Design Guide for Twisting

Moments in Slabs
Reported by Join t ACI-ASCE Committee 447
ACI 447R-18

@Seismicisolation
@Seismicisolation
 First Printing
April 2018
ISBN: 978-1-64195-010-7

Design Guide for Twisting Moments in Slabs

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 http://concrete.org/Publications/
DocumentErrata.aspx. 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 application and use of this information.

All information in this publication is provided “as is” without warranty of any kind, either express or
implied, including 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 consequential 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 development 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, 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 the ACI
Collection of Concrete Codes, Specifications, and Practices.

38800 Country Club Drive

Farmington Hills, MI 48331
Phone: +1.248.848.3700
Fax: +1.248.848.3701
www.concrete.org

@Seismicisolation
@Seismicisolation
 ACI 447R-1 8
Design Guide for Twisting Moments in Slabs
Reported by Joint ACI-ASCE Committee 447
Ganesh Thiagarajan, Chair Jian Zhao, Secretary

Riadh S. Al-Mahaidi Gianluca Cusatis Song F. Jan Dan Palermo

Gangolu Appa Rao Mukti L. Das Ioannis Koutromanos Guillermo Alberto Riveros
Ashra f S. Ayoub James B. Deaton Laura N. Lowes Mohammad Shara fbayani
Zdenĕk P. Bažant Jason L. Draper Yong Lu Hazim Sharhan
Allan P. Bommer Serhan Guner Yi-Lung Mo Sri Sritharan
Mi-Geum Chorzepa Trevor D. Hrynyk Abbas Mokhtar Zadeh
Carlos Arturo Coronado John F. Jakovich Wassim I. Naguib

Consulting Members
Ahmet Emin Aktan Wai F. Chen Robert Iding Syed Mizanur Rahman
Sarah L. Billington Christopher H. Conley Anthony R. Ingra fea Victor E. Saouma
Johan Blaauwendraad Robert A. Dameron Feng-Bao Lin Frank J. Vecchio
Oral Buyukozturk Filip C. Filippou Christian Meyer Kaspar J. Willam
Ignacio Carol Kurt H. Gerstle Hiroshi Noguchi
Luigi Cedolin Walter H. Gerstle Gilles Pijaudier-Cabot

This guide assists practitioners in understanding: 1) twisting CONTENTS

moments in two-way slabs, when twisting moments are an essential
consideration; 2) methods that can be used to account for twisting
moments in design; and 3) the options available for each method
of the various system geometries. Descriptions of twisting moments
are provided theoretically and visually in the guide, and six methods
of accounting for twisting moments in design are discussed. Appli-
cability of the various methods is evaluated through a comparison
of designs resulting from each method for a variety of two-way
slab types and geometries. The theories described in the guide also
apply to the design of two-way wall and two-way dome systems.
Keywords: fnite element analysis; shell design; slab design; torsion; twist;
twisting moments; wall design.
ACI Committee Reports, Guides, and Commentaries are
intended for guidance in planning, designing, executing, and
inspecting construction. This document is intended for the use
o f individuals who are competent to evaluate the signifcance
and limitations o f its content and recommendations and who
will accept responsibility for the application o f the material it
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 there from.
Re ference to this document shall not be made in contract
documents. If items found in this document are desired by
the Architect/Engineer to be a part o f the contract documents,
they shall be restated in mandatory language for incorporation
by the Architect/Engineer.
@Seismicisolation
@Seismicisolation
CHAPTER 1 —INTRODUCTION AND SCOPE, p. 2
1 .1 —Introduction, p. 2
1 .2—Scope, p. 2
CHAPTER 2—NOTATION AND DEFINITIONS, p. 2
2.1 —Notation, p. 2
2.2—Defntions, p. 3
CHAPTER 3—BACKGROUND, p. 3
3.1 —Qualitative introduction to twisting moments in
slabs, p. 3
3.2—Behavior o f linear-elastic isotropic slabs, p. 4
3.3—Equilibrium in slabs, p. 4
3.4—Principal axes, p. 4
3.5—Orthogonal reinforcement and equilibrium for
twisting moments, p. 5
3.6—E fects o f slab geometry on twisting moments, p. 5
3.7—Traditional slab design methods, p. 6
3.8—Finite element analysis (FEA)-based slab design
resultants, p. 6
ACI 447R-1 8 was adopted and published April 201 8.
Copyright © 201 8, American Concrete Institute.
All rights reserved including rights o f reproduction and use in any form or by
any means, including the making o f copies by any photo process, or by electronic
or mechanical device, printed, written, or oral, or recording for sound or visual
reproduction or for use in any knowledge or retrieval system or device, unless
permission in writing is obtained from the copyright proprietors.

1
2 DESIGN GUIDE FOR TWISTING MOMENTS IN SLABS (ACI 447R-1 8)
CHAPTER 4—AVAILABLE DESIGN METHODS, p. 7
4.1 —Finite element analysis (FEA)-based design ignoring
twist, p. 7
4.2—Design using the Wood and Armer method, p. 7
4.3—Design using the sandwich model, p. 7
4.4—Design using element nodal forces, p. 8
4.5—Design using twist- free analysis, p. 9
CHAPTER 5—COMPARISON OF DESIGN
METHODS, p. 10
5.1 —Sensitivity to angle o f principal axes, p. 1 0
5.2—Typical design conditions, p. 11
CHAPTER 6—TWO-WAY WALLS, p. 21
6.1 —General considerations, p. 21
6.2—Impact o f twisting moment on walls exhibiting two
adjacent fxed edges, p. 21
CHAPTER 7—SHELL STRUCTURES, p. 24
7.1 —General considerations, p. 24
7.2—Typical bulk material storage hemisphere, p. 24
7.3—Typical loading conditions, p. 25
7.4—Typical design regions, p. 25
CHAPTER 8—REFERENCES, p. 29
Authored documents, p. 29
CHAPTER 1 —INTRODUCTION AND SCOPE
1.1 —Introduction
Section 8.2.1 o f ACI 31 8-1 4 allows slabs to be designed
by any procedure that satisfes equilibrium and geometric
compatibility, and requires that, at each section, the design
strength exceeds the required strength and serviceability
requirements are fulflled.
Traditional strip design methods for slabs are based
on approximate analysis and provide neither a complete
equilibrium load path or satis fy geometric compatibility.
Nonetheless, these methods have been used success fully
for many years to design slabs with supports arranged in a
rectangular grid.
From 1 995 to 201 5, design engineers transitioned from
predominantly using traditional slab analysis methods to
using fnite element analysis (FEA). More recently, engi-
neers use FEA to assist in the structural design o f two-way
concrete members. Twisting moments in two-way slabs can
require additional reinforcement from those proportioned
for bending moments, yet they are o ften misunderstood
and sometimes ignored, neglected, or both, by practitio-
ners in design. This is most likely due to their lack o f being
discussed comprehensively in design codes and frequent
exclusion from college concrete design course curricula.
Although FEA solutions provide a full equilibrium load
path and satis fy geometric compatibility, they determine
load paths that require twisting moments for equilib-
rium (Shin et al. 2009). Many designers using FEA have
ignored these twisting moments—a possible unconserva-
@Seismicisolation
@Seismicisolation
tive assumption where twisting moments are high (Park and
Gamble 2000). To provide designers with guidance related
to this issue, methods for explicitly incorporating twisting
moments determined from FEA in the design o f slabs are
discussed in this guide.
The purpose o f this design guide is to provide advice
to design engineers who analyze slab systems with fnite
element methods and who need to ensure their designs are
satis factory for the twisting moments predicted by the anal-
ysis. This guide provides background information regarding
twisting moments and describes multiple approaches for
consideration o f twisting moments in design. It also provides
advice for designers o f walls and shells with twisting moment
conditions similar to those in slabs.
1.2—Scope
This design guide applies to slabs o f both uni form and
nonuni form thicknesses, including drop caps and drop
panels, except where noted in the text. This guide does not
apply to wafe slabs, or the beams o f beam-and-slab f oor
systems. Chapters 3 through 6 address slabs and walls in
which the response is determined purely by bending. Chapter
7 addresses shells for which the response is determined by
bending and membrane action. Chapter 6 and the theory
sections o f this guide are applicable to walls. Chapter 7 and
the theory sections o f this guide are applicable to shells, with
the caveat that equations presented in Chapter 3 are not valid
for curved shells.
CHAPTER 2—NOTATION AND DEFINITIONS
2.1 —Notation
ci, j = fraction for consideration o f sections partially
crossing element to apply to forces in local node j
in element i
D = f exural rigidity o f plate, in.-lb (N· mm)
E = Young’s modulus, psi (MPa)
F = force vector, lb (N)
fi, j = nodal force vector for local node j in element i
h = thickness o f slab or plate, in. (mm)
L = width o f design section, in. (mm)
M = bending moment, or moment vector, in.-lb (N· mm)
Md = design bending moment, in.-lb (N· mm)
Mi = bending moment from isotropic analysis, in.-lb
(N· mm)
Mtf = bending moment from twist- free analysis, in.-lb
(N· mm)
Mu = design moment for slab cross section, in.-lb (N· mm)
m i, j = nodal moment vector for local node j in element i,
in.-lb/in. (N· mm/mm)
mr = bending moment causing stresses parallel to r-axis,
per unit length o f slab or plate, in.-lb/in. (N· mm/
mm)
m rs = twisting moment relative to r- s-axes per unit length
o f slab or plate, in.-lb/in. (N· mm/mm)
ms = bending moment causing stresses parallel to s-axis,
per unit length of slab or plate, in.-lb/in. (N·mm/mm)
 DESIGN GUIDE FOR TWISTING MOMENTS IN SLABS (ACI 447R-1 8) 3

Fig. 3.1—Twisting moment example free body diagram.

m ux = design moment causing stresses parallel to x-axis, vx = transverse shear on x- face per unit length, lb/in.
per unit length o f slab or plate, in.-lb/in. (N· mm/ (N/mm)
mm) vy = transverse shear on y- face per unit length, lb/in.
m ux+ = positive design moment causing stresses parallel (N/mm)
to x-axis, per unit length o f slab or plate, in.-lb/in. w = transverse de fection, in. (mm)
(N· mm/mm) xi,j = distance vector from section centroid to local node
m ux– = negative design moment causing stresses parallel to j in element i, in. (mm)
x-axis, per unit length o f slab or plate, in.-lb/in. ν = Poisson’s ratio
(N· mm/mm)
m uy = design moment causing stresses parallel to y-axis, per 2.2—De f ntions
unit length of slab or plate, in.-lb/in. (N·mm/mm) ACI provides a comprehensive list o f defnitions through
m uy+ = positive design moment causing stresses parallel an online resource, ACI Concrete Terminology. Defnitions
to y-axis, per unit length o f slab or plate, in.-lb/in. provided herein complement that source.
(N· mm/mm) anticlastic bending —curvature caused by the Poisson
m uy– = negative design moment causing stresses parallel e fect and curvature about a perpendicular axis.
to y-axis, per unit length o f slab or plate, in.-lb/in. strip design method —a method o f designing slabs by
(N· mm/mm) dividing them into two sets o f approximately perpendicular
mx = bending moment causing stresses parallel to x-axis, strips, with each strip analyzed and designed independently
per unit length of slab or plate, in.-lb/in. (N·mm/mm) from each other.
m xy = twisting moment relative to x- y-axes per unit length
o f slab or plate, in.-lb/in. (N· mm/mm) CHAPTER 3—BACKGROUND
my = bending moment causing stresses parallel to y-axis,
per unit length of slab or plate, in.-lb/in. (N·mm/mm) 3.1 —Qualitative introduction to twisting moments
nx = membrane tension in x-axis direction per unit in slabs
length, lb/in. (N/mm) Twist exists in most every slab, except those theoretical-
n xy = membrane in plane-shear in x- y-axes direction per case-only slabs whose moments at any point are identical
unit length, lb/in. (N/mm) about any axis. Figure 3.1 illustrates an extreme case o f
ny = membrane tension in y-axis direction per unit twist; a square slab with supports at three corners and a load
length, lb/in. (N/mm) at the fourth corner.
q = transverse load per unit area, lb/in. 2 (N/mm 2 ) From equilibrium, it can be shown that the bending
T = torsional moment, in.-lb (N· mm) moment about the A-A and B-B axes in the fgure is zero,
V = shear force, lb (N) although this slab is clearly supporting a load and needs
Vd = design shear force, lb (N) to be reinforced. Looking at the C-C axis, whose bending
@Seismicisolation
@Seismicisolation moment is nonzero, gives us insight to the load-carrying
4 DESIGN GUIDE FOR TWISTING MOMENTS IN SLABS (ACI 447R-1 8)

mechanism o f the slab. What appears as twist about A-A

and B-B is bending about C-C and D-D, as shown in the
de f ected shapes along C-C and D-D.

