Professional Documents
Culture Documents
SP344 04 AlternativeDesignProceduresforTorsion
SP344 04 AlternativeDesignProceduresforTorsion
net/publication/344811217
CITATIONS READS
4 4,236
4 authors, including:
Evan Bentz
University of Toronto
120 PUBLICATIONS 4,611 CITATIONS
SEE PROFILE
All content following this page was uploaded by Edvard P.G. Bruun on 22 October 2020.
Allan Kuan, Edvard P.G. Bruun, Evan C. Bentz, and Michael P. Collins
Synopsis: Although the torsion design procedures in ACI 318-19 are simple and broadly applicable, the resulting
designs tend to be overly conservative. To address this, clause 9.5.4.6 in ACI 318-19 permits the use of an alternative
design procedure when designing members with an aspect ratio ≥ 3 for torsion, provided that the alternative procedure
has been shown to agree with the results of comprehensive tests. This paper evaluates and compares the torsion design
procedures in CSA A23.3:19 and the PCI Design Handbook 8th edition with those in ACI 318-19. Each of the three
methods are found to show good agreement with 282 tests found in the literature. A comparison of the three concludes
that the designs obtained using the ACI method generally require the most reinforcement. More economical designs
for members subjected to relatively low and high torques can be obtained by using the PCI and CSA methods
respectively. A design example of a spandrel beam using the three methods is presented, and then further conclusions
are stated to guide practicing engineers on the relative strengths and weaknesses of each procedure.
Keywords: reinforced concrete; prestressed concrete; torsion; design codes; shear; moment
1
Alternative Design Procedures for Torsion in ACI 318-19: A Comparative Study
ACI Student Member Allan Kuan is a Ph.D. candidate in the Department of Civil & Mineral Engineering at the
University of Toronto. He received his bachelor’s degree in Engineering Science from the University of Toronto in
2015. His research interests include developing design procedures and nonlinear analysis tools for reinforced and
prestressed concrete members subject to shear and torsion.
ACI Young Professional Member Edvard P.G. Bruun, P.Eng, is a PhD student in the Civil and Environmental
Engineering Department at Princeton. He received his bachelor and master’s degree from the University of Toronto,
after which he worked as a structural engineer at Arup. He is the 2019 recipient of the ACI’s Daniel W. Falconer
Memorial Fellowship for graduate studies. His research interests include reinforced concrete mechanics, finite element
modelling, generative design, and robotic fabrication.
ACI Member Evan C. Bentz, FACI, is a Professor of civil engineering at the University of Toronto. He received his
bachelor’s degree from the University of Waterloo, Waterloo, ON, Canada, in 1994, and his PhD from the
University of Toronto in 2000. He is former Chair of ACI Committee 365, Service Life Modeling, and is a member
of Joint ACI-ASCE Committee 445, Shear and Torsion.
ACI Honorary Member Michael P. Collins is a University Professor of structural engineering at the University of
Toronto. He is a member and former Chair of Joint ACI-ASCE Committee 445, Shear and Torsion. His research
interests include developing rational but simple shear design procedures for both reinforced and prestressed concrete
structures, failure investigations, and evaluating and strengthening concrete structures in distress.
INTRODUCTION
Designing reinforced and prestressed concrete members to safely resist torsion in combination with other actions is a
complex task which involves carefully proportioning the cross section and internal reinforcement to resist the resulting
circulating shear stresses. Although historically neglected as a secondary effect, the introduction of extensive design
provisions for torsion into the 1971 ACI code1 necessitated engineers learning many complex methods in order to
consider the effects of torsion when designing concrete structures. By contrast, the current provisions in ACI 318-19
are simple and broadly applicable which has made designing for torsion comparable in difficulty to designing for axial
load, moment and shear2. However, prioritizing simplicity has come at the cost of constructability and efficiency, with
designs for torsion developed using the ACI code generally being overly conservative3. One way that ACI 318-19 has
addressed this drawback has been through inclusion of clauses which permit the use of alternative design procedures
if certain criteria are met, such as the following4:
“9.5.4.6 For solid sections with an aspect ratio h/bt ≥ 3, it shall be permitted to use an alternative design procedure,
provided the adequacy of the procedure has been shown by analysis and substantial agreement with results of
comprehensive tests. The minimum reinforcement requirements [of ACI 318-19] need not be satisfied, but the detailing
requirements [of ACI 318-19] apply.”
Although this clause appears to encourage using tailor-made approaches to design specific types of members, such as
slender precast members in parking structures, it opens up the possibility of using any experimentally validated
approach to design members that satisfy the aspect ratio requirement. Some examples of design methods which can
be used to satisfy this clause include several found in the academic literature, such as those by Rahal5-6 and Collins
and Mitchell7, as well as other codes and standards, such as the Canadian concrete design code CSA A23.3:198,
AASHTO LRFD-89 and the PCI Design Handbook 8th edition3.
This purpose of this paper is to inform practicing engineers of possible safe and economical alternatives to the standard
ACI torsion design procedures and explain how to use them. The alternatives considered in this study are CSA
A23.3:19 and the PCI Design Handbook 8th edition as they are major “codes” used throughout North America for the
design of reinforced and prestressed concrete structures. An overview of the alternatives is presented first, where the
relevant aspects of designing for torsion are explained for each method. This is followed by a validation study of each
method against 282 experiments found in the literature. Different aspects of each approach are then compared with
each other followed by an illustrative design example showcasing their application.
2
Kuan et. al.
Although the ACI 318-19, CSA A23.3:19 and PCI Design Handbook 8th edition approaches differ in terms of their
theoretical basis and overall application, each contain the following five main components:
1. A threshold torsion limit, Tth, below which the effects of torsion may be neglected
2. Requirements for minimum reinforcement
3. An upper limit on the permissible shear and torsion which can be resisted by the member
4. Equations to calculate the nominal torsional strength of the member
5. Various requirements or equations which describe how torsion interacts with moment and shear
Note that each of the equations are presented in US customary units. Instances where a specific unit is required (i.e.
stresses in psi) are noted when describing the relevant parameter. Conversion factors are provided for SI units as well.
𝐴.)- 𝑓-)
𝑇"# = 𝜙𝜆'𝑓)* + 0 11 + (1)
𝑝)- 4𝜆'𝑓)*
Where pcp is the outside perimeter of the cross section, Acp is the area enclosed by pcp, fc’ is the specified concrete
cylinder compressive strength in psi, fpc is the compressive stress in the concrete at the centroid of the cross section
after accounting for all prestress losses in psi, λ is a factor to account for low density concrete taken here as 1.0 and ϕ
is a reduction factor for shear and torsion equal to 0.75. If MPa units are used instead of psi for fc’ and fpc in Eq. (1)
then each instance of '𝑓)* should be multiplied by 0.083.
2. Minimum reinforcement requirements – The minimum quantity of closed transverse reinforcement which must
be provided if torsion is considered is:
𝑏
⎧0.75'𝑓)* L
𝐴9 + 2𝐴" ⎪ 𝑓M,"
8 < = max (2)
𝑠 =>? ⎨ 50 L𝑏
⎪ 𝑓M,"
⎩
Where Av is the area of shear reinforcement, At is the area of one leg of closed transverse torsion reinforcement, s is
the spacing of the transverse reinforcement, fy,t is the yield strength of the transverse reinforcement in psi and bw is the
minimum effective web width.
The minimum quantity of longitudinal torsional reinforcement which must be provided is calculated using Eq. (3).
