Engineering Structures 168 (2018) 759–769

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Ductility of RC beams under torsion T

⁎
M.M. Teixeira, L.F.A. Bernardo
University of Beira Interior, C-MADE – Centre of Materials and Building Technologies, Covilhã, Portugal

A R T I C LE I N FO A B S T R A C T

Keywords: In this article, the torsional ductility of reinforced concrete (RC) beams with rectangular cross section is studied.
Reinforced concrete For this, the experimental results of several RC beams found in the literature were compiled and analyzed. A
Beams torsional ductility index was used to characterize the torsional ductility of the studied beams. The following
Torsion variables study were considered: compressive concrete strength, torsional reinforcement ratio and cross section
Ductility
type (plain or hollow). The influence of each variable study on the torsional ductility is studied and important
Codes of practice
findings are pointed out which could help for the design of RC beams under torsion. An additional comparative
analysis with the rules from some codes of practice in use is also performed. It is shown that, in general, the
codes are too much restrictive as far as the maximum torsional reinforcement is concerned. As a consequence,
this can leads to the unacceptance of several beams with ductile behavior.

1. Introduction plastic twist capacity in their critical sections [6,7]. These studies also
show that, as for bending and shear force, some requirements are need
Nowadays, it is well known that plastic deformation capacity con- to ensure torsional ductility in RC beams. For instance, a minimum and
stitutes an important requirement to ensure the redistribution capacity maximum amount of torsional reinforcement must be provided, in
of internal forces in structures, in order to prevent a sudden and pro- order to avoid a brittle and premature failure due to the yielding of the
gressive structural collapse. This property, which depends on the duc- reinforcement right after cracking or a brittle failure due to compressive
tility of the critical sections of reinforced concrete (RC) members, is concrete crushing. When compared with the bending case, these studies
important because it is related with the structural safety for the ulti- seems to show that the range for the amount of torsional reinforcement
mate limit states. compatible with ductility is much narrower. Additionally, special care
Over the last decades, many experimental studies have shown that must also be given to the detailing of the torsional reinforcement. For
the critical cross sections of RC members, namely beams, can provide instance, spacing between steel bars cannot be exaggerated so that
sufficient ductility after the yielding of the reinforcement to ensure the cracks are always intersected by bars.
redistribution capacity of the internal forces for the ultimate state. This Older experimental studies also show that RC hollow beams are less
is particularly accepted for RC beams under bending, including high- ductile when compared to similar RC plain beams [8]. Since hollow
strength concrete beams [1–3]. For RC members under shear force, beams are commonly used, this observation becomes very important.
doubts about the ductile behavior may still exist. However, experi- In order to provide sufficient ductility in structural members,
mental researches show that RC members under shear, including high- structural engineers usually follow the rules from codes of practice. For
strength concrete panels, can also present good levels of ductility [4,5]. RC beams under bending and shear forces, codes of practice in use
This is because the softening effect (influence of diagonal cracking on provide sufficient information. However, for torsion specific rules to
the strength and deformation capacity of concrete in compression) ensure ductility are much scarcer. For instance, in some codes the
leads to internal energy dissipation through appreciable levels of plastic minimum amount of torsional reinforcement, which can be considered
shear deformation. a basic requirement, is still inexistent. Additionally, the maximum
The above considerations about RC members under shear can also torsional reinforcement can indirectly be satisfied by checking the
be extended to RC beams under torsion, due to the prevalence of an maximum compressive stress allowed in the concrete struts. This limit
internal shear stress state imposed by the external torsional moments. is usually given in the codes. However, some codes continue to refer to
Few previous researches on RC beams under torsion have shown that some ductility rules for bending and shear forces, as far as longitudinal
these structural elements can also present appreciable ductility and and transverse torsional reinforcement is concerned.

⁎
Corresponding author at: University of Beira Interior, Department of Civil Engineering and Architecture, Edifício II das Engenharias, Calçada Fonte do Lameiro, 6201-001 Covilhã,
Portugal.
E-mail address: lfb@ubi.pt (L.F.A. Bernardo).

https://doi.org/10.1016/j.engstruct.2018.05.021
Received 16 December 2017; Received in revised form 13 April 2018; Accepted 7 May 2018
0141-0296/ © 2018 Elsevier Ltd. All rights reserved.
M.M. Teixeira, L.F.A. Bernardo Engineering Structures 168 (2018) 759–769

Table 1
Properties of reference beams.
a
Beam x y t x1 y1 Asl Ast / s ρl ρt fly fty mb fc εc3 θy,th θu,th μ θ,th θy,exp θu,exp μ θ,exp
cm cm cm cm cm cm2 cm2/m % % MPa MPa MPa ‰ °/m °/m °/m °/m

