ACI 445.1R-12 - Report On Torsion in Structural Concrete (2013)

ACI 445.
1R-12
Report on Torsion
in Structural Concrete
Reported by Joint ACI-ASCE Committee 445

First Printing
April 2013
American Concrete Institute
Advancing concrete knowledge
Report on Torsion in Structural Concrete
Copyright by the American Concrete Institute, Farmington Hills, MI. All rights reserved. This material may not be
reproduced or copied, in whole or part, in any printed, mechanical, electronic, film, or other distribution and storage
media, without the written consent of ACI.
The technical committees responsible for ACI committee reports and standards strive to avoid ambiguities, omissions,
and errors in these documents. In spite of these efforts, the users of ACI documents occasionally find information
or requirements that may be subject to more than one interpretation or may be incomplete or incorrect. Users who
have suggestions for the improvement of ACI documents are requested to contact ACI via the errata website at
www.concrete.org/committees/errata.asp. Proper use of this document includes periodically checking for errata for
the most up-to-date revisions.
ACI committee documents are intended for the use of individuals who are competent to evaluate the significance and
limitations of its content and recommendations and who will accept responsibility for the application of the material
it contains. Individuals who use this publication in any way assume all risk and accept total responsibility for the ap-
plication and use of this information.
All information in this publication is provided as is without warranty of any kind, either express or implied, includ-
ing but not limited to, the implied warranties of merchantability, fitness for a particular purpose or non-infringement.
ACI and its members disclaim liability for damages of any kind, including any special, indirect, incidental, or con-
sequential damages, including without limitation, lost revenues or lost profits, which may result from the use of this
publication.
It is the responsibility of the user of this document to establish health and safety practices appropriate to the specific
circumstances involved with its use. ACI does not make any representations with regard to health and safety issues
and the use of this document. The user must determine the applicability of all regulatory limitations before applying
the document and must comply with all applicable laws and regulations, including but not limited to, United States
Occupational Safety and Health Administration (OSHA) health and safety standards.
Participation by governmental representatives in the work of the American Concrete Institute and in the develop-
ment of Institute standards does not constitute governmental endorsement of ACI or the standards that it develops.
Order information: ACI documents are available in print, by download, on CD-ROM, through electronic subscription,
or reprint and may be obtained by contacting ACI.
Most ACI standards and committee reports are gathered together in the annually revised ACI Manual of Concrete
Practice (MCP).

38800 Country Club Drive
Farmington Hills, MI 48331
U.S.A.
Phone: 248-848-3700
Fax: 248-848-3701
www.concrete.org
ISBN-13: 978-0-87031-810-8
ISBN: 0-87031-810-1
ACI 445.1R-12

Reported by Joint ACI-ASCE Committee 445
Daniel A. Kuchma, Chair Robert W. Barnes Jr., Secretary
Perry Adebar Hakim Bouadi Neil M. Hawkins Stavroula J. Pantazopoulou

Neal S. Anderson Michael D. Brown Thomas T. C. Hsu* Maria A. Polak
Robert B. Anderson Michael P. Collins Gary J. Klein Julio A. Ramirez
Mark A. Ascheim David Darwin Zhongguo John Ma Karl-Heinz Reineck
Oguzhan Bayrak Walter H. Dilger* Adolfo B. Matamoros David H. Sanders*
Zdenek P. Baant Marc O. Eberhard Denis Mitchell Raj Valluvan
Abdeldjelil Belarbi* Catherine E. French Yi-Lung Mo* James K. Wight
Evan C. Bentz Robert J. Frosch Lawrence C. Novak
John F. Bonacci Gary G. Greene* Carlos E. Ospina
*
Subcommittee members who produced this report.

Subcomittee Chair.
The committee would like to thank the following individuals for their contribution to this report: Mohammad Ali, Neal S. Anderson, Shri Bhide, Michael D. Collins, Maria Cristina
Vidigal de Lima, Leonard Elfgren, Christos Karayannis, Liang-Jenq Leu, Mohammad Mansour, Basile Rabbat, Khaldoun Rahal, and Paul Zia.
A clear understanding of the effects of torsion on concrete The behavior of members subjected to torsion combined with
members is essential to the safe, economical design of reinforced bending moment, axial load, and shear is discussed. This report
and prestressed concrete members. This report begins with a brief deals with design issues, including compatibility torsion, span-
and systematic summary of the 180-year history of torsion of drel beams, torsional limit design, open sections, and size effects.
structural concrete members, new and updated theories and their The final two chapters are devoted to the detailing requirements
applications, and a historical overview outlining the development of transverse and longitudinal reinforcement in torsional members
of research on torsion of structural concrete members. Historical with detailed, step-by-step design examples for two beams under
theories and truss models include classical theories of Navier, torsion using ACI (ACI 318-11), European (EC2-04), and Cana-
Saint-Venant, and Bredt; the three-dimensional (3-D) space truss of dian Standards Association (CSA-A23.3-04) standards. Two design
Rausch; the equilibrium (plasticity) truss model of Nielson as well examples are given to illustrate the steps involved in torsion design.
as Lampert and Thrlimann; the compression field theory (CFT) Design Example 1 is a rectangular reinforced concrete beam under
by Collins and Mitchell; and the softened truss model (STM) by pure torsion, and Design Example 2 is a prestressed concrete
Hsu and Mo. girder under combined torsion, shear, and flexure.
This report emphasizes that it is essential to the analysis of torsion
in reinforced concrete that members should: 1) satisfy the equi- Keywords: combined action (loading); compatibility torsion; compression
librium condition (Mohrs stress circle); 2) obey the compatibility field theory; equilibrium torsion; interaction diagrams; prestressed concrete;
condition (Mohrs strain circle); and 3) establish the constitutive reinforced concrete; shear flow zone; skew bending; softened truss model;
spandrel beams; struts; torsion detailing; torsion redistribution; warping.
relationships of materials such as the softened stress-strain rela-
tionship of concrete and smeared stress-strain relationship of
steel bars. CONTENTS
CHAPTER 1INTRODUCTION AND SCOPE, p. 2

1.1Introduction, p. 2
ACI Committee Reports, Guides, and Commentaries are
intended for guidance in planning, designing, executing, and 1.2Scope, p. 3
inspecting construction. This document is intended for the use
of individuals who are competent to evaluate the significance CHAPTER 2NOTATION AND DEFINITIONS, p. 3
and limitations of its content and recommendations and who 2.1Notation, p. 3
will accept responsibility for the application of the material it 2.2Definitions, p. 5
contains. The American Concrete Institute disclaims any and
all responsibility for the stated principles. The Institute shall
not be liable for any loss or damage arising therefrom. ACI 445.1R-12 was adopted and published April 2013.
Reference to this document shall not be made in contract Copyright 2013, American Concrete Institute.
documents. If items found in this document are desired by All rights reserved including rights of reproduction and use in any form or by any
the Architect/Engineer to be a part of the contract documents, means, including the making of copies by any photo process, or by electronic or
mechanical device, printed, written, or oral, or recording for sound or visual reproduc-
they shall be restated in mandatory language for incorporation
tion or for use in any knowledge or retrieval system or device, unless permission in
by the Architect/Engineer. writing is obtained from the copyright proprietors.
1
2 REPORT ON TORSION IN STRUCTURAL CONCRETE (ACI 445.1R-12)
CHAPTER 3HISTORICAL OVERVIEW OF 9.4Design Example 1: solid rectangular reinforced

TORSION THEORIES AND THEORETICAL concrete beam under pure torsion, p. 67
MODELS, p. 5 9.5Design Example 2: Prestressed concrete box girder
3.1Naviers theory, p. 5 under combined torsion, shear, and flexure, p. 74
3.2Thin-tube theory, p. 5
3.3Historical development of theories for reinforced CHAPTER 10REFERENCES, p. 86
concrete members subjected to torsion, p. 6
3.4Concluding remarks, p. 13 CHAPTER 1INTRODUCTION AND SCOPE
1.1Introduction
CHAPTER 4BEHAVIOR OF MEMBERS Accounting for the effects of torsion is essential to the
SUBJECTED TO PURE TORSION, p. 13 safe design of structural concrete members, requiring a
4.1General, p. 13 full knowledge of the effects of torsion and a sound under-
4.2Plain concrete, p. 13 standing of the analytical models that can easily be used
4.3Reinforced concrete, p. 15 for design. For over three decades, considerable research
4.4Prestressed concrete, p. 17 has been conducted on the behavior of reinforced concrete
4.5High-strength concrete, p. 18 members under pure torsion and torsion combined with other
4.6Concluding remarks, p. 19 loadings. Likewise, analytical models have been developed
based on the truss model concept. Several of these models
CHAPTER 5ANALYTICAL MODELS were developed to predict the full load history of a member,
FOR PURE TORSION, p. 20 whereas others are simplified and used only to calculate
5.1General, p. 20 torsional strength. Many models developed since the 1980s
5.2Equilibrium conditions, p. 20 account for softening of diagonally cracked concrete.
5.3Compatibility conditions, p. 20 This report reviews and summarizes the evolution of torsion
5.4Stress strain relationships, p. 22 design provisions in ACI 318, followed with a summary of
5.5Compression field theory, p. 23 the present state of knowledge on torsion for design and
5.6Softened truss model, p. 25 analysis of structural concrete beam-type members. Despite
5.7Graphical methods, p. 26 a vast amount of research in torsion, provisions of torsion
design did not appear in ACI 318 until 1971 (ACI 318-71),
CHAPTER 6MEMBERS SUBJECTED TO TORSION although ACI 318-63 included a simple clause regarding
COMBINED WITH OTHER ACTIONS, p. 28 detailing for torsion. Code provisions in 1971 were based
6.1General, p. 28 on Portland Cement Association (PCA) tests (Hsu 1968b).
6.2Torsion and flexure, p. 29 These provisions were applicable only to rectangular
6.3Torsion and shear, p. 33 nonprestressed concrete members. In 1995, ACI 318-95
6.4Torsion and axial load, p. 36 adopted an approach based on a thin-tube, space truss model
6.5Torsion, shear, and flexure, p. 37 previously used in the Canadian Standards Association
(CSA-A23.3-77) code and the Comit Euro-International
CHAPTER 7ADDITIONAL DESIGN ISSUES du Bton (CEB)-FIP code (1978). This model permitted
RELATED TO TORSION, p. 39 treatment of sections with arbitrary shape and prestressed
7.1General, p. 39 concrete (Ghoneim and MacGregor 1993; MacGregor and
7.2Compatibility torsion and torsional moment redistri- Ghoneim 1995). The ACI 318-02 code extended the appli-
bution, p. 39 cation of the (ACI 318) 1995 torsion provisions to include
7.3Precast spandrel beams, p. 47 prestressed hollow sections. ACI 318 allows the use of alter-
7.4Torsion limit design, p. 48 native design methods for torsional members with a cross
7.5Treatment of open sections, p. 51 section aspect ratio of 3 or greater, like the procedures of
7.6Size effect on the strength of concrete beams in pre-1995 editions of ACI 318 or the Prestressed Concrete
torsion, p. 53 Institute (PCI) method (Zia and Hsu 1978).
This report reviews and summarizes the present state
CHAPTER 8DETAILING FOR TORSIONAL of knowledge on torsion and reviews their use as a frame-
MEMBERS, p. 53 work for design and analysis of structural concrete beam-
8.1General, p. 53 type members. Chapter 3 presents a historical background
8.2Transverse reinforcement, p. 55 outlining the development of research on torsion of struc-
8.3Longitudinal reinforcement, p. 57 tural concrete members. The general behavior of reinforced
8.4Detailing at supports, p. 58 and prestressed concrete members under pure torsion is
discussed in Chapter 4. In Chapter 5, the compression field
CHAPTER 9DESIGN EXAMPLES, p. 59 theory (CFT) and softened truss model (STM) are presented in
9.1Torsion design philosophy, p. 59 detail. Chapter 5 also includes a description of two graphical
9.2Torsion design procedures, p. 59 methods (Rahal 2000a,b; Leu and Lee 2000). The behavior of
members subjected to torsion combined with shear, flexure,
REPORT ON TORSION IN STRUCTURAL CONCRETE (ACI 445.1R-12) 3
and axial load is discussed in Chapter 6. Chapter 7 introduces D = cross-sectional depth used in fracture mechanics
additional design issues related to torsion, such as precast calculations, mm (in.)
spandrel beams, torsion limit design, size effect, open sections, D0 = size effect constant for computing sN for plain
and torsional moment distribution. Detailing of torsional concrete section
members is described in Chapter 8. Chapter 9 covers detailed D1 = normalized constant to represent characteristic
design examples of several beams subjected to torsion using structural dimensions used in fracture mechanics
ACI 318, EC2-04, and CSA-A23.3-04 design equations, and calculations
additional graphical design methods reported by researchers. Db = size effect constant for computing sN for reinforced
concrete section
1.2Scope Dc = total energy dissipated on discontinuous concrete
Theories presented in this report were developed and verified yield surface
for building members of typical size. For application to large- Ds = total energy dissipated by reinforcement
scale members, size effects should be considered. They could e = moment arm for torsion, mm (in.)
present a serious safety issue when using the shear strength Ec = modulus of elasticity of concrete, MPa (psi)
equations provided in the design standard, which cannot take Eps = modulus of elasticity of prestressed reinforcement
into account the shear strength reduction in large-scale members in flexural tension zone, MPa (psi)
caused by loss of aggregate interlock behavior. Experimental Eps = tangential modulus of Ramberg-Osgood curve at
information on large-scale torsional members is lacking. zero load MPa (psi)
Es = modulus of elasticity of reinforcement and struc-
CHAPTER 2NOTATION AND DEFINITIONS tural steel, MPa (psi)
The material presented is a summary of research carried EJw = rigidity of beam under warping torque, Nm2 (lb-in.2)
out worldwide and spanning more than four decades, fc = characteristic concrete cylinder compressive
making unification of the symbols and notations used by the strength, MPa (psi)
various researchers and design codes a challenge. In some fc* = concrete effective (plastic) compressive stress,
cases, mostly for graphs and figures, the notation is kept as MPa (psi)
originally published. fck = characteristic compressive strength of concrete,
MPa (psi); fck = fcm 8 MPa (fck = fcm 1200 psi)
2.1Notation fcm = mean compressive strength of concrete, MPa (psi)
a = moment arm for bending, mm (in.) fd = diagonal concrete stress, MPa (psi)
ac = geometric property index fds = diagonal concrete stress corresponding to strain eds,
ao = depth of equivalent rectangular stress block in MPa (psi)
concrete strut of torsional member, mm (in.) f = reinforcement stress in direction, MPa (psi)
A = area of yield surface, mm2 (in.2) fp = prestressing reinforcement stress in the l direction,
Acp = area enclosed by outside perimeter of concrete MPa (psi)
cross section, mm2 (in.2) fy = specified yield strength of longitudinal reinforce-
A = total area of longitudinal reinforcement to resist ment, MPa (psi)
torsion, mm2 (in.2) fp = stress in prestressing reinforcement; fp becomes fp
Ao = gross area enclosed by shear flow path, mm2 (in.2) or ftp when applied to longitudinal and transverse
(noted as Atb in Eq. (7.2.6)) reinforcement, respectively, MPa (psi)
Aoh = area enclosed by centerline of outermost closed fp0.1 = characteristic yield strength of prestressing rein-
transverse torsional reinforcement, mm2 (in.2) forcing strands, MPa (psi); fp0.1 = 0.9fu
Aps = area of prestressing reinforcement in flexural fpc = compressive stress in concrete due to prestress,
tension zone, mm2 (in.2) MPa (psi)
As = area of nonprestressed longitudinal tension rein- fpk = characteristic tensile strength of prestressing rein-
forcement, mm2 (in.2) forcing strands, MPa (psi); fpk = fpu
As = area of longitudinal compression reinforcement, fpo = effective prestress after losses in prestressing rein-
mm2 (in.2) forcement, MPa (psi)
At = area of one leg of a closed stirrup resisting torsion fpu = specified tensile strength of prestressing reinforce-
within spacing s, mm2 (in.2) (noted as Atb in Eq. ment, MPa (psi)
(7.2.6)) fp,ud = design ultimate strength of prestressing reinforcing
b = width of compression face of member, mm (in.) strands, MPa (psi); fp,ud = fpk/gs (gs = 1.15)
bc = width of stirrups, mm (in.) fr = modulus of rupture of concrete, MPa (psi)
B = integral of Tw ft = reinforcement stress in t direction, MPa (psi)
C = cross-sectional constant to define torsional proper- ft = uniaxial tensile strength of concrete, MPa (psi)
ties of a beam ft* = concrete effective (plastic) tensile stress, MPa (psi)
dv = distance between top and bottom longitudinal rein- ftp = prestressing reinforcement stress in t direction,
forcement, mm (in.) MPa (psi)
fty = specified yield strength of transverse reinforcement, To = pure torsional strength of section, Nm (in.-lb)
MPa (psi) Ts = nominal torsional strength provided by reinforce-
fy = specified yield strength of reinforcement, MPa (psi) ment, Nm (in.-lb)
fyd = design yield strength reinforcing steel, MPa (psi); Tu = factored torsional moment at section, Nm (in.-lb)
fyd = fy/gs (gs = 1.15) Tw = warping torsional moment, Nm (in.-lb)
fy = yield strength of the torsional longitudinal rein- Txu = factored balanced torsional strength, Nm (in.-lb)
forcement, MPa (psi) Txub = balanced torsional strength, Nm (in.-lb)
fyv = torsional hoop yield strength reinforcement, MPa T xub = nondimensional balanced torsional strength, Nm
(psi) (in.-lb)
G = shear modulus, MPa (psi) v = shearing stress due to shear, MPa (psi)
h = overall thickness or height of a member, mm (in.) v* = plastic flow rate (Chapter 7)
Ho = horizontal force in radial direction, N (lb) (Chapter 7) vu = ultimate shear stress, MPa (psi)
Ip = polar moment of inertia, mm4 (in.4) V = applied shear force at section, N (lb)
k1 = ratio of average stress to peak stress Vc = nominal shear strength provided by concrete, N (lb)
K = value from Mohr-Coulomb yield criterion Vo = pure shear strength of section, N (lb)
Kf = flexural stiffness of floor beams, Nm2 (lb-in.2) Vu = factored shear force at section, N (lb)
Kts = torsional stiffness of spandrel beam, Nm/rad w = ultimate distributed load on helical stair, N/m (lb/ft)
(in.-lb/rad) (Chapter 7)
= span length of beam, mm (in.) W = external work, N/m (lb/ft)
f = length of flexural beam, mm (in.) x = shorter overall dimension of rectangular part of
q = width of shear flow q along top wall (Fig. 4.2(a) cross section, mm (in.)
and (b)), mm (in.) x1 = distance section centroid and an infinitesimally
m = ratio of effective (plastic) compressive stress to small area of yield surface, mm (in.)
effective (plastic) tensile stress of concrete y = longer overall dimension of rectangular part of
M = applied flexural moment at section, Nm (in.-lb) cross section, mm (in.)
Mo = pure flexural strength of section, Nm (in.-lb) z = distance along axis of beam, mm (in.)
n = integer value a, b = Saint-Venants coefficients for homogeneous
nR = number of redundants torsional section
nV = coefficient describing an under-reinforced, partially a*, b* = rotational angles in beam subjected to torsion
under-reinforced, or completely over-reinforced section (Chapter 7)
N = applied axial load at section, N (lb) a1 = stress block factor given as ratio of fd to fc (Chapter 5)
No = pure axial strength of section, N (lb) b = factor relating effect of longitudinal strain on shear
ph = perimeter of centerline of outermost closed trans- strength of concrete (American Association of State
verse torsional reinforcement, mm (in.) Highway and Transportation Officials (AASHTO)
po = perimeter of outer concrete cross section, mm (in.) LRFD (general message)
(sometimes noted as pcp) b1 = factor relating depth of equivalent rectangular
P = applied concentrated load, N (lb) compressive stress block to neutral axis depth; also,
q = shear flow, N/m (lb/in.) block factor given as ratio of ao to td (Fig. 4.5)
r = ratio of top-to-bottom yield forces of the longitu- g1 = angle along helical stair (in plan) at which maximum
dinal reinforcement torsional moment is assumed to occur
r = size effect constant for computing sN g2 = angle along helical stair (in plan) at which vertical
R = shape parameter used in Ramberg-Osgood moment is assumed to be zero
s = center-to-center spacing of longitudinal and trans- gt = shear strain
verse reinforcements, mm (in.) ed = strain in d direction
sl = center-to-center spacing of longitudinal reinforce- edec = strain in prestressing reinforcement at decompres-
ment, mm (in.) sion of concrete
st = center-to-center spacing of transverse reinforce- eds = maximum strain at concrete strut surface (Fig. 4.3)
ment, mm (in.) eh = strain in hoop direction e
t = wall thickness of hollow section, mm (in.) ey = yield strain in direction
td = thickness of shear flow zone, mm (in.) eo = strain at peak compressive stress fc in concrete
T = applied torsional moment at section, Nm (in.-lb) ep = peak strain in concrete
Tc = nominal torsional strength provided by concrete, er = strain in r direction
Nm (in.-lb) es = strain in nonprestressed reinforcement; es becomes
Tcr = torsional cracking resistance of cross section, Nm e or et when applied to longitudinal or transverse
(in.-lb) reinforcement, respectively
Tf = applied torsional moment, Nm (in.-lb) (Chapter 9) et = strain in t direction
Tmax = maximum torsional moment, Nm (in.-lb) (Chapter 7) ety = yield strain in t direction
Tn = nominal torsional moment strength, Nm (in.-lb)
euk = characteristic total elongation of reinforcing steel at F = angle of twist in torsional beam, radians/m
ultimate load (radians/in.)
ex = longitudinal strain at midheight of concrete section F = second derivative of rotation with respect to beams
z = softening coefficient of concrete strut axis z
h = normalized reinforcement ratio of longitudinal F = third derivative of rotation with respect to beams
reinforcement axis z
hb = balanced normalized reinforcement ratio of longi- Y = bending curvature of concrete strut
tudinal reinforcement w = reinforcement index in direction
ht = normalized reinforcement ratio of transverse steel ws = functional indicator of an index of reinforcement
reinforcement ws = reinforcement ratio index
htb = balanced normalized reinforcement ratio of trans- wt = reinforcement index in t direction
verse steel reinforcement
q = angle between axis of strut, compression diagonal, 2.2Definitions
or compression field and tension chord of the ACI provides a comprehensive list of definitions through
member; also, the angle between -t direction/axis an online resource, ACI Concrete Terminology, http://
and d-r direction/axis, radians terminology.concrete.org.
x = coefficient equal to 1 for rectangular sections and
to p/4 for circular cross sections; x can be taken CHAPTER 3HISTORICAL OVERVIEW OF
as unity for all shapes of cross sections with only TORSION THEORIES AND THEORETICAL MODELS
negligible loss of accuracy for Ao and po 3.1Naviers theory
r = reinforcement ratio in direction A theory for torsion of elastic homogeneous members
rp = prestressing reinforcement ratio in direction was first developed by C. L. Navier (1826) for circular cross
rt = reinforcement ratio in t direction sections. His theory, which was based on equilibrium condi-
rtp = prestressing reinforcement ratio in t direction tions, compatibility conditions, and a linear stress-strain
s = compressive stress acting in combination with relationship like Hookes Law, has guided the development
torsional moment, psi (MPa) of various theories about the behavior of reinforced concrete
s0 = nominal torsional strength according to the current members subjected to torsion after cracking.
code specifications based on plastic limit analysis,
MPa (psi) 3.2Thin-tube theory
sd = principal stress in d direction for concrete struts, Naviers torsion theory for members of circular sections
MPa (psi) was followed by Saint-Venants (1856) solution for rectan-
s = normal stress in longitudinal direction for reinforced gular sections. Saint-Venants torsional constants considered
concrete, MPa (psi) warping of rectangular cross sections. According to Saint-
smax = maximum principal tensile stress, MPa (psi) Venants circulatory shear flow theory, the most efficient
sN = nominal strength of structure, MPa (psi) cross section to resist torsion is a thin tube. Bredt (1896) was
sr = principal stress in r direction for the concrete struts, able to derive simple equations for thin tubes. His thin-tube
MPa (psi) theory states that the shear stress multiplied by wall thick-
st = normal stress in the transverse direction for rein- ness has a constant value around the perimeter and that this
forced concrete, MPa (psi) shear flow is found by dividing the torsion by twice the area
s = strength of plain beams according to elastic analysis enclosed by the shear flow path. Bredts theory has served as
with maximum stress limited by material strength, the basis for modern theories of cracked reinforced concrete
MPa (psi) members subjected to Saint-Venant torsion.
t = shearing stress due to torsion and shear, MPa (psi) 3.2.1 Two- and three-dimensional plane truss models
tmax = maximum shear stress, MPa (psi) The first theoretical models for shear in cracked reinforced
tt = applied shear stress in -t coordinate for reinforced concrete members date back to the turn of the century when
concrete, MPa (psi) Ritter (1899) and Mrsch (1902) formulated the two-dimen-
n = uniform plastic effectiveness factor (Chapter 7) sional (2-D) plane truss model concept, where reinforced
nc = plastic effectiveness factor for compression concrete members were modeled as an assembly of two
(Chapter 7) types of linear elementsconcrete struts and reinforcement
nt = plastic effectiveness factor for tension (Chapter 7) ties. The axis of concrete struts in the model was assumed
j = friction angle to be inclined at 45 degrees to longitudinal members, and
f = strength reduction factor shear strength was assumed to be controlled by the yielding
fc = strength reduction factor for concrete (0.65 for of transverse reinforcement ties. By extending the 2-D
cast-in-place, 0.70 for precast concrete) plane truss model, Rausch (1929) developed a three-dimen-
fp = strength reduction factor for prestressing tendons sional (3-D) space truss model for torsion that consisted of
(0.90) longitudinal and hoop reinforcement-resisting tension and
fs = strength reduction factor for nonprestressed rein- concrete struts-resisting compression. He also assumed that
forcing bars (0.85)
the shear flow path would follow the centerline of the hoop
reinforcement.
3.2.2 Skew-bending and space truss theoriesIn later
years, research on torsion followed two theoretical tracks
skew-bending and space truss. Lessig (1959) first proposed a
skew-bending theory for reinforced concrete members with
two modes of failure, Mode 1 and Mode 2, as explained in
3.3.5. The skew-bending theory used only equilibrium equa-
tions and assumed that all reinforcement yielded before
failure. Lessigs research was followed by the skew-bending
theory of Walsh et al. (1966) and Collins et al. (1968a,b), who
proposed a third failure mode, Mode 3, and used all three
modes to derive nondimensional torsion-bending moment
interaction equations (Walsh et al. 1967), as described in Fig. 3.3.2ACI Committee 438Torsion, Mexico City,
3.3.5. Based on three modes of failure, a nondimensional October 1976: Tom Hsu, Lennart Elfgren, Phil Ferguson,
interaction surface of torsion, shear, and flexure was derived Art McMullen, Emory Kemp, Gordon Fisher, Paul Zia, and
by Elfgren (1972a,b) and Elfgren et al. (1974a,b). Rauschs Michael Collins.
space truss theory for torsion was generalized by Lampert
and Thrlimann (1969, 1971), who showed how the angle relationships) have been the basis for research in torsion
of inclination of the compression diagonals at failure could of reinforced concrete members. (Equation notation in this
be determined from equilibrium if both the hoops and longi- section are provided in Chapter 2.)
tudinal reinforcement were assumed to yield. Lampert and 3.3.2 Twentieth centuryIn the first 60 years of the twen-
Collins (1972) showed that predictions of skew-bending and tieth century, progress in reinforced concrete theories was
space truss theories were in close agreement. made primarily on flexure. Early flexural theories for rein-
3.2.3 Compression field theory (CFT)The truss model forced concrete assumed plane sections remained plane
with linear elements developed by Rausch was replaced in the and stress-strain relationships of concrete and reinforce-
1960s by a new type of truss model with membrane elements ment were linear. Equilibrium conditions for longitudinal
that were subjected to in-plane normal and shearing stresses. stresses were used to determine the location of the neutral
In determining the torsional strength of members where some axis and stresses in concrete and reinforcement caused by
reinforcement does not yield, consider compatibility conditions. the moment. The contribution of concrete tensile stresses
Such conditions were introduced by Baumann (1972) for shear was disregarded if concrete cracking was expected. Later
and by Collins (1973) for torsion. Mitchell and Collins (1974) flexural theories accounted for the nonlinear stress-strain
incorporated compatibility conditions in their CFT, which also response of the concrete and steel reinforcement so that the
relied on equilibrium equations and nonlinear material models complete moment-curvature relationship for a section could
for concrete and reinforcement. Unlike previous models, the be predicted. In terms of shear and torsion research, a signifi-
CFT calculates cracked member torsional behavior up to the cant achievement was made with the development of truss
peak torque. A compatibility condition derived by minimizing models (Ritter 1899; Mrsch 1902; Rausch 1929). Research
the strain energy in the system is used to calculate the angle of in torsion of reinforced concrete underwent significant
inclination in the truss model struts. advances during the last 40 years of the twentieth century.
3.2.4 Softened truss model (STM)In 1985, Hsu and Mo Two theories were developedskew-bending and truss
(1985a,b,c) proposed the STM by softening the concrete models with membrane elements. Skew-bending includes
stress-strain curve. All of the aforementioned models satisfy the theories of Lessig (1959), Yudin (1962), Collins et al.
Naviers theory. Earlier models overestimated test strengths (1968a), Hsu (1968a), and Elfgren (1972a,b). Truss models
(Hsu 1968c), whereas the CFT, which uses spalling of include the theories of Nielsen (1967), Lampert and Thr-
concrete cover, and STM, which uses softening of concrete, limann (1968, 1969), Collins (1973), Mitchell and Collins
have been shown to predict test results accurately (McMullen (1974), Elfgren et al. (1974a,b), Collins and Mitchell (1980),
and El-Degwy 1985). and Hsu and Mo (1985a,b,c). Several researchers involved
in the development of theories for torsion were members of
3.3Historical development of theories for ACI Committee 438 for Torsion, which is now the Joint ACI-
reinforced concrete members subjected to torsion ASCE Committee 445 for Shear and Torsion (Fig. 3.3.2).
3.3.1 GeneralSection 3.1 summarizes the historical Development of these modern truss models was based on
models developed to describe reinforced concrete members the same three principles of mechanics, which, in terms of
subjected to torsion, covering almost two centuries of torsion and shear, include the softened stress-strain relation-
research from 1826 to the early twenty-first century (2007). ship of concrete.
Classical theories include Navier (1826), Saint-Venant 3.3.3 Classical torsion theory for homogeneous members
(1856), Bredt (1896), and Bach (1911). This review shows Navier (1826) derived a theory for torsion of homogeneous
that the three principles of mechanics of materials (equilib- elastic members with circular cross sections. His theory is
rium, compatibility conditions, and materials stress-strain based on the three principles of mechanics of materials:
equilibrium, compatibility conditions, and Hookes Law.

Naviers work also includes the linear theory for flexure. His
book is recognized as the first on the mechanics of mate-
rials The three principles of the mechanics of materials
have become well known as Naviers theory. His theory
defines the torsional rigidity for circular sections as GIp. By
extending the formulas for the polar moment of inertia of
circular sections to square sections, Navier found that the
calculated strength for specimens tested by Duleau (1820)
overestimated measured values by approximately 20 percent.
He acknowledged that the formulae for square members do
not depict as accurately the behavior as those for circular
members. This inconsistency was explained three decades Fig. 3.3.3Torsion of thin tube and lever arm area Ao (Hsu
later by Saint-Venant (1856), who recognized that Naviers 1993 after Bredt 1896).
polar moment of inertia could not reflect the warping defor-
mation of rectangular cross sections. To obtain the correct Navier, Saint-Venant, Bredt, and Bach are applicable to rein-
solution, Saint-Venant developed the semi-inverse method forced concrete beams before cracking. They also laid the
to solve all 15 differential and algebraic equations in the foundation for developing theories to calculate the behavior
theory of elasticity developed by Cauchy (1828). By satis- of cracked reinforced concrete members subjected to torsion.
fying equilibrium, compatibility, and Hookes Law at each 3.3.4 Theories for reinforced concrete under flexure,
differential element of a member, Saint-Venant developed a shear, and torsionReinforced concrete was first developed
solution that considered the warping displacements of rect- in 1867 when Joseph Monier obtained a patent for rein-
angular cross sections. In Saint-Venants rigorous derivation, forcing his concrete flowerpots with wrought iron wires. The
torsional rigidity is defined as GC. The torsional constant C concept of using steel reinforcement to overcome the weak-
is taken as bx3y, where the coefficient b is a function of the ness of concrete in tension was quickly adapted to build-
ratio y/x and varies between 0.141 (y/x = 1) and 0.333 (y/x ings and bridges, making the use of reinforced concrete for
= ). The maximum shear stress tmax occurs on the outside construction a widely accepted application in the last quarter
face of the rectangular section at midpoint of each long side of the nineteenth century. Such growth in applications gave
and is equal to T/ax2y, where a is a coefficient that varies rise to the demand for theories to analyze and design rein-
between 0.208 (y/x = 1) and 0.333 (y/x = ). The relationship forced concrete structures.
between a and b is a = b/k, where 3.3.4.1 Flexure theoryAs reported by Delhumeau
(1999), the first flexure theory to emerge was the linear
8 1 flexure theory developed by Hennebiques firm near the
k = 1
p2 n =1,3, 5,... 2 npy end of the nineteenth century. This theory served as the
n cosh basis for the allowable stress design method addressed in
2x
the first ACI code [National Association of Cement Users
(NACU) 1910]. Theories for nonlinear flexure occupied the
According to Saint-Venants circulatory shear flow pattern,
attention of researchers until 1963 when strength design
the maximum shear stresses occur at the outer periphery of
was incorporated in ACI 318 (ACI 318-63). Both linear and
a cross section, and the most efficient cross section to resist
nonlinear theories for flexure satisfy the three principles of
torsion is a thin tube, as shown in Fig. 3.3.3.
the mechanics of materials: equilibrium of parallel coplanar
Bredt (1896) was able to derive a simple equilibrium
forces, Bernoullis linear strain compatibility, and the consti-
equation for thin tubes by assuming the entire tube to be
tutive laws of materials. A linear stress-strain curve of rein-
uniformly and fully stressed
forcement and concrete is used for the linear flexure theory
and a nonlinear stress-strain curve of concrete is used for the
T
q= (3.3.3) nonlinear flexure theory.
2 Ao 3.3.4.2 Torsion and shear theoryFollowing the adop-

tion of flexural strength design in ACI 318-63, the attention
The area Ao is formed by sweeping the lever arm (symbol of researchers turned to more complex problems of torsion
a in Fig. 3.3.3) around the axis of twist, a term later called and shear in beams. Computer analysis of structures was
the lever arm area (Hsu 1988). The torsional constant C for becoming available, making it feasible for design engineers
thin tubes with constant thickness was also simplified to C = to compute the magnitude of torsions in their buildings. In
4Ao2t/po. Bachs formula (1911) is a simplification of Saint- addition, concrete box-girder bridges, which were becoming
Venants theory for thin-walled open sections, such as T, L, a competitive bridge type, needed to be designed for torsion.
and I sections. Because the coefficient b for each rectangular Shear is essentially a two-dimensional problem, requiring
component of such sections can be approximated as 1/3, the an understanding of the interaction of two principal stresses
torsional constant for the entire section can be taken as the and strains in a membrane element. Torsion is compli-
sum (S) of the components (C = 1/3Sx3y). The theories of cated because it is a three-dimensional problem involving
Fig. 3.3.4.2Number of published papers on torsion from 1900 to 2007 (plot prepared by
A. Belarbi).
the shear problem of membrane elements in a tube and the as a truss with two types of linear elements: struts made
warping of tube walls that cause flexure in concrete struts. out of concrete and ties made out of steel reinforcement.
Figure 3.3.4.2 shows that the number of globally published The Ritter and Mrsch model represents the struts and ties
papers on torsion began to surge around 1960 and peaked as lines without cross-sectional dimensions, where forces
around 1970. The principles of equilibrium, compatibility satisfy equilibrium at points of intersectiona model with
conditions, and materials stress-strain relationships that the advantage of conceptual clarity. Extending the 2-D plane
were needed to solve torsion problems in reinforced concrete truss model to a 3-D space truss model, Rausch (1929) devel-
members were primarily developed between 1960 and 1985 oped a theory for torsion of reinforced concrete. Rauschs
(Lampert and Thrlimann 1971; Lampert and Collins 1972; space truss model, as shown in Fig. 3.3.5a, is made up of
Elfgren et al. 1974a,b; Collins and Mitchell 1980; Hsu and 45-degree diagonal concrete struts, longitudinal reinforcing
Mo 1983). Theories and tests produced before 1980 are bars, and hoop reinforcing bars connected at the joints by
summarized in detail by Hsu (1984). hinges. Torsional moment is carried by the concrete struts
By 1985, researchers solved the basic problems of reinforced in axial compression (dotted lines), and by the straight rein-
concrete design by applying Naviers theory. Further research forcing bars in axial tension (solid lines) in the longitudinal
was necessary to refine the constitutive laws of materials for (horizontal) and lateral (hoop) directions. Equilibrium of
torsion and shear. The experimental work needed to generate the joints in the longitudinal, lateral, and radial directions
new advancements is tedious and requires highly sophisticated requires that the forces in the longitudinal bars (X), in the
testing equipment. Only two universities in North America hoop bars (Q), and in the inclined struts (D) should be evenly
the University of Toronto and University of Houstonare distributed among all cells and joints. To satisfy equilibrium,
capable of studying the behavior of reinforced concrete shell the relationship between these forces should be X = Q =
and panel elements subjected to in-plane shear and normal D/2. As shown in Fig. 3.3.5a, the series of hoop forces Q
stresses. The study of softened concrete in shear elements, at the joints constitute a shear flow q = Q/s. Using Bredts
which has been the subject of extensive research worldwide lever arm area concept, T can be related to q (or Q/s) by
in the last three decades, continues to be a major research 2Ao, as expressed by Bredts equation (Eq. (3.3.3)). The term
topic. The need for larger and more complex specimens has Ao refers to the area enclosed by a series of straight lines
increased ongoing work at both universities. The high cost of connecting joints of the cross section.
experimental research needed for new developments in torsion Assuming that ultimate torque is reached when the forces
imposes a limiting constraint on new research. For example, in the transverse reinforcement reach the yield stress, then q
studying the behavior of full-scale girders with open sections = Q/s = Atfty/st and Eq. (3.3.3) becomes
that involve Saint-Venant and warping torsion is expensive. The
future of torsion research is largely tied to available equipment At fty
or to the combined efforts of many institutions, or both. Tn = 2 Aoh (3.3.5)
st
3.3.5 Space truss model using struts and tiesThe first
theory for shear design of reinforced concrete was developed
at the turn of the twentieth century when Ritter (1899) and Although the space truss model has the advantage
Mrsch (1902) formulated the concept of plane trusses with of conceptual clarity in terms of simple assemblage of
struts and ties. They modeled a reinforced concrete member compression struts and tension ties, Rauschs equation (Eq.
Fig. 3.3.5aRauschs space truss model with struts and ties

(Hsu 1984 after Rausch 1929).
(3.3.5)) was found to greatly overestimate the experimen-

tally measured torsional strengths of reinforced concrete
members. As a result, empirical formulas proposed to
correct limitations of Rauschs equation were incorporated
in the design codes of several countries until the 1990s.
Figure 3.3.5b compares Tn with AohAtfty/st for a series of
Fig. 3.3.5bExperimental Tn versus Aoh(Atfty/st) curve
PCA tests (Hsu 1968b,c). The straight solid line represents
compared with Rauschs formula and 1971 ACI code (Hsu
calculated strengths using Eq. (3.3.5) (that is, at = 2.0) and
1984).
the dashed line represents calculated strengths using the
torsion provisions of ACI 318-71. The lever arm area Ao is lated with Saint-Venants theory (Tn = ax2yft), assuming the
defined by the centerline of closed stirrups and denoted as principal tensile stress reaches a value of ft. To find a reason
Aoh. In view of the unconservative nature of the values calcu- for this discrepancy, Hsu (1968a) used high-speed photog-
lated with Eq. (3.3.5) (Fig. 3.3.5b), in particular when Tn is raphy to record failure process and observed that plain
large, an empirical formula was proposed for ACI 318-71. concrete members failed abruptly in a skew-bending mode.
This formula is simply a modification of Eq. (3.3.5), which Based on the skew-bending failure, the torsional strength
reduced the slope of 2, shown in Fig. 3.3.5b, down to 1.2, was derived as Tn = (1/3)x2y(0.85fr). As a result, the constant
and added a concrete contribution Tc, which represents the 1/3 is an approximation of Saint-Venants coefficient a
vertical intercept of the experimental curve. The background when the x/y ratio becomes large. This coefficient was used
of concrete contribution Tc, which is included in early ACI as the calibration for the torsional strength of beams without
Codes (1971 through 1995) and later dropped, is provided reinforcement, and for the concrete contribution of the
in Hsu (1997). From the experimental tests shown in Fig. torsional strength of beams with web reinforcement (Hsu
3.3.5b, Lampert and Thrlimann deduced that Ao should be 1968b,c). Formulas based on the parameter (1/3)x2yft were
less than Aoh. They recommended that Ao be taken as the area incorporated into the ACI Code from 1971 to 1995.
enclosed by a line joining the centers of the longitudinal bars 3.3.6.2 Beams with web reinforcementSkew-bending
in the corners of the hoops. With this change in the definition theory was also applied to concrete beams reinforced with
of Ao, the agreement between predictions and experimental both longitudinal and transverse reinforcement subjected to
results was considerably improved. This new definition of torsion, shear, and flexure. Such beams could fail in three
Ao was subsequently incorporated into the CEB-FIP code modes (Fig. 3.3.6.2). The first mode has a compression zone
(1978). A different approach, to reconcile Rauschs space along the top face and failure is caused by yielding of the
truss model with test results, was achieved in 1983 with bottom longitudinal reinforcement and transverse reinforce-
development of the STM (Hsu and Mo 1983) for torsion, ment along the three remaining faces. The second mode has
using a softened stress-strain relationship of concrete devel- a compression zone along a side face and failure is caused
oped by Vecchio and Collins (1981). A more recent STM is by the yielding of the longitudinal reinforcement on the
provided in 5.6. opposite side face, and in the transverse reinforcement along
3.3.6 Skew-bending theory for torsion the three faces in tension. The third mode has a compres-
3.3.6.1 Beams without web reinforcementWhen beams sion zone along the bottom face where failure is caused by
without transverse reinforcement are subjected to torsion, yielding along the top longitudinal reinforcement and trans-
they fail quickly and abruptly at cracking. Their torsional verse reinforcement along the three remaining faces.
behavior is similar to that of plain concrete beams and, there- Lessig (1959) was the first to propose a skew-bending
fore, can be predicted by Saint-Venants theory. The experi- theory in connection with Mode 1 and Mode 2 failures.
mental torsional strength of a plain concrete beam, however, Equilibrium equations were established by taking moment
was found to be greater than the theoretical strength calcu- equilibrium about an axis AB (Fig. 3.3.6.2) parallel to the
Fig. 3.3.6.2Three failure modes in skew-bending theories (Lessig 1959; Yudin 1962;
Collins et al. 1968a; Elfgren et al. 1974a,b).
skew compression zone and force equilibrium about the axis combined torsion and flexure, Lampert and Collins (1972)
perpendicular to the center of the compression zone. The showed that both the skew-bending theory and the space
crack angles defining the skew failure surfaces were deter- truss theory predict the same parabolic shape for the interac-
mined by minimizing the torsional resistance. The resulting tion curve. Elfgren et al. (1974a,b) then derived a nondimen-
equations were complicated. sional interaction surface for combined torsion, shear, and
Yudin (1962) took a different approach by deriving three flexure based on the three modes of skew-bending failure. If
equilibrium equations: moment equilibrium about the longi- the inclination angle of cracks in each wall is adjusted so that
tudinal and transverse axes through the compression zone both longitudinal and transverse reinforcement can yield, the
center, and force equilibrium about the axis perpendicular to interaction surface becomes identical to that derived from
the compression zone center. Assuming all the crack angles the equilibrium (plasticity) truss model (Elfgren 1972a,b),
to be 45 degrees, he showed that his equation for ultimate as explained in 5.5.
torque was the same as Rauschs. Lessig and Yudins research 3.3.7 Truss models with membrane elements
was followed by the skew-bending theory of Walsh et al. 3.3.7.1 Equilibrium (plasticity) truss model for torsion
(1966, 1967) and Collins et al. (1968a,b), who also identified Rauschs space truss model with struts and ties (1929) was
Mode 3 failure. The three modes of failure1, 2, and 3are replaced by a new and more realistic type of truss model
governed by the dominant effect of positive flexure moment, in the 1960s. As shown in Fig. 3.3.7.1a, the new model is
torsion shear, and negative flexure moment, respectively. For made up of membrane elements treated as trusses made up
Fig. 3.3.7.1aMembrane element in shear (Hsu 1993).
of struts and ties after cracking. Three equilibrium equations = Atfty/st), Lampert and Thrlimanns Eq. (3.3.7.1e) looks
for the elements were first derived by Nielsen (1967) and identical to Rauschs Eq. (3.3.5), except the term Ao is
Lampert and Thrlimann (1968). defined differently. The former is based on the centerline
of longitudinal reinforcement and the latter is based on
s = sdcos2q + srsin2q + r f (3.3.7.1a) the centerline of hoop reinforcement. Equation (3.3.7.1e)
was first used in the European CEB-FIP code in 1978 and
s t = s d sin 2 q + s r cos2 q + rt ft (3.3.7.1b) ACI 318-95. By combining Eq. (3.3.7.1e) with equations

for shear and flexure, Elfgren (1972a,b) and Elfgren et al.
tt = (sd + sr)sinqcosq (3.3.7.1c) (1974a) established a nondimensional interaction surface
as shown in Fig. 3.3.7.1b, based on the equilibrium (plas-
Because Eq. (3.3.7.1a) through (3.3.7.1c) describe the ticity) truss model as explained in 6.5. Elfgren also showed
transformation of stresses, they form a system of stresses the interaction surface derived from the truss model is iden-
that should satisfy Mohrs stress circle, a type of equilib- tical to the skew-bending theory. Actual testing has shown,
rium condition called Mohrs circular stress equilibrium. however, that Elfgrens interaction surface overestimates the
The longitudinal and transverse reinforcement stresses are experimental results in the region adjacent to torsion and
assumed to yield at failure (f = fy, ft = fty), and the concrete shear because softening concrete was not considered.
tensile stress is neglected (sr = 0). Equations (3.3.7.1a) 3.3.7.2 Compatibility truss model for torsion
through (3.3.7.1c) were simplified to yield the shear stress tt 3.3.7.2.1 GeneralAccording to Navier, the strain
in the case of pure shear without normal stresses (s = st = 0) compatibility (or geometric) condition should be considered
if the angle of twist is to be related to the torsional moment.
Assuming the membrane element to behave as a truss as
t t = r f y rt fty (3.3.7.1d)
shown in Fig. 3.3.7.1a(b), three strain compatibility equa-
tions have been derived by Baumann (1972) and Mitchell
A reinforced concrete tube subjected to torsion can be and Collins (1974).
visualized as an assembly of shear elements similar to the
one shown in Fig. 3.3.3. e = e d cos2 q + e r sin 2 q (3.7.2.1a)
The torsional moment at yielding, Tn, can then be derived
by substituting the shear stress tt (tt = q/t) into Eq. (3.3.3)
and substituting r = A/st and rt = At/stt e t = e d sin 2 q + e r cos2 q (3.7.2.1b)

