ACI STRUCTURAL JOURNAL TECHNICAL PAPER

Title no. 85-856

Softened Truss Model Theory for Shear and Torsion

by Thomas T. C. Hsu

The softened truss model theory, which has recently been developed It is generally agreed by researchers in recent years
for shear and torsion of reinforced concrete members, is summarized that the truss model theory provides a more promising
in a systematic and unified manner. Eleven equations involving jour-
teen variables are derived from equilibrium, compatibility, and ma-
way to treat shear ·and torsion. The original truss model
terials conditions to solve the shear problem. An additional six equa- concept was first proposed to treat shear problems by
tions involving six more variables are required to treat the torsion Ritter2 and Morsch 3 at the turn of the 20th century. It
problem. The theory was successfully applied to structures where was extended to treat torsion problems by Rausch 4 in
shear behavior predominates, such as low-rise shearwalls, framed wall 1929. These theories hold that a concrete element rein-
panels, deep beams, and shear transfer strengths. It also worked very
well for members subjected to torsion.
forced with orthogonal steel bars and subjected to shear
Efficient algorithms are proposed to solve the simultaneous equa- stresses would develop diagonal cracks at an angle in-
tions for different types of structures. The theoretical predictions are clined to the steel bars. These cracks would separate the
in good agreement with the test results in all cases. The prediction in- concrete into a series of diagonal concrete struts, which
cludes not only the shear and torsional strengths, but also the defor- are assumed to resist axial compression. Together with
mations of structures throughout their post-cracking loading history.
the steel bars, which are assumed to take only axial
tension, they form a truss action to resist the applied
shear stresses. For simplicity, the concrete struts are as-
Keywords: angle of twist; beams (supports); compatibility methods; deep
beams; deformation; equilibrium methods; failure mechanisms; frames; rein-
sumed to be inclined at 45 deg to the steel bars. Con-
foreed eonerete; reinforcing steels; shear properties; shearwalls; strains; sequently, these theories are known as the 45 deg truss
strength; stresses; stress-strain relationships; structural analysis; torsion; trusses model.
walls.
The rudimentary truss model of Ritter, Morsch, and
Rausch is very elegant, and the equations derived from
the equilibrium conditions are simple. Unfortunately,
A reinforced concrete member may be subjected to the predictions from these equations did not agree with
four types of actions; namely, axial load (compression the test results. For the case of pure torsion, the theory
and tension), bending, shear, and torsion. The first two may overestimate the test strengths by 30 percent. For
types of actions, axial loads and bending, are quite well the case of low-rise shearwalls, the overestimation may
understood; and the design methods are essentially the exceed 50 percent.
same for different nations. In contrast, the last two To improve the predictions of the truss model, the
types of actions, shear and torsion, are not well under- theory had undergone three major developments. The
stood; and the empirical design methods used in the first important development was the generalization of
codes and specifications are very different around the the angle of inclination of the concrete struts by Lam-
world. pert and Thurlimann. 5 They assumed that the angle of
In the past, there were two basic approaches used to inclination may deviate from 45 deg. On this basis,
analyze shear and torsion problems in reinforced con- three basic equilibrium equations had been derived,
crete; namely, the mechanism method and the truss which could explain why longitudinal and transverse
model method. The mechanism method is the basis of steel with different percentages can both yield at fail-
the current shear and torsion provisions in the ACI ure. Their theory was known as the variable-angle truss
Building Code. 1 By fitting the mechanism method to
the test results, the ACI method becomes empirical or
at best semi-empirical. From a theoretical point of
Received Mar. 3, 1987, and reviewed under Institute publication policies.
view, this method cannot satisfy the compatibility con- Copyright © 1988, American Concrete Institute. All rights reserved, including
dition, and in certain cases, even the equilibrium con- the making of copies unless permission is obtained from the copyright propri-
etors. Pertinent discussion will be published in the September-October 1989 ACI
dition. Structural Journal if received by May I, 1989.

624 ACI Structural Journal I November-December 1988

 ened truss model theory to emphasize the importance of
Thomas T. C. Hsu, FACI, is a professor in the civil engineering department at
the University of Houston. Dr. Hsu is the author of many technical publica- taking into account the softening of the concrete struts.
tions, and in 1965 was the corecipient of ACI's Wason Medal for Materials The discovery and the quantification of this softening
Research. He is a member of ACI Committees 215, Fatigue; 358, Concrete phenomenon has solved the nagging mystery that has
Guideways; and joint ACI-ASCE Committees 343, Concrete Bridge Design; and
445, Shear and Torsion. plagued the researchers for almost half a century. The
purposes of this paper are: (1) to show how the treat-
ment of shear and torsion can be unified on the basis
model. Since plasticity was assumed at failure, it could of the softened truss model theory; and (2) to show
also be called the plasticity truss model. The second de- how this theory can be applied to a variety of practical
velopment was the derivation of the compatibility problems, including shear transfer, deep beams, low-
equation by Collins6 to determine the angle of inclina- rise shearwalls, framed wall panels, and members sub-
tion of the concrete struts. Since this angle was as- jected to torsion. 9- 16
sumed to coincide with the angle of inclination of the
principal compression stress and strain, this theory was
known as the compression field theory. The author, GOVERNING EQUATIONS FOR SHEAR
however, preferred to call it the compatibility truss A reinforced concrete element is subjected to shear
model, because the average strain condition should sat- stresses and normal stresses as shown in Fig. l(a). The
isfy Mohr's strain circle in this theory. The stress in the directions of the longitudinal and transverse steel bars
concrete struts should also satisfy Mohr's stress circle. are designated as the£ and t-axes, respectively, consti-
The third development was the discovery of the soft- tuting the £-t coordinate system. Accordingly, the nor-
ening of concrete struts by Robinson and Demorieux7 mal stresses are a1 and a" and the shear stresses are r1,.
and the quantification of this phenomenon by Vecchio After the development of diagonal cracks, the con-
and Collins. 8 Vecchio and Collins proposed a softened crete struts are subjected to compression and the steel
stress-strain curve, in which the softening effect de- bars act as tension links, thus forming a truss action.
pended on the ratio of the two principal strains. The compression struts are oriented in the d-axis, which
Combining the equilibrium, compatibility and soft- is inclined at an angle a to the longitudinal steel bars.
ened stress-strain relationships, a theory was developed This direction is also assumed to be the direction of the
that can predict with good accuracy the test results of principal compressive stress and strain. Taking the di-
various types of reinforced concrete structures sub- rection perpendicular to the d-axis as the r-axis, we
jected to shear or torsion. The theory can predict not have a d-r coordinate system in the direction of the
only the shear and torsional strengths but also the de- principal stresses and strains. The normal principal
formations of a structure throughout its post-cracking stresses in the d and r directions are ad and a, respec-
loading history. This theory has been called the soft- tively.