B3 [8] P 25.4 38.1 – 21.6 34.3 11.36 10.16 1.17 1.17 328 320 1.02 28.1 1.75 2.14 3.60 1.69 2.37 3.86 1.63
B4 [8] P 25.4 38.1 – 21.6 34.3 15.48 14.01 1.60 1.62 320 323 0.98 29.2 1.75 1.91 3.37 1.77 2.56 4.63 1.81
B5 [8] P 25.4 38.1 – 21.6 34.3 20.39 18.47 2.11 2.13 332 321 1.02 30.6 1.75 1.76 3.22 1.83 2.85 5.10 1.79
B6 [8] P 25.4 38.1 – 21.6 34.3 25.81 22.58 2.67 2.61 332 323 1.05 28.8 1.75 1.63 2.99 1.83 3.05 4.41 1.44
B9 [8] P 25.4 38.1 – 21.6 34.3 11.36 4.66 1.17 0.54 319 343 2.03 28.8 1.75 1.58 3.94 2.50 1.73 4.34 2.51
C4 [8] P 25.4 25.4 – 21.6 21.6 11.36 13.11 1.76 1.76 337 328 1.03 27.2 1.75 2.12 3.74 1.76 3.68 5.91 1.61
C5 [8] P 25.4 25.4 – 21.6 21.6 15.48 17.67 2.40 2.37 328 329 1.01 27.2 1.75 1.95 3.52 1.80 3.98 8.83 2.22
C6 [8] P 25.4 25.4 – 21.6 21.6 20.39 23.91 3.16 3.20 316 328 0.95 27.6 1.75 1.81 3.35 1.85 4.43 7.30 1.65
G3 [8] P 25.4 50.8 – 21.6 47.0 11.36 8.29 0.88 0.88 339 328 1.03 26.8 1.75 1.85 3.50 1.89 2.03 3.89 1.91
G4 [8] P 25.4 50.8 – 21.6 47.0 15.48 11.29 1.20 1.20 326 321 1.01 28.3 1.75 1.95 3.31 1.70 2.47 3.82 1.55
G5 [8] P 25.4 50.8 – 21.6 47.0 20.39 15.05 1.58 1.60 331 328 1.00 26.9 1.75 1.70 2.99 1.76 2.53 4.49 1.78
G7 [8] P 25.4 50.8 – 21.6 47.0 12.00 8.84 0.93 0.94 319 323 0.98 31.0 1.75 1.77 3.83 2.17 1.67 3.96 2.38
G8 [8] P 25.4 50.8 – 21.6 47.0 17.03 12.32 1.32 1.31 322 329 0.99 28.3 1.75 1.87 3.23 1.72 2.26 4.42 1.96
I3 [8] P 25.4 38.1 – 21.6 34.3 11.36 10.16 1.17 1.17 343 334 1.03 44.8 1.75 1.98 4.71 2.38 1.62 5.01 3.09
I4 [8] P 25.4 38.1 – 21.6 34.3 15.48 14.01 1.60 1.62 315 326 0.96 45.0 1.75 2.12 4.18 1.97 1.61 2.68 1.67
I5 [8] P 25.4 38.1 – 21.6 34.3 20.39 18.47 2.11 2.13 310 326 0.94 45.0 1.75 1.99 3.84 1.93 2.24 5.20 2.32
I6 [8] P 25.4 38.1 – 21.6 34.3 25.81 22.58 2.67 2.61 326 329 1.01 45.8 1.75 1.83 3.64 1.99 2.35 5.16 2.19
J1 [8] P 25.4 38.1 – 21.6 34.3 5.16 4.66 0.53 0.54 328 346 0.94 14.3 1.75 2.21 3.12 1.41 2.38 5.20 2.19
J2 [8] P 25.4 38.1 – 21.6 34.3 8.00 7.21 0.83 0.83 320 341 0.93 14.6 1.75 1.88 2.82 1.50 2.62 5.20 1.98
J4 [8] P 25.4 38.1 – 21.6 34.3 15.48 14.01 1.60 1.62 324 332 0.97 16.8 1.75 1.61 2.61 1.62 3.15 5.21 1.66
K2 [8] P 15.2 49.5 – 11.4 45.7 7.74 6.77 1.03 1.03 336 338 0.99 30.6 1.75 2.91 5.41 1.86 2.30 3.55 1.54
K3 [8] P 15.2 49.5 – 11.4 45.7 12.00 10.42 1.59 1.58 316 321 0.99 29.0 1.75 2.61 4.69 1.80 3.39 5.88 1.73
K4 [8] P 15.2 49.5 – 11.4 45.7 17.03 15.05 2.26 2.28 344 340 1.00 28.6 1.75 2.32 4.28 1.85 3.92 8.66 2.21
M2 [8] P 25.4 38.1 – 21.6 34.3 11.36 6.77 1.17 0.78 329 357 1.38 30.6 1.75 1.95 3.89 2.00 2.05 3.71 1.81
N1 [8] P 15.2 30.5 – 13.0 28.3 2.84 3.50 0.61 0.62 352 341 1.01 29.5 1.75 2.85 7.53 2.64 2.48 10.64 4.29
N1a [8] P 15.2 30.5 – 13.0 28.3 2.84 3.50 0.61 0.62 346 345 0.98 28.7 1.75 2.87 7.33 2.56 2.53 10.63 4.21
N2 [8] P 15.2 30.5 – 13.0 28.3 5.16 6.35 1.11 1.13 331 338 0.96 30.4 1.75 3.41 5.80 1.70 3.89 8.18 2.10
N2a [8] P 15.2 30.5 – 13.0 28.3 1.61 6.21 1.11 1.10 333 361 0.93 28.4 1.75 3.29 5.58 1.69 3.84 10.40 2.71
N3 [8] P 15.2 30.5 – 13.0 28.3 4.26 5.08 0.92 0.90 352 352 1.02 27.3 1.75 3.43 5.81 1.69 3.60 9.75 2.71
N4 [8] P 15.2 30.5 – 13.0 28.3 6.58 7.98 1.42 1.42 341 356 0.96 27.3 1.75 2.94 5.13 1.74 4.12 9.33 2.26
VB2 [10] P 44.0 24.0 – 42.0 22.0 7.01 5.84 0.66 0.71 541 541 0.94 26.4 1.75 2.37 3.85 1.63 3.42 4.18 1.22
VB3 [10] P 44.0 24.0 – 42.0 22.0 7.01 5.84 0.66 0.71 541 541 0.94 39.1 1.75 2.77 4.71 1.70 3.20 4.50 1.40
VB4 [10] P 44.0 24.0 – 42.0 22.0 7.01 5.84 0.66 0.71 541 541 0.94 49.8 1.75 2.93 5.29 1.81 2.93 4.24 1.45
VM2 [10] P 44.0 24.0 – 42.0 22.0 6.60 5.32 0.63 0.65 432 436 0.96 36.1 1.75 2.31 5.03 2.18 2.18 3.43 1.57
VM3 [10] P 58.7 32.0 – 56.1 29.4 12.84 7.14 0.68 0.65 461 442 1.10 40.0 1.75 1.71 3.91 2.29 1.82 3.66 2.02
VQ1 [10] P 32.4 32.4 – 30.4 30.4 3.46 2.88 0.33 0.33 557 557 0.99 19.0 1.75 2.69 3.78 1.41 2.68 3.24 1.21
VQ3 [10] P 58.0 18.6 – 56.0 16.6 4.27 3.05 0.40 0.41 433 433 0.96 17.6 1.75 2.89 4.23 1.46 2.52 2.98 1.18
VQ9 [10] P 80.6 14.0 – 78.6 12.0 5.08 2.82 0.45 0.45 441 441 0.99 19.5 1.75 3.56 5.31 1.49 3.82 4.59 1.20
VS2-VQ2 [10] P 44.0 24.0 – 42.0 22.0 3.66 3.05 0.35 0.37 433 433 0.94 19.0 1.75 2.33 4.12 1.77 1.77 2.70 1.52
VS3 [10] P 44.0 24.0 – 42.0 22.0 5.49 4.55 0.52 0.55 433 433 0.94 19.5 1.75 2.42 3.59 1.48 3.70 3.97 1.08
VS4-VQ5 [10] P 44.0 24.0 – 42.0 22.0 7.32 6.10 0.69 0.74 433 433 0.94 19.0 1.75 2.11 3.28 1.55 3.46 3.99 1.15
VS9 [10] P 44.0 24.0 – 42.0 22.0 3.48 2.90 0.33 0.35 571 571 0.94 17.6 1.75 2.66 3.77 1.42 2.97 3.74 1.26
VS10-VB1 [10] P 44.0 24.0 – 42.0 22.0 6.96 5.80 0.66 0.70 571 571 0.94 19.0 1.75 2.12 3.29 1.55 3.37 3.86 1.14
A2 [11] P 25.4 25.4 – 22.2 22.2 5.16 7.82 0.80 1.08 380 285 0.99 38.2 1.75 2.11 5.87 2.78 2.12 4.17 1.96
A3 [11] P 25.4 25.4 – 21.9 21.9 8.00 8.94 1.24 1.22 352 360 1.00 39.4 1.75 2.75 4.99 1.81 2.36 7.49 3.17
A4 [11] P 25.4 25.4 – 21.9 21.9 11.36 12.42 1.76 1.69 351 360 1.02 39.2 1.75 2.42 4.50 1.86 2.98 4.34 1.46
B3 [11] P 17.8 35.6 – 14.3 32.1 8.00 8.60 1.27 1.26 352 360 0.98 38.6 1.75 3.01 5.40 1.79 3.61 7.32 2.03
B4 [11] P 17.8 35.6 – 14.3 32.1 11.36 11.76 1.80 1.73 351 360 1.02 38.5 1.75 2.59 4.89 1.89 3.50 6.54 1.87
B5UR1 [12] P 20.3 30.5 – 16.5 26.7 5.16 6.56 0.83 0.92 386 373 0.94 39.6 1.75 2.65 5.97 2.25 1.33 3.75 2.82
B7UR1 [12] P 20.3 30.5 – 16.5 26.7 5.16 6.56 0.83 0.92 386 399 0.88 64.6 1.84 2.42 7.49 3.10 0.58 2.35 4.06
B9UR1 [12] P 20.3 30.5 – 16.5 26.7 5.16 6.56 0.83 0.92 386 373 0.94 75.0 1.98 2.38 8.42 3.53 1.87 3.93 2.10
B12UR1 [12] P 20.3 30.5 – 16.5 26.7 5.16 6.56 0.83 0.92 386 399 0.88 80.6 2.06 2.35 8.39 3.57 0.52 1.80 3.48
B12UR2 [12] P 20.3 30.5 – 16.5 26.7 5.16 6.95 0.83 0.97 386 386 0.86 76.2 2.00 2.33 8.03 3.45 0.31 2.07 6.58
B12UR3 [12] P 20.3 30.5 – 16.5 26.7 6.58 7.46 1.06 1.04 380 386 1.00 72.9 1.95 2.52 7.40 2.94 1.05 3.31 3.14
B12UR4 [12] P 20.3 30.5 – 16.5 26.7 7.74 7.88 1.25 1.10 373 386 1.10 75.9 2.00 2.64 7.12 2.70 1.09 3.06 2.81
B12UR5 [12] P 20.3 30.5 – 16.5 26.7 8.00 10.13 1.29 1.41 380 386 0.90 76.7 2.01 2.64 6.53 2.47 1.54 3.85 2.51
B14UR1 [12] P 20.3 30.5 – 16.5 26.7 5.16 6.56 0.83 0.92 386 386 0.91 93.9 2.24 2.29 9.13 3.99 0.04 2.99 70.45
H-12–12 [13] P 35.0 50.0 – 30.0 45.0 20.65 14.19 1.18 1.22 410 440 0.90 78.5 2.03 1.68 4.01 2.39 1.30 4.43 3.40
H-14–10 [13] P 35.0 50.0 – 30.0 45.0 17.03 16.13 0.97 1.38 500 360 0.98 68.4 1.89 1.65 3.91 2.37 1.29 4.90 3.80
N-12–12 [13] P 35.0 50.0 – 30.0 45.0 20.65 14.19 1.18 1.22 410 440 0.90 35.5 1.75 1.62 2.88 1.78 2.09 6.51 3.11
N-14–10 [13] P 35.0 50.0 – 30.0 45.0 17.03 16.13 0.97 1.38 500 360 0.98 33.5 1.75 1.60 2.84 1.77 2.61 5.95 2.27
NBS-43–44 [14] P 35.0 50.0 – 30.0 45.0 7.60 5.09 0.43 0.44 400 385 1.03 35.0 1.75 1.45 4.86 3.34 0.91 4.17 4.59
D3 [8] H 25.4 38.1 6.4 21.6 34.3 11.36 10.16 1.17 1.17 341 333 1.02 28.4 1.75 2.02 3.57 1.77 2.22 3.44 1.55
D4 [8] H 25.4 38.1 6.4 21.6 34.3 15.48 14.01 1.60 1.62 330 333 0.98 30.6 1.75 1.91 3.38 1.77 2.27 3.27 1.44
T1 [9] H 50.0 50.0 8.0 45.4 45.4 18.10 10.28 0.72 0.75 357 357 0.97 35.3 1.75 1.22 3.13 2.56 1.53 2.32 1.51
T2 [9] H 50.0 50.0 8.0 43.0 43.0 18.10 10.28 0.72 0.71 357 357 1.02 35.3 1.75 1.22 3.14 2.57 1.71 4.00 2.34
VH1 [10] H 32.4 32.4 8.0 30.4 30.4 3.46 2.88 0.33 0.33 447 447 0.99 17.2 1.75 2.46 3.48 1.41 2.36 2.86 1.21
VH2 [10] H 32.4 32.4 8.0 30.4 30.4 6.91 5.76 0.66 0.67 447 447 0.99 17.2 1.75 1.88 2.91 1.55 1.89 3.67 1.94
A1 [6] H 60.0 60.0 9.8 53.7 54.7 6.53 3.14 0.18 0.19 696 637 1.05 48.4 1.75 1.44 5.30 3.68 0.78 2.69 3.45
A2 [6] H 60.0 60.0 10.7 53.8 53.1 13.95 6.28 0.39 0.37 672 696 1.00 47.3 1.75 1.68 3.33 1.99 1.69 3.74 2.22
A3 [6] H 60.0 60.0 10.9 53.5 53.5 18.10 8.27 0.50 0.49 672 715 0.96 46.2 1.75 1.66 2.79 1.68 1.46 2.14 1.47
A4 [6] H 60.0 60.0 10.4 52.0 52.5 23.75 11.22 0.66 0.65 724 715 1.03 54.8 1.75 1.54 2.79 1.82 1.58 2.42 1.53
A5 [6] H 60.0 60.0 10.4 52.8 52.8 30.66 14.14 0.85 0.83 724 672 1.11 53.1 1.75 1.38 2.44 1.77 1.52 2.20 1.45
(continued on next page)