3.2—Behavior of linear-elastic isotropic slabs

Like other components in concrete structures, the design
forces for slabs are typically determined using some form o f
a linear-elastic analysis. For slabs that are not thick (span-to-
thickness ratio o f more than 1 0), linear-elastic slab behavior
can be predicted by well-known plate theory.
The de f ection o f a thin, linear-elastic, isotropic plate
subjected to loads perpendicular to its plane can be expressed
by a fourth-order partial di ferential equation (Timoshenko
and Woinowski-Krieger 1 959)

∂4w ∂4w ∂4w q

+2 + = (3.2a)
∂x 4
∂x ∂y
2 2
∂y 4 D
Fig. 3. 3—An in fnitesimal plate element shown with resulting
where w = w(x, y) is the transverse (out-o f-plane) displacement
shear forces and moments due to transverse loading.
feld in the direction o f loading; q = q(x, y) per unit area; and D
is the fexural rigidity o f the plate, which is expressed as 3.3—Equilibrium in slabs
By considering the rotational equilibrium o f the slab
Eh 3 element shown in Fig. 3.3, Eq. (3.3), describing the inter-
D= (3.2b)
1 2 (1 − ν 2 ) relationship o f m x, m y, and m xy, can be derived.

in which E is the modulus o f elasticity, h is the thickness o f ∂ 2 mx ∂ 2 m xy ∂2my

+2 + = −q (3.3)
the plate, and ν is Poisson’s ratio. ∂x2 ∂ x∂ y ∂y 2
The de formations o f the plate can be interpreted as two
Equation (3.3), which holds for all slabs regardless o f
∂2 w w ∂
2
∂2w material behavior, reinforcement, or cracking, shows that
curvatures and , and one twist . Corre-
∂x 2 ∂y 2
∂ x∂ y any change in m xy must be o fset by some change in m x, m y,
sponding to these three de formations, there are two bending or both. Hence, m xy can only be reduced to zero by changing
moments and one twisting moment as shown in Eq. (3.2c). the load path o f the slab.

 ∂2 w 
3.4—Principal axes
  The values o f m x, m y, and m xy are related via Mohr’s circle
∂x2  (Timoshenko and Woinowski-Krieger 1 959); the moments
 mx   D νD 0  
  ∂ w  m r, m s, and m rs for any set o f perpendicular r-s axes can be
2
  
m
 y = −

νD D 0
  ∂y 2  (3.2c) determined from m x, m y, and m xy, as shown in Eq. (3.4a)
m   0 0 (1 − ν ) D / 2   
 xy   ∂2w  through (3.4c).
2 
 ∂ x∂ y  1 1
mr = ( mx + m y ) + ( m x − m y ) cos 2 θ + m xy sin 2 θ (3.4a)
2 2

∂2 w 1 1
From Eq. (3.2c), note that unless
∂ x∂ y
is zero, there will ms = (mx + m y ) − ( m x − m y ) cos 2 θ − m xy sin 2 θ (3.4b)
2 2
∂2w
be twisting moments in the plate. In general, is zero in 1
∂ x∂ y mrs =− ( m x − m y ) sin 2 θ + m xy cos 2 θ (3.4c)
2
only a small subset o f locations in a slab, so twisting
moments will exist almost everywhere in a linear-elastic
isotropic slab. For every point in a plate, there is one set o f principal
axes, where m rs is zero and about which the slab is in pure
bending. The orientation o f these axes changes from point to
point in the slab, and the principal axes are typically more
@Seismicisolation
@Seismicisolation valuable for general understanding than for determining
 DESIGN GUIDE FOR TWISTING MOMENTS IN SLABS (ACI 447R-1 8)
design forces. Figure 3.4 illustrates the construction o f
Mohr’s circle for Eq. (3.4a) through (3.4c).
3.5—Orthogonal reinforcement and equilibrium for
twisting moments
How can reinforcement be used to resist twisting
moments? This question is most conveniently addressed by
considering the reinforcement required to resist the moments
about the principal axes. Figure 3.5 illustrates the conceptual
approach.
Figures 3.5(a) and 3.5(b) illustrate the slab di ferential
element to be reinforced, and its principal moments m 1 and
Fig. 3. 4—Mohr’s circle for situation where mx and my are
both positive and mx > my.
@Seismicisolation
@Seismicisolation
Fig. 3. 5—Orthogonal reinforcement from twisting moments.
5
m 2 . The values o f m 1 and m 2 cause tension in the bottom
o f the slab. Figure 3.5(c) shows bottom reinforcement
forces aligned with the principal axes, selected using stan-
dard bending design approaches, and the x- and y-axis force
vectors the reinforcement provides. Figure 3.5(d) shows
the reinforcement forces along the 1 - and 2-axes replaced
by reinforcement forces along the x- and y-axes that, when
combined, provide vector forces equivalent to the 1 - and
2-axes reinforcement forces.
3.6—Effects of slab geometry on twisting
moments
The slab in Fig. 3.6a illustrates how di ferent slab regions
and geometries afect the importance o f the consideration o f
twisting moments when designing a slab. The fgure high-
lights three slab regions that are discussed as follows. The
slab is assumed to be designed with reinforcing bars parallel
to the x- and y-axes.
Region 1 o f the slab is regular with all columns arranged
in a rectangular grid. Regions similar to this are not twist-
sensitive. Ignoring twisting moments in these regions can
be proved safe through yield line theory. Although yield line
theory is an upper-bound approach, for a slab with regular
supports such as those in Region 1 , the number o f yielding
patterns to consider is small. If top reinforcement is clus-
tered at the columns, only the folded-plate pattern in each
6 DESIGN GUIDE FOR TWISTING MOMENTS IN SLABS (ACI 447R-1 8)

is approximately 40 percent less than calculated. This

percentage is calculated by comparing integrated slab strip
bending moments using isotropic plates and using ortho-
tropic plates with zero-twist sti fness. This calculation
assumes that the slab has no reliable twist capacity if twist
has not been considered in the design o f reinforcement (Mtf/
Mi = 1 .4). The strip design methods re ferred to in this docu-
ment are those commonly used by practicing engineers,
not the more sophisticated strip approaches discussed in
academic literature.