This reinforcement must be uniformly distributed around the inside perimeter of the transverse reinforcement and be
spaced less than 305 mm (12 in) apart.
3
Alternative Design Procedures for Torsion in ACI 318-19: A Comparative Study
*
⎧ 5'𝑓) 𝐴)- − 8𝐴" < 𝑝 𝑓M,"
⎪ 𝑓M,O 𝑠 # 𝑓M,O
𝐴O,=>? = min (3)
⎨5'𝑓)* 𝐴)- 25𝑏L 𝑓M,"
⎪ −+ 0 𝑝#
⎩ 𝑓M,O 𝑓M," 𝑓M,O
Where ph is the perimeter of the centerline of closed transverse reinforcement, Al is the area of longitudinal torsion
reinforcement and fy,l is the yield strength of the longitudinal reinforcement in psi. If MPa units are used for fc’, fy,l and
fy,t instead of psi, then the 0.75 becomes 0.0623 and the 50 becomes 0.345 in Eq. (2). Similarly, the 5 becomes 0.415
and the 25 becomes 0.172 in Eq. (3).
3. Upper limit on permissible shear and torsion – To control cracking and avoid crushing of the cross section prior
to yielding of the reinforcement, the magnitude of the applied shear force Vu and torsion Tu is limited by the following
equation for solid sections:
.
𝑉 . 𝑇𝑝 𝑉
18 U < + + U #. 0 ≤ 𝜙 8 ) + 8'𝑓)* < (4)
𝑏L 𝑑 1.7𝐴W# 𝑏L 𝑑
Where Vc is the shear strength attributed to the concrete, bw is the minimum effective web width, d is the effective
depth of the member and Aoh is the area enclosed by the centerline of the closed torsion reinforcement. For non-
prestressed members containing at least minimum transverse reinforcement and not subjected to axial load, Vc can be
taken as 2'𝑓)* 𝑏L 𝑑 and hence the upper limit can be rewritten as:
.
𝑉 . 𝑇𝑝
18 U < + + U #. 0 ≤ 𝜙10'𝑓)* (5)
𝑏L 𝑑 1.7𝐴W#
If MPa units are used instead of psi for fc’ in Eq. (4) and (5), then the coefficients 8 and 10 become 0.664 and 0.83
respectively. Definitions for Acp, pcp, Aoh and ph are shown in Figure 1.
4. Nominal torsional strength – The nominal torsional resistance Tn shall exceed the factored demand Tu and is
calculated as follows:
𝜙𝑇? ≥ 𝑇U (6)
𝐴" 𝑓M,"
⎧2𝐴W cot 𝜃
𝑠
𝑇? = min (7)
⎨2𝐴 𝐴O 𝑓M,O tan 𝜃
⎩ W 𝑝#
𝐴W = 0.85𝐴W# (8)
Where Ao is the area enclosed by the shear flow path and θ is the angle of diagonal compressive stresses in the concrete.
Although ACI 318-19 permits the use of θ ranging from 30° to 60°, it suggests taking θ = 45° for non-prestressed
members and 37.5° for prestressed members in lieu of a more detailed analysis.
5. Interaction with moment and shear – Under combined moment and torsion, a member may fail due to yielding
of the longitudinal flexural reinforcement, since this steel is required to carry the tension arising from both the torsion
and bending moment. When analyzing a member subjected to moment and torsion which fails in this manner, the
torsional resistance may be calculated as:
4
Kuan et. al.
Where As’ and As are the areas of longitudinal reinforcement in the flexural compression and flexural tension zones
respectively, jd is the flexural lever arm and ω is the ratio Mu/Tu. It should also be noted that under combined bending
and torsion, the flexural compression counteracts the tension caused by torsion in the flexural compression zone.
Because of this, the ACI code permits the quantity of reinforcement required in that region to be reduced by an amount
equal to Mu/(0.9d × fy,l).
Under combined shear and torsion, the transverse steel provides both shear capacity Vs and torsional strength Tn. From
a design perspective, Eq. (10) can be used to determine the required quantity of transverse reinforcement required to
carry the applied shear and torsion:
𝑉U − 𝜙𝑉) 𝑇U
𝐴9 + 2𝐴" 𝑑 + 𝐴 tan 𝜃
W
≥ (10)
𝑠 𝜙𝑓M,"
Although the design provisions in ACI 318-19 are formulated in a way which simplifies the process of proportioning
the torsional reinforcement, using them to evaluate the torsional strength of an existing member design subjected to
both shear and torsion is not a straightforward task. To address this, the following equation has been derived to directly
calculate the torsional strength of a member containing two-leg stirrups subjected to combined shear and torsion.
Where ξ is the ratio of the shear force to the torsion, Vu/Tu. When combinations of shear, moment and torsion are
present, the torsional resistance is taken as the minimum of Tn calculated from Eq. (9) and (11).
𝐴.)- 𝜙- 𝑓-)
𝑇"# = 1.14𝜆𝜙) '𝑓)* + 0 11 + (12)
𝑝)- 4.58𝜆𝜙) '𝑓)*
In the above equation, ϕc is a reduction factor for concrete and ϕp is a reduction factor for prestressing steel, taken as
ϕc = 0.65 and ϕp = 0.90 respectively. If MPa units are used when using Eq. (12), then 0.095 should be used in lieu of
the 1.14 and 0.38 instead of 4.58.
5
Alternative Design Procedures for Torsion in ACI 318-19: A Comparative Study
2. Minimum reinforcement requirements – When torsion needs to be considered, the minimum amount of
transverse reinforcement which needs to be provided is:
𝐴9 + 2𝐴" 𝑏L
8 < = 0.723'𝑓)* (13)
𝑠 =>? 𝑓M,"
If MPa units are used instead of psi for fc’ and fy,t in Eq. (13), then the 0.723 becomes 0.06. Unlike ACI 318-19, CSA
A23.3:19 does not require a minimum amount of longitudinal torsional reinforcement. Instead, the longitudinal
torsional reinforcement is explicitly proportioned for longitudinal tension using Eq. (19) and (20).
3. Upper limit on permissible shear and torsion – The CSA A23.3:19 upper limit to avoid crushing for members
with solid cross sections is given as:
. .
𝑉U 𝑇U 𝑝#
18 < ++ 0 ≤ 0.25𝜙) 𝑓)* (14)
0.9𝑏L 𝑑 1.7𝐴.W#
4. Nominal torsional strength – In the CSA code, the nominal torsional resistance Tn must exceed the demand Tu
as shown in Eq. (15). Ao is calculated in the same manner as AC 318-19 using Eq. (8) and ϕs is a reduction factor for
reinforcing steel taken as ϕs = 0.85.
𝜙` 𝑇? ≥ 𝑇U (15)
𝐴" 𝑓M,"
𝑇? = 2𝐴W cot 𝜃 (16)
𝑠
The angle of diagonal compressive stresses in the concrete, θ, is related to the longitudinal strain at the mid-height of
the member, εx, and is calculated as follows:
𝑀U . 0.9𝑇U 𝑝# .
+ 1d𝑉k − 𝜙𝑉- e + l
0.9𝑑 2𝐴W m + 0.5𝑁U − 𝐴- 𝑓-W
𝜀i = ≤ 0.003 (18)
2d𝐸` 𝐴` + 𝐸- 𝐴- e
Where Mu is the applied bending moment, Vu is the applied shear force, Vp is the component of the unfactored effective
prestressing force in the direction of the applied shear, Nu is the axial force, Ap is the area of prestressed reinforcement,
fpo is the stress in the prestressing steel when the strain in the surround concrete is equal to zero and Es and Ep are the
Young’s moduli of conventional and prestressed reinforcement respectively. In certain cases, an alternative value of
εx should be used instead of the value calculated by Eq. (18). For example, if there is enough axial tension in the
member to crack the flexural compression zone, the coefficient of 2 in the denominator is taken as 1 instead.