760
M.M. Teixeira, L.F.A. Bernardo Engineering Structures 168 (2018) 759–769

Table 1 (continued)
a
Beam x y t x1 y1 Asl Ast / s ρl ρt fly fty mb fc εc3 θy,th θu,th μ θ,th θy,exp θu,exp μ θ,exp
cm cm cm cm cm cm2 cm2/m % % MPa MPa MPa ‰ °/m °/m °/m °/m

B2 [6] H 60.0 60.0 10.8 53.3 53.4 14.58 6.70 0.41 0.40 672 696 0.99 69.8 1.91 1.61 3.02 1.88 1.68 2.89 1.73
B3 [6] H 60.0 60.0 10.9 53.5 53.7 23.75 11.22 0.66 0.67 724 715 1.00 77.8 2.02 1.76 3.10 1.76 1.42 1.98 1.39
B4 [6] H 60.0 60.0 11.2 52.3 53.6 32.17 15.08 0.89 0.89 724 672 1.09 79.8 2.05 1.67 2.79 1.67 1.25 1.66 1.32
B5 [6] H 60.0 60.0 11.7 51.8 51.8 40.21 18.85 1.12 1.09 724 672 1.11 76.4 2.00 1.48 2.57 1.73 1.15 1.56 1.36
C1 [6] H 60.0 60.0 9.7 54.0 54.9 6.53 3.14 0.18 0.19 696 637 1.04 91.7 2.21 1.43 7.28 5.09 1.13 2.03 1.80
C2 [6] H 60.0 60.0 10.0 53.2 53.3 13.95 6.28 0.39 0.37 672 696 1.01 94.8 2.26 1.58 2.77 1.75 1.41 2.18 1.55
C3 [6] H 60.0 60.0 10.3 54.5 54.0 23.75 10.47 0.66 0.63 724 715 1.06 91.6 2.21 1.91 3.32 1.73 1.37 1.63 1.19
C4 [6] H 60.0 60.0 10.3 54.6 54.5 30.66 14.14 0.85 0.86 724 672 1.07 91.4 2.21 1.88 2.96 1.58 1.47 1.77 1.21
C5 [6] H 60.0 60.0 10.4 54.0 54.3 36.69 17.40 1.02 1.05 724 672 1.05 96.7 2.28 1.79 2.82 1.57 1.35 1.63 1.20
C6 [6] H 60.0 60.0 10.4 53.3 52.9 48.25 22.62 1.34 1.34 724 672 1.08 87.5 2.16 1.51 2.57 1.71 1.13 1.47 1.30
C065a [16] H 49.5 78.1 8.5 43.5 72.1 20.00 9.93 0.52 0.59 338 376 0.78 78.8 2.04 0.76 4.63 6.08 0.13 0.86 6.59
C100a [16] H 49.9 72.3 12.7 43.9 66.3 28.39 12.90 0.79 0.79 466 447 1.04 90.6 2.20 1.16 3.79 3.26 0.96 2.23 2.31
D090a [16] H 50.1 72.2 10.5 44.1 66.2 28.39 12.90 0.79 0.79 466 447 1.04 105.7 2.41 1.11 3.96 3.58 0.73 2.06 2.81

a
P – Plain section; H – Hollow section.