3.7—Traditional slab design methods

Fig. 3. 6a—Representative regions of a slab foor plan. Traditionally, slab design is performed using design forces
Region 1 is twist-sensitive, Region 2 moderately twist-sensi- determined by simplifed analysis methods that idealize the
tive, and Region 3 highly twist-sensitive. slab as a set o f intersecting strips (e fectively supports and
wide beams) in two perpendicular directions. The required
reinforcement is calculated using the strip bending moments
and standard beam bending design approaches.
These methods have two primary f aws. The frst is that
equilibrium is not fully satisfed because there is no consis-
tency between the load paths o f the two sets o f strips. The
second f aw is that de formation compatibility between both
parallel and perpendicular strips is ignored.
Despite these f aws, slabs designed using these methods
have generally performed well. For slabs with supports
arranged in a rectangular grid it is o ften shown, using yield-
line theory or lower-bound methods, that the traditional strip
Fig. 3. 6b—Folded plate yield lines. methods provide adequate capacity when reinforcement is
distributed appropriately. (Burgoyne 2004; Kennedy and
Goodchild 2003). For those slabs with supports not arranged
direction shown in Fig 3.6b need be considered (Kennedy
in a rectangular grid, engineering judgment is necessary
and Goodchild 2003 ). The folded plate yield line patterns
to determine if the traditional methods will produce a safe
show that any reinforcement design that supports the total
design.
span static moment (wL 2 /8 for a uni form load) in each direc-
tion is safe.
Region 2 o f the slab is where there is a major shi ft in
3.8—Finite element analysis (FEA)-based slab
the column layout. Regions similar to this are moderately
design resultants
To complete slab design, an engineer should determine
twist-sensitive. The slab tends to span along a 45-degree
the quantity o f reinforcement required for each design
line between the closest columns. This diagonal spanning
cross section. Although design cross section locations and
reduces the bending moments along the x- and y-axes while
lengths are o ften guided by code rules, in general, sections
increasing twisting moments. Ignoring twisting moments in
are needed at peak stress locations and their lengths based
design o f similar regions could lead to a slab load capacity
on the extent o f the slab that can be assumed to act as a
that is approximately 20 percent less than calculated. This
unit in resisting internal forces. The width o f each section
percentage is calculated by comparing integrated slab strip
should be chosen so that the moment distribution along the
bending moments using isotropic plates and using ortho-
section is reasonably uni form, does not change sign, and can
tropic plates with zero twist sti fness. This calculation
be resisted by uni formly distributed reinforcements.
assumes that the slab has no reliable twist capacity if twist
When using FEA to support slab design, the engineer
has not been considered in the design o f reinforcement, and
should convert the slab analysis element results to resultant
is performed by dividing the bending moment determined by
forces and moments acting on these sections. At this stage,
a twist- free analysis, Mtf, by the bending moment determined
the engineer should frst trans form the results o f FEA into a
by an isotropic analysis, Mi. For Region 2, Mtf/Mi = 1 .2.
coordinate system orthogonal to the section, and then inte-
Region 3 o f the slab is a cantilevered corner. Regions
grate/sum all forces and moments acting on the section to
similar to this are highly twist-sensitive. The slab tends to
determine the design moment. Consideration o f twisting
span between, and cantilever o f, the line between the two
moments is the most di f cult aspect o f this conversion.
columns. Although this diagonal cantilevering reduces the
ACI 31 8 does not explicitly address twisting moments, nor
bending moments along the x- and y-axes, it adds large
does the commentary provide guidance on their consid-
twisting moments. Ignoring twisting moments in design
eration. However, Section 6.1 .1 o f ACI 31 8.2-1 4, which
o f similar regions could lead to a slab load capacity that
@Seismicisolation
@Seismicisolation addresses shell reinforcement, states that reinforcement
 DESIGN GUIDE FOR TWISTING MOMENTS IN SLABS (ACI 447R-1 8)
must be designed to resist tension from bending and twisting
moments.
Design can also be performed considering the FEA results
at each point in the slab, independently and foregoing the
integration along a section. A zero-length section should
still be considered to provide the correct axes for the design
resultants and reinforcement. Design code requirements,
such as those in ACI 31 8, can be di f cult to apply to these
point results, as the codes are written for section-based
results. Another potential di f culty with using FEA results
at a point for design is that the results at certain locations,
such as under a point load, will approach infnity as the mesh
is refned. In general, the approaches described in this guide
apply to either point-based or section-based resultants.
CHAPTER 4—AVAILABLE DESIGN METHODS
This chapter discusses various options to consider for
twisting moments. Chapters 5 through 7 evaluate some
o f these options in sample structures. Design methods
discussed in this chapter are typically used with the results
from linear-elastic analyses.
4.1 —Finite element analysis (FEA)-based design
ignoring twist
A simple, commonly-used approach to determine the
design moment for a slab cross section is to integrate the
FEA-predicted moment about the axis o f the section along
the length o f the section.
Mu = ∫m rds (4.1 )
In Eq. (4.1 ), m r is the moment per unit length along the
section. This approach ignores twisting moments in the equi-
librium load path and may signifcantly underestimate the
design demand.
The e fect o f twisting moments can be safely ignored
in this manner when sections are approximately aligned
orthogonally to the principal bending directions at all loca-
tions, as the twist on the section vanishes in this case. An
instructive rule o f thumb is that twist can be ignored if it
is smaller than 1 0 percent o f the primary bending moment
(Deaton 2005). If this criterion is not satisfed, neglecting the
e fect o f twist could lead to unconservative results.
4.2—Design using the Wood and Armer method
Wood (1 968) and Armer (1 968) (re ferred to in this docu-
ment as Wood and Armer) proposed one o f the most popular
methods for explicitly incorporating twisting moments in
slab design. The method seeks to prevent yielding in all direc-
tions and was developed by considering the normal moment
and Johansen’s yield criteria (Park and Gamble 2000). At
any point in the slab, for any arbitrary direction, the design
moment determined from FEA results must not exceed the
ultimate normal resisting moment in that direction. The ulti-
mate normal resisting moments calculated from the rein-
forcement in the x- and y-directions are typically compared
to adjusted design moments m ux and m uy. Design moments
@Seismicisolation
@Seismicisolation
based on the Wood and Armer method are computed
7
m ux = m x ± | m xy| (4.2a)
m uy = m y ± | m xy| (4.2b)
where all plus signs are used to compute required bottom
reinforcement, all minus signs are used to compute required
top reinforcement, and m ux and m uy will be negative when the
top reinforcement is in tension. This assumes a sign conven-
tion where positive moment causes tension on the bottom
surface o f the slab and negative moment causes tension on
the top surface o f the slab. For non-slab systems such as
walls and domes, top and bottom should be defned such that
they are consistent throughout and consistent with loading
assumptions.
Where m ux or m uy is found to have the opposite sign from
what is expected (negative for bottom reinforcement or
positive for top reinforcement), the design moment can be
conservatively set to zero. A less conservative calculation
option for this case is shown in Table 4.2.
The Wood and Armer approach requires that a fne regular
mesh be used to produce accurate slab moment and twist
predictions. Also, because moments, twists, or both, can be
theoretically infnite at concentrated loads or reactions (such
as slab column connections), integration o f the moments or
twists at these locations is di f cult to perform accurately.
Previous studies have indicated that the Wood and Armer
method could produce unconservative results for slabs with
high reinforcement ratios (approximately more than 0.75
percent) at regions o f signifcant twisting moments, espe-
cially near restrained slab corners (May and Lodi 2005).
Note also that this method cannot be directly applied to slabs
with beams or drop panels. The Wood and Armer method
has been modifed and implemented in more conservative
forms in both CSA-A23.3-04 and EN 1 992-1 -1 :2004 .
4.3—Design using the sandwich model
Designs based on the normal moment yield criterion, such
as the Wood and Armer method, do not account for trans-
verse shear or membrane forces within the slab. The sand-
wich model approach allows for the computation o f rein-
forcement to resist not only bending and twisting moments,
but also the e fects o f shear and membrane stresses. In this
approach, membrane e fects (n x, n y, and n xy) and bending and
twisting moments (m x, m y, and m xy) are resisted by the sand-
wich exterior layers, whereas the shear e fects (vx and vy)
are resisted via the sandwich core. Figure 4.3 (Marti 1 990)
shows this concept.
To design by using the sandwich model, a slab section
is divided into three layers; the depth o f each layer can be
determined such that the middle planes o f the outer layers
coincide with the center o f the top and bottom reinforce-
ment layers. Then, bending moments are decomposed into
a couple o f tensile and compressive normal forces, and
twisting moments are decomposed into a couple o f in-plane
shear forces acting at top and bottom layers o f the slab. The
sandwiched inner layer is used to resist transverse shear.
These normal and shear forces due to moments are combined
with membrane (in-plane) forces. Finally, slab reinforce-
8 DESIGN GUIDE FOR TWISTING MOMENTS IN SLABS (ACI 447R-1 8)

Table 4.2—Less conservative design moments (Wood 1 968; Armer 1 968)

m x > | m xy| m x ≤ | m xy| m x < –| m xy|
m uxbot = 0
m uxbot = m x + | m xy| m xy
2

m uxbot = m x + | m xy| m xy
2 mux
top
= mx −
my
m uxtop = 0 mux
top
= mx −
m y > | m xy| m uybot = m y + | m xy|
my
2
m xy
m uytop = 0 m uybot = m y + | m xy| muy
bot
= my +
mx
m uytop = 0
m uytop =0
m uxbot = m x + | m xy| m uxbot = 0
m uxtop = 0 m uxbot = m x + | m xy| m uxtop = m x – | m xy|
m uybot = m y + | m xy| m uxtop = m x – | m xy| 2
m y ≤ | m xy| 2 m uybot = m y + | m xy| muy
bot
= my +
m xy
m xy mx
top
muy = my − m uytop = m y – | m xy|
mx
m uytop = m y – | m xy|
2
m xy
bot
mux = mx + m xy
2
my
mux
bot
= mx + m uxbot = 0
my
=0 m uxtop = m x – | m xy|
m y < –| m xy| m uxtop
m uybot = 0
m uybot =0 m uxtop = m x – | m xy|
m xy
2 m uybot = 0 m uytop = m y – | m xy|
top
muy = my − m uytop = m y – | m xy|
mx

Fig. 4. 4. 2—Design section forces derived from element

nodal forces.

