete
[28] Fu CC, Tang Y. Torsional analysis for prestressed concrete multiple cell box. J beams. ACI Struct J 2004;101(3):304–13.
Eng Mech ASCE 2001;127(1):45–51. [36] Website: http://staffweb.ncnu.edu.tw/chjeng/conR.pdf.
[29] Hsu TTC. Torsion of structural concrete—behavior of reinforced concrete [37] Bairan JM, Mari AR. Multiaxial-coupled analysis of RC cross-sections subjected
rectangular members. Torsion Struct Concr SP-18 ACI 1968;261–306. to combined forces. Eng Struct 2007;29(8):1722–38.

Вам также может понравиться