(a) DEFINITIONS OF STRESSES AND COORDINATE SYSTEM

O"t

-1891;,-
T
REINFORCED CONCRETE REINFORCEMENT
CONCRETE

(b) SUPERPOSITION OF CONCRETE STRESSES AND STEEL STRESSES

Fig. 1-Stress condition in reinforced concrete element

ACI Structural Journal I November-December 1988 625
 Compatibility equations
From the compatibility condition of the truss model,
it can also be shown 17 that the average strains (or
Eq(7a) smeared strains) satisfy Mohr's strain circle, giving

(4)

(5)

'Ytr = 2(ed - er) sina cosa (6)

Ed where
average strains in the £ and t directions, re-
(a) Compression Stress-Strain Relationship spectively (positive for tension)
'Ytr average shear strains in £-t coordinate (posi-
tive as shown in Fig. 1 for r11)
average principal strains in the d and r direc-
tions, respectively (positive for tension)
Eq(9a)
Material laws
The stress and strain of concrete in the d-direction is
assumed to obey the following material law proposed
by Vecchio and Collins 8 for the softened concrete

(b) Tension Stress-Strain Relationship

Fig. 2-Stress-strain relationship for softened concrete

Equilibrium equations
Eq. (7a) and (7b) are plotted in Fig. 2(a). The stress
From the three equilibrium conditions of the truss 1: is the maximum compressive stress of a nonsoftened
model, it can be shown 17 that the stresses in the con- standard cylinder. The strain f is defined as the strain
0

at the maximum compressive stress of nonsoftened

crete satisfy Mohr's stress circle. Assuming that the
concrete and can be taken as - 0.002. The factor tis a
steel bars can resist only axial stresses, then the super-
position of concrete stresses and steel stresses as shown softened coefficient suggested to be
in Fig. 1(b) results in

(1) (8)

The softening coefficient t. which is less than unity,

is the reciprocal of the coefficient A. given in Reference
r1, = (ud - ur) sina cosa (3)
8. The Poisson ratio p, in Eq. (8) is taken as 0.3. The
strains and stresses ed, e0 , ud and 1: in Eq. (7) and (8)
are negative for compression and er is positive for ten-
where
sion.
normal stresses in the £ and t directions, re- The stress-strain relationship in the r-direction can be
spectively (positive for tension)
expressed by
shear stresses in the £-t coordinate (positive as
shown in Fig. 1)
(9a)
principal stresses in the d and r directions, re-
spectively (positive for tension)
angle of inclination of the d-axis with respect where
to £-axis Ee initial modulus of elasticity of concrete, taken
Ptt Pr = reinforcement ratios in the £ and t directions, to be - 2/; /e with f = -0.002
0 0

respectively Eer strain at cracking of concrete taken to befe/Ee

j,,J, steel stresses in the £ and t directions, respec- fer stress at cracking of concrete assumed to be
tively 4.Jl:, where/; and .JJ: are expressed in psi
626 ACI Structural Journal I November-December 1988
 fer
a, = --~-=c==== (9b) uv
1 + f, - fer

0.005 la u
'
T
/ /

~ """"
'/
Eq. (9a) and (9b) are plotted in Fig. 2b. 1/
'/

The stress-strain relationships for the longitudinal L '

/
'/
""

1
'/ /
and transverse steel bars are assumed to be elastic-per- '/
r
/
/ "
' / "
fectly plastic

ft ~ fly f, = f,y (lOa) ~ .'-.. II

Dv \
l
ft < fty j, = Esft (lOb)
Critical Zone (b)

E1 ~ E1y j, = /,y (11a)

E1 < E1y J, = Esft (11b)

where
(a)
Es modulus of elasticity of steel bars
f,y, /,y = yield stresses of longitudinal and transverse Fig. 3-Push-off test specimen for shear transfer
steel bars, respectively
yield strains of longitudinal and transverse
steel bars, respectively Ed, f"a, and r. By selecting a value of one unknown
variable, usually Ed, the other five unknowns can be
The 11 governing equations for shear, Eq. (1) to (11), solved.
contain 14 unknown variables. These unknown vari-
ables include seven stresses (a1, a,, Tw ad, a" J,, J,) and APPLICATIONS OF SHEAR EQUATIONS
five strains (E1, e, "(1, ed, E,) as well as the angle a and Shear transfer strength
the material coefficient r. If 3 unknown variables are The application of the softened truss model theory to
given, then the remaining 11 unknown variables can be predict the shear transfer strength of reinforced con-
solved by the 11 equations. For example, in the case of crete was made by Hsu, Mau, and Chen. 9 A typical
pure shear, the known variables are a1 = 0 and a1 = 0. specimen is shown in Fig. 3(a). 18 Under the applied ex-
The third known variable can be chosen for conve- ternal load V, shear stresses are created in the shear
nience, say Ed. For each given value of ed, the remaining plane, which coincides with the plane of action of the
11 unknowns can be solved. The series of solutions for line load. For an initially uncracked specimen, the load
various Ed values allow us to trace the post-cracking V will create diagonal cracks in the vicinity of the shear
loading history. plane. This cracked region is called the critical zone, in
At first sight, the solution of 11 nonlinear equations which failure will occur. In this zone the concrete struts
appears to be complex. However, careful examination and the longitudinal and transverse steel bars will form
shows that these 11 equations can be drastically re- a truss action to resist the applied load. Therefore, the
duced because they exhibit three characteristics. First, softened truss model theory is expected to be applica-
the 3 equations for material laws of concrete, Eq. (7) ble.
through (9), are expressed in terms of six unknown An element isolated from the critical zone is shown
variables, including the four unknown stress and strain in Fig. 3(b). With the imposed i-t coordinate system,
variables in the d-r coordinate ad, a" ed, and e, plus a the stresses acting on the element are a, and 7 11 • Assum-
and r. Second, Eq. (1) and (2) are coupled to the com- ing that the shear stresses are uniformly distributed
patibility and materials equations. f, in Eq. (1) and/, in along the shear plane and the transverse normal stresses
Eq. (2) can be expressed in terms of Ed, En and a are uniformly distributed on a plane perpendicular
through Eq. (10), (11), (4), and (5). Accordingly, Eq. thereto, then Tu = VI Lb, where L and bare defined in
(1) and (2) can also be expressed in terms of the six un- Fig. 3. Since these two stresses a1 and r1, are directly
known variables, when a1 and a1 are given. Third, ex- proportional to the external applied load V, they can be
amination of Eq. (3) through (6), (10), and (11) shows related by a proportional constant K, i.e.
that the six unknown stress and strain variables in the
i-t coordinate T 11 , E1, E,, "(1,, j,, and/,, can each be inde- (12)
pendently expressed as a function of some of these six
unknowns. Therefore, out of the set of 11 equations we where K = Lie. A sensitivity analysis 9 has shown that
can first solve a set of 5 equations - Eq. (1), (2), (7), the shear transfer strength is not sensitive to the con-
(8), and (9) - involving only the six unknowns ad, a" stant K. The shear strength will be changed by less than
ACI Structural Journal I November-December 1988 627
 sym.
v v obtained for the selected Ed value. Otherwise, a new E,
r-a is assumed, and Steps 2 through 7 are repeated.