A f y At fty g t
Tu = 2 Ao (3.3.7.1e) = ( e d + e r )sin q cos q (3.7.2.1c)
s st 2

Because Eq. (3.7.2.1a) through (3.7.2.1c) describe the
Equation (3.3.7.1e) was first derived by Lampert and
transformation and distribution of strains, they should satisfy
Thrlimann (1968, 1971). In the case of equal yield force
Mohrs strain circle. This type of compatibility condition is
in both longitudinal and transverse reinforcement (Afy/s
therefore called Mohrs circular strain compatibility.
Collins and Chockalingam (1979) and Vecchio and Collins

(1981, 1982, 1986) showed that the concrete softening coef-
ficient in the principal compressive direction is related to the
principal tensile strain, not the principal tensile stress. Soft-
ening was further quantified experimentally by researchers
at the University of Houston (Belarbi and Hsu 1994, 1995;
Pang and Hsu 1995; Hsu et al. 1995a,b; Zhang and Hsu
1998), who confirmed the importance of principal tensile
strain and identified concrete strength as an additional
essential variable. An STM (Hsu and Mo 1985a,b,c; Hsu
1993) was developed based on a softened concrete stress-
strain curve. Details of this theory are provided in 4.6. In the
STM (Hsu and Mo 1985a,b,c), the softened concrete stress-
strain curve is used in the flexural compression zone of the
concrete struts to determine the neutral axis position. The
Fig. 3.3.7.1bNondimensional interaction surface for
distance from the neutral axis to the extreme compression
torsion, shear, and flexure (Hsu 1984). (The numbering of
fiber is known as thickness of shear flow zone, td; the lower
the failure modes is different than in Fig. 3.3.6.2.)
the softening coefficient, the greater the thickness td. Because
In addition, there are two compatibility equations for a the softening coefficient varies from approximately 1/4 to
reinforced concrete tube. One relates the angle of twist, F, 1/2, the thickness td, obtained from the softened stress-strain
to the shear strain gt of the membrane element curve, is expected to be approximately two to four times the
thickness from the non-softened stress-strain curve. Based
po on Vecchio and Collinss softened stress-strain curve (1981),
F= g t (3.7.2.1d) Hsu and Mo (1985a,b,c) studied the torsional behavior of
2 Ao
cracked reinforced concrete members. They found that
Rauschs Eq. (3.3.5) is valid if the thickness td and the lever
The other describes the flexure of diagonal concrete struts. arm area Ao are calculated by a truss model that incorporates
The curvature of concrete struts Y at an angle q is related to a softened stress-strain curve. Hsu and Mos STM satisfies
the angle of twist F by the Mohrs stress circle, the Mohrs strain circle, and the
softened stress-strain curve of concrete. It was further shown
Y = Fsin2q (3.7.2.1e) to provide accurate estimates of experimental results avail-
able at the time (Hsu and Mo 1985a,b,c), including the PCA
3.3.7.2.2 Compression field theory (CFT)In a major tests in Fig. 3.3.5b. When accounting for the modifications
development of the truss model for torsion, Mitchell and proposed by Hsu and Mo, the equation by Rausch becomes
Collins (1974) established the CFT to calculate the complete a curve that starts at the origin, closely tracing the PCA test
torque-twist response of reinforced concrete members, irre- points. This occurs because Ao decreases with increasing Tn
spective of sectional shape or amount of reinforcement. and an increasing percentage of reinforcement. The depen-
The CFT combines four equilibrium equations [Eq. (3.3.3), dency of td and Ao on Tn is expressed by simple equations
(3.3.7.1a), (3.3.7.1b), and (3.3.7.1c)], five compatibility (Hsu 1990, 1993)
equations [Eq. (3.7.2.1a) through (3.7.2.1e)], and constitu-
tive equations of concrete and reinforcement. Concrete diag- 4Tn
onals thickness td and the area enclosed by the shear flow, td = (3.3.7.2.3a)
Acp fc
Ao, are determined as described in 5.5. The assumption that
the diagonal compressive strains are reduced linearly with
depth below the surface was experimentally verified (Collins t d po t d pcp 2Tn pcp
Ao = Acp Acp = Acp (3.3.7.2.3b)
1973). Because the concrete cover is assumed to have spalled 2 2 Acp fc
before the section reaches the maximum strength, this theory
is also known as the spalling model. Details of the CFT are
provided in 5.5. As indicated by Eq. (3.3.7.2.3b), the lever arm area Ao
3.3.7.2.3 Softened truss modelA reinforced concrete is decreased by increases in the thickness td. Because the
membrane element subjected to shear is a two-dimensional torsional strength of a reinforced concrete member is directly
problem; the shear stress can be resolved into a principal proportional to Ao, the strong effect of the softening coef-
tension stress and a principal compressive stress in the ficient on the torsional strength is evident, explaining why
45-degree direction. Robinson and Demorieux (1972) found Rauschs equation for Tn (Eq. (3.3.5)) overestimated the test
that when a membrane element is subjected to a biaxial results, as shown in Fig. 3.3.5b. Because the shear flow zone
stress condition, the principal compressive stress is reduced thickness td occupies such a crucial position in the STM,
(or softened) by the principal tensile stress in the perpen- torsion research is now focused on a better understanding of
dicular direction. Performing biaxial tests of shear elements, the shear flow zone (Alkhrdaji and Belarbi 2003).
3.3.8 Derivation of ACI 318: torsion design equationsAs torsion proved to be much more difficult because they were,
discussed previously, ACI equations for designing members respectively, two-dimensional and three-dimensional prob-
under torsion are based on the space truss model theory. lems. From a historical perspective, the development of the
There are two basic design equations: one for torsional current torsional theories spanned over 40 years. Equilib-
hoop reinforcement, and another for torsional longitudinal rium equations were developed in the 1960s, compatibility
reinforcement. conditions were added in the 1970s, and the formulation and
From the truss model (Fig. 3.3.7.1a(b)), equilibrium of the expressions of softening coefficients were introduced in the
top face of the element relates the shear flow q to the force 1980s and 90s.
in the torsional hoop reinforcement per unit length (Atft/s) Since 1960, the bulk of work has focused on beams
through the equation subjected to monotonic torsional loading. Early tests on
members subjected to reverse cyclic loading were conducted
q = At ft cotq (3.3.8a) by Collins and Chockalingam (1979) on reinforced concrete
s beams subjected to pure torsion and by Stevens et al. (1991)
on membrane elements subjected to reversed cyclic shear up
Substituting Eq. (3.3.8a) into Bredts Eq. (3.3.3) and to reinforcement yielding. Significant recent advances have
assuming yielding of the hoop reinforcement renders the ACI applied rational models such as the STM and the modified
318 requirement for transverse reinforcement for torsion. CFT to shear elements subjected to cyclic loading. Vecchio
(1999) and Palermo and Vecchio (2003, 2004) applied the
MCFT for cyclic loading up to yielding of the reinforcement.
At Tu
= (3.3.8b) Mansour et al. (2001), Zhu et al. (2001), Zhu and Hsu (2002),
s f2 Ao f yv cot q and Mansour and Hsu (2005a,b) studied cyclic torsion and

shear, including the unloading and reloading branches of the
Similarly, from the equilibrium of the left face of the softened stress-strain curves (hysteretic loop) and the effect
element, shown in Fig. 3.3.7.1a(b), the shear flow q can be of Poissons ratio in a membrane element. The latter was
related to the force in the torsional longitudinal reinforce- crucial in determining the descending branch of the softened
ment per unit length (Af/s) through the equation stress-strain curve.
Belarbi and Greene tested several full-scale box girders
A f under cyclic torsion and combined torsion and shear
q= tan q (3.3.8c) (Belarbi and Greene 2004; Greene and Belarbi 2008). The
s
cyclic shear tests of membrane elements (Mansour and
Hsu 2005a,b) permitted the establishment of cyclic stress-
The term Al is defined in ACI 318 as the total area of strain curves of concrete and reinforcement. These cyclic
torsional longitudinal reinforcement in the cross section, constitutive laws laid the foundation for the Cyclic Soft-
assuming that po = ph, where ph is the perimeter of the ened Membrane Model (Mansour and Hsu 2005a,b), which
centerline of the outermost hoop bars. Equation (3.3.8c) then could be used for the design of seismic-resistant reinforced
becomes concrete structures (Hsu et al. 2006).
q=
A f
tan q (3.3.8d) CHAPTER 4BEHAVIOR OF MEMBERS
ph SUBJECTED TO PURE TORSION

4.1General
The torsional strength of concrete beams depends on many
An expression for the amount of torsional longitudinal
factors, such as the amount and distribution of transverse
reinforcement in ACI 318 is obtained by equating the shear
and longitudinal reinforcement, concrete strength, cross
flow q of Eq. (3.3.8d) and (3.3.8a)
section shape, and, for rectangular cross sections, aspect
ratio (depth-to-width ratio). Concrete strength and amount
A f yv
A = t ph cot 2 q (3.3.8e) of reinforcement can change the failure characteristics. The
s f y failure mechanism varies from ductile to sudden and brittle,

depending on the longitudinal and transverse reinforce-
3.4Concluding remarks ment volumetric ratios. Chapter 4 details the behavior of
This historical review identifies research studies that have plain concrete, reinforced concrete, and prestressed concrete
provided the foundation for the current understanding of under pure torsion. The behavior of high-strength concrete
torsion in reinforced concrete members. It took almost the members is also discussed. The effects of other stress resul-
entire twentieth century for researchers to successfully apply tants, such as shear and flexure, are discussed in Chapter 6.
Naviers theory to cracked reinforced concrete. The applica-
tion of Naviers theory to the flexure theory of reinforced 4.2Plain concrete
concrete was relatively easy because of the one-dimensional Saint-Venants elastic theory accurately describes the
condition of equilibrium and compatibility, combined with torsional behavior of plain concrete members at low torque
the non-softened constitutive laws of materials. Shear and before cracking. This theory assumes that the cross section
the inclined face due to flexure reaches a reduced modulus

of rupture of concrete (Hsu 1968b). The formula derived is
Tu = (1/3)x2y(0.85fr). The torsional stress factor is a constant
1/3, which lies between those of the elastic theory and the
plastic theory, as shown in Fig. 4.2. The skew-bending
theory accurately describes the failure of concrete elements
with rectangular cross sections, but it should be considered
as an assembly of component rectangles when applied to
flanged sections (Hsu 1968b, 1984). For thin-walled tubes,
the relationship between torsional shear stress t (= q/td) and
the torsional moment T can be found from Bredts Eq. (4.2).
Collins (1973) suggested that if a solid section was replaced
by a tube with the same exterior dimensions but with a wall
thickness of 0.75Ao/po, the cracking torques could still be
determined from Eq. (4.2). Figure 4.2 compares the torsional
Fig. 4.2Torsion stress factors for solid rectangular section shear stress t for solid rectangular sections predicted by
(Collins 1973). elastic theory, plastic theory, skew-bending, and thin-walled
tube model. By further assuming that the wall thickness td =
shape remains unchanged after twisting, the angle of twist 0.75Acp/pcp and that Ao = (2/3)Acp, the torsional stress t can
per unit length is constant along the element length, warping be related to T by the parameter (Acp2/pcp)
deformation perpendicular to the cross section is identical
throughout the member length, and failure occurs when the T T
t= = (4.2)
maximum principal stress equals the direct concrete tensile 2 Ao t d ( Acp2 / pcp )

strength. Member shear stresses resist applied torsional
moment. For solid circular shapes, the shear stresses are
Equation (4.2) was embraced by CSA-A23.3-77 and later
zero at the member center, increasing linearly to a maximum
adopted by ACI 318-95. For example, the ACI code uses it to
at the outside surface. In solid rectangular shapes, the shear
address compatibility torsion of spandrel beams. When the
stresses increase from zero at the center to a maximum at
nominal torsional stress tn needed to form a torsional plastic
the center of the longer faces, but they decrease along the
hinge was found to be 0.33fc MPa (4fc psi) (Hsu and
edge and are zero again at the corners. When considering a
Burton 1974; Hsu and Hwang 1977), the design torsional
small element in the member subjected to pure torsion, only
shear stresses develop and, therefore, the principal tension moment Tn of the spandrel beam was 0.33 fc (Acp2/pcp)
and compression stresses equal the shear stress. Applying MPa (equivalent to 4 fc (Acp2/pcp) psi).
Saint-Venants theory to concrete, the member would fail in In addition to the aforementioned theories, there are
a brittle manner once the maximum shear stress equals the approaches to the torsional behavior of concrete elements
concrete tensile cracking strength. that are based on truss models. Because these models are
After cracking, the elastic theory does not accurately developed on modeling the flow of forces in a cracked rein-
describe the concrete behavior because various nonlineari- forced concrete element, as stated by Karayannis (1995),
ties exist, such as the stress-strain behavior in compression, these approaches are more suitable for the description of the
post-peak softening in tension, and tensile cracking. When post-elastic behavior of reinforced concrete elements, with
the elastic theory is used, it considerably underestimates the rectangular cross sections reinforced in the longitudinal and
torque at failure of a plain concrete member because concrete transverse directions.
exhibits a complex structural response (Hsu 1984; Karay- Karayannis and Chalioris (2000) confirmed that a plain
annis 2000). Other theories, such as the plastic theory and concrete member subjected to pure torsion fails in tension
skew-bending theory, were developed to estimate the failure due to shear. The behavior of concrete elements in torsion
strength of plain concrete members. The plastic theory is not is therefore primarily governed by the material tensile
theoretically satisfactory because it assumes that concrete response, particularly its tensile cracking characteristics.
develops plasticity. As in the elastic theory, failure is Tension cracking makes the plain concrete behavior in
assumed to occur when the maximum principal tensile stress torsion nonlinear. Before the peak torque is reached, some
of concrete reaches the tensile strength. Concrete, however, microcracking and softening regions are developed.
does not develop significant plastic behavior in tension and The cracking is a localized phenomenon, and its boundary
the torsional failure of plain concrete is brittle (Hsu 1984). is the strain softening region called the fracture process
Unlike the elastic theory, the plastic theory tends to overesti- zone. Baant and Oh (1983) regarded this tension-softening
mate the failure strength. The skew-bending theory assumes region as being caused not only by microcracking but also by
that plain concrete members subjected to torsion fail by bond rupture; therefore, the fracture process zone has to be
bending about an axis parallel to the wider cross-sectional assumed wider than the region of visible microcracks. The
face and inclined at an angle of 45 degrees with respect to the zone width can be considered as a material property that,
longitudinal axis. Failure occurs when the tensile stress on except for the influence of the stress distribution, depends
on the nature and size of the aggregate. The role of the size
effect on the fracture process zone width should also be
considered. Between the microcracks of this zone, there are
less damaged or even elastic parts; this can be considered a
result of the strong tendency for eventual localization of the
damage and the simultaneous recovery of the undamaged or
less-damaged parts between the final cracks.
4.3Reinforced concrete
Before the concrete cracks, a member under pure torsion
is assumed to behave elastically with a nearly linear torque-
twist curve. After cracking, the resistance mechanism
changes and stirrups and longitudinal reinforcement carry
most of the tensile load whereas the concrete carries the
compression. The longitudinal reinforcement tensile strains Fig. 4.3aDiagonal cracks in members under pure torsion
cause the member to lengthen. Increased torque causes a (Mitchell and Collins 1974).
nearly linear increase in the beam length. A longitudinal
restraint acts as a compressive prestress force and increases
the torsional strength. the surface, crack width, beam strength and stiffness, and
Members reinforced for torsion typically have closed stir- failure mode. The reinforcement ratios have a strong influ-
rups and at least one longitudinal bar in each stirrup corner, ence on the inclined crack angles and the size of the cracks.
as originally required in ACI 318-63. Members with only The crack width is related to the principal tensile strain.
longitudinal reinforcement behave similarly to a plain When the volumetric ratios rl and rt are equal, q is approxi-
concrete member in regards to torsional rigidity and torsional mately 45 degrees and the cracks have the smallest width
strength, and they fail after cracking in a brittle manner. because the strains in the longitudinal and transverse direc-
An uncracked beam loaded in pure torsion has only shear tions are also equal. For beams with rt larger than rl, the
stresses on a plane perpendicular to the longitudinal axis. angle is greater than 45 degrees relative to the horizontal and
The associated principal tensile stress causes inclined cracks the strain larger in the longitudinal direction, which causes
on the beam surface in the principal compressive stress wider cracks. Similarly, in beams with rt less than rl, the
direction. These cracks spiral around the beam. The tensile angle is less than 45 degrees relative to the horizontal, the
cracks are due to strain in the principal tensile stress direc- transverse strain is larger, and the cracks are also wider than
tion, and they affect the concrete acting in compression. The when q is 45 degrees (Mitchell and Collins 1974; Hsu 1993).
tensile strains have the effect of weakening the concretes The volumetric ratios of longitudinal and transverse rein-
ability to withstand compression in the perpendicular direc- forcement to concrete affect the failure mechanism. Large
tion. This is known as softening of the concrete compres- rl and rt result in reinforcement stresses less than yield
sive strength. Tests performed by Vecchio and Collins (1982, when the concrete reaches its ultimate compressive strain at
1986) identified the principal tensile strains as the primary the surface. In this condition, the concrete is crushed and
variable influencing tension softening and quantified them. the beam experiences sudden and brittle failure. Beams of
Continued work by Belarbi and Hsu (1994, 1995), Pang and this type are called over-reinforced. Figure 4.3b shows an
Hsu (1995), Hsu and Zhang (1997), Hsu and Zhu (2002), example of beam failure boundaries (Leu and Lee 2000).
and Hsu and Mo (2010) has improved the understanding The condition known as balanced failure point is met when
of compression softening. Other studies, such as those of concrete reaches its ultimate strain at the point where both
Vecchio and Collins (1993) and Koutchoukali and Belarbi transverse and longitudinal reinforcement begin to yield
(2001), have also been reported based on further calibration simultaneously. Beams with rl and rt less than the rein-
of their original models. After cracking, the longitudinal and forcement ratios at the balanced condition are called under-
transverse reinforcement are in tension, and the concrete reinforced (Region I). Under-reinforced beams are capable
struts between the diagonal cracks are in compression, as of continued twist as the reinforcement yields, producing
shown in Fig. 4.3a. The faces of the beam are warped as a ductile failure. Over-reinforced beams are in Region IV
the beam is twisted, causing bending stresses, in addition where the failure is brittle. For lower ratios of either r or rt,
to compressive stresses, in the concrete struts (Mitchell the longitudinal or transverse reinforcement yields before the
and Collins 1974). The concrete between inclined cracks concrete is crushed. Beams of this type are called partially
is capable of acting in tension and will increase the overall under-reinforced (Regions II and III).
beam torsional stiffness. This effect, known as tension stiff- The volumetric ratios of longitudinal and transverse rein-
ening, is most predominant immediately after the first cracks forcement have a strong influence on the beams postcracking
appear and decreases with increased torsion. torsional stiffness and strength as shown in Fig. 4.3c. For
The volumetric ratios rl and rt have a significant effect on under-reinforced beams with equal volumetric ratios r
the behavior of reinforced concrete beams subject to pure and rt, an increase in the total reinforcement increases the
torsion. The ratios affect the angle of the inclined cracks on torsional rigidity, which is the slope of the torque-twist curve
Fig. 4.3dEffects of concrete cover on the torque-twist

behavior (Mitchell and Collins 1974). (Note: 1 in.-kip =
0.113 kNm.)
Fig. 4.3bFailure boundaries for fc = 4.0 ksi (27.5 MPa) shows that the inner core of solid members is not effective
and fy = 45.2 ksi (311.9) MPa (Leu and Lee 2000). after cracking. The cracking strength of hollow members is
less than that of solid members because, before cracking, the
shearing stresses are acting over the entire section including
the inner core. After cracking, however, the outer perimeter
of the cross section resists the larger part of the torque with
minimum contribution from the inside core; torsional strength
and general behavior are similar in solid and hollow sections
after cracking as long as wall thickness is large enough to
accommodate the shear flow zone (Hsu 1968a). Leonhardt
and Schelling (1974) confirmed these findings when they
performed an extensive test series on beams in pure torsion
where they systematically varied the cross section type,
including rectangular with different aspect ratios, circular,
and hollow box, T-shaped, and inverted L-shaped beams.
They made two observations: 1) the importance of detailing
by properly anchoring the stirrups and using small stirrup
spacings; and 2) the importance of softening concrete in the
inclined struts.
High concrete shearing stresses outside of the stirrups
create potential for concrete spalling under high torques.
To satisfy equilibrium, the concrete compressive stress
that resists torsion after cracking changes directions at the
Fig. 4.3cTorque-twist curves of beams with various outside corners of the beam. The resultant tensile stress in
percentages of reinforcement (Hsu 1968a). (Note: 1 in.-kip the concrete causes the concrete cover to spall off when the
= 0.113 kNm; 1 in. = 0.0254 m.) tensile stress is high enough. Typically the spall occurs at the
stirrups. The resulting section is smaller and has a reduced
after cracking. It also increases the torsional yield strength, torsional strength (Mitchell and Collins 1974). Mitchell and
which is the amount of torque the beam can sustain at the Collins (1974) tested two beams, PT5 and PT6, to inves-
onset of yielding. As the total amount of reinforcement tigate the effects of spalling. The beams had similar rein-
increases, however, the post-yield twist decreases. There- forcing cages and concrete strength, but PT5 had a cover of
fore, an increase in the total amount of reinforcement results 1.5 mm (1/16 in.) and PT6 had a cover of 40 mm (1-9/16 in.).
in a stiffer and stronger beam, but at the cost of decreased The results of the tests showed that, although the concrete
ductility. The minimum amount of reinforcing steel is the outside the hoop reinforcement had a significant effect on
amount causing the reinforcement to yield at the same the cracking torque, it had little effect on the peak torque
torque that causes cracking. Where this occurs, the total rein- after significant spalling occurred (Fig. 4.3d).
forcement ratio, including both longitudinal and transverse In tests performed by Rahal and Collins (1995b), speci-
reinforcement, is approximately 0.01 for normal-strength mens with small covers did not experience spalling until
concrete (Hsu 1984). A beam with less reinforcement than after the torsional strength had been reached. Beams with
the minimum fails in a brittle manner after cracking. The larger covers spalled before reaching the torsional strength.
behavior of solid members compared with hollow ones
Fig. 4.4aCowans failure criteria (Hsu 1984).
In these tests, the beams had stirrup dimensions of 245 x

545 mm (9.65 x 21.5 in.), and cover thicknesses of 22.5 and
42.5 mm (0.89 and 1.637 in.) for the thin and thick concrete
covers, respectively. The concrete cover thickness is the
critical parameter for spalling. Experiments have shown
that spalling occurs if the cover thickness is greater than
30 percent of the ratio of area to the perimeter of the cross
section. For this calculation, the sectional area and perimeter
do not consider the area removed by voids in hollow sections
(Rahal and Collins 1996).
4.4Prestressed concrete
Adding longitudinal prestressing to a concrete member
subjected to torsion will increase its cracking torque and,
to a lesser extent, its failure torque. The longitudinal
prestressing force creates a compression-shear biaxial state
of stress that delays cracking of concrete. This effect is illus-
trated by Cowans failure criteria, shown in Fig. 4.4a. The
Mohr-Coulomb failure criterion has been simplified into two
straight lines in Fig. 4.4a(a). If the Mohr circle of stresses
touches the straight line BD described by t = c tanj, where
j is the friction angle (a typical value for concrete is 37
degrees), failure is by sliding. If the Mohr circle touches the
tension cutoff line DE described by s = ft, failure is by sepa- Fig. 4.4bUltimate torque-versus-reinforcement factor for
ration. In Fig. 4.4a(b), smax = OE = FE FO = FP FO. The prestressed and reinforced beams under pure torsion (Hsu
variable smax can then be expressed as 1984). (Note: 1 in.-kip = 0.113 kNm.)
2
f pc f pc only the prestress and the concrete tensile strength. In effect,
s max = + t 2
2 2 (4.4a) the longitudinal prestress increases the cracking strength,
as shown in Fig. 4.4b. Expressions for prestressed member
torsional strength are similar to those of reinforced concrete
f pc members with a factor that increases the strength due to the
t = ft 1 + (4.4b) prestress. Pure torsion tests have shown that the prestressed
ft
beams appear to require approximately the same minimum
torsional reinforcement as the nonprestressed beams (Hsu
Cowans failure envelope using Mohrs circle is a simple 1984). Similar to beams without prestressing, prestressed
and concise model for explaining the compression-shear state beams without web reinforcement fail shortly after cracking.
of stress. Equation (4.4b) for shear stress is derived from the Testing has shown that as the amount of longitudinal rein-
geometry of Cowans failure envelope, and it is a function of forcement and prestress increase, the torque at cracking,
of cracks is fewer in HSC. Wider cracks in HSC could also

be the result of reduced aggregate interlock. Shortly after
the longitudinal reinforcement yield, the width of the largest
crack increases rapidly. After this point, all further defor-
mation occurs at this crack. This behavior is exaggerated
in high-strength beams due to the smooth-faced cracks that
are less effective in transmitting shear (Koutchoukali and
Belarbi 2001). Rasmussen and Baker (1995) observed diag-
onal cracks in both NSC and HSC beams on the larger faces
of the cross section because the shear stresses are largest on
these faces. This is similar to the assumptions of the elastic
theory. After first cracking, the number of cracks parallel to
the initial cracks increases as the applied torque increases.
In this way, characteristic spiral cracks develop around the
beam. Cracking of an HSC beam is more brittle than that
of an NSC beam because in HSC, not only the matrix but
also the aggregate splits (the paste in HSC is strengthened
relative to the aggregate), producing a more brittle response.
That is, cracks in NSC matrix propagate around the aggre-
gate, whereas cracks in HSC also pass through the aggre-
gate. Before cracking, surface compressive principal strains
for HSC beams are smaller compared with those of NSC
Fig. 4.4cEffect of longitudinal reinforcement (Mitchell beams because of the higher elastic modulus for HSC.
and Collins 1978). (Note: 1 in.-kip = 0.113 kNm; 1 in. = Whereas HSC delays beam cracking, after cracking the
0.0254 m.) shape and angle of inclination of cracks are the same for all
concrete strengths (Rasmussen and Baker 1995). An HSC
hoop yield, and ultimate increase, whereas the twist at failure beam has a higher torsional stiffness than an NSC beam for a
decreases. This effect is illustrated by the experimental work given torque and a given cross section because of the higher
shown in Fig. 4.4c (Mitchell and Collins 1978). Prestressing modulus of elasticity for HSC. Figure 4.5 shows the torque-
does not influence ultimate torsional strength if the longitu- twist diagrams for an NSC (36.3 MPa [5.1 ksi]) beam and
dinal reinforcement does not yield before failure, as shown an HSC (109.8 MPa [15.9 ksi]) beam under pure torsion,
in Specimen P5 of Fig. 4.4c. Prestressed beams under reported by Rasmussen and Baker (1995). They showed that
sustained, long-term torque exhibit increased rotation due to the torque-twist curves are linear in the uncracked range
concrete creep. Beams tested to failure after being subjected and the HSC beam has greater torsional stiffness. After
to sustained torque showed no reduction in strength or stiff- cracking, the torsional stiffness is reduced significantly in
ness. After a sustained torque, the ratio of beam rotation to both beams. The post-cracking stiffness of an HSC beam
the instantaneous rotation increases with larger values of remains almost linear up to failure, whereas the NSC beam
torque and decreases with increased amounts of transverse curve is nonlinear.
reinforcement (Allos and Rashid 1989). The surface compression strain at failure for an HSC beam
is slightly less than that for an NSC beam, but because of the
4.5High-strength concrete higher modulus of rupture and consequent higher torsional
High-strength concrete (HSC) is more brittle than normal- stiffness for HSC, the twist at peak torque for both NSC
strength concrete (NSC). In some ways, NSC can be treated and HSC beams is approximately the same (Rasmussen and
as a perfectly plastic material, whereas HSC very much Baker 1995). The increase in both longitudinal and trans-
resembles a perfect elastic and brittle material (Rasmussen verse reinforcement stresses immediately after cracking is
and Baker 1995; Iravani 1996; ACI Committee 363 1992). larger for an HSC beam than an NSC beam, resulting in
The ascending portion of the stress-strain curve is steep and brittle cracking of HSC. For the same torque, the reinforce-
approximately linear, whereas the descending curve almost ment stresses in an HSC beam are less than in an NSC beam.
vanishes and shows no strain softening. High-strength Therefore, for a given torque, the required amount of beam
concrete has a peak load strain that is higher than the typical reinforcement is less using HSC. A general comparison of
0.0020 of NSC and an ultimate load strain less than the the ratio of transverse reinforcement stresses and longitu-
typical 0.0035 (Wafa et al. 1995). dinal reinforcement stresses at failure shows that this ratio
In the case of pure torsion, the use of HSC allows a higher increases when concrete strength increases. For HSC, the
service stress in the reinforcing bars and produces a higher bond between concrete and reinforcement is better due
stiffness in the structure, enhancing both the strength and to denser concrete. Therefore, the vertical component of
stiffness of a member. For a given amount of reinforcement, diagonal concrete compression is easily transferred to the
the initial crack width is larger for HSC because larger rein- transverse reinforcement so that it attracts a relatively larger
forcement strains are induced at cracking, but the number amount of torque using HSC. The lower bond between the
stirrup and concrete using NSC results in a portion of the

vertical component of the diagonal concrete compression
being carried by longitudinal reinforcement bending and
shear (Rasmussen and Baker 1995). Beams elongate because
of increased cracking state under increasing torque. The
longitudinal strains at failure increase as concrete strength
increases. For the beams tested by Rasmussen and Baker
(1995), the longitudinal elongation at failure for the HSC
beam (B110.1) was approximately 35 percent larger than for
the NSC beam (B30.3). For a given torque and given cross
section, however, the longitudinal elongation of an HSC
beam was approximately 50 percent lower than that of an
NSC beam. Accounting for the spalling effect is essential for
a correct evaluation of the ultimate torque and failure mech-
anism of over-reinforced HSC beams. In the case of some
over-reinforced NSC concrete beams, the phenomenon of
spalling does not take place, and the torsional strength is Fig. 4.5Torque-twist relation for NSC and HSC beams
evaluated considering the entire cross section. In the case (Rasmussen and Baker 1995). (Note: 1 m = 39.4 in., 1
of over-reinforced HSC beams, the cover is reasonably kNm= 8.85 in.-kip; 1 MPa = 0.145 ksi.)
neglected. In fact, due to the higher load-carrying strength
the plain concrete member fails in a brittle manner once the
of HSC with respect to NSC, spalling of the concrete cover
maximum shear stress equals the concrete tensile cracking
from the stirrups appears to occur more frequently (Cerioni
strength. For reinforced concrete members, because the elastic
et al. 1998). This is because, even though HSC is stronger,
theory does not accurately describe the member behavior after
the tensile strength does not increase in the same propor-
cracking, other theories such as truss model and skew-bending
tion as the compressive strength. Based on the results of 14
were developed to estimate the failure strength.
HSC prestressed beams tested under pure torsion, Wafa et
In members with mild reinforcement, the longitudinal and
al. (1995) concluded that increasing the concrete strength
transverse reinforcement are in tension after cracking and the
and the prestressing force increases the cracking strength,
concrete struts between diagonal cracks are in compression.
ultimate torsional strength, and torsional stiffness. Wafa et
The beam faces are warped as the beam is twisted, causing
al. reported that among the several models usedthe space
bending and compressive stresses in the concrete strut. The
truss model with softening of concrete (Hsu and Mo 1985a),
total amount and the relative amount of longitudinal and
space truss model with spalling of concrete cover (Mitchell
transverse steel reinforcement have a significant effect on
and Collins 1974), and skew-bending theory (Hsu 1984)
the behavior of reinforced beams subject to pure torsion.
the space truss theory with softening of concrete gave the
The behavior of solid members compared with hollow ones
best estimate of the test beams torsional strength. Wafa et
shows the inner core of solid members is ineffective after
al. (1995) observed that the tensile strength increase is not in
cracking, making the cracking strength of hollow members
direct proportion to the compressive strength. The ratios of the
less than solid ones. Torsional strength and general behavior,
modulus of rupture to the compressive strength were 9 percent
however, are similar in solid and hollow sections. The most
and 11.4 percent, respectively, for the HSC and NSC.
significant effect of adding longitudinal prestressing to a
The post-cracking torsional stiffness, as represented by
concrete member subjected to torsion is the increase in its
the slope of the torque-twist curve, is influenced by the
cracking strength. Testing has shown that as the amount of
concrete strength, the amount of torsional reinforcement,
longitudinal reinforcement and the prestress force increase,
and the aspect ratio. The higher the concrete compressive
the torque at cracking, hoop yield, and ultimate strength all
strength, the higher the stiffness is, irrespective of the aspect
increase whereas the twist at failure decreases. The use of
ratio and the prestressing force. Torsional strength increases
HSC in torsional members allows a higher service stress in
with an increase in concrete strength and prestressing level
the reinforcing bars and produces a higher sectional stiff-
and with a decrease in aspect ratio. All tested beams exhib-
ness in the structure. Cracking of an HSC beam is more
ited compression zones located on the longer sides of the
brittle than that of an NSC beam; in HSC, both the matrix
beams. Tension cracks initially formed on one longer face
and aggregate split, producing a more brittle response. An
and progressed to form spirals. Crushing on the other longer
increase in twist immediately after cracking is more signifi-
face characterized the failure pattern.
cant for HSC beams because of the higher energy released,
with more brittle crushing as concrete strength increases.
4.6Concluding remarks
The post-cracking stiffness of an HSC beam remains almost
Saint-Venants elastic theory accurately describes plain
linear up to failure, whereas the curve for an NSC beam is
concrete members torsional behavior at low torque and
nonlinear.
before cracking. Applying Saint-Venants theory to concrete,
CHAPTER 5ANALYTICAL MODELS

FOR PURE TORSION
5.1General
In 1929, using space truss concepts, Rausch developed
an analytical model to predict the torsional strength of rein-
forced concrete members. Unfortunately, Rauschs equa-
tion was unconservative by more than 30 percent for under-
reinforced beams (Hsu 1968a,b). Although Rauschs model
assumed the lever arm area Ao to be the area within the hoop
reinforcement centerline, Aoh, the actual Ao would be much
smaller in members with a high percentage of reinforcement
when concrete softening is considered. The correct determi-
nation of the lever arm area Ao and the centerline of shear
flow depend on the means used to determine the thickness of
the shear flow zone, td. Fig. 5.2.1Hollow box subjected to torsion (Hsu 1993).
Since the late 1960s, the truss model theory has undergone
four major developments. First, Lampert and Thrlimann and Collins (1974), but given herein in the most general
(1968, 1969) introduced the variable angle truss model and form (Hsu 1993)
discovered the bending phenomenon in diagonal concrete
struts. Second, Collins (1973) derived compatibility equa- s = s d cos2 q + s r sin 2 q + r f + r p f p (5.2.1a)
tions to determine the angle of the diagonal concrete struts. It
should be noted that Mitchell and Collins (1974) developed a
space truss model with concrete cover spalling to determine s t = s d sin 2 q + s r cos2 q + rt ft + rtp ftp (5.2.1b)

the shear flow zone thickness. Third, the softening phenom-
t t = ( s d + s r ) sin q cos q
enon in the concrete struts, discovered by Robinson and
Demorieux (1972), was quantified by Vecchio and Collins (5.2.1c)
(1981) using a softened coefficient. Fourth, combining the
equilibrium, compatibility, and softened stress-strain rela- In Eq. (5.2.1a) through (5.2.1c) the reinforcement ratios
tionships, Hsu and Mo (1985a) developed a softened truss rl, rt, rlp, and rtp, should be taken with respect to the shear
model (STM) theory to determine the shear and torsional flow zone thickness td as described in 5.3. These three equa-
behavior of reinforced concrete members throughout the tions are expressed in the form of stress transformation to
post-crack loading history up to the peak strength. Using the show that they satisfy Mohrs stress circle. Because Eq.
STM theory, shear flow zone thickness td can be determined (5.2.1a) and (5.2.1b) involve the reinforcement ratios r, rt,
accurately to calculate the torsional strength of reinforced rlp, and rtp, they are coupled with the compatibility equa-
concrete members. The test data for solid and hollow beams tions through the variable td. The thickness td is a geometric
suggests that, once cracking has occurred, concrete in the variable that is determined by equilibrium conditions as
center member has little effect on torsional strength of the well as compatibility conditions. The determination of td is
cross section and is ignored (Hsu 1968b; Lampert and Thr- similar to that of the flexural neutral axis, which requires
limann 1968; Leonhardt and Schelling 1974). Beams are plane section compatibility conditions.
considered equivalent tubular members. This observation is 5.2.2 Bredts equilibrium equationEquations (5.2.1a)
the basis of the torsion design procedures introduced in ACI through (5.2.1c) were derived from the equilibrium of a
318-95 (MacGregor and Ghoneim 1995). member element in the shear flow zone. To maintain equi-
librium of the whole cross section, a fourth equation derived
5.2Equilibrium conditions by Bredt (1896) should be satisfied. For a shear flow zone
5.2.1 Shear elements in shear flow zoneA prismatic thickness td, the shear stress tlt is
reinforced concrete member subjected to an external torque
T is shown in Fig. 5.2.1(a). This external torque is resisted T
t t =
by an internal torque formed by the circulatory shear flow q 2 Ao t d (5.2.2)
along the cross section periphery. This shear flow q occupies
a zone, called the shear flow zone, with a thickness denoted
by td. The thickness is a variable determined from equilib- This also introduces an additional variable, the torque
rium and compatibility conditions; it is different than the T. The shear flow zone thickness td is incorporated in Eq.
given wall thickness t of a hollow member. (5.2.2) explicitly, as well as implicitly through Ao, which is
An element A in the shear flow zone (Fig. 5.2.1(a)) is a function of td.
subjected to a shear stress tlt = q/td, as shown in Fig. 5.2.1(b).
The in-plane equilibrium of this element should satisfy Eq. 5.3Compatibility conditions
(5.2.1a) through (5.2.1c) as originally derived by Nielsen 5.3.1 Shear elementsAs shown in Fig. 5.2.1(a), the
(1967), Lampert and Thrlimann (1968, 1969), and Mitchell element A in the shear flow zone is subjected to a shear
stress. The in-plane deformation of this element should

satisfy three compatibility equations (Eq. (5.3.1a) through
(5.3.1c)) as originally derived by Baumann (1972) and
Mitchell and Collins (1974). The equations are in the form
of strain transformation to show they satisfy Mohrs strain
circle (Hsu 1993)
e = e d cos2 q + e r sin 2 q (5.3.1a)

e t = e d sin 2 q + e r cos2 q (5.3.1b)

g t
= ( e d + e r ) sin q cos q (5.3.1c)
2
Element A is subjected to in-plane shear as well as out-of-

plane deformation as demonstrated (5.3.2).
5.3.2 Shear strain due to twistingWhen a tube is
subjected to torsion, the relationship between the shear
strain gt in the tube wall and the member angle of twist, F,
can be derived from the compatibility condition of warping
deformation (Bredt 1896).
po
F= g t (5.3.2)
2 Ao

It is clear from Eq. (5.3.2) that the angle of twist, F, will

produce a shear strain gt in the shear flow zone elements.
This shear strain gt will induce the reinforcement strains e
and et in the d-r direction. The relationships between the
strains in the -t direction (e, et, and gt) and the strains in the
d-r direction (ed and er) are described by the three transfor-
mation equations, Eq. (5.3.1a) through (5.3.1c).
5.3.3 Bending of diagonal concrete strutsIn a torsional
member, the angle of twist, F, also produces warping in
Fig. 5.3.2Bending of a concrete strut in
member walls that in turn causes bending in the concrete
the wall of a box section subjected to torsion
struts. The concrete struts are therefore subjected to compres-
(Mitchell and Collins 1974; Hsu and Mo 1985a).
sion due to circulatory shear and bending due to warping of
walls. The relationship between the angle of twist, F, and the
bending curvature of concrete struts, Y, is given in Eq. (5.3.3). Equation (5.3.3) was originally derived by Mitchell and
Figure 5.3.2(a) shows a box member with four walls Collins (1974) using Mohrs circle, as shown in Fig. 5.3.2(c).
of thickness t subjected to torsional moment T. Each wall The illustration in Fig. 5.3.2 and the derivation of bending
contains a shear flow zone with a thickness of td. The perim- curvature are based on Hsu (1993). Although the imposed
eter of the centerline of shear flow q has a width of q along curvature is illustrated by a rectangular box section, this
the top wall. The length of the member in the longitudinal equation is applicable for any arbitrary, bulky section with
direction is taken to be qcosq, so that the diagonal line in multiple walls.
the center plane of the shear flow in the top wall OABC has 5.3.4 Strain distribution in concrete strutsThe curvature
an angle of inclination q with respect to the longitudinal Y derived in Eq. (5.3.3) produces a nonuniform strain distri-
axis. When this member is subjected to an angle of twist, bution in the concrete struts. Figure 5.3.4(a) shows a concrete
F, the center plane OABC becomes a hyperbolic paraboloid strut unit width in a hollow section with a wall thickness of
surface OADC, as shown in Fig. 5.3.2(b). The plane edge t. The tension area in the cross section inner portion is disre-
CB rotates to the position CD through an angle Fqcosq. garded. The area in the outer portion, which is in compres-
The curve OD has a curvature of Y that is related to F sion, is considered effective in resisting the shear flow. The
and q by compression zone depth from the neutral axis to the extreme
compression fiber is defined as the shear flow zone thickness
Y = Fsin2q (5.3.3) td. Within this thickness td, the strain distribution is assumed
to be linear, as shown in Fig. 5.3.4(b), based on Bernoullis
Fig. 5.3.4Strains and stresses in concrete struts (Hsu and Mo 1985a).
plane section hypothesis used in bending theory (Mitchell Descending branch:

and Collins 1974). The thickness td can therefore be related
to the curvature Y and the maximum surface strain eds by the e / ze 1 2 ed
simple relationship s d = zfc 1 d o
> 1 (5.4.1a(b))
2 / z 1 ze o
e ds (5.3.4a)
td =
Y where z is a softening coefficient. Section 5.6.1 discusses the
softening coefficient as presented in Eq. (5.6.2d) with values
The average strain ed can be related to the maximum strain ranging between 0.2 and 0.6.
eds, as shown in Fig. 5.3.4(d), by The stress-strain relationship of concrete in tension is
irrelevant, if the concrete tensile stress sr is assumed to be
e ds (5.3.4b) zero in the equilibrium Eq. (5.2.1a) through (5.2.1c).
ed = 5.4.2 Mild reinforcementWhen the tensile strength of
2
the concrete is neglected, the stress-strain relationship of
mild reinforcement is taken as the elastic perfectly plastic
Equations (5.3.2), (5.3.3), (5.3.4a), and (5.3.4b) are the relationship, expressed as
four additional compatibility equations for torsion. They
introduce four additional variables, F, Y, td, and eds. f = Ese e < ey (5.4.2a(a))
5.4Stress strain relationships f = fy e ey (5.4.2a(b))