In this article, the torsional ductility of RC beams with rectangular to the all set from the previous studies were discarded because of the
cross section is studied. For this, several experimental results of RC applied criteria. The criteria to select the beams for this study were the
beams tested under pure torsion and collected from literature are used. following;
A torsional ductility index is used to characterize the ductility of the
studied beams under torsion. The influence of the following variables – The complete experimental curve Torque (T ) – Twist (θ) should be
study is evaluated: compressive concrete strength, torsional reinforce- given for the beam. This is because the torsional ductility is char-
ment ratio and cross section type (plain or hollow). Additionally, a acterized from this curve (Section 4);
comparative analysis with the rules from some codes of practice in use – The beam should have a typical behavior under torsion. To check
is also performed, in order to check their adequacy to guarantee the this, the experimental T −θ curve was qualitatively analyzed. This
ductility of RC beams under torsion. curve should show a typical shape with all the behavioral stages
It should be referred that in real structures torsion is usually com- until failure, such as described in [8];
bined with other internal forces (bending moments, shear and axial – In addition to the previous criterion, basic and current requirements
forces) in the critical sections. However, in current structures torsion were also checked by using ACI Code [17]. This code was used
can be one of the primary internal forces for the design. Moreover, as because it constitutes one of the codes in use with higher number of
usually required by the codes of practice, the design for combined specific rules for torsion, in order to impose a good behavior for a
loading requires to design the cross section separately for each internal RC beam under torsion. These requirements included: proper de-
force and, eventually, complement the analysis with interaction re- tailing and proper solution for the torsional reinforcement, max-
quirements. From this point of view, the study of RC beams under pure imum spacing between bars, minimum diameter for the bars and
torsion can be considered important. minimum size for the walls (hollow beams). Some beams that do not
respected requirements related with the minimum and maximum
2. Research significance amount of torsional reinforcement were considered because they are
of interest for this study (Section 6);
Specific studies on the torsional ductility of RC beams are very – The beam should have a minimum ductile failure. This was checked
scarce [6,7]. Moreover, these studies only focused RC hollow beams. In by visualizing qualitatively the experimental T −θ curve in the ulti-
the literature, no additional studies were found on the torsional duc- mate stage. This one should show inelastic deformations for the
tility of RC plain beams, neither comparative analyses between similar peak load and eventually also a post-peak descending branch. This
plain and hollow RC beams. In [7] it was shown, from numerical shape indicates that the test was carried out with sufficient control
analyses with continuous and curved RC beams, that the redistribution to register the experimental inelastic deformations in the ultimate
capacity of torsional moments is as much important as for bending. This stage. In general, such behavior occurs when the torsional re-
requires that sufficient torsional ductility must be ensured in the critical inforcement yields before the maximum torque is reached.
sections. However, ductile behavior was also observed for some beams for
The lack of specific studies on this subject is reflected in the limited which this condition did not occurred. This will be discussed in the
rules for most of codes of practice in use. For instance, when compared next section.
with the ductility rules for bending, the number of ductility rules di-
rectly related with torsion is much lesser or even inexistent. Table 1 presents some of the geometrical and mechanical properties
For these reasons, a specific study on the torsional ductility of RC for the reference beams that were effectively used in this study, namely:
beams can be considered important. the width ( x ) and the height ( y ) of the cross section, the thickness of the
walls for hollow sections (t ), the total area of longitudinal reinforce-
3. Reference beams ment ( Asl ), the distributed area of the transverse reinforcement ( Ast / s ,
with Ast the area for one bar of the hoop and s the longitudinal spacing
From the literature, several RC beams with rectangular cross section between hoops), the longitudinal reinforcement ratio (ρl = Asl / xy ), the
and tested under torsion were collected for this study. These beams transverse reinforcement ratio (ρt = Ast u/ Ac s , with u = 2(x1 + y1) and
come from the following studies: Hsu in 1968 [8], Lampert and Thur- Ac = xy , being x1 and y1 the average width and height of the transverse
limann in 1969 [9], Leonhardt and Schelling in 1974 [10], McMullen reinforcement), the average yielding stress for the longitudinal and
and Rangan in 1978 [11], Koutchkali and Belarbi in 2001 [12], Fang transverse reinforcement ( fly and fty , respectively), the effective balance
and Shiau in 2004 [13], Chiu et al. in 2007 [14], Bernardo and Lopes in between longitudinal and transverse reinforcement for the ultimate
2009 [6], Peng and Wong in 2011 [15] and Jeng in 2015 [16]. Beams state (mb = (ρl fly )/(ρt fty ) = [Asl s /(At 2(x1 + y1))](fly / fty ) ) and the average
from other studies found in literature and some of the beams belonging compressive strength for concrete ( fc ). Parameter mb represents the

761
M.M. Teixeira, L.F.A. Bernardo Engineering Structures 168 (2018) 759–769

ratio between the mechanical ratios of the longitudinal reinforcement beams with unbalanced torsional reinforcement, longitudinal and
to the transverse reinforcement. The other parameters presented in transverse reinforcement do not yield at the same time. In this case two
Table 1 will be explained later. Some of the properties for the materials yielding points far apart exist in the T −θ curve (or even only one if the
are usually not given by the authors, namely the average Young’s reinforcements are highly unbalanced). In this case, the chosen cri-
Modulus for steel and concrete (Es and Ec , respectively). For concrete, terion was defined from the observation of the shape of the T −θ curve
the Young’s Modulus was computed from the European code EC 2 [18] immediately after the first yielding point. If a pronounced nonlinear
by correlation with fc . For steel bars, the Young’s Modulus was con- shape starts to develop right after this point (deformation rate starts to
sidered to be 200 GPa [18]. increase when compared with torque rate), then the effective yielding
Some of the properties presented in Table 1 are related with the point is attributed to the first yielding point (or to the unique one if only
variables study to be analyzed (Section 5) and are also need to compute one of the torsional reinforcement yielded). These situations are ex-
both the experimental and theoretical ductility indexes (Section 4). emplified in Fig. 1 for Beams B9 and M2 [8]. If the nonlinear shape
starts only after the second yielding point, either this second point is
4. Torsional ductility index considered the effective yielding point, either an average point is de-
fined if the two yielding points are not too far apart.
In this study, the same torsional ductility index μ θ defined and used It should be referred that the theoretical T −θ curve was sometimes
by Bernardo and Lopes in 2015 [6] was used. This index, based on the used to check the location of the yielding key point for the reference
angular deformation per unit meter (twist), is defined as follows: beams for which the authors did not provide sufficient data to identify
clearly this point. This can be done because it is known that the
θu
μθ = GSVATM provides good predictions, including for the cracking and
θy (1) maximum torque [19]. For such beams, the location of the experi-
where mental yielding point was obtained from an interpolation between the
cracking and maximum torque in the experimental T −θ curve, by
θu = ultimate twist (corresponding to the ultimate torque Tu ); knowing the relative location of this point between the cracking and
θy = yielding twist (corresponding to the yielding torque Ty ). maximum torque in the theoretical T −θ curve. In the end, the location
of the yielding point in the experimental T −θ curve was confirmed from
In this study, the experimental and theoretical values for μ θ are a qualitative analysis of the shape of the curve right after this point. In
computed for the reference beams. The experimental values are ob- Fig. 1, it can be seen that this procedure is quite acceptable from
tained from the experimental T −θ curves. The theoretical values are comparison between the experimental and theoretical T −θ curves,
obtained from the theoretical T −θ curves computed from the namely for the torque levels and the relative location of the yielding
Generalized Softened Variable Angle Truss Model (GSVATM). This point between the cracking and maximum torque (see Beams G4 and B9
analytical model was chosen because it provides good predictions for in Fig. 1).
the behavior of RC beams under torsion [19]. Still related with the yielding point of the T −θ curve, it was ob-
The yielding twist θy is defined from the key point of the T −θ curve served that for some reference beams a ductile behavior (inelastic de-
corresponding to the yielding of the torsional reinforcement (point with formations) rises before the yielding of the torsional reinforcement.
coordinates (θy ;Ty ) ). For beams with balanced torsional reinforcement This situation is exemplified in Fig. 1 for Beams I5 and K3 [8]. For this
(mb ≈ 1), both longitudinal and transverse reinforcements generally last one, the torsional reinforcement did not yielded. This occurs par-
yield at the same time (or almost). In this situation, only one yielding ticularly in beams with normal strength concrete and with moderate or
point is defined in the T −θ curve. If two closely yielding points are high torsional reinforcement ratio. This behavior is not new and can be
identified (one for each torsional reinforcement), then an average point explained due to the nonlinear behavior of the compressive concrete
is defined. This situation is exemplified in Fig. 1 for Beam G4 [8]. For before the peak stress due to the initiation of the crushing phenomena.