rt'~T
9. Select another value of Ed and repeat Steps 1
t r------.,
I through 8. In this way, a series of solutions for various
··O"t
L
I
I
I
1T111 0
..._

~
~Tit
I
I
Ed values can be obtained.
10. The stress and strains in the i-t coordinates r1,, E,
L I I

11 1
I f-O"t I
I ______ .J Et, 'Ytt• ft, and !t can be calculated from Eq. (3),. (4), (5),
L
(6), (10), and (11), respectively. The relationship be-
-lbl- tween any two variables, such as ru and -y,, can be
plotted.
v v Twenty push-off specimens tested by Mattock et
al. 18"20 have been compared to the softened truss model
Fig. 4-Stress condition in web of deep beam theory. 9 The mean value of the ratio of the measured
shear strength to calculated shear strength is 1.054 and
± 10 percent when K is doubled or halved. the standard deviation is 0.069.
Eq. (12) provides the first given condition. The sec-
ond condition is simply Deep beams
The prediction of deep beams using the theory was
made by Mau and Hsu. 10 A typical deep beam Fig. 4 is
Ut = 0 (13)
subjected to a two-point loading V and is simply sup-
ported. The shear span-to-height ratio a/h must be less
Using Eq. (12), (3), (11), and (5), then Eq. (2) can be
than 2. Under the applied load V, diagonal cracks will
expressed as follows
occur in the shear span, inducing a truss action. The
moment in the shear span is assumed to be resisted by
(ud - u,) K sina cosa (14a) the top and bottom bars while the shear force is taken
= ud sin2 a + u, cos 2a + p,j,Y by the truss action in the web. The web is defined by
the dotted rectangle in Fig. 4, the height of which is d,.
Er < Ery (ud - u,) K sina cosa d, is the distance between the centroids of the top and
= ud sin2 a + u, cos2a (14b) bottom steel. To develop the truss action, the web must
+ p,E. (Ed sin a + 2 E, cos a)
2 have transverse and longitudinal steel. The longitudinal
steel could include the top and bottom bars as long as
Also, using Eq. (13), (10), and (4), then Eq. (1) be- they do not yield prior to the shear failure in the web.
comes Fig. 4 also shows the stress condition of an element
in the web of the shear span. The element is subjected
to a shear stress r11 and a transverse (vertical) normal
stress u,. ru is defined as V/bdv. Since ru and u, are both
ud cos2 a + u, sin2a (15b) proportional to the applied load V, they can be related
+ p,E. (Ed sin2a + E, cos 2a) = 0 by a proportional constant K as expressed in Eq. (12).
Similarly, u1 = 0 in Eq. (13) is also applicable. With
Eq. (14) and (15), as well as Eq. (7) through (9) for these two identical conditions given, the solution pro-
the material laws of concrete, are all expressed in terms cedures for deep beams are identical to those derived
of the six unknown variables ud, u, Ed, E, a, and By r. for the shear transfer strength.
The only difference between these two physical
giving a known value to the variable Ed, the remaining
five unknown variables can be solved by the five equa- problems (deep beams versus shear transfer) is the def-
tions. Ed is selected because it is expected to vary mon- inition of the proportional constant K. K is equal to
otonically during the loading history. Lie in the shear transfer problem, but should become a
The five simultaneous equations can be easily solved function of the shear span-to-height ratio a/h in the
by an iteration procedure: deep beam problem. An approximate expression of K
1. Select a value for Ed. for deep beams is given in Reference 10
2. Assume a value for E,.
3. Finds from Eq. (8). K = 2dv a
h
0 < h~ 0.5 (16a)
4. Find ud from Eq. (7).
5. Find u, from Eq. (9).
6. Solve a from Eq. (15). To determine whether Eq. a
0.5 < h ~ 2 (16b)
(15a) or Eq. (15b) has been correctly used, check E1 us-
ing Eq. (4). ,
7. Calculate E, from Eq. (14b) if E, < Err If E1 ~ E1y, In Eq. (16b) K varies from 2dvfh (about 1.6) to zero
check Eq. (14a). E, can be calculated using Eq. (5). when a!h varies from 0.5 to 2. When a/h exceeds 2, K
8. If E, is close enough to the assumed value or if Eq. remains zero because the beam has reached the geo-
(14a) is satisfied, a set of SOlution Ud, U, E, a, and Sis metric proportion of a slender beam. The sensitivity of
628 ACI Structural Journal I November-December 1988
K on the shear strength has also been studied. A varia- a:
i5
tion of the K value by ± 0.25 does not change the shear
strength by more than 10 percent.
gt
z-
0
HOR. DIR.