ties from which slab design forces are calculated. Another

Fig. 4. 3—Sandwich model.
ment is proportioned to satis fy equilibrium conditions at
each top and bottom layer in the two orthogonal directions.
Further descriptions o f this method have been presented by
fb Bulletin 45 (Comité Euro-International du Béton 2008),
Brøndum-Nielsen (1 985), Lourenço and Figueiras (1 995),
and the CEB-FIP Model Code (Comité Euro-International
du Béton 1 993 ).
The sandwich model is not typically used for slab design
for commercial structures or generally available in slab
design so ftware. The assumed lever arm between the outer
layers is very conservative for thin slabs, as the ratio o f true
ultimate lever arm to assumed lever arm is signifcantly
greater than 1 .0 (although this conservatism is reduced as
the slab thickness is increased).
4.4—Design using element nodal forces
4.4.1 Nodal forces approach —The methods discussed
@Seismicisolation
in 4.1 , 4.2, and 4.3 use slab stresses as the analysis quanti-
@Seismicisolation
approach for slab design using FEA results is to compute
results for the design cross section from element nodal
forces and moments. Using this approach, the section
design forces are guaranteed to be in equilibrium with the
externally applied nodal loads (Deaton 2005). Slab design
methods based on element nodal forces have been imple-
mented in various FEA so ftware. These methods are attrac-
tive because results are relatively accurate (even for very
coarse or irregular meshes), as they can be used for slabs
containing beams or drop panels and are easily extended to
design post-tensioned f oors.
4.4.2 Calculation of nodal forces —The nodal forces
approach trans forms all element nodal forces and moments
acting at nodes on the design cross section into resultant
forces and moments acting at the centroid o f the section, as
shown in Fig. 4.4.2. Because trans formation o f nodal forces
to the centroid considers the eccentricity o f the nodes from
the centroid, equilibrium o f all acting forces and moments
is provided.
The equilibrium equations in vector form are:
 DESIGN GUIDE FOR TWISTING MOMENTS IN SLABS (ACI 447R-1 8)
Fig. 4.4.4—Assumed shear stress distribution causing torsion.
? ?
F = ∑∑ci , j ⋅ fi , j
i j
? ? ? ?
M = ∑∑ci , j ⋅ ( mi , j + xi , j × fi , j )
i j
where i is the iterator over the elements crossed; j is the
iterator over the local nodes in element i (on one side o f the
? ?
section); F and M are the force and moment vectors at the
? ?
section centroid, respectively; fi , j and mi , j are nodal force
?
and moment vectors, respectively; xi , j is a vector distance
from the section centroid to node i,j; and ci,j is a multiplier
that equals 1 .0 unless the section does not fully cross element
i, in which case it varies between 0.0 and 1 .0, typically based
on the fraction o f the element crossed.
The forces and moments to be used in the slab design
? ?
are found by trans forming the F and M vectors into the
coordinate system o f the design cross section. While six
resultants are determined, typically only transverse shear V,
bending moment M, and torsion T are o f interest. There is no
resultant quantity that directly corresponds to the twisting
moments on the section. Methods available for consider-
ation o f the calculated torsion are as follows.
4.4.3 Torsion as bending —One approach to incorporating
T is to directly combine M with ±T, as shown in Eq. (4.4.3).
E fectively, this method uses T as a proxy for integrated
twist and then applies the Wood and Armer approach.
Md = M ± T
This approach can lead to unconservative design moments
in two conditions:
(1 ) Condition 1 —Twisting moments along the cross
section change sign and partially cancel in their integration
(2) Condition 2—Torsion due to shear forces is o f the
opposite sign as torsion due to twist
Condition 1 can be detected visually and avoided by
breaking the design section into two design sections;
however, this can be di f cult when considering numerous
loading conditions, as the location o f the sign change could
vary. Condition 2 rarely occurs without Condition 1 also
occurring. When neither Condition 1 or 2 occur, the design
moments derived from considering torsion as bending can
be conservative. For a simple torsional cantilever slab, the
design moments are twice those derived from the Wood and
Armer approach (Den Hartog 201 4).
4.4.4 Torsion as shear—The total torsion T acting on the
design cross section can be decomposed into an assumed
@Seismicisolation
@Seismicisolation
linearly varying transverse shear per unit length, as shown in
9
Fig. 4.4.4. This torsional shear stress has a maximum value
(4.4.2a)
equal to 6 T/L 2 , where L is the width o f the section—that is,
the width o f section cut in Fig. 4.4.2. This is analogous to
the bending stress in a rectangular beam being calculated as
(4.4.2b)
6 M/bh 2 .
By multiplying this maximum shear stress assumed due to
torsion by the width o f the section, an equivalent, but very
conservative, resultant design shear force due to torsion,
which is equal to 6 T/L , can be considered during the shear
design o f the section, as is shown in Eq. (4.4.4).
6T 6T
Vd =V± 2
L =V± (4.4.4)
L L
Considering torsion as shear is generally not applicable
when signifcant twisting moments exist. However, applying
the torsion-as-shear design approach to the results o f a twist-
free analysis (3.5) has the potential for providing improved
vertical shear designs while simultaneously fully consid-
ering twisting moments.
4.4.5 Nodal forces with twisting moment integrations —
While the nodal forces approach does not provide twisting
moments, it can be supplemented with element twisting
moment integrations to provide a full set o f design forces
suitable for the Wood and Armer method or other design
methods. The beneft o f using such a hybrid approach is to
gain accuracy o f the nodal forces approach for all quantities
to which it applies.
(4.4.3) 4.5—Design using twist-free analysis
Slab analysis and design so ftware capable o f analyzing
orthotropic slabs (with di ferent properties in two orthogonal
directions) typically allows engineers to separately modi fy
the sti fness corresponding to each o f the bending moments
about the two orthogonal axes and the twisting moment
(Shin et al. 2009). Setting the twist sti fness to zero leads to
an analytical solution and load path in which all the twisting
moments are zero. The behavior and load path o f a twist-
free slab is somewhat similar to that o f a grillage o f beams.
The design philosophy behind twist- free analysis is similar
to that for the compatibility torsion approach that has long
been accepted in ACI 31 8.
Bending and twisting moments are interrelated through
a Mohr’s circle equilibrium constraint. Because o f this
constraint, the twist sti fness can only be set to zero about a
single set o f perpendicular axes at each location. The most
practical axes for this purpose are the axes of the design cross
sections, which are perpendicular to the reinforcement. In a
typical slab that is reinforced in only two perpendicular direc-
10 DESIGN GUIDE FOR TWISTING MOMENTS IN SLABS (ACI 447R-1 8)
Fig. 5. 1a—(a) Uniform moment feld; (b) de fected shape; and (c) design sections to rotate
in evaluation of twist-sensitivity.
tions, the entire slab can be given the same modifed sti fness.
For slabs where the reinforcement directions vary from region
to region, multiple sti fness modifcations are required.
Twist- free slab analyses result in larger de f ections than
those predicted by analyses with nonzero twist sti fness. This
is to be expected, as the load path with nonzero twist sti f-
ness is more e f cient in terms o f elastic energy than one with
zero twist sti fness. Comparing slab de f ections computed by
analyses with and without twist sti fness is a quick means
o f gauging how much the forces will be redistributed to
mobilize the twist- free load path. I f the de f ections from the
twist- free analysis are signifcantly greater than those from
the analysis with twist sti fness— for example, by a factor o f
2—the potential for twist-related cracking should be inves-
tigated. Twist- free analysis can be used in combination with
nodal force or moment resultant methods.
CHAPTER 5—COMPARISON OF DESIGN
METHODS
This chapter compares the available twisting moment
design approaches in various scenarios. Although numer-
ical comparisons are provided, the intent is to qualitatively
illustrate how well or poorly each design method considers
support confgurations and design section orientations. The
slabs investigated in this chapter are thin (8 in. [200 mm]),
with small bars (0.5 in. [1 2 mm]) and moderate cover (1 in.
[25 mm]); other slab parameters have little inf uence over
the comparative results.
@Seismicisolation
@Seismicisolation ensures a safe design.
5.1 —Sensitivity to angle of principal axes
This section investigates the sensitivity o f the considered
design methods to the angle between the design section and
the principal zero-twist axes for an isotropic linear-elastic
slab. Figure 5.1 a shows a slab with uni form moments (and
zero twist) about the x- and y-axes (one moment causing
de f ection upward and the other causing de f ection down-
ward), causing uni form twist and zero bending moments
along the 45-degree axis.
The design sections shown in Fig. 5.1 a(c) are rotated
about their crossing point and the design quantities and
results investigated. At angle zero, there is pure bending and
no twist. As shown in Fig. 5.1 b, all methods, except for the
sandwich method, lead to the same results. The sandwich
method requires additional reinforcement due to its conser-
vative lever arm assumption. At 45 degrees, there is pure
twist and no bending. Between zero and 45 degrees, there is
a combination o f twist and bending.
Figure 5.1 b shows the relative total quantities o f rein-
forcement required as the design sections are rotated from
zero to 45 degrees. The y-axis is normalized by the required
reinforcement for the no-twist zero angle. Although this
plot shows reinforcement quantities for a particular design
scenario, general behaviors can be observed.
The Wood and Armer method and the torsion-as-bending
method have similar curves that approach a ratio o f 2 as the
slab approaches pure twist. As explained in 3.2, the Wood
and Armer approach has a strong theoretical basis that
 DESIGN GUIDE FOR TWISTING MOMENTS IN SLABS (ACI 447R-1 8)
Fig. 5.1b—Relative calculated total reinforcement versus
section angle.
It is expected that the torsion-as-bending approach will give
similar results to the Wood and Armer approach, provided that
the total torsion is approximately the same as the total twist.
Figure 5.1 b shows that the torsion-as-bending gives the same
design as Wood and Armer for pure twist and pure bending,
and similar designs for cases between those two extremes. The
diferences between Wood and Armer and torsion-as-bending
in the plot are caused by the o f the ci,j factor o f Eq. (4.4.2b) not
exactly matching the m xy stress integration results.
Ignoring torsion and treating torsion-as-shear have iden-
tical curves that approach zero as the slab approaches pure
twist. In this case, the torsional moments were not large
enough for the torsion-as-shear method to require transverse
reinforcement. There fore, it is expected that its design will
be the same as ignoring torsion. Both methods are unsafe for
the pure twist condition.
Although the sandwich model follows a similar pattern
to Wood and Armer, it requires more reinforcement due to
its conservative lever-arm assumption. The test slab is thin
(8 in. [200 mm]), which accentuates this conservatism. As
explained in 4.3, the sandwich method has a strong theo-
retical basis that ensures a safe design.
The twist- free method provides a clear, safe equilibrium
load path and is e f cient for angles less than approximately
25 degrees. However, at angles greater than 25 degrees, the
reinforcement layout creates an ine f cient load path, where
it becomes questionable if this load path can be attained
without some failure or serviceability distress. At 45 degrees,
the requirements theoretically approach infnity and no equi-
librium is found.
Figure 5.1 c shows the predicted relative maximum de f ec-
tions as sections (and the slab behavior in the twist- free case)
are rotated from zero to 45 degrees. The y-axis is normalized
by the isotropic de f ection value. For all methods, except
twist- free analysis, the predicted de f ections are constant,
as the design method does not afect the isotropic analysis.
For twist- free analysis, the slab orthotropic properties are
oriented parallel to the design sections, and the slab becomes
increasingly more f exible as the axes approach the pure-
twist axes o f 45 degrees. Above approximately 25 degrees,
@Seismicisolation
@Seismicisolation
11
Fig. 5.1c—Relative maximum defection versus section
angle.
the de f ections become very large, and signifcant cracking
would be expected for the slab to de form from the uncracked
isotropic load path to the cracked orthotropic load path. This
plot shows that the ratio o f twist- free de f ection to isotropic
de f ection can be used as a proxy to determine qualitatively
the amount o f redistribution and cracking necessary to reach
a twist- free load path.
5.2—Typical design conditions
This section investigates regions o f slabs with plan geom-
etries that occur frequently, comparing how the considered
design methods di fer for each o f the geometries. Isotropic
and orthotropic (twist- free) linear-elastic behaviors are
investigated. While the slab confgurations discussed do not
cover every scenario, they provide a range o f conditions that
expose the strengths and weaknesses o f each design method.
Figure 5.2 locates the typical conditions on a slab plan and
lists the section numbers where the conditions are discussed.
Note that while many design codes, including ACI 31 8,
would suggest using column strips and middle strips for slab
design, the design sections investigated in this section have
been chosen for illustrative purposes and might not follow
any particular code requirements.
The slab regions analyzed for in 5.2 are simplifed and are
intended to represent typical behavior patterns. Only gravity
loading e fects are evaluated. Lines o f symmetry have been
used where the slab continues into an adjacent region.
Notes on fgures and tables used in 5.2 —Sections
5. 2 . 1
5.2.2 through 5.2.7 each contain one fgure with four images.
The legend for each set o f four images is the same and
presented in Fig. 5.2.1 .
The tables presented in 5.2.2 through 5.2.7 contain the
same rows and columns presented for each o f the design
conditions. Each row reports for a design method, and the
meaning o f the columns is:
a) M represents the bending moment at the cross section.
This will be the same for all methods except for twist- free.
The value is normalized by the isotropic result.
b) T represents the torsion (due to both twist and eccen-
tric shear) at the cross section. This will be the same for all
methods except for twist- free. The value is normalized by
the isotropic M value.
12 DESIGN GUIDE FOR TWISTING MOMENTS IN SLABS (ACI 447R-1 8)

Fig. 5. 2—Key map for typical design conditions discussions.