5.4.1 Concrete strutsThe stress-strain relationship of
concrete in compression, as shown in Fig. 5.3.4(e), can be ft = Eset et < ety (5.4.2b(a))
expressed analytically by two branches of parabolic curves
(Hsu and Mo 1985a) ft = fty et ety (5.4.2b(b))
Ascending branch: 5.4.3 Prestressing reinforcementIn the case of

prestressed concrete, the nonlinear stress-strain relationship
e e 2 ed of prestressed strands is given (Mattock 1979)
s d = zfc 2 d d < 1 (5.4.1a(a))
ze o ze o ze o
fp = Eps(edec + es) fp 0.7fpu (5.4.3a(a))
E ps (e dec + e s )
fp =
{ ( E (e ) }
1/ R
1+ + e s ) / f pu
R
fp > 0.7fpu
ps dec
(5.4.3a(b))
In structures where prestress is applied in both direc-

tions, such as nuclear containment vessels, Eq. (5.4.3a(a))
and (5.4.3a(b)) are applicable to prestressing reinforcement
in both the longitudinal and transverse directions. When
applied to longitudinal reinforcement, fp is equal to flp, and es
is equal to e. When applied to transverse reinforcement, fp is
equal to ftp, and es is equal to et.
The equilibrium equations, compatibility conditions, and
stress-strain relationships described are used by two main
theories for torsional analysis, namely, the compression field
theory (CFT) (Mitchell and Collins 1974) and the STM (Hsu
and Mo 1985a). The primary difference between these two
theories lies in the stress-strain relationship of concrete in
compression. The STM uses a softened stress-strain curve
as shown in Eq. (5.4.1a(a)) and (5.4.1a(b)), where z is a
softening coefficient that varies from approximately 0.2 to
0.6. In CFT, the stress-strain curve is not softened, and the
softening coefficients z in Eq. (5.4.1a(a)) and (5.4.1a(b)) are Fig. 5.5aSpalling of the concrete cover due to torsion
taken as unity. To compensate for the unconservative nature (Mitchell and Collins 1974).
of this non-softening assumption, CFT also assumes spalling
concrete cover before reaching the peak torque. As a result, direction; therefore, it can be derived from compatibility Eq.
CFT is also called the spalling theory. The CFT and the (5.3.1a) and (5.3.1b), by combining them to form
STM are described in detail in 5.5 and 5.6, respectively,
by combining the stress-strain relationships of concrete e + ed
and reinforcement given in Eq. (5.4.1a(a)) and (5.4.1a(b)), tan 2 q = (5.5a)
et + ed
and through (5.4.3a(b)) with the equilibrium Eq. (5.2.1a)
through (5.2.1c) and (5.2.2) and compatibility Eq. (5.3.1a)
through (5.3.4b). The solution algorithms of the two theories Equation (5.5a) shows the angle q can be expressed as a
are also different. function of the strain in the concrete diagonals (ed), the strain
in the longitudinal reinforcement (e), and the strain in the
5.5Compression field theory transverse reinforcement (et). The full behavioral response
Collins and Mitchell (1980) and Mitchell and Collins of reinforced concrete members subjected to torsion can be
(1974, 1978) proposed the CFT for structural concrete in predicted by using compatibility (Eq. (5.5)), equilibrium (Eq.
torsion. This section provides a basic summary of the theory. (5.2.1a) through (5.2.1c)), the stress-strain relationships of
Before the truss analogy equilibrium equations can be used concrete [Eq. (5.4.1a(a)) and (5.4.1a(b))], and those for mild
to design a member for torsion, the diagonal struts inclina- reinforcement [Eq. (5.4.2a(a)) through (5.4.2b(b))]. Besides
tion should be determined from the compatibility equations. these fundamental equations, additional torsional aspects
Wagner (1929) dealt with an analogous problem in studying were considered in the CFT. In resisting the torsion, not all
the post-buckling shear resistance of thin-webbed metal of the concrete is effective in providing diagonal compres-
girders. He assumed that after buckling, the thin webs would sive stresses. Estimating the equilibrium of a corner element
not resist compression and the shear would be carried by a for a beam in torsion (Fig. 5.5a) reveals that the compres-
field of diagonal tension. To determine the angle of inclina- sion in the concrete tends to push off the corner whereas the
tion of the diagonal tension, Wagner considered the systems tension in the hoops holds it in place. Because concrete is
deformations. He assumed that the angle of inclination of weak in tension, the concrete outside of the hoops spalls off
the diagonal tensile stress would coincide with the angle at higher torsions. Because of concrete cover spalling, the
of inclination of the principal tensile strain. This approach effective outer surface of the concrete is assumed to coincide
became known as Tension Field Theory. Applying Wagners with the hoop centerline.
approach to reinforced concrete, the concrete is assumed to Examining the deformed shape of the twisted beam in
be unable to carry tension, and the shear is assumed to be Fig. 5.5b reveals that the diagonal concrete stresses vary in
carried by a diagonal compression field. The inclination q magnitude over the thickness of the effective concrete tube,
of the diagonal compression represents the principal strain from zero at the inside to a value fds corresponding to the
strain eds at the effective outer surface. As in the case of
given in 5.2 and 5.3. The following solution algorithm is the

approach used by Rahal (2000a,b). From the equilibrium of
an element in the shear flow zone, shown in Fig. 5.2.1(b),
two equations can be derived
DN At ft
q= (5.5d)
po s

and
At ft po
tan q = (5.5e)
s DN
Substituting Eq. (5.5d) into Bredts Eq. (5.2.2), the

torsional moment T can be related to the longitudinal and
transverse reinforcement forces, DN and Atft, as
DN At ft
T = 2 Ao (5.5f)
po s

Like flexure, the compression zone depth is a function of

reinforcement tensile forces, and is derived from equilibrium
Fig. 5.5bEffective wall thickness of a twisted beam conditions. In a shear element, however, the reinforcement
(Collins and Mitchell 1980). includes both longitudinal and transverse reinforcement
forces, DN and Atft. Eliminating q from Eq. (5.2.1a) through
flexure, this actual stress distribution can be replaced by a (5.2.1c) results in
uniform stress of a1fc = fd acting over a depth of b1td = ao,
where the stress block factors a1 and b1 depend on the shape DN Af
of the concrete stress-strain curve and the value of surface ao = + t t (5.5g)
a fcpo a1 fcs
compression strain eds. The centerline dimensions of the
resulting tube of uniformly stressed concrete of thickness ao
are assumed to define the shear flow path q. This path lies Once the compression block depth ao is known, the terms
ao/2 inside the centerline of the hoop, as shown in Fig. 5.5b. Ao and po in Eq. (5.5b) and (5.5c) can be calculated.
Given the shear flow path, the terms Ao (the area enclosed To determine the longitudinal and transverse beam strains,
by the shear flow) and po (the perimeter of the shear flow el and et, which correspond to the chosen value of eds, the
path) can be determined using Eq. (5.5b) and (5.5c), which basic compatibility Eq. (5.3.1a) through (5.3.4a) were
assume a thin tube manipulated to develop two expressions
ao (5.5b) a b f A s
Ao = Aoh ph e t = 1 1 c oh 1 e ds (5.5h)
2
2 ph At ft
po = ph 4ao (5.5c)
a b f A p
e = 1 1 c oh o 1 e ds (5.5i)
Based on the equilibrium of an element in the shear flow 2 ph DN
zone, shown in Fig. 5.2.1(b), the shear flow, q, in a box
section can be expressed in terms of the longitudinal rein- A trial-and-error process can be used to solve the eight
forcement force, DN, and the transverse reinforcement force, variables (q, ao, et, e, Ao, po, DN, and Atft) using Eq. (5.5a)
Atft. The longitudinal reinforcement force DN is assumed to through (5.5c), (5.5e), and (5.5h) through (5.5i). First, select
be distributed uniformly along the shear flow path po. Equi- a value of eds and assume a pair of forces, Atft and DN, from
librium Eq. (5.2.1a) through (5.2.1b) are simplified in three which the four variables q, ao, et, and e, can be calculated
respects: 1) for pure torsion, s = st = 0; 2) concrete tensile from Eq. (5.5e), (5.5g), (5.5h), and (5.5i), respectively.
stress is neglected, sr = 0; and 3) when no prestressing rein- Given the value of ao, the variables Ao and po can be calcu-
forcement is used, rp = rtp = 0. In the CFT, several approxi- lated from Eq. (5.5b) and (5.5c). Based on the values of et,
mate algorithms have been developed to plot the torque-twist e, and ed (related to eds by geometry), another q can be calcu-
curve using the equilibrium and compatibility relationships lated from Eq. (5.5a). If the q from Eq. (5.5a) is not equal
to the q from Eq. (5.5e), assume another pair of Atft and DN, This equation is valid for HSC up to 100 MPa (14.5 ksi).
and repeat the cycle until Eq. (5.5a) and Eq. (5.5e) produce As proposed by Belarbi and Hsu (1995), for NSC up to 42
the same q. MPa (6.0 ksi), z becomes
Once these eight variables are solved, the torque T can be
calculated from Eq. (5.5f) and the angle of twist, F, from 0.9
z= (5.6.2d)
Eq. (5.3.3). In Eq. (5.3.3), the curvature of the concrete 1 + 400 e r
struts, Y, can be obtained from Eq. (5.3.4a), where the neutral
axis depth td is equal to ao/b1, and eds is the chosen value. The
torque-twist response (T-versus-F curve) is then determined The softening coefficient z usually varies in the range of
by repeating these calculations for a series of eds values. 0.2 (0.045) to 0.6 (0.003).
5.6.3 Centerline of shear flow zone and formulas for Ao
5.6Softened truss model and poAs shown in Fig. 5.3.4(c), the resultant force of
5.6.1 GeneralThe softened truss model is built on satis- the compression stress block C is located at a distance k2td
fying the equilibrium Eq. (5.2.1a) through (5.2.2), the compat- from the surface, where the coefficient k2 defines the loca-
ibility Eq. (5.3.1a) through (5.3.4b), the stress-strain materials tion of the resultant force C. By integrating the concrete
relationships in Eq. (5.4.1a(a)) and (5.4.1a(b)) for concrete, stress-strain curve given in Eq. (5.4.1a(a)) and (5.4.1a(b)),
and Eq. (5.4.2a(a)) through (5.4.2b(b)) for reinforcement. the coefficient k2 is found to vary in the range of 0.40 to 0.45,
The properties of concrete stress block and shear flow zone depending on concrete strength. To simplify, the centerline
are derived from the softened stress-strain relationships of of the shear flow is assumed to coincide with the centerline
concrete. of the shear flow zone, located at a distance 0.5td from the
5.6.2 Concrete stress blockAs shown in Fig. 5.3.2, the extreme compression fiber. This assumption also results in a
diagonal concrete struts in the shear flow zone are subjected compatible agreement between theory and tests.
to axial stress and bending. The compressive strains in the Based on this simple centerline assumption of the shear
concrete struts are assumed to vary linearly from eds at the flow, formulas for calculating the lever arm area Ao and the
surface to zero at the neutral axis, as shown in Fig. 5.3.4(b). perimeter po for a thick tube, whose thickness is usually
The compression zone depth td is defined as the shear flow quite large with respect to the overall dimension of the cross
zone. The distribution of compressive stresses within the section and different from the thin tube assumption in CFT
thickness td is shown by the solid curve in Fig. 5.3.4(c), when the softening of concrete is considered, are
based on the softened stress-strain relationship shown in Fig.
5.3.4(e). The peak stress is zfc and the average compressive Ao = Ac
1
pc t d + t d2 (5.6.3a)
stress is sd. The average stress sd of the concrete stress block 2
in Fig. 5.3.4(c) can be expressed as follows
po = pc 4td (5.6.3b)
sd = zfc (5.6.2a)
When the softening coefficient z of concrete is consid-
where the coefficient k1 is the ratio of the average stress to ered, the shear flow zone thickness td is usually large with
the peak stress. By integrating the stress-strain curve in Eq. respect to the overall cross section dimensions. The formula
(5.4.1a(a)) and (5.4.1a(b)), the coefficient k1 can be expressed for calculating Ao, therefore, is expressed in Eq. (5.6.3a) and
should include the third term td2. The thin tube formula used
e ds 1 e ds e ds in the CFT (Eq. (5.5b)) is not used because it neglects the
k1 = 1 3 ze 1 (5.6.2b(a))
ze o ep ao2 term.
o
5.6.4 Solution algorithmThe 18 governing equations
for a torsional member are: four equilibrium, (Eq. (5.2.1a)
z 2 1 e ds z2 e ds 1 e ds through (5.2.2); seven compatibility, (Eq. (5.3.1a) through
k1 = 1
2
1 + 1
( 2 z) 3 ze o ( 2 z) ze o 3 ze o
2 (5.3.4b)); seven constitutive (Eq. (5.4.2a(a)) through
(5.4.3a(b)) and (5.6.2a) through (5.6.2d)).
e ds For a member subjected to pure torsion, the normal
>1
ep stresses sl and st acting on an element in the shear flow zone
(5.6.2b(b)) are equal to zero (sl = st = 0). When the concrete tensile
strength is neglected, sr = 0. Disregarding these three vari-
The coefficient k1 is tabulated as a function of eds and z in the ables (sl, st, and sr) in the analysis, the 18 equations contain
work by Hsu (1993), and equals approximately 0.8. The soft- 19 unknown variables as follows: seven stress or force vari-
ening coefficient was first proposed by Zhang and Hsu (1998) ables, including tlt, sd, fl, ft, flp, ftp, and T; 10 strain or geom-
as a function of tensile strain er and concrete strength fc etry variables, including el, et, glt, ed, er, q, F, Y, td, and eds;
and two material coefficients, including z and k1. If one of
5.8 1 5.8 the unknown variables is selectedfor example, edthen
z= 0.9
fc(MPa ) 1 + 400 e r and fc(MPa ) (5.6.2c) the remaining 18 unknown variables can be solved using the
18 equations.
Next, select the concrete strain ed as the independent vari- Equations (5.6.8a) and (5.6.8b) will facilitate solution
able because it varies monotonically from zero to maximum. procedures (5.6.9).
The solutions for a series of ed values make it possible to 5.6.9 Solution procedure and flow chartThe solution
trace the loading history. A solution algorithm is given by procedure is illustrated by a flow chart shown in Fig. 5.6.9.
Hsu (1993) and is briefly described. This method requires This solution procedure uses 12 equations, Eq. (5.4.2a)
the derivations of Eq. (5.6.5) through (5.6.8b). through (5.4.3b) and (5.6.2a) through (5.6.8b), to solve 12
5.6.5 Thickness td as a function of strainThe thickness unknown variables (z, k1, sd, f, ft, fp, ftp, e, et, er, eds, and
of the shear flow zone td can be expressed in terms of strain td) for each selected value of ed. From these stresses and
using the seven compatibility equations: Eq. (5.3.1a) through strains, the remaining six variables (q, tt, T, gt, F, Y) can
(5.3.4b). The substitutions and manipulations result in be calculated using the appropriate equilibrium or compat-
ibility equations. Selecting a series of ed values produces the
Ao ( e d ) ( e r e d ) entire loading history of the 18 variables. The efficiency of
td = (5.6.5) this solution procedure arises from the fact that the angle q
po ( e e d ) ( e t e d )
is eliminated in the expressions of Eq. (5.6.5) through (5.6.7)
and is, therefore, not involved in the iteration process of td.
The variable td is expressed in terms of strains in all d, r, In addition, the derivation of Eq. (5.6.6) and (5.6.7) allows
, and t directions (ed, er, e, et). The compressive strain is the iteration to include nonlinear stress-strain relationships
defined as negative and tensile as positive. The variable td (Eq. (5.4.3a) and (5.4.3b)) of prestressing strands, therefore
is also involved in equilibrium equations through the terms making this solution algorithm applicable to prestressed
Ao, po, r, and rt. Therefore, the variable td should first be concrete beams. A hand-calculation example illustrating
assumed and then checked by Eq. (5.6.5). this solution procedure is given by Hsu (1993). The STM
5.6.6 Longitudinal strain el as a function of longitudinal theory was used by Hsu and Mo (1983, 1985a) to calculate
stresses f, fpThe strain e can be related to the reinforce- the strength and behavior of 108 torsional beams. For the
ment stresses f and fp by eliminating the angle a from the 61 under-reinforced beams with stirrup spacings within ACI
equilibrium Eq. (5.2.1a) using compatibility equations (Eq. code limits, the experiment-to-calculated torsional strengths
(5.3.1c) to (5.3.4b)). These manipulations result in had a mean value of 1.014 and a standard deviation of 0.051.
The post-cracking deformations (angle of twist and strains
Ao ( e d ) ( s d ) in reinforcement and concrete) were also correctly evalu-
e = ed +

( A f + A f ) (5.6.6)
p p
ated. McMullen and El-Degwy (1985) compared the STM
(softened theory) and the CFT (spalling theory) using their
extensive torsion tests. They concluded that the space truss
The three unknown variables e, f, and fp can be solved model with softening of concrete gives a better prediction of
using Eq. (5.6.6) and the stress-strain relationships, Eq. maximum torque for the beams tested in this investigation
(5.4.2a), (5.4.2b), (5.4.3a), and (5.4.3b), for longitudinal (average Tu,exp/Tu,th = 1.03 and standard deviation = 0.091).
reinforcement. The space truss theory with spalling of concrete cover gave
5.6.7 Transverse strain et as a function of transverse an average Tu,exp/Tu,th of 1.10 and a standard deviation of
stresses ft, ftpSimilarly, the strain et can be related to the 0.129. McMullen and El-Degwy also observed that concrete
reinforcement stresses ft and ftp by eliminating the angle q cover spalled only after maximum torque was reached.
from equilibrium Eq. (5.2.1b) using the compatibility equa-
tions (Eq. (5.3.1b) through (5.3.4b)). 5.7Graphical methods
5.7.1 Rahals methodRahal developed a simplified
Ao s ( e d ) ( s d ) method for design and analysis of reinforced concrete panels
et = ed +

( )
po At ft + Atp ftp (5.6.7)
subjected to pure shear (Rahal 2000a) and extended the
method to reinforced concrete beams subjected to torsional
moments (Rahal 2000b). The original method for calculating
The three unknown variables et, ft, and ftp can be solved shear relates the normalized ultimate shear strength vu/fc of
using Eq. (5.6.7) and the two stress-strain relationships, Eq. reinforced concrete panels to the reinforcement indexes wt
(5.4.2c) and (5.4.2d) and (5.4.3a) and (5.4.3b), for transverse and wL in the transverse and longitudinal directions, respec-
reinforcement. tively. The reinforcement indexes are given by
5.6.8 Useful compatibility equationsBy combining
compatibility Eq. (5.3.1a) and (5.3.1b), the two variables er and rt fty rl f l y
wt = ; wL = (5.7.1a)
q can be expressed directly in terms of strains e, et, and ed as fc fc

e r = e + e t e d (5.6.8a)
Figure 5.7.1 gives the relationship between the indices and
the normalized shear strength obtained using the results of
e + ed the modified CFT (Collins and Mitchell 1991). Each curve
tan 2 q =
e t + e d (5.6.8b)
in the figure represents the relationship between vu/fc and wL

at a given wt.
At relatively low wt values, the strains in the transverse rein-
forcement exceed the yield strains before the ultimate strength
is reached. Beyond a specific level of reinforcement, the
concrete is crushed before the reinforcement yields. Figure
5.7.1 shows a curve passing through those points beyond
which concrete is crushed before the transverse reinforce-
ment yields (over-reinforced case). The figure also shows a
similar curve for the over-reinforced case in the longitudinal
direction. The two yield curves divide Fig. 5.7.1 into four
regions. The relative position of a point of coordinates (wt;
wL), with respect to these curves or regions, indicates the
expected failure mode of an element with these reinforcement
ratios. Four failure modes are possible:
1) Partially over-reinforced; only longitudinal reinforce-
ment yields (Zone III)
2) Only transverse reinforcement yields (Zone II)
3) Completely over-reinforced; concrete crushing before
reinforcement yielding (Zone IV) or
4) Completely under-reinforced; longitudinal and trans-
verse reinforcement yield (Zone I).
Because vu/fc, wt, and wL are dimensionless, they
provide values that can be applied to any system of
units. The results of this method have been compared
with experimental results from 46 reinforced concrete
panels and were found to give accurate calculations
of the ultimate shear strength and failure mode. The
average ratio of experimental shear strength to predicted
strength was 1.01 with a coefficienct of variation (COV)
of 0.125 (Rahal 2000a). This method is applied to beams
subjected to torsion by idealizing the section as a hollow
tube and adopting simplified assumptions regarding the
hollow tube thickness and shear flow zone size. Tube
walls are treated as the shear panels to which Fig. 5.7.1
applies. Based on assumptions of the method, the rein-
forcing indexes are given by Fig. 5.6.9Flow chart for torsion analysis using STM.
At fty po
wt = (5.7.1b)
0.42 sAcp fc

A f y + Aps f py
wL = (5.7.1c)
0.375sAcp fc

The ultimate torsional moment is also related to ultimate

shear strength in walls
Acp2
Tu = 0.67 vu (5.7.1d)
pcp
Fig. 5.7.1Normalized shear strength curve for reinforced
concrete members (Rahal 2000a).
In contrast to the STM and CFT, Rahals method is used
to calculate approximate strength only and not the entire
perimeter, and therefore can be applied to closed sections of
torque-twist relationship. Equations (5.7.1b), (5.7.1c), and
various shapes.
(5.7.1d) depend on the gross area and outer cross section
5.7.2 Leu and Lees MethodLeu and Lee (2000)
proposed a graphical solution to the STM. Their strength
76 11, 000
hb = (fy in MPa) or hb = (f in psi)
200 + f y 30, 000 + f y y

(5.7.2c)
76 11, 000
htb = (fy in MPa) or htb = (f in psi)
100 + f y 15, 000 + f y y

(5.7.2d)
3. Calculate the reinforcement indexes wL and wt
h
wL = (5.7.2e)
h b

ht
wt = (5.7.2f)
htb

4. Calculate the nondimensional balanced torsional

Fig. 5.7.2Strength contour (ws) diagram (Leu and Lee strength that corresponds to the case when both longitudinal
2000). and transverse reinforcements yield simultaneously with
crushing of diagonal struts, Txub
contour diagram, shown in Fig. 5.7.2, is constructed to give
the torsional strength of a rectangular member when their
140 20, 000
longitudinal and transverse reinforcement indexes, wL and Txub = (fy in MPa) or Txub = (f in psi)
wt, are calculated. Because this graph is built on the failure 300 + f y 42, 000 + f y y

mode diagram of Hsu (1993), it can also predict the four
failure modes: (5.7.2g)
1) Under-reinforced
2) Partially under-reinforced in the transverse direction 5. Calculate the balanced torsional strength Txub
3) Partially under-reinforced in the longitudinal direction
4) Over-reinforced. The diagram was constructed based Txub fcAcp2
Txub = (5.7.2h)
on STM and, therefore, its accuracy is as good as STM pcp
prediction.
In Fig. 5.7.2, a = 0.4 and a = 0.6 correspond to reinforce-
ment grades of 40 and 60 ksi (280 and 420 MPa), respec- 6. Use wL and wt to determine the value of ws from Fig.
tively. The diagrams are almost identical for different rein- 5.7.2. The coordinate (wL, wt) in the four failure mode
forcement grades, except in the over-reinforced region of regions also determines the member failure mode; and
failure mode of Region IV. Fortunately, an over-reinforced 7. Determine the torsional strength.
design is usually not permitted.
1. Calculate the normalized reinforcement ratios h and ht Txu = wsTxub (5.7.2i)
The graphical solution using Fig. 5.7.2 is not exact. Its

f y A
h = accuracy is affected by the aspect ratio of the cross section and
fc Acp (5.7.2a) the difference in the yield stresses, fy and fty, when calculating

fy = (fy + fty)/2. Thirty-eight specimens from Hsu (1968a)
were used to evaluate the accuracy of the design charts. The
f y At pcp mean and maximum errors of torsional strength were only 1.5
ht =
fc Acp s (5.7.2b) percent and 4 percent, respectively, when compared with the

strengths of those obtained using the STM.
where fy = (fy + fty)/2 is the average yield strength of longitu-
CHAPTER 6MEMBERS SUBJECTED TO
dinal and transverse reinforcement bars, fy and fty.
TORSION COMBINED WITH OTHER ACTIONS
2. Calculate the balanced normalized reinforcement ratios
6.1General
when both longitudinal and transverse reinforcements yield
Torsion rarely occurs in concrete structures unaccompa-
simultaneously with crushing of diagonal struts hlb and htb
nied by other stress resultants. In the general case of loading,
a beam cross section can be subjected to a maximum of six

stress resultants, in the orthogonal system of coordinates,
as shown in Fig. 6.1. Three forces and three moments are
possible: one torsional moment, two shearing forces along
the major and minor axes, two bending moments along
the major and minor axes, and an axial force (tension or
compression).
Early torsion research concerned with pure torsion was
later extended to torsion combined with flexure (Hsu 1968d).
This combination has attracted more attention than all other
combinations of stress-resultants combined. The torsion and
shear combination attracted less attention regardless of its
practical importance in design. Torsion and compression is
the least studied regardless of its occurrence in design, like Fig. 6.1General case of loading a beam cross
in the case of prestressed members or columns subjected section.
to earthquake loading. Early experimental work, which
focused on the ultimate strength and stiffness of concrete
beams, mostly recommended interaction curves between
torsion and flexure and, in some cases, with shear. Although
early theoretical work focused on the ultimate strength and
shape of interaction curves, most research was applicable
only to under-reinforced members in which all the reinforce-
ment yields at ultimate state. The first model to consider
strain compatibility and that was capable of obtaining the
full response of concrete members subjected to pure torsion
was the CFT by Mitchell and Collins (1974). This model
was further developed by Onsongo and Collins (Onsongo
1978) to calculate the full response under combined torsion,
flexure, and axial load. These two models apply to both
under-reinforced and over-reinforced sections, a feature
unavailable in previous models concerned only with ultimate
strength and not strain compatibility. Various models are
capable of calculating the full response of sections subjected
to some combination of the six stress resultants (Fig 6.1).
One such model was developed in the 1970s by Rabbat Fig. 6.2.1Test setup for combined torsion and flexure.
and Collins (1977, 1978) in parallel with the development
of CFT for torsion. Another model developed by Rahal and
two reasons: first, researchers understood that such studies
Collins (1995a), considers information on torsion and shear
would be prudent. Second, the beams subjected to combined
published in the 1980s and 1990s, like tension stiffening and
T and M are easier than other combinations in testing. In
concrete softening. Both models provide a unique perspec-
the absence of shearing forces, a four-point loading arrange-
tive on the interaction between torsion and various stress
ment is easily modified to apply constant torsional moment
resultants. Ewida and McMullen (1981) also developed a
along with the bending moment in the central portion of a
model to calculate the post-cracking response of sections
beam specimen. Figure 6.2.1 shows that the central loads in
subjected to combined torsion, shear, and flexure. Their
a four-point load test setup can be applied at an eccentricity
model explained other phenomena like shear lag and effects
e, subjecting the test region of the specimen to a uniform
of reinforcement on the interaction curve. Chapter 6 summa-
combination of torsion and flexure. Torsional moment to
rizes work on the interaction between torsion and other stress
bending moment ratio is controlled by distances e and a, as
resultants. Although members reinforced only in the longi-
shown in Fig. 6.2.1. The most significant factors affecting
tudinal direction are uncommon in practice, their behavior
the behavior of members subjected to combined torsion and
is discussed because in some cases they behave similarly to
flexure are the presence of transverse reinforcement, the
members with transverse and longitudinal reinforcement.
torque-to-bending ratio (T/M), and the amount and distribu-
tion of the longitudinal reinforcement. Other factors include
6.2Torsion and flexure
the cross section shape and concrete strength.
6.2.1 GeneralThe combination of torsional (T) and flex-
The presence of transverse reinforcement causes a consid-
ural (M) moments can be critical in design. Aside from the
erable change in the behavior of members subjected to
case of pure torsion, this combination has prompted most of
torsion. Sections 6.2.1 and 6.2.2 discuss the T-M interaction
the torsion research combined with other stress-resultants for
for members with and without this reinforcement.
Fig. 6.2.3aTorque-twist relationships at various T/M

ratios for sections with symmetrical longitudinal reinforce-
ment (McMullen and Warwaruk 1967). (Note: 1 in.-kip =
0.113 kNm; 1 in. = 0.0254 m.)
The interaction for concentrically prestressed members is

Fig. 6.2.2Normalized T-M interaction curves for members similar to that of nonprestressed members. For eccentrically
without transverse reinforcement. prestressed beams, the induced flexural moment increases
torsional strength when the eccentric prestressing force
6.2.2 Members without transverse reinforcementThe counteracts tension caused by the bending moment. ACI
T-M interaction in members containing only longitudinal 318 gives no direct information on the T-M interaction in
reinforcement has limited practical importance, attracting members without transverse reinforcement. The minimum
a significant amount of experimental research, mainly due amounts of longitudinal reinforcement for torsion and
to its importance in understanding the behavior of members flexure and transverse reinforcement for torsion required
with transverse reinforcement. for the combination are usually calculated independently
Behavior and ultimate strength of these members is signif- for each stress resultant, and the sole interaction is from
icantly influenced by the shape of the beam cross section, superposition of the amount of longitudinal reinforcement
the amount and distribution of the longitudinal reinforce- required for each stress-resultant.
ment, and the T-M ratio (Zia 1970). In prestressed members, 6.2.3 Members with transverse reinforcementMembers
the amount and eccentricity of the prestressing also affect reinforced in the longitudinal and transverse directions
the behavior because it causes a strain variation similar to exhibit significant post-cracking strength and ductility
that due to a compressive force or a flexural moment (Hsu compared with members reinforced in the longitudinal
1968e). direction only. Behavior of these members also depends
Research on the T-M interaction in members containing significantly on the distribution and amount of longitudinal
only longitudinal reinforcement is mainly experimental. reinforcement. Cases of symmetrically and unsymmetrically
Figure 6.2.2 shows examples of the interaction curves recom- reinforced beams are discussed later. Figure 6.2.3a shows the
mended by various researchers. For square and rectangular torque-twist relationship of four 152 x 305 mm (6 x 12 in.)
sections, Kemp et al. (1961) and Hsu (1968a) independently under-reinforced solid reinforced concrete specimens with
suggested the trilinear interaction. Victor and Ferguson (1968) symmetrical longitudinal reinforcement tested by McMullen
suggested a similar trilinear interaction for L-sections, and a and Warwaruk (1967) at variable T/M ratios ranging from
square interaction for T-beams. Lim and Mirza (1968) also pure torsion to T/M = 0.25. As shown in Fig. 6.2.3a, the addi-
proposed a square interaction curve for T-beams. Based on tion of flexural moment significantly reduced the torsional
evaluation of numerous experimental results, Zia (1970) strength, ductility, and post-cracking stiffness.
observed that significant reduction in the torsion strength Figure 6.2.3b shows the experimentally observed normal-
occurs at M/Mo between 0.5 and 1.0, and a circular interac- ized T-M interaction from two series of symmetrically rein-
tion curve serves as a lower bound for the larger portion of forced specimens. Five under-reinforced beams, each 152
the experimental results. The scatter in the results, however, x 305 mm (6 in. x 12 in.), were tested by McMullen and
is relatively large. Some test results have suggested a linear Warwaruk (1967). The torque-twist results of four of those
interaction (McMullen and Woodhead 1973). beams are shown in Fig. 6.2.3b. In addition, five symmetri-
The interaction between T and M in prestressed members cally prestressed beams, each 305 x 432 mm (12 x 17 in.),
depends largely on the eccentricity of the prestressing force. were tested by Mardukhi and Collins (Mardukhi 1974). This
Fig. 6.2.3cTorque-twist relationships at various T/M ratios

for sections with unsymmetrical longitudinal reinforcement
Fig. 6.2.3bNormalized T-M interaction curves in symmet-
(Onsongo 1978). (Note: 1 kNm= 8.85 in-kip; 1 m = 39.4 in.)
rically longitudinally reinforced members with transverse
reinforcement.
The peak in the interaction curve depends on the ratio r
study showed that a flexural moment equal to 60 percent of and if the beam is under-reinforced or over-reinforced in
the ultimate flexural strength caused only about a 10 percent the transverse direction. An increase of up to 30 percent in
reduction in the torsional strength. Similarly, a torsional torsional strength was observed with the addition of flexural
moment equal to 40 percent of the ultimate pure torsion moment equal to 40 percent of the pure flexural strength
strength caused an approximately 20 percent reduction in in the under-reinforced tests by McMullen and Warwaruk
the flexural strength. (1967). Onsongo (1978), however, observed a 25 percent and
The behavior of unsymmetrically reinforced beams differs 6 percent increase in under-reinforced and over-reinforced
significantly from the previous case. Figure 6.2.3c shows beams, respectively. The interaction curve seems to be less
the torque-twist relationship of three under-reinforced affected by the ratio r in over-reinforced members, where
hollow reinforced concrete specimens tested by Onsongo Onsongo reported only a 6 percent increase in torsional
and Collins (Onsongo 1978) at variable T/M ratios from strength in spite of the significantly low value of r (approxi-
0.63 to 4.27. Figure 6.2.3d shows the observed interac- mately 0.1). Zararis and Penelis (1986), who tested unsym-
tion from four series of unsymmetrically reinforced beams, metrically reinforced T-beams under combined flexure and
one series tested by McMullen and Warwaruk (1967), two torsion, observed a trend similar to that shown in Fig. 6.2.3d.
by Onsongo and Collins (Onsongo 1978), and one by the Due to weakness in the smaller top reinforcement, the pres-
Zurich group (Lampert and Thrlimann 1968, 1969). The ence of a flexural moment increased torsional strength by
ratio of compression to tension longitudinal reinforcement approximately 18 percent. Presence of a flexural moment
yield force r = (Asfy/Asfy) ranged from 0.1 to 0.27. reduces the torsional ductility of a member with symmetrical
Figures 6.2.3c and 6.2.3d show that the addition of a small or unsymmetrical longitudinal reinforcement, as shown in
flexural moment can significantly increase the torsional Fig. 6.2.3a and 6.2.3c.
strength and post-cracking stiffness of unsymmetrically The torque-twist diagrams show that the ultimate torsional
reinforced beams. In pure torsion, the additional bottom strength was reached at a smaller twist. Similarly, the pres-
longitudinal reinforcement available in unsymmetrically ence of relatively small flexural moment, T/M = 4.27 in Fig.
reinforced sections does not increase the ultimate strength 6.2.3c, also affected the curvature of unsymmetrically rein-
because the weaker top reinforcement is critical. The addi- forced members subjected to torsion. Figure 6.2.3e shows
tion of a flexural moment introduces compression in the the moment-curvature diagrams of two unsymmetrically
weaker top reinforcement and increases its resistance to reinforced hollow specimens tested by Onsongo and Collins
torsional shear stresses. Based on the same experimental (Onsongo 1978). The flexural moment was predominant in
study (Onsongo 1978), it appears that a small axial compres- one specimen (T/M = 0.63), whereas torsion was predomi-
sive force substantially increases post-cracking stiffness of nant in the other (T/M = 4.27). Torsional shearing stresses
a torsionally cracked member with strength also enhanced. were equilibrated by tensile stresses in the longitudinal
This is because the compressive axial load counteracts the direction. Upon cracking, the weaker top reinforcement
widening of spiral cracks. This point may be particularly elongates more than the stronger bottom reinforcement,
relevant in the case of structures under seismically-induced causing greater reverse curvature than that caused by less-
torsion. predominant flexural moment in the torsion predominant
beam (T/M = 4.27). Therefore, the section is subjected to an
Fig. 6.2.3eObserved flexural moment-curvature

relationships at various T/M ratios (Onsongo 1978).
(Note: 1 kNm= 8.85 in.-kip; 1 m = 39.4 in.)
2
T M
r + = 1 (6.2.3a)
To Mo
Fig. 6.2.3dNormalized T-M interaction curves in unsym-
metrically longitudinally reinforced members with trans-
verse reinforcement. Equation (6.2.3b) applies when the weaker top longitu-
dinal reinforcement yields along with the stirrups
overall curvature opposite in direction to the relatively small 2
flexural moment. T 1 M
The T/M ratio affects the diagonal compression angle and T r M = 1 (6.2.3b)
o o

the beam crack pattern. The presence of flexural moment
introduces tensile and compressive strains in the bottom and
Figure 6.2.3g shows the interaction curves for members
top faces of the section, respectively. Compression in the top
with r values of 0.3, 0.5, and 1. The increase in torsional
face delays cracking, in some cases until ultimate strength is
strength calculated using both equations is larger than the
reached (Johnston and Zia 1975). In the case of pure torsion,
experimentally-observed increase shown in Fig. 6.2.3d.
the crack angle on the bottom face is diagonal, and in the
Both equations give acceptable results when pure torsion
case of pure flexure, the crack angle becomes normal to
strength To is taken as the conservatively calculated value, as
the beam longitudinal axis. The angle on the sides is steep
demonstrated by Lampert and Collins (1972).
near the bottom of the section due to tensile strains and
Since 1971, ACI code procedures for the design of members
becomes flatter near the top from the compressive strains
subjected to flexure and torsion are based on the superposi-
caused by flexural moment. Numerous theories (McMullen
tion of reinforcement required to resist torsion (longitudinal
and Warwaruk 1967; Walsh et al. 1967; Lampert and Collins
and transverse) and flexural moment (longitudinal). Experi-
1972; Elfgren et al. 1974a) recognize the phenomenon shown
ence has proved that these procedures are conservative and
in Fig. 6.2.3d. Interaction curves recommended by Lampert
suitable for design. Calculating cross section strength using
and Collins (1972) have been simplified. They suggested that
the code equation is not a simple task, especially in common
in under-reinforced sections, an increase in torsional strength
cases where shear is present and the longitudinal reinforce-
is related to the ratio of the yield force of the compression
ment is not symmetrically placed in the top and bottom sides
reinforcement to tension reinforcement, r = (Asfy/Asfy). The
of the cross section.
flexural moment creates a tensile force M/dv in the bottom
Research by Greene and Belarbi (Greene and Belarbi
stringer and an equal compressive force in the top stringer.
2006a; Greene 2006) expands the STM described in Chapter 5
The torsion induces a total tensile force of Tpo/2Aocotq in the
to members under torsion combined with shear and flexure.
longitudinal reinforcement, where dv is the distance between
The combined-action STM can be used to predict the load-
the top and bottom longitudinal reinforcement, po and Ao are
deformation response for hollow or solid reinforced concrete
the shear flow perimeter and area, respectively, and q is the
members and also to generate torque-flexure interaction
diagonal crack angle. The longitudinal reinforcement forces
curves for members under torsion combined with shear and
are additive, as shown in Fig. 6.2.3f.
flexure. The model considers a softened stress-strain relation-
Two simple equations (Eq. (6.2.3a) and (6.2.3b)) are
ship for concrete and the tension-stiffening effect of concrete
derived based on this theory. Equation (6.2.3a) is for the case
surrounding the reinforcement to predict service-level defor-
where the bottom longitudinal reinforcement yields along
mation (Greene and Belarbi 2006b). Figure 6.2.3h shows the
with the stirrups
interaction curves created using the combined-action STM
Fig. 6.2.3fSuperposition of forces due to torsion and bending (Hsu 1993).
Fig. 6.3.1Shearing stress due to shear, torsion, and

combined shear and torsion.
action curve. ACI code equations were used to calculate To

and Mo for use in Elfgrens model. Figure 6.2.3h(a) shows
that for this test series, ACI 318-11 is unconservative in its
calculation of To, which affects the calculation of the interac-
tion curves.
Fig. 6.2.3gNormalized T-M interaction curves for
members with unsymmetrical longitudinal reinforcement 6.3Torsion and shear
(Lampert and Collins 1972). 6.3.1 GeneralTorsion and shear cause shear stresses
across a section, as shown in Fig. 6.3.1. The shear stresses t
due to the torsional moment T circulate around the section,
whereas the vertical shear force V induces shear stress v. The
side of the section where the stresses are additive (t + v) is
critical in design due to the great intensity of the shearing
stresses. Intense stresses increase tensile strains in transverse
and longitudinal steel reinforcement and compressive and
tensile strains in the concrete.
In spite of its practicality, work on T-V interaction is
relatively limited mainly because shear cannot be induced
without causing a flexural moment along the beams test
region. Therefore, the torsion-shear condition is achieved
only at the point of inflection. Otherwise, an experimentally
observed T-V interaction curve is only a projection of a part
Fig. 6.2.3hInteraction curves predicted by combined- of the T-V-M curve on the T-V coordinate plane. Addition-
action STM and Elfgren Model for McMullen and Warwaruk ally, the pure shear strength of a beam cannot be determined
(1967). Series 1 and 2 (Greene 2006). (Note: 1 kNm= 8.85 experimentally. Typically, this maximum should be defined
in.-kip.) from a calculated value.
6.3.2 Members without transverse reinforcement
for two series of beams under combined torsion and flexure Nylander (1945) reported that tests on members containing
tested by McMullen and Warwaruk (1967). Members of only longitudinal reinforcement showed considerable scatter,
Series 1 in Fig. 6.2.3h(a) were symmetrically reinforced and with most tests falling between a linear and circular interac-
members of Series 2 in Fig. 6.2.3h(b) had more reinforce- tion curve. Other tests by Birkland (1965), Hamilton (1966),
ment on the flexural tension side. For comparison, the model and Ersoy and Ferguson (1968) have shown that a circular
by Elfgren et al. (1974a) was also used to predict the inter- interaction is more accurate. Nylanders model (1945) and
Fig. 6.3.3bVarious normalized T-V interaction curves in

the literature.
Fig. 6.3.3aNormalized T-V linear and circular interaction 6.3.3a). Because the pure shear strength Vo is significantly
curves for members with transverse reinforcement. affected by the accompanying bending moment, it cannot
be accurately represented and a calculated value is used.
those proposed by other researchers conservatively recom-
Different researchers use varying definitions of Vo, making
mend a linear interaction.
it difficult to compare the suggested curves in Fig. 6.3.3b.
Mirza and McCutcheon (1968) tested quarter-scale speci-
Figure 6.3.3a shows that the presence of torsional moment
mens and found that the longitudinal reinforcement has a
reduces shear strength, especially if the torque is more than
significant effect on the interaction curve. The results
25 percent of pure torsional strength. The amount of trans-
scatter was considerable and a lower-bound linear interac-
verse reinforcement is considered the main factor affecting
tion was recommended. McMullen and Woodhead (1973)
the shape of the curve.
tested eccentrically prestressed beams under various combi-
Based on an experimental testing program, Klus (1968)
nations of torsion, shear, and flexure and found that Eq.
suggested a bilinear interaction curve as shown in Fig.
(6.3.2) adequately fit the results
6.3.3b. The theoretical model and the experimental results
2
of Ewida and McMullen (1981) showed the amount of rein-
V T forcement has a considerable effect on the interaction curve.
+ = 1 (6.3.2)
Vo To They suggested a simplified equation