A total of 64 deep beams have been tested by Smith v

...J

and Vantsiotis; 21 Kong, Robins, and Cole; 22 and De- --·~.-----------rr~

Paiva and Siess. 23 The test results of these beams are
compared to the theory. 10 The ratio of the experimental
to theoretical maximum shear forces has a mean value
of 0.975 and a standard deviation of 0.089.
(a) GENERAL VIEW (b) WALL ELEMENT A
Low-rise shear walls
The theory was applied to low-rise shearwalls by Hsu
and Mo 11 and Mau and Hsu. 12 A low-rise shearwall,
shown in Fig. 5(a), is defined as a wall subjected to a
horizontal force V and having a height-to-width ratio
less than unity. For such a structure, the sectional mo-
ments are resisted by the flanges (or boundary ele-
ments), while the shear forces are taken by the web. As
a whole, the deformation is governed primarily by (c) SECTION I- I
shear.
After diagonal cracking, a truss action develops in Fig. 5-Low-rise shearwall
the web. An element A is isolated in Fig. 5(b) on which of concrete, i.e., (J, = 0. Hence, Eq. (9) for tensile
the £-t coordinate system is imposed. The vertical axis stress-strain relationship of concrete becomes irrele-
is defined as the longitudinal £-axis and the horizontal
vant. Since f, = 0, .ft in Eq. (2) should also become
axis is the transverse t-axis. The diagonal cracks are,
zero. Without the term p,j,, Eq. (2) is now uncoupled
therefore, inclined at an angle a to the £-axis. On the
from the compatibility and stress-strain equations just
four edges of the element exists a shear stress Tm which
as Eq. (3). The only equilibrium equation that is still
is defined as Vlbdv. dv is taken to be the distance be-
coupled to the compatibility and stress-strain equations
tween the steel area centroids of the two flanges, Fig.
is Eq. (1). Substituting Eq. (20) and (10) into Eq. (1)
5(c).
gives
The particular boundary characteristics of a low-rise
shearwall is the large, rigid foundation. Because of its
large constraint, the transverse strain in the web can be ft ;;;;: fey r2 = qe - Pefey

ad
(22a)
assumed to be negligible, i.e
a, - P 1 E, f 1
f, = 0 (17) fe < fey r2 = ad
(22b)

With this given condition, the equilibrium and compat- In Eq. (22) q 1 is a given constant which may be zero if
ibility equations can be further simplified. no vertical loads are present.
Compatibility equation-Using f 1 = 0 and substitut- Eq. (7}, (21}, and (22) contain four unknowns fd, f 1,
ing Eq. (5) into Eq. (4) gives (Jd, and !;. By assuming a value of fd, the other three
unknowns can be solved. The solution can be obtained
(18) by a simple iterative procedure as follows:
1. Select a value of fd.
Summing Eq. (4) and (5), it can easily be shown that 2. Assumes and calculate f 1 and ad from Eq. (21) and
(7), respectively.
(19) 3. Substitute f 1 and (Jd into Eq. (22) to calculate !;.
4. If r is close to the assumed value, a solution is ob-
Substituting Eq. (17) through (19) into Eq. (8) (making tained for the three unknown ft, ad, and r. If is not r
also a slight change by taking p. equal to zero) we ob- close enough to the assumed value, repeat steps (2) and
tain a very simple form of r (3).
5. Select another value of fd and repeat Steps (2)
r= cosa (20) through (4). fd can be selected with a suitable increment
up to 0.0035. This will give a set of solutions for the
Substituting Eq. (20) into Eq. (18) gives four unknowns, fd, ft, ad, and r.
6. Calculate a, f" T1, 'Yw andft from Eq. (20), (19),
(21) (3}, (6), and (10}, respectively. The relationship of any
two variables, such as T 11 versus 'Yw can be plotted.
Equilibrium equation-Equilibrium Eq. (1) through Twenty-four low-rise shearwalls have been tested by
(3) can be simplified by neglecting the tensile strength Barda, Hanson, and Corley; 24 Benjamin and Wil-
ACI Structural Journal I November-December 1988 629
 much greater than the longitudinal (vertical) restraint.
--~=~~~=~~Frame If the transverse restraint is assumed to be infinite, then
Beam
E1 = 0.

The longitudinal compressive stress a1 can be as-

- - ~=~#.~=~~ Frame
Column sumed a known value uniformly distributed along the
horizontal width of a wall panel. Since plastic hinges
--~=~~~=~-Element are observed at the ends of the beams when failure is
in Panel approached, a1 can be calculated as the uniform stress,
which produces a plastic mechanism in the frame
(a) Framed Shearwall beams. In addition, the weight of the upper structures
should also be taken into account. With the two given
f conditions E, = 0 and a1 calculated independently, the
solution of the framed wall panel becomes identical to
/Frame Beam
that of the low-rise shearwalls. Hence, the solution
Wall Panel procedures will not be repeated here.
I- Frame
Column
The theories are compared to the test results of eight
framed wall panels by Tomii and Esaki which have very
large frame beams. 27 As the boundary condition of E, =
0 is nearly satisfied, the mean ratio of the test strength
(b) Framed Wall Panel to the calculated strength is 0.996 and the standard dt,
(c) Element in Panel viation is 0.037. A comparison is also made with the 14
specimens tested by Yamada, Kawamura, and Katagi-
hara with smaller frame beams. 28 The mean test-to-cal-
Fig. 6-Frame shearwa/1 and frame wall panel culated shear strength ratio is 1.087 and the standard
deviation is 0.149.
liams; 25 and Galletly. 26 A comparison of these tests with
the theory was made. 11 All the shearwalls are subjected GOVERNING EQUATIONS FOR TORSION
to horizontal loads without the presence of vertical For members subjected to torsion, new equilibrium,
loads. The ratio of experimental to theoretical shear compatibility, and material equations will have to be
strength has a mean value of 1.027 and a standard de- derived in addition to the 11 for shear. The derivations
viation of 0.147. The shear deformations are also cor- have been given by Hsu and Mo. 14"16 In this paper,
rectly predicted. however, the equations for torsion will be systemati-
cally introduced, and a new and simpler algorithm will
be developed for solving the simultaneous equations.
Framed wall panels
Framed shearwalls as shown in Fig. 6{a) and (b) were
analyzed theoretically by Mau and Hsu. 13 A framed Equilibrium equations
shearwall is a beam-and-column frame inserted with A rectangular section subjected to a torque T is
wall panels to increase the strength and stiffness of the shown in Fig. 7(a). On the cross section a shear flow q
frame to resist horizontal loads. In seismic regions the will develop along an outer ring area. This area will be
wall panels serve as sacrificial elements to absorb en- called the shear flow zone. According to Bredt's the-
ergy during earthquakes so as to insure the safety of the ory/9·17 the shear flow q should be constant along the
frame system. They can also be rapidly repaired after- centerline of the shear flow zone and can be related to
ward. This design concept is very popular in Japan, the torque T by a very simple equilibrium equation q =
China, and Southwest Asia, but is seldom used in the T/2A"' where A a is the area within the centerline of the
United States. shear flow.
When a framed shearwall is subjected to horizontal After diagonal cracking, a series of spiral concrete
load, the wall panel will be subjected to shear stresses struts are created that interact with the longitudinal
r 1, as shown in Fig. 6{c). After diagonal cracking due to bars and the hoop bars to form a space truss. For an
shear forces, the wall panel will act like a truss so that element within the shear flow zone, a plane truss action
the diagonal concrete struts are in compression and the is developed similar to that for two-dimensional struc-
longitudinal (vertical) and transverse (horizontal) rein- tures discussed previously. If the element is isolated
forcements are in tension. The tensile strains in the steel from the zone in thef-t coordinate [Fig. 7(b)], it will be
produces an expansion of the panel. This expansion is subjected to a shear stress r 1, on its four edges. This
restrained by the frame, causing longitudinal and shear stress is acting on the thickness of the shear flow
transverse compressive stresses in the panel. zone td and can be expressed as
As the height-to-length ratio of a wall panel is usu-
ally in the range of Y2 to \14, the flexural stiffness of the
columns will be much greater than that of the beams. T
(23)
Hence, the transverse (horizontal) restraint will be Ter = 2 Ao td

