Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
OF REINFORCED C O N C R E T E
By Thomas T. C. Hsu, 1 Fellow, ASCE
INTRODUCTION
The basic formula for calculating the torsional strength of reinforced con-
crete members was developed by Rausch (1929) using the space truss con-
cept. Unfortunately, Rausch's equation may be unconservative by more than
30% for under-reinforced beams (Hsu 1968a, 1968b). The error is traced to
the incorrect determination of the centerline of the circulating shear flow,
resulting in the overestimation of the lever arm area A0. The correct deter-
mination of the centerline of shear flow depends on a logical way to find
the thickness of the shear flow zone, td.
Since the late 1960s, the truss model theory for shear and torsion has
undergone four major developments. First, the introduction of the variable-
angle truss model and the discovery of the bending phenomenon in the di-
agonal concrete struts were made by Lampert and Thurlimann (1968, 1969).
Second, compatibility equation was derived by Collins (1973) to determine
the angle of the diagonal concrete struts. Third, the softening phenomenon
in the concrete struts was discovered by Robinson and Demorieux (1972),
and this behavior was quantified by Vecchio and Collins (1981), using a
softening coefficient. Fourth, combining the equilibrium, compatibility and
softened stress-strain relationships, a softened truss model theory was de-
veloped (Hsu 1988), which was able to analyze the shear and torsional be-
havior of reinforced concrete members throughout the post-cracking loading
history.
Using the softened truss model theory, the thickness of the shear flow
zone td can expeditiously be calculated for the torsional strength of reinforced
concrete members. This method is presented in this study. In addition, a
simple expression for td is proposed for practical design.
'Prof., Dept. of Civ. and Envir. Engrg., Univ. of Houston, Houston, TX 77204-
4791.
Note. Discussion open until April 1, 1991. To extend the closing date one month,
a written request must be filed with the ASCE Manager of Journals. The manuscript
for this paper was submitted for review and possible publication on January 27, 1988.
This paper is part of the Journal of Structural Engineering, Vol. 116, No. 11,
November, 1990. ASCE, ISSN 0733-9445/90/001 l-3206/$l.00 + $.15 per page.
Paper No. 25246.
3206
T = qads (1)
where a = the lever arm of shear flow q measured from the center of twist,
O. The product ads is represented graphically by twice the shaded triangular
area shown. Therefore, ads is twice the area within the centerline of shear
flow, and will be denoted as 2A0. A0 will be called the lever arm area and
is proportional to the square of the level arm a. Substituting 2A0 into Eq. 1
gives
T = q(2A0) (2)
Eq. 2 is Bredt's thin-tube theory (1896), but it should also be applicable to
a thick tube, if the position of the centerline of shear flow can be determined.
In a reinforced concrete member, after cracking, as shown in Fig. 2(a),
an element isolated from the tube defined by the shear flow zone with a
thickness td, Fig. 2(c), can be represented by a truss model in Fig. 2(d).
The element has a vertical length as well as a horizontal length of unity.
The diagonal lines representing the cracks are inclined at an angle a. Taking
equilibrium of forces on the horizontal face and assuming yielding of steel
gives:
A
'fiy .
(3)
q= cot a.
s
o.8Cf'c
?. Shear Flow
N.A.
Stress Strain
Diagram Diagram
(in a Direction of Concrete Struts)
N.A.
Center Line of Shear Flow
H-Ho.85f^ u
N.A. _J*
3207
(b) Shear Flow Zone (c) Shear Flow Zone (d) Truss Model of
Element in d-r Element in l~~t Element in
Coordinate Coordinate Shear Flow Zone
Eq. 4 is the fundamental equation for torsion in the variable-angle truss model.
It reduces to the well-known Rausch's equation (1929) when a is taken as
45. When equilibrium of forces is taken on the vertical face of the truss
model element in Fig. 2(d), q and Tn can be expressed in terms of the lon-
gitudinal steel, i.e., q = (A,f,y/p0) tan a and T = [(A,fly/p0) tan a](2A0).
The analysis of torsion shown is analogous to the analysis of bending in
a prismatic members shown in Fig. 1(b). Taking the moment about the cen-
troid of the tension steel, the external moment, M, is resisted by the internal
moment
M = S(ab)adc = C(jd) (5)
where C = the resultant of stresses o- in the compression zone and jd = the
lever arm of the resultant. Comparing Eq. 5 to Eq. 2, the term of twice the
lever arm area 2A0 in Eq. 2 is equivalent to the resultant lever arm jd in Eq.