nV
T V
When torsion design provisions were introduced in + = 1 (6.3.3a)
ACI 318-71, a circular interaction between Vc and Tc was To Vo

assumed. This approach slightly increased the design time in
members with transverse reinforcement. Abandoned in ACI
where
318-95, concrete contribution to torsion, Tc, was disregarded
nV = 1.2 for under-reinforced sections in which all rein-
to simplify design. Considerable scatter produced by test
forcement (both longitudinal and transverse) yields
specimens without transverse reinforcement confirms the
when the section reaches ultimate strength
sound principle that members subjected to such combined
= 1.75 for partially under-reinforced sections in which
actions should contain a minimum amount of transverse and
only stirrups yield or only the longitudinal reinforce-
longitudinal reinforcement.
ment yields when the section reaches ultimate strength
6.3.3 Members with transverse reinforcementDevelop-
= 3.0 for completely over-reinforced sections in
ment of models for torsion and shear has made it possible to
which concrete crushes before yielding in any of
calculate shear strength with accuracy similar to the flexure
the reinforcement.
theorys. Early work on T-V interaction was largely experi-
All three curves are shown in Fig. 6.3.3b. For under-
mental. Figure 6.3.3a shows some of the experimental results
reinforced sections (nV = 1.2), the curve is almost linear,
and Fig. 6.3.3b presents some proposed interaction curves.
whereas for completely over-reinforced sections, the inter-
Experimental results were scattered, with most strength
action is closer to a circular curve. Kluss (1968) bilinear
values falling between linear and circular curves (Fig.
interaction curve lies between the linear and circular curves.
Elfgren et al. (1974b) developed a theoretical model for the

ultimate strength of members subjected to combined torsion,
shear, and flexure as discussed in 6.5. The T-V interaction
curve is circular when Failure Modes 1 and 2 govern, but
it is close to a straight line when Failure Mode 3 governs.
These three failure modes were introduced in 3.3.5. Ewida
and McMullen (1981) observed that the addition of a rela-
tively small shear force (20 percent of the shear strength
calculated using ACI equations) caused a slight torsional
strength increase of 3 percent, a phenomenon they called
shear lag. Rahal and Collins (1995a) confirmed the signifi-
cant effect the amount of reinforcement has on the interac-
tion curve. They also showed that in completely over-rein-
forced sections, the addition of shear causes redistribution
in the shear stresses and a subsequent slight increase in the
torsional strength. ACI 318 adopts a superposition approach
in which the reinforcement required to resist T is added to
that required to resist V. To control the diagonal cracks and
ensure that the stirrups yield before crushing occurs in the
concrete, ACI 318 sets an upper limit on the shear stresses
due to torsion and shear.
Figure 6.3.3c (Rahal and Collins 2003b) shows a compar-
ison between the calculations of ACI 318-02 and the 1998
AASHTO LRFD Bridge Design Specifications with test
results from a series of large-scale beams subjected to a Fig. 6.3.3cComparison between observed and calculated
combination of T, V, and a relatively small M (Rahal and T-V interaction curves (Rahal and Collins 2003b). (Note: 1
Collins 1995b). The test setup for this study ensured failure kNm= 8.85 in.-kip; 1 MPa = 0.145 ksi.)
near the point of inflection to accurately represent the inter-
action of torsion and shear. Loads were introduced to the
test region through wing beams instead of by typical concen-
trated loads applied at the beam top. This loading minimized
disturbance in the test region, representing the actual situa-
tion better than the typical setup.
ACI design provisions give conservative results if the
recommended value of q = 45 degrees is used for the angle
of inclination of the diagonal struts. Using the lowest allow-
able value of q = 30 degrees gives less consistent results,
overestimating failure loads at high torsion-to-shear ratios.
Nonetheless, the AASHTO (1998) provisions resulted in a
calculated value for q of approximately 36 degrees for the
series of beams studied. This value compared well with
the inclinations of cracks observed in the tests, producing
better correlation with failure loads. Calculations from ACI
and AASHTO resulted in a nearly linear interaction curve,
with a cut-off horizontal plateau defined by the limit set on
torsional strength to avoid concrete crushing before the rein- Fig. 6.3.3dVariation of angle of diagonal compressive
forcement yields. Variation in the intensity and direction of stresses depending on T/V (Rabbat and Collins 1978).
shear stresses (Fig. 6.3.1) affects the concrete crack patterns
Stresses in the transverse and longitudinal reinforcement
and reinforcement strains. Figure 6.3.3d shows the angle
are also significantly larger on the side where the shear
of diagonal compressive stresses and crack orientation at
stresses are additive (t + v) rather than subtractive (t v).
variable T/V ratios calculated by Rabbat and Collins (1977,
The difference in longitudinal stress magnitude on vertical
1978) using a Variable Angle Space Truss Model. In pure
sides of the section shown in Fig. 6.3.1 causes a lateral
shear, the side cracks are parallel, where in pure torsion, they
curvature, as shown in Fig. 6.3.3e, tested by Rahal and
are nearly perpendicular. Crack orientation on the side where
Collins (1995b). They observed lateral curvatures in two of
shear stresses are subtractive (t v) depends on the rela-
the specimens tested at a T/V of 76 mm (3.0 in.) (Specimen
tive intensity of stresses t and v. This trend was confirmed
RC2-4) and 156 mm (6.14 in.) (Specimen RC2-1). Calcula-
in tests by Rahal and Collins (1995b), Greene (2006), and
tions of the model (Rahal and Collins 1995a) plotted with
Greene and Belarbi (2008).
Fig. 6.3.3eObserved and calculated lateral curvature in Fig. 6.4aTorsion-axial compression interaction in rein-
sections subjected to combined torsion and shear (Rahal forced concrete (tests from Bishara and Peir [1968]). (Note:
and Collins 1995a). 1 in.-kip = 0.113 kNm.)
dashed lines show that available theoretical models capture the descending branch plotted in Fig. 6.4a demonstrate suit-
this phenomenon. ability to capturing the behavior of Bisharas tests. The results
provided (Bishara and Peir 1968), however, were less conser-
6.4Torsion and axial load vative than experimental results from prestressed beams
Elements subjected to combined torsion and compressive without stirrups tested by Humphreys (1957) and Zia (1961).
forces, N, include bridge piers under gravity and horizontal The square root factor in Eq. (6.4a) was labeled the
loads, and prestressed concrete beams subjected to torsion. prestressed factor by Hsu (1968b). He showed that this
Cracked spandrel beams subjected to torsion tend to elon- prestressed factor can be derived theoretically from an
gate. Columns and walls restraining the elongation induce element subjected to shear and compression, assuming
beam compression acting in combination with the compat- the principal tensile stress reached the tensile strength of
ibility torsion and other stress-resultants. Though the combi- concrete. The prestressed factor can also be derived from the
nation of torsional moment (T) and axial force (N) can be skew-bending theory of a beam in pure torsion. To ensure
critical in design for these cases, it has not been adequately best fit with the test results, the constant 12 in the prestressed
studied, particularly in experimental research on nonpre- factor was reduced to 10, resulting in Eq. (6.4b).
stressed elements and combined torsion and axial tension.
Lack of adequate studies is due to the relative rarity of this 1 + 10 ( s fc) (6.4b)
combination compared with others discussed. Torsion tests
on concrete beams subjected to various prestressing levels
simulate conditions similar to those of reinforced concrete The torsional strength based on a 45-degree angle of
beams subjected to combined torsion and axial compres- the compression diagonals and spalled dimensions of the
sion. Experimental results on members without stirrups by section described previously is 2.1 kNm (18.6 in.-kip)38
Humphreys (1957) and Zia (1961) showed trends similar to percent of experimental strength. This strength is increased
those observed by Bishara and Peir (1968), which are given to 4.62 kNm (40.9 in.-kip)84 percent of experimental
in Fig. 6.4a. A drop in torsional strength beyond the trans- strengthif unspalled dimensions are used. For prestressed
formation point was clear only in Zias tests on prestressed members, ACI 318 allows a 37.5 degrees value for the
concrete beams subjected to torsion. Humphreys reported angle of the compression diagonals, leading to a strength
a 170 percent increase in torsional strength and Zia a 190 of 6.05 kNm (53.3 in.-kip). This value, which is plotted
percent increase in torsional strength due to prestressing. in Fig. 6.4a, remains considerably smaller than the experi-
Based on their experimental findings, Bishara and Peir (1968) mentally observed maximum torque of 17.2 kNm (151.9
recommended Eq. (6.4) to calculate an increase in torsional in.-kip) acting in combination with a compressive stress of
strength in the presence of an axial compressive load approximately 0.65fc. The general procedure for the torsion
and shear design method in AASHTO (1998) specifications
s accounts for the effect of compression and prestressing on
T = To 1 + 12 (6.4a) torsional strength.
fc
Figure 6.4a shows calculations for the AASHTO LRFD
method (general method) based on unspalled dimensions,
The descending part is taken as a line joining the transfor- which are more likely for small concrete cover (19 mm [0.75
mation point (defined as the point at s = 0.65fc and Tu = 3To) in.]) used in Bishara and Peirs (1968) tests. Although this
and pure compression strength. Results from Eq. (6.4) and method underestimates an increase in torsional strength at
Fig. 6.4cTorque-axial tension calculated

Fig. 6.4bTorque-twist relationships of reinforced concrete using diagonal compression field theory (Rahal
columns under different compressive stresses (Bishara and and Collins 1995a).
Peir 1968). (Note: 1 in.-kip = 0.113 kNm; 1 in. = 0.0254 m.)
higher compressive loads, it produces slightly better predic-

tions than ACI 318. Figure 6.4a also shows the results of a
detailed analysis based on the modified compression field
theory (CFT) (Rahal and Collins 1995a, 2003a). Similar to
the AASHTO LRFD method (general method), this model
underestimates the torsional strength increase at high
compressive loads while providing generally conservative
results. Figure 6.4b shows the torque-twist relationships for
some tests by Bishara and Peir (1968). Compressive stress
increased torsional strength and torque at peak strength, but
decreased torsional ductility in members with compressive
stresses greater than approximately 0.4fc.
Combined torsion and axial tension in beams and columns
is uncommon. The rarity of this combination and difficul-
ties testing reinforced concrete sections under this forced Fig. 6.5aInteraction surface for torsion, shear, and flexure
combination account for little experimental data on the (Hsu 1993).
subject. Figure 6.4c shows the torsion-axial tension inter- fail in one of three modes. In first mode failure, the bottom
action calculated using diagonal CFT (Rahal and Collins longitudinal reinforcement and transverse reinforcement
1995a, 2003a) for an under-reinforced section. Axial tension yield on the side where shear and torsional stresses are addi-
reduced torsional strength most significantly after tensile tive. In second mode failure, the top longitudinal reinforce-
forces reach 85 percent of pure tensile strength. ment and transverse reinforcement yield on the additive side.
Third mode failure occurs when longitudinal and transverse
6.5Torsion, shear, and flexure reinforcement on the additive side yield. Nondimensional-
In most cases of practical importance, torsion acts in combi- ized interaction relationships for M, V, and T for the three
nation with shear and flexure. Research on pure torsion, torsion failure modes are given by Eq. (6.5a) through (6.5c). Figure
combined with flexure, and torsion combined with shear in 6.5a illustrates the three interaction relationships for the case
members with or without reinforcement is aimed at under- of r = 1/3, where r = (Asfy/Asfy). A systematic derivation of
standing the general behavior of the three stress-resultants Elfgrens nondimensional interaction surface is found in
T, M, and V. Similar to work on the combinations discussed Hsus (1993) book.
above, early research (Hsu 1968a; Johnston 1971; McGee and
Zia 1973; Elfgren et al. 1974a) focused on studying ultimate 2 2
M V T
strength and sought to produce three-dimensional interac- Mode 1: + r + r = 1 (6.5a)
tion surfaces. Further work created the ability to obtain the Mo Vo To

full response of reinforced and prestressed concrete beams
subjected to various stress combinations (Rabbat and Collins 2 2
1978; Rahal and Collins 1995a, 2003a). 1 M V T
Mode 2: + + = 1 (6.5b)
Elfgren et al.s (1974a) theory assumes that a rectangular r M o Vo To

box section subjected to torsion, shear, and flexure could
2 2 and Gabrielsson (1999). Both theoretical and experimental

V T VT 2 dv 1 + r
Mode 3: + + 2 = work has been carried out by Teutsch (1980) and Kordina
VoTo po
(6.5c)
Vo To 2r and Teutsch (1983, 1985).
The beam shown in Fig. 6.5c was loaded at midspan
Researchers in Zrich carried out work in this area. Their with an eccentric point load acting downward. The beam
results are reported in Lchinger (1977), Mller (1976, was rectangular, with dimensions 100 x 200 x 3300 mm
1978), and Thrlimann (1978). Based on their work, Elfgren (3.94 x 7.87 x 130 in.) (width x height x length), and the
(1979) provided a kinematics model that gives the same equa- stirrup strength was 0.236 MN/m (13.5 kip/in.). The rela-
tions, Eq. (6.5a) through (6.5c), derived earlier with equilib- tionship among torsional moment T, vertical shear force V,
rium methods (Elfgren 1972a,b; Elfgren et al. 1974a,b). The and flexural moment M in the failure section was M:T:V =
kinematics model is illustrated in Fig. 6.5b and Fig. 6.5c. 0.1:0.5:0.2 (Elfgren 1979). The numerals shown along the
Beam deformations under combined torsion, shear, and cracks in Fig. 6.5c refer to the applied load when this part of
flexure were studied by Petersson (1972), Karlsson (1973), the crack became visible (units in multiples of 10 kN [2.25
kip]). In the left beam, two failure cracks, ABC and FED,
are indicated as well as a rotation hinge AD (compare with
Fig. 6.5b).
The combined-action STM (Greene and Belarbi 2006a,
2009a,b) can also generate torsional-flexural moment inter-
action curves for members under torsion combined with
shear and flexure. Figure 6.5d shows the interaction curves
for two series of beams under combined torsion, shear, and
flexure tested by McMullen and Warwaruk (1967). Members
of Series 5 and 6 were unsymmetrically reinforced with more
longitudinal reinforcement in the bottom side. For compar-
ison, the model by Elfgren et al. (1974a) was also used to
predict the interaction curve. ACI 318 code equations were
used to calculate To, Mo, and Vo for use in Elfgrens model.
Figure 6.5d(a) shows that for this test series, ACI 318-11
is unconservative in its calculation of To, which affects the
calculation of the interaction curves, similar to the observa-
tion related to Fig. 6.2.3h(a).
Leonhardt et al. (1968) performed large-scale tests on two
prestressed concrete box beams representing hollow box
bridges loaded in torsion, shear, and flexure. Figure 6.5e
shows crack patterns after failure. The beams had a depth
of 0.78 m (2.6 ft) and a span of 6.00 m (19.7 ft), and web,
top, and bottom flange widths of only 80 mm (3.1 in.). The
beams were eccentrically loaded at a midspan diaphragm,
Fig. 6.5bKinematics failure model: (a) General view; and the bottom slab reinforcement was varied between an
(b) Model seen from above; (c) Deformations in bottom; orthogonal mesh and mesh inclined at 45 degrees. In both
(d) Bending moment diagram (Elfgren 1979). beams, the web near the eccentrically applied load failed in
Fig. 6.5cCrack pattern and failure mechanism for a beam loaded in combined torsion,
shear, and flexure (Elfgren 1979).
inclined compression (Fig. 6.5e(a)), confirming the impor-

tance of determining the strength of struts considering the
web strain and cracking conditions.
CHAPTER 7ADDITIONAL DESIGN ISSUES

RELATED TO TORSION
7.1General
Chapter 7 addresses additional issues encountered in the
design of structural concrete members subjected to torsion.
When considering torsion, it is necessary to distinguish
between when the torsional moment cannot be reduced by
redistribution of internal forces (equilibrium torsion) and
Fig. 6.5dInteraction curves predicted by combined-action when the torsional moment can be reduced by redistribu-
TM and the Elfgren model for McMullen and Warwaruk tion of internal forces (compatibility torsion) after cracking.
(1967) Series 5 and 6 (torsion, shear, and flexure) and Spec- To allow torsional moment redistribution, a limit analysis
imen 1-1 (pure torsion) (Greene 2006). (Note: 1 kNm= 8.85 method similar to moment redistribution is addressed. This
in.-kip; 1 m = 39.4 in.) method requires the formation of torsional plastic hinges at
critical sections. Also addressed are precast spandrel beams,
torsion limit design, treatment of open concrete sections
subjected to torsion, and the size effect on torsional members.
7.2Compatibility torsion and torsional moment

redistribution
7.2.1 Basic concepts of torsional moment redistribution
7.2.1.1 GeneralTorsion and flexural moment redistribu-
tion are similar in their dependency on adequate ductility in
plastic hinge regions. No fundamental difference, therefore,
exists between flexural analysis for the case of pure flexure
and of torsion. Three standard cases of torsion redistribution
have been reported:
1) A spandrel beam loaded by a floor beam
2) An L-beam
3) A spandrel beam in a frame subjected to lateral load.
7.2.1.2 Spandrel beam loaded by floor beam (Collins and
Lampert 1973; Hsu and Burton 1974; Hsu and Hwang 1977;
Abul Mansur and Rangan 1978)Figure 7.2.1.2 shows a
portion of a three-dimensional structural frame, including
columns, spandrel, and floor beams. A load P, applied to the
floor beam, produces end rotation, which in turn produces
a torsional moment in the spandrel beam. The interaction
of floor beam and spandrel beam can be studied using the
shaded portion in the shape of a T-specimen. This T-specimen
is cut at the flexural inflection points, which can be simulated
by hinges. When the T-specimen is loaded, a concentrated
torsional moment will occur at the spandrel beam midspan.
Whereas cracking reduces flexural stiffness by a factor of
approximately 2, it reduces torsional stiffness by a factor of
approximately 10. Therefore, a significant redistribution of
internal actions begins to occur at cracking. When the span-
drel beam develops torsional hinges at its ends, additional
load will result in higher flexural moment in the floor beam.
This response is equivalent to moment redistribution from
the spandrel to floor beam. Designing the spandrel beam
based on this redistribution could be economical.
7.2.1.3 L-beam (Bishara and Londot 1979)Figure
Fig. 6.5eCrack pattern and failure of prestressed hollow 7.2.1.3 shows a floor system with an L-shaped spandrel beam.
box-beam BM 2 with high shear and torsion loaded eccen- Bishara and Londot (1979) studied the response of L-beams
trically (Leonhardt et al. 1968). under combined torsion and flexure. Due to torsional stiff-
Fig. 7.2.1.2Spandrel beam under concentrated

torsion (Hsu and Burton 1974).
Fig. 7.2.1.4aSpandrel beam under lateral

sway (Pantazopoulou and Moehle 1990).
tioned eccentric to the centroidal axes of the beam elements

and connected to the beams through short rigid segments.
For any curvature distribution along the longitudinal or main
Fig. 7.2.1.3Floor system with spandrel beams as identified in the figure, this configuration creates
beams. a corresponding elongation at mid-depth of the slab. Truss
forces are transferred to the transverse beam by modeled
ness of the spandrel beam, the slab cannot be assumed to truss elements in the slab, creating a torsional lever arm in
be hinged at the spandrel beam, and vertical loading on the the transverse (spandrel) beam.
slab causes distributed compatibility torsion in the spandrel The result of these stresses on the spandrel beam (the
beam. When torsion redistribution occurs, a plastic hinge transverse beam in Fig. 7.2.1.4b), is twofold: 1) flexural
forms at a yield line that defines the slab-beam connection. moments in the slab plane develop in the spandrel beam;
This could potentially result in a torsional moment that and 2) torsional moments develop in the spandrel beam as
affects the spandrel beam and that could be lower than that a result of the eccentric action of the slab relative to the
computed by elastic analysis. These tests also showed that shear center of the spandrel beam cross section. Torsion and
before yielding, torsional cracking caused a reduction in flexure in the spandrel beam are maximum at the column,
both torsional and flexural stiffnesses, leading to redistribu- and decrease with distance from the joint. This combina-
tion of internal forces. Reductions in flexural stiffness range tion of torsion and flexure in the spandrel beams of exte-
from 40 to 60 percent to as much as 80 to 90 percent for rior connections is commonly responsible for the diagonal
torsional stiffness in beams with transverse reinforcement. cracking often observed on the exterior and interior faces
7.2.1.4 Spandrel beam in a frame subjected to lateral of such beams. These actions can also be tested as shown in
load (Pantazopoulou and Moehle 1990)When a typical Fig. 7.2.1.4c. Two separate actions can cause moment redis-
building frame is subjected to lateral sway, beam flexural tribution in this case:
action results in contraflexure along the span length, as shown 1. Plasticity could occur in the slab under axial load, a
in Fig. 7.2.1.4a. Therefore, in each bay, there is a region condition often idealized as plastic deformation in the truss
where the slab is in tension and another where the slab is in members modeling the slab of Fig. 7.2.1.4b. Alternatively, the
compression. The beam is likely to respond asymmetrically. slab could reach its maximum load strength and redistribu-
The surface of the floor beam or main beam elongates as the tion could occur before reinforcement in the spandrel beam
beam is subjected to negative moment. The same elongation yields (Collins and Lampert 1973; Pantazopoulou and Moehle
develops in the slab at the slab-beam interface. Elongation 1990). This redistribution could result in less torsion on the
distorts the slab in shear to satisfy deformation compat- spandrel beam than that computed from elastic analysis; and
ibility and the slab reinforcement develops tensile stresses 2. A torsional plastic hinge could form at a location across
that are transferred to the spandrel (transverse) beams. These the spandrel beam axis, resulting in redistribution of torsion
actions are illustrated by an equivalent structure shown in along the spandrel beam axis. For example, formation of the
Fig. 7.2.1.4b. Slab action is modeled as truss elements posi- plastic hinge at the connection between the spandrel beam and
Fig. 7.2.1.4bIdealization of spandrel beam under lateral sway (Pantazopoulou and

Moehle 1990).
Fig. 7.2.1.4cTesting of spandrel beams under lateral sway (Pantazopoulou and Moehle 1990).
the column (location of maximum torsion in the elastic anal- 7.2.4 Existing code statusIn ACI 318, a flexural moment
ysis) would reduce the torsion affecting the spandrel beam. redistribution formula has been maintained since 1971.
7.2.2 Design benefits from torsion redistributionPlastic In 1977, ACI 318 introduced the torsional limit design,
hinges and moment redistribution permit full strength use allowing the torsional plastic hinge to have a torsional shear
of the strength of the flexural members cross section. With stress of 0.33 fc MPa (4 fc psi). Alternatively, Section
torsion, elastic analysis typically results in high torsional 8.6.1 of ACI 318-77 specifies that any set of reasonable
moments, which in turn require both stirrups and longitu- assumptions can be used for computing relative torsional
dinal reinforcement beyond minimum requirements. Torsion stiffnesses of structural elements. Commentary indicates
redistribution can reduce design torsional moments, allowing that member stiffnesses should reflect the degree of cracking
for more economical designs. and inelastic action occurring along the member length
7.2.3 Disregarding structural torsionAlthough one
before yielding. The stress limit of 0.33 fc MPa (4 fc
common practice is to ignore spandrel beams torsional stiff-
psi), however, has been extended to prestressed and nonpre-
ness and similar members under gravity loads, effectively
stressed members subjected to an axial tensile or compres-
assuming full torsional redistribution, it is also common to
sive force in the subsequent editions, including ACI 318-11.
disregard spandrel beams torsional moments resulting from
In commentary Section 8.6.1 of ACI 318-77, in the case of
wind or earthquake actions. Sections 7.2.4 through 7.2.6
compatibility torsion (torsion not required to satisfy equilib-
address codes, experimental evidence, and discussions of
rium), torsional stiffness can be disregarded. Section 8.6.1
their limitations.
commentary effectively permits up to 100 percent torsional
moment redistribution from the elastic analysis without M (7.2.5.1b)

relating it to specific ductility requirements to satisfy a high T=
2
redistribution value. The gap between flexural and torsional
redistribution provisions is due primarily to limited research
in literature for torsional moment redistribution. There are In each series, the spandrel beam torsional moment is
three areas where the gap is manifested: calculated assuming varying ratios between the spandrel
1. Testing relating torsional redistribution to member beam torsional stiffness and the floor beam flexural stiffness.
ductility is limited These ratios varied from zero to 100 percent torsional redis-
2. Factors affecting torsional ductility and, therefore, the tribution. The latter condition is where torsional moment
permissible torsional redistribution, are not understood does not affect the spandrel beam, but is redistributed to
3. The concept of permissible torsional redistribution not positive flexure in the floor beam. Figure 7.2.5.1a shows the
affecting service load deflections has not been experimen- results for Series A tests and Fig. 7.2.5.1b for Series B.
tally verified. Three pairs of companion specimens from this testing
In the following sections, literature available in these three program are examined: A1 and B1, A3 and B2, and A5 and
areas is reviewed and specific recommendations for needed B3. Specimens A1 (Fig. 7.2.5.1a) and B1 (Fig. 7.2.5.1b)
research are given. were designed assuming uncracked stiffness of both floor
7.2.5 Torsion redistribution experiments and spandrel beams. Specimens A3 and B2 were designed
7.2.5.1 Early testsCollins and Lampert (1973) tested six according to the limit design theory, assuming the joint
spandrel beam specimens with a concentrated load applied moment to be 48 percent of that calculated from the elastic
at midspan of the floor beam. They found that specimens moment distribution using the uncracked stiffnesses of the
designed assuming the members had zero torsional stiffness floor and spandrel beams (and therefore 52 percent torsional
behaved as satisfactorily as specimens designed assuming moment redistribution). Specimens A5 and B3 were
uncracked stiffness values. They also showed that the ratio designed by assuming 100 percent redistribution, meaning
of torsional-to-flexural stiffness will drop at cracking, the spandrel beam in these two specimens was not designed
causing redistribution of the torsion and flexural moments. to sustain any torsion.
They concluded that for compatibility torsion, design should 7.2.5.2 Strength and adequate values of torsion redistri-
be for a twist and not a torque, and the main function of butionWhen comparing torsional ductility and redistribu-
torsional reinforcement is to distribute the cracks caused tion among the three specimen sets, review the experimental
by twist. The compatibility torsion magnitude is over- load-torque relationships. These curves are presented in
estimated if gross stiffness is used. If zero torsional stiff- Sets A (Fig. 7.2.5.1a) and B (Fig. 7.2.5.1b). Curves for Spec-
ness is assumed, design procedures become simple, and imens A1 and B1 (no torsional redistribution) can be divided
only minimum torsional reinforcement is needed to ensure into three stages:
ductility and limit crack width. 1. The first stage represents elastic behavior before cracking.
Hsu and Burton (1974) focused on addressing three ques- 2. The second stage is where crack development caused
tions for spandrel beams subjected to torsion: the torsional stiffness to decrease and the torsional moment
1. Would torsional strength be reached if a spandrel beam remained essentially constant whereas the load increased.
were designed by the ACI philosophy of neglecting torsional When the load reached a magnitude at which the floor beam
bottom longitudinal reinforcement yielded, as indicated by
stiffness and assuming 100 percent redistribution?
in the figures, the third stage began.
2. Can other values of torsion redistribution be used
3. In the third stage, a torsional plastic hinge developed
effectively?
under the load and accelerated the twist deformation of the
3. How serious is service load cracking?
spandrel beam. A new load-carrying mechanism evolved in
To address these questions, Hsu and Burton (1974) tested
which the load was primarily sustained by the spandrel beam
two series (A and B) of spandrel beams:
torsional strength. The torsional moment again increased,
1. Series A: Concentrated load at midspan of the floor beam.
primarily resulting in a slight increase in the load-carrying
2. Series B: Uniform loads on the floor beam simulated by
strength of the assembly. Spandrel beam stirrups and longi-
four concentrated loads. Based on elastic stiffness analysis,
tudinal bars then yielded before reaching assembly failure.
the torsional moment distribution to spandrel beam at its joint
Specimens A3 and B2 (52 percent redistribution) exhib-
with the floor beam is calculated according to Eq. (7.2.5.1a)
ited behavior similar to that of Specimens A1 and B1,
3 despite the assumed higher value of torsion redistribution.
P f In both Specimens A3 and B2, the load-torque curve had the
M = 16 (7.2.5.1a) same three stages exhibited in Specimens A1 and B1. Both
3 Kf
1+ specimens also exhibited the anticipated level of torsion
4 Ku redistribution. Although A3 and B2 failed at almost the same
ultimate failure loads as Specimens A1 and B1, the failure
Assuming that spandrel beam is torsionally fixed at both load corresponded to about 50 percent less torsion at the
ends, the spandrel beam torsional moment magnitude is joint between the floor beam and the spandrel beam. These
results indicate the torsional redistribution assumptions for
www.TeraStandard.com
American Concrete Institute Copyrighted Materialwww.concrete.org
Fig. 7.2.5.1aSeries A (Hsu and Burton 1974). (Note: 1 in.-kip = 0.113 kNm; 1 kip =
4.45 kN.)
Fig. 7.2.5.1bSeries B (Hsu and Burton 1974). (Note: 1 in.-kip = 0.113 kNm; 1 kip =
4.45 kN.)
these two specimens closely matched the design assump- load increased significantly, almost doubling, at the same
tion of 52 percent torsional redistribution. Specimens A5 torsion value until failure. Failure occurred at a load far
and B3 were designed for 100 percent or complete torsion below design strength, indicating that the design assumption
redistribution, as implied in ACI 318 commentary. As shown for torsion redistribution was invalid. Figures 7.2.5.2a and
in Fig. 7.2.5.1a and 7.2.5.1b, torsion resisted at the joint 7.2.5.2b show the torque-versus-measured twist deformation
stayed constant after cracking occurred. After cracking, the for Series A and B, respectively. Both figures support conclu-
Fig. 7.2.5.2aSeries A torque-twist relationship (Hsu and Burton 1974). (Note: 1 in.-kip
= 0.113 kNm; 1 in. = 0.0254 m.)
Table 7.2.5.3Maximum crack width at service increase may be within the measurement tolerance. Speci-
load mens B1 and B2 had the same crack widths. These tests indi-
Design torsion Maximum crack cate that significant torsion redistribution can be performed
Specimen redistribution, % width, in. (mm) without affecting serviceability. Nevertheless, the results
A1 0 0.004 (0.10) were not sufficient to support firm conclusions.
A3 52 0.005 (0.13) Following Hsu and Burtons (1974) approach, Hsu and
A5 100 Hwang (1977) tested T-shaped specimens with short span-
B1 0 0.005 (0.13) drel beams to study the moment distributions between floor
B2 52 0.005 (0.13) beam and spandrel in the case of high shear-to-torsion
B3 100 ratio. They found that the torsional limit design method
can be applied to spandrel beam end regions, adjacent to
the columns, where high shear stresses are combined with
sions similar to those previously mentioned. Specimens A1 torsional moments. As a result, the torsional limit design
and B1 exhibited high ductility, as indicated by the torque- was added to ACI 318-77 code. Torsional hinge moment was
twist relationship. Specimens A3 and B2 exhibited a twist
calculated based on a torsional shear stress of 0.33 fc in
angle 50 percent greater at failure than that of Specimens A1
and B1, which roughly corresponds to the designed torsion MPa (4 fc in psi). Conclusions reached by Hsu and Burton
redistribution. Specimens A5 and B3, designed for 100 (1974) were validated by Abul Mansur and Rangan (1978),
percent torsion redistribution, exhibited far lower twisting who conducted experiments on seven beams with configu-
ductility. In fact, both specimens failed at 0.3 103 degrees/ rations shown in Fig. 7.2.5.3. In a related set of experi-
mm (8 103 degrees/in.), far less than the angles exhibited ments, Abul Mansur and Rangan (1978) studied the effect of
by Specimens A1 and B1. applying a concentrated load directly on the joint. This work
7.2.5.3 Cracking service loadHsu and Burton (1974) showed that applying a concentrated load in this fashion did
defined service load as one-half the ultimate design load or not affect torsion redistribution results, which were gener-
actual ultimate test load, whichever was larger. Table 7.2.5.3 ally in agreement with Hsu and Burtons work (1974).
shows the maximum crack widths measured at the service 7.2.6 Factors affecting torsion redistributionSection or
load for specimen Series A and B. No crack width readings member ductility is a primary factor affecting permissible
were recorded while testing Specimens A5 and B3 because redistribution. With flexure, ACI 318-95 relates ductility
these two specimens failed prematurely. to the amount of longitudinal reinforcement present in the
A comparison of Specimens A1 and A3 shows that section. A similar torsional ductility reinforcement ratio
maximum crack widths increased only slightly for the 52 would be appropriate, accounting for transverse and longi-
percent torsional redistribution assumption. This slight tudinal reinforcement areas coupled with torsion support
Fig. 7.2.5.2bSeries B torque-twist relationship (Hsu and Burton 1974). (Note: 1 in.-kip
= 0.113 kNm; 1 in. = 0.0254 m.)
conditions. Before 1995, ACI 318 adopted the skew-bending

theory for idealizing member behavior under torsion. The
approach was complex and produced several inconsistencies,
which led to the introduction of space truss analogy to ACI
318-95. Before this change in philosophy, when the skew-
bending theory was prevalent, Chakraborty (1977) proposed
a balanced torsional reinforcement concept. He used a space-
truss analogy for developing the balanced reinforcement
concept, referring to early work on the space-truss analogy
by Mitchell and Collins (1974). Chakrabortys work was Fig. 7.2.5.3Test configuration (Abul Mansur and Rangan
the subject of commentary in several professional journal 1978). (Note: 1 mm = 0.0394 in.)
discussions, most notably by Collins and Mitchell (1978).
Discussions focused primarily on the difficulty of using balanced reinforcement, was not adopted by ACI 318 when
Chakrabortys approach and an apparent difficulty in deriving the committee changed to the space-truss analogy in 1995.
similar conclusions from the specimens tested. These difficul- Numerous articles have reviewed ductility and service-
ties stemmed primarily from the fact that Chakrabortys (1977) ability, and their relationship to the angle q. The crack
approach considered both the truss-analogy from the modified width model of Thrlimann et al. (1983) is shown in Fig.
compression field theory (CFT) of Mitchell and Collins, and 7.2.6a. This model minimizes strains in the entire concrete
compatibility between the concrete and reinforcement. This medium. Thrlimann also suggested that within these limits
approach resulted in a complex equation (Eq. (7.2.6)) for the of strut inclination, the maximum diagonal compressive
balanced stirrup reinforcement area for one leg. The equation stress is independent of the angle of inclination. Further, the
proposed by Chakraborty is minimum crack strain corresponds to an inclination angle
of 45 degrees that in turn corresponds to the simultaneous
0.85 fcb1 yielding of both the longitudinal (main) and transverse (stir-
Atb ph 0.003Es rups) reinforcement (Fig. 7.2.6b).
= 2
Their analysis was based on shallow beams to which
Ao s fty b
0.003Es 1 1 po (7.2.6) Bernoullis hypothesis of plane strain is applicable. In such
2
fty + beams, the elastic stress distribution implies that maximum
ph diagonal compression stress is inclined at 45 degrees to the
beam axis. Another interpretation of angle limits suggested
The above approach of combining both compatibility and by Thrlimann et al. (1983) is that as the truss-model stress
the space-truss analogy to compute torsional strength, or field deviates from the elastic stress field that the crack strains,
Fig. 7.2.6cEffect of torsion-flexure interaction on ductility

Fig. 7.2.6aCrack width model (Thrlimann et al. 1983). (Bishara and Londot 1979).
Fig. 7.2.7Torsion in spandrel beam

Fig. 7.2.6bCracking strain and inclination angle q (Thr- due to lateral load (Pantazopoulou and
limann et al. 1983). Moehle 1990). (Note: 1 in.-kip = 0.113
kNm; 1 in. = 0.0254 m.)
thereby increasing crack widths. Ali and White (1997) showed
that for disturbed (D) regions in general, where the Bernoulli beams, however, is needed to understand the relationship
hypothesis is not applicable, large deviations from the elastic between permissible moment redistribution and q. Under-
stress distribution result in loss of ductility and serviceability standing the direct relationship between q and ductility is
due to increased concrete strains. This deviation corresponds also essential. It is important to include the interaction
to large deviations from the 45-degree angle between diag- among torsion, shear, and flexure and its effect on ductility.
onal compression and longitudinal reinforcement for shallow The beam configuration in Fig. 7.2.1.2 (described in more
beams. The interaction between torsion and flexure has been detail in Hsu and Burton (1974)) is ideally suited for such a
shown to affect flexural ductility of members subjected to study. Curves similar to those in Fig. 7.2.6c, once developed,
both torsion and flexure. An extensive testing program of could be readily used in code provisions. Simple provisions
L-beams conducted by Bishara and Londot (1979) showed for calculating the slab-action torsion developed in a span-
that increased torsional loads resulted in reduced flexural rota- drel beam under lateral loads (Pantazopoulou and Moehle
tions. This reduced flexural rotation is shown in Fig. 7.2.6c, 1990) would be equally beneficial.
where the flexural rotation is plotted with respect to a normal- Tests by Pantazopoulou and Moehle (1990) showed that,
ized torsional stress. due to slab action, torsional moments can develop in the
7.2.7 Needed researchAs in the case of flexure, a spandrel beam. Figure 7.2.7 shows flexural and torsional
preferred design procedure would specify the relationship moment distribution and the corresponding deflection and
between permissible moment redistribution and measure of twist along the longitudinal axis of a spandrel beam due to
ductility. To allow for complete yielding of reinforcement, slab action under lateral loads. The left side shows measured
ACI 318 limits the angle q in the range of 30 to 60 degrees. flexure with moment at the top and deflection at the bottom,
The angle q is generally taken as 45 degrees for reinforced and the right side shows torque with torque and twist at
concrete beams and 37.5 degrees for prestressed beams. the bottom. The supported end is to the right of each graph
A comprehensive study of prestressed and nonprestressed whereas the free end is to the left.
Fig. 7.3.1aTypical L-shaped spandrel beam (Lucier et al. 2007). (Note: 1 in. = 0.0254 m.)
7.3Precast spandrel beams