630 ACI Structural Journal I November-December 1988

 (a) General View

Cracks

(b) Shear Flow Zone (c) Shear Flow Zone

Element in .t -t Element in d-r
Coordinate Coordinate

Fig. 7-Member subjected to torsion

The thickness td will be determined later by the com- ory, 29 •17 () and 'Ytt are related by the compatibility con-
patibility conditions. dition
Using the principle of stress transformation in the
concrete struts, an element in the d-r coordinate [Fig. () Po
= 2 A 'Yet (24)
7(c)] is subjected to a principal compressive stress ud 0

along the d-axis and a principal tensile stress u, along

the r-axis. Therefore, the equilibrium equations, Eq. After diagonal cracking and the formation of the
(1), (2), and (3), should be applicable, except that the truss action, the shear strain -y1, in the wall will cause an
percentages of steel p, and p1 are now based on the area expansion due to the tensile strains in the longitudinal
of the shear flow zone, rather than the total area of and transverse reinforcement E, and E1• Using the strain
concrete. p1 = A/potd where A, is the total area of lon- transformation principle, a principal compressive strain
gitudinal steel in the cross section and Po is the perime- Ed is present along the d-axis and a principal tensile
ter of the centerline of shear flow. p, = A/std where A, strain E, along the r-axis. Consequently, the compatibil-
is the area of one leg of a hoop bar and s is the spacing ity equations, Eq. (4) through (6), should be applica-
of the hoop bars. ble. The a angle in these equations is identical to that
In short, for a section subjected to torsion, equilib- used in the equilibrium equations because the principal
rium conditions will provide four equations, namely, strain axes are assumed to coincide with the principal
Eq. (1) through (3) and (23). The additional Eq. (23) stress axes.
introduces an additional new variable T. Being a func- In addition to the strain Ed in the d-direction, the di-
tion of td, Eq. (1), (2), and (23) are now coupled to a agonal concrete struts will also be subjected to a bend-
new set of compatibility equations that determines td. ing action resulting from the warping of the surface on
the shear flow tube. The curvature of the concrete
Compatibility equations struts 1/; can be related by geometry to the angle of twist
When a girder is subjected to torsion each cross sec- ()5,14,17

tion will rotate producing an angle of twist () and a

shear strain -y11 in the wall. According to Bredt's the- 1/; = () sin2a (25)
ACI Structural Journal I November-December 1988 631
 HOOP LONGIT.
BAR BAR

.:::----
CENTERLINE OF
SHEAR FLOW ~
2

STRAIN ACTUAL
DIAGRAM STRESS
BLOCK
UNIT WIDTH OF
CONCRETE STRUT

Fig. 8-Strain and stress distributions in concrete struts

Due to the bending of the concrete struts, the tension

area in the inner portion of the cross section will be ne-
glected. The area in the outer portion that is in
compression will be considered as effective to resist the
shear flow. The depth of the compression zone is
therefore defined as the thickness of the shear flow
zone td. Within the thickness td the strain distribution is Based on Eq. (29), k, has been tabulated as a function
assumed to be linear. The thickness td can therefore be r
of Eds and in Table 7.4 of Reference 17.
related to the maximum strain at the surface eds and the Besides Eq. (28) and (29), Eq. (8), (10}, and (11) for
curvature if; by concrete softening and steel characteristics remain ap-
plicable in the case of torsion, giving a total of five
equations. The additional equation, Eq. (28), also in-
(26) troduces a new material variable k, in addition to the
softening coefficient r.
It should be pointed out that Eq. (7), which consid-
Moreover, because of the linear strain distribution, the
ers only axial compressive stress in the concrete struts,
maximum strain eds should be related to the average
is no longer applicable to torsion. It has been replaced
strained by
by Eq. (29), which takes into account bending as well
as axial compression. Furthermore, the tensile strength
(27)
of concrete is neglected in torsion study, i.e., u, = 0,
and Eq. (9) becomes irrelevant.
Eq. (24) through (27) are the four compatibility
The total number of equations for equilibrium, com-
equations required for torsion in addition to the three
patibility, and material laws, as well as the total num-
compatibility equations (4) through (6). These four ad-
bers of variables for stresses (or forces), strains (or
ditional equations also introduce four additional strain
geometric quantities), and material characteristics are
and geometric variables 0, if;, Ewn and td.
summarized in Table 1. It can be seen that 16 equa-
tions are available involving 19 variables. If 3 variables
Material laws
are given, the remaining 16 variables can be solved.
Being subjected to axial stress and bending, the dis-
tribution of the compressive stresses within the thick- APPLICATION TO PURE TORSION
ness td is shown by the solid curve in Fig. 8. Based on For the case of pure torsion, two variables are
the softened stress-strain relationship described in Eq. known; namely, u1 = 0 and u1 = 0. For the third
(7), the peak stress is t J: .The average compressive known variable, eds (and therefore Ed = Ed/2) is se-
stress ud, as shown by the dotted line, can be defined as lected, since eds increases monotonically with increasing
loading.
(28) In this paper, an efficient algorithm is presented
which reduces the 16 equations to 6 equations involv-
where k, is the ratio of the average stress to peak stress
in the stress block. The k,-ratio can be obtained by in-
ing 7 variables ud, Ed, e, a, td, r.
and k,. The 6 equa-
tions include the 3 equations for materials - Eq. (8),
tegration of Eq. (7) and should be a function of the (28), and (29); 2 equilibrium equations derived from
maximum strain f:ds and the softening coefficient r Eq. (1) and (2); as well as one compatibility equation
for td derived from the new compatibility equations,
(29a) (24) through (27). Substituting Eq. (10) and (4) into Eq.
(1) gives
632 ACI Structural Journal I November-December 1988
 Table 1 - Equations and variables tor torsion
Variables Equations
Strain,
Stress deformation,
or or
Category force geometry Material Equilibrium Compatibility Material
u, f, r Eq.(l) Eq. (4) 'Eq. (t')
u, f, (2) (5) (8)
T, -y, (3) (6) *(Ji')
For (10)
(Jd fd
shear
·~o f, (11)
j, a
j,
T 8 k, (23) (24) (28)
Additional (25) '(29)
for
torsion "'
td (26)
fds (27)
Total 7 + 10 + 2 4 + 7 + 5
for
torsion 19 16
*When tensile strength of concrete is neglected, u, = 0 and Eq. (9) become ITrelevant.
**Eq. (7) used in shear is replaced by Eq. (29) applicable to torsion.