Furthermore, εx is permitted to be taken as a larger value than the value calculated by Eq. (18) if desired, although εx
must still be less than or equal to 0.003. Being able to use alternative means of calculating εx in lieu of Eq. (18) can be
particularly important when designing for torsion and is explained in greater detail in subsequent sections in this paper.
5. Interaction with moment and shear – The CSA A23.3:19 method explicitly considers if yielding of the longitudinal
reinforcement in the flexural tension and flexural compression zones govern the strength of the member. This allows
it to capture the interaction between torsion, axial force and moment in a relatively straightforward manner by using
the following equations:
𝑀U . 0.45𝑇U 𝑝# .
𝐹O" = + 0.5𝑁U + 1d𝑉U − 0.5𝜙` 𝑉` − 𝜙- 𝑉- e + 8 < cot 𝜃 ≤ 𝜙` 𝐴` 𝑓M,O + 𝜙- 𝐴- 𝑓-M (19)
0.9𝑑 2𝐴W
6
Kuan et. al.
𝑀U . 0.45𝑇U 𝑝# .
𝐹O) = − + 0.5𝑁U + 1d𝑉U − 0.5𝜙` 𝑉` − 𝜙- 𝑉- e + 8 < cot 𝜃 ≤ 𝜙` 𝐴*` 𝑓M,O + 𝜙- 𝐴*- 𝑓-M (20)
0.9𝑑 2𝐴W
Where Flt and Flc are the longitudinal tension forces which must be carried by the flexural tension and flexural
compression reinforcement respectively and Vs is the shear strength provided by shear reinforcement, calculated as:
0.9𝐴9 𝑓M," 𝑑
𝑉` = cot 𝜃 (21)
𝑠
When designing for combined shear and torsion, the transverse steel is assumed to be used in the same manner as in
the ACI code. The quantity of transverse reinforcement required to resist the shear and torsion is calculated using Eq.
(22), which differs slightly from ACI 318-19 approach due to the definition of Vn in the CSA code:
𝑉U − 𝜙) 𝑉) − 𝜙- 𝑉- 𝑇U
𝐴9 + 2𝐴" 8 0.9𝑑 + 𝐴 < tan 𝜃
W
≥ (22)
𝑠 𝜙` 𝑓M,"
Vc is calculated in the Canadian code using Eq. (23) for members which contain at least minimum transverse
reinforcement. If MPa units are used instead of psi for fc’ in this equation, the coefficient 4.32 becomes 0.36.
4.32
𝑉) = 𝜆 '𝑓 * 𝑏 𝑑 (23)
1 + 1500𝜀i ) L
When analyzing a member containing stirrups with two legs subjected to combined shear and torsion, the torsional
resistance may be calculated as:
In situations where shear, torsion and moment are all present, the torsional resistance is first checked using Eq. (24).
If Flt and Flc exceed the capacity of either the top or bottom steel respectively, then Tn is limited by yielding of the
longitudinal reinforcement and must be reduced until both Eq. (19) and Eq. (20) are satisfied.
Where x and y are the shorter and longer dimensions respectively of a rectangular component of the cross section, as
shown in Figure 1, ϕ is the same reduction factor used in ACI 318-19 and γ is a prestress factor, calculated as:
𝑓-)
𝛾 = 11 + 10 (26)
𝑓)*
If MPa units are used instead of psi in Eq. 25, the 0.5 becomes 0.0664
2. Minimum reinforcement requirements – When torsion needs to be considered, the minimum amount of
transverse reinforcement required is determined using the following equation:
7
Alternative Design Procedures for Torsion in ACI 318-19: A Comparative Study
If MPa units are used instead of psi for fy,t, the 50 and 200 in Eq. (27) should be 0.345 and 1.38 respectively.
Although the PCI method does not explicitly contain a requirement for the minimum amount of longitudinal
reinforcement, it can be obtained by using Eq. (37), which calculates how much longitudinal steel is needed given the
quantity of provided transverse reinforcement, together with Eq. (27).
3. Upper limit on permissible shear and torsion – The upper limit in the PCI method to avoid crushing prior to
yielding of the reinforcement follows a different formulation than ACI 318-19 and CSA A23.3:19 and is shown below:
1
𝐶𝛾'𝑓)* ∑ 𝑥 . 𝑦
𝑇U ≤ 3 (28)
𝐶𝛾𝑉U .
11 + l
30𝐶" 𝑇U m
C is a factor accounting for prestress and Ct is a factor which relates the geometric properties used for shear and
torsion:
𝑓)-
𝐶 = 12 − 10 * (29)
𝑓)
𝑏L 𝑑
𝐶" = (30)
∑ 𝑥.𝑦
Eq. (28) should be multiplied by 0.083 if MPa units are used instead of psi for fc’.
4. Nominal torsional strength – Unlike ACI 318-19 and CSA A23.3:19, the torsional strength in the PCI method,
Tn, is the sum of the torsional strength contributions from both the concrete, Tc, and the transverse reinforcement, Ts.
This strength must exceed the factored torsion in the same manner as ACI 318-19, which is shown in Eq. (6).
𝑇? = 𝑇) + 𝑇` (31)
𝑇)*
𝑇) = (32)
.
11 + 8𝑇) ′ 𝑉U <
𝑉) ′ 𝑇U
𝑉) ′
𝑉) = (33)
.
11 + 8 𝑉) ′ 𝑇U <
𝑇) ′ 𝑉U
In Eq. (32) and (33), Vc’ is the concrete shear strength in the absence of torsion as calculated using ACI 318-19 and
Tc’ is the pure torsional strength attributed to the concrete, calculated as:
The torsional strength provided by the transverse torsion reinforcement, Ts, is calculated as:
8
Kuan et. al.
Where x1 and y1 are the short and long dimensions of the closed torsion reinforcement respectively, shown in Figure
1, and α is a factor determined by the aspect ratio of the member:
0.33𝑦y
𝛼 = 0.66 + ≤ 1.5 (36)
𝑥y
Eq. (34) should be multiplied by 0.083 if MPa units are used instead of psi for fc’. Furthermore, when using Eq. (36)
for solid sections which satisfy the requirements of clause 9.5.4.6 in ACI 318-19, α will typically be taken as 1.5.
5. Interaction with moment and shear – Because the PCI method is intended to be used to proportion the transverse
reinforcement after the design of the flexural reinforcement has been completed, it does not contain an explicit
interaction between moment and torsion. However, the following equations are used to determine the quantity of
longitudinal torsion reinforcement, Al:
𝑥y + 𝑦y
⎧ 2𝐴" 8 <
⎪ 𝑠
𝐴O = max (37)
⎨z400𝑥𝑠 z 𝑇U { − 2𝐴" { 8𝑥y + 𝑦i <
⎪ 𝑓M," 𝑉 𝑠
𝑇U + 3𝐶U
⎩ "
If the second equation in Eq. (37) governs Al, then the term 2At shall be taken to be less than:
50𝑏L 𝑠 𝜎 200𝑏L 𝑠
2𝐴" ≤ 81 + 12 * < ≤ (38)
𝑓M," 𝑓) 𝑓M,"
If MPa units are used instead of psi, the 50, 200 and 400 in Eq. (37) and (38) become 0.345, 1.38 and 2.76 respectively.