80 Fig. 1. Experimental and theoretical T −θ

curves for some reference beams.

70 Exp. B9
Theor. B9
60
Exp. G4
Theor. G4
50
Exp.I5
T [kN.m]

40 Theor. I5
Exp. K3
30 Theor. K3
Exp. M2
20 Theor. M2
Equiv. Yielding
10
Yielding Asl
Yielding Ast
0
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0
θ [º/m]

762
M.M. Teixeira, L.F.A. Bernardo Engineering Structures 168 (2018) 759–769

In current situations with over-reinforced sections in bending, an ap- appreciable differences exist between μ θ,th and μ θ,exp for some of the
propriate design based on the strain control for the reinforcement can reference beams. This is because the theoretical and experimental T −θ
maximize this favorable behavior of the compressive concrete [20]. For curves usually show different shapes in the ending. This occur because
the reference beams that show this behavior, this ductile behavior must GSVATM incorporates criteria to stop the calculations, namely con-
be considered and a criterion is need to define an equivalent yielding ventional limits for the strains in the materials. For this reason, in this
point in the T −θ curve. The chosen criterion was to attribute for the study the theoretical values for the torsional ductility index are only
equivalent yielding point, the point in the T −θ curve for which the used to validate the experimental trends observed with the same ex-
strain in the concrete in compression reaches the strain corresponding perimental values (Section 5), which are expected to present higher
to the peak stress. This was done with the help of the theoretical results dispersion. Since the theoretical T −θ curves are computed with the
from the GSVATM, since experimental data for the strain in the com- same stopping criteria for all the reference beams, it is assumed that the
pressive concrete in the principal direction of stresses are not usually obtained theoretical trends are valid to check the experimental ones.
provided in the studies. To compute a reference value for the strain
corresponding to the peak stress for the concrete in compression, three
values from EC 2 [18] were tested: εc1 from a nonlinear stress (σ ) – 5. Analysis of the torsional ductility index
strain (ε ) relationship, εc2 from a parabola-rectangle σ−ε relationship
and εc3 from a bilinear σ−ε relationship. After the calculations and a The aim of this section is to analyze qualitatively, with scatter
comparative analysis between the concerned referred beams, it was graphs, the evolution of the torsional ductility index as function of some
found that εc3 from a bilinear σ−ε relationship is more realistic to es- variables study related with the characteristics of the reference beams.
timate the equivalent yielding point in the T −θ curve. Table 1 presents The analysis of the results from previous experimental researches car-
the values for εc3 computed from EC 2 [18] for the reference beams. For ried out over the last decades allow to state that the ultimate behavior
all the remaining reference beams, it was checked that the equivalent of RC beams under torsion, with rectangular cross section, is influenced
yielding point (concrete) is located to the right of the yielding point by the following variables [7,8,10,21]:
(torsional reinforcement) in the T −θ curves.
The location and verification of the key point corresponding to εc3 in – concrete compressive strength ( fc );
the experimental T −θ curve was obtained in a similar way as previously – total ratio of torsional reinforcement ρtot (beams with balanced re-
explained for the yielding point. The location of the equivalent yielding inforcement);
point is exemplified in Fig. 1 for Beams I5 and K3 [8]. – section type (plain or hollow);
From the referred above, and by knowing the location of the – unbalance of torsional reinforcement (characterized with parameter
yielding point (or equivalent yielding point), the value for θy was ob- mb );
tained for both the theoretical and experimental T −θ curves (θy,th and – height to width ratio ( y / x ) of the rectangular cross section.
θy,exp , respectively). These values are presented in Table 1.
The ultimate twist θu is defined from the key point of the T −θ curve, Among these variables, the first three are those that show notable
in the post-peak descending branch, for which the beam no longer re- influence in the ultimate behavior of the beams, namely for the ultimate
sists high torques. For this study, the same criterion discussed and used deformations, provided that mb and y / x are not very far from unity
by Bernardo and Lopes in 2009 [6] is adopted. This criterion assumes (common situation for beams in pure torsion). This is also true for the
that the ultimate twist corresponds to the key point with 80% of the torsional ductility.
maximum torque. This situation applies to Beam G4 [8] illustrated in The aim of this study is to investigate the influence of each of the
Fig. 1. For the beams for which the last point of the T −θ curve corre- previously referred variables study on the torsional ductility of the re-
sponds to a torque higher than 80% of the maximum torque, the key ference beams. For this and from Table 1, groups of beams are created
point to define the ultimate twist is assumed to be the last point of the according to the chosen variable to be studied. In each group, the
T −θ curve (point with maximum twist). For instance, this situation variable study to be analyzed varies while the other variables are kept
applies for Beam B9 illustrated in Fig. 1. constant (or approximately). To study the influence of the variables fc ,
From the referred above, and by knowing the location of the key ρtot , mb and y / x on the torsional ductility index, the reference beams
point to define the ultimate twist, the value for θu was obtained for both were also separated into two groups: plain and hollow cross section.
the theoretical and experimental T −θ curves (θu,th and θu,exp , respec- For each group of beams, a minimum number of beams must exist to
tively). These values are presented in Table 1. provide a sufficient number of points in the graphs to show reliable
Table 1 also presents, for each reference beam, the theoretical and trends. This condition somewhat limited the number of analysis that
experimental torsional ductility index, μ θ,th and μ θ,exp respectively, could be performed.
computed from Eq. (1). From Table 1, it should be noted that
μθ,exp

μ θ,th

4.5

4.0

3.5
A: 0.00-0.99%
3.0 B: 1.00-1.99%
2.5 C: 2.00-2.99%

2.0 D: 3.00-3.99%
E: 4.00-4.99%
1.5
F: 5.00-5.99%
1.0
0 20 40 60 80 0 20 40 60 80 100
fc [MPa]
Fig. 2. Influence of fc in μ θ for groups of RC beams with similar ρtot (plain beams).