5, and the shear flow q is similar to the resultant of compressive stresses C.
After cracking of the flexural member, the truss model concept in bending
is reduced to the so-called internal couple concept. Assuming the yielding
of steel gives
C = A,fv (6)
Substituting C from Eq. 6 into Eq. 5:
M = AJ^jd) (7)
Eq. 7 shows that the bending moment capacity, M, is equal to the longi-
tudinal steel force, Asfsy, times the resultant lever arm, jd. Similarly, in Eq.
3208
3209
& ) ( A , ( , (In.*)
FIG. 3. Comparison of Rausch's Formula and ACI Code Formula with Tests (1
in. = 25.4 mm; 1 in.-kip = 113 N-m)
(8)
s
where a, = 0.66 + 0.33 yi/jci =s 1.5; xx = shorter center-to-center dimension
of a rectangular closed stirrup; yi = longer center-to-center dimension of a
rectangular closed stirrup; Tc = nominal torsional strength contributed by
concrete = 0.%x2y\/f7c where x and y = the shorter and longer sides, re-
spectively, of a rectangular section.
Two modifications of Rausch's equation are made in Eq. 8 based on tests.
First, a smaller lever arm area (a,/2)Ai is specified, where a, varies from 1
to 1.5. Second, a new term Tc is added. This term represents the vertical
intercept of a straight line in the T versus (A,fv/s)(Ai) diagram (Fig. 3).
Although the addition of Tc allows the test curve to be closely approximated
by a straight line in the under-reinforced region, the complexity that is gen-
erated by Tc is certainly undesirable.
The definitions of the lever arm areas, Au A2 or (a,/2)A1( all have a com-
mon weakness. They are not related to the thickness of the shear flow zone
or the applied torque. A logical way to define A0 must start with the deter-
mination of the thickness of shear flow zone.
Compatibility Equations
When a hollow reinforced concrete member is subjected to torsion as shown
in Fig. 2, each cross section will rotate, producing an angle of twist, 9, in
the member and a shear strain 7,, in the shear flow tube. According to Bredt's
theory for circulatory torsion (see Chapter 1 in Hsu 1984) 6 and 7,, are related
by the compatibility condition:
><2<d
ill.
Downloaded from ascelibrary.org by Matthew Foreman on 07/22/13. Copyright ASCE. For personal use only; all rights reserved.
SHEAR FLOW
STRAIN ACTUAL
DIAGRAM STRESS
BLOCK
UNIT WIDTH OF
CONCRETE STRUT
fc = Y (12)
Po = Pc - 4&d (16)
3212
P-P;
= (j 0.003 "d
Material Law
Being subjected to axial stress and bending, the distribution of the com-
pressive stresses in a diagonal concrete strut within the thickness td is shown
by the solid curve in Fig. 5. This stress-strain relationship is based on a
softened stress-strain curve, Fig. 6, proposed by Vecchio and Collins (1981).
Their concrete test panels were reinforced in both the longitudinal and trans-
verse directions and were subjected to pure shear at the edges. Their tests
clearly show that after diagonal cracking the stress and the strain in the con-
crete struts, <jd and ed, are softened by the tensile strain in the perpendicular
direction, er. The softened coefficient t, is a function of the ratio r/ed:
I- J-*- 6
(17)
V *d - r
The softening coefficient , which is less than unity, is the reciprocal of the
coefficient X given by Vecchio and Collins (1981). In their paper, ed in the
denominator is multiplied by a constant (1 u,), where u, is Poisson's ratio
for concrete. The omission of JJL produces negligible difference (Hsu 1984).
Note also that d is negative and er = e, + e, - erf is positive.
Based on the softened stress-strain relationship in Fig. 6, the peak stress
is /c and the average compressive stress, ad, can be defined as
3213
e & varies from 0.0015 to 0.0030, and varies from 0.35 to 0.50. Within
those ranges, the table shows that kx varies in a narrow range from 0.85 to
0.77. Taking an average value of k{ = 0.80, and treating/^ as positive, then
ad becomes
ad = -O.mfc (19)
Substituting the softening coefficient from Eq. 17 into Eq. 14, td can be
expressed in terms of and a:
A02
(20)
p0 sin a cos a
It is interesting to note that td in Eq. 20 is no longer a function of the strains
ed or e r . Physically, this means that td is independent of the loading history.