When designing a member carrying shear and torsion, Vc and Tc must first be calculated using Eq. (32) and Eq. (33)
respectively. The quantity of transverse reinforcement needed to resist the applied shear and torsion can then be
determined using Eq. (39).
𝑉U − 𝜙𝑉) 𝑇 − 𝜙𝑇
𝐴9 + 2𝐴" + 2 U𝛼𝑥 𝑦 )
𝑑 y y
≥ (39)
𝑠 𝜙𝑓M,"
From an analysis perspective, the torsional resistance of a member containing stirrups with two legs subjected to
combined shear and torsion can be calculated as:
9
Alternative Design Procedures for Torsion in ACI 318-19: A Comparative Study
Lampert and Thürlimann19 (6 beams), McMullen and Rangan20 (10 beams), McMullen and Warwaruk21 (3 beams),
Mitchell and Collins22 (6 beams), Mardukhi and Collins23 (1 beam), Pandit and Warwaruk24 (4 beams), Rasmussen
and Baker25 (12 beams) and Wafa et al.26 (14 beams). These tests comprehensively examine many variables which
affect torsional behaviour, such as aspect ratio, quantity of reinforcement, distribution of longitudinal reinforcement,
concrete compressive strength, clear cover, presence of voids and influence of prestressing. The summary of results
shown in Table 1 provide statistics about the ratio of the experimentally observed ultimate torque to the predicted
failure torque for each of the three methods.
ACI 318-19 tends to be conservative, with none of the 157 experiments failing at a torque of less than 80% of the
predicted strength and the average Ttest/Tpred being the highest of the three methods at 1.28. The predictions of strength
are even more conservative for prestressed members, with an average Ttest/Tpred of 1.66 for the 19 prestressed concrete
beams analyzed, compared to an average Ttest/Tpred of 1.23 for the 138 non-prestressed beams analyzed. All modes of
failure, such as yielding by the transverse reinforcement, yielding of the longitudinal reinforcement or crushing of the
cross section are generally well-predicted, albeit conservatively.
The predictions made using CSA A23.3:19 tend to be less conservative and have a larger spread than those made
using ACI 318-19. Unlike ACI 318-19, prestressed members were predicted with comparable accuracy as the
reinforced members (Ttest/Tpred = 1.29 for the prestressed beams compared to 1.14 for the non-prestressed beams).
Although CSA predicts failures caused by yielding of the reinforcement well, members which failed due to crushing
of the cross section were poorly predicted by CSA’s upper limit, shown in Eq. (14). If the 32 beams which failed due
to crushing during testing are removed from the set, the average Ttest/Tpred and coefficient of variation become 1.25
and 20.4% respectively, which is better than the ACI predictions and comparable to the PCI method. The issues
attributed to CSA’s upper limit are discussed further in subsequent sections.
The PCI 8th edition method shows excellent results overall, having the smallest coefficient of variation and working
equally well for both reinforced and prestressed members which failed due to crushing or yielding.
Overall each of the methods are comparable in performance, generally work well and tend to be yield conservative
predictions, particularly for beams which fail by yielding of the transverse and/or longitudinal reinforcement.
10
Kuan et. al.
Compared to the predictions for beams tested in pure torsion, there is significantly more spread in the results for each
of the methods used, particularly for lower M/T ratios. As with the pure torsion experiments, ACI 318-19 is the most
conservative with only 2 of the 121 beams being having a lower strength than 80% of the predicted value, though this
is offset by the fact that the largest observed Test/Pred is almost four. Based on the summary statistics and plots, CSA
A23.3:19 and the PCI 8th edition methods are comparable, with the CSA method generally being slightly more
conservative and consistent, with a higher average Test/Pred and lower coefficient of variation.
To evaluate the capabilities of each method further, a more detailed study of the torsion-bending interactions predicted
by each method was done by examining two series of beams tested over a variety of M/T ratios by Mardukhi23 and
Onsongo31.
11
Alternative Design Procedures for Torsion in ACI 318-19: A Comparative Study
Each method predicts the experimental behaviour reasonably well, with ACI 318-19 being the most conservative,
CSA A23.3:19 being somewhat less conservative and the PCI 8th edition method being accurate in for low M/T ratios
but unconservative for high M/T ratios. For low M/T ratios, ACI predicts that yielding of the stirrups at θ = 45° governs
Tn, resulting in a flat plateau before a subsequent reduction in torsional strength once yielding of the longitudinal steel
begins to govern the failure. The CSA interaction is more accurate and less overly conservative than the ACI prediction
because θ varies according to M/T and is generally less than 45°. This has the effect of predicting a higher Tn when
failure is governed by yielding of the stirrups, gradually reducing Tn as M/T increases due to the changing θ, and more
severely reducing Tn once yielding of the longitudinal reinforcement occurs. The PCI method predicts the strength of
TB4, TB1 and TB2 very well, but becomes unconservative in situations of high moment and low torsion, like the case
12
Kuan et. al.
of TB3. This is because of the Tc term, which is unaffected by the presence of the bending moment, was calculated to
be substantially higher than the failure torsion of TB3, (Tc = 46.5 kNm (412 kip-in) vs Tult = 27.8 kNm (246 kip-in)).
As seen in Figure 9, the torsion-bending interaction predictions by the ACI code and PCI method are comparable and
conservative over the full range of M/T ratios. Both incorrectly predict that the strength of the member is independent
of moment for low M/T ratios – in fact, the conservative upper limit in ACI 318-19 predicts that the member fails due
to crushing of its cross section, whereas PCI predicts that yielding of the transverse reinforcement governs the ultimate
strength. However, for large M/T ratios, both the ACI and PCI methods correctly predict that failure is governed by
yielding of the flexural tension steel.
The CSA A23.3:19 torsion-bending interaction is more accurate than ACI and PCI in predicting the strength of the
experiments and also correctly captures a wide variety of failure modes, which results in an unusual shape. Because
it was known that the flexural compression side of the section would be carrying a large tension because of the torsion,
the longitudinal strain at mid-height εx was calculated as the average of the strains in the top and bottom reinforcement
(εx,top and εx,bot respectively) instead of with Eq. (18):
𝑀U . 0.9𝑇U 𝑝# .
− + 1d𝑉k − 𝜙- 𝑉- e + l + 0.5𝑁U − 𝐴*- 𝑓-W
0.9𝑑 2𝐴W m
𝜀i,"W- = ≥0 (41)
𝐸` 𝐴`* + 𝐸- 𝐴-*
13
Alternative Design Procedures for Torsion in ACI 318-19: A Comparative Study
.
𝑀U
1d𝑉k − 𝜙- 𝑉- e + l0.9𝑇U 𝑝# m + 0.5𝑁U − 𝐴- 𝑓-W
.
+
0.9𝑑 2𝐴W
𝜀i,}W" = ≥0 (42)
𝐸` 𝐴` + 𝐸- 𝐴-
𝜀i,"W- + 𝜀i,}W"
𝜀i = ≤ 0.003 (43)
2
When using the CSA code to calculate the torsion-bending interaction for the experiments, four distinct regions of
behaviour were observed. In the first region, which corresponds to very low M/T ratios, failure was governed by
yielding of the top steel when checking Flc using Eq. (20). The second region, corresponding to the flat portion of the
graph, is where failure was governed by yielding of the stirrups but εx was restricted to 0.003 and θ to 50°. For
increasing moment, there is a corresponding increase in torsional strength as the presence of moment reduces εx, which
lowers θ and hence increases Tn. Finally, the sharp decrease in torsional strength for high M/T ratios is due to failure
being governed by yielding of the bottom steel when checking Flt using Eq. (19).