763
M.M. Teixeira, L.F.A. Bernardo Engineering Structures 168 (2018) 759–769

5.1. Concrete compressive strength capacity to redistribute the shear stresses (from the external to the
center of the cross section). This allow plain beams to sustain high
Fig. 2 shows the influence of the concrete compressive strength, fc , torques during a larger range for the deformations at the ultimate stage.
in the torsional ductility index μ θ for some groups with plain beams In hollow beams, the lowest capacity do redistribute shear stresses leads
having equal or similar total reinforcement ratio, ρtot . A range of 1% for to a faster degradation of the cross section and to a faster loss of the
ρtot was considered to provide a sufficient number of beams for each torque capacity. As a consequence, hollow beams show lower torsional
group. As previously referred, variables mb and y / x show to have a ductility.
small effect in the ultimate stage, namely in the torsional ductility, The results of this section show that, although it is accepted that, for
provided that they are not very far from unity. For this reason, and to similar plain and hollow beams, the concrete core has negligible in-
provide a sufficient number of beams and groups for the analysis, these fluence on the torsional resistance (as observed in previous studies, for
variables were allowed to vary in the following ranges: instance [8]), this is no longer true as far as torsional ductility is con-
0.90 ⩽ mb ⩽ 1.10 and 0.50 ⩽ y / x ⩽ 2.00 . The studied groups include the cerned.
following beams: Group A – VS9 and VS2-VQ2 [10]; Group B – J1 and
J2 [8] + B5UR1, B7UR1, B9UR1 and B12UR1 [12]; Group C: B3 and I3
5.2. Torsional reinforcement ratio
[8] + B12UR3 and B12UR5 [12]; Group D: J4, B4 and I4 [8]; Group E:
B5 and I5 [8]; Group F: B6 and I6 [8]. Fig. 2 presents both the ex-
Fig. 4 shows the influence of the total reinforcement ratio ρtot in the
perimental and theoretical torsional ductility indexes for the same re-
torsional ductility index μ θ (for both the experimental and theoretical
ference beams (μ θ,exp and μ θ,th , respectively).
values) for some groups with plain beams having equal or similar
As expected, Fig. 2 shows larger dispersion of the results for the
concrete compressive strength, fc . A range of 10 MPa for fc was con-
experimental torsional ductility indexes, when compared to the theo-
sidered for each group. Again, the following ranges were considered:
retical ones. Some dispersion is also observed for the theoretical results.
0.90 ⩽ mb ⩽ 1.10 and 0.50 ⩽ y / x ⩽ 2.00 . The groups include the fol-
This is probably related with the size of the interval that was considered
lowing beams: Group A – N1a, G3, N3, N2a, G4, G8, N4 and G5 [8];
for the torsional reinforcement ratio, ρtot , for each group. Fig. 2 also
Group B – A2, A3 and A4 [11]; Group C: I3, I4, I5 and I6 [8]; Group D:
shows that, for a given range of ρtot , torsional ductility tends to increase
B9UR1, B12UR3 and B12UR5 [12].
as the concrete compressive strength fc increases, regardless of being
From Fig. 4, it can be observed that, for a given range of the con-
normal or high strength concrete. For this latter, this constitutes a po-
crete compressive strength, the torsional ductility indexes, both ex-
sitive observation. This trend is observed for both the experimental and
perimentally (although with higher dispersion) and theoretically, de-
theoretical torsional ductility indexes. For this reason, the observed
creases notably as the total torsional reinforcement ratio increases, up
experimental trend can be considered reliable.
to a certain limit. This one seems to be in the range 2.0–3.0%. After this
Fig. 3 presents the same graphs as in Fig. 2, but now for RC hollow
limit, few beams continue to show a torsional ductility index greater
beams. The groups include the following beams: Group A – A2, A3, B2
than unity (this one corresponds to non-ductility). This can be due to
and C2 [6]; Group B – A4, A5, B3, B4, C3 and C4 [6]; Group C – B5, C5
the influence of the nonlinear concrete behavior under compression, as
and C6 [6]. Fig. 3 shows a different trend for hollow beams, that is, the
discussed in Section 4. According to the theoretical results (with less
torsional ductility (both experimentally and theoretically) tends to de-
dispersion), the value of the referred limit seems to depend on the level
crease as the concrete compressive strength increases. This result was
of concrete strength. This limit seems to increase, within the previously
already observed by Bernardo and Lopes 2009 [6] with the same hollow
referred range (2.0–3.0%), as the concrete strength increases.
beams and is reproduced here for comparison with the result for plain
Fig. 5 presents the same graphs as in Fig. 4, but now for RC beams
beams.
with hollow cross section. The groups include the following beams:
From Figs. 2 and 3, it is observed that, for a given range of ρtot , plain
Group A – A1, A2 and A3 [6]; Group B – B3 and B4 [6]; Group C – C1,
beams present higher torsional ductility when compared to hollow
C2, C3, C4 and C5 [6]. For hollow beams, Fig. 5 shows a similar trend as
beams. This observation confirms an important point previously dis-
the one previously observed for plain beams, and also similar to the
cussed in Section 1. In fact, from the T −θ curves, it is observed that for
trend observed in a previous study with the same beams [6]. However,
hollow beams the torque capacity decreases faster after the peak
the limit of ρtot for which the torsional ductility starts to stabilize (with a
torque, comparatively to plain beams. From the previous results, it
small value) seems to be less, in the range 1.0–1.5%. As for plain beams,
seems also that, with regard to torsional ductility, high strength con-
this limit seems to increase, within this range, as the concrete strength
crete is favorable for plain beams and unfavorable for hollow beams.
increases.
These observations are probably related to the influence of the
From Figs. 4 and 5, it seems also that the range for ρtot compatible
concrete core in plain beams for the ultimate stage, namely for the
with a notable ductile behavior is much larger for plain beams, when
deformation capacity. In such beams, concrete core allows for higher
compared with hollow beams. Again, this observation can be explained
μθ,exp

3.0
μ θ,th

2.5

2.0

A: 0.00-0.99%
1.5
B: 1.00-1.99%
C: 2.00-2.99%
1.0
0 20 40 60 80 100 0 20 40 60 80 100 120
fc [MPa]

Fig. 3. Influence of fc in μ θ for groups of RC beams with similar ρtot (hollow beams).

764
M.M. Teixeira, L.F.A. Bernardo Engineering Structures 168 (2018) 759–769

μθ,exp
4.5

μθ,th
4.0

3.5

3.0

2.5 A: 20.0-29.9 MPa

2.0 B: 30.0-39.9 MPa

C: 40.0-49.9 MPa
1.5
D: 70.0-79.9 MPa
1.0
0.0 1.0 2.0 3.0 4.0 5.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0
ρtot [%]

Fig. 4. Influence of ρtot in μ θ for groups of RC beams with similar fc (plain beams).

due to the influence of the concrete core, as previously discussed 5.4. Other variables study
(Section 5.1).
Finally, when compared with the results in Section 5.1, it can be For the remaining variables study, mb and y / x , it was not possible to
stated that the total reinforcement ratio, ρtot , shows higher influence on form groups with a sufficient number of beams, such that the variable
the torsional ductility, when compared to the concrete compressive to be study varied and the other ones were kept constant (or approxi-
strength, fc . mately). For this reason, no clear tendencies were observed, neither
theoretically nor experimentally. Moreover, high dispersion were ob-
served with only few points in the graphs, for both experimental and
5.3. Cross section type theoretical results. The authors believe that the observed high disper-
sion can be explained because the influence of the variables mb and y / x
To visualize better the influence of the concrete core in the torsional in the torsional ductility is very small. It should be negligible when
ductility, Fig. 6 shows the influence of the section type (plain or hollow) compared with the influence of the variables previously studied. For
in the torsional ductility index μ θ (for both the experimental and the- these reasons, no dispersion charts are presented in this section. It
oretical values) for beams with mb ≈ 1 and y / x = 1 (no values different should be referred that this point also contributed to reduce even more
from unity were found for hollow beams under torsion, see Table 1). In the number of reference beams effectively used in this study (see
the previous section it was observed that the total reinforcement ratio Table 1), when compared to the initial set of reference beams collected
ρtot shows higher influence on the torsional ductility, when compared to from literature (around 120).
the concrete compressive strength fc . For this reason, the results in
Fig. 6 are presented in terms of ρtot .The graphs include the following
beams: Plain beams – C4, C5 and C6 [8] + A2, A3 and A4 [11]; Hollow 6. Comparative analysis with codes of practice
beams – T1 and T2 [9] + VH1 and VH2 [10] + A1, A2, A3, A4, A5, B2,
B3, B4. C1, C2, C3, C4, C5 and C6 [6]. This section presents a comparative analysis between the maximum
In general, Fig. 6 confirms the conclusions stated in the previous and minimum values of the torsional reinforcement ratio that are pro-
Sections 5.1 and 5.2. A notable decrease of the torsional ductility is posed by the codes of practice, with the effective torsional reinforce-
observed for low torsional reinforcements until a limit is reached, ment ratio of the reference beams, in light of their torsional ductility
which is in the following range: 1.0–1.5% (for hollow beams) and indexes μ θ . Since it was previously observed (Section 5) that the ex-
2.0–3.0% (for plain beams). From this limit, some beams continue to perimental trends are reliable (they matched with the theoretical ones),
show some ductility (with μ θ in the range 1.5–2.0). The experimental then only the experimental torsional ductility indexes are used here.
results follow this trend for hollow beams. For plain beams, the ten- The aim is to check the suitability of the code’s rules to ensure the
dency is not so clear due to the dispersion of the points. torsional ductility for RC beams, in order to avoid premature or brittle
In general, it can be confirmed that, for a given reinforcement ratio, torsional failures due to insufficient or excessive torsional reinforce-
torsional ductility is higher for plain beams. ment.
The following reference codes of practice are considered: American
code ACI 318R-14 [17], European Codes MC 10 [22] and EC 2 [18],
and Canadian Standard CAN3-A23.3-14 [23].
μθ,exp