(30a)
Pc = outer perimeter of concrete cross section
~ a coefficient equal to 1 for rectangular section
and 1r/4 for circular section, ~ can be taken as
unity for all shapes of cross sections with only
(30b) negligible loss of accuracy for Ao and Po
In Eq. (30) p1 = A/po td
Eq. (33) and (34) are derived by assuming that the cen-
Substituting Eq. (11) and (5) into Eq. (2) gives two terline of shear flow lies midway in the thickness of the
possibilities shear flow zone td.
The thickness td can be solved from Eq. (32) by a
(31a) small trial-and-error procedure, if the variables Ed, E"
and a are given. The iteration method for solving the
six equations - (8), (28), (29), (30), (31), and (32) is
(ud + p 1EsEd) tan2 a
Er = (31b) 1. Select a value for Eds and Ed = Ed/2.
PtEs 2. Assume a value of E,.
3. Calculate t from Eq. (8).
In Eq. (31a) and (31b) p, = A,!std.
4. Calculate k 1 from Eq. (29).
The thickness td in Eq. (26) can be expressed in terms 5. Calculate ud from Eq. (28).
of Ed, E, and a by a series of substitution: (a) substitute
6. Assume a value of td and calculate Ao and Po from
'Yet from Eq. (6) into Eq. (24); (b) substitute 8
Eq. (33) and (34).
as per Eq. (24) from Eq. (24) into Eq. (25); (c) substi-
7. Calculate a from Eq. (30). To determine whether
tute 1/; from Eq. (25) into Eq. (26); and (d) substitute Eds
Eq. (30a) or (30b) is valid, check E1 by Eq. (4).
from Eq. (27) into Eq. (26). The resulting expression is
8. Calculate td from Eq. (32). If td is close to the
value assumed, the set of td and a is obtained. If td is
(32) not close enough to the assumed value, repeat Steps 6
and 7.
9. Calculate E, from Eq. (31b) if E1 <E1r E1 can be cal-
Notice that Ao and Po are also functions of td
culated from Eq. (5). If E1 ~ E1y check Eq. (31a).
10. If E, is close enough to the assumed value or if
(33) Eq. (31a) is satisfied, a set of solutions for E, t. kl> ud,
a, and td is obtained for the selected Eds value. Other-
wise, a new E, is assumed and Steps 2 through 9 are re-
(34) peated.
11. Select another value of Eds and repeat Steps 1
where through 10. Eds can be selected with a suitable incre-
Ac = area bounded by the outer perimeter of con- ment up to 0.0035. This will give a set of solutions for
crete cross section Seven VariableS Ed, Ero S• kl, Ud, a, and fd.
ACI Structural Journal I November-December 1988 633
 12. Calculate the other strain or deformation vari- Bericht No. 6506-2, Institut fiir Baustatik, ETH, Zurich, June 1968,
ables e1, e" 'Ytu 0, and 1/; from Eq. (4), (5), (6), (24), and 101 pp., and "Torsion-Bending Tests on Reinforced Concrete Beams
(25), respectively; and the stress or force variables r1,, j,, (Torsion-Biege-Versuche an Stahlbetonbalken)," Bericht No. 6506-3,
Institut fiir Baustatik, ETH, Ziirich, Jan. 1969, 116 pp.
j,, and T from Eq. (3), (10), (11), and (23), respec-
6. Collins, M. P., "Torque-Twist Characteristics of Reinforced
tively. Concrete Beams," Inelasticity and Non-Linearity in Structural Con-
The procedures just mentioned are illustrated by an crete, Study No. 8, University of Waterloo Press, 1973, pp. 211-231.
example in Appendix A. 7. Robinson, J. R., and Demorieux, J. M., "Essais de Traction-
The softened truss model theory has been used in Compression sur Modeles d'ame de Poutre en Beton Arme,"
IRABA Report, Institut de Recherches Applliquees du Beton Armae,
Reference 14 to predict the strengths and behaviors of Part I, June 1968, 44 pp., and Part 2, "Resistance Ultimate Du Be-
108 torsional beams available in the literature. 5•30•36 The ton De L'ame de Poutres en Double Teen Beton Arme," May 1972,
algorithm is more complicated than the one given here 53 pp.
but the results should be identical. For the 61 under- 8. Vecchio, F., and Collins, M. P., "Stress-Strain Characteristics
reinforced beams with stirrups spacing within the ACI of Reinforced Concrete in Pure Shear," Final Report, IABSE Col-
loquium on Advanced Mechanics of Reinforced Concrete (Delft,
Code limitation, the experimental-to-calculated tor- 1981), International Association for Bridge and Structural Engineer-
sional strengths have a mean value of 1.014 and a stan- ing, Zurich, pp. 211-225.
dard deviation of 0.051. The post-cracking deforma- 9. Hsu, Thomas T. C.; Mau, S. T.; and Chen, Bin, "Theory on
tions (angle of twist, strains in steel and concrete) are Shear Transfer Strength of Reinforced Concrete," ACJ Structural
Journal, V. 84, No. 2, Mar.-Apr. 1987, pp. 149-160.
also correctly evaluated.
10. Mau, S. T., and Hsu, Thomas T. C., "Shear Strength Predic-
CONCLUSIONS tion for Deep Beams with Web Reinforcement," ACI Structural
Journal, V. 84, No. 6, Nov.-Dec. 1987, pp. 513-523.
I. The softened truss model theory is shown to unify
II. Hsu, Thomas T. C., and Mo, Y. L., "Softening of Concrete