As with pure torsion and combined bending and torsion, the ACI code is the most conservative of the three methods.
The CSA code is less conservative when torsion is considered because θ is calculated to be lower than the 45° assumed
by ACI 318-19. Both interactions have a flat segment for low V/T ratios which is caused by the lack of a Tc. In this
region, Vc exceeds the shear demand and the strength is instead governed by the stirrups being fully utilized to resist
the torsion. The PCI method produces a comparable interaction to the ACI and CSA codes, although there is no cut-
off for low V/T ratios due to the presence of a Tc which smoothly interacts with Vc. Although the CSA method produces
the best predictions, all three methods give similar results and generally underestimate the strength of the four tested
beams over the full V/T spectrum.
14
Kuan et. al.
Although all three methods are applicable to prestressed members, the majority of this discussion will focus on non-
prestressed members.
1. Threshold torsion
The ACI and CSA codes have identical formulations for calculating the threshold torsion, though if reduction factors
are not considered, the CSA threshold is about 15% higher than the ACI threshold to account for the fact that the ACI
reduction factor ϕ is 15% higher than the CSA equivalent ϕc. On the other hand, the PCI method uses a different
formulation for the threshold torsion which is generally higher than the ACI and CSA values depending on the aspect
ratio of the member. Figure 11, which shows the normalized Tth with resistance factors included and plotted over a
range of aspect ratios for each method, illustrates that for small aspect ratios, the threshold torsion permitted by PCI
can be up to double the ACI limit. However, for aspect ratios of 3 or greater, a necessary condition for using an
alternative method, the PCI threshold torsion is up to 33% higher than the ACI and CSA values.
All three methods agree that prestressing increases Tth, although the factor to describe the influence of prestressing on
the threshold torsion is different in the PCI method compared to the ACI and CSA approaches. However, the effect it
has on Tth is virtually identical as the factors in the ACI and CSA codes.
15
Alternative Design Procedures for Torsion in ACI 318-19: A Comparative Study
For prestressed members, the minimum reinforcement requirements of the PCI method increase linearly as a function
of the fpc/fc’. As the minimum reinforcement requirements in the ACI and CSA codes are unrelated to the amount of
prestress, this means that designs obtained using the PCI method may require more transverse reinforcement compared
to designs obtained using the ACI and CSA codes for heavily prestressed members. For example, up to four times
more reinforcement would be required by the PCI method if fpc/fc’ is equal to or greater than 0.25.
Figure 14 shows the relationship between the upper limit on torsion and the concrete compressive strength fc’ for all
three methods when applied to members loaded in pure torsion. Also shown in the figure are the normalized failure
torques of 32 beams in the literature which failed prior to yielding of the reinforcement. Although there is disagreement
16
Kuan et. al.
between each of the methods and the experimental data, the ACI equation is the most conservative of the three. It can
be seen that the CSA upper limit equation is unconservative, especially when predicting the crushing of high strength
concrete members. The poor limit on crushing is the reason why many of the over-reinforced tests in pure torsion were
overpredicted in terms of strength by CSA A23.3:19, suggesting that improvements to the equation are needed. It is
unclear if the linear relation with fc’ is incorrect, as recent work done by Proestos et al.37 have shown the CSA upper
limit equation to be superior than the ACI code for predicting the brittle failure of members loaded in shear in the
absence of torsion.
Also shown in Figure 14 is the theoretical upper limit obtained by using a recently developed sectional analysis
program capable of predicting the full torque-twist response of members loaded in pure torsion and shown to give
excellent results38. Its prediction, based on an the “average” aspect ratio of the tests of 1.5, lies in between the ACI,
CSA and PCI methods, suggesting that effect of increasing fc’ on the upper limit on torsion should be a more generous
than the ACI and PCI methods, but less than the Canadian code. However, the large scatter in the experimental data
and substantial disagreement with each of the methods presented here suggest that more research is needed to obtain
a better method of predicting crushing.
For prestressed members, the ACI 318-19 and the PCI 8th edition upper limits increase as a function of the prestress
provided, fcp. In contrast, the CSA upper limit is not affected by the effects of prestressing.
Figure 13 – Shear-Torsion upper limit interaction Figure 14 – Effect of fc’ on upper limit in pure torsion
Note: 1 MPa = 145 psi
Figure 15 illustrates how the pure torsion strength of a typical member is affected by increasing the quantity of
transverse reinforcement from 0 to 1% while the quantity of longitudinal reinforcement is kept constant at 1%. The
predicted torsional response from the sectional analysis program is shown in addition to the three design methods to
act as a theoretical reference. As expected, for a given value of At/s, CSA predicts a larger benefit of adding stirrups
than the ACI code since θ was calculated to be around 30° instead of the 45° used by ACI. PCI predicts the smallest
benefit of adding transverse reinforcement, although this drawback is offset by the presence of Tc. The ACI and PCI
methods predict a continuous increase in strength as At/s is increased until Tn is limited due to crushing at normalized
torques of 4.22 and 5.40 MPa respectively (612 and 783 psi). On the other hand, the CSA predictions made using Eq.
(18) to calculate εx predicts that the benefit of adding stirrups stops once yielding of the longitudinal steel occurs at a
17
Alternative Design Procedures for Torsion in ACI 318-19: A Comparative Study
normalized torque of 3.31 MPa (480 psi). The prediction from the sectional analysis procedure reflects aspects of all
three methods, with the addition of stirrups predicted to be very beneficial for low quantities of At/s (predicted by
CSA), but with failure being governed by crushing for higher values of At/s (predicted by ACI and PCI).
Although this example may suggest that designs obtained using CSA A23.3:19 require more longitudinal
reinforcement in order to fully utilize the stirrups, the CSA provisions can be used in an alternative manner to better
engage the provided transverse reinforcement if yielding of the longitudinal steel is limiting Tn. The red dashed line
in Figure 15 shows the influence of adding transverse reinforcement if a larger value of εx is used instead of the one
calculated using Eq. (18). By increasing εx and therefore θ, the demand on the longitudinal reinforcement is reduced,
allowing the stirrups to reach their yield strength prior to yielding of the longitudinal reinforcement. This permits Tn
to further increase as more transverse steel is provided until Tn is cut off by the upper limit to avoid crushing.
Based on the results of this example, the CSA and PCI methods predict a higher torsional strength Tn for a given
quantity of transverse reinforcement compared to ACI 318-19. Both methods also show more agreement with the
theoretical sectional analysis results than to the ACI code, which is both conservative and underestimates the gain in
strength by adding more transverse torsional reinforcement. Although the CSA and PCI methods predict comparable
values of Tn for larger values of At/s, for smaller design torsions, designs obtained using the PCI method may require
less reinforcement due to the presence of Tc, especially if the member is prestressed.
Figure 15 – Influence of transverse steel on torsional strength. Note 1 mm = 0.0394 in and 1 MPa = 145 psi.
Figure 16, which shows the predicted behaviour by ACI 318-19, CSA A23.3:19, the PCI 8th edition method and the
sectional analysis, shows how the pure torsion strength of a member increases as the quantity of longitudinal
reinforcement varies from 0 to 2% while the quantity of transverse reinforcement is kept constant at 0.33%. All of the
methods agree that increasing the quantity of longitudinal reinforcement increases the torsional strength until failure
of the member is governed by yielding of the stirrups instead of by yielding of the longitudinal reinforcement. Since
the ACI method uses a fixed value of θ when calculating Tn, increasing Al does not increase the overall strength once
18
Kuan et. al.
yielding of the stirrups governs. Similarly, for the PCI method, providing additional longitudinal reinforcement does
not provide any further benefits once the required threshold limit has been reached. Although the benefit of increasing
Al is more pronounced in the ACI code compared to the PCI method, Tn obtained by the PCI method is generally
higher due to Tc.