Fig. 5. 2. 1 —Legends for @Seismicisolation

@Seismicisolation
fgures in 5. 2. 2 to 5. 2. 7.
 DESIGN GUIDE FOR TWISTING MOMENTS IN SLABS (ACI 447R-1 8)
c) Twist represents the integral o f the absolute value o f
| |
twist (∫| m xy|) along the cross section. This will be the same
for all methods except for twist- free. The value is normal-
ized by the isotropic M value.
d) De f represents maximum de f ection in the slab region.
This will be the same for all methods except for twist- free.
The value is normalized by the isotropic de f ection value.
e) Md represents design moment. The value is normalized
by the Wood and Armer design moment. Design moments
are not provided for the sandwich model, as it reduces the
design demand to a design force instead o f a design moment.
f) Ab represents the bottom rein forcement area. The value
is normalized by the maximum o f the Wood-Armer top and
bottom reinforcement areas.
g) At represents the top reinforcement area. The value is
normalized by the maximum o f the Wood-Armer top and
bottom reinforcement areas.
h) Av represents the requirement for shear reinforcement.
The only method that can require shear reinforcement is
torsion-as-shear.
i) Appl summarizes the applicability o f the method to the
confguration. The values used are:
i. Yes — The method is applicable, and within 5 percent o f
the most e f cient applicable method.
ii. Cons — The method is applicable, and within 20
percent o f the most e f cient applicable method.
iii. Ex. Cons — The method is applicable, although it is 20
percent or more conservative than the most e f cient appli-
cable method.
iv. Serv ? — The method is applicable for strength, but
there are serviceability concerns (this value only applies to
the twist- free analysis approach).
v. ? — The reinforcement calculated is not less than one o f
the applicable methods, but the assumed mechanisms o f the
method do not match the slab behavior, so the applicability
o f the method to similar confgurations is questionable.
vi. No — The method is not applicable.
The discussion in 5.2.2 through 5.2.7 re ferences the correct
quantity o f reinforcement. In the context o f those discussions,
the correct quantity o f reinforcement is the minimum quantity
o f reinforcement required by a method that is known to be
appropriate for the conditions being discussed.
5.2.2 Regular slab – interior panel at column —Figure
5.2.2 shows an interior panel o f a regular slab and the loca-
tion o f a critical design section at the support in a region with
large bending moments and signifcant twist.
Table 5.2.2 notes the normalized key design quantities and
results for the design section shown in Fig. 5.2.2.
While there is a signifcant twist peak near the support,
overall the integrated twist for this section is only 9 percent
o f the integrated moment. Due to symmetry o f the slab
confguration, torsion on the section is zero. Note the
following regarding design methods:
a) Ignore twist: Ignoring twist results in the correct quan-
tity o f reinforcement, although it is questionable, as it does
not guarantee a mechanism exists to resist the twist.
b) Wood and Armer: This approach is applicable, although
@Seismicisolation
@Seismicisolation
it results in a 9 percent excess reinforcement penalty.
13
c) Torsion as bending: Condition 1 discussed in 4.4.3
applies here; the twisting moments change sign along the
section, so torsion-as-bending is not expected to apply.
Treating torsion as bending results in the correct quantity o f
reinforcement. Due to symmetry, however, the total torsion
is zero; there fore, torsion is a poor proxy for integrated twist
and torsion-as-bending questionable.
d) Torsion-as-shear: Treating torsion as shear results in
the correct quantity o f reinforcement. However, it does not
guarantee a mechanism for resisting the twist that exists;
there fore, torsion-as-shear is questionable.
e) Sandwich model: Although this approach is appli-
cable, it results in a 32 percent excess reinforcement penalty.
This excess would be smaller in thicker slabs.
f) Twist-free analysis: This approach results in de f ec-
tion predictions that are 4 percent larger than a regular
(isotropic) analysis. This increase is very small, so there are
no concerns o f signifcant cracking required to achieve the
twist- free load path. Twist- free analysis is the most e f cient
applicable method for this case.
5.2.3 Regular slab – edge panel at column —Figure 5.2.3
shows an edge panel o f a regular slab and the location o f a
design section at the support in a region with large bending
moments and signifcant twist.
Table 5.2.3 notes the normalized key design quantities and
results for the design section shown in Fig. 5.2.3.
While there is a signifcant twist peak near the support,
overall the integrated twist for this section is only 1 4 percent
o f the integrated moment. Due to support at one end o f
the design section, torsion on this section is large—1 49
percent o f the integrated moment. This torsion is almost
entirely caused by eccentric shear and not twisting. Note the
following regarding design methods:
a) Ignore twist: Ignoring twist results in the correct quan-
tity o f reinforcement, although it is questionable as it does
not guarantee a mechanism exists to resist the twist.
b) Wood and Armer: This approach is applicable, although
it results in a 1 5 percent excess reinforcement penalty.
c) Torsion-as-bending: Neither Condition 1 or 2 discussed
in 4.4.3 applies here, so torsion-as-bending is expected to
apply. Treating torsion-as-bending results in a very conser-
vative quantity o f reinforcement, as torsion in this case is a
very conservative proxy for integrated twist.
d) Torsion-as-shear: Because the torsion is primarily
caused by eccentric shear, torsion-as-shear largely matches
the actual slab behavior and provides a much more appro-
priate shear design than the other approaches. However,
although this approach results in the correct quantity o f rein-
forcement, it does not provide a mechanism for resisting the
1 0 percent o f torsion caused by twist; there fore, torsion-as-
shear is considered questionable.
e) Sandwich model: This approach is applicable but
results in a 40 percent excess reinforcement penalty. This
excess would be smaller in thicker slabs.
f) Twist-free analysis: This approach results in de f ection
predictions that are 6 percent larger than a regular (isotropic)
analysis. The increase is small, so there are no concerns o f
signifcant cracking required to achieve the twist- free load
14 DESIGN GUIDE FOR TWISTING MOMENTS IN SLABS (ACI 447R-1 8)

Table 5.2.2—Analysis and design quantities for regular slab interior panel
Design
approach M T | twist| Def. Md Ab At Av Appl.

Ignore twist 1 .00 0.00 0.09 1 .00 0.92 0.00 0.92 No ?

Wood and
1 .00 0.00 0.09 1 .00 1 .00 0.00 1 .00 No Cons.
Armer
Torsion as
1 .00 0.00 0.09 1 .00 0.92 0.00 0.92 No ?
bending
Torsion as
1 .00 0.00 0.09 1 .00 0.92 0.00 0.92 No ?
shear
Sandwich
1 .00 0.00 0.09 1 .00 N/A 0.00 1 .21 No Ex. Cons
model
Twist- free
1 .00 0.00 0.00 1 .04 0.92 0.00 0.92 No Yes
analysis

Fig. 5. 2. 2—Regular slab, interior panel.

path. Twist- free analysis is the most e f cient applicable Table 5.2.4 notes the normalized key design quantities and
method for this case. results for the design section shown in Fig. 5.2.4.
5.2.4 Regular slab – corner panel at column —Figure There is a signifcant twist peak near the support and
5.2.4 shows a corner panel o f a regular slab and the location the bending moment is very small; overall, the integrated
o f a critical design section at the support in a region with twist for this section is 1 57 percent o f the integrated (small)
small bending moments and signifcant twist. moment. Due to the support at one end o f the design section,
@Seismicisolation
@Seismicisolation
 DESIGN GUIDE FOR TWISTING MOMENTS IN SLABS (ACI 447R-1 8) 15

Table 5.2.3—Analysis and design quantities for regular slab, edge panel
Design
approach M T | twist| Def. Md Ab At Av Appl

Ignore twist 1 .00 1 .49 0.1 4 1 .00 0.87 0.00 0.87 No ?

Wood and
1 .00 1 .49 0.1 4 1 .00 1 .00 0.00 1 .00 No Cons
Armer
Torsion as
1 .00 1 .49 0.1 4 1 .00 2.1 8 0.42 2.23 No Ex. Cons
bending
Torsion as
1 .00 1 .49 0.1 4 1 .00 0.87 0.00 0.87 No ?
shear
Sandwich
1 .00 1 .49 0.1 4 1 .00 N/a 0.00 1 .22 No Ex. Cons
model
Twist- free
1 .00 1 .43 0.00 1 .06 0.87 0.00 0.87 No Yes
analysis

Fig. 5. 2. 3—Regular slab, edge panel.

the torsion on the section is very large at 525 percent o f the
integrated moment. Although torsion is mostly caused by
eccentric shear, the twisting component is signifcant. Due
to the corner column being much less sti f than the slab, the
overall reinforcement demands at the corner column are
@Seismicisolation
@Seismicisolation
very small; minimum reinforcement provisions can govern
in this region. Note the following regarding design methods:
a) Ignore twist: Ignoring twist results in the correct quan-
tity o f reinforcement, although it is questionable as it does
not guarantee a mechanism exists to resist the twist.
16 DESIGN GUIDE FOR TWISTING MOMENTS IN SLABS (ACI 447R-1 8)

Table 5.2.4—Analysis and design quantities for regular slab, corner panel
Design
approach M T | twist| Def. Md Ab At Av Appl.

Ignore twist 1 .00 5.25 1 .57 1 .00 0.39 0.00 0.39 No ?

Wood and
1 .00 5.25 1 .57 1 .00 1 .0 0.22 1 .00 No Ex. Cons
Armer
Torsion as
1 .00 5.25 1 .57 1 .00 2.44 1 .67 2.47 No Ex. Cons
bending
Torsion as
1 .00 5.25 1 .57 1 .00 0.39 0.00 0.39 No ?
shear
Sandwich
1 .00 5.25 1 .57 1 .00 N/a 0.34 1 .29 No Ex. Cons
model
Twist- free
0.99 5.1 0 0.00 1 .1 0 0.38 0.00 0.38 No Yes
analysis

Fig. 5. 2. 4—Regular slab, corner panel.

b) Wood and Armer: This approach is applicable, although vative quantity o f rein forcement, as torsion in this case is a
it results in a 220 percent excess reinforcement penalty. very conservative proxy for integrated twist.
c) Torsion-as-bending: Neither Condition 1 or 2 discussed d) Torsion-as-shear: Because the torsion is primarily
in 4.4.3 applies here, so torsion-as-bending is expected to caused by eccentric shear, the torsion-as-shear approach
apply. Treating torsion as bending results in a very conser- largely matches the actual slab behavior and provides a much
@Seismicisolation
@Seismicisolation
more appropriate shear design than the other approaches.
 DESIGN GUIDE FOR TWISTING MOMENTS IN SLABS (ACI 447R-1 8) 17

Table 5.2.5—Analysis and design quantities for slab with misaligned supports, interior panel
Design
approach M T | twist| Def. Md Ab At Av Appl.

Ignore twist 1 .00 0.00 0.11 1 .00 0.90 0.00 0.89 No No

Wood and
1 .00 0.00 0.11 1 .00 1 .00 0.00 1 .00 No Yes
Armer
Torsion as
1 .00 0.00 0.11 1 .00 0.90 0.00 0.89 No No
bending
Torsion as
1 .00 0.00 0.11 1 .00 0.90 0.00 0.89 No No
shear
Sandwich
1 .00 0.00 0.11 1 .00 N/a 0.00 1 .29 No Ex. Cons
model
Twist- free
1 .21 0.00 0.00 1 .1 8 1 .09 0.00 1 .09 No Cons
analysis

However, it does not guarantee a mechanism for resisting
twist, which causes 30 percent o f the torsion; there fore,
torsion-as-shear is considered questionable.
e) Sandwich model: This approach is applicable but
results in a 340 percent excess reinforcement penalty.
f) Twist-free analysis: This approach results in de f ec-
tion predictions that are 1 0 percent larger than a regular
(isotropic) analysis. This increase is small, so there are no
concerns o f signifcant cracking to achieve the twist- free
load path. Twist- free analysis is the most e f cient applicable
method for this case.
5.2.5 Slab with unaligned supports – interior panel at
column —Figure 5.2.5 shows an interior panel o f an irregular
slab with supports staggered by hal f the span length and the
location o f a critical design section at the support in a region
with large bending moments and signifcant twist.
Table 5.2.5 notes the normalized key design quantities and
results for the design section shown in Fig. 5.2.5.
While there is a signifcant twist peak near the support,
overall the integrated twist for this section is only 11 percent
o f the integrated moment. Due to the symmetry o f the
slab confguration, torsion on the section is zero. Note the
following regarding design methods:
a) Ignore twist: Ignoring twist neither provides the correct
quantity o f reinforcement or a mechanism to resist the twisting
moments, so it is not applicable for this confguration.
b) Wood and Armer: The Wood and Armer method is the
most e f cient applicable method for this case.
c) Torsion-as-bending: Condition 1 discussed in 4.4.3
applies here; the twisting moments change sign along the
section, so torsion-as-bending is not expected to apply.
Treating torsion as bending does not provide the correct
amount o f reinforcement, as torsion in this case is a poor
proxy for integrated twist. Torsion-as-bending is not appli-
cable for this confguration.
d) Torsion-as-shear: Treating torsion as shear neither
provides the correct quantity o f reinforcement nor a mecha-
nism to resist the twisting moments, so it is not applicable
for this confguration.
e) Sandwich model: This approach is applicable but
results in a 21 percent excess reinforcement penalty. This
excess would be smaller in thicker slabs. @Seismicisolation
@Seismicisolation
f) Twist-free analysis: The twist- free analysis approach
for this slab results in de f ection predictions that are 1 8
percent larger than a regular (isotropic) analysis. This
increase is small, so there are no concerns o f signifcant
cracking to achieve the twist- free load path. However, twist-
free analysis results in a 9 percent rein forcement penalty.
5.2.6 Two-way cantilever slab —Figure 5.2.6 shows
a two-way cantilever and the location o f a critical design
section at the support in a region with very large bending
moments and signifcant twist.
Table 5.2.6 notes the normalized key design quantities and
results for the design section shown in Fig. 5.2.6.
While there is a signifcant twist peak near the support,
overall, the integrated twist for this section is only 4 percent
o f the integrated moment. Due to the length o f the canti-
lever (creating a large moment), the torsion on the section
is also small: 3 percent o f the integrated moment. Note the
following regarding design methods:
a) Ignore twist: Ignoring twist results in the correct quan-
tity o f reinforcement, although it is questionable as it does
not guarantee a mechanism exists to resist the twist.
b) Wood and Armer: This approach is also applicable,
although it results in a 4 percent excess reinforcement penalty.
c) Torsion-as-bending: Condition 1 discussed in 4.4.3
applies here; the twisting moments change sign along the
section, so torsion-as-bending is not expected to apply.
Treating torsion as bending results in a safe quantity o f rein-
forcement, but the total torsion is a little less than integrated
twist. There fore, torsion-as-bending is questionable.
d) Torsion-as-shear: Treating torsion as shear results in
the correct quantity o f reinforcement, but eccentric shear
only causes 25 percent o f total torsion. There fore, torsion-
as-shear is questionable.
e) Sandwich model: This approach is applicable but
results in a 24 percent excess reinforcement penalty. This
excess would be smaller in thicker slabs.
f) Twist-free analysis: Unexpectedly, the twist- free anal-
ysis for this slab results in a peak de f ection prediction that
is 1 5 percent less than a regular (isotropic) analysis. One
asks the question o f how can this be; how can removing a
slab sti fness reduce the de f ection? The answer is that the
average de f ection for the twist- free analysis is 1 2 percent
18 DESIGN GUIDE FOR TWISTING MOMENTS IN SLABS (ACI 447R-1 8)