the treatment of shear and torsion in reinforced con- in Low-Rise Shear Walls," ACI JOURNAL, Proceedings V. 82, No.6,
crete. Eleven equations involving fourteen variables Nov.-Dec. 1985, pp. 883-889.
were derived from equilibrium and compatibility con- 12. Mau, S. T., and Hsu, Thomas T. C., "Shear Design and
ditions as well as the material laws to solve the shear Analysis of Low-Rise Structural Walls," ACI JouRNAL, Proceedings
V. 83, No. 2, Mar.-Apr. 1986, pp. 306-315.
problem. Six additional equations involving six addi-
13. Mau, S. T., and Hsu, Thomas, T. C., "Shear Behavior of
tional variables are required to solve the torsion prob- Reinforced Concrete Framed Wall Panels with Vertical Loads," ACI
lem. The most efficient algorithms for solving the si- Structural Journal, V. 84, No.3, May-June 1987, pp. 228-234.
multaneous equations are presented for various types of 14. Hsu, Thomas T. C., and Mo, Y. L., "Softening of Concrete
structure. in Torsional Members-Theory and Tests," ACI JOURNAL, Proceed-
ings V. 82, No. 3, May-June 1985, pp. 290-303.
2. The theory is applicable to a variety of structures 15. Hsu, Thomas T. C., and Mo, Y. L., "Softening of Concrete
subjected to shear, such as low-rise shearwalls, framed in Torsional Members-Design Recommendations," ACI JOURNAL,
wall panels, deep beams, and the shear transfer prob- Proceedings V. 82, No. 4, July-Aug. 1985, pp. 443-452.
lems. It is also quite suitable for torsional members. In 16. Hsu, Thomas T. C., and Mo, Y. L., "Softening of Concrete
all these cases, the theoretical predictions agree very in Torsional Members-Prestressed Concrete," ACI JouRNAL, Pro-
ceedings V. 82, No. 5, Sept.-Oct. 1985, pp. 603-615.
well with the test results.
17. Hsu, Thomas T. C., Torsion of Reinforced Concrete, Van
3. The theory is able not only to predict the shear Nostrand Reinhold Co., New York, 1984, 544 pp.
and torsional strength of a structure, but can also pre- 18. Mattock, Alan H., "Effect of Aggregate Type on Single Di-
dict the deformation, such as shear deflection, angle of rection Shear Transfer Strength in Monolithic Concrete," Report No.
twist, and steel and concrete strains throughout the SM 74-2, Department of Civil Engineering, University of Washing-
ton, Seattle, Aug. 1974, 72 pp.
post-cracking loading history.
19. Mattock, Alan H.; Li, W. K.; and Wang, T. C., "Shear
ACKNOWLEDGMENT Transfer in Lightweight Concrete," Journal, Prestressed Concrete
The research is sponsored by the National Science Foundation Institute, V. 21, No. I, Jan.-Feb. 1976, pp. 20-39.
through Grant No. ECE-8511876. 20. Hofbeck, J. A.; Ibrahim, I. 0.; and Mattock, Alan H., "Shear
Transfer in Reinforced Concrete," ACI JouRNAL, Proceedings V. 66,
REFERENCES No. 2, Feb. 1969, pp. 119-128.
I. ACI Committee 318, "Building Code Requirements for Rein- 21. Smith, K. N., and Vantsiotis, A. S., "Shear Strength of Deep
forced Concrete (ACI 318-83), American Concrete Institute, Detroit, Beams," ACI JouRNAL, Proceedings V. 79, No. 3, May-June 1982,
1983, Ill pp. pp. 201-213.
2. Ritter, W., "Die Bauweise Hennebique," Schweizerishe Bauz- 22. Kong, Fung-Kew; Robins, Peter J.; and Cole, David F., "Web
eitung (Zurich), V. 33, No.7, Feb. 1899, pp. 59-61. Reinforcement Effects on Deep Beams," ACI JOURNAL, Proceedings
3. Morsch, E., Der Eisenbetonbau, seine Anwendung und Theo- V. 67, No. 12, Dec. 1970, pp. 1010-1017.
rie, 1st Edition, Wayss and freytag, A.G., Im Selbstverlag der Firma, 23. de Paiva, H. A. Rawdon, and Siess, Chester P., "Strength and
Neustadt a d. Haardt, May 1902, 118 pp.; Der Eisenbetonbau, seine Behavior of Deep Beams in Shear," Proceedings, ASCE, V. 91, ST5,
Theorie und Anwendung, 2nd Edition, Verlag von Konrad Wittmer, Part I, Oct. 1965, pp. 19-41.
Stuttgart, 1906, 252 pp.; 3rd Edition translated into English by E. P. 24. Barda, Felix; Hanson, John M.; and Corley, W. Gene, "Shear
Goodrich, McGraw-Hill Book Co., New York, 1909, 368 pp. Strength of Low-Rise Walls with Boundary Elements," Reinforced
4. Rausch, E., Design of Reinforced Concrete in Torsion (Bere- Concrete Structures in Seismic Zones, SP-53, American Concrete In-
chnung des Eisenbetons gegen Verdrehung), Technische Hochschule, stitute, Detroit, 1977, pp. 149-202. Also, Research and Development
Berlin, 1929, 53 pp. 2nd Edition, 1938. 3rd Edition, titled Drillung, Bulletin No. RD043.01D, Portland Cement Association.
Schub und Scheren in Stahlbetonbau, Deutscher lngenieur-Verlag 25. Benjamin, Jack R., and Williams, Harry A., "The Behavior of
GmbH, Dusseldorf, 1953, 168 pp. One-Story Reinforced Concrete Shear Walls," Proceedings, ASCE,
5. Lampert, P., and Thurlimann, B., "Torsion Tests of Rein- V. 83, ST3, May 1957, pp. 1254-1 to 1254-49.
forced Concrete Beams (Torsionsversuche an Stahlbetonbalken)," 26. Galletly, G. D., "Behavior of Reinforced Concrete Shear Walls