Increasing the quantity of longitudinal steel using the CSA code and Eq. (18) provides relatively small gains in
torsional strength for small values of Al compared to the ACI and PCI methods. This is because the longitudinal strain
εx is inversely proportionate to the quantity of longitudinal steel, and so for low values of Al, θ is greater than the 45°
used by ACI 318-19. However, increasing the Al gradually lowers θ, which permit a much higher value of Tn to be
obtained for beams that are heavily reinforced in the longitudinal direction because the stirrups become more efficient.
If a value of εx which is greater than the one obtained from Eq. (18) is used, the dashed red line is obtained, which
shows that increasing Al provides gains in strength comparable to the PCI method and greater than the ACI code.
Based on the results of this example, the CSA code predicts the highest increase in Tn as Al was increased, which was
76% higher than Tn predicted by the ACI code, 21% higher than Tn predicted by the PCI method and closest to the
prediction made by the sectional analysis. In addition, it also predicts the largest incremental gain in strength obtained
by increasing Al compared to the ACI and PCI methods. A caveat to this conclusion is the PCI method may still require
the lowest amount of reinforcement for low values of Tu, especially for prestressed members which have a large Tc.
Figure 16 – Influence of longitudinal steel on torsional strength. Note 1 mm = 0.0394 in and 1 MPa = 145 psi.
DESIGN EXAMPLE
To illustrate the different designs which can be produced using the three methods, an example design of a non-
prestressed spandrel beam is described in this section. The spandrel beam, shown in Figure 17, is simply supported
with a clear span of 8 m (26.3 ft) and carries a factored line load of 75 kN/m (5.175 kips/ft). To provide a fair
comparison between all three procedures, the concrete compressive strength was chosen so that the cross-section
dimensions would satisfy the limit on crushing for all three design methods. The loading was selected to simulate
realistic conditions, have a factored torque exceeding the threshold torsion for each of the methods and lead to designs
whose reinforcement arrangement was not governed by minimum reinforcement requirements. Hence the main
purpose of this example is to determine and compare how much longitudinal and transverse reinforcement is necessary
to carry the applied moment, torsion and shear using the ACI, CSA and PCI methods.
19
Alternative Design Procedures for Torsion in ACI 318-19: A Comparative Study
If 2-leg stirrups are used, then Av = 200 mm2 (0.31 in2) and hence s ≥ 200/1.674 = 119 mm (4.69 in). Selecting s = 115
mm (4.53 in), ϕTn can then be checked using Eq. (11):
The required quantity of longitudinal torsional reinforcement Al is then calculated by rearranging the second
expression in Eq. (7):
𝑇? 𝑝# 55.5 × 2600
𝐴O = cot 𝜃 = cot 45° = 1632 mm. (2.53 in. )
2𝐴W 𝑓M,O 2 × 110,500 × 400
This longitudinal reinforcement must be distributed uniformly around the perimeter of the member, resulting in 816
mm2 (1.26 in2) each in the top and bottom halves of the beam. The amount in the flexural compression zone can be
reduced by Mu/(0.9 × d × fy,l) = 732 mm2 (1.13 in2), meaning that 2-10M bars can be used as top steel (As’ = 200 mm2
or 0.31 in2). In the flexural tension region, the required area of steel is Mu/(0.9 × d × fy,l) + Al/2 = 1548 mm2 (2.40 in2)
which is less than what is already provided. Additional bars need to be provided so that the spacing of longitudinal
torsional reinforcement around the inside perimeter of the stirrups does not exceed 305 mm (12 in).
20
Kuan et. al.
The longitudinal strain at mid-height εx must first be calculated using Eq. (18). Before using the full equation, it is
instructive to compare the relative magnitudes of the terms in the numerator:
Since the term with Mu is smaller than the term containing Vu and Tu, the CSA code predicts the flexural compression
zone to crack, and hence Eq. (18) should be used with a 1 instead of a 2 in the denominator. εx is then calculated as:
In a similar fashion to the ACI design method, Vc is then calculated using Eq. (23), and then the required quantity of
reinforcement is found using Eq. (22).
0.36
𝑉) = √45 × 152 × 975 = 93.5 kN (21.0 kips)
1 + 1500 × 1.885 × 10ˆ‰
The selected spacing is then 200/1.571 = 127 mm (5.00 in). The factored torsional strength, ϕsTn, using that spacing
should then be checked using Eq. (24):
127 × 0.65 × 93.5 + 0.85 × 1.8 × 100 × 400 × 975 cot 42.4°
𝑇? = 110,500 = 47.6 kNm (35.1 kip-ft)
127(0.9 × 975 + 110,500 × 0.006163)
After the torsional resistance attributed to the transverse reinforcement is calculated, the longitudinal reinforcement
must be checked to ensure that their tensile capacity is not exceeded. Performing this check requires first calculating
Vs with Eq. (21) and then using Eq. (19) and (20) to calculate Flt and Flc:
𝜙) 0.65
𝑉` = 𝜉𝑇? − 𝑉) = 0.006163 × 47.6 − × 93.5 = 232 kN (52.2 kips)
𝜙` 0.85
21
Alternative Design Procedures for Torsion in ACI 318-19: A Comparative Study
As the capacity of the bottom steel is equal to 0.85 × 2100 × 400 = 714 kN (161 kips), no additional longitudinal
reinforcement is required at the bottom of the beam. In the flexural compression zone, reinforcement is required as
the net force is positive, but Flc is small enough that 2-10M bars (As’ = 200 mm2 or 0.31 in2) will suffice.
The second design is obtained by repeating the procedure with As = 4000 mm2 (6.20 in2) which reduces the longitudinal
strain by approximately half to become εx = 0.990 × 10-3. This flattens θ to 35.9° which provides a substantial increase
in the required stirrup spacing from 127 mm to 170 mm. The smaller angle however increases the demand on the
longitudinal reinforcement, which in turn increases the amount of reinforcement in the flexural compression zone
from 2-10M to 2-15M bars (As’ = 400 mm2 or 0.62 in2).
Note that CSA A23.3:19 does not require additional longitudinal bars to be placed around the perimeter to reduce the
spacing of longitudinal torsional reinforcement to be less than 305 mm (12 in) like the ACI and PCI methods do.
𝑇*) = 0.8 × 0.083 × √45 × 32.95 × 10Š = 14.68 kNm (10.82 kip-ft)
14.68
𝑇) = = 12.78 kNm (9.43 kip-ft)
.
‡1 + l14.68 × 257 m
165 40.4
165
𝑉) = = 81.3 kN (18.29 kips)
.