7.3.1 GeneralTo this point, the discussion has focused
on monolithic, cast-in-place concrete structures in which
torsional moment redistribution could occur when members
subjected to torsion are designed with adequate ductility.
The situation is different in precast structures where precast
members, either reinforced or more commonly prestressed,
are usually simply supported and, without continuity, there
is no torsional moment redistribution. These members are
designed for the full torsional moment to maintain equilib-
rium. Precast L-shaped spandrel beams, which are common
to their type, are characterized by a plate- or wall-like web
with a continuous ledge running along the bottom of one web
side. This ledge provides deck beam support, as shown in Fig.
7.3.1a. A variation on the L-shaped spandrel beam is the corbel
Fig. 7.3.1bEquilibrium torsion in T-beam with unbalanced
spandrel beam on which the continuous ledge is replaced by
load.
a series of discrete corbels that support deck beams. Another
variation is the pocketed spandrel beam in which the discrete widely accepted as an alternative to cast-in-place concrete
corbels are replaced by a series of pockets (recesses) cast in construction. Before publication of a paper by Zia and
the spandrel that support deck beams. All these spandrel beam McGee (1974), the primary guidance on torsion design for
types are subjected to significant torsion caused by the series engineers were recommendations developed by former ACI
of large, concentrated, eccentric loads along the span. Committee 438 in 1969. The ACI 438 report formed the
Typically, L-shaped spandrel beams are 1.5 to 2.1 m (5 to basis for the first comprehensive torsion design provisions to
7 ft) deep with spans ranging from 9.1 to 15.2 m (30 to 50 ft) be included in the ACI code (ACI 318-71). The Zia-McGee
in length. They are usually simply supported at the columns. paper extended the ACI code torsion provisions to prestressed
Beam ends are connected to columns to prevent torsional concrete. In 1978, Zia and Hsu updated the Zia and McGee
rotation. Deck beams are often connected to the spandrel (1974) paper to conform to ACI 318-77 code format and
web to provide lateral restraint along the span. These end and also introduced refinements. Papers by Zia and Hsu (1978,
intermediate connections greatly affect the L-shaped span- 2004) were widely used by the precast prestressed concrete
drel beam torsional behavior. Eccentrically applied loads industry for nearly 30 years for its reliability and ease of use.
on unsymmetrical L-shaped spandrel beams cause vertical The ACI 318 code and the Zia-Hsu approach present a unified
displacement as well as significant lateral displacement and method for the torsion and shear design of prestressed and
rotation. The torsional and shear effects are largest near the nonprestressed concrete flexural members following similar
spandrel end. This complex structural behavior, coupled design procedures. The ACI method was developed with a
with heavy loadings, often results in heavy reinforcement primary focus on compact closed sections generally found
in the end regions. Similar significant torsion effects may in cast-in-place reinforced concrete structures and was based
occur in inverted T-beams with severely unbalanced loads, on a thin-walled tube, space-truss analogy. The Zia-Hsu
as shown in Fig. 7.3.1b. method was developed with a primary focus on noncompact
7.3.2 Torsion designDesign for torsion in precast flanged sections more common in precast structural systems
concrete members received much attention in the late 1950s and the skewed-bending concept. In cast-in-place structures,
and early 1960s when precast prestressed concrete became the torsional effect is often minimized due to stress redis-
tribution in an indeterminate system, whereas in precast 4. Reinforcement requirements on the inside face and
structures, the torsional effect is more severe and connec- outside face may be considered separately; shear increases
tion details can significantly affect the torsional behavior the required reinforcement on the inside face and decreases
of structural members (Klein 1986; Raths 1984). The ACI it on the outside face.
torsion design procedure produces design results compa-
rable to those offered by the Zia-Hsu method (Prestressed 7.4Torsion limit design
Concrete Industry (PCI) 1999), which was updated based 7.4.1 Basic concepts of limit designThere are two
on the Zia-McGee method (PCI 1978) developed from the fundamental limit design theories (Gvozdev 1938, 1960;
tests of small laboratory specimens. The aspect ratio of their Melan 1938; Horne 1949; Lubliner 1990):
component rectangles rarely exceeded 1:3, in contrast to 1. Lower-bound approach
large precast members such as L-shaped spandrel beams. 2. Upper-bound approach
Large spandrel beams may develop plate-bending action in The lower-bound approach requires a stress field that
addition to torsion, and therefore behave differently from the satisfies equilibrium everywhere and that does not violate
small laboratory specimens. The Zia-Hsu method has been appropriate yield conditions or material failure criterion at
calibrated by two tests (Klein 1986) of full-size members any location.
with support conditions similar to those used in practice. The classic example of lower-bound approach is the truss
Torsion behavior of slender spandrel beams in these tests model in which a stress field is assumed and the ultimate
was much different from compact beams. Spandrel beams strength is calculated accordingly. For members made of
showed no signs of cover spalling or stirrup debonding, for plastic materials (those that satisfy the basic theory of plasticity
which closed stirrups are required. End region behavior of assumptions), the resulting ultimate strength is easily proved to
slender precast spandrel beams is dictated by connection provide a conservative, lower-bound solution to the true ulti-
details. Upper horizontal reaction prevents torsional rotation mate strength of the structural member under consideration.
and causes out-of-plane web bending. Vertical shear force In the upper-bound approach, a strain field or failure
from the reaction is carried across the same diagonal crack. mechanism is developed that does not violate strain compat-
Out-of-plane bending caused by horizontal forces is essen- ibility conditions or appropriate yield conditions and that
tially the same as bending produced by a concentrated load does not exceed material failure criterion at any location.
or reaction in the corner of a structural slab. Tests conducted A classic example of the upper-bound approach is the
by Logan (2007) and by researchers at North Carolina State yield line theory of slabs in which a failure mechanism is
University (Lucier et al. 2007; Hassan et al. 2007) confirmed assumed and the ultimate strength is calculated accordingly.
that out-of-plane bending dominates torsional response in For members made of materials satisfying basic theory of
slender spandrel beams. Lucier et al. (2007) evaluated the plasticity assumptions, the resulting ultimate strength can
need for closed ties in slender spandrel beams using full- easily be proved to provide an upper-bound, or unconserva-
scale tests. Twelve precast spandrel beams, each nearly 14 m tive, solution to the true ultimate strength of the structural
(15.3 yd) long, were tested to failure in the laboratory. End member under consideration. Exact solutions are obtained
regions of the beams were designed for plate-bending and if the results of the upper-bound and lower-bound solutions
shear, with and without closed ties. Several beams failed in are identical.
flexure or ledge punching. The ledge punching failures were The advantage of applying either the lower- or upper-
primarily due to interaction with global tension and shear bound approach lies in their simplicity. Relaxation of either
forces in the ledge. Beams with reinforcement to prevent equilibrium or strain compatibility conditions in solving the
such flexure and ledge failures failed due to combined torsion governing equations significantly simplifies the solution
and shear. All torsion/shear failures occurred just inside the procedure. There are, however, disadvantages:
bearing reaction along a failure plane inclined at an angle of 1. The potential exists for large over- or under-estimation
approximately 45 degrees. Although the beams with closed of true ultimate strength; and
ties performed somewhat better, all of the beams sustained 2. No information is provided for important quantities,
test loads well in excess of their calculated nominal strengths. such as deflections and rotations at service or failure load
These test results suggest that slender spandrel beams can be limit states; these are quantities that require classical deflec-
designed based on the following four principles: tion analysis considering all conditions of equilibrium,
1. Torsion acting on the 45-degree failure plane can be compatibility, and constitutive relations governing the mate-
divided into plate-bending and twist components, each equal rial behavior.
to Tu/2. Research in limit design faces the following challenges:
2. Plate-bending component of torsion requires equal 1. The appropriate stress or strain field needs to be derived
amounts of longitudinal and transverse reinforcement evenly and validated from ultimate strength calculations that are
distributed over the height of the beam. neither too conservative nor too unconservative for the cases
3. The twist component of torsion is resisted by out-of- of lower and upper bounds, respectively.
plane shear stresses, which are greatest near the sections top 2. Parameters should be developed for the stress or strain
and bottom, as predicted by Saint-Venants equations and the field such that optimization techniques can be used to achieve
soap bubble analogy. Closed ties are required only where realistic lower and upper ultimate strength estimations.
the concrete section cannot safely resist the twist component.
3. Both the lower- and upper-bound approaches should

be recognized as valid for plastic materials; any approach
is valid if it satisfies the basic assumptions of the theory of
plasticity of either perfectly rigid plastic or linear-elastic,
perfectly-plastic materials. In both cases, the material should
ideally exhibit unlimited strains under the same yield stress.
Reinforcementin particular, mild reinforcementexhibits
behavior closely satisfying these requirements. Certain
assumptions and limitations, however, should be applied to
achieve proper application of the plasticity theory to struc-
tural concrete members.
The first two challenges are explored in 7.4.2 and 7.4.3
because they pertain to reinforced concrete members
Fig. 7.4.2.2Elevation of helical stair (Rangan et al.
subjected to torsion. The third challenge is far-reaching, as it
1978).
relates to the general applicability of the theory of plasticity to
plain concrete and reinforced concrete members. A thorough
discussion of this challenge is provided in Chen (1982). loading. Elastic theory indicated that a helical stair is inde-
7.4.2 Lower-bound approach terminate to the sixth degrees. When the loading is symmet-
7.4.2.1 GeneralThe lower-bound method requires a rical, however, the resultant-stress field is fully described in
stress field that satisfies equilibrium everywhere and does terms of only two redundancies: Mo and Ho. The term Mo
not violate the appropriate yield conditions at any point. is the flexural moment acting in a tangential plane, and Ho
From a design perspective, this approach means that if every is the horizontal force in the radial direction, both acting at
region of a structure is proportioned and detailed for a stress- midspan, as shown in Fig. 7.4.2.2.
resultant field that satisfies all equilibrium conditions, then For this problem, six equilibrium equations were first
the structure should safely carry the design ultimate load. formulated in terms of Mo and Ho, along with reactions at the
A well-known example of this approach is the Hillerborgs two supports. In their paper, Rangan et al. (1978) assumed the
(1960) strip method for reinforced concrete slab design. maximum torsional moment to occur at an angle F = g1 on the
Rangan et al. (1978) proposed a closed-form, lower-bound helical stair. The maximum To is therefore expressed as
method for the torsion design of reinforced concrete struc-
x2 y
tures with the following steps: (T f )F = g 1 = To = 0.33 fc ( N-mm)
1. Set up all equilibrium equations in a statically indeter- 3
minate structure in terms of the redundancy. Let the number x2 y (7.4.2.2a)
of redundancies be represented as n. (T f )F = g 1 = To = 4 fc (in.-llb)
3
2. Assign the maximum torsional moment, which is
normally near a torsionally-restrained support, equal to the
Development of another equilibrium equation required
limiting value To defined as
solving for the redundancies. The moment Mrf was assumed
to be zero at F = g2 (inflection point). The resulting equa-
To = 0.33 fc( x 2 y / 3) (N-mm) tions were in the form
(7.4.2.1)
To = 4 fc( x 2 y / 3) (in.-lb)
w = f(g1,g2) (7.4.2.2b)
Here, x and y are the shorter and longer overall rectangular For any selection of angle g1 or g2, the resulting ultimate
cross section dimensions, respectively, and fc is the charac- load provided a lower-bound solution to the true ultimate
teristic concrete cylinder compressive strength. As an alter- strength of the stair. The most realistic estimate of ultimate
native, the value of To can be set by the designer. This condi- load is obtained when the ultimate load in Eq. (7.4.2.2b) is
tion will result in an equation to solve the redundancies. maximized, or when
3. To obtain the other (nR 1) equations, introduce (nR
1) known quantities in the stress-resultant field. For this w w
purpose, the position of inflection points and the maximum = 0 and = 0 (7.4.2.2c)
g 1 g 2
values of any stress resultants other than the torsional
moment can be assumed. All or some of these can then be
used to obtain the required (nR 1) equations. Substituting Eq. (7.4.2.1) and (7.4.2.2a) through (7.4.2.2c)
4. The above nR equations can be solved simultaneously, into the equilibrium equations for the helical stairs and
and the resultant stress field is completely formulated. proceeding with additional mathematical derivations
7.4.2.2 Helical stair exampleRangan et al. (1978) (Rangan et al. 1978), an equation for the ultimate helical
applied the four steps (7.4.2.1) to the design of a helical stair strength was provided as a function of the stair dimen-
stair with fixed support, subject to uniformly distributed sions and the characteristic concrete cylinder compressive
strength fc.
Fig. 7.4.2.3Curved guideway structure.
7.4.2.3 Curved girder exampleCollapse load cannot be

determined using a closed-form solution for all torsional
cases, except for possibly the simplest cases, such as the Fig. 7.4.3Square prismatic reinforced concrete torsional
helical stair case solved by Rangan. A trial-and-error proce- member (Wang and Hsu 1997).
dure, therefore, is usually employed to solve for more
complex members subjected to torsion, such as the case of cability to reinforced concrete structures are numerous. For
curved girders. Geometrical techniques have been presented example, Wang (1995) and Wang et al. (1993) have discussed
for the trial-and-error procedure (Boulton and Boonsukha its applicability to reinforced concrete deep beams. Wang
1959; Jordaan et al. 1974), but most published studies use and Hsu (1997) extended that work to reinforced concrete
computer analyses (Yoo and Heins 1972) to obtain the beams subjected to pure torsion. Both deep beams and
collapse load. Representative solutions from the literature torsional members can be viewed as special cases of shear
provide various iterative procedures. For example, Yoo and failure. In the case of pure torsion, shear stresses develop
Heins (1972) determined the sequence of hinge formation in the member to resist the applied torque. The application
leading to collapse using an elastoplastic analysis and finite- of plasticity theory may then successfully lead to analytical
difference techniques. Badawy and Jordaan (1977) used a solutions of ultimate torsional strength of prismatic elements
methodical technique to set up a system of nonlinear, simul- made of homogeneous and isotropic materials. One analyt-
taneous equations for each failure mode. The equations ical solution method is the sand-heap analogy developed by
satisfied both equilibrium and yield conditions, and provide Sadowsky (1949), which is based on the theory of plasticity.
a lower-bound estimate of ultimate strength. Badawy and This theory has proven effective for estimating the ultimate
Jordaan (1977) indicated that: torsional strength of metal elements. The same approach for
In some cases, the number of equations that result structures made of reinforced concrete, however, is complex
from a lower bound analysis is less than the number and not as straightforward.
of unknowns, and the required solution should Wang and Hsu (1997) used the modified Coulomb-Mohr
correspond to the maximum load. The problem is failure criterion as the constitutive law of concrete and
generally reduced to one requiring an optimization assumed that the plastic flow of concrete after yielding is
type procedure, in which the load is maximized associated with this failure surface. They employed a gener-
under the conditions or constraints of equilibrium alized energy dissipation rate formulation along a yield line
and yield criterion. in accordance with the failure criterion, as well as the upper-
A typical optimization technique normally employed to bound approach, to calculate the ultimate strength of a beam
satisfy those constraints simultaneously, and therefore solve subjected to pure torsion. They compared their approach
for the ultimate carrying strength, is the Lagrange multi- to the results of experiments and found substantial agree-
plier method as described in Badawy and Jordaan (1977). ment with them. Expanding on this work, Wang and Hsu
The curved girder (Fig. 7.4.2.3) and associated torsion solu- (1997) considered a square prismatic reinforced concrete
tion problem represent a condition examined by several beam subjected to pure torsion, as shown in Fig. 7.4.3. By
researchers because of its frequent occurrence in design. introducing a permissible failure mechanism, as illustrated
Separate studies by Boulton and Boonsukha (1959), Jordaan in Fig. 7.4.3, the work equation along the yield surface is
et al. (1974), Yoo and Heins (1972), and Badawy and Jordaan formulated as
(1977) all shared a similar general lower-bound framework
as applied to a curved girder subjected to torsion. v* = F x1 sec(a* b* ) (7.4.3a)
7.4.3 Upper-bound approachThe use of upper-bound
approaches in the design of reinforced concrete structures
is common, particularly in slab design using the yield line Refer to Fig 7.4.3 for identification of a* b*.
method. Studies of the general upper-bound approach appli- Therefore, on an infinitesimally small area dA of the yield
surface, Eq. (7.4.3b) applies
dA = 2 x1 sec b* dx1 (7.4.3b) T = 0 (7.4.3i)

The solution of Eq. (7.4.3i) results in an expression of the
By counting all reinforcement bars on the relevant section, form
the total energy dissipated by the reinforcement is repre-
sented by
T = f (m, ft* , w s , ac ) (7.4.3j)

Ds = 2bc2 F rl fly tan(a* b* ) + rt fty tan b* (7.4.3c)
Where the first two functional terms represent concrete
properties, ws is a reinforcement index, and the last term
The energy dissipated on the discontinuous concrete yield represents geometric properties. Wang and Hsu (1997)
surface can also be found by integration, resulting in the obtained a closed form solution for Eq. (7.4.3j), which they
equation compared with experimental results. Thirty-nine test results
from previous empirical studies conducted at the University
F 3 of Houston were analyzed and compared with the proposed
Dc = b sec(a* b* ) ft* (sin a* + K cot a* cos a* ) (7.4.3d)
3 theoretical solution. Researchers found that the best fit was
obtained when a uniform plastic effectiveness factor for
K is obtained from the Mohr-Coulomb yield criterion and concrete under torsion was used, as given by
takes the form
n = 0.75 + 0.125w s (7.4.3k)

1
K = m + 2(1 m + 1) (7.4.3e) This equation indicates that an increase in reinforcement
4
ratio provides better confinement to the concrete medium,
allowing higher ductility and, therefore, higher plastic effec-
where
tiveness. The measured-versus-predicted ultimate torsional
strength of the 39 tested beams averaged 0.98 with a stan-
fc*
m = * (7.4.3f) dard deviation of 0.09, which is a good indicator of the
ft applicability of the upper-limit approach derived by Wang

and Hsu (1997).
* *
The quantities fc and ft are related to the concrete 7.4.4 Approach comparisonMarti and Kong (1987)
compressive and tensile strength, fc and ft, using plastic provided a similar yield line approach to predict the overall
effectiveness factors nc and nt load-deformation response of orthogonally reinforced
concrete slabs subjected to pure torsional moments in the
f* n f reinforcement direction. In a companion paper, Marti et al.
m = c* = c c (7.4.3g) (1987) presented the results of torsion tests for nine orthogo-
ft nt ft
nally reinforced concrete slabs, comparing the results to
various theoretical predictions, including the yield line
Although a discussion on plastic effectiveness factors is approach provided by Marti and Kong (1987) . They found
generic to concrete structures and outside the scope of this the measured ultimate resistances to be 5 to 46 percent below
report, it is noteworthy that plastic effectiveness factors are yield line theory predictions. They also found that a close
direct functions of concrete ductility. The higher the strains prediction was achieved by a lower-bound limit analysis
withstood by the concrete medium, the higher the plastic approach with an assumed effective plastic concrete strength
effectiveness factors. Many formulas give the plastic effec- equal to 45 percent of the measured uniaxial compressive
tiveness factors as inversely proportional to concrete strengths strength. No similar research comparing lower- and upper-
because ductility decreases as concrete strength increases. bound approaches to each other or to existing codes for
Another factor affecting plastic effectiveness is the beams subjected to torsion, either pure or combined with
degrees of concrete confinement. Higher confinement levels, flexure, has been published. Such a study would have shed
due either to applied triaxial pressure or to the presence of more light on the validity of upper- and lower-bound assump-
confining reinforcement, increases concrete ductility and tions, as well as the degrees of conservatism or unconserva-
increases the plastic effectiveness factors. For a complete tism in the lower- and upper-bound approaches, respectively.
discussion of the plastic effectiveness factors, refer to Chen
(1982). Equations (7.4.3c) through (7.4.3g) are related to 7.5Treatment of open sections
applied torsion by the equation of externally applied work 7.5.1 Types of torsional resistanceWhen subjected to
to the dissipated work, where the external work is given as torsion, a member develops two types of torsional resis-
tanceSaint-Venant and warping. For solid and hollow
W = TF (7.4.3h) members with a relatively bulky cross section, Saint-Venant
(1856) torsion normally predominates with warping torsion
The lowest upper-bound solution can then be found by generally neglected. In contrast, for members with a thin-
Hwang and Hsu (1983) used the Fourier Series method to

solve the warping torsion problem. Under torsional stresses,
the angle of rotation was assumed to be a continuous func-
tion between the supports, expressed as
np np
F( z ) = Bo + Bn sin z + Bn cos z (7.5.2c)
n =1 L n =1 L
where Bo, Bn, and Bn are constant in the Fourier series expan-
sion of a beams curvature under warping.
Both the Saint-Venant torsion (1856) and warping torsion
Fig. 7.5.2Typical B-F curve (Hwang and Hsu 1983). expressed the basic governing equation for mixed torsion as
walled open section, warping torsional resistance predomi- T ( z ) = GC( z ) F (z ) EJ w F (z ) (7.5.2d)
nates and the Saint-Venant torsional resistance does not
occur. Past research has usually ignored secondary warping
phenomenon (Vlasov 1961; Zbirohowski-Koscia 1968), The solution for Eq. (7.5.2d) was calculated using matrix
mainly due to the mathematical difficulties encountered in notation, which is discussed in Hwang (1978). Before
formulating this type of torsional resistance and the very cracking, the GC and EJw terms are constants for any given
narrow thickness of metal structures. The wall thickness of section and the solution is easily identified. After cracking,
reinforced concrete open sections, however, is often substan- however, both GC and EJw vary as functions of the load
tial compared with overall cross-sectional dimensions. level, making it essential to employ a trial-and-error proce-
Therefore, this secondary warping may play a less signifi- dure to obtain the solution for mixed torsion (Hwang 1978).
cant role in reinforced concrete structures. Krpan and Collins (1981a) also developed an analytical
The torsional behavior of a homogeneous elastic member procedure using Vlasovs theory and incorporating CFT
with a thin-walled open section is predicted by Vlasovs to predict the elastic pre- and post-cracking response of
theory. The theory was later generalized by Hwang (1978) to thin-walled reinforced concrete beams. The post-cracking
predict the torsional behavior of inelastic reinforced concrete response was categorized as either a cracked elastic response
members with open sections in the post-cracking stage. The or an inelastic warping torsion response. The authors recom-
mathematical formulation proposed by Hwang (1978) and mended using the post-cracking section properties rather
Hwang and Hsu (1983) is given Eq. (7.5.2a). than the uncracked elastic when analyzing the overall struc-
7.5.2 Consideration of warping torsion-formulation ture values to achieve a more accurate assessment of the
Applying the concept of sectorial area, Vlasov (1961) actual force distributions in the structure.
derived the basic equation for warping torsion, which relates 7.5.3 Experimental verificationTo verify the proposed
the warping torque Tw to the third derivative of the rotation solution, Hwang and Hsu (1983) designed two reinforced
with respect to z, F concrete beam specimens with open sections and tested their
torsional behavior. The results were used to evaluate the
Tw = EJwF (7.5.2a) Fourier Series method proposed for mixed torsion analysis.
A comparison of experimental and predicted data revealed
The quantity EJw is used to characterize rigidity of the that the Saint-Venant torsional resistance had to be modified
beam subjected to warping torsion. The term Jw is called the by certain empirical factors, as addressed in Hwang (1978).
sectional moment of inertia and is calculated by applying Incorporating these modification factors into the Saint-
procedures summarized by the original work of Vlasov. Venant torsional portion, they found that a mixed torsion
Integrating Eq. (7.5.2a) results in the equation analysis using the Fourier Series method predicted with
reasonable accuracy both the pre- and post-cracking stages
B = EJwF (7.5.2b) of the T-F curves of the two tested specimens.
Figure 7.5.3 shows test results and analytical predictions for
where B is the integral of Tw. a thin-walled reinforced concrete specimen tested by Krpan
A typical B-F curve is shown as a solid line in Fig. 7.5.2. and Collins (1981b). That work indicated that to accurately
This B-F curve can be constructed by a method proposed predict the strain in transverse reinforcement, it was necessary
by Hwang (1978) and Hwang and Hsu (1983) based on a to account for the interaction between Saint-Venants torsion
bimaterial model. A point on the curve signifies a certain load and warping torsion. Test results obtained by the authors,
level. At the specific load level, the slope of a straight line shown as numbered points in Fig. 7.5.3, were compared with
connecting this point with the origin is defined as the post- analytical predictions made according to the method proposed
cracking warping rigidity. The presence of an end diaphragm by Krpan and Collins (1981a). Figure 7.5.3 shows these
provides a warping restraint, which is considered in their results and confirms the ability of the developed analytical
model by incorporating an applied bidirectional moment. method to predict measured torque and twist. This method
also accurately predicted the elastic pre- and post-cracking
response of thin-walled reinforced concrete beams. The post-

cracking response for the cracked elastic response, or inelastic
warping torsion response, was also well predicted.
7.6Size effect on the strength of concrete beams

in torsion
Concrete size effect (Baant 1984; Baant and Planas
1998; RILEM Committee QFS 2004) is explained by energy
release due to stress redistribution caused by a large crack
or cracking zone formed before the point of maximum
load. This size effect is transitional between the small-size
asymptotic case of no size effect corresponding to plastic
limit analysis, and the large-size asymptotic size effect
corresponding to linear elastic fracture mechanics (LEFM),
in which the fracture process is assumed to take place at one Fig. 7.5.3Comparison of predicted and observed torque-
point only at the tip of a sharp crack. In small concrete struc- twist relationship (Krpan and Collins 1981b).
tures, typical of normal laboratory tests, the size effect is
normally almost undetectable, whereas in extrapolations to represent the size effect for similar fractures according to
large structures in the field, the size effect may be strong. LEFM, which is the strongest possible size effect. The data
Size effect is expected for all the brittle failures of concrete trend closely approaches this slope, which means that the
structures, which include torsion. For reinforced concrete size effect is strong for plain concrete beams and beams with
beams with little or no shear reinforcement, size effect for longitudinal reinforcement without stirrups.
shear has been observed (Baant and Yu 2005a,b; Bentz Based on the works of Baant and Planas (1998) and Baant
2005). Torsional failure is a similar phenomenon, making (2002), theoretically-justified formulas were developed to
size effect expected in that case as well. There is signifi- account for the size effect in beams subjected to torsion.
cantly less experimental data on size effect in torsion than The equation for reinforced concrete beams, which fail after
for that of shear failure, although some experimental data do significant fracture growth, is given in simplified form as
exist (Humphreys 1957; Hsu 1968a; McMullen and Daniel
1975). These were collected and evaluated by Baant and sN = s0(1 + D/D0)1/2 (7.6a)
Sener (1987) and reproduced in Fig. 7.6a (top and middle)
as plots of log sN/s1 versus log D/D1. Term D is the charac- and for plain concrete beams, which fail at fracture initia-
teristic structural dimension, chosen here as the cross section tion, is given in simplified form as
size, whereas s1 and D1 are normalizing constants that are
different for each plot. The term sN is the nominal strength sN = s(1 + rDb/D) 1/r (7.6b)
of the structure; for beams it is geometrically similar in three
dimensions. It is defined as sN = Tmax/D3, where Tmax is the where s0 is the nominal torsional strength according to the
maximum torsional moment, and D is the cross section depth current code specifications based on plastic limit analysis;
defined in Baant et al. (1988). The size range of these data, s is the plain beams strength according to elastic analysis
which does not exceed 1:2.7, is too limited comparing to the with the maximum stress limited by material strength; and
scatter width. If the plastic limit analysis approach to torsion D0, Db, and r are constants. Simple prediction of these three
was valid, the size effect would have to be absent and the constants and their dependence on the longitudinal rein-
trends in Fig. 7.6a would be horizontal. Clearly this is not the forcement ratio, stirrup ratio, stirrup spacing, longitudinal
case. Although these data are too scattered to confirm any and transverse prestress (if any), and some other geometrical
particular formula, they nevertheless indicate the downward ratios is lacking at present. Experiments should be conducted
trend of size effect. Baant et al. (1988) obtained further test to verify and calibrate the theory to predict these constants.
data, with strict geometrical scaling in three dimensions and For very large beams, the Weibull statistical size effect
a greater size range (1:4). They used square beams made should be supplanted to Eq. (7.6b), as shown in general in
of microconcrete with a maximum aggregate size of 4.8 Baant (2002).
mm (0.18 in.) and cross-sectional side dimensions of 25.4,
50.8, and 101.6 mm (1, 2, and 4 in.) (Fig. 7.6b). They used CHAPTER 8DETAILING FOR TORSIONAL
standard reduced-scale deformed reinforcement taken from MEMBERS
PCA tests (Hsu 1968a) and tested both plain concrete beams 8.1General
and those with longitudinal reinforcement without stirrups Torsional moment in a reinforced concrete member is
(except at beam ends). Results are shown in the bottom of resisted by a circulatory shear flow in a tube along the cross
Fig. 7.6a. There was much less scatter observed in this study section periphery. The tube can be idealized as a space truss
than the data in the top and middle of Fig. 7.6a. Decreasing made up of reinforcement ties and concrete struts, as shown
strength with the increase of size is clearly confirmed due to in Fig. 3.3.5a. The shear flow induces tensile forces in both
size effect. The straight lines of slope (1/2) in these plots the hoop reinforcement and longitudinal reinforcement.
Fig. 7.6aNominal torsional failure strengths of beams with rectangular cross section and various sizes. Left: plain
concrete beams. Right: longitudinally reinforced concrete beams without stirrups (Baant and Sener 1987). Top: test
data of Humphreys (1957). Middle: data of Hsu (1968a) and of McMullen and Daniel (1975). Bottom: Data on micro-
concrete beams tested by Baant et al. (1988).
Good reinforcement detailing is required to ensure that the a longitudinal bar at the corners. Enclosure of the longitudinal
hoop and longitudinal reinforcement can develop their yield reinforcement by the transverse reinforcement provides the
strength to resist circulatory shear flow. necessary equilibrium at the joint in the three principal direc-
Good detailing demands consideration of the interaction tions, where the three-dimensional force flow is equilibrated.
between the member longitudinal and transverse reinforce-
ment. Although each member type brings about different 8.2Transverse reinforcement
detailing conditions, the designer should be mindful of this 8.2.1 GeneralOnce proportioned for torsion and shear, the
overall force interaction in the member. Transverse reinforce- transverse reinforcement is laid out at a specific longitudinal
ment, oriented either horizontally or vertically, should contain spacing along the member span. The objective of transverse
Fig. 7.6bPlain and longitudinally reinforced microconcrete beams tested by Baant et al.
(1988) and the typical mode of torsional failure.
reinforcement for torsion and shear is to provide the reinforce- member cross sectionstirrups are provided in a closed rect-
ment around the perimeter to enclose the member core. Typi- angular shape to encase the rectangular member core. The
cally, this reinforcement has a smaller diameter than the longi- hooks of the closed stirrup are developed into the core with
tudinal reinforcement due to spacing, placement, bending, 135-degree bends. These bends ensure the hooks are well-
and proportioning needs. The transverse reinforcement should anchored to the member core and prevent hook pullout under
enclose the perimeter as closely as possible while maintaining high torsional loads. Figure 8.2.1a provides an example of a
clear cover requirements. A closed stirrup is imperative for simple rectangular closed stirrup.
torsional detailing. In the simplest casea basic rectangular
Other common examples of cast-in-place member types supplemental ties or stirrups developed back into the rectan-
subject to torsional loads are shown in Fig. 8.2.1b. The key gular core of the individual member.
to providing transverse reinforcement in a member subject As shown in Fig. 8.2.1c, good detailing usually dictates that
to torsion is to start with the largest rectangular cross section additional ties or stirrups reinforce any protrusions. These
and provide a rectangular closed stirrup in that section. Alter- ties also have a semi-closed detail with 135-degree or greater
nately, multiple-leg configurations can also be used for this hooks developed into the core, which is the region enclosed
purpose with single or multi-leg pieces or bar layouts to rein- by the closed stirrup shape. In addition to the closed stirrups
force the cross section. Any protrusions, apertures, ledges, and longitudinal bars shown in Fig. 8.2.1b and 8.2.1c, local
corbels, or other geometric outcroppings are provided with reinforcement in the disturbed regions or D-regions should
accommodate specific load concentrations. To be effective in
any size member subjected to torsion, spacing between the
closed ties should not exceed about one-half of the smallest
dimension of the member, except for slender precast span-
drel beams, such as those used in parking structures. In these
members, torsional forces cause out-of-plane bending in the
web. As described in 7.3, limited testing of such members
has not shown signs of spalling or stirrup debonding for
which closed stirrups are required. In load tests, slender
precast spandrel beams have performed exceptionally well
without closed ties. The current state of practice on spandrel
beam behavior is contained in a recent study at North Caro-
lina State University (Lucier et al. 2010).
8.2.2 Hooks and development considerationsStirrups
or ties are best terminated with 135- or 180-degree bends.
Hooks should be developed into the main core of the
member, where greater confinement is present. This detail
is important in isolated members, where hook confinement
is only provided by the member core, and no other external
geometric conditions provide confinement. Practical consid-
erations might dictate the use of simpler stirrup geometry,
Fig. 8.2.1aA typical closed stirrup usually employing 90-degree hooks. When 90-degree hooks
used in a simple rectangular cross are used, confinement should be provided at locations where
section. a slab frames into the beam side or elsewhere as needed.
Fig. 8.2.1bExamples of transverse torsional detailing in cast-in-place concrete members.

Fig. 8.2.1cExamples of transverse torsional detailing in precast concrete members.
Fig. 8.2.2aRecommended two-piece closed single and multiple U-stirrups for members
subjected to torsion (ACI Committee 315 2004).
In Fig. 8.2.2a, examples from the ACI Detailing Manual

(ACI Committee 315 2004) suggest using 90-degrees hooks cross section perimeter. In beam regions or B-regions, special
under various side confinement conditions. Examples of details need not be provided aside from equal spacing or
poor detailing are reproduced in Fig. 8.2.2b. These recom- proportioning of the reinforcement around the perimeter.
mendations are adopted from the research of Mitchell and Splices are proportioned in accordance with ACI require-
Collins (1976). ments. At the end of a cast-in-place member, the perimeter
longitudinal reinforcement may have to be developed into
8.3Longitudinal reinforcement a column or other type of rigid vertical member providing
Longitudinal reinforcement is also proportioned according torsional restraint.
to torsional requirements and provided around the member CSA-A23.3-04 also includes the requirement that A
longitudinal reinforcing bar or bonded prestressing tendon
Fig. 8.2.2bIneffective closed stirrup types for members subjected to torsion (ACI
Committee 315 2004).
shall be placed in each corner of closed transverse steel rein-

forcement required for torsion. The nominal diameter of the
bar or tendon shall not be less than s/16. The corner bars
help to support the outward thrusts in the zones between
the hoops (Mitchell and Collins 1976). Precast concrete
members usually have semi-rigid connections consisting of
field-welded angles and plates. Although this is not a full
torsionally restrained connection, the end region should be
examined along with subsequent development of the longi-
tudinal, perimeter reinforcement at the member ends. One
common way of developing reinforcement is to use conven-
tional hooks at the member end. In some cases, the hooks
might not fit in typically thin precast concrete members.
Another common detail is the use of U-bars placed hori-
zontally at the member end and lapped with the longitudinal
reinforcement. The U-bars provide additional end confine-
ment to a given precast member. Likewise, such reinforce-
ment can provide sufficient confinement around the connec-
tion plate studs or tail bar reinforcement.
8.4Detailing at supports
Precast concrete members require other special detailing
considerations due to their horizontal support conditions.
Torsional forces in precast members are often equilibrated
by out-of-plane, horizontal, or sometimes vertical, reactions
at discrete locations along the member depth. The member
end conditions and subsequent details are highly dependent
on the support configuration.
Figure 8.4 illustrates a common precast spandrel beam
and the horizontal force couple that is typically developed
at the end. Additional reinforcement is thereby required at
Fig. 8.4Support detailing requirements in a precast the member end near the top to accommodate a potential
spandrel member, dependent on the support connection 45-degree crack that typically develops at the end location
locations (Raths 1984). due to the couple resisting torsion, as shown in Fig. 8.4(a)
and 8.4(b). A possible reinforcement scheme to address this analogy provides the basic concept for torsion design assuming
condition is shown in Fig. 8.4(c) and 8.4(d). Similar condi- that the tension contribution in concrete is neglected and the
tions often exist in other precast members due to their specific diagonal compression struts spiraling around the member
support and horizontal restraint conditions. Although these with variable inclined angle that depend on the loading condi-
are D-region locations, they are complicated by the three- tion and reinforcement ratio. The bending moment and longi-
dimensional or out-of-plane nature of the problem. tudinal forces due to torsion and shear are considered resis-
tant to four chords, one in each corner of the space truss and
CHAPTER 9DESIGN EXAMPLES the shears by the shear flows in the walls. Dimensions of a
9.1Torsion design philosophy cross section are limited to prevent crushing of the diagonal
Design philosophy for torsion in the ACI 318-11 building compression struts. In addition to this AASHTO LRFD
code is based on a thin-walled tube, space truss analogy in method (general method), CSA-23.3-04 provides a simplified
which compression diagonals wrap around the tube and the method for a restricted group of structural members, which
tensile contribution of concrete is neglected. Both solid and states that the inclined angle of diagonal concrete compres-
hollow members are considered tubes in accordance with sion strut is fixed at 35 degrees.
Saint-Venants circulatory shear flow pattern both before and
after cracking. The outer part of the cross section centered 9.2Torsion design procedures
along the stirrups is assumed to provide torsional resistance. 9.2.1 Torsion design in ACI 318-11According to Saint-
The contribution of core concrete cross section is neglected. Venants circulatory shear flow pattern, the most efficient
Once a reinforced concrete beam has cracked in torsion, the cross section to resist torsion is tube-shaped. Therefore,
torsional resistance is provided primarily by closed stirrups torsion of a reinforced concrete member is a three-dimen-
and longitudinal bars located near the members surface and sional (3-D) problem because it involves the shear in a
diagonal compression struts. The inclined angle of the diag- reinforced concrete two-dimensional (2-D) wall element of
onal compression struts is permitted to be taken as 45 degrees a hollow tube and the out-of-wall bending of the concrete
for nonprestressed and lightly prestressed members, and 37.5 struts. In ACI 318-11, two simplifications are made. First, the
degrees for most prestressed members. Accordingly, ACI concrete strut bending is neglected and the amount of hoop
318-11 makes the specific assumptions in torsion design that: steel required in the tube determined from Bredts (1896)
a) Concrete tensile strength in torsion is neglected equilibrium equation of a cross section
b) Torsion has no effect on the shear strength of concrete
c) Torsion stress determination is based on the closed thin- qy = Tu/2Ao (9.2.1a)
walled tube with uniform stress distribution and specific
thickness, called shear flow where the symbol qy is the shear flow at yield (N/mm [lb/
d) The torsional, flexural, and shear strength are accounted in.]); Tu is the torsional moment (Nmm [in.-lb]); and Ao
for by adding longitudinal reinforcement calculated for [mm2 (in.2)] is the lever arm area enclosed by the centerline
torsion and flexure of the shear flow.
e) The longitudinal reinforcement are calculated for Second, the hoop and longitudinal steel are assumed to
torsion and shear yield at ultimate strength. To design steel reinforcement in
The design of torsional resistance in Section 6.3 of a 2-D shear element, it is possible to use only three equilib-
EC2-04 is also based on a truss model using the thin-walled rium equations (Hsu 1993). Combining the three equations
closed section theory with an effective wall thickness. creates a simple equation for yield shear flow
The angle between the concrete compression strut and the
members longitudinal axis, q, may be taken between 22 q y = ( At f y / st )( A f y / s ) (9.2.1b)
and 45 degrees. Both the solid and hollow cross section can
be modeled by an equivalent hollow section neglecting the
core concrete contribution to calculate the torsional resis- where fy is yield stress of hoop steel and longitudinal steel
tance, which is limited by the strength of the concrete struts. (MPa [psi]); At, A are area of hoop steel and longitudinal
The longitudinal and transverse reinforcement contributions steel (mm2 [in.2]), respectively; and st, s are spacing of hoop
to torsional resistance are accounted for after the thin-wall steel and longitudinal steel (mm [in.]), respectively. Substi-
cracks. Effects of combined torsion and shear may be super- tuting the shear flow qy into Bredts (1896) equation gives
imposed assuming the same value for the strut inclination
angle. The required longitudinal and transverse reinforce- Tu = 2 A0 ( At f y / st )( Al f y / sl ) (9.2.1c)

ment for torsion should be added to the existing longitudinal
reinforcement for bending and transverse reinforcement for which is the essence of the ACI code provision.
shear, respectively. The lever arm area Ao (mm2 [in.2]) is formed by sweeping
The Canadian code (CSA-A23.3-04) provides a General the lever arm of shear flow one full circle around the axis
Design Method for torsion derived from the modified of twist. The centerline of shear flow was taken by Rausch
compression field theory (CFT) and represents solid cross (1929) to be the centerline of the hoop steel bar, and the
sections by an equivalent thin-walled tube with a wall thick- corresponding lever arm area is denoted as Aoh (mm2 [in.2]).
ness determined by cross section dimensions. The space truss
However, this definition of area Aoh was found to overesti- by an equivalent hollow section from which the torsional
mate the torsional strength by as much as 30 percent. There- resistance is calculated. Complex shapes, such as T-sections,
fore, the ACI code provides a simple, approximate formula can be divided into a series of subsections modeled as an
for calculating the lever arm area as equivalent thin-walled section, and the total torsional resis-
tance taken as the sum of the capacities of each individual
Ao = 0.85Aoh (9.2.1d) element. The effects of combined torsion and shear for both
hollow and solid members can be superimposed assuming
To provide a more accurate formula for the ultimate the same value for the strut inclination angle q.
torsional strength, consider the softening of concrete struts A common value for angle q is 45 degrees. Eurocode 8
in the reinforced concrete 2-D wall elements of a tube. Under (EN 1998-1:2004) determines that: In the critical regions
a biaxial tension-compression stress condition, the compres- of primary seismic beams, the strut inclination q in the
sive stress-strain curve of the 2-D elements should be multi- truss model shall be 45 degrees (Paragraph 5.5.3.1.2(2)
plied by a softening coefficient. This softening coefficient of EC2-08). However, in members not designated to resist
is a function of the principal tensile strain (Zhang and Hsu seismic actions, a reduced value of angle q could be consid-
1998) and varies from approximately 0.25 to 0.50. Applying ered to decrease the required transverse reinforcement and
this softened stress-strain curve of concrete to the study of required longitudinal reinforcement. This way, fewer stir-
reinforced concrete tubes under torsion (Hsu 1990, 1993), rups and more longitudinal bars could be provided. Required
the thickness td (mm [in.]) of the shear flow zone and lever torsional reinforcement is added to the required stirrups and
arm area can be determined as bars calculated from the shear and flexural design, respec-
tively. Strength of materials used in EC2-04 is based on
t d = 4Tu / Acp fc characteristic values and depend on whether the value is
(9.2.1e) used for strength or stiffness. The characteristic value used
Ao = Acp (2Tu pcp / Acp fc)
to calculate strength corresponds to the 95 percent fractile
of strength from material tests. The characteristic value for
where Acp is the area enclosed by the outer boundary of cross stiffness corresponds to mean strength from material tests.
section (mm2 [in.2]); and pcp is the periphery of the outer Design values are based on multiplying the characteristic
boundary (mm [in.]). These formulas are given in the ACI value for resistance by the safety factors a and b.
code commentary, and the background was given in a paper Design procedure in accordance with EC2-04:
by Hsu (1997). - Step 1: Calculation of the equivalent thin-walled section
9.2.2 Torsion design in EC2-04Section 6.3 of EC2-04 characteristics such as tef, Ak, and uk (also refer to Fig. 6.11
requires a full design procedure for a reinforced concrete of EC2-04 for notation)
member under torsion covering both ultimate and service-
ability limit states in cases where the static equilibrium of tef = A/u 2c and, in the case of a hollow section, tef < treal
the structure depends on torsional resistance of the elements.
In conventional statically indeterminate reinforced concrete where
structures, torsion arises from consideration of compat- tef = effective wall thickness of the equivalent thin-
ibility and it is normally unnecessary to consider torsion walled section [mm (in.)]
at the ultimate limit state. However, even if torsion arises A = total area of the cross section within the outer
from consideration of compatibility only, it may lead to circumference, including inner hollow areas (for
excessive cracking in the serviceability limit state. There- example, A = bh in a rectangular cross section with
fore, a minimum reinforcement of stirrups and longitudinal width and height equal to b and h, respectively)
bars should be provided to prevent excessive cracking, as [mm2 (in.2)]
indicated in EC2-04 for cracking control (Section 7.3) and u = outer circumference of the cross section (for
detailing beams (Section 9.2). example, u = 2(b + h) in a rectangular cross section)
In normal slab-and-beam or framed structures, specific c = distance between edge and center of the longitu-
calculations for torsion are usually unnecessary when dinal reinforcement (centroid cover) [mm (in.)]
torsional cracking is being adequately controlled by shear treal = real thickness of a hollow section [mm (in.)]
and minimum flexural reinforcement. Where torsion is Ak = area enclosed by the centerlines of connecting
essential for equilibrium of the structure, EC2-04 should be walls, including inner hollow areas (for example,
consulted. One example of this is when structure arrange- Ak = (b tef)(h tef) in a rectangular cross section)
ment is such that loads are imposed mainly on one face of a [mm2 (in.2)]
beam without corresponding rotational restraints provided. uk = perimeter of the area Ak (for example, uk = 2(b +
The design of torsional resistance moment is based on a h 2tef) in a rectangular cross section)
truss model using a thin-walled closed section theory with - Step 2: Assume the value of angle of compression struts,
inclined angle q between the concrete compression strut and q, based on the expression: 1 cotq 2.5 (45 degrees q
the beam axis. The angle q should be limited and recom- 22 degrees). For combined shear and torsion, the same value
mended limits are: 1 cotq 2.5 (45 degrees q 22 of q should be assumed and the common value is 45 degrees.
degrees). With a solid section, the section can be modeled
- Step 3: Check the maximum resistance of the member - Step 4: Calculation of the required cross-sectional area
subjected to torsion and shear. This is limited by the strength of the longitudinal reinforcement for torsion, SAs:
of the concrete struts. If the following relationship is not
satisfied, the member cross section dimensions, the concrete TEd uk cot q
compressive strength, or both, should be increased As = (9.2.2f)
2 Ak f y d

TEd V
+ Ed 1 (9.2.2a) where fyd is the design yield stress of the longitudinal rein-
TRd.max VRd.max
forcement [MPa (psi)].
Notes: The longitudinal reinforcement for torsion should
where be added to the required longitudinal reinforcement for
TEd = design torsional moment [Nm (in.-lb)] flexure. The longitudinal reinforcement should generally be
TRd.max = design torsional resistance moment according to distributed over the length of side, zi, (zi is the side length of
the following relationship [Nm (in.-lb)] wall i defined by the distance between intersection points
with the adjacent walls [refer to Fig. 6.11 of EC2-04]), but
TRd,max = 2vacwfcdAktefsinqcosq (9.2.2b) for smaller sections it may be concentrated at the ends of this
length. According to EC2-04 provisions (Section 9.2.3(4)),
VEd = design shear force [N (lb)] longitudinal bars for torsion should be arranged such that
VRd.max = maximum design shear resistance according to the there is at least one bar at each corner, with the others being
following relationship [Nm (in.-lb)] distributed uniformly around the inner periphery of the links,
with a spacing not greater than 350 mm (14 in.).
a c bw zvfcd - Step 5: Calculation of the required cross-sectional area
VRd , max = (9.2.2c)
(cot q + tan q) of the transversal reinforcement for torsion
Asw TEd
bw = width of the web of the cross section [mm (in.)] = (9.2.2g)
z = inner lever arm, for a member with constant depth, s 2 Ak f ywd cot q

corresponding to the bending moment in the element
under consideration. In the shear analysis of rein- where
forced concrete without axial force, the approximate Asw = cross-sectional area of the transversal reinforce-
value z = 0.9d may normally be used (d is the effec- ment (stirrups) [mm2 (in.2)]
tive depth of the cross section) [mm (in.)] s = spacing of the stirrups [mm (in.)]
v = strength reduction factor for concrete cracked in fywd = design yield stress of transversal reinforcement
shear, recommended values (values for use in a [MPa (psi)]
country may be found in its National Annex): Notes: The transversal reinforcement for torsion should
be added to the existing transverse reinforcement for shear.
v = 0.6(1 fck / 250) [ fck in MPa] The torsion links (stirrups) should be closed and anchored
(9.2.2d)
v = 0.6(1 fck / 36.26) [ f y in ksi]] by means of laps or hooked ends and form an angle of 90
degrees with the axis of the structural element. Refer to Fig.
9.6 of EC2-04 for recommended shapes. According to provi-
acw = coefficient taking into account the state of compres- sions of EC2-04 (Section 9.2.3(3)), longitudinal spacing of
sive stress the torsion stirrups should not exceed u/8, or the require-
ments about the maximum longitudinal spacing between
1 non-prestressed shear assemblies (Section 9.2.2(6) of EC2-04) or the lesser
1+ s / f 0 < s cp 0.25 fcd
cp cd dimension of the beam cross section.
a cw = (9.2.2e) - Step 6: Check the value of the angle of compression
1.25 0.25 fcd < s cp 0.5 fcd
struts, q, based on the calculated and provided longitudinal
2.5(1 s / f ) 0.5 fcd < s cp fcd
cp cd and transversal reinforcement from Steps 4 and 5
scp = mean compressive concrete stress due to design Asw uk f ywd

qcalc = tan 1
axial force (measured positive) [MPa (psi)] s Asl f y d (9.2.2h)
fck = characteristic compressive concrete strength [MPa
(psi)]
fcd = design compressive concrete strength (= fck/gc, Note: In case of a significant difference between the calcu-
where gc is the partial factor for concrete equal to lated and the initially assumed angle of compression struts,
1.5 for ultimate limit state and persistent and tran- reassume the angle q (Step 2) and recalculate Step 3 through 6.
sient design situations) [MPa (psi)] 9.2.3 Torsion design in CSA-A23.3-04The Canadian
code stipulates that the effect of torsion should be considered
in design only if the torsion due to factored loads, Tf [Nm The longitudinal strain ex is affected by the bending
(in.-lb)], exceeds 0.25Tcr. The cracking torque Tcr [Nm (in.- moment: shear, torsion, and if present, by axial load and
lb)] is assumed to be reached when the principal tensile stress prestressing in the member. In the presence of bending
fl [MPa (psi)] (equivalent to the shear stress v in pure torsion) moment, shear, and torsion, the strain at mid-depth of the
equals the factored tensile strength of the concrete, fcr [MPa section, ex, is computed from the expression
(psi)]. For the calculation of Tcr, the following assumptions
are made: e x = ( M f / d + V f2 + [0.9 ph T f / 2 Ao ]2 ) / (2 Es As ) (9.2.3f)
- Solid cross sections are represented by an equivalent
thin-walled tube with a wall thickness [mm (in.)]
where
tc = 0.75Ac/pc (9.2.3a) Mf = moment due to factored loads [Nm (in.-lb)]
Vf = shear force due to factored loads [N (lb)]
- Bredts classical equation for tubular section applies ph = perimeter of the centerline of the closed transverse
[MPa (psi)] reinforcement [mm (in.)]
Es = modulus of elasticity of the reinforcement [MPa (psi)]
v = Tf/(2Aotc) (9.2.3b) As = area of flexural reinforcement on the flexural
tension side of the member [mm2 (in.2)]
- Area enclosed by shear flow path [mm2 (in.2)] q = 29 + 7000ex (degrees)
For a given angle q, the transverse reinforcement to resist the
Ao = 2/3Ac (9.2.3c) factored torque Tf is derived from equilibrium and given by
- Factored design tensile strength for normal concrete Tf s

[MPa (psi)] At = (9.2.3g)
2 Ao f s f y cot q

fcr = 0.38fc fc [ fc in MPa ]
(9.2.3d) where
fcr = 0.38fc 12 fc [fc in psi] s = spacing of transverse reinforcement measured

parallel to the axis of the member [mm (in.)]
For non-prestressed concrete members, this results in the fs = resistance factor for non-prestressed reinforcing bars
following expression fy = specified yield strength of transverse reinforcement
[MPa (psi)]
Additional longitudinal reinforcement is required to resist
Tcr = ( Ac / pc )0.38fc fc [ fc in MPa ]
the longitudinal forces generated by torsion. As usual, the
(9.2.3e)
Tcr = ( Ac / pc )0.38fc 12 fc [ffc in psi] transverse reinforcement due to torsion should be added to

the shear reinforcement.
Dimensions of the cross section of the member have to
The symbols not defined in the above equations are as
be such that crushing of the diagonal compression struts is
follows:
prevented. This is achieved if the combined stress due to shear
Ac = area enclosed by outside perimeter pc of concrete
and torsion does not exceed 25 percent of the factored compres-
section [mm2 (in.2)]
sive strength of the concrete. This is expressed by Eq. (9.2.3h)
fc = specified compressive strength of concrete
fc = resistance factor for concrete (= 0.65)
Vf T f ph
If torsion is not negligible (Tf > Tcr), torsion reinforcement + 2
0.25fc fc (9.2.3h)
should be provided. The General Design Method (CSA- bw dv 1.7 Aoh

A23.3-04) for torsion was derived from the MCFT, which
represents a holistic approach for both shear and torsion 9.2.4 A comparison of torsion design procedures for ACI,
design. For torsion, the basic concept is the space truss EC2, and CSAThe design philosophy and procedures for
analogy, originally envisioned by Rausch (1929), assuming pure torsion and combined loads including bending, shear,
a 45-degree angle for the compression struts. This AASHTO and torsion are discussed previously according to ACI
LRFD (general method), originally developed for shear, 318-11, EC2-04, and CSA-A23.3-04. All the design proce-
requires a longitudinal strain indicator ex and the level of dure and equation citations from these codes are summa-
normalized shear stress vu/fc to estimate q and b. In the case rized and compared in Table 9.2.4.
of a member subjected to pure torsion, it is not necessary to
consider b. For torsion, the General Design Method (CSA- 9.3Introduction to design examples
A23.3-04) assumes that: Two examples were selected to illustrate the steps involved
a) Concrete in the cracked member carries no tension in torsion design: 1) a solid reinforced concrete rectangular
b) The angle q of the diagonal compression struts spiraling beam under pure torsion; and 2) A prestressed box girder
around the member is variable and depends on longitudinal under combined loading including torsion. Although this
strain at mid-depth of section, ex

Table 9.2.4Comparison of torsion design procedures for ACI, EC2, and CSA codes
Pure torsion design procedure ACI 318-11 EC2-04 CSA-A23.3-04
Section 11.5.1, Section 11.2.9.1, Eq. 11-2
2
2
f A A c
c cp 0.38lf c
Tcr = fc [ fc in MPa ]
Determine the factored torsional Tu < fl c p c
[ f in MPa ]
12 p cp
1 moment and if torsion effects can be
2
disregarded A cp
2
Tu < fl [ fc in psi ] A c
fc Tcr = 0.38lf 12 f [ f in psi ]
c c c
p cp p c
Section 6.3.2(1)
Eq. 6-26 and 6-27
Calculate properties of the equivalent
2 A / p c
TEd
thin-walled section t ef = max t r t ef =

2c t
2 Ak VEd = t t t ef z
,
Section 11.5.3.1,
Eq. (11-18) and Section 11.2.1.1, Eq.
(11-3)
Tu ph Vc 2 fc
2
f + [ fc in MPa ]
1.7 Aoh bw d 3 Section 6.3.2(4)
Section 11.3.10.4,
Eq. 6-29 and 6-30
2l fc Eq.11-19
Determine if dimensions of the cross pure torsion
Vc TEd VEd
3 = [ fc in MPa ] + 1 TEd <= TRd,max 2 2
section are adequate TRd , max VRd,max f

V V T p p f h
bw d 12 b d + 1.7 A 0.25f f 2 c c
w v oh
TRd , max = 2na c fcd Ak t ef sin q cos q
Tu ph V c
2 c c
f + 8 f [ f in psi ]
1.7 A b d
oh
w
Vc
= 2l fc [ fc in psi]
bw d
Calculate the amount of stirrups Section 11.3.10.3,
required for pure torsion 11.3.8.1, and 11.3.6.4
Eq.11-1, 11-12, 11-13, and 11-17
Av bw
Section 11.5.3.5,
Section 6.3.2 (2) and 9.2.3
= 0.06 fc [ fc in MPa ]
Eq. (11-20) and Section 11.5.3.6, s min fyv
Eq. (11-21) transformed into Eq. 9.5N
v
f A bw
At Tu c
= 0.72 fc [ fc in psi ]
t yk c
r = 0.08 / f [ f in MPa ]
REPORT ON TORSION IN STRUCTURAL CONCRETE (ACI 445.1R-12)

12 s min
fyv
s f2 fyt Ao cot q
and Section 11.5.5.2, Eq. (11-23) can be rt = 0.08 fc / fyk [ fc in psi ] q = 29 + 7000 e x
4 ( )
Select and check the stirrups details expressed by 2
and Eq. 9.6N Mf 2 0.9 p T h f
for the cross section A t
0.75 fc bw + (Vf Vp ) + + 0.5 N f Ap f po
= [ fc in MPa ] u/8
dv 2 A o
s min
2 12 fyv ex =
2[ E s As + E p Ap ]
smax = min 0.75d (1 + cot a )
At 0.75 bw
= fc [fc in psi] min(b , h ) At Tf
w

s

2
min
f yv
s 2 Ao f s fy cot q
0.7 d v
smax = min
63
600 mm (23.62 in.)