5.5
μθ,th

5.0
4.5
4.0
3.5
3.0
2.5 A: 40.0-49.9 MPa
2.0 B: 70.0-79.9 MPa
1.5
C: 90.0-99.9 MPa
1.0
0.0 0.5 1.0 1.5 2.0 0.0 0.5 1.0 1.5 2.0 2.5
ρtot [%]

Fig. 5. Influence of ρtot in μ θ for groups of RC beams with similar fc (hollow beams).

765
M.M. Teixeira, L.F.A. Bernardo Engineering Structures 168 (2018) 759–769

μθ,exp

μθ,th
5.5
5.0
4.5
4.0
3.5
3.0
2.5
2.0
Plain
1.5
Hollow
1.0
0.0 2.0 4.0 6.0 0.0 2.0 4.0 6.0 8.0
ρtot [%]

Fig. 6. Influence of the section type in μ θ .

RC beams with a brittle and premature torsional failure due to in- distributed area for the transverse reinforcement. Table 2 also presents
sufficient reinforcement should not be accepted by codes of practice. the effective transverse reinforcement ratio (ρt ) and the experimental
This should be guaranteed through the imposition of a minimum limit ductility index (μ θ,exp ). From Table 2 it is observed that, except for
for the torsional reinforcement. From the studied codes, only the ACI CAN3-A23.3-14 and ACI 318R-14, the codes of practice did not pre-
code explicitly imposes a minimum amount for the torsional long- dicted a brittle and premature failure due to insufficient reinforcement
itudinal reinforcement. Both ACI and Canadian codes impose a for any of the reference beams. This is consistent with the real failure of
minimum amount for the torsional transverse reinforcement. The other the reference beams used in this study. For few beams, CAN3-A23.3-14
ones (European codes) refer in general to the minimum amount of re- and ACI 318R-14 (only for one beam) failed because it predicted such a
inforcement for bending (longitudinal reinforcement) and shear failure. For all the codes of practice, some reference beams with real
(transverse reinforcement). ductile behavior are expected to have a brittle failure due to concrete
RC beams with a brittle failure due to crushing of the compressive crushing. For these beams, the codes failed their predictions.
concrete in the struts should also not be accepted by codes of practice. For each code, the results from Table 2 are analyzed below with the
This should be guaranteed through the imposition of a maximum limit help of graphs.
for the torsional reinforcement. This requirement can be directly or Figs. 7–10 present, for each of the analyzed codes of practice and
indirectly imposed. In general, codes of practice impose a maximum regardless of the concrete strength, the experimental torsional ductility
limit for the stress in the compressive concrete in the struts. This limit index, μ θ,exp , with the transverse reinforcement ratio, ρt . Graphs of
can be used to compute the maximum torsional reinforcement to Figs. 7–10 also show, for each range of values for ρt,min and ρt,max
guarantee this requirement. computed from the codes’ rules, the average (vertical lines) and the
In this study and for the aim of the present section, only beams with width of the range with the maximum and minimum boundaries values.
approximately balanced reinforcement condition are considered: Since the range for the values of ρt,max is large when all the reference
0.90 ⩽ mb ⩽ 1.10 (the same ones considered in Sections 5.1 and 5.2). beams are included, the beams are grouped according to some chosen
Moreover, the limits for the torsional reinforcement are applied only for range. This option allow to visualize better how many beams don’t
transverse reinforcement. comply with the upper code’s limit (ρt,max ). The range of values for ρt,min
In RC beams mainly under torsion the torsional reinforcement is is narrow, regardless of the used code. For this reason, a single range is
usually balanced, i.e. mb = 1. For such beams, both longitudinal and presented for ρt,min . Additionally, in Figs. 7–10 plain and hollow beams
transverse reinforcements participate equally for the torsional behavior are distinguished by using “plain and hollow markers”.
of the beams. For example as referred in Section 4, both reinforcements Fig. 7 shows that the interval defined by the maximum and the
will yield at the same time (before or after the concrete crushing in the minimum transverse reinforcement ratio computed from ACI 318R-14
struts). If too much transverse reinforcement exists, which can lead to a is very restrictive, mainly for ρt,max and for plain beams. Several plain
beams (almost regardless of ρt,max ) and some hollow beams do not
brittle failure, then too much longitudinal reinforcement also exist.
comply with the maximum limit from the code. In both cases, these
Then, for beams with balanced reinforcement, either longitudinal or
beams show a ductile behavior (mainly for plain sections). These ob-
transverse torsional reinforcement can be used as a key parameter to
servations contradict the code, for which a brittle failure due to con-
analyze the torsional ductility. In RC beams under torsion the shear
crete crushing should be expected.
effect is relevant and it seems to be more logical to relate its behavior
For MC 10, Fig. 8 shows similar results as the ones previously re-
with the transverse reinforcement. These reasons justify partially why
ported from Fig. 7, namely that several plain beams and some hollow
the transverse torsional reinforcement was chosen for this section.
beams do not comply with ρt,max computed from the code. It is observed
In addition to the previous justifications, for this section the
that the interval defined by ρt,min and ρt,max found by MC 10 is also very
minimum and maximum normative values for the allowed quantity of restrictive, mainly for ρt,max and for plain beams. When compared with
reinforcement are need. As far as specific rules for torsion are con- ACI 318R-14, MC 10 slightly restricts less ρt,min .
cerned, the majority of them are defined only for the transverse re- Fig. 9 shows that, when compared to MC 10 and ACI 318R-14, EC 2
inforcement. For instance, for the torsional longitudinal reinforcement, restricts less the transverse reinforcement ratio, mainly the maximum
ACI code only specifies the minimum quantity. The other analyzed limit (ρt,max ). As a result, more reference beams comply with the limits
codes don’t specify specific rules for the torsional longitudinal re- imposed by the code. Only few plain beams, with some ductile beha-
inforcement. From the aforementioned, one can conclude that it is not vior, do not comply with ρt,max . All hollow beams with ductile behavior
possible to perform complete comparative analyzes based only on the comply with the code’s limits. From this perspective, EC 2 is better.
torsional longitudinal reinforcement. Finally, Fig. 10 shows that, for CAN3-A23.3-14, the results are si-
Table 2 presents, for each reference beam and for each code of milar to the ones observed for EC 2, although it is more restrictive for
practice, the minimum and maximum transverse reinforcement ratio ρt,max and less ductile beams comply when compared to EC2. Again,
(ρt ,min and ρt,max ). These are computed from the minimum and maximum some plain beams and some hollow beams do not comply with ρt,max