634 ACI Structural Journal I November-December 1988

Under Static Load," Report, Department of Civil and Sanitary En- 2. Assume E, = 0.00425
gineering, Massachusetts Institute of Technology, Cambridge, Aug.
1952, 123 pp. t = ~= I - 0·0015 = 0.5320 Eq. (8)
27. Tomii, M., and Esaki, F., "Design Method of Reinforced ~~ ~0.7(-0.0015) -0.00425
Concrete Framed Shear Walls to Sustain Vertical Loads after Shear
Failure," Proceedings, 8th World Conference on Earthquake Engi- 3. Table 7.4 of Reference 17 gives k, = 0.7881 or Eq. (29)
neering, San Francisco, 1984, V. 5, pp. 581-588.
28. Yamada, M.; Kawamura, H.; and Katagihara, K., "Rein- ad= k,rf: = 0.7881(0.5320)(-4440) = -1861 psi Eq. (28)
forced Concrete Shear Walls without Openings; Test and Analysis,"
Shear in Reinforced Concrete, SP-42, American Concrete Institute, 4. Assume td = 2.46 in.
Detroit, 1974, pp. 539-558.
29. Bredt, R., "Kritische Bemerkungen zur Drehungselastizitat," A, = A,-(t./2)p,+t~ = 150-(2.46/2)(50)+2.46' Eq. (33)
Zeitschrift des Vereines Deutscher lngenieure, V. 40, No. 28, July 11, = 94.5 in.'
1896, pp. 785-790, and No. 29, July 18, 1896, pp. 813-817.
30. Hsu, Thomas T. C., "Torsion of Structural Concrete-Behav- p, = p, - 4td = 50 - 4(2.46) = 40.16 in. Eq. (34)
ior of Reinforced Concrete Rectangular Members," Torsion of
Structural Concrete, SP-18, American Concrete Institute, Detroit, = ~ = 2.4
1968, pp. 261-306. p, p, td 40.16 (2.46) = 0.02429
31. Leonhardt, Fritz, and Schelling, Gunther, "Torsionversuche an
Stahlbetonbalken," Bulletin No. 239, Deutscher Ausschuss fiir - p,E,E, Eq. (30b)
Stahlbeton, Berlin, 1974, 122 pp. cos'a
ad - p, E, (E, - Ed)
32. McMullen, Arthur E., and Rangan, B. Vijaya, "Pure Torsion
in Rectangular Sections-A.Re-Examination," ACI JOURNAL, Pro- -0.02429 {29 X 10') (0.00425)
ceedings V. 75, No. 10, Oct. 1978, pp. Sll-519. -1861 - 0.02429 (29 X 10')(0.00425 + 0.0015)
33. Mitchell, D., and Collins, M.P., "The Behavior of Structural = 0.5065
Concrete Beams in Pure Torsion," Publication No. 74-06, Depart-
ment of Civil Engineering, Univeristy of Toronto, Mar. 1974, 88 pp. a = 44.6 deg
34. Wyss, A. N.; Garland, J. B.; and Mattock, A. H., "A Study
of the Behavior of 1-Section Prestressed Concrete Girders Subjected E, = ed cos' a + e, sin'a = 0.0015 (0.5065) Eq. (4)
to Torsion," Structures and Mechanics Report No. SM69-1, Depart- + 0.00425 (0.4935) = 0.001338
ment of Civil Engineering, University of Washington, Seattle, Mar. < e,, = 0.00165 O.K.
1969, 46 pp.
35. McMullen, A. E., and Warwaruk, J., "The Torsional Strength A, ( Ed )
Eq. (32)
of Rectangular Reinforced Concrete Beams Subjected to Combined td = p, sin' a cos' a Ed - e,
Loading," Report No. 2, Department of Civil Engineering, Univer- 94.5 ( -0.0015 )
sity of Alberta, Edmonton, 1967, 162 pp. = (40.16) (0.4935) (0.5065) -0.0015 - 0.00425
36. Bradburn, J. H., "An Investigation of Combined Bending and = 2.456 = assumed 2.46 in. O.K.
Torsion in Rectangular Reinforced Concrete Members," PhD disser-
tation, Department of Civil Engineering, North Carolina State Uni- E, = ed sin' a + e, cos' a = -0.0015 (0.4935) Eq. (5)
versity, Raleigh, 1968, 174 pp. + 0.00425 (0.5065) = 0.001412 < e,,
= A, = 0 ' 20 = 0.002243
p, std 3.625 (2.46)
APPENDIX - CALCULATIONS FOR BEAM 04 IN
REFERENCE 30 -(ad + p, E, Ed) tan' a Eq. (31b)
The geometric and material properties of Beam D4 follow: E,
p,E,
10 in. - [( -1861)+0.02243(29 X 10')( -0.0015)](0.9743)
b
h 15 in. (0.02243) (29 X 10')
A, 2.4 in.' = 0.004249 = assumed 0.00425 O.K.
A, 0.20 in.'
s 3.625 in. T1, (ad - a,) sin a cos a Eq. {3)
.!. 47,900 psi ( -1861 - 0) (0.702)(0.712) = -930 psi
f,, 48,300 psi
J: 4440 psi "{, 2 (Ed- E,) sin a cos a= 2(-0.0015 Eq. (6)
E, 29 x 10' psi - 0.00425)(0.702) (0.712) = -0.00575
E, -0.002
T = 2A, td T,, Eq. (23)
Solution 2 (94.5) (2.46) (- 930) = -432,000 in.-lb
The calculations for plotting the torque-twist curve are usually done
by a personal computer. Computer calculations show that maximum (J = p, 'Ytt = 40.16 ( -0.00575) Eq. (24)
torque is reached when e., reaches - 0.003 for Beam D4. The follow- 2A, 2 (94.5)
ing hand calculations are made to illustrate the procedures for com- = - 0.001222 rad.lin. = -0.070 deg.lin.
puting the maximum torque and the corresponding angle of twist.
The test values of T and fJ are 424,000 in.-lb and 0.0877 deg.lin., re-
1. Select e.,= -0.003 and ed = ed,/2 = -0.0015 spectively.

ACI Structural Journal I November-December 1988 635