64
Section 11.5.3.7,
Calculate the longitudinal bars At fyt 2
required for pure torsion Al = ph cot q
s fyl Section 6.3.2(3) and 9.2.1 Section 11.3.10.6 and 11.3.9.2, Eq. 11-14
Eq. (11-22):
Tu uk cot q and 1-21 can be transformed into
5 fc Acp A t
f yt Al 2
A ,min = ph [fc in MPa ] 2 Ak fyd Mf 2 0.45 p T h f
5 s Eq. 6-28 F t = + cot q (Vf 0.5Vs Vp ) +

12 f y f y dv 2 A o
fctm
and Section 11.5.5.3, A , min = 0.26 bt d 0.45 ph Tf
Select and check the longitudinal bar = cot q
Eq. (11-24) fyk 2 Ao
details for the cross section and Eq. 9.1N
5 fc Acp At fyt
A ,min = ph [ fc in psi ]
fyl s

fyl
Design procedure for prestressed concrete members

under combined loadings
(Bending, shear, and torsion) ACI 318-11 EC2-04 CSA-A23.3-04
Concrete compressive strength: Section 3.1.2 and 3.1.6

Concrete tensile strength: Section 3.1.2, 3.1.6 and 3.1.7
Determine the material and cross- Reinforcing steel properties: Section 2.4.2.4 and 3.2.7
1
sectional properties Pre-stressed reinforcing strands:
Section 2.4.2.4, 3.3.6 and 3.3.3
Concrete covers: Section 4.4.1.2
Determine the factored loads and

Section 8.3 and
2 calculate factored shear, torque, and Section 2.4.3 and 5.1.3 Annex C
9.2.1
bending moment
Section 11.5.1 Section 11.2.9.1,

2
fc Acp Eq. 11-2
Tu [ fc in MPa ] 2
g
f p fcp
< fl (1.5 A )
Determine if torsion effects can be 12 pcp Tcr = 0.38lf c fc 1 + [ fc in MPa ]
3 pc 0.38lf c fc
disregarded
2
2
A cp (1.5 A )g
f p fcp
Tu < fl fc [ fc in psi ] Tcr = 0.38lf c 12 fc 1 + [ fc in psi ]
pc 0.38lf c 12 fc
p cp

Section 11.2.1.1,
Eq. (11-3), Section 11.3.2, Eq. (11-9),
and Section 11.5.3.1, Eq. (11-18)
Vc 2l fc
= [ fc in MPa ]
bw d 12
Vc
= 2l fc [ fc in psi ]
bw d Section 6.3.2(4)
Eq. 6-14, 6-29, and 6-30 Section 11.3.10.4,
f c
V d u TEd VEd Eq. 11-18
Determine if dimensions of the cross + 700(0.00689) a c bw zvfcd
c w c
V = 0.6 l b d [ f in MPa ]
4 12 VRd , max = + 1
u
M Vf Vp Tf
section are adequate ( cot q + tan q) TRd , max VRd , max + 0.25f c fc
Vu d bw d v 1.7 Aoh t
fc + 700 TRd , max = 2na c fcd Ak t ef sin q cos q
w
Vc = 0.6 l b d [ fc in psi ]
Mu
Tu ph Vc 2 fc
2
f + [ fc in MPa ]
1.7 Aoh b w
d 3
Tu ph V c

2
f +8 fc [ fc in psi ]
1.7 Aoh b d w

Shear resistance of concrete: Section
11.2.1.1, Eq. (11-3) and Section 11.3.2,
Eq. (11-9) The angle of diagonal compression strut:
Vc 2l fc Section 11.3.6.4, Eq. 11-12 and 11-13
= [ fc in MPa ] 2
Mf 2 h
0.9 p T f
bw d 12 + + 0.5 N A f
f p po
(Vf Vp ) +
, dv 2 A o
Vc ex =
= 2l fc [ fc in psi ] 2[ E s As + E p Ap ]
Design shear resistance of a member without shear reinforce-
bw d
ment: Section 6.2.2(1). If not adequate, At Tf
f c
V d Required shear reinforcement: Section 6.2.3(4); Eq. 6-13 and
u
c
V = 0.6 l + 700(0.00689) b d [ f in MPa ] w c 6-14 s 2 Ao f s fy cot q
12 M u
Asw Asw VEd Shear resistance of concrete: Section 11.3.4,

Shear design to determine Vu d VRd , s = zfywd cot q VEd
s s zf cot q Eq. 11-6 and 11-11
ywd
5 transverse reinforcement Vc = 0.6 l fc + 700
bw d [ fc in psi ]
Mu Vc = f c lb fc bw d v [ fc in MPa ]
for shear force
a c bw zvfcd
Required shear reinforcement: Section 11.4.7 VRd , max =
Av Vu fVc ( cot q + tan q) Vc = f c lb12 fc bw d v [fc in psi]
= Minimum shear reinforcement: Section 9.2.2(5);
s fdfyv 0.40 1300
Maximum effective cross-sectional area of the shear rein- b=
forcement: Section 6.2.3(3) 1 + 1500 e x 1000 + sze
Minimum shear reinforcement: Section
11.4.6.3, Eq. 11-13 Required shear reinforcement: Section
fc bw 11.3.5.1, Eq. 11-7 can be transformed into
Av , min = 0.75 [ fc in MPa ] Av Vs
12 fyt =
s f s fy d v cot q
bw
Av , min = 0.75 fc [ fc in psi]
65
f yt
66
Assume the angle of diagonal compres- Calculate the terms of the equivalent thin-walled section for
Required transverse reinforcement for torsion:
sion strut. Required transverse reinforce- torsion design: Section 6.3.2(1)
Section 11.3.10.3, Eq. 11-17 can be trans-
ment for torsion: Section 11.5.3.6 and Required transverse reinforcement for torsion: Section
Torsion design to determine transverse formed into
6 11.5.6, Eq. (11-21) transformed into 6.3.2(2)
reinforcement for torsion At Tf
At Tu Asw TEd
= s 2 Ao f s fy cot q
s f2 Ao fyv cot q s 2 Ak fywd cot q
Determine the total transverse rein-

forcement for shear and torsion Add the shear and torsion transverse rein- Minimum transverse reinforcement:
forcement in one web: Section 11.5.3.8 Section 11.2.8.2, Eq. 11-1
Minimum transverse reinforcement: Add the shear and torsion transverse reinforcement in one
Av , min bw
Section 11.5.5, Eq. (11-23) web: Section 6.3.2(2) = 0.06 fc [ fc in MPa ]
Maximum longitudinal spacing of the stirrups for shear: Eq. s fy
t v
A 1 A fc bw
= 0.375 [ fc in MPa ] 9.6N in EC2-04
+ = 90
s 2 s min
12 fyt Av , min bw
smax = 0.75d (1 + cot a ) a smax = 0.75d = 0.72 fc [ fc in psi ]
7 At 1 Av bw s fy
Maximum longitudinal spacing of the stirrups for torsion:
s + 2 s = 0.375 fc f [ fc in psi]
min yt Section 9.2.3(3) Maximum longitudinal spacing of transverse
Select and check transverse reinforce- u/8 reinforcement: Section 11.3.8.1
Maximum longitudinal spacing of
ment details for cross section
transverse reinforcement for shear: ACI smax = min 0.75d (1 + cot q ) 0.7 d v
318-11, Section 11.4.5 smax = min
min( b, h ) 600 mm (23.62 in.)
Maximum longitudinal spacing of
transverse reinforcement for torsion: ACI Add the shear and torsion transverse rein-
318-11, Section 11.5.6 forcement in one web
Determine longitudinal reinforcement

Required torsional longitudinal reinforce-
for torsion
ment: Section 11.5.3.7, Eq. (11-22)
At f yv Required torsional longitudinal reinforce-
Al = ph cot 2 q
s f yl ment: Section 11.3.10.6 11.3.9.2, Eq. 11-14
Required torsional longitudinal reinforcement: Section
and 11-21
Minimum longitudinal reinforcement for 6.3.2(3), Eq. 6-28

can be transformed into
torsion: Section 11.5.5.3, Eq. (11-24) 2
TEd uk cot q
SAs Mf 2 0.45 p T h f
8
5 fc Ag A t
fyv 2 Ak fywd F t = + cot q (Vf 0.5Vs Vp ) +
ph [ fc in MPa ] dv 2 A o
Select and check longitudinal reinforce- A , min = 12 f s
yl
fyl
ment details Spacing of longitudinal reinforcement for torsion: Section
F
for cross section 5 fc Ag
9.2.3 (4) Ap = t
t
A f yv
A , min = p [ f in psi ]
h c
f p f pr
fyl s f yl
Spacing of longitudinal reinforcement for

torsion: Section 11.5.6.2 and 11.5.6.3
Fig. 9.3Design process flow chart for combined shear and torsion effects.
9.4Design Example 1: solid rectangular

report focuses on recent torsion developments and theories,
reinforced concrete beam under pure torsion
the design examples are solved by three major building codes:
9.4.1 Design problem statementAs shown in Fig. 9.4.1,
ACI 318-11, EC2-04, and CSA-A23.3-04. Example 1 was
the cross-sectional dimensions of beam are bw = 300 mm
also solved by the two graphical methods proposed by Rahal
(12 in.) and h = 500 mm (20 in.). The characteristic concrete
(2000b) and Leu and Lee (2000). These graphical methods
cylinder compressive strength is fc = 20 MPa (2900 psi)
are not suitable to treat prestressed concrete sections under
[Class C20/25], and the characteristic steel yield strength is
combined loads and, therefore, were not used for Example 2.
fy = 420 MPa (60,000 psi). The applied torsional moment is
Following each design example, a comparison table summa-
Tu = 30 kNm (266 in.-kip). The mean cylinder strength of
rizes results of the various codes and methods. Design equa-
the concrete: fcm = fc + 8 MPa fcm = 28 MPa (4000 psi).
tions and expressions may vary slightly from one code to
Assume 40 mm (1.5 in.) from exterior face to stirrup center-
another, but the general design procedure is the same. In
line typical.
the case of torsion combined with shear, design procedures
9.4.2 Solution according to ACI 318-11
for all three codes follow the flow chart given in Fig. 9.3.
1. Determine if torsion effects may be disregarded (ACI
Hsu (1997) provides background on ACI 318-95 that is still
318-11, Section 11.5.1).
applicable to the updated ACI 318-11. His paper includes a
The torsion effects can be disregarded if the following
detailed design example of a prestressed hollow box girder
expression is valid
subjected to torsion, shear, and flexure that constitutes
Example 2.
ph = 2(xo + yo) = 1280 mm (50.4 in.)

xo, yo = horizontal and vertical dimension of the centerline
of outermost closed transverse torsional reinforce-
ment, respectively
xo = (300 2 40) = 220 mm (9 in.)
yo = (500 2 40) = 420 mm (17 in.)
Aoh = area enclosed by centerline of outermost closed
transverse torsional reinforcement, mm (in.)
Aoh = xoyo = 92,400 mm2 (153 in.2)
Therefore
Tu ph 30 106 1280
= = 2.65 MPa (347.6 psi)
2
1.7 Aoh 1.7 92, 400 2
Fig. 9.4.1Rectangular
reinforced concrete
cross section subjected whereas
to pure torsion.
5 fc 5 20 2.80 MPa (403 psi)
fc Acp2 f = 0.75 =
Tu < fl 6 6

12 pcp

Because the expression in Eq. (11-18) (ACI 318-11) is
valid, the cross section dimensions are adequate.
where
3. Calculate the amount of stirrups required for pure
f = strength reduction factor for torsion and shear = 0.75
torsion (ACI 318-11, Section 11.5.3.3).
l = modification factor of lightweight concrete
To calculate transverse reinforcement for torsion, Eq.
(normalweight concrete, l = 1.0)
(11-20) and (11-21) in ACI 318 can be transformed into
pcp = perimeter of outer concrete cross section, mm (in.)
= 2 (bw + h) = 1600 mm (64.0 in.)
At Tu
Acp = total area enclosed by the outside perimeter of
concrete cross section, mm (in.) = bw h = 150,000 s f2 f yt Ao cot q

mm2 (240 in.2)
Therefore, where
fc Acp2 20 150, 000 2 At = area of one leg of a closed stirrup resisting torsion
fl = 0. 75(1. 0 ) 12 1600 Nmm s = spacing of the stirrups
12 pcp
fyt = design strength of torsion transverse reinforcement
= 3.93 kNm (36.3 in.-kip) = 420 MPa (60,000 psi)
Tu = 30.0 kNm (266 in.-kip). Ao = gross area enclosed by shear flow path = 0.85Aoh
Torsion effects must be considered. q = angle of compression diagonals in truss analogy for
2. Determine if dimensions of the cross section are torsion = 45 degrees
adequate (ACI 318-11, Section 11.5.3.1). Therefore,
Dimensions of the cross section are adequate if
30 10
6
At Tu
=
2
Vu Tu ph
2
V fc s f2 fyt Ao cot q 0.75 2 420 0.85 92, 400 1
+
b d 1.7 A2 f b d + 8 12
c
w oh w = 0.61 mm2/mm (0.0227 in.2/in.)

The minimum transverse reinforcement is computed by
In case of pure torsion, the expression becomes Eq. (11-23) in ACI 318.
In the case of pure torsion, this minimum is expressed by
Tu ph 5 fc V 2l fc
f , taking c =
0.75 fc bw 0.75 20 300
2

1.7 Aoh 6 bw d 12 At
s = 2 12 f = 24 420
min yv

where
= 0.1 mm2/mm (0.0040 in.2/in.)
ph = perimeter of centerline of outermost closed trans-
verse torsional reinforcement, mm (in.)
but shall not be less than
At 0.35 bw 0.175 300 A = total area of the outside perimeter of the concrete cross
s = 2 f = 420
section
min yv
= bwh = 300 500 = 150,000 mm2 (240 in.2)
= 0.125 mm2/mm (0.005 in.2/in.) pc = outside perimeter of the concrete cross section
= 2(bw + h) = 2(300 + 500) = 1600 mm (64.0 in.)
The maximum spacing of the stirrups is given by (ACI c = distance between edge of member and center of the
318-11, Section 11.5.6.1) longitudinal reinforcement
= 50 mm (2.0 in.)
p /8 The area enclosed by the centerlines of the connecting
smax = min h = 160 mm (6.50 in.) thin-walls is calculated by
300 mm
Ak = (bw tef)(h tef) = 200 400 = 80,000 mm2 (124.0 in.2)
For stirrups 8 (ds = 8.0 mm [0.315 in.]), At = 50.3 mm 2
(0.078 in.2) and s 81 mm (3.20 in.). Select: s = 80 mm The perimeter of the area enclosed by the centerlines of
(3.15 in.). the connecting thin-walls is determined by
For stirrups No. 3 (ds = 9.5 mm (0.375 in.)), At = 71.0 mm2
(0.11 in.2) and s 115 mm (4.55 in.). Select: s = 110 mm
(4.50 in.).
( ) ( )
uk = 2 bw tef + h tef = 2(200 + 400)
4. Calculate the longitudinal bars required for pure torsion = 1200 mm (48.0 in.)
(ACI 318-11, Section 11.5.3.7).
Total longitudinal reinforcement for torsion (Al) (Eq. 2. Determine if dimensions of the cross section are
(11-22) in ACI 318) is calculated by adequate (evaluate the strength of concrete struts) (EC2-04,
Section 6.3.2(4)).
At f yt Dimensions of the cross section are adequate if (Eq. (6-29)
A = ph cot 2 q = 0.610 1280 1 1 in EC2-04)
s f y

= 780.8 mm2 (1.18 in.2) TEd V
+ Ed 1 pure torsion
TEd <= TRd , max
TRd , max VRd,max
Minimum longitudinal reinforcement (Eq. (11-24) in ACI
318) is calculated by
where
5 f A A f 5 20 150, 000 TEd = design torsional moment = Tu
= p = 0.610 1280 1
c cp t yt
A VEd = design shear force = Vu
, min
12 f s f
y
h
y
12 420
TRd, max = design torsional resistance moment
< 0 mm2 VRd, max = design shear resistance force
The design torsional resistance moment is given as (Eq.
Therefore, A = 780.8 mm2 (1.18 in.2) (6-30) in EC2-04)
At least one longitudinal bar is required in each stirrups
corner and bar spacing distributed around the perimeter of the TRd , max = 2 va c fcd Ak tef sin q cos q

closed stirrups is 305 mm (12 in.). Therefore, the number of
longitudinal bars is at least six and each bar requires780.8/6
where
= 130 mm2 (0.20 in.2).
v = strength reduction factor for concrete cracked in
For six longitudinal bars 14 (ds = 14.0 mm [0.551 in.]),
2 2 shear
Al = 6 153.9 = 923 mm (1.43 in. ).
= 0.6(1 fc/250) = 0.6(1 20/250) = 0.552 (fc in MPa)
For six longitudinal bars No. 5 (ds = 15.9 mm [0.625 in.]),
ac = 1 (for non-prestressed elements)
Al = 6 200 = 1200 mm2 (1.86 in.2).
q = angle of compression diagonals in truss analogy for
9.4.3 Solution according to EC2-04 code
torsion
1. Calculate the terms of the equivalent thin-walled section
1 cotq 2.5; assume q = 35 degrees
(EC2-04, Section 6.3.2(1)).
fcd = design compressive strength of concrete
Effective wall thickness is calculated by
= fck/gc = 20/1.5 = 13.3 MPa (1933 psi)
Therefore,
A / pc 150,000/1600 = 94 mm
tef = max = max
2 c 2 50 = 100 mm TRd , max = 2 va c fcd Ak tef sin q cos q
tef = 100 mm (4.0 in.) = 2 0.552 1 13.3 80, 000 100 sin 35 cos 35(10 6 )
= 55.2 kNm (513.4 in.-kip)
where TRd , max = 55.2 kNm (513.4 in.-kip) > Tu = 30 kNm (266 in.-kip)

Because the above expression is valid, dimensions of the 4. Calculate the longitudinal bars required for pure torsion
cross section are adequate. (EC2-04, Section 6.3.2(3)).
3. Calculate the amount of stirrups required for pure The total longitudinal reinforcement needed for torsion
torsion (EC2-04, Section 6.3.2 (2)). (SAl) is given as (Eq. (6-28) in EC2-04)
The amount of stirrups required is calculated using the
equation Tu uk cot q 30 106 1200 cot 35
A = A
2 Ak f yd 2 80, 000 365.2
At Tu
880 mm2 (1.37 in.2)
s 2 Ak f ywd cot q

Minimum longitudinal reinforcement is determined by
where (Eq. (9.1N) in EC2-04)
At = area of one leg of a closed stirrup resisting torsion
s = spacing of the stirrups fctm
fywd = design strength of transverse reinforcement A , min = 0.26 bt d 0.0013bt d
f yk
= fty/gs = 420/1.15 = 365.2 MPa (52,174 psi)
Therefore,
where
Tu 30 10
6 fctm = mean tension strength of concrete
= = 0.30fck2/3 = 0.3(202/3) = 2.21 MPa (319 psi)
2 Ak f ywd cot q 2 80, 000 365.2 cot 35
bw = mean width of tension zone = 300 mm (12 in.)
= 0.359 mm2/mm (0.014 in.2/in.) Therefore,
At 0.26 ( 2.21 / 420) 300 450 = 184.7 mm

2
0.359 mm2/mm (0.014 in.2/in.) A = max
s
0.0013 300 450 = 175 mm
2
The ratio of the required transverse reinforcement is given = 185 mm2 (0.287 in.2)
by
Longitudinal bars are arranged with at least one bar at each
At A 0.359 corner of the stirrups and the others distributed uniformly
rt = a
= 90
rt = t = = 0.0012 around the inner periphery of the torsion links (closed stir-
sbw sin a sbw 300
rups) with a maximum spacing of 350 mm (13.8 in.). There-
fore, the number of longitudinal bars is at least six and each
where a is angle between the stirrups and longitudinal axis. bar requires 880/6 = 147 mm2 (0.23 in.2).
The ratio of the minimum stirrups is (Eq. (9.5N) in EC2-04) For six longitudinal bars 14 (ds = 14 mm [0.55 in.]), A
= 6 154 = 924 mm2 (1.43 in.2).

( )
rt = 0.08 fc / f yk = 0.08 20 / 420 = 0.00085 For six longitudinal bars No. 5 (ds = 15.8 mm [0.625 in.]),
A = 6 198 = 762 mm2 (1.88 in.2).
9.4.4 Solution according to CSA-A23.3-04 code
The maximum longitudinal spacing of the stirrups is given 1. Determine if torsion effects may be disregarded (CSA-
as (Eq. (9.6N) and Section 9.2.3(3) in EC2-04). A23.3-04, Section 11.2.9.1).
If the magnitude of the torsion, Tf, satisfies the following
u/8 expressions (Eq. (11-2) in CSA-A23.3-04), torsional effects

smax = min 0.75d (1 + cot a ) need not be considered
min(b , h )
w Tf < 0.25Tcr
1, 600 / 8 = 200
A2 A2
smax = min 0.75 450 = 338 = 200 mm (8.0 in.) Tcr = c 0.38lfc fc 1 +
f p fcp
= c 0.38lfc fc
pc 0.38lfc fc pc
300
For stirrups 8 (ds = 8.0 mm [0.315 in.]), At = 50.3 mm2 150, 000 2
= 0.38 1 0.65 20 Nmm
1600
(5.0 in.).
For stirrups No. 3 (ds = 9.5 mm [0.375 in.]), At = 71 mm2 = 15.5 kNm (144 in.-kip)
(7.1 in.). where
Ac = Acp = 150,000 mm2 (240 in.2)
pc = pcp = outside perimeter of concrete cross section = where

1600 mm (64.0 in.) At = area of one leg of a closed stirrup resisting torsion
l = factor to account for low-density concrete s = spacing of the stirrups
= 1.0fc = resistance factor for concrete = 0.65 Ao = gross area enclosed by shear flow path = 0.85Aoh
fc = specified compressive strength of concrete, which q = angle of inclination of compression stresses to the
is the same as the characteristic concrete cylinder longitudinal member axis
compressive strength in previous definition = 20 4. Calculate angle of inclination of compression strut
MPa (2.9 ksi) (CSA-A23.3-04, Section 11.3.6.4).
Therefore, The angle of inclination of the diagonal compression strut
is given by the expression (Eq. (11-12) in CSA-A23.3-04)
Tf (= 30.0 kNm) > 0.25Tcr (= 0.25 15.5 = 3.9 kNm)
q = 29 + 7000ex
Torsion effect, therefore, must be considered.
2. Determine if dimensions of the cross section are The longitudinal strain indicator ex is defined by (Eq.
adequate (CSA-A23.3-04, Section 11.3.10.4). (11-13) in CSA-A23.3-04)
Dimensions of the cross section are adequate if the equa-
tion below (Eq. (11-19) in CSA-A23.3-04) is satisfied 2
Mf 0.9 ph T f
+ (V f Vp )2 + + 0.5 N f Ap f po
dv 2 Ao
V f Vp T f ph
2 2
ex =
b d + 1.7 A2 0.25fc fc 2[ Es As + E p Ap ]
w v oh

For non-prestressed sections subjected to pure torsion, the
In case of pure torsion, the expression is simplified to expression for ex is simplified to
T f ph
2
0.25fc fc 0.9 ph T f
1.7 Aoh 2 Ao 0.9 ph T f
ex = =
2 Es As 4 Es As Ao

With a concrete cover of 40 mm (1-1/2 in.) and 10M (No. 3)
stirrups (diameter 12 mm [1/2 in.]), the following calcula-
with Ao = 0.85Aoh = 0.85 84,860 = 72,130 mm2 (120 in.2)
tions/values apply
and the longitudinal reinforcement As = 413 mm2 (0.64 in.2)
(established in Section 6 of this example), the following is
Aoh = (300 2 46)(500 2 46) = 84,864 mm2 (141.0 in.2)
obtained
ph = 2[(300 2 46) + (500 2 46)] = 1232 mm (50.0 in.2)
0.9 1232 30 106
ex = = 0.00140
Therefore, 4 200, 000 413 72,130

Tu ph 30 106 1232 Therefore

= = 3.02 MPa (394 psi)
2
1.7 Aoh 1.7 84, 860 2

q = 29 + 7000 0.00140 = 38.8 degrees
0.25 f c f c = 0.25 0.65 20 = 3.25 MPa (471 psi)
5. Calculate stirrups required for pure torsion.
The required stirrup area per unit length is calculated
Because 3.02 MPa < 3.25 MPa, dimensions of the cross using the equation (Eq. (11-17) in CSA-A23.3-04)
section are adequate.
3. Calculate the stirrups required for pure torsion (CSA- At Tf 30 106
A23.3-04, Section 11.3.10.3); =
s 2 Ao f s f y cot q 2 72,130 0.85 420 cot 38.8
The equation of nominal torsional strength Tn is same in
all codes except for differences in strength reduction factors. = 0.468 mm2/mm (0.0175 in.2/in.)
For design, the equation for Tn is rearranged to express the
required area of transverse reinforcement per unit length The minimum transverse reinforcement is (Eq. (11-1) in
(Eq. (11-7) in CSA-A23.3-04) CSA-A23.3-04)
At Tf Av bw 300
s = 0.06 fc f = 0.06 20 420
s 2 Ao f s f y cot q min yv

= 0.192 mm2/mm (0.00775 in.2/in.)
The maximum spacing of the stirrups is (CSA-A23.3-04, 1. Determine if torsion effects may be disregarded.
Section 11.3.8.1) The torsion effects can be disregarded if this expression is
valid: Tu 0.25fTcr
0.7dv
smax = min = 277 mm (11.2 in.)
600 mm Ac2 150, 000 2
Tcr = 0.328 fc = 0.328 20 Nmm
pc 1600

where = 20. 6 kNm (193.87 in.-kip)
dv = effective shear depth = max(0.9d, 0.72h) = max(396
mm, 360 mm) [(16.0 in., 14.4 in.)] The torque is disregarded if Tu = 30 kNm (266 in.-kip) is
For stirrups 10M (ds = 11.3 mm [0.444 in.]), At = 100 mm2 smaller than 0.25fTcr = 0.25 0.75 20.6 = 3.9 kNm (35.8
(0.155 in.2) s 210 mm (8.32 in.). Select s = 200 mm in.-kip). Torsion effects, therefore, must be considered.
(8.0 in.). 2. Calculate normalized shear stress and determine if size
For stirrups No. 3 (ds = 9.5 mm [0.375 in.]), 71.0 mm2 (At of cross section is adequate.
= 0.110 in.2) s 150 mm (5.89 in.). Select: s = 150 mm Shear stress in the walls of the cross section is calculated by
(5. 9 in.).
6. Calculate the amount of longitudinal bars required for Tu 30 106
pc 1600
pure torsion (CSA-A23.3-04, Section 11.3.9.2 and 11.3.10.6). v f 0.75
Total longitudinal reinforcement is calculated using the = = = 0.212
fc 0.67 Ac2 fc 0.67(150, 000)2 20
following equation (Eq. (11-21) and (11-14) in CSA-A23.3-04)
2 The normalized shear stress fits well within Region I in

Mf 0.45 ph T f 0.45 ph T f
F t = + cot q (V f 0.5Vs Vp )2 + = cot q Fig. 5.7.1. Therefore, the section can be designed as under-
dv 2 Ao 2 Ao
reinforced, and the section dimensions are adequate.
3. Calculate the required reinforcement.
where The most straightforward design of an under-reinforced
Ft = required tension force in longitudinal reinforcement = section uses equal amounts of longitudinal and transverse
fAlfy reinforcement indexes (w = wt = v/fc). Therefore, w = 0.212
Therefore and wt = 0.212. From Eq. (5.7.1b) and (5.7.1c), the longitu-
dinal and transverse reinforcement are
0.45 ph T f 0.45 1232 30 106
A cot q = cot 38.8 At 0.42 Ac fc (0.42)(150, 000)(20)
2f s Ao f y 2 0.85 72,130 420 = wt = 0.212
s fty pc (420)(1600)
= 402 mm2 (0.61 in.2)
= 0.40 mm2/mm (0.016 in.2/in.)
One longitudinal reinforcing bar should be placed in each
corner of the closed transverse reinforcement required for 0.375 Ac fc (0.375)(150, 000)(20)
A = wl = 0.212
torsion. The nominal diameter of the corner bars should be f y 420

no less than s/16. If 10M stirrups are used, the bar nominal
= 568 mm2 (0.92 in.2)
diameter can be no less than 200/16 = 12.5 mm (8/16 = 0.5
in.), and for No. 3: 150/16 = 9.38 mm (6/16 = 0.375 in.).
The stirrups maximum spacing is taken as 160.0 mm (6.30
Select: four longitudinal bars 15M (ds = 16.0 mm [0.628
in.), which is the smaller of ph/8 = 2(220 + 420)/8 = 160.0 mm
in.]) at the corners and two 10M bars at mid-depth: A = 4
(6.50 in.) and 300.0 mm (12.0 in.), as per ACI code.
200 + 2 100 = 1000 mm2 (1.55 in.2).
For stirrups 8 (ds = 8.0 mm [0.315 in.]), At = 50.3 mm2
Or, six longitudinal bars No. 3 (ds = 9.5 mm [0.375 in.]),
(0.08 in.2), the maximum spacing is calculated as s 124.5
A = 6 71 = 426 mm2 (0.66 in.2)
mm (4.9 in.), and it can be used: s = 120 mm (4.50 in.).
Or, six longitudinal bars No. 4 (ds = 12.7 mm [0.50 in.]),
For stirrups No. 3 (ds = 9.5 mm [0.375 in.]), At = 71.0 mm2
A = 6 127 = 762 mm2 (1.18 in.2).
(0.11 in.2), the maximum spacing is calculated as s 175.7
9.4.5 Rahals graphical methodRahals graphical
mm (6.9 in.), and it can be used: s = 160 mm (6.50 in.).
method uses the ACI general requirements, such as an upper
To provide four longitudinal corner bars and limit the spacing
limit on spacing of transverse and longitudinal reinforce-
to 300 mm (12 in.), six bars are needed (ACI design in 8.2.2).
ment, an upper limit on yield strength of reinforcing rebars,
The minimum bar area is of 578/6 = 96 mm2 (0.15 in.2).
and a minimum of four corner longitudinal bars. Based on
For six longitudinal bars 12 (ds = 12.0 mm [0.472 in.]),
ACI requirements, it also disregards the torque effect if it
A = 6 113 = 678 mm2 (1.05 in.2).
is smaller than 25 percent of the cracking torque. The cross
For six longitudinal bars No. 4 (ds = 12.7 mm [0.50 in.]),
section outer perimeter and area enclosed within this perim-
A = 6 129 = 774 mm2 (1.20 in.2).
eter were calculated in the ACI design example 9.4.1 as: Ac =
9.4.6 Leu and Lees graphical method
150,000 mm2 (240 in.2) and pc = 1600 mm (64.0 in.).
1. Calculate the required torsional strength.
Table 9.4.6Summary of design solution of Example 1 using all five solution methods
Code Transverse reinforcement Longitudinal reinforcement
required: 0.61 mm2/mm (0.0227 in.2/in.) 780.8 mm2 (1.18 in.2)
2 2
minimum: 0.125 mm /mm (0.0042 in. /in.) 0 mm2 (0 in.2)
ACI 318
8/80 mm (No. 3 at 4.50 in.) 614 (6 No. 5)
provided:
0.625 mm2/mm (0.0244 in.2/in.) 923 mm2 (1.86 in.2)
required: 0.359 mm2/mm (0.014 in.2/in.) 880 mm2 (1.37 in.2)
2 2
minimum: 0.261 mm /mm (0.0103 in. /in.) 185 mm2 (0.287 in.2)
EC2-04
8/125 mm (No. 3 at 7.10 in.) 614 (6 No. 5)
provided:
0.402 mm2/mm (0.0155 in.2/in.) 924 mm2 (1.88 in.2)
minimum: 0.192 mm2/mm (0.00775 in.2/in.)
CSA A23.3-04 Four 15M + two 10M
10M/200.0 mm (No. 3 at 5.90 in.)
provided: (6 No. 3 or 6 No. 4)
0.500 mm2/mm (0.0186 in.2/in.)
1000 mm2 (0.66 or 1.18 in.2)
required: 0.40 mm2/mm (0.0160 in.2/in.) 568.1 mm2 (0.92 in.2)
minimum:
Rahal (2000b)
8/120.0 mm (No. 3 at 6.50 in.) 612 (6 No.4)
provided:
0.417 mm2/mm (0.0183 in.2/in.) 678 mm2 (1.18 in.2)
minimum:
Leu and Lee (2000)
8/100.0 mm (No. 3 at 5.50 in.) 612 (6 No. 4)
provided:
0.503 mm2/mm (0.0200 in.2/in.) 678 mm2 (1.18 in.2)
Nondimensional balanced torsional strength is calculated Required reinforcement indexes (Fig. 5.7.2): w and wt
using Eq. (5.7.2g) For convenience, assume w = wt.
Referring to (Fig. 5.7.2), ws 0.73 w = wt 0.70.
140 140 Balanced normalized reinforcement ratios hb and htb are
Txub = ( f y in MPa ) = = 0.196
300 + f y 300 + 420 calculated by Eq. (5.7.2c) and (5.7.2d)

76
Balanced torsional strength Txub is determined by Eq. hb = (f in MPa) = 76
= 0.125
200 + f y y 200 + 420
(5.7.2h)
Txub fcAc2 0.196 20 150, 000 2 76

Txub = = Nmm htb = (f in MPa) = 76
= 0.135
pc 1600 100 + f y y 100 + 420

= 55.1 kNm (512 in.-kip)
Required normalized reinforcement ratios h and ht are
where determined using Eq. (5.7.2e) and (5.7.2f)
Ac = total area of the outside perimeter of the concrete
cross section h = w h b 0.70 0.125 h 0.0861

= bwh = 300 500 = 150,000 mm2 (240 in.2)
pc = outside perimeter of the concrete cross section
ht = w t htb 0.70 0.135 ht 0.105
= 2(bw + h) = 2(300 + 500) = 1600 mm (64.0 in.)
The torsional strength indicated by Eq. (5.7.2h) is greater
than required Required longitudinal bars are calculated with Eq. (5.7.2a)
Tu f y A A
w s Txub w s 55.1 30 fs 0.73w hl 0.0868 0.0868 420 0.0861
f 0.75 fc Acp 20 150, 000
A 615 mm2 (1.0 in.2)
where
Tu = applied torsional moment = 30.0 kNm (266 in.-kip) At least one longitudinal bar is placed at each corner of
f = strength reduction factor for shear and torsion; the stirrups with the others distributed uniformly around
which is assumed to be the same as that in ACI 318 the inner periphery of the closed stirrups with a maximum
= 0.75ws = strength contour value (Fig. 5.7.2) spacing of 300 mm (12 in.), as indicated in ACI 318. There-
2. Calculate the longitudinal bars and stirrups required for fore, the number of longitudinal bars is at least six and each
pure torsion. bar requires 615/6 = 102.5 mm2 (0.17 in.2).
For six longitudinal bars 12 (ds = 12.0 mm [0.472 in.]), load, which consists of two truckloads located symmetri-
A = 6 113 = 678 mm2 (1.05 in.2). cally at a distance 3200 mm (10 ft 6 in.) from midspan. Each
For six longitudinal bars No. 4 (ds = 12.7 mm [0.500 in.]), axle load is taken as 24 percent of the crush live load (513.8
A = 6 127 = 762 mm2 (1.18 in.2). kN/4) with 100 percent impact and a maximum side shift
Required stirrups are calculated using Eq. (5.7.2b) of 914 mm (36.0 in.). The self-weight of the girder is 34.4
kN/m (2.36 kip/ft). The girder is also subjected to a superim-
f y At pcp posed dead load caused by the track rails weight, rail plinth
ht 0.1036 0.1036 pads, power rail, guard rail, cableway, acoustic barrier, and
fc Acp s
other permanent loads. At derailment, this superimposed
420 1600 At dead load is assumed to produce a uniform vertical load of
0.1036
20 150, 000 s 12.8 kN/m (0.88 kip/ft) and a uniformly distributed torque of
3.16 kN-m/m (0.71 ft-kip/ft). This torque is neglected in the
At
0.470 mm2/mm (0.019 in.2/in.) calculation because the magnitude of the distributed torque
s is small, and the torque is acting in a direction opposite to
the derailment torque.
Using ACI 318, the maximum stirrups spacing is calcu- 9.5.2 Solution according to ACI 318-11
lated to be 160 mm (6.3 in.). 1. Determine the factored forces for (ACI 318-11, Section
For stirrups 8 (ds = 8.0 mm [0.315 in.]) At = 50.3 mm2 9.2.1).
(0.08 in.2) and s 107 mm (4.2 in.) and it can be used: s = Factored dead and live loads
100 mm (4.0 in.). The load factor for live loads is taken as 1.6.
For stirrups No. 3 (ds = 9.5 mm [0.375 in.]) At = 71 The derailment load per axle is calculated by
mm2 (0.11 in.2) and s 151 mm (5.9 in.) and it can be used:
s = 150 mm (5.5 in.). 513.8
Pu,L = 16 2 = 411 kN/axle (92.4 kip/axle)
4
9.5Design Example 2: Prestressed concrete box
girder under combined torsion, shear, and flexure
The derailment torque per axle is calculated by
9.5.1 Design problem statement
9.5.1.1 Design problem descriptionDesign the shear and
torsional reinforcement of a box girder. A 3658 mm (12 ft) Tu,L = 16 513.8 2 0.914
wide and 1270 mm (4 ft 2 in.) deep box girder with over- 4
hanging flanges (Fig. 9.5.1.1(a)) was designed as an alterna- = 375.7 kNm/axle (277 ft-kip/axle)
tive to the double-tee girder in Dade County, FL (Hsu 1997).
The standard prestressed box girder is simply supported, 24.00 The load factor for dead loads is taken as 1.2.
m (79.00 ft) long, and prestressed with 64 strands at 1860 MPa The girder weight is calculated by
(270 ksi), 13.0 mm (1/2 in.), seven-wire strands as shown in
Fig. 9.5.1.1(b). Total prestress force is 6076 kN (1366 kips) wu,g = 1.2(34.4) = 41.3 kN/m (2.83 kip/ft)
after prestress loss. The design of flexural reinforcement is
omitted for simplicity. The concrete cover is 40 mm (1.5 in.), The superimposed dead weight is calculated by
and material strengths are normalweight concrete: fc = 48.0
MPa (7000 psi) and fy = 420 MPa (60,000 psi). wu,s = 1.2(12.8) = 15.4 kN/m (1.05 kip/ft)
9.5.1.2 Sectional properties
L = 24.00 m (79.00 ft) Factored shear, torque, and bending moment
h = 1270 mm (50.00 in.) The Vu, Tu, and Mu at 0.3L from the support are
d = 1016 mm (40.00 in.) at 0.3L from support Vu = (wu,g + wu,s)(0.2L) + 2Pu,L = (41.3 + 15.3) 0.2
t = 251 mm (9.88 in.) (average of stem width) 24.00 + 2 411 = 1094 kN (246 kips)
bw = 502 mm (20 in.) Tu = 2Tu,L = 2 375.7 = 752 kNm (554 ft-kip)
A = 1.523 106 mm2 (2361.4 in.2) Mu = 0.5(wu,g + wu,s)(L 0.3L)(0.3L) + 2Pu,L(0.3L)
I = 319.8 109 mm2 (768,336 in.4) = 0.5(41.3 + 15.4)(24.00 7.2) 7.2 + 2 411 7.2
yt = 516 mm (20.34 in.) = 9347 kNm (6922 ft-kip)
yb = 753 mm (29.66 in.) 2. Determine if torsion effects may be disregarded.
l = modification factor of lightweight concrete (l = 1.0) Check outstanding flanges
9.5.1.3 Loading criteriaThe standard girders are designed As indicated by Fig. 9.5.1.1(a), the parameter Acp2/pcp is
to carry a train of cars, each 22.86 m (75 ft 0 in.) long. Each determined by (disregard overhanging flanges)
car has two trucks with a center-to-center distance of 16.46 Acp = 1854 203 + 0.5(1854 + 1791) 1067 = 2.32 106
m (54 ft 0 in.). Each truck consists of two axles 1981 mm mm2 (3597 in.2)
(6 ft 6 in.) apart. The crush live load of each car is 513.8 kN pcp = 1854 + 1791 + 2 1270 = 6185 mm (243.5 in.)
(115.5 kip). The maximum web reinforcement amount was Acp2/pcp =(2.32 106)2/6185 = 8.70 108 mm3 (53,090 in.3)
obtained at section 0.3L from the support under a derailment Check threshold torque (ACI 318-11, Section 11.5.1)
Fig. 9.5.1.1Cross section and elevation of box girder.
Ag = 2.32 106 (1270 419)(1320) = 1.20 106 mm2 Assume a clear concrete cover of 40 mm (1.5 in.) and 13
(1855 in.2)(fTcr/4) mm No. 4 bars for web reinforcement
= Aoh = 0.5[(1854 93) + (1791 93)](1270 93)
= 2.04 106 mm2 (3177 in.2)
Ag2 f pc Ag2 6076 / A ph = (1854 93) + (1791 93) + 2(1270 93)
f(0.083)l fc 1 + = f(0.083)l fc 1 + = 5813 mm (229.5 in.)
pcp 0.33 fc pcp 0.33 fc
9754 7224

( ) e = (753 127) 508
2
1.20 106 6076 / 1523 6 9754
=0.75(0.083)(1.0) 48.0 1 + 10 = 496 mm (19.47 in.) at 0.3L from support
6185 0.33 48.0
d = yt + e = 516 + 496 = 1012 mm (39.81 in.) at 0.3L
= 276 kNm (202 ft-kip) < 752 kNm (554 ft-kip) from support
Factored torsional moment should be considered in design. d = 0.8h = 0.8 1270 = 1016 mm (40.00 in.) governs
3. Determine if dimensions of the cross section are bw = 2t = 2 251 = 502 mm (20.0 in.)
adequate (ACI 318-11, Section 11.5.3.1). bwd = 502 1016 = 510 103 mm2 (790 in.2)
Check cross section The interaction equation for hollow box sections is (Eq.
(11-18) in ACI 318-11)
Vu Tu Vc = 2.043 mm2/mm (0.0818 in.2/in.)