@Seismicisolation
@Seismicisolation
Fig. 5. 2. 5—Slab with misaligned supports, interior panel.
 DESIGN GUIDE FOR TWISTING MOMENTS IN SLABS (ACI 447R-1 8) 19

Table 5.2.6—Analysis and design quantities for two-way cantilever

Design
approach M T | twist| Def. Md Ab At Av Appl.

Ignore twist 1 .00 0.03 0.04 1 .00 0.96 0.00 0.96 No ?

Wood and
1 .00 0.03 0.04 1 .00 1 .00 0.00 1 .00 No Yes
Armer
Torsion as
1 .00 0.03 0.04 1 .00 0.98 0.00 0.98 No ?
bending
Torsion as
1 .00 0.03 0.04 1 .00 0.96 0.00 0.96 No ?
shear
Sandwich
1 .00 0.03 0.04 1 .00 N/a 0.00 1 .1 9 No Ex. Cons
model
Twist- free
1 .00 0.03 0.00 0.85 0.96 0.00 0.96 No Yes
analysis

Fig. 5.2.6—Two-way cantilever.

more than for the regular (isotropic) analysis; however, the 5.2.7 Reentrant corner—Figure 5.2.7 shows a reentrant
load path is di ferent and there is less o f a peak de f ection at unsupported slab corner with the design section at the corner
the cantilever tip. Overall, there are no concerns o f signif- crossing a region o f moderate bending moments and signif-
cant cracking to achieve the twist- free load path. Twist- free cant twists.
@Seismicisolation
analysis is the most e f cient applicable method for this case.
@Seismicisolation
20 DESIGN GUIDE FOR TWISTING MOMENTS IN SLABS (ACI 447R-1 8)

Table 5.2.7—Analysis and design quantities for reentrant corner

Design
approach M T | twist| Def. Md Ab At Av Appl.

Ignore twist 1 .00 0.47 0.30 1 .00 0.77 0.76 0.00 No No

Wood and
1 .00 0.47 0.30 1 .00 1 .00 1 .00 0.00 No Yes
Armer
Torsion as
1 .00 0.47 0.30 1 .00 1 .1 3 1 .1 3 0.00 No Cons.
bending
Torsion as
1 .00 0.47 0.30 1 .00 0.77 0.76 0.00 No No
shear
Sandwich
1 .00 0.47 0.30 1 .00 N/a 1 .23 0.00 No Ex. Cons
model
Twist- free
0.87 0.09 0.00 1 .38 0.66 0.66 0.00 No Serv?
analysis

Fig. 5. 2. 7—Reentrant corner.

Table 5.2.7 notes the normalized key design quantities and this section is 30 percent o f the integrated moment. Torsion,
results for the design section shown in Fig. 5.2.7. 47 percent o f the integrated moment, is caused primarily by
There is a signifcant twist peak near the notch and the twist, but also has a signifcant component due to eccentric
@Seismicisolation
bending moment is moderate; overall, the integrated twist for shear. Note the following regarding design methods:
@Seismicisolation
 DESIGN GUIDE FOR TWISTING MOMENTS IN SLABS (ACI 447R-1 8)
Table 5.2.8—Summary of method applicability
Regular slab Regular slab
Design approach interior Regular slab edge
Ignore twist ? ?
Wood and Armer Cons. Cons
Torsion as bending ? Ex. Cons
Torsion as shear ? ?
Sandwich model Ex. Cons Ex. Cons
Twist- free analysis Yes Yes
a) Ignore twist: Ignoring twist neither provides the
correct quantity o f reinforcement or a mechanism to resist
the twisting moments, so it is considered not applicable for
this confguration.
b) Wood and Armer: The Wood and Armer method is the
most e f cient applicable method for this case.
c) Torsion-as-bending: Neither Condition 1 or 2 discussed
in 4.4.3 applies here, so torsion-as-bending is expected to
apply. Treating torsion as bending is applicable for this
case, although it requires 1 3 percent excess reinforcement.
Torsion is a conservative proxy for twist.
d) Torsion-as-shear: Treating torsion as shear neither
provides the correct quantity o f reinforcement nor a mecha-
nism to resist the twisting moments, so it is considered not
applicable for this confguration. While eccentric shear
causes signifcant torsion, twisting moments are still large.
e) Sandwich model: This approach is applicable but
results in a 23 percent excess reinforcement penalty. This
excess would be smaller in thicker slabs.
f) Twist-free analysis: The twist- free analysis approach
for this slab requires the least rein forcement but results
in de f ection predictions that are 38 percent larger than
a regular (isotropic) analysis. This increase is somewhat
large, so there are some concerns o f noticeable cracking to
achieve the twist- free load path. That makes twist- free anal-
ysis questionable for this confguration. Adding diagonal
reinforcement in the twist-critical reentrant corner might
be adequate to ease the serviceability concerns while still
requiring less reinforcement than the other methods. This,
however, requires engineering judgment beyond the scope
o f this guide.
5.2.8 Summary of typical design conditions —Table 5.2.8
displays a summary o f the applicability o f design methods
for each condition discussed in 5.2. Note that no method is
optimal for all cases, but that some conclusions can be drawn.
Although the Wood and Armer method is always appli-
cable, it can be more conservative than required.
The sandwich model is always applicable but is very
conservative for thin slabs such as the one investigated. For
signifcantly thicker slabs, the sandwich model will have
results similar to the Wood and Armer method. The strengths
o f the sandwich model are generally not relevant for typical
slab design.
Twist- free analysis, where applicable, is usually among
the most e f cient methods. Twist- free analysis is applicable
to most confgurations; if the twist- free de f ection is less
@Seismicisolation
@Seismicisolation
21
Unaligned Two-way
corner supports cantilever Reentrant corner
? No ? No
Ex. Cons Yes Yes Yes
Ex. Cons No ? Cons.
? No ? No
Ex. Cons Ex. Cons Ex. Cons Ex. Cons
Yes Cons Yes Serv?
than 25 percent greater than the isotropic de f ection, twist-
free analysis is generally applicable.
The other methods, while appropriate in some circum-
stances, are either inappropriate or excessively conservative
in too many cases to be used indiscriminately.
While not investigated in detail in this chapter, applying
the torsion-as-shear design approach to a twist- free analysis
is applicable whenever the twist- free analysis approach is
applicable. This combination approach has the potential for
improving vertical shear design while retaining the advan-
tages o f the twist- free analysis approach.
CHAPTER 6—TWO-WAY WALLS
6.1 —General considerations
The discussion o f twisting moments to this point has
focused on two-way slabs supported by discrete columns.
Twisting moments in two-way walls can also impact the out-
o f-plane design moments. Generally, walls are fundamen-
tally di ferent than slabs because: 1 ) out-o f-plane f exure in
the wall results from horizontal loads rather than vertical
loads; and 2) boundary conditions are typically continuous
rather than discrete. The former di ference is largely irrel-
evant to this discussion; however, the later di ference can
have an impact on the importance o f twisting moment on the
magnitude o f design moments.
Multiple combinations o f boundary and loading condi-
tions acting on walls can result in twisting moments. Many
infrastructure projects involve wall elements with contin-
uous boundary conditions along two adjacent edges and
out-o f-plane loading along one face, resulting in a behavior
similar to that shown in Fig. 3.1 . This geometry is common
in dam structures involving piers, as shown in Fig. 6.1 a.
Tanks o ften exhibit walls that are continuously supported on
three sides, as shown in Fig. 6.1 b. This geometry can also
create the conditions that result in high twisting moments
occurring at locations o f high bending moment.
6.2—Impact of twisting moment on walls
exhibiting two adjacent f xed edges
The pier shown in Fig. 6.1 a demonstrates a geometry
and boundary conditions common in water control struc-
tures such as dam spillways. In this section, the impacts
o f twisting moment will be explored for wall-like spillway
piers. Three di ferent aspect ratios, as shown in Fig. 6.2a, are
considered to quanti fy the impact o f geometry. Walls with a
22 DESIGN GUIDE FOR TWISTING MOMENTS IN SLABS (ACI 447R-1 8)

Fig. 6.1a—Dam pier and resulting defected shape of the twist-sensitive element.