‡1 + l 165 40.4
14.68 × 257 m
The quantity of transverse reinforcement to carry the shear and torsion is then calculated using Eq. (39):
The corresponding maximum stirrup spacing is 200/2.15 = 93 mm (3.66 in). Using a spacing of 90 mm instead
provides a factored torsional strength ϕTn which is calculated using Eq. (40):
The quantity of longitudinal steel was found to be governed by the first equation in Eq. (37), requiring Al to be:
92 + 1004
𝐴O = 2 × 100 × 8 < = 2436 mm. (3.78 in. )
90
On the flexural compression side, the required area of steel is 2436/2 - Mu/(0.9 × d × fy,l) = 486 mm2 (0.75 in2), which
can be provided by at least five 10M bars distributed in this region (As’ = 500 mm2 (0.76 in2). On the flexural tension
side, the required area of steel is 2436/2 + Mu/(0.9 × d × fy,l) = 1950 mm2 (3.02 in2), which is already present to carry
22
Kuan et. al.
the moment at midspan. Like the ACI design, additional bars must be placed around the perimeter so that the spacing
of longitudinal torsion reinforcement does not exceed 305 mm (12 in).
Table 3 summarizes the designs obtained by each method, which are all heavily reinforced in the transverse direction.
The two CSA designs both required the least amount of transverse reinforcement, especially Design #2 which
contained more longitudinal reinforcement. The design obtained using the PCI method requires the most amount of
transverse reinforcement because of two reasons. The first is due to the large ratio of Tu/Tc, which greatly reduced Vc
and hence increased the amount of reinforcement needed to carry the shear force through Vs. The second reason is
because the PCI formulation of Ts, related to αx1y1, uses the stirrups less efficiently than the ACI and CSA methods,
which are both related to 2Aocotθ. This difference in efficiency is particularly pronounced in this example because the
area enclosed by the ledge of the spandrel is not included when calculating αx1y1, making it much smaller than 2Aocotθ
and hence requiring more reinforcement to obtain the same amount of torsional strength. Given the relatively high
cost of placing closed stirrups compared to using longitudinal steel, both CSA designs are more economical for this
particular example. The results of this design example reflect the observations made earlier in this paper about the
relative influences of transverse and longitudinal reinforcement on predicted torsional strength using the three
methods.
CONCLUSION
When permitted, using the alternative design clause in ACI 318-19 can produce more economical and constructible
designs compared to simply applying the basic torsion provisions in the ACI code. Having thoroughly validated and
compared the torsion design provisions in ACI 318-19, CSA A23.3:19 and the PCI Design Handbook 8th edition in
this paper, the following conclusions can be made to inform practicing engineers who might be interested in using
these alternative procedures:
• All three design methods show good agreement with tests found in the literature, particularly with specimens
which were under-reinforced and hence failed by yielding. The predictions made using the three design
procedures are generally more accurate when applied to instances of pure torsion and tend to be conservative
for tests involving torsion in combination with moment and/or shear.
• The CSA method can capture more failure modes attributed to torsion compared to the ACI and PCI
approaches, particularly those associated with non-symmetric sections under combined moment and torsion.
• ACI and CSA use the same threshold torsion value in design. The threshold torsion used by PCI is generally
higher than in CSA and ACI, although this difference becomes smaller for large aspect ratios. Each code
predicts a similar influence of prestressing on Tth.
• ACI 318-19 has the most stringent requirements for minimum transverse reinforcement for non-prestressed
members. CSA A23.3:19 has more lenient minimum reinforcement requirements for low concrete strengths
and the PCI 8th edition method has more lenient minimum reinforcement requirements if the member is non-
prestressed.
• The CSA upper limit on torsion is considerably less conservative than the one used in ACI and PCI. However,
the scatter in the data suggests that more research is required to find a better upper limit for all of the codes.
23
Alternative Design Procedures for Torsion in ACI 318-19: A Comparative Study
• If yielding of the longitudinal reinforcement governs the design of a member, using a value of εx which is
greater than the default value calculated from Eq. (18) can produce more optimal designs using the CSA
code.
• Designs obtained using the ACI code generally require the most reinforcement. Designs obtained using the
CSA code generally require less reinforcement than both ACI and PCI methods, especially when Tu is
relatively large.
• For smaller values of Tu, the PCI approach likely requires the least amount of reinforcement due to presence
of Tc which can contribute considerably to the overall strength of the member, especially if it is prestressed.
• The benefit of requiring less reinforcement for prestressed members designed using the PCI method may be
offset by the more stringent minimum reinforcement requirements.
NOTATION
Acp = Area enclosed by outside perimeter of concrete cross section
Al = Total area of longitudinal reinforcement to resist torsion
Ao = Area enclosed by shear flow path
Aoh = Area enclosed by centerline of closed transverse reinforcement
Ap = Area of prestressed reinforcement on the flexural tension side of member
Ap’ = Area of prestressed reinforcement on the flexural compression side of member
As = Area of longitudinal reinforcement on the flexural tension side of member
As’ = Area of longitudinal reinforcement on the flexural compression side of member
At = Area of one leg of closed transverse torsion reinforcement
Av = Area of shear reinforcement within spacing s
bt = Width of part of cross section containing closed stirrups resisting torsion
bw = Effective web width within depth d
C = Factor accounting for prestress
Ct = Factor relating shear and torsional stress properties
d = Distance from the extreme compression fibre to centroid of longitudinal tension reinforcement
Es = Young’s modulus of reinforcing steel
Ep = Young’s modulus of prestressed reinforcement
fc’ = Specified compressive strength of concrete
Flc = Force in the longitudinal reinforcement on the flexural compressive side of the section
Flt = Force in the longitudinal reinforcement on the flexural tension side of the section
fpo = Stress in prestressed reinforcement when strain in surrounding concrete is zero
fpy = Specified yield strength of prestressed reinforcement
fy,l = Specified yield strength of longitudinal reinforcement
fy,t = Specified yield strength of transverse reinforcement
fpc = Compress in concrete at the centroid of the cross section after considering all prestress losses
h = Overall height of member
jd = Flexural lever arm
Mn = Nominal flexural strength of the member
Mu = Moment demand at section
Nu = Axial load demand at section
pcp = Outside perimeter of the concrete cross section
ph = Perimeter of the centerline of the closed transverse torsion reinforcement
s = Spacing of transverse reinforcement
Tc = Torsional strength attributed to concrete
Tc’ = Torsional strength attributed to concrete in the absence of shear
Tn = Nominal torsional strength of the member
Ts = Torsional strength provided by closed transverse torsion reinforcement
Tth = Threshold torsion below which the effects of torsion may be neglected
Tu = Torsion demand at section
Vc = Shear strength attributed to concrete
Vc’ = Shear strength attributed to concrete in the absence of torsion
Vp = Unfactored component if the direction of the applied shear of the effective prestressing force
Vs = Shear strength provided by shear reinforcement
24
Kuan et. al.
REFERENCES
[1] ACI Committee 318 “Building Code Requirements for Reinforced Concrete (ACI 318-71),” American
Concrete Institute, Detroid, MI, 1971, 144 pp.
[2] MacGregor, J.G. and Ghoneim, M.G., “Design for Torsion”, ACI Structural Journal, Vol. 92, No. 2, March-
April 1995, pp. 211-218.
[3] PCI Design Handbook – Precast and Prestressed Concrete, Eighth Edition, Precast/prestressed Concrete
Institute, Chicago, IL, 2017.
[4] ACI Committee 318 “Building Code Requirements for Structural Concrete (ACI 318-19) and Commentary
(ACI 318R-19),” American Concrete Institute, Farmington Hills, MI, 2019, 623 pp.
[5] Rahal, K.N., “Combined Torsion and Bending in Reinforced and Prestressed Concrete Beams Using
Simplified Method for Combined Stress-Resultants,” ACI Structural Journal, Vol. 104, No. 38, July-August,
2007, pp. 402-411.
[6] Rahal, K.N., “Torsional strength of normal and high strength reinforced concrete beams,” Engineering
Structures, Vol. 56, 2013, pp. 2206-2216.