Aoh
b d + 1.7 A t f b d + 8 0.083 fc when t < p Av , min fc bw
w oh w h > = 0.75 = 0.51 mm2/mm (0.021 in.2/in.)
s 12 f yt
Aoh smax = 305.0 mm (12.00 in.) for torsion governs
= 2.04 10 = 351 mm (13.85 in.) > t = 251 in.
6
ph Design of torsional hoop reinforcement (ACI 318-11,

5829
Section 11.5.3.6)
2Tu pcp 2 752 6185 (1000)
2
Ao = Acp = 2.32 106
Vu Tu 1094 1000 ffcAcp 0.75 48 2.32 106
751.4 106
+ = +

b d 1.7 A t
w oh 510 103 ( )
1.7 2.04 106 251 = 2.21 106 mm2 (3582.7 in.2)
Assume q = 37.5 degrees, as recommended by the code
= 2.15 + 0.86 = 3.01 MPa (437 psi) provision for prestressed members:
752 (1000)
2
Ats Tu
Vu d = =
Vc = 0.6 0.083l fc + 700 0.0689
M w
b d, s ( )
f2 Ao f yv cot q 0.75 2 2.21 106 420 1.303
u
= 0.414 mm2/mm (0.0166 in.2/in.)

smax = ph/8 = 5813/8 = 727 mm (28.7 in.) > 305 mm (12
Vu d
where 1 (Eq. (11-9) in ACI 318-11) in.) (ACI 318-11, Section 11.5.6);
Mu These calculations indicate that 305 mm (12 in.) spacing
governs.
Transverse reinforcement for vertical walls (ACI 318-11,
Vu d 1094 1016 Section 11.5.3.8)
= = 0.119 < 1 OK Transverse reinforcement in the vertical walls is contrib-
Mu 9347 1000
uted by both torsion and shear:
At 1 Av
+ = 0.414 + 0.5 2.043
Vc = (0.6 0.083(1.0) 48.0 + 700 0.00689 0.119) s 2 s
510 103 = 1.44 mm2/mm (0.0575 in.2/in.)
1000 = 469 kN (105.4 kip)
At 1 Av fc bw
s + 2 s = 0.375 12 f
Vc,min = 2 0.083l fc bwd min yt

= 0.259 mm2/mm (0.01 in.2/in.)
510 103 < 1.44 mm2/mm (0.0575 in.2/in.) OK
= 2 0.083(1.0) 48.0
1000 (Eq. (11-23) in ACI 318-11)
(Eq. (11-3) in ACI 318-11) Select two layers of 18 mm bars (No. 6 bars) in each
vertical wall at 305 mm (12 in.) spacing
= 587 kN (132.13 kip) governs
2 ( 254)
= 1.666 mm2/mm (0.0656 in.2/in.)
V 588 1000 (305)
f c + 8 0.083 fc = 0.75 + 8 0.083 48.0
bw d 510 103 > 1.440 mm2/mm (0.0575 in.2/in.) OK
Transverse reinforcement for horizontal walls

= 0.75(1.15 + 4.61) = 4.31 MPa (627 psi) > 3.01 MPa (437 Transverse reinforcement in horizontal walls is contrib-
psi) OK uted by torsion only:
4. Calculate number of transverse bars required
Design of shear reinforcement (ACI 318-11, Section At
11.4.7) = 0.414 mm2/mm (0.0166 in.2/in.)
s
Vc = Vc,min = 587 kN (132.2 kip)
Av Vu fVc (1094 0.75 587)1000
= = Select two layers of 10 mm bars (No. 3 bars) in each
s fdf yv 0.75 1016 420 horizontal wall at 305 mm (12 in.) spacing:

2 78.5 - Required average compressive strength (psi)

= 0.515 mm2/mm (0.0203 in.2/in.)
305
fcr = fc + 1400 (for fc > 5000 psi)
> 0.259 mm2/mm (0.01 in.2/in.)
The transverse reinforcement in the top wall should be fcr = 7000 + 1400 = 8400 psi (57.9 MPa)
added to the flexural reinforcement required in the top flange
acting as a transverse continuous slab. - Mean compressive strength (based on fcr)
5. Calculate number of longitudinal bars required.
Design of torsional longitudinal reinforcement (Eq. fcm = 57.9 MPa (8400 psi)
(11-22) in ACI 318-11)
- Characteristic compressive strength (MPa) (EC2-04,
A f yv Section 3.1.2(5))
A = t ph cot 2 q = 0.414 5813 1 1.3032
s f y fc = fcm 8 fck = 49.9 MPa (7240 psi)

= 4085.9 mm2 (6.46 in.2)
- Strength class for concrete according to Section 3.1.2
Check minimum limitation for At/s and A,min (Eq. (11-23) (Table 3.1) in EC2-04 corresponding to mean compressive
in ACI 318-11) strength
At b Strength Class C50, therefore: fc = 50 MPa (7250 psi)

= 0.375 fc w = 0.259 mm2/mm (0.01 in.2/in.)
s f yt
- Design compressive strength (EC2-04 , Section 3.1.6(1))
< 0.414 mm2/mm (0.0166 in.2/in.)
fc 50
fcd = a cc = 1 = 33.3 MPa (4833 psi)
gc 1.5

5 fcAg A f yv
A ,min = t ph
f yl s f yl - Mean compressive strength according to EC2-04

(Eq. (11-24) in ACI 318-11) (Section 3.1.6(5)) and the strength class of concrete
fc = fcm + 8 fcm = 58.0 MPa (8410 psi)
=
(
5 0.083 48.0 1.2 10 6
)
0.414 5813 1 Concrete tensile strength (EC2-04, Section 3.1.6 (2)
420 - Mean tensile strength
= 8215 2407 = 5808 mm2 (9.17 in.2) governs
ft = 0.3fc2/3 = 4.07 MPa (590 psi)
Select 36 16 bars (No. 5 bars) longitudinal bars
- Design tensile strength
A = 36 201 = 7236 mm2 (11.22 in.2) > 5808 mm2 (9.17 in.2)
fctk ,0.05 0.7 ft
fctd = a ct = 1 = 1.89 MPa (275 psi)
Arrangement of reinforcing bars gc 1.5
The arrangement of the reinforcing bars for torsion and shear
is summarized in Table 9.4.6. This reinforcement arrangement
could be conservatively used throughout the girder length. - Modulus of elasticity
9.5.3 EC2-04 code 0.3
9.5.3.1 Material properties f
Em = 22 cm = 37.28 GPa (5405 ksi) (EC2-04, Section
9.5.3.1.1 Concrete 10

Concrete compressive cylinder strength (EC2-04, Section
3.1.3, Table 3.1)
3.1.2 and 3.1.6)
- Specified compressive strength (The characteristic
Parabola-rectangle diagram for concrete under compres-
concrete cylinder compressive strength)
sion, as shown in Fig. 9.5.3.1.1 (EC2-04, Section 3.1.7)
fc = 48.0 MPa (7000 psi)
Area As = 99 mm2 (0.153 in.2)

Characteristic tensile strength fpk = fpu = 1860 MPa (270 ksi)
Design ultimate strength (EC2-04, Section 2.4.2.4)
f pk 1860
f p,ud = = = 1617 MPa (234.78 ksi)
gs 1.15

Characteristic yield strength (EC2-04, Section 3.3.3(1))
fp0.1k = 0.9fpu = 1674 MPa (243 ksi)

Fig. 9.5.3.1.1Parabola-rect-
angle diagram for concrete under Design yield strength (EC2-04, Section 3.3.6(6))
compression.
f p 0.1k 1674
f pd = = = 1456 MPa (211 ksi)
e
n
gs 1.15
s c = fcd 1 1 c
e c2
for
ec < ec2
Modulus of elasticity (EC2-04, Section 3.3.6(3))
sc = fcd for ec ec2
where Ep = 196 GPa (28,420 ksi)
90 fc
4
n = 1.4 + 23.4
100
Strain at design yield strength
= 2.00 f pd
e c 2 = (2 + 0.085 ( fc 50) ) / 100
0.53
e p, yd = = 0.0074
Ep
= 0.0020
90 fc
4
e cu 2 = 2.6 + 35 / 100 Ultimate design strain (EC2-04, Section 3.3.6(7))

100
= 0.0035 ep,ud = 0.0200
9.5.3.1.2 Reinforcing steel
Type: Ribbed reinforcing bars (high bond) 9.5.3.2 Sectional areas and concrete cover
Diameter for longitudinal bars = 12 mm [12] (0.472 in.) 9.5.3.2.1 Gross concrete area
Diameter for closed stirrups = 10 mm [10] (0.394 in.) - Area of the gross cross section including overhanging
Characteristic yield strength fy = 420 MPa (60 ksi) flanges
Design yield strength (EC2-04, Section 2.4.2.4)
= 2.65 106 mm2 (4107 in.2)
fy 420
f yd = = = 365 MPa (52 ksi)
g s 1.15 - Area of the gross cross section (disregarding overhanging

flanges)
Modulus of elasticity (EC2-04, Section 3.2.7(4))
= 2.32 106 mm2 (3596 in.2)
Es = 200 GPa (29,000 ksi)
- Area of the concrete cross section (disregarding over-
hanging flanges)
Strain at design yield strength
Ac = 1.20 106 mm2 (1860 in.2)
f yd
ey = = 0.0018
Es 9.5.3.2.2 Concrete covers (EC2-04, Section 4.4.1.2);

- Minimum cover with regard to bond
Ultimate design strain (EC2-04, Section 3.2.7(2))
cmin,b = diameter of bar = 12 mm (0.47 in.) (reinforcing steel)
eud = 0.9euk = 0.0200 = 13 mm (0.51 in.) (prestressing steel)
9.5.3.1.3 Prestressing reinforcing strands - Exposure class related to environmental conditions

Type: Low relaxation, 270K, seven-wire strands
Exposure Class XD3 (Cyclic wet and dry. Parts of bridges
Diameter = 13.0 mm (0.50 in.)
exposed to spray containing chlorides.)
- Structural class (XD3) = 4 + 2 (service life of 100 years) 9.5.3.4 Prestressed tendons requirements
1 (concrete class C45) = 5 Minimum longitudinal reinforcement (EC2-04, Section
9.2.1.1)
- Minimum cover with regard to durability
ft
cmin,dur = 50 mm (1.97 in.) (reinforcing steel) A = 0.26 bt d 0.0013bt d
f yk
= 60 mm (2.36 in.) (prestressing steel)
- Minimum cover where bt = 470 mm (18.5 in.) mean width of tension zone,
and d = 1172 mm (46.1 in.) distance from extreme top fiber
cmin = max(cmin,b; cmin,dur; 10 mm [0.39 in.]) to the centroid of the reinforcement.
= 50 mm (1.97 in.) for reinforcing steel Therefore,
= 60 mm (2.36 in.) for prestressing steel
A,min = 716 mm2 (1.1 in.2)
- Nominal cover
The area of the prestressed tendons is
cnom = cmin + 10 mm (0.39 in.)
= 60 mm (2.36 in.) for reinforcing steel AP = 64 99 = 6336 mm2 (9.82 in.2) As,min OK
= 70 mm (2.76 in.) for prestressing steel
9.5.3.5 Shear design
- Distance from center of longitudinal reinforcing bars to 9.5.3.5.1 Design shear resistance of a member without
extreme concrete fiber (cover from bars centroid) shear reinforcement (EC2-04, Section 6.2.2(1))
(
= 66 mm (2.56 in.) > cmin + 12 mm/2 (0.47 in./2)
VRd ,c = C Rd ,c k (100r fck )

1/ 3
)
+ k1s cp bw d vmin + k1s cp bw d
- Distance from center of prestressed tendons to extreme
bottom fiber (cover from tendons centroid) where
d = 517 + 494 = 1011 mm (39.8 in.) at 0.3L = 7.2 m
= 102 mm (3.94 in.) > cmin + 13 mm/2 (0.51 in./2) (23.7 ft) (9.5.1.1(b))
bw = 470 mm (18.50 in.)
9.5.3.3 Factored shear, torque, bending moment, and prestress CRd,c = 0.18/gc = 0.18/1.5 = 0.12
Factored shear force k = 1+ 200 d 2 k = 1.44
Vu = 1.4 (34.4 + 12.8) 0.2 24.0 + 1.5 (513.8/2) 2 Ap
r = 0.02 r = 0.0133
= 1089 kN (245 kip) bw d
N Ed Pt
Factored torsional moment s cp = = = 5.1 MPa (734.4 psi) (compressive)
Ac Ac
Tu = 1.5 (513.8/2)2 0.914 = 705 kNm (520 ft-kip)
k1 = 0.15
vmin = 0.035k3/2fck1/2 vmin = 0.428
Factored bending moment Therefore
Mu = 0.5 1.4 (34.4 + 12.8) 0.7 0.3 (24.00)2 VRd,c = 698 kN (156.2 kip)
+ 1.5(513.8/2)(2 7.2) = 9561 kNm (7074 ft-kip)
Because
Prestress force at time t =
VRd,c < VEd = 1089 kN (245 kip),
Pt = 6076 kN (1366 kip) shear reinforcement must be provided
Total prestress losses 9.5.3.5.2 Minimum shear reinforcement (EC2-04, Section
9.2.2(5))
20 percent or w = Pt/Po = 0.80
The ratio of the minimum stirrups is
( )
Therefore, prestress force at time t = 0
rw,min = 0.08 fck f yk = 0.08 49.9 / (1.2 420) = 0.00113
Po = Pt/w = 7595 kN (1708 kip)
Therefore
Asw A Asw A VEd A 1089 103

0.00113 sw 0.5311 mm 2 /mm (0.0209 in.2 /in.) VRd , s = zf yd cot q VEd sw sw
sb s min s s zf yd cot q s 910 365 cot q
w min
9.5.3.5.3 Maximum effective cross-sectional area of the q = 45 degrees

shear reinforcement (EC2-04, Section 6.2.3(3))
Asw 3.28 mm2/mm (0.131 in.2/in.),

Asw.max f yd 1 A a vb f s
a c vfcd sw c w cd
sbw 2
s max 2 f yd

q = 35 degrees
where
Asw 2.30 mm2/mm (0.092 in.2/in.), and
v = 0.6(1 fck/250) = 0.48
ac = 1 + scp/fcd = 1.154 (because scp 0.25fcd) s
Therefore
q = 22 degrees
Asw
s 11.86 mm /mm (0.47 in. /in.)
2 2
Asw 1.32 mm2/mm (0.053 in.2/in.).
max
s
9.5.3.5.4 Required shear reinforcement (EC2-04, Section

Check
6.2.3(4))
Asw Asw Asw
VRd , s s < s < s OK
VRd = min VEd
VRd ,max min max

9.5.3.6 Torsion design
a c bw zvfcd
VRd , max = (Eq. (6.14) in EC2-04) 9.5.3.6.1 Terms of the equivalent thin-walled section
(cot q + tan q) (EC2-04, Section 6.3.2(1));
With the above data and A 375 mm

tef = max u treal = max 178 mm
2c 134 mm
z = 0.9d z = 910 mm (35.8 in.)
tef = 178 mm (7 in.)
fcd = 33.3 MPa (4830 psi)
where
1 cotq 2.5 22 degrees q 45 degrees A = 2.32 106 mm2 (3597 in.2) total area of the cross
(mean value: q = 35 degrees) section within the outer circumference, including
inner hollow areas (disregarding overhanging flanges)
1.154 470 910 0.48 33.3
VRd, max = u = 6185 mm (243.5 in.) outer circumference of the
(cot q + tan q) cross section (disregarding overhanging flanges)
c = 66 mm (2.56 in.) distance from edge to center of the
For longitudinal reinforcement
q = 45 degrees treal = 178 mm (7 in.) minimum wall thickness of the real
concrete thin-walled section.
VRd,max = 3945 kN (886 kips) > VEd, The continuous area and perimeter enclosed by centerlines
of the connecting thin-walls, as shown in Fig. 9.5.3.6.1, are:
q = 35 degrees Ak = 1.8 106 mm2 (2794 in.2) and uk = 5486 mm (216 in.).
9.5.3.6.2 Determine if cross-sectional dimensions (for
VRd,max = 3707 kN (832 kips) > VEd, and example, strength of concrete struts under torsion) are
adequate (EC2-04, Section 6.3.2(4))
q = 22 degrees Dimensions of the cross section are adequate if (Eq. (6.29)
in EC2-04)
VRd,max = 2740 kN (615 kips) > VEd.
TEd VEd
+ 1
Thus for (Eq. (6.13) in EC2-04) TRd , max VRd , max

VRd = VRd,s VEd
Fig. 9.5.3.6.1Assumed effective thin-wall section.
The design torsional resistance moment is calculated as TEd V 705 1089

+ Ed = + = 0.57 1 OK
(Eq. (6.30) in EC2-04) TRd ,max VRd ,max 4085 2740

TRd,max = 2vacfcdAktefsinqcosq
Dimensions of the cross section, therefore, are adequate
for every case (every value of q).
where
9.5.3.6.3 Calculate the required stirrup area for torsion.
v = 0.6(1 fck/250) = 0.48 and ac = 1 + scp/fcd = 1.154 (EC2-04, Section 6.3.2(2))
The required stirrup area per unit length is calculated by
and for
q = 45 degrees Asw TEd A 705 106
sw
s 2 Ak f yd cot q s 2 1.8 106 365 cot q

TRd,max = 5915 kNm (4356 ft-kip),
For
q = 35 degrees
q = 45 degrees
TRd,max = 5559 kNm (4093 ft-kip)
Asw
0.54 mm2/mm (0.021 in.2/in.),
q = 22 degrees s
TRd,max = 4109 kNm (3026 ft-kip) q = 35 degrees
Design shear resistance force is determined by (Eq. (6.14) Asw

in EC2-04) 0.38 mm2/mm (0.015 in.2/in.), and
s
a c bw zvfcd
VRd , max = q = 22 degrees
(cot q + tan q)
Asw
0.22 mm2/mm (0.009 in.2/in.)
Thus for s
q = 45 degrees
where Asw and s are the area of one leg of a closed stirrup
TEd V 705 1089 resisting torsion and the spacing of the stirrups, respectively.
+ Ed = + = 0.39 1 OK
TRd , max VRd , max 5910 3945 9.5.3.6.4 Add stirrup areas for torsion and shear, and

select the stirrups (EC2-04, Section 6.3.2(2))
q = 35 degrees
Asw, S + T 1 Asw, S Asw,T
= + (using two single-legged stirrups)
TEd VEd 705 1089 s 2 s s
+ = + = 0.42 1 OK
TRd , max VRd , max 5554 3707
For
q = 45 degrees
q = 22 degrees
Asw, S + T q = 45 degrees
2.18 mm2/mm (0.087 in.2/in.),
s
SAs 2943 mm (4.62 in.2),
q = 35 degrees q = 35 degrees
Asw, S + T SAs 4204 mm2 (6.60 in.2), and

1.53 mm2/mm (0.061 in.2/in.), and
s
q = 22 degrees
q = 22 degrees
SAs 7285 mm2 (11.50 in.2).
Asw, S + T
0.88 mm2/mm (0.035 in.2/in.). Longitudinal bars shall be arranged so that at least one
s bar is placed at each corner of the stirrups and the others
are distributed uniformly around the inner periphery of the
Maximum longitudinal spacing of the stirrups for shear torsion links (closed stirrups) with a maximum spacing of
(Eq. (9.6N) in EC2-04) 350 mm (13.8 in.) (EC2-04, Section 9.2.3(4)). Therefore, the
number of the longitudinal bars is at least 36.
smax = 0.75d (1 + cot a ) a
= 90
smax = 0.75d For

q = 45 degrees
= 758 mm (29.8 in.)
3810 SAs = 2985 mm (4.63 in.2) > 2943 mm (4.63 in.2),
where a is the inclination of the stirrups.
Maximum longitudinal spacing of the stirrups for torsion q = 35 degrees
(EC2-04, Section 9.2.3(3))
3812 SAs = 4298 mm2 (6.66 in.2) > 4204 mm2 (6.61 in.2),
u/8 773 mm and

smax = min 0. 75 d( +
1 cot q ) = min 758 mm
min(b, h ) 235 mm q = 22 degrees

= 235 mm (9.25 in.) 3816 SAs = 7640 mm2 (11.84 in.2) > 7285 mm2 (11.58 in.2).
For two single-legged 10 stirrups, spacing is calculated Final selection of longitudinal bars (q = 35 degrees)
from shear and torsion
For 12 (0.47 in.) As = 113 mm2 (0.175 in.2).
q = 45 degrees
Due to uniform distribution and symmetry
s 72 mm (2.8 in.) < smax OK
4012 SAs = 4524 mm2 (7.0 in.2).
q = 35 degrees
The torsional longitudinal reinforcement is in addition
s 102 mm (4.0 in.) < smax OK to the prestressing tendons. Low values of angle q (q = 22
degrees) lead to a design with lower area requirements of
q = 22 degrees transverse reinforcement (10/175 mm) and higher area
requirements of longitudinal reinforcement (4016). High
s 178 mm (7.0 in.) < smax OK values of angle q (q = 45 degrees) lead to higher area require-
ments of stirrups (10/70 mm) and lower area requirements
Total selection (q = 35 degrees): two single-legged stir- of longitudinal bars (4010).
rups of 10/100 mm (diameter 0.39 in. at 4.0 in.). The selection of q = 45 degrees maximizes the concrete
9.5.3.6.5 Longitudinal reinforcement required for torsion strength components, such as VRd,max = 3945 kN and TRd,max
(EC2-04, Section 6.3.2(3)) = 5910 kNm. This value of angle q could be used when
Total longitudinal reinforcement for torsion (Eq. (6-28) in checking the adequacy of the cross section dimensions.
EC2-04) The following relationship between the strength of
concrete struts under torsion is satisfied
TEd uk cot q 705 106 5486 cot q
SAs SAs
2 Ak f yd 2 1.8 106 365 TEd V
+ Ed 1
TRd , max VRd , max

For
Fig. 9.5.3.6.6Reinforcement configuration (EC2-04).
9.5.3.6.6 Arrangement of reinforcing barsTotal rein- The superimposed dead weight (with a load factor of 1.25)
forcement of the member with a hollow cross section under is calculated by
combined loading of prestressing, torsion, shear, flexure,
and axial force is depicted in detail in Fig. 9.5.3.6.6. wu,s = 1.25 12.8 = 16.0 kN/mm (1.10 kip/ft)
9.5.4 Design solution using CSA-A23.3-04 codeLoad
factors of the CSA-A23.3-04 code are used here to establish Factored shear, torque, and bending moment
the forces due to factored loads. These factors are 1.25 for At distance 0.3L from support, the following values are
dead loads and 1.50 for live loads. The cross-sectional area obtained
of the prestressing strands (fpu = 1860 MPa [270 ksi]) is 6336
mm2 (9.82 in.2), the effective prestressing force is 6076 kN Vu = (43.0 + 16.0)(0.2 24.00) + 2 385 = 1053 kN (237 kip)
(1366 kip), and the average prestress is 3.99 MPa (589 psi).
The concrete strength is 48.0 MPa (7000 psi), and the yield Tu = 2 352.3 = 705 kNm (520 ft-kip)
strength of the non-prestressed reinforcement is 420 MPa
(60,000 psi). Mu = 0.5(43.0 + 16.0)(24.00 7.2)7.2 + 2 385 7.2
= 9112 kNm (6759 ft-kip)
1. Determine the factored forces (CSA-A23.3-04, Annex C).
Factored dead and live loads
2. Determine if torsion effects can be disregarded (CSA-
The derailment load per axle is calculated by
A23.3-04, Section 11.2.9.1).
Threshold torque
513.8 For a hollow section with a wall thickness of less than
Pu,L = 1.5
4
2 = 385 kN/axle (86.6 ft/axle)
0.75Ac/pc (= 281 mm > 178 mm), torsion must be considered
if the torque due to factored loads, Tf, exceeds 0.25Tcr
The derailment torque per axle is determined by
(1.5 A )
2
g f p fcp
513.8 Tcr = 0.38lfc fc 1 +
Tu,L = 1.5 pc 0.38lfc fc
4
(2 0.914) = 352 kNm (259.8 ft-kip)

The girder weight (with a load factor of 1.25) is deter- where

mined by Ac = Acp = 2.32 106 mm2 (3597 in.2)
pc = pcp = outside perimeter of concrete cross section =
wu,g = 1.25 34.3 = 43.0 kN/mm (2.95 kip/ft) 6185 mm (243.5 in.)
Ag = gross area of section (without flanges) = 1.20 106
mm2 (1855 in.2)
l = factor to account for low-density concrete = 1.0
fc = resistance factor for concrete = 0.70 for precast The angle of inclination of the diagonal compression strut
concrete is given by the expression (Eq. (11-12) in CSA-A23.3-04)
fp = resistance factor for prestressing reinforcement =
0.90 q = 29 + 7000ex
fc = specified compressive strength of concrete = 48.0
MPa (7.0 ksi) In the absence of an axial load normal to the cross section,
fcp = compression stress in concrete due to effective the strain at mid-depth of the section is defined by (Eq.
prestress = 3.99 MPa (0.6 ksi) (11-13) in CSA-A23.3-04)
With this information, the following calculation can be made
2
Mf 0.9 ph T f
(1.5 1.20 10 ) 0.9 3.99 + (V f Vp )2 + Ap f po
2 Ao
6 2
Tcr = 0.38 0.70 48.0 1 + dv
6185 0.38 0.70 48.0 ex =
2[ Es As + E p Ap ]
= 1658 kNm (1216 ft-kip)
Because 0.25Tcr = 415 kNm (304 ft-kip) < 705 kNm (520 The terms not defined above are
ft-kip), torsion must be considered in the design. Mf = Mu = moment due to factored loads = 9166 kNm
3. Determine if dimensions of the cross section are adequate. (6761 ft-kip)
Check cross-sectional dimensions (CSA-A23.3-04, Section Ao = 0.85Aoh = 0.85 2.04 106 mm2 = 1.74 106 mm2
11.3.10.4) (2697 in.2)
For box sections with a wall thickness of less than Aoh/ph, Ap = area of prestressing reinforcement = 6336 mm2
the cross-sectional dimensions must satisfy the following (9.8 in.2)
criterion (Eq. (11-19) in CSA-A23.3-04) fpo = stress in prestressing tendons (may be taken as
0.7fpu = 1302 MPa [189.0 ksi])
V f Vp Tf Es = 200,000 MPa (29,000 ksi)
+ 0.25fc fc As = area of non-prestressed reinforcement in tension
bw dv 1.7 Aoh t
zone (assume fourteen 15M bars with 2800 mm2
(4.34 in.2)
where Ep = 190,000 MPa (28,000 ksi)
Vf = Vu = shear force due to factored loads = 1054 kN With this information, the following calculation can be
(237.2 kip) made
Vp = shear force due to prestressing factored by fp
= 0.9(6076)508/9754 = 284 kN (64.0 kip) 9112 106
2
0.9 5801 705 106
+ [(1054 284) 103 ]2 + 1302 6336
Aoh = [(1854 + 1791)/2 2 48)](1270 2 48) = 2.03 914 2 1.74 106
ex = = 0.00086
2(200, 000 2800 + 190, 000 6336)
106 mm2 (3142 in.2)
ph = (1854 96) + (1791 96) + 2(1270 96) = 5801
mm (228.4 in.) Therefore
Aoh/ph = 350 mm (13.8 in) > 178 mm (7.0 in.) for bottom
flange q = 29 + 7000 0.00086 = 35.0
> 235 mm (9.25 in.) for web
> 203 mm (8 in.) for top flange The shear force resisted by the concrete is (Eq. (11-6) in
dv = larger of 0.9d = 0.9 1016 = 914 mm and 0.72h = CSA-A23.3-04)
0.72 1270 = 914 mm (36 in.)
t = minimum wall thickness = 178 mm (7.0 in.) Vc = fc lb fcbw dv

Tu = torque due to factored loads = 705 kNm (520 ft-kip)
Therefore,
where (Eq. (11-11) in CSA-A23.3-04)
(1054 284) 103 705 106
+ 0.40 1300
235 2 914 1.7 2.03 106 235 b=
1 + 1500 e x 1000 + sze
= 2.66 MPa (386 psi) < 0.25 0.70 48.0 = 8.40 MPa
(1223 psi) OK
Because minimum transverse reinforcement is provided,
Because the web governs design, the above equation sze = 300 mm (12 in.)
was the web thickness t = 235 mm, not the bottom flange Therefore
thickness.
4. Calculate q and b (CSA-A23.3-04, Section 11.3.6.4). b=
0.40

1300
= 0.175
Angle of diagonal compression strut and shear resistance 1 + 1500(0.00086) 1000 + 300
of concrete
and Design of longitudinal reinforcement for torsion (CSA-

A23.3-04, Section 11.3.10.6 and 11.3.9.2)
Vc = 0.70 1.0 0.175( 48.0 )235 2 914 N Longitudinal reinforcement on the tension side of the girder
= 365 kN (80.7 kip) shall have dimensions such that its factored resistance shall
not be less than that given by the following expression. In
5. Calculate transverse reinforcement. the absence of an axial load normal to the cross section, the
Design of transverse reinforcement for shear (CSA-A23.3- required longitudinal force due to torsion, shear, and flexure is
04, Section 11.3.5.1)
Shear with Vs = Vf Vc Vp = 1054 365 284 = 405 kN Mf 0.45 ph T f
2
(92.6 kip) F t = + cot q (V f 0.5Vs Vp ) + 2
dv 2 Ao

Av Vs 405 103
= =
s f s f y dv cot q 0.85 420 914 cot 35.0 9112 106 405
2
0.45 5801 705 106
2
= + cot 35.0 1054 284 (103 ) +

914 2 2 1.74 106
= 0.87 mm2/mm (0.0356 in.2/in.)
Design of transverse reinforcement for torsion (CSA- = 9969 + 1108 = 11,077 kN (2503 kip)
A23.3-04, Section 11.3.10.3)
The required transverse reinforcement for torsion is given Factored tension resistance provided by the prestressing
by (Eq. 11-17 in CSA-A23.3-04) reinforcement with fpr = 0.96fpu = 1786 MPa (259 ksi) (CSA-
A23.3-04, Section 18.6.2)
At Tu 705 106
= = F tr = f p Ap f pr = 0.9 6336 1786 10 3
s 2 Ao f s f y cot q 2 1.74 106 0.85 420 cot 35.0

= 10,184 kN (2290 kips)
= 0.40 mm2/mm (0.0160 in.2/in.)
The non-prestressed reinforcement required in tension
This amount of transverse reinforcement must be provided
zone is calculated by
in the top and bottom slabs.
Total transverse reinforcement in one web is calculated as
(11, 077 10,184)(103 )
As = = 2501 mm2 (4.26 in.2)
0.85 420
At Av = 0.40 + 0.87/2 = 0.835 mm2/mm (0.0338 in.2/in.)
+ These calculations indicate that fourteen 15M bars are
s 2s
required, providing As = 2800 mm2. This reinforcement is to
be distributed in the tension zone.
Because the compression due to moment is much greater
The selection of two 10M bars (Abar = 100 mm2) per web
than the tension due to torsion and shear, there is no need to
yields the following bar spacing
check the tension reinforcement required in the compres-
s = 2(100)/0.835 = 240 mm; select s = 225 mm (9.17 in.) sion zone.
9.5.5 Comparison of design solutions according to ACI
Minimum transverse reinforcement (CSA-A23.3-04, 318, CSA-A23.3-04, and EC2-04The required longitu-
Section 11.2.8.2) dinal and transverse reinforcement calculated using ACI
The minimum transverse reinforcement is established 318, CSA-A23.3-04, and EC2-04 is shown in Table 9.5.5.
with the maximum web thickness at the top of the web, bw = The longitudinal reinforcement (L Bars) given is the total
534 mm (21.0 in.) amount required for the entire cross section. The required
transverse reinforcement is given for the top wall (T-1
Bars), side walls (T-2 Bars), and bottom walls (T-3
min Av b 534
= 0.06 fc w = 0.06 48.0
420
Bars). The transverse reinforcement in the top wall should
s fy be added to the flexural reinforcement required in the top

= 0.528 mm2/mm (0.0212 in.2/in.) wall acting as a transverse continuous slab.
< 0.835 mm2/mm (0.0338 in.2/in.) Arrangement of reinforcing bars
Arrangement of the reinforcing bars for torsion and shear
For Tf > 0.25Tcr, max s = 0.35dv = 0.35(914) = 320 mm is shown in Fig. 9.5.5 and summarized in Table 9.5.5.
(12.6 in.) > 225 mm. (9.17 in.) OK The general location of longitudinal and transverse rein-
6. Calculate the longitudinal reinforcement. forcement in the section, which is shown in Fig. 9.5.5, is
based on all three design codes.
Canadian Standards Association