Fig. 6.1b—Rectangular tank and resulting defected shape of the twist-sensitive element.

the results o f the examples in Chapter 5. However, only the

Fig. 6.2a—Twist-sensitive wall examples.
ratio close to 1 :1 generally exhibit two-way behavior. As the
aspect ratio increases, the wall starts behaving more like a
one-way element near the free edge.
The walls in this analysis are subject to a uni form lateral
load. To provide simpler behavior that can lead to better
understanding, Poisson’s ratio has been set to 0 to elimi-
@Seismicisolation
nate anticlastic bending. Results are produced similar to
@Seismicisolation moment is still largely controlled by moment at the fxed
following design approaches are considered: 1 ) ignore twist;
2) Wood and Armer; and 3) twist- free. For Chapter 5, the
assumed design strip is the width o f a slab cross section and
for Chapter 6, the point values are used to produce the design
moment. Results o f these analyses for the three di ferent
aspect ratios are shown in Fig. 6.2b. Location 1 exhibits
the largest twist and Location 2 exhibits the largest bending
moment. De f ection and required steel areas are determined
using the forces obtained in these analyses. The contours
shown in Fig. 6.2b are similar to the assumptions outlined in
Fig. 5.2.1 , with the exception that 0 is centered on the scale
allowing for negative values in this chapter.
The forces at Locations 1 and 2 changed signifcantly
between the 1 :1 and 2:1 aspect ratios for the twist analysis.
However, they change little between the 2:1 and 3:1 aspect
ratios.
The de f ections and required area o f steel for the design
moments are presented in Table 6.2 for the three walls and
three design approaches. Whereas the largest twist occurs
at Location 1 for each aspect ratio, the controlling design
 DESIGN GUIDE FOR TWISTING MOMENTS IN SLABS (ACI 447R-1 8) 23

Fig. 6.2b—Twist-sensitive wall study showing twisting moment trends.

boundary condition at Location 2. There fore, the values that
populate Table 6.2 are at Location 2.
The following is a summary o f each o f the design
approaches considering all three aspect ratios:
(1 ) Ignore twist : For the examples o f this study, ignoring
twist was only unconservative by approximately 5 percent.
However, other wall geometries and boundary conditions
can create locations o f higher twist and moment acting at
the same location. There fore, accounting for twist in the
element sti fness formulation and then ignoring the results
is not recommended. @Seismicisolation
@Seismicisolation
(2) Wood and Armer : This approach was applicable for
all aspect ratios. However, the resulting design for the ignore
twist method starts to approach the Wood and Armer method
for larger aspect ratios. Twisting sti fness enables a more
f
e cient load distribution because the twist is largely concen-
trated at locations o f lower moment for uni formly rein forced
walls. There fore, this approach is generally recommended as
f
a safe and e cient design method.
(3) Twist-free analysis : This approach is excessively
conservative until you approach the 3:1 ratio, where values
o f twist are small and the moment increases less where
24 DESIGN GUIDE FOR TWISTING MOMENTS IN SLABS (ACI 447R-1 8)

Table 6.2—Analysis and design quantities for various aspect ratios

1:1 ratio design
approach M Twist Def. Md As Applicability
Ignore twist 1 .0 0.08 1 .00 0.92 0.92 No
Wood and Armer 1 .0 0.08 1 .00 1 .00 1 .00 Yes
Twist- free analysis 1 .6 0.00 2.26 1 .44 1 .43 Ex. Cons

2:1 ratio design

approach M Twist Def. Md As Applicability
Ignore twist 1 .0 0.02 1 .00 0.98 0.98 No
Wood and Armer 1 .0 0.02 1 .00 1 .00 1 .00 Yes
Twist- free analysis 1 .4 0.00 1 .52 1 .36 1 .41 Ex. Cons

3:1 ratio design

approach M Twist Def. Md As Applicability
Ignore twist 1 .0 0.005 1 .00 1 .00 0.995 Yes
Wood and Armer 1 .0 0.005 1 .00 1 .00 1 .000 Yes
Twist- free analysis 1 .1 0.000 1 .1 2 1 .09 1 .087 Cons

twisting sti fness does not contribute to redistribution o f the
force. Although it could be used for large aspect ratios, it
is recommended to avoid this method. The large de f ection
multiple for the 1 :1 ratio wall and, to a lesser extent, the 3:1
wall, is indicative o f a large alteration o f the load path. This
alteration in load path can lead to service cracking.
For the 1 :1 aspect ratio, behavior observed is close to a
pure twist (Fig. 3.1 ) in which the f ow o f force is most e f -
cient at a 45-degree angle to the edges and assumed rein-
forcement direction. By eliminating twisting sti fness, the
f ow o f force is confned to the far more ine f cient load path,
which is orthogonal to the edges and, basically, creating two
cantilevers. For larger aspect ratios, the most e f cient f ow
o f force is naturally more closely aligned to the edges and
assumed reinforcement direction. In these situations, the
ignore twist analysis starts to produce designs similar to the
Wood and Armer design method.
In conclusion, the Wood and Armer design approach is
generally recommended for all wall analysis with two adja-
cent fxed boundary conditions and orthogonally placed
reinforcement. Twist can be ignored for larger aspect ratios,
but this is generally not recommended.
CHAPTER 7—SHELL STRUCTURES
7.1 —General considerations
While shells, which are curved thin structures, behave
very di ferently than f at slabs, they do exhibit twisting
moments similar to f at slabs. Shell behavior is typically
dominated by membrane actions, although f exural behavior
(including twisting moments) can be signifcant at boundary
conditions, near loading points, and at openings and other
discontinuities. These twisting moments should be consid-
ered in design, as required in Section 6.1 .1 o f ACI 31 8.2-1 4.
This chapter investigates twisting moment considerations in
@Seismicisolation
@Seismicisolation
hemispherical domes, which are commonly shell-shaped.
The conclusions may or may not be relevant for other shell
shapes, in which twisting may be more or less important in
the load paths.
7.2—Typical bulk material storage hemisphere
Hemispherical shells are commonly used as an e f cient
structural form for bulk granular material storage. This
chapter considers the e fects o f twisting moments in the
design o f a typical bulk storage hemisphere. The hemisphere
studied has a thickness to radius ratio o f 0.01 .
7.2.1 Finite element modeling —The fnite element model
o f the hemisphere that was studied uses quadrilateral shell
elements (Fig. 7.2.1 a). Isotropic elements are used to
evaluate the ignore twist, and the Wood and Armer design
approaches. Orthotropic elements are used to evaluate the
twist- free design approach. The analyses performed are all
linear-elastic.
One signifcant di ference between shells and slabs is the
varied orientation o f the shell local axes from location-to-
location; there fore, it is not parallel to any global axes. To
account for this, a clear axes convention needs to be estab-
lished. For the hemispheric shell case investigated, the local
axes are defned in Fig. 7.2.1 b. Except in the apex region,
reinforcement is placed parallel to the local axes shown in
the fgure.
The base o f the hemisphere is constrained against vertical
translation (local x-axis direction) and against translation
tangential to the radius from the center vertical axis o f the
hemisphere (local y-axis direction). The base is allowed to
translate radially (local z-axis direction). Figure 7.2.1 c illus-
trates constraints at the base o f the dome near the opening.
The fgure shows the unde formed and de formed shape o f
the model. Arrows at the base o f the hemisphere depict the
constraint against translation used in the model. The cylin-
drical coordinate system is provided to clari fy the boundary
conditions. The model is constrained against translation in
 DESIGN GUIDE FOR TWISTING MOMENTS IN SLABS (ACI 447R-1 8) 25

Fig. 7.2.1a—Finite element model (geometry and mesh).

the θ- and Z-directions, as shown in the fgure. The fgure
illustrates how the model is only allowed to translate radially.
While the region o f the opening is shown, the constraints are
typical all around the base. The ring foundation continues
under the opening to provide the constraint in the θ direction.

7.3—Typical loading conditions

Three common loadings are considered for the hemi-
sphere: 1 ) sel f-weight; 2) pseudo-hydrostatic; and 3) a point
load at the apex.
7.3.1 Self-weight loading —The sel f-weight o f the hemi-
sphere due to gravity is a signifcant load for this type o f
structure. The e fects o f this loading are considered in 7.4.1
and 7.4.2.
7.3.2 Pseudo-hydrostatic loading —Bulk granular mate-
rials stored in the hemisphere apply a load to the shell similar
to a hydrostatic load. This loading is di ferent from a hydro-
static pressure in that it acts horizontally and not perpendicu-
larly to the shell. Figure 7.3.2 illustrates the load considered.
The e fects o f this loading are considered in 7.4.1 and 7.4.2.
7.3.3 Apex point loading —Point loading at the apex o f Fig. 7.2.1b—Local shell axes.
the hemisphere is typical for bulk material storage facilities.
This load comes from the equipment to fll the hemisphere.
The e fects o f this loading are considered in 7.4.3.

7.4—Typical design regions

Three regions o f the hemisphere are considered: 1 ) the
apex; 2) an opening at the base; and 3) a typical region at the
base. Figure 7.4 identifes the three regions.
For each region the ignore-twist, Wood and Armer, and
twist- free analysis approaches are evaluated. All results are
normalized by the Wood and Armer results; bending and
twisting analysis results are normalized together so the rela-
tive magnitude o f twist and bending is apparent. The reported
de f ections are the maximum de f ection in any direction. Fig. 7.2.1c—Hemisphere base boundary conditions and
Note that scales are intentionally omitted from the plots. The cylindrical coordinate system.
importance o f the fgure is to show qualitative behaviors. numerical rounding in the FEA analysis. The section cut o f
Re fer to Fig. 5.2.1 for scale information. the area analyzed is shown in the fgure. The length o f the
7.4.1 Typical base region —At the typical base section
analyzed strip is approximately 20 percent o f the radius o f
away from the opening, the results are observed to be similar curvature o f the shell. The analyzed strip location was chosen
to those o f a pure hemisphere with no discontinuities. As at the location o f highest Mx moment. This is di ferent than
the loadings are all radially symmetric, no signifcant twist previous chapters because the twisting moment is very small.
is observed. Figure 7.4.1 shows this region subjected to the The calculation o f the values in Table 7.4.1 is calculated
sel f-weight and hydrostatic loadings; the contours shown in the same way as in previous chapters. Note the following
@Seismicisolation
Fig. 7.4.1 (b) are all nearly zero and are primarily a result o f
@Seismicisolation regarding design methods:
26 DESIGN GUIDE FOR TWISTING MOMENTS IN SLABS (ACI 447R-1 8)
Fig. 7. 3. 2—Pseudo-hydrostatic loading.
a) Ignore twist: Ignoring twist results in the correct quan-
tity o f reinforcement, although it is questionable as it does
not guarantee a mechanism exists to resist the twist.
b) Wood and Armer: This approach is applicable,
although it results in a 4 percent excess reinforcement
penalty.
c) Twist-free analysis: This approach results in de f ec-
tions that are similar to a regular (isotropic) analysis. There
are no concerns o f signifcant cracking required to achieve
the twist- free load path. Twist- free analysis is the most e f -
cient applicable method for this case.
7.4.2 Opening base region subject to self-weight
loading —Figure 7.4.2 shows this region subjected to the
sel f-weight loading. The section cut o f the area analyzed
is shown in the fgure. The length o f the analyzed strip is
approximately 5 percent o f the radius o f curvature o f the
shell. The analyzed strip location was chosen at the loca-
tion o f highest twisting moment.
@Seismicisolation
@Seismicisolation penalty.
Fig. 7. 4—Design regions.
Table 7.4.1 —Comparisons for typical base section
Design
approach M |twist| Md Defection Applicable
Ignore twist 1 .00 0.03 0.97 1 .00 ?
Wood and
1 .00 0.03 1 .00 1 .00 Cons
Armer
Twist- free
0.99 0.00 0.96 1 .00 Yes
analysis
The area around the opening o f the curved shell shows
a large amount o f twisting moment. This is illustrated in
Table 7.4.2.
a) Ignore twist: This approach is not applicable in this
area, as the twisting moment is signifcant and the isotropic
load path requires twisting moments. The ignore twist design
moment is only 44 percent o f the most e f cient applicable
design method and, hence, could lead to unsa fe designs.
b) Wood and Armer: This method is excessively conser-
vative because it results in a 47 percent excess reinforcement
penalty.
c) Twist-free analysis: This approach results in de f ec-
tion predictions that are 1 5 percent larger than a regular
(isotropic) analysis. This increase is relatively small, so there
are no concerns o f signifcant cracking required to achieve
the twist- free load path. Twist- free analysis is the most e f -
cient applicable method for this case.
7.4.3 Opening base region subject to pseudo-hydro-
static loading —Figure 7.4.3 shows this region subjected
to pseudo-hydrostatic loading. The section cut o f the area
analyzed is shown in the fgure. The length o f the analyzed
strip is approximately 5 percent o f the radius o f curvature o f
the shell. The analyzed strip location was chosen at the loca-
tion o f highest twisting moment.
The area around the opening o f the curved shell shows a
noticeable amount o f twisting moment. This is illustrated in
Table 7.4.3.
a) Ignore twist: This approach is not applicable in these
areas, as the twisting moment is signifcant and the isotropic
load path requires twisting moments. The ignore twist design
moment is only 58 percent o f the most e f cient applicable
design moment and, there fore, could lead to unsa fe designs.
b) Wood and Armer: This method is excessively conser-
vative because it results in a 75 percent excess reinforcement
 DESIGN GUIDE FOR TWISTING MOMENTS IN SLABS (ACI 447R-1 8) 27

Fig. 7.4.1—Typical section of hemisphere at base.