[7] Collins, M.P. and Mitchell, D., "Shear and Torsion Design of Prestressed and Non-Prestressed Concrete
Beams", Journal of the Prestressed Concrete Institute, Vol. 25, No. 5, September-October. 1980, pp. 32-100.
[8] CSA A23.3:19, “Design of Concrete Structures,” Canadian Standards Association, Mississauga, ON, Canada,
2019, 295 pp.
[9] American Association of State Highway and Transportation Officials, “AASHTO LRFD Bridge Design
Specifications and Commentary,” 8th Edition, Washington, D.C., 2017, 1781 pp.
[10] EN 1992-1-1:2004, “Eurocode 2: Design of Concrete Structures- Part 1-1: General Rules and Rules for
Buildings,” European Committee for Standardization, Brussels, Belgium, 2004, 225 pp.
[11] BD-002 “AS 3600:2018 Concrete Structures,” Standards Australia, Sydney, Australia, 2018, 259 pp.
[12] Zia, P. and Hsu, T.C., “Design for Torsion and Shear in Prestressed Concrete Flexural Members”, PCI
Journal, Vol. 49, No. 3, May-June, 2004, pp. 34-42.
25
Alternative Design Procedures for Torsion in ACI 318-19: A Comparative Study
[13] Zia, P., and McGee, W.D., “Torsion Design of Prestressed Concrete”, PCI Journal, Vol. 19, No. 2, Mar-Apr,
1974, pp. 46-65.
[14] Bernardo, L.F.A. and Lopes, S.M.R., “Torsion in High-Strength Concrete Hollow Beams: Strength and
Ductility Analysis”, ACI Structural Journal, Vol. 106, No. 1, January-February, 2009, pp. 39-48.
[15] Fang, I.K. and Shiau, J.K., “Torsional Behaviour of Normal- and High-Strength Concrete Beams”, ACI
Structural Journal, Vol. 101, No. 3, May-June 2004, pp. 304-313.
[16] Hsu, Thomas T. C., "Torsion of Structural Concrete - Behavior of Reinforced Concrete Rectangular
Members," Torsion of Structural Concrete, SP-18, American Concrete Institute, Detroit, 1968, pp. 261-306.
[17] Jeng, C.H., Peng, S.F., Chiu, H.J. and Hsiao, C.K, “New Torsion Experiment on large-Sized Hollow
Reinforced Concrete Beams,” ACI Structural Journal, Vol. 111, No. 6, November-December 2014, pp. 1469-
1480.
[18] Koutchoukali, N.E. and Belarbi, A., “Torsion of High-Strength Reinforced Concrete Beams and Minimum
Reinforcement Requirement”, ACI Structural Journal, Vol. 98, No. 4, July-August 2001, pp. 462-469.
[19] Lampert, P., and Thurlimann, B., “Torsionsversuche an Stahlbetonbalken,” Bericht Nr. 6506-2, Institut fur
Baustatik, ETH Zurich, Germany, Jan. 1968.
[20] McMullen, A.E. and Rangan, B.V., “Pure Torsion in Rectangular Sections – A Re-Examination,” ACI
Journal, Vol. 75, No.10, October 1978, pp. 511-519.
[21] McMullen, A. E., and Warwaruk, J., “The Torsional Strength of Rectangular Reinforced Beams Subjected
to Combined Loading,” Report No. 2, Civil Engineering Department, University of Alberta, Alberta, Canada,
1967, 162 pp.
[22] Mitchell, D. and Collins, M.P., "Influence of Prestressing on Torsional Response of Concrete Beams,"
Journal of the Prestressed Concrete Institute, Vol. 23, No. 3, May-June 1978, pp. 54-73.
[23] Mardukhi, J., “The Behaviour of Uniformly Prestressed Concrete Box Beams in Combined Torsion and
Bending”, M.A.Sc. Thesis, 1974, 72 pp.
[24] Pandit, G.S. and Warwaruk, J., “Reinforced Concrete Beams in Combined Bending and Torsion,” American
Concrete Institute Special Publication, SP-18, 1968, pp. 133-164.
[25] Rasmussen, L.J. and Baker, G., “Torsion in Reinforced Normal and High-Strength Concrete Beams – Part 1:
Experimental Test Series”, ACI Structural Journal, Vol. 92, No. 1, January-February 1995, pp. 56-62.
[26] Wafa, F.F., Shihata, S.A., Ashour S.A. and Akhtaruzzaman, A.A., “Prestressed High-Strength Concrete
Beams under Torsion,” Journal of Structural Engineering, Vol. 121, No. 9, September 1995, pp. 1280-1286.
[27] Collins, M.P., Walsh, P.F., Archer, F.E., and Hall, A.S., "The Ultimate Strength of Reinforced Concrete
Beams Subjected to Combined Torsion and Flexure," American Concrete Institute Special Publication, SP-
18, 1968, pp. 379-402.
[28] Goode, C.D. and Helmy, M.A., “Ultimate Strength of Reinforced Concrete Beams in Combined Bending and
Torsion,” American Concrete Institute Special Publication, SP-18, 1968, pp. 357-378.
[29] Kemp, E.L., “Behavior of Concrete Members Subject to Torsion and to Combined Torsion, Bending, and
Shear,” American Concrete Institute Special Publication, SP-18, 1968, pp. 179-202.
[30] Iyengar, K.T.S.R. and Rangan, B.V., “Strength and Stiffness of Reinforced Concrete Beams Under Combined
Bending and Torsion,” American Concrete Institute Special Publication, SP-18, 1968, pp. 403-440.
26
Kuan et. al.
[31] Onsongo, W. M., “The Diagonal Compression Field Theory for Reinforced Concrete Beams Subjected to
Combined Torsion, Flexure, and Axial Load,” PhD thesis, Department of Civil Engineering, University of
Toronto, Toronto, Ontario, Canada, 1978, 246 pp.
[32] Lampert, P., and Thurlimann, B., “Torsions-Biege-Versuche an Stahlbetonbalken,” Bericht Nr. 6506-3,
Institut fur Baustatik, ETH Zurich, Germany, Jan. 1969.
[33] Collins, M.P. and Lampert, P., "Torsion, Bending and Confusion - An Attempt to Establish the Facts," Journal
of the American Concrete Institute, Vol. 69, No. 8, August 1972, pp. 500-504.
[34] Rahal, K.N. and Collins, M.P., “Experimental Evaluation of ACI and AASHTO-LRFD Design Provisions
for Combined Shear and Torsion,” ACI Structural Journal, Vol. 100, No.3, May-June 2003, pp 277-282.
[35] Klein, G. J., “Design of Spandrel Beams,” PCI Specially Funded Research Project No. 5, Precast/ Prestressed
Concrete Institute, Chicago, IL, 1986.
[36] Rahal, K.N. and Collins, M.P., "Effect of Thickness of Concrete Cover on the Shear-Torsion Interaction - An
Experimental Investigation," ACI Structural Journal, Vol. 92, No. 3, May-June 1995, pp. 334-342.
[37] Proestos, G.T., Bentz, E.C. and Collins, M.P., “Maximum Shear Capacity of Reinforced Concrete Members”,
ACI Structural Journal, Vol. 115, No. 5, September 2018, pp. 1463-1474.
[38] Kuan, A., Bruun, E.P.G., Bentz, E.C., Collins, M.P., “Nonlinear sectional analysis of reinforced concrete
beams and shells subjected to pure torsion,” Computers and Structures Vol. 222, October 2019, pp. 118-132.
27