The substitution of the softening coefficient t, from Eq. 17 into Eq. 14
involves an assumption. Since Eq. 17 is obtained from tests of reinforced
concrete panels subjected to pure shear alone, the strains ed and e r in this
equation represent the uniform in-plane strains of an element without bend-
ing. By contrast, Eq. 14 is derived from an element in a concrete strut sub-
jected to in-plane strains as well as bending, so that ed and e r represent the
average strains in the mid depth of the thickness td. Therefore, Eq. 20 is
obtained by assuming that the softening of a concrete strut subjected to
compression and bending is identical to the softening of a concrete strut
subjected to the average compression strain without bending. This assump-
tion has yet to be proven by tests, but it should provide a very good ap-
proximation.
The thickness of shear flow zone, td, can be solved by Eq. 20 in con-
junction with two equilibrium equations.
Equilibrium Equations
From the truss model of a reinforced concrete element shown in Fig. 2(d)
it can be demonstrated that the stresses in the concrete satisfy Mohr's stress
circle (Hsu 1984). Assuming that the steel will yield at failure (for under-
reinforced members) and the concrete cannot resist tension in the direction
perpendicular to the cracks, i.e., <rr = 0, then the superposition of the con-
crete stresses and steel stresses gives the following three equilibrium equa-
tions:
3214
t, = (24)
0.80/^
Substituting ad = 0.S0iC,f'c from Eq. 19 into Eq. 21 and utilizing Eq.
24 give:
cos 2 a = (25)
Solution Method
The compatibility Eq. 20 and the two equilibrium Eqs. 24 and 25 provide
three equations involving three unknown variables, td, , and a. The solution
of these three simultaneous equations can be obtained by a simple trial-and-
error procedure as follows:
Once a solution is obtained, the ultimate shear stress T;, can be calculated
from Eqs. 23 and 19, the torsional strength T can be obtained from Eq. 4.
An example problem showing the solution procedures is given in Appen-
dix I.
1.452 200.2
B3 2.341* 366.4* 332.0 0.906
B4 2.493* 428.9* 419.0 0.977
B7 1.979* 252.3 236.0 0.935
B8 2.206* 275.3* 288.0 1.046
B9 1.887* 265.0* 264.0 0.996
BIO 2.253* 276.2* 304.0 1.101
Dl 1.510 203.5 198.0 0.973
D3 2.345* 371.7* 346.0 0.931
D4 2.494* 429.4* 424.0 0.987
Ml 1.678 252.7 269.0 1.064
M2 2.101 340.1 359.0 1.055
M3 2.770 394.2 388.0 0.984
11 2.275 181.3 190.0 1.048
02 1.636 339.1 357.0 1.053
G3 2.272 444.7 439.0 0.987
G4 2.581* 543.4* 574.0 1.056
G6 1.669 337.0 346.0 1.027
Q7 2.095 466.3 466.0 0.999
Nl 1.058 75.1 80.5 1.072
Nla 1.038 75.4 79.6 1.055
N2 1.473 119.1 128.0 1.075
N2a 1.558 119.2 117.0 0.981
N3 1.426 102.9 108.0 1.049
Kl 1.115 131.4 136.0 1.035
K2 1.565 207.9 210.0 1.010
K3 1.831* 237.2* 252.0 1.062
12 1.425 320.6 319.0 0.995
13 1.808 429.3 404.0 0.941
14 2.147 529.3 514.0 0.971
15 2.424* 611.7* 626.0 1.023
3216
ment so that the beam will not fail brittlely at cracking, i.e. T > T. T
for a hollow beam can be taken as 2Act(2.5\(fl), where t is the actual wall
thickness. For solid beams, t can be taken as Ac/pc; (2) the member should
be under-reinforced so that both the longitudinal bars and the stirrups will
yield at failure. To achieve this purpose, a should be greater than 12 +
3 3 [ T / / ^ ( 0 . 2 7 - 45e,v)] but less than 78 - 3 3 [ T / / ; ( 0 . 2 7 - 45e)]; (3)
stirrup spacing should not be excessive to cause significant drop of torsional
strength, i.e., s should be less than p,/8 or 12 in. (30 cm); (4) concrete
cover should not be too thick to cause spalling before the maximum torque
is reached. In other words, the distance c measured from the concrete surface
to the inner face of the transverse hoop bars should be less than 0.75td.