Fig. 7.4.2—Base at opening subject to self-weight.

Table 7.4.2—Comparisons for base at opening Table 7.4.3—Comparisons for base at opening
subject to self-weight subject to pseudo-hydrostatic loading
Design Design
approach M |twist| Md Defection Applicable approach M |twist| Md Defection Applicable

Ignore twist 1 .00 2.33 0.30 1 .000 No Ignore twist 1 .00 2.02 0.33 1 .00 No
Wood and Wood and
1 .00 2.33 1 .000 1 .000 Ex. cons 1 .00 2.02 1 .000 1 .00 Ex. cons
Armer Armer
Twist- free Twist- free
2.26 0.000 0.68 1 .1 5 Yes 1 .72 0.000 0.57 1 .11 Yes
analysis analysis

c) Twist-free analysis: This approach results in de f ec-

tion predictions that are 11 percent larger than a regular
(isotropic) analysis. This increase is relatively small, so there
are no concerns o f signifcant cracking required to achieve
the twist- free load path. Twist- free analysis is the most e f -
cient applicable method for this case. @Seismicisolation
@Seismicisolation
7.4.4 Apex region —Due to the radial symmetry o f the
loadings and the structure in the region o f the apex (that is,
the base opening is far away), the behavior in this region is
largely radially symmetric. There fore, there are no signif-
cant twisting moments about the meridional and hoop axes.
28 DESIGN GUIDE FOR TWISTING MOMENTS IN SLABS (ACI 447R-1 8)

Fig. 7. 4. 3—Base at opening subject to pseudo-hydrostatic loading.

Table 7.4.4—Analysis and design comparisons for

apex loading
Design
approach M |twist| Md Defection Applicable

Ignore twist 1 .000 0..27 0..79 1 .000 ?

Wood and
1 .000 0.27 1 .00 1 .000 Cons
Armer
Twist- free
1 .07 0.00 0.84 1 .03 Yes
analysis

in this region. Note that moments, normal and twisting, are

Fig. 7. 4. 4—Apex of hemisphere under point loading.
The reinforcement at the apex, however, cannot be placed
in a radially symmetric pattern due to constructability
constraints. Rein forcement in this region is placed along
orthogonal axes that are approximately in a horizontal plane.
When the moments about the radial axes are trans formed to
the reinforcement axes, signifcant twisting moments arise.
Figure 7.4.4 shows the apex region. Only the apex point
@Seismicisolation
@Seismicisolation
loading is considered, as that is the most signifcant loading
theoretically infnite under a point load.
The section cut o f the area analyzed is shown in the
fgure. The length o f the analyzed strip is approximately 1 0
percent o f the radius o f curvature o f the shell. The analyzed
strip is slightly o f center o f the apex. This is so that the
twisting moment is not exactly zero along the length o f the
analysis strip.
Table 7.4.4 notes the normalized key design quantities and
results for the apex region shown in Fig. 7.4.4.
a) Ignore twist: This approach is not applicable in these
areas, as the twisting moment is signifcant and the isotropic
load path requires twisting moments. The ignore twist design
moment is only 94 percent o f the most e f cient applicable
design moment and, there fore, could lead to unsa fe designs.
b) Wood and Armer: This approach is applicable,
although it results in a 1 9 percent excess reinforcement
penalty.
c) Twist-free analysis: This approach results in de f ec-
tion predictions that are 3 percent larger than a regular
(isotropic) analysis. This increase is very small, so there are
no concerns o f signifcant cracking required to achieve the
twist- free load path. Twist- free analysis is the most e f cient
applicable method for this case.
 DESIGN GUIDE FOR TWISTING MOMENTS IN SLABS (ACI 447R-1 8)
Table 7.4.5—Summary of method applicability
Design approach Typical Opening with in-plane loading
Ignore twist ? No
Wood and Armer Cons Ex. cons
Twist- free analysis Yes Yes
7.4.5 Summary —A hemispherical shell is modeled with
several di ferent loading scenarios. The response o f the
shell under a point load at the apex, near an opening with
a pseudo-hydrostatic force, and in a typical section with
pseudo-hydrostatic forces is investigated. These loads are
additional to the load caused by the sel f-weight o f the struc-
ture. Generally, the twisting moments in this shell are more
infuenced by geometric discontinuities than by loading
discontinuities. The following is a written summary o f the
behavior o f the shell for twisting moments, which is also
summarized in Table 7.4.5.
Ignore twist: The analyses found that the ignore twist
method was not a valid methodology for all the investi-
gated loading scenarios. The stresses in the shell away from
openings do not exhibit large twisting moments because
the membrane action dominates the behavior o f the shell in
these areas. However, it does not guarantee a mechanism
exists to resist the twisting moments.
Wood and Armer: The Wood and Armer method is an
appropriate analysis method for shells in all locations.
However, it also results in excess reinforcement in areas o f
geometric or loading discontinuities.
Twist-free analysis: The twist- free analysis is an appro-
priate analysis method. The de f ections are very similar to
the de f ections produced when considering twist. This small
di ference in de f ections create no concern o f signifcant
cracking required to achieve the twist- free load path. Di fer-
ences between the shells and slabs are due to the membrane
capabilities o f the shells. The twist- free analysis results in
di ferent membrane forces in the shell that need to be consid-
ered in design.
CHAPTER 8—REFERENCES
Committee documents are listed frst by document number
and year o f publication followed by authored documents
listed alphabetically.
ACI 31 8-1 4—Building Code Requirements for Structural
Concrete and Commentary
ACI 31 8.2-1 4—Building Code Requirements for Concrete
Thin Shells and Commentary
CSA Group
CSA-A23.3-04—Design o f Concrete Structures
European Committee for Standardization
EN 1 992-1 -1 :2004—Eurocode 2: Design o f Concrete
Structures, General Rules and Rules for Buildings
@Seismicisolation
@Seismicisolation
29
Opening with out-of-plane loading Point load at apex
No ?
Ex. cons Cons
Yes Yes
Authored documents
Armer, G. S. T., 1 968, “The Rein forcement o f Slabs in
Accordance with a Pre-Determined Field o f Moments,”
by R. H. Wood, discussion o f re ference 1 2.1 5, Concrete
(London) , V. 2, No. 8, Aug., pp. 31 9-320.
Brøndum-Nielsen, T., 1 985, “Optimization o f Rein-
forcement in Shells, Folded Plates, Walls, and Slabs,” ACI
Journal Proceedings , V. 82, No. 3, May-June, pp. 304-309.
Burgoyne, C., 2004, “Are Structures Being Repaired
Unnecessarily?” The Structural Engineer , V. 82, No. 1 , Jan.,
pp. 22-26.
Comité Euro-International du Béton fb , 1 993, “CEB-FIP
Model Code 1 990,” Lausanne, Switzerland, 460 pp.
Comité Euro-International du Béton fb , 2008, “Practi-
tioners’ Guide to Finite Element Modeling o f Rein forced
Concrete Structures,” Bulletin 45, Lausanne, Switzerland,
344 pp.
Deaton, J. B., 2005, “A Finite Element Approach to Rein-
forced Concrete Slab Design,” MS thesis, Georgia Institute
o f Technology, Atlanta, GA, 1 70 pp.
Den Hartog, J. P., 201 4, Advanced Strength of Materials ,
Courier Corporation, Dover Publications Inc., New York,
379 pp.
Kennedy, G., and Goodchild, C., 2003, Practical Yield
Line Design , British Cement Association, 1 71 pp.
Lourenço, P. B., and Figueiras, J. A., 1 995, “Solution
for the Design o f Rein forced Concrete Plates and Shells,”
Journal of Structural Engineering , V. 1 21 , No. 5, May, pp.
81 5-823. doi: 1 0.1 061 /(ASCE)0733-9445(1 995)1 21 :5(81 5)
Marti, P., 1 990, “Design o f Concrete Slabs for Transverse
Shear,” ACI Structural Journal, V. 87, No. 2, Mar.-Apr., pp.
1 80-1 90.
May, I. M., and Lodi, S. H., 2005, “Defciencies o f the
Normal Moment Yield Criterion for RC Slabs,” Proceedings
of the Institution of Civil Engineers—Structures and Build-
ings , V. 1 58, No. 6, Dec., pp. 371 -380.
Park, R., and Gamble, W. L., 2000, Reinforced Concrete
Slabs , second edition, John Wiley and Sons, New York,
71 6 pp.
Shin, M.; Bommer, A.; Deaton, J. B.; and Alemdar, B. N.,
2009, “Twisting Moments in Two-Way Slabs,” Concrete
International , V. 31 , No. 7, July, pp. 35-40.
Timoshenko, S., and Woinowski-Krieger, S., 1 959, Theory
of Plates and Shells , second edition, McGraw-Hill Book Co.
Wood, R. H., 1 968, “The Rein forcement o f Slabs in Accor-
dance with a Pre-Determined Field o f Moments,” Concrete
(London) , V. 2, No. 2, pp. 69-76.
@Seismicisolation
@Seismicisolation
@Seismicisolation
@Seismicisolation
@Seismicisolation
@Seismicisolation
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, 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 to discuss and share concrete knowledge and fellowship.

38800 Country Club Drive

Farmington Hills, MI 48331
Phone: +1.248.848.3700
Fax: +1.248.848.3701
www.concrete.org

@Seismicisolation
@Seismicisolation
38800 Country Club Drive
Farming ton H ills, M I 48331 USA
+1 . 248.848. 3700
www.concrete.org

The American Concrete Institute (ACI) is a leading authority and resource

worldwide for the development and distribution of consensus-based
sta n d a rd s a n d tech n ica l resou rces, ed u ca tion a l prog ra m s, a n d certifca tion s

for individuals and organizations involved in concrete design, construction,

and materials, who share a commitment to pursuing the best use of concrete.

Individuals interested in the activities of ACI are encouraged to explore the

ACI website for membership opportunities, committee activities, and a wide
variety of concrete resources. As a volunteer member-driven organization,
ACI invites partnerships and welcomes all concrete professionals who wish to
be part of a respected, connected, social group that provides an opportunity
for professional growth, networking and enjoyment.

@Seismicisolation
@Seismicisolation 9 781641 950107