CSA-23.3-77Design of Concrete Structures
CSA-23.3-04Design of Concrete Structures
Abul Mansur, M., and Rangan, B., 1978, Torsion in
Spandrel Beams, Journal of the Structural Division, V. 104,
pp. 1061-1075.
ACI Committee 315, 2004, ACI Detailing Manual, SP-66,
American Concrete Institute, Farmington Hills, MI, 166 pp.
ACI Committee 363, 1992, Report on High-Strength
Concrete (ACI 363R-92), American Concrete Institute,
Farmington Hills, MI, 55 pp.
ACI Committee 438, 1969, Tentative Recommendations
Fig. 9.5.5Reinforcement locations for box girder under for the Design of Reinforced Concrete Members to Resist
combined torsion, shear, and flexure. Torsion, ACI Journal, V. 66, No. 1, Jan., pp. 1-8.
Ali, M. A., and White, R. N., 1997, On Extending ACI
318 to High-Strength Concrete, Proceedings of the First
Table 9.5.5Summary of design solution of International Conference on High Strength Concrete, Kona,
Example 2 using design codes ACI 318-11, HI, July 13-18, pp. 554-567.
EC2-04, and CSA-A23.3-04 Alkhrdaji, T., and Belarbi, A., 2003, Shear Flow Zone
Design code in Rectangular RC Members Subjected to Pure Torsion,
EC2-04 Proceedings of the International Conference on the Perfor-
Reinforcing (45, 35, and 22 mance of Construction Materials in the New Millennium,
bars ACI 318-11 degrees) CSA-A23.3-04 Cairo, Egypt, Feb. 17-20, pp. 555-564.
5808 mm2 4204 mm2 2503 mm2 Allos, A. E., and Rashid, A. H., 1989, Prestressed
L bars
(9.2 in.2) (6.61 in.2) (4.26 in.2)
Concrete Rectangular Beams Subjected to Sustained
2.18, 1.53,
Torque, ACI Structural Journal, V. 86, No. 4, July-Aug.,
T-1 and T-3 0.414 mm2/mm 0.88 mm2/mm 0.40 mm2/mm
bars (torsion) (0.0166 in.2/in.) (0.087, 0.061, (0.0160 in.2/in.) pp. 469-472.
0.035 in.2/in.) American Association of State Highway and Transporta-
T-2 bars 2.18, 1.53, tion Officials, 1998, AASHTO LRFD Bridge Design Speci-
(torsion + shear) 0.88 mm2/mm 0.835 mm2/mm fications and Commentary, SI units, second edition, Wash-
1.44 mm2/mm (0.087, 0.061, (0.0338 in.2/in.) ington DC, 1091 pp.
(0.0575 in.2/in.) 0.035 in.2/in.) Bach, B., 1911, Elastizitt und Festigkeit, sixth edition.
(in German)
CHAPTER 10REFERENCES Badawy, H., and Jordaan, I., 1977, Plastic Analysis of
Committee documents are listed first by document number Curved Beams under Point Loads, Journal of the Structural
and year of publication followed by authored documents Division, V. 103, pp. 1429-1455.
listed alphabetically. Baumann, T., 1972, Zur Frage der Netzbewehrung von
Flachentragwerken (On the Problem of Net Reinforcement
American Concrete Institute of Surface Structures), Der Bauingenieur, V. 47, No. 10,
318-63Building Code Requirements for Reinforced pp. 367-377.
Concrete Baant, Z. P., 1984, Size Effect in Blunt Fracture:
318-71Building Code Requirements for Reinforced Concrete, Rock, Metal, Journal of Engineering Mechanics,
Concrete V. 110, No. 4, pp. 518-535.
318-77Building Code Requirements for Reinforced Baant, Z. P., 2002, Scaling of Structural Strength, Hermes
Concrete Penton Science, Ltd. (Kogan Page Science), London, UK,
318-95Building Code Requirements for Structural 336 pp.
Concrete Baant, Z. P., and Oh, B. H., 1983, Crack Band Theory
318-02Building Code Requirements for Structural for Fracture of Concrete, Matriauxet Constructions, V. 16,
Concrete May-June, pp. 155-177.
318-08Building Code Requirements for Structural Baant, Z. P., and Planas, J., 1998, Fracture and Size
Concrete (ACI 318-08) and Commentary Effect in Concrete and Other Quasibrittle Materials, CRC
318-11Building Code Requirements for Structural Press, Boca Raton, FL and London, UK, 640 pp.
Concrete (ACI 318-11) and Commentary Baant, Z. P., and Sener, S., 1987, Size Effect in Torsional
Failure of Concrete Beams, Journal of Structural Engi-
European Committee for Standardization
neering, V. 113, No. 10, pp. 2125-2136.
Eurocode 2-04Design of Concrete StructuresPart 1-1
Baant, Z. P.; Sener, S.; and Prat, P. C., 1988, Size Effect
Eurocode 8-04Design of Structures for Earthquake
Tests of Torsional Failure of Plain and Reinforced Concrete
ResistancePart 1
Beams, Materials and Structures, V. 21, pp. 425-430.
Baant, Z. P., and Yu, Q., 2005a, Designing Against Size in Structural Concrete, SM Study No. 8, University of
Effect on Shear Strength of Reinforced Concrete Beams Waterloo Press, Waterloo, ON, Canada, pp. 211-231.
Without Stirrups: I. Formulations, Journal of Structural Collins, M. P., and Chockalingam, S., 1979, Reinforced
Engineering, V. 131, No. 12, pp. 1877-1885. Concrete Beams Under Reversed Cyclic Torsional Loading,
Baant, Z. P., and Yu, Q., 2005b, Designing Against Size Structural Concrete Under Seismic Actions, CEB Bulletin D
Effect on Shear Strength of Reinforced Concrete Beams Information, No. 132, pp. 181-187.
Without Stirrups: II. Verification and Calibration, Journal Collins, M. P., and Lampert, P., 1973, Redistribution
of Structural Engineering, V. 131, No. 12, pp. 1886-1897. of Moments at CrackingThe Key to Simpler Torsion
Belarbi, A., and Greene, G. G., 2004, Reinforced Design? Analysis of Structural Systems for Torsion, SP-35,
Concrete Box Girders Under Cyclic Torsion, Proceedings American Concrete Institute, Farmington Hills, MI, pp.
of the 13th World Conference on Earthquake Engineering, 343-383.
Vancouver, BC, Canada, Aug. 1-6, 2004, 15 pp. Collins, M. P., and Mitchell, D., 1978, Torsional-Balanced
Belarbi, A., and Hsu, T. T. C., 1994, Constitutive Laws Steel in Concrete Beams Discussion by Michael P. Collins
of Concrete in Tension and Reinforcing Bars Stiffened by and Denis Mitchell, Journal of the Structural Division, V.
Concrete, ACI Structural Journal, V. 91, No. 4, July-Aug., 114, pp. 1684-1687.
pp. 465-474. Collins, M. P., and Mitchell, D., 1980, Shear and Torsion
Belarbi, A., and Hsu, T. T. C., 1995, Constitutive Laws Design of Prestressed and Non-Prestressed Concrete
of Softened Concrete in Biaxial Tension-Compression, ACI Beams, PCI Journal, V. 25, No. 5, pp. 32-100.
Structural Journal, V. 92, No. 5, Sept.-Oct., pp. 562-573. Collins, M. P., and Mitchell, D., 1991, Prestressed
Bentz, E. C., 2005, Empirical Modeling of Reinforced Concrete Structures, Prentice Hall, Englewood Cliffs, NJ,
Concrete Shear Strength Size Effect for Members without 766 pp.
Stirrups, ACI Structural Journal, V. 102, No. 2, Mar.-Apr., Collins, M. P.; Walsh, P. F.; Archer, F. E.; and Hall, A. S.,
pp. 232-241. 1968a, Ultimate Strength of Reinforced Concrete Beams
Birkland, C. J., 1965, Strength of Rectangular Beams with Subjected to Combined Torsion and Bending, Torsion of
Longitudinal Reinforcement in Combined Torsion, Bending, Structural Concrete, SP-18, American Concrete Institute,
and Shear, Masters thesis, Department of Civil Engineering, Farmington Hills, MI, pp. 279-402.
University of Washington at Seattle, Seattle, WA. Collins, M. P.; Walsh, P. F.; Archer, F. E.; and Hall, A. S.,
Bishara, A., and Londot, L., 1979, Flexural Rotational 1968b, Reinforced Concrete in Torsion, UNICIV Report No.
Capacity of Spandrel Beams, Journal of the Structural R-31, University of New South Wales, NSW, Australia, 338 pp.
Division, V. 105, No. 1, pp. 147-161. Delhumeau, G., 1999, LInvention du Beton arme:
Bishara, A., and Peir, J. C., 1968, Reinforced Concrete Hennebique, 1890-1914, Horma, Institut Francais
Rectangular Columns in Torsion, Journal of the Structural dArchitecture, Paris, France, 344 pp.
Division, V. 94, Dec, pp. 2913-2934. Duleau, A., 1820, Essai Theorique et Experimental sur la
Boulton, N., and Boonsukha, B., 1959, Plastic Collapse Resistance du Fer Forge, Paris, France (in French).
Loads for Circular Arc Girders, Proceedings of the Institu- Elfgren, L., 1972a, Reinforced Concrete Beam Loaded
tion of Civil Engineers, June, London, UK. in Torsion, Bending and Shear, Publication 71:3, Division
Bredt, R., 1896, Kritische Bemerkungen zur Drehungs- of Concrete Structures, Chalmers University of Technology,
elastizitt, Zeitschrift des Vereines Deutscher Ingenieure, Gteborg, Sweden, 249 pp.
Band 40, No. 28, July 11, pp. 785-790; No. 29, July 18, pp. Elfgren, L., 1972b, Reinforced Concrete Beams Loaded
813-817. (in German) in Combined Torsion, Bending and Shear. A Study of the
Cauchy, A., 1828, Sur les Equations qui Experimen- Ultimate Load-Carrying Capacity, PhD thesis, Publication
tent les Conditions dEquilibre ou les Lois de Mouvement 71:1, Division of Concrete Structures, Chalmers University
Interieur dun Corps Solide, Exercices de Mathematique, of Technology, second edition, Gteborg, Sweden, 230 pp.
Paris, France. (in French) Elfgren, L., 1979, Torsion-Bending-Shear in Concrete
CEB-FIP, 1978, Model Code for Concrete Structures, Beams: A Kinematic Model, Colloquium Copenhagen
CEB-FIP International Recommendation, third edition, 1979: Plasticity in Reinforced Concrete, IABSE Reports,
Comite Euro-International du Beton (CEB), 348 pp. Zrich, V. 29, pp. 111-118.
Cerioni, R.; Ferreti, D.; and Iori, I., 1998, Torsional Elfgren, L.; Karlsson, I.; and Losberg, A., 1974a, Torsion-
Behavior of Prestressed High-Strength Concrete Beams, Bending-Shear Interaction for Concrete Beams, Journal of
Computational Modeling of Concrete Structures, pp. the Structural Division, V. 100, No. 8, pp. 1657-1676.
797-806. Elfgren, L.; Karlsson, I.; and Losberg, A., 1974b, Nodal
Chakraborty, M., 1977, Torsional-Balanced Steel in Forces in the Analysis of the Ultimate Torsional Moment for
Concrete Beams, Journal of the Structural Division, V. 103, Rectangular Beams, Magazine of Concrete Research, V. 26,
pp. 2181-2191. No. 86, pp. 21-28; discussion in V. 27, No. 90, pp. 42-45.
Chen, W., 1982, Plasticity in Reinforced Concrete, Ersoy, U., and Ferguson, P. M., 1968, Concrete Beams
McGraw-Hill Book, New York, 474 pp. Subjected to Combined Torsion and ShearExperimental
Collins, M. P., 1973, Torque-Twist Characteristics of Trends, Torsion of Structural Concrete, SP-18, American
Reinforced Concrete Beams, Inelasticity and Non-Linearity Concrete Institute, Farmington Hills, MI, pp. 441-460.
Ewida, A. A., and McMullen, A. E., 1981, Torsion-Shear tional Association for Bridges and Structural Engineering,
Interaction in Reinforced Concrete Members, Magazine of Zurich, Switzerland.
Concrete Research, V. 23, No. 115, pp. 113-122. Horne, M. R., 1949, Fundamental Propositions in the
Gabrielsson, H., 1999, Ductility in High Performance Plastic Theory of Structures, Journal of the Institute of Civil
Concrete Structures: An Experimental Investigation and Engineers (London), V. 34, pp. 174-177.
a Theoretical Study of Prestressed Hollow Core Slabs and Hsu, T. T. C., 1968a, Torsion of Structural Concrete
Prestressed Cylindrical Pole Elements, PhD thesis 1999:15, Plain Concrete Rectangular Sections, Torsion of Structural
Division of Structural Engineering, Lule University of Concrete, SP-18, American Concrete Institute, Farmington
Technology, Lule, 283 pp. Hills, MI, pp. 203-238.
Ghoneim, M. G., and MacGregor, J. G., 1993, Evalu- Hsu, T. T. C., 1968b, Torsion of Structural Concrete
ation of Design Procedure for Torsion in Reinforced and Behavior of Reinforced Concrete Rectangular Members,
Prestressed Concrete, Report No. 184, Department of Civil Torsion of Structural Concrete, SP-18, American Concrete
Engineering, University of Alberta, Edmonton, AB, Canada, Institute, Farmington Hills, MI, Jan., pp. 261-306.
301 pp. Hsu, T. T. C., 1968c, Ultimate Torque of Reinforced
Greene, G., and Belarbi, A., 2009a, Model for Reinforced Rectangular Beams, Journal of the Structural Division,
Concrete Members under Torsion, Bending, and Shear. I: V. 94, pp. 485-510.
Theory, Journal of Engineering Mechanics, V. 135, No. 9, Hsu, T. T. C., 1968d, Torsion of Structural Concrete
pp. 961-969. Interaction Surface for Combined Torsion, Shear and
Greene, G., and Belarbi, A., 2009b, Model for Reinforced Bending in Beams without Stirrups, ACI Journal, V. 65, No.
Concrete Members under Torsion, Bending, and Shear. II: 1, Jan., pp. 51-60.
Model Application and Validation, Journal of Engineering Hsu, T. T. C., 1968e, Torsion of Structural Concrete
Mechanics, V. 135, No. 9, pp. 970-977. Uniformly Prestressed Rectangular Members without Web
Greene, G. G., 2006, Behavior of Reinforced Concrete Reinforcement, PCI Journal, V. 13, No. 2, pp. 34-44.
Girders under Cyclic Torsion and Torsion Combined with Hsu, T. T. C., 1984, Torsion of Reinforced Concrete, Van
Shear: Experimental Investigation and Analytical Models, Nostrand Reinhold, New York, pp. 516-523.
PhD thesis, Department of Civil, Architectural, and Environ- Hsu, T. T. C., 1988, Softening Truss Model Theory for
mental Engineering, University of Missouri at Rolla, Rolla, Shear and Torsion, ACI Structural Journal, V. 85, No. 6,
MO, 227 pp. Nov.-Dec., pp. 624-635.
Greene, G. G., and Belarbi, A., 2006a, Softened Truss Hsu, T. T. C., 1990, Shear Flow Zone in Torsion of Rein-
Model for RC Torsional Members Under Combined Action, forced Concrete, Journal of Structural Engineering, V. 116,
SEI Structures Congress, St. Louis, MO, 8 pp. No. 11, pp. 3206-3226.
Greene, G. G., and Belarbi, A., 2006b, Tension-Stiffened Hsu, T. T. C., 1993, Unified Theory of Reinforced Concrete,
Softened Truss Model for RC Members under Torsion, CRC Press, Boca Raton, FL, 313 pp.
Proceedings of Second FIB Congress, Naples, Italy, June Hsu, T. T. C., 1997, ACI Shear and Torsion Provisions for
5-8, 12 pp. Prestressed Hollow Girders, ACI Structural Journal, V. 94,
Greene, G. G., and Belarbi, A., 2008, Performance of Box No. 6, Nov.-Dec., pp. 787-799.
Bridge Girders under Cyclic Torsion and Shear, Proceed- Hsu, T. T. C.; Belarbi, A.; and Pang, X. B., 1995a, A
ings of PCI National Bridge Conference, Orlando, FL, Paper Universal Panel Tester, Journal of Testing and Evaluation,
No. 73, Oct. 4-7, 25 pp. V. 23, No. 1, pp. 41-49.
Gvozdev, A. A., 1938, Determination of the Value of Hsu, T. T. C., and Burton, K., 1974, Design of Reinforced
Failure Load for Statically Indeterminate Systems Subject to Concrete Spandrel Beams, Journal of the Structural Divi-
Plastic Deformations (in Russian), Conference on Plastic sion, V. 100, No. 1, pp. 209-229.
Deformations 1936, B. G. Galerkin, ed., Akademia Nauk Hsu, T. T. C., and Hwang, C. S., 1977, Torsional Limit
SSSR, Moscow and Leningrad, pp. 19-38. Design of Spandrel Beams, ACI Journal, V. 74, No. 2, Feb.,
Gvozdev, A. A., 1960, The Determination of the Value pp. 71-79.
of the Collapse Load for Statically Indeterminate Systems Hsu, T. T. C.; Mansour, M. Y.; Mo, Y. L.; and Zhong, J.,
Subject to Plastic Deformations, International Journal of 2006, Cyclic Softened Membrane Model for Nonlinear
Mechanical Sciences, V. 1, No. 1, pp. 322-335. Finite Element Analysis of Concrete Structures, Finite
Hamilton, M. E., 1966, Strength of T- and L- Section Element Analysis of Reinforced Concrete Structures,
Beams with Longitudinal Reinforcement in Combined SP-237, American Concrete Institute, Farmington Hills, MI,
Torsion, Bending, and Shear, Masters thesis, Department pp. 71-98.
of Civil Engineering, University of Washington at Seattle, Hsu, T. T. C., and Mo, Y. L., 1983, Softening of Concrete
Seattle, WA. in Torsional Members, Research Report No. ST-TH-001-83,
Hassan, T.; Lucier, G.; Rizkalla, S.; and Zia, P., 2007, Department of Civil Engineering, University of Houston,
Modeling of L-Shaped Precast Prestressed Concrete Span- Houston, TX, Mar., 107 pp.
drels, PCI Journal, V. 52, No. 2, pp. 78-92. Hsu, T. T. C., and Mo, Y. L., 1985a, Softening of Concrete
Hillerborg, A., 1960, A Plastic Theory for the Design of in Torsional MembersTheory and Tests, ACI Journal,
Reinforced Concrete Slabs, Sixth ABSE Congress Interna- V. 82, No. 3, May-June, pp. 290-303.
Hsu, T. T. C., and Mo, Y. L., 1985b, Softening of Concrete Structures, Chalmers University of Technology, Gteborg,
in Torsional MembersPrestressed Concrete, ACI Journal, Sweden, 203 pp.
V. 82, No. 5, Sept.-Oct., pp. 603-615. Kemp, E. L.; Sozen, M. A.; and Siess, C. P., 1961, Torsion
Hsu, T. T. C., and Mo, Y. L., 1985c, Softening of Concrete of Reinforced Concrete, Structural Research Series No.
in Torsional MembersDesign Recommendations, ACI 226, University of Illinois, Urbana, IL.
Journal, V. 82, No. 4, July-Aug., pp. 443-452. Klein, G. J., 1986, Design of Spandrel Beams, PCI
Hsu, T. T. C., and Mo, Y. L., 2010, Unified Theory of Specially Funded R & D Program Research Project No. 5
Concrete Structures, John Wiley and Sons, Ltd, London, (PCISFRAD #5), Prestressed Concrete Institute, Chicago,
UK, pp. 60-69. IL; summary paper in PCI Journal, V. 31, No. 5, Sept.-Oct.,
Hsu, T. T. C., and Zhang, L. X., 1997, Nonlinear Anal- pp. 76-124.
ysis of Membrane Elements by Fixed- Angle Softened-Truss Klus, J., 1968, Ultimate Strength of Reinforced Concrete
Model, ACI Structural Journal, V. 94, No. 5, Sept.-Oct., pp. Beams in Combined Torsion and Shear, ACI Journal, V. 65,
483-492. No. 3, Mar., pp. 210-216.
Hsu, T. T. C.; Zhang, L. X.; and Gomez, T., 1995b, A Kordina, K., and Teutsch, M., 1983, Bemessung von
Servo-Control System for Universal Panel Tester, Journal Stahlbeton- und Spannbetonbalken unter kombinierter
of Testing and Evaluation, V. 23, No. 6, pp. 424-430. Beanspruchung infolge von Biegung, Querkraft und
Hsu, T. T. C., and Zhu, R. R. H., 2002, Softened Torsion, Beton- und Stahlbetonbau, V. 78, Heft 1, pp. 1-6.
Membrane Model for Reinforced Concrete Elements in (in German)
Shear, ACI Structural Journal, V. 99, No. 4, July-Aug., pp. Kordina, K., and Teutsch, M., 1985, Versuche an
460-469. Konstruktionsleichtbetonbauteilen unter kombinierter
Humphreys, R., 1957, Torsional Properties of Prestressed Beanspruchung aus Torsion, Biegung und Querkraft,
Concrete, The Structural Engineer, V. 35, No. 6, pp. Deutscher Ausschu fr Stahlbeton (DafStb), H. 362, W.
213-224. Ernst u. Sohn, Berlin, 83 pp.
Hwang, C. S., 1978, Warping Torsion of Reinforced Koutchoukali, N. E., and Belarbi, A., 2001, Torsion of
Concrete Beams with Thin Walled Open Sections, PhD High-Strength Reinforced Concrete Beams and Minimum
dissertation, Department of Civil Engineering, University of Reinforcement Requirement, ACI Structural Journal, V. 98,
Miami, Coral Gables, FL, 286 pp. No. 4, July-Aug., pp. 462-469.
Hwang, C. S., and Hsu, T. T. C., 1983, Mixed Torsion Krpan, P., and Collins, M. P., 1981a, Predicting Torsional
Analysis of Reinforced Concrete Channel Beams-A Fourier Response of Thin-Walled Open RC Members, Journal of
Series Approach, ACI Journal, V. 80, No. 5, Sept.-Oct., pp. the Structural Division, V. 107, pp. 1107-1127.
377-386. Krpan, P., and Collins, M. P., 1981b, Testing Thin-Walled
Iravani, S., 1996, Mechanical Properties of High-Perfor- Open RC Structure in Torsion, Journal of the Structural
mance Concrete, ACI Materials Journal, V. 93, No. 5, Division, V. 107, pp. 1129-1140.
Sept.-Oct., pp. 416-426. Lampert, P., and Collins, M. P., 1972, Torsion, Bending,
Johnston, D. W., 1971, Hollow Prestressed Concrete and ConfusionAn Attempt to Establish the Facts, ACI
Beams Under Combined Torsion, Bending and Shear, PhD Journal, V. 69, No. 8, Aug., pp. 500-504.
thesis, Department of Civil Engineering, North Carolina Lampert, P., and Thrlimann, B., 1968, Torsionsversuch
State University, Raleigh, NC, 177 pp. an Stahlbetonbalken (Torsion Tests of Reinforced Concrete
Johnston, D. W., and Zia, P., 1975, Prestressed Box Beams), Bericht Nr. 6506-2, 101 pp. (in German)
Beams Under Combined Loading, Journal of the Structural Lampert, P., and Thrlimann, B., 1969, Torsion-Biege-
Division, V. 101, No. 7, pp. 1313-1331. Versuche an Stahlbetonbalken (Torsion-Bending Tests on
Jordaan, I.; Khalifa, M.; and McMullen, A. E., 1974, Reinforced Concrete Beams), Bericht Nr. 6506-3, Institute
Collapse of Curved Reinforced Concrete Beams, Journal of Baustatik, ETH, Zurich, Switzerland (in German).
of the Structural Division, V. 100, No. 11, pp. 2255-2269. Lampert, P., and Thrlimann, B., 1971, Ultimate Strength
Karayannis, C. G., 1995, Torsional Analysis of Flanged and Design of Reinforced Concrete Beams In Torsion and
Concrete Elements with Tension Softening, Computers & Bending, Publications, International Association of Bridge
Structures, V. 54, No. 1, pp. 97-110. and Structural Engineering (Zurich), V. 31-I, pp. 107-131.
Karayannis, C. G., 2000, Smeared Crack Analysis for Leonhardt, F., and Schelling, G., 1974, Torsionsversuche
Plain Concrete in Torsion, Journal of Structural Engi- an Stahlbetonbalken, Deutscher Ausschu fr Stahlbeton
neering, V. 126, No. 6, pp. 638-645. (DAfStb), Heft 239, W. Ernst u. Sohn, Berlin, 122 pp.
Karayannis, C. G., and Chalioris, C. E., 2000, Experi- Leonhardt, F.; Walther, R.; and Vogler, O., 1968, Torsions-
mental Validation of Smeared Analysis for Plain Concrete in und Schubversuche an vorgespannten Hohlkastentrgern,
Torsion, Journal of Structural Engineering, V. 126, No. 6, Deutscher Ausschu fr Stahlbeton (DAfStb), Heft 202,
pp. 646-653. Wilhelm Ernst & Sohn, Berlin, pp. 1-78.
Karlsson, I., 1973, Stiffness Properties of Reinforced Lessig, N. N., 1959, Determination of Load-Carrying
Concrete Beams Loaded in Combined Torsion, Bending and Capacity of Rectangular Reinforced Concrete Elements
Shear, PhD thesis, Publication 73:1, Division of Concrete Subjected to Flexure and Torsion, Trudy No. 5, Institut
Betona i Zhelezobetona (Concrete and Reinforced Concrete
Institute), Moscow, pp. 5-28 (in Russian). Translated by Mattock, A. H., 1979, Flexural Strength of Prestressed
Portland Cement Association, Foreign Literature Study No. Concrete Sections by Programmable Calculator, PCI
371 (Available from S. L. A. Translation Center, John Crerar Journal, V. 24, No. 1, pp. 32-54.
Library Translation Center, Chicago, IL). McGee, W. D., and Zia, P., 1973, Prestressed Concrete
Leu, L. J., and Lee, Y. S., 2000, Torsion Design Charts Members Under Torsion, Shear, and Bending, Research
for Reinforced Concrete Rectangular Members, Journal of Report, Department of Civil Engineering, North Carolina
Structural Engineering, V. 126, No. 2, pp. 210-218. State University, Raleigh, NC, 212 pp.
Lim, S. N., and Mirza, M. S., 1968, discussion of Rein- McMullen, A. E., and Daniel, H. R., 1975, Torsional
forced Concrete T-Beams without Stirrups under Combined Strength of Longitudinally Reinforced Concrete Beams
Moment and Torsion, by D. J. Victor and P. M. Fergusson, Containing an Openings, ACI Journal, V. 72, No. 8, Aug.,
ACI Journal, V. 65, No. 7, July, pp. 560-563. pp. 415-420.
Logan, D. R., 2007, L-Spandrels: Can Torsional Distress McMullen, A. E., and El-Degwy, W. M., 1985, Prestressed
Be Induced by Eccentric Vertical Loading? PCI Journal, V. Concrete Tests Compared with Torsion Theories, PCI
52, No. 2, Mar.-Apr., pp. 46-61. Journal, V. 30, No. 5, pp. 96-127. (discussion by Hsu and
Lubliner, J., 1990, Plasticity Theory, Macmillan Publishing Mo, V. 31, No. 5, Oct., pp. 160-164).
Co., New York, 495 pp. McMullen, A. E., and Warwaruk, J., 1967, The Torsional
Lchinger, P., 1977, Bruchwiderstand von Kastentrgern Strength of Rectangular Reinforced Concrete Beams
aus Stahlbeton unter Torsion, Biegung und Querkraft (Ulti- Subjected to Combined Loading, Report No. 2, Department
mate Strength of Box-Girder Beams in Reinforced Concrete of Civil Engineering, University of Alberta, Edmonton, AB,
loaded in Torsion, Bending and Shear), Institut fr Baustatik Canada, 162 pp.
und Konstruktion, ETH Zrich, Bericht Nr 69, Birkhuser McMullen, A. E., and Woodhead, H. R., 1973, Experi-
Verlag, Basel und Stuttgart, 107 pp. mental Study of Prestressed Concrete Under Combined
Lucier, G.; Rizkalla, S.; Zia, P.; and Klein, G., 2007, Torsion, Bending, and Shear, PCI Journal, V. 18, No. 5,
Precast Concrete, L-Shaped Spandrels Revisited: Full- pp. 85-100.
Scale Tests, PCI Journal, V. 52, No. 2, pp. 62-76. Melan, E., 1938, Der Spannungszustand eines Miese-
Lucier, G.; Walter, C.; Rizkalla, S.; and Zia, P., 2010, schen-Henckyschen Kontinuums bei vernderlicher Belas-
Development of a Rational Design Methodology for tung. Sitzungsberichte, Akademie der Wissenschaften in
Precast Slender Spandrel Beams, Research Report No. Wien, Abt. II/a, V. 147, pp. 73-87.
IS-09-10, Construction Facilities Laboratory, Department of Mirza, M. S., and McCutcheon, J. O., 1968, discussion
Civil, Construction, and Environmental Engineering, North of Torsion of Structural ConcreteInteraction Surface for
Carolina State University, Raleigh, NC. Combined Torsion, Shear and Bending in Beams without
MacGregor, J. G., and Ghoneim, M. G., 1995, Design for Stirrups,, ACI Journal, V. 65, No. 7, July, pp. 567-570.
Torsion, ACI Structural Journal, V. 92, No. 2, Mar.-Apr., Mitchell, D., and Collins, M. P., 1974, Diagonal
pp. 211-218. Compression Field TheoryA Rational Model for Struc-
Mansour, M., and Hsu, T. T. C., 2005a, Behavior of tural Concrete in Pure Torsion, ACI Journal, V. 71, No. 8,
Reinforced Concrete Elements under Cyclic Shear: Part I Aug., pp. 396-408.
Experiments, Journal of Structural Engineering, V. 131, Mitchell, D., and Collins, M. P., 1976, Detailing for
No. 1, pp. 44-53. Torsion, ACI Journal, V. 73, No. 9, Sept., pp. 506-511.
Mansour, M., and Hsu, T. T. C., 2005b, Behavior of Mitchell, D., and Collins, M. P., 1978, Influence of Pre-
Reinforced Concrete Elements under Cyclic Shear: Part II stressing on Torsional Response of Concrete Beams, PCI
Theoretical Model, Journal of Structural Engineering, V. Journal, V. 23, No. 3, pp. 54-73.
131, No. 1, pp. 54-65. Mrsch, E., 1902, Der Eisenbetonbau, seine Anwendung
Mansour, M.; Lee, J. Y.; and Hsu, T. T. C., 2001, Cyclic und Theorie, Wayss and Freytag, A. G., Im Selbstverlag der
Stress-Strain Curves of Concrete and Steel Bars in Membrane Firma, Neustadt, a. d.; Haardt, 118 pp. (English translation by
Elements, Journal of Structural Engineering, V. 127, No. E. P. Goodrich, McGraw-Hill Book Co., New York, 368 pp.).
12, pp. 1402-1411. Mller, P., 1976, Failure Mechanisms for Reinforced
Mardukhi, J., 1974, The Behaviour of Uniformly Concrete Beams in Torsion and Bending, Publications,
Prestressed Concrete Box Beams in Combined Torsion and International Association for Bridge and Structural Engi-
Bending, Masters thesis, University of Toronto, Toronto, neering (IABSE), V. 36-II, Zrich, pp. 147-163.
ON, Canada, 73 pp. Mller, P., 1978, Platische Berechnung von Stahlbeton-
Marti, P., and Kong, K., 1987, Response of Reinforced scheiben und Balken (Plastic Analysis of Walls and Beams of
Concrete Slab Elements to Torsion, Journal of Structural Reinforced Concrete), Institut fr Baustatik und Konstruk-
Engineering, V. 113, No. 5, pp. 976-993. tion, ETH Zrich, Bericht Nr 83, Birkhauser Verlag, Basel
Marti, P.; Leesti, P.; and Khalifa, W., 1987, Torsion Tests und Stuttgart, 160 pp.
on Reinforced Concrete Slab Elements, Journal of Struc- National Association of Cement Users, 1910, Standard
tural Engineering, V. 113, No. 5, pp. 994-1010. Building Regulations for the Use of Reinforced Concrete,
NACU Standard No. 4, American Concrete Institute, Farm-
ington Hills, MI, pp. 349-361.
Navier, C. L., 1826, Rsume des Leons Donnes a Rahal, K. N., 2000b, Torsional Strength of Reinforced
lEcole des Ponts et Chausses sur lApplication de la Mca- Concrete Beams, Canadian Journal of Civil Engineering,
nique a lEtablissement des Constructions et des Machines, V. 27, No. 3, pp. 445-453.
Premire partie, Continuant les Leons sur la Rsistance Rahal, K. N., and Collins, M. P., 1995a, Analysis of
des Matriaux et sur lEtablissement des Constructions en Sections Subjected to Combined Shear and TorsionA
Terre, en Maonnerie et en Charpente, Firmin Didet, Paris Theoretical Investigation, ACI Structural Journal, V. 92,
(in French). No. 4, July-Aug., pp. 459-469.
Nielsen, M. P., 1967, Om Forskydningsarmering i Jern- Rahal, K. N., and Collins, M. P., 1995b, Effect of the
betonbjaelker (On Shear Reinforcement in Reinforced Thickness of Concrete Cover on the Shear-Torsion Inter-
Concrete Beams), Bygningsstatiske Meddelelser, V. 38, No. action An Experimental Investigation, ACI Structural
2, pp. 33-58. Journal, V. 92, No. 3, May-June, pp. 334-342.
Nylander, H., 1945, Torsion and Torsional Restraint Rahal, K. N., and Collins, M. P., 1996, Simple Model for
of Concrete Structures (Vrindning och Vridningsinspan- Predicting Torsional Strength of Reinforced and Prestressed
ning vid Betong-Konstruktioner), Bulletin No. 3, Statens Concrete Sections, ACI Structural Journal, V. 93, No. 6,
Kommittee for Byggnadsforskning, Stockholm, Sweden. Nov.-Dec., pp. 658-666.
Onsongo, W. M., 1978, The Diagonal Compression Rahal, K. N., and Collins, M. P., 2003a, Combined
Field Theory for Reinforced Concrete Beams Subjected to Torsion and Bending in Reinforced and Prestressed Concrete
Combined Torsion, Flexure, and Axial Load, PhD thesis, Beams, ACI Structural Journal, V. 100, No. 2, Mar.-Apr.,
Department of Civil Engineering, University of Toronto, pp. 157-165.
Toronto, ON, Canada, 246 pp. Rahal, K. N., and Collins, M. P., 2003b, An Experimental
Palermo, D., and Vecchio, F. J., 2003, Compression Field Evaluation of the ACI and AASHTO-LRFD Design Provi-
Modeling of Reinforced Concrete Subjected to Reversed sions for Combined Shear and Torsion, ACI Structural
Loading: Formulation, ACI Structural Journal, V. 100, No. 5, Journal, V. 100, No. 3, May-June, pp. 277-282.
Sept.-Oct., pp. 616-625. Rangan, B. V.; Natarajan, P. R.; and Parameswaran, V.
Palermo, D., and Vecchio, F. J., 2004, Compression Field S., 1978, Lower Bound Limit Design of Concrete Struc-
Modeling of Reinforced Concrete Subjected to Reversed tures, Journal of the Structural Division, V. 104, No. 12,
Loading: Verification, ACI Structural Journal, V. 101, No. pp. 1901-1906.
2, Mar.-Apr., pp. 155-164. Rasmussen, L. J., and Baker, G., 1995, Torsion in Rein-
Pang, X. B., and Hsu, T. T. C., 1995, Behavior of Rein- forced Normal and High-Strength Concrete BeamsPart 1:
forced Concrete Membrane Elements in Shear, ACI Struc- Experimental Test Series, ACI Structural Journal, V. 92,
tural Journal, V. 92, No. 6, Nov.-Dec., pp. 665-679. No. 1, Jan.-Feb., pp. 55-63.
Pantazopoulou, S., and Moehle, S., 1990, Truss Model Raths, C. H., 1984, Spandrel Beam Behavior and
for 3-D Behavior of R.C. Exterior Connections, Journal of Design, PCI Journal, V. 29, No. 2, pp. 62-131.
Structural Engineering, V. 116, No. 2, pp. 298-315. Rausch, E., 1929, Design of Reinforced Concrete in
PCI, 1978, PCI Design Handbook, second edition, Torsion and Shear, (Berechnung des Eisenbetons gegen
Prestressed Concrete Institute, Chicago, IL. Verdrehung [Torsion] und Abscheren), Technische Hoch-
PCI, 1999, PCI Design Handbook, fifth edition, Prestressed schule, Berlin, Germany, 53 pp. (in German); second edition
Concrete Institute, Chicago, IL, 540 pp. in 1938; third edition in 1953, Drillung (Torsion) Schub
Petersson, T., 1972, Shear and Torsion in Reinforced und Scheren in Stahlbelonbau, Deutcher Ingenieur-Verlag
Concrete Beams. A Study of Critical Load and Torsional GmbH, Dusseldorf, Germany, 168 pp.
Stiffness, Document D10.1972, National Swedish Building RILEM Committee QFS, 2004, Quasibrittle Fracture
Research, Stockholm, Sweden, 276 pp. Scaling and Size Effect, Materials and Structures, V. 37,
Rabbat, B. G., and Collins, M. P., 1977, The Computer No. 272, pp. 547-586.
Aided Design of Structural Concrete Sections Subjected to Ritter, W., 1899, Die Bauweise Hennebique, Schweizeri-
Combined Loading, Computers & Structures, V. 7, No. 2, sche Bauzeitung, V. 33, No. 7, pp. 59-61.
pp. 229-236. Robinson, J. R., and Demorieux, J. M., 1972, Essais de
Rabbat, B. G., and Collins, M. P., 1978, Variable Angle Traction-Compression sur Modles dAme de Poutre en
Space Truss Model for Structural Concrete Members Bton Arme, IRABA Report, Institut de Recherches Appli-
Subjected to Complex Loading, Douglas McHenry Sympo- ques du Bton de LAme, Part 1, June 1968, 44 pp; Resis-
sium on Concrete and Concrete Structures, SP-55, American tance Ultime du Bton de Lme de Poutres en Double Te en
Concrete Institute, Farmington Hills, MI, pp. 547-587. Bton Arme, Part 2, May, 53 pp.
Rahal, K. N., 2000a, Shear Strength of Reinforced Sadowsky, M., 1949, An Extension of the Sand Heap
Concrete, Part I: Membrane Elements Subjected to Pure Analogy in Plastic Torsion Application to Cross-Section
Shear, ACI Structural Journal, V. 97, No. 1, Jan.-Feb., pp. Having One or More Holes, Journal of Applied Mechanics,
86-93 (closure to discussion, V. 97, No. 6, Nov.-Dec., pp. V. 62.
910-913.) Saint-Venant, B., 1856, Mmoire sur la torsion des
prismes (lu lAcadmie le 13 juin 1853). Mmoires Des
Savants Etrangers, Mmoires prsentes par divers savants a
lAcadmie des Sciences, de linstitut Imperial de France et tions, The Institution of Engineers, Australia, V. CE9, No.
imprime par son ordre, V. 14, Imprimerie Impriale, Paris, 2, pp. 313-320.
pp.233-560. Walsh, P. F.; Collins, M. P.; Archer, F. E.; and Hall, A. S.,
Stevens, N. J.; Uzumeri, S. M.; and Collins, M. P., 1991, 1966, The Ultimate Strength Design of Rectangular Rein-
Reinforced Concrete Subjected to Reversed Cyclic Shear forced Concrete Beams Subjected to Combined Torsion,
Experiments and Constitutive Model, ACI Structural Bending and Shear, Civil Engineering Transactions, The
Journal, V. 88, No. 2, Mar.-Apr., pp. 135-146. Institution of Engineers, Australia, V. CE8, No. 2, pp.
Teutsch, M., 1980, Trag- und Verformungsverhalten von 143-157.
Stahlbeton- und Spannbetonbalken mit rechteckigem Quer- Wang, W., 1995, Shear Strength and Fatigue Crack Prop-
schnitt unter kombinierter Beanspruchung aus Biegung, agation in Concrete by Energy Methods, PhD dissertation,
Querkraft und Torsion, dissertation, TU Carolo Wilhelmina Department of Civil and Environmental Engineering, New
Braunschweig, Fakultt fr Bauwesen, 181 pp. Jersey Institute of Technology, Newark, NJ.
Thrlimann, B., 1978, Plastic Analysis of Reinforced Wang, W., and Hsu, T. T. C., 1997, Limit Analysis of
Concrete Beams, Introductory Report, IABSE Colloquium, Reinforced Concrete Beams Subjected to Pure Torsion,
Reports of the Working Commissions, International Asso- Journal of Structural Engineering, V. 123, No. 1, pp. 86-94.
ciation for Bridges and Structural Engineering (IABSE), Wang, W.; Jiang, D.; and Hsu, T. T. C., 1993, Shear
Zrich, Switzerland, V. 28, Oct., pp. 71-90. Strength of Reinforced Concrete Deep Beams, Journal of
Thrlimann, B.; Marti, P.; Pralong, J.; Ritz, P.; and Structural Engineering, V. 119, No. 8, pp. 2294-2312.
Zimmerli, B., 1983, Anwendung der Plastizitatstheorie auf Yoo, C., and Heins, C., 1972, Plastic Collapse of Hori-
Stahlbeton (Plasticity in Concrete Structures), Vorlesung, zontally Curved Girders, Journal of the Structural Division,
ETH-Zurich. V. 98, No. 4, pp. 899-913.
Vecchio, F. J., 1999, Towards Cyclic Load Modeling of Yudin, V. K., 1962, Determination of the Load-Carrying
Reinforced Concrete, ACI Structural Journal, V. 96, No. 2, Capacity of Rectangular Reinforced Concrete Elements
Mar.-Apr., pp. 193-201. Subjected to Combined Torsion and Bending, Beton i
Vecchio, F. J., and Collins, M. P., 1981, Stress-Strain Zhelezobeton (Concrete and Reinforced Concrete), Moscow,
Characteristic of Reinforced Concrete in Pure Shear, IABSE No. 6, pp. 265-269.
Colloquium, Advanced Mechanics of Reinforced Concrete, Zararis, P. D., and Penelis, G. G., 1986, Reinforced
Delft, Final Report, International Association of Bridge and Concrete T-Beams in Torsion and Bending, ACI Journal,
Structural Engineering, Zurich, Switzerland, pp. 221-225. V. 83, No. 1, Jan.-Feb., pp. 145-155.
Vecchio, F. J., and Collins, M. P., 1982, The Response of Zbirohowski-Koscia, K. F., 1968, Stress Analysis of
Reinforced Concrete to In-Plane Shear and Normal Stresses, Cracked Reinforced Concrete Thin Walled Beams and
Publication No. 82-03, Department of Civil Engineering, Shells, Magazine of Concrete Research, V. 20, No. 65, pp.
University of Toronto, Toronto, ON, Canada, pp. 332. 213-220.
Vecchio, F. J., and Collins, M. P., 1986, The Modi- Zhang, L. X., and Hsu, T. T. C., 1998, Behavior and
fied Compression Field Theory for Reinforced Concrete Analysis of 100 MPa Concrete Membrane Elements,
Elements Subjected to Shear, ACI Journal, V. 83, No. 2, Journal of Structural Engineering, V. 124, No. 1, pp. 24-34.
Mar.-Apr., pp. 219-231. Zhu, R. R. H., and Hsu, T. T. C., 2002, Poisson Effect of
Vecchio, F. J., and Collins, M. P., 1993, Compression Reinforced Concrete Membrane Elements, ACI Structural
Response of Cracked Reinforced Concrete, Journal of Journal, V. 99, No. 5, Sept.-Oct., pp. 631-640.
Structural Engineering, V. 119, No. 12, pp. 3590-3610. Zhu, R. R. H.; Hsu, T. T. C.; and Lee, J. Y., 2001, Rational
Victor, D. J., and Ferguson, P. M., 1968, Reinforced Shear Modulus for Smeared-Crack Analysis of Reinforced
Concrete T-Beams without Stirrups under Combined Concrete, ACI Structural Journal, V. 98, No. 4, July-Aug.,
Moment and Torsion, ACI Journal, V. 65, No. 1, Jan., pp. pp. 443-451.
29-36. Zia, P., 1961, Torsional Strength of Prestressed Concrete
Vlasov, V. Z., 1961, Thin Walled Elastic Beams; second Members, ACI Journal, V. 57, No. 10, pp. 1337-1359.
revised and augmented edition (1985) (in Russian); English Zia, P., 1970, What Do We Know About Torsion in
translation, National Technical Information Service, Spring- Concrete Members? Journal of the Structural Division,
field, VA, 493 pp. V. 96, No. 6, pp. 1185-1199.
Wafa, F. F.; Shihata, S. A.; Akhtaruzzaman, A. A.; and Zia, P., and Hsu, T. T. C., 1978, 2004, Design for Torsion
Ashour, S. A., 1995, Prestressed High-Strength Concrete and Shear in Prestressed Concrete, Preprint No. 3424, a
Beams under Torsion, Journal of Structural Engineering, paper presented at ASCE Convention, Chicago, IL, 17 pp.;
V. 121, No. 9, pp. 1280-1286. Updated version: PCI Journal, 2004, Precast/Prestressed
Wagner, H., 1929, Ebene Blechwandtrger mit sehr Concrete Institute, V. 49, No. 3, pp. 34-42.
dnnem Stegblech, Zeitschrift fr Flugtechnik und Motor- Zia, P., and McGee, W. D., 1974, Torsion Design of
luftschiffahrt, V. 20, No. 8 to 12, Berlin, Germany. Prestressed Concrete, PCI Journal, V. 19, No. 2, pp. 46-65.
Walsh, P. F.; Collins, M. P.; and Archer, F. E., 1967, The
Flexure-Torsion and Shear-Torsion Interaction Behaviour of
Reinforced Concrete Beams, Civil Engineering Transac-
As ACI begins its second century of advancing concrete knowledge, its original chartered purpose
remains to provide a comradeship in finding the best ways to do concrete work of all kinds and in
spreading knowledge. In keeping with this purpose, ACI supports the following activities:
Technical committees that produce consensus reports, guides, specifications, and codes.
Spring and fall conventions to facilitate the work of its committees.
Educational seminars that disseminate reliable information on concrete.
Certification programs for personnel employed within the concrete industry.
Student programs such as scholarships, internships, and competitions.
Sponsoring and co-sponsoring international conferences and symposia.
Formal coordination with several international concrete related societies.
Periodicals: the ACI Structural Journal and the ACI Materials Journal, and Concrete International.
Benefits of membership include a subscription to Concrete International and to an ACI Journal. ACI
members receive discounts of up to 40% on all ACI products and services, including documents, seminars
and convention registration fees.
As a member of ACI, you join thousands of practitioners and professionals worldwide who share a
commitment to maintain the highest industry standards for concrete technology, construction, and
practices. In addition, ACI chapters provide opportunities for interaction of professionals and practitioners
at a local level.

38800 Country Club Drive
Farmington Hills, MI 48331
U.S.A.
Phone: 248-848-3700
Fax: 248-848-3701
www.concrete.org
The AMERICAN CONCRETE INSTITUTE
was founded in 1904 as a nonprofit membership organization dedicated to public

service and representing the user interest in the field of concrete. ACI gathers
and distributes information on the improvement of design, construction and
maintenance of concrete products and structures. The work of ACI is conducted by
individual ACI members and through volunteer committees composed of both members
and non-members.
The committees, as well as ACI as a whole, operate under a consensus format,

which assures all participants the right to have their views considered. Committee
activities include the development of building codes and specifications; analysis of
research and development results; presentation of construction and repair
techniques; and education.
Individuals interested in the activities of ACI are encouraged to become a member.

There are no educational or employment requirements. ACIs membership is
composed of engineers, architects, scientists, contractors, educators, and
representatives from a variety of companies and organizations.
Members are encouraged to participate in committee activities that relate to their

specific areas of interest. For more information, contact ACI.
www.concrete.org


ACI 445.1R-12 - Report On Torsion in Structural Concrete (2013)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

ACI 445.1R-12 - Report On Torsion in Structural Concrete (2013)

Uploaded by

Copyright:

Available Formats

ACI 445.

Reported by Joint ACI-ASCE Committee 445

Report on Torsion in Structural Concrete

American Concrete Institute

Report on Torsion in Structural Concrete

Daniel A. Kuchma, Chair Robert W. Barnes Jr., Secretary

Perry Adebar Hakim Bouadi Neil M. Hawkins Stavroula J. Pantazopoulou

CHAPTER 1INTRODUCTION AND SCOPE, p. 2

CHAPTER 3HISTORICAL OVERVIEW OF 9.4Design Example 1: solid rectangular reinforced

equilibrium, compatibility conditions, and Hookes Law.

Fig. 3.3.5aRauschs space truss model with struts and ties

(3.3.5)) was found to greatly overestimate the experimen-

Fig. 3.3.7.1aMembrane element in shear (Hsu 1993).

Collins and Chockalingam (1979) and Vecchio and Collins

the inclined face due to flexure reaches a reduced modulus

Fig. 4.3dEffects of concrete cover on the torque-twist

Fig. 4.4aCowans failure criteria (Hsu 1984).

In these tests, the beams had stirrup dimensions of 245 x

of cracks is fewer in HSC. Wider cracks in HSC could also

stirrup and concrete using NSC results in a portion of the

CHAPTER 5ANALYTICAL MODELS

stress. The in-plane deformation of this element should

e = e d cos2 q + e r sin 2 q (5.3.1a)

e t = e d sin 2 q + e r cos2 q (5.3.1b)

Element A is subjected to in-plane shear as well as out-of-

It is clear from Eq. (5.3.2) that the angle of twist, F, will

Fig. 5.3.4Strains and stresses in concrete struts (Hsu and Mo 1985a).

plane section hypothesis used in bending theory (Mitchell Descending branch:

5.4Stress strain relationships f = fy e ey (5.4.2a(b))

Ascending branch: 5.4.3 Prestressing reinforcementIn the case of

In structures where prestress is applied in both direc-

given in 5.2 and 5.3. The following solution algorithm is the

Substituting Eq. (5.5d) into Bredts Eq. (5.2.2), the

Like flexure, the compression zone depth is a function of

in the figure represents the relationship between vu/fc and wL

The ultimate torsional moment is also related to ultimate

3. Calculate the reinforcement indexes wL and wt

4. Calculate the nondimensional balanced torsional

The graphical solution using Fig. 5.7.2 is not exact. Its

a beam cross section can be subjected to a maximum of six

Fig. 6.2.3aTorque-twist relationships at various T/M

The interaction for concentrically prestressed members is

Fig. 6.2.3cTorque-twist relationships at various T/M ratios

Fig. 6.2.3eObserved flexural moment-curvature

Fig. 6.2.3fSuperposition of forces due to torsion and bending (Hsu 1993).

Fig. 6.3.1Shearing stress due to shear, torsion, and

action curve. ACI code equations were used to calculate To

Fig. 6.3.3bVarious normalized T-V interaction curves in

Elfgren et al. (1974b) developed a theoretical model for the

Fig. 6.4cTorque-axial tension calculated

higher compressive loads, it produces slightly better predic-

2 2 and Gabrielsson (1999). Both theoretical and experimental

inclined compression (Fig. 6.5e(a)), confirming the impor-

CHAPTER 7ADDITIONAL DESIGN ISSUES

7.2Compatibility torsion and torsional moment

Fig. 7.2.1.2Spandrel beam under concentrated

Fig. 7.2.1.4aSpandrel beam under lateral

tioned eccentric to the centroidal axes of the beam elements

Fig. 7.2.1.4bIdealization of spandrel beam under lateral sway (Pantazopoulou and

moment redistribution from the elastic analysis without M (7.2.5.1b)

conditions. Before 1995, ACI 318 adopted the skew-bending

Fig. 7.2.6cEffect of torsion-flexure interaction on ductility

Fig. 7.2.7Torsion in spandrel beam