The theory presented here has been rigorously derived. The only major
inaccuracy introduced is the approximation of kt = 0.80. This approximation
should be quite good after the maximum fiber strain erfj reaches well into
the descending branch of the softened stress-strain curve, i.e. eds > 1.5e0
where e0 is taken as 0.002. Therefore, the theory is very suitable for finding
the torsional strength. At the low load stages when erfl < e0, however, k{
= 0.80 would not be sufficiently accurate, and a more general method of
solution (Hsu 1988) should be used.
The thickness of the shear flow zone given in Eq. 20 is suitable for the
analysis of torsional strength. It is, however, not convenient for the design
of torsional members. In design, the thickness of shear flow zone td should
be expressed in terms of the torsional strength, T. This approach will now
be introduced.
The stress in the diagonal concrete struts, ad, can be related to the thick-
ness td and the shear flow q using the equilibrium Eq. 23:
<rd = (26)
td sin a cos a
3217
td = (27)
lA<Pd,max sin a cos a
Downloaded from ascelibrary.org by Matthew Foreman on 07/22/13. Copyright ASCE. For personal use only; all rights reserved.
A0 = Ac ~~Pc (28)
Substituting A0 from Eq. 28 into Eq. 27 and multiplying all the terms by
2pc/A2c results in:
2
^ ) - 2 K ) + ^ i =0 (29)
Ac I \AC I Ac o-d,max sm a cos a
Define:
* td0 = Ac/pc
T
;i ~ * nPc/Ac
~ &<l,v sin a cos a
Eq. 29 becomes
= 0. (30)
(-)*-<-) +
\W \tdJ
When td/tM is plotted against T/T_max in Fig. 7, Eq. 30 represents a para-
bolic curve. Solving td from Eq. 30 gives:
U ~ tdl 1 - A 1 (31)
This approach of determining the thickness of the shear flow zone, first pro-
posed by Collins and Mitchell (1980) and later adopted by the Canadian
Code ("Design" 1984), gives:
Tpi 1
1 1 - tan a + (32)
Pi L 0.74>/;AI tan a,
In Eq. 32 Ac andpc are replaced by Aj andplt respectively, since the concrete
cover is considered ineffective. o-imax is assumed to be 0.7<$>cf'c, in which
the material reduction factor c can be taken as 0.6.
Eqs. 32 and 31 clearly show that the thickness ratio, td/td0, is primarily
a function of the shear stress ratio, T/f'c. The thickness ratio td/td0 is also
a function of the crack angle a, but is not sensitive when a varies in the
vicinity of 45.
Eq. 3 1 , T < T?max represents the case of under-reinforcement, while T >
T,max means over-reinforcement. The case of over-reinforcement cannot
be expressed by Eq. 3 1 , because it gives a complex number (VT). Fig.
3218
Parabolic Curve
Not Applicable
when t d / t d o > 1
Note: p; and p, are the reinforced ratios of longitudinal steel and transverse hoop steel, respectively, based on total cross-
sectional area Ac. Cross section 10 in. x 15 in. fy ~ 47,000 psi; f'c = 4,000 psi; 1 in. = 25.4 mm; 1 psi = 6.89 kPa.
1 1 1 1
o Hsu
a = tan = or tan
A Lampert & Thiirlimann /* 3 5
_ + Leonhardt a Schelling Z^^a=45
a McMullen S Rangan
v McMullen & Warwaruk
_ * Bradburn & Zia o^*^\*^
$*^\(
~tdo =(0.082+3.405-^-) V
'c Sin2a
Eq.{33)
^- = 4><Eq.36)
'do Ic
-
FIG. 8. Thickness Ratio td/td0 as Straight Line Functions of Shear Stress Ratio,
Tjf'c
wieldy for practical design. In the next section a simplified expression for
td is proposed. The simplicity is obtained with a small sacrifice in accuracy.
A simple expression for the thickness of shear flow zone, td, can be ob-
tained directly from Eq. 2, noting that q = 7td and T = T:
td = - (34)
2A0T
td = C (35)
Acf c
3220
- -
Downloaded from ascelibrary.org by Matthew Foreman on 07/22/13. Copyright ASCE. For personal use only; all rights reserved.
/ o Hsu (1968)
A LeonhardI A Shelling (1974)
s McMullen & Rangan (1978)
D Mitchell & Collins (1974)
Bradburn (1968)
o McMullen & Warwaruk (1967)
q/ a Lampert & Thurilmann (1969)
1 1 1 1 1 A
DO/
-
500 -
q /
^
-
a
A
8
300 -- a/
o Hsu (1968)
6r A Leonhardt & Shelling (1974)
200 J&5 0 McMullen & Rangan (1978)
Mitchell & Collins (1974)
X Bradburn (196B)
o McMullen & Warwaruk (1967) _
1 i i I I I
200 300 400
Tn, calc (in-kips)
FIG. 10. Comparison of Test Strengths with Calculated Strengths Using Pro-
posed td (Eq. 36)Expended Scale for Lower Portion of Fig. 9
3221
A0 = AC- = (37)
Ac J c
ACKNOWLEDGMENT
The writer would like to express his deep appreciation to National Science
Foundation for supporting this study through Grant No. ECE-8511876. He
also wishes to thank X. B. Pang, graduate student in the civil engineering
department, University of Houston, for the computation of Table 1, and the
plotting of Figs. 9 and 10.
A0 = AC- pc(-j +td = 200 - 60( - j + (2)2 = 144 sq in. (929 cm2)
(*&) + fA'M
V Po / U 1.656+ 1.628
U.43/1
* o.Wct 0.80(4.49)(2)
and Eq. 25.
(A,f\
\po) 1.656
\ 1.656+1.628
(M) + (k
V Po / \ s/
2
sin a = 0.4958
Aa? 144(0.4571)2
d 2 2
po sin a cos a 52(0.5042)(0.4958)
= 2.314 in. (5.88 cm) > 2 in. (5.08 cm)
N.G. repeat cycle assuming td = 2.10 in. (5.22 cm)
Now
td = 2.094 in. (5.32 cm) *= 2.10 in. (5.33 cm) o.k.
and finally Eq. 4.
3223
Design the reinforcement for the hollow box beam with the trapezoidal
cross section as shown in Fig. 11. The beam should be able to resist a tor-
sional moment of 7,400 in.-kip (836 kN-m). The net concrete cover is 1.5
in. (3.81 cm) and the material strengths a r e / ; = 4,000 psi (27.6 MPa) and
fy = 60,000 psi (413 MPa).
Downloaded from ascelibrary.org by Matthew Foreman on 07/22/13. Copyright ASCE. For personal use only; all rights reserved.
Solution
For the given outer cross-sectional dimensions shown in Fig. 11
A (3 + 4)(3)(12)2
Ac = = 1>512 sq in, ( 9 ) 755 c m 2),
4T 4(7,400)
td = - : = = 4.89 in. (12.4 cm) < 5 in. (12.7 cm)
AJ'C 1,512(4) ,
wall thickness o.k.
, Pch 157(4.89)
A0 = AC = 1,512 = 1,128 sq in. (7,277 cm2)
2 2
0.5ft 3 ft '0.5ft'
3224
The minimum a is
TnPc
= 12 + 33 2
A J'C(0.21 - 45e)J
7,400(157)
12 + 33
_(1,512)2(4)(0.27 - 45 0.00207)
= 35.7 < 45. Under-reinforced.
Select a = 45 for best crack control.
A,
- = 0.0547(1) = 0.0547 sq in./in. (0.139 cm 2 /cm)
s
0.44
Select No. 6 bars ,s = = 8.04 in. (20.4 cm)
0.0547
Check stirrup spacing
0.75
157 - 4(2)1 !-5 +
Pl v 2
= 17.75 in. (45.1cm)
i = 8.04 < 12 in. (30 cm) < 17.75 in. (45.1 cm) o.k.
Use No. 6 transverse hoop bars at 8 in. (20.3 cm) spacing.
Actual A, = 13(0.60) = 7.80 sq in. (50.3 cm2) > 7.51 sq in. (48.4 cm2) o.k.
3225
3226