766
M.M. Teixeira, L.F.A. Bernardo Engineering Structures 168 (2018) 759–769

Table 2
Transverse reinforcement ratios from codes of practice.
ACI 318R-14 MC 10 EC 2 CAN3-A23.3–14

Beam μ θ,exp ρt mb ρt,min ρt,max ρt,min ρt,max ρt,min ρt,max ρt,min ρt,max
% % % % % % % % %

B3 [8] 1.63 1.17 1.02 1.33 7.10(b) 0.63 3.83(b) 0.83 16.69 2.46 9.41(b)
B4 [8] 1.81 1.62 0.98 1.37 7.45(b) 0.66 4.11(b) 0.88 17.78 2.57 10.06(b)
B5 [8] 1.79 2.13 1.02 1.31 7.35(b) 0.65 4.15(b) 0.87 17.83(b) 2.53 10.15(b)
B6 [8] 1.44 2.61 1.05 1.32 7.03(b) 0.63 3.82(b) 0.84 16.67(b) 2.47 9.43(b)
C4 [8] 1.61 1.76 1.03 1.30 5.53(b) 0.59 3.07(b) 0.66 13.29 2.36 7.21(b)
C5 [8] 2.22 2.37 1.01 1.33 5.65(b) 0.61 3.12(b) 0.68 13.57(b) 2.42 7.37(b)
C6 [8] 1.65 3.20 0.95 1.38 5.99(b) 0.64 3.36(b) 0.71 14.47(b) 2.53 7.86(b)
G3 [8] 1.91 0.88 1.03 1.29 7.51(b) 0.59 3.86(b) 0.87 17.27 2.33 9.72
G4 [8] 1.55 1.20 1.01 1.34 8.01(b) 0.63 4.22(b) 0.94 18.79 2.49 10.65(b)
G5 [8] 1.78 1.60 1.00 1.32 7.73(b) 0.60 3.99(b) 0.89 17.80 2.39 10.02(b)
G7 [8] 2.38 0.94 0.98 1.38 8.59(b) 0.69 4.75(b) 1.02 20.85 2.66 11.94
G8 [8] 1.96 1.31 0.99 1.36 8.07(b) 0.64 4.25(b) 0.95 18.95 2.52 10.74(b)
I3 [8] 3.09 1.17 1.03 1.54 8.55(b) 0.81 5.45(b) 1.08 23.55 2.97 14.31
I4 [8] 1.67 1.62 0.96 1.68 9.40(b) 0.88 6.00(b) 1.18 25.91 3.24 15.75
I5 [8] 2.32 2.13 0.94 1.71 9.56(b) 0.90 6.11(b) 1.20 26.36 3.30 16.03(b)
I6 [8] 2.19 2.61 1.01 1.64 9.03(b) 0.86 5.73(b) 1.15 24.99 3.17 15.27(b)
J1 [8] 2.19 0.54 0.94 1.33 5.10 0.35 1.97(b) 0.47 9.09 1.76 4.83
J2 [8] 1.98 0.83 0.93 1.37 5.26(b) 0.37 2.05(b) 0.49 9.42 1.82 5.01(b)
J4 [8] 1.66 1.62 0.97 1.35 5.58(b) 0.42 2.33(b) 0.56 10.63(b) 1.92 5.71(b)
K2 [8] 1.54 1.03 0.99 0.78 5.00(b) 0.65 4.96(b) 0.66 13.97 1.51 6.91
K3 [8] 1.73 1.58 0.99 0.83 5.15(b) 0.66 4.98(b) 0.68 14.12 1.56 6.94(b)
K4 [8] 2.21 2.28 1.00 0.76 4.74(b) 0.60 4.56(b) 0.62 12.92(b) 1.42 6.33(b)
N1 [8] 4.29 0.62 1.01 0.74 4.61 0.35 2.40(b) 0.53 10.91 1.41 6.25
N1a [8] 4.21 0.62 0.98 0.76 4.62 0.35 2.37(b) 0.53 10.84 1.41 6.19
N2 [8] 2.10 1.13 0.96 0.79 4.97(b) 0.38 2.62(b) 0.58 11.91 1.52 6.85
N2a [8] 2.71 1.10 0.93 0.79 8.45 0.36 4.92(b) 0.55 17.50 1.46 11.25
N3 [8] 2.71 0.90 1.02 0.75 4.36(b) 0.33 2.17(b) 0.51 10.03 1.36 5.70
N4 [8] 2.26 1.42 0.96 0.77 4.54(b) 0.34 2.26(b) 0.52 10.44 1.40 5.93(b)
VB2 [10] 1.22 0.71 0.94 1.40 4.69(b) 0.29 1.86(b) 0.49 9.82 2.51 6.02
VB3 [10] 1.40 0.71 0.94 1.58 5.71(b) 0.37 2.73(b) 0.64 13.74 3.05 8.92
VB4 [10] 1.45 0.71 0.94 1.78 6.44 0.43 3.15(b) 0.74 16.65 3.44 11.36
VM2 [10] 1.57 0.65 0.96 1.90 6.77 0.44 3.12(b) 0.76 15.85 3.68 10.17
VM3 [10] 2.02 0.65 1.10 2.50 8.56 0.59 3.96(b) 1.02 20.58 4.83 13.53
VQ1 [10] 1.21 0.33 0.99 1.00 3.97 0.29 1.32(b) 0.39 7.09 1.52 4.32
VQ3 [10] 1.18 0.41 0.96 2.31 4.19 0.20 1.35(b) 0.40 7.80 3.38(c) 4.40
VQ9 [10] 1.20 0.45 0.99 3.14(c) 3.46 0.18 1.34(b) 0.37 7.22 4.84(c) 3.82
VS2-VQ2 [10] 1.52 0.37 0.94 1.75 4.98 0.28 1.68(b) 0.48 9.13 2.66 5.43
VS3 [10] 1.08 0.55 0.94 1.75 5.03 0.28 1.72(b) 0.49 9.32 2.70 5.56
VS4-VQ5 [10] 1.15 0.74 0.94 1.75 4.98(b) 0.28 1.68(b) 0.48 9.13 2.66 5.43(b)
VS9 [10] 1.26 0.35 0.94 1.33 3.63 0.20 1.18(b) 0.34 6.44 1.94 3.81
VS10-VB1 [10] 1.14 0.70 0.94 1.33 3.77(b) 0.21 1.27(b) 0.36 6.92 2.02 4.11(b)
A2 [11] 1.96 1.08 0.99 1.28 6.94(b) 0.56 4.04(b) 0.73 18.05 2.48 10.73
A3 [11] 3.17 1.22 1.00 1.41 6.41(b) 0.66 3.77(b) 0.81 17.13 2.71 10.05
A4 [11] 1.46 1.69 1.02 1.41 6.35(b) 0.66 3.71(b) 0.81 16.92 2.72 9.93(b)
B3 [11] 2.03 1.26 0.98 0.97 5.77(b) 0.66 4.86(b) 0.74 16.12 1.88 8.97
B4 [11] 1.87 1.73 1.02 0.98 5.68(b) 0.66 4.73(b) 0.75 15.83 1.89 8.81(b)
B5UR1 [12] 2.82 0.92 0.94 1.03 5.78(b) 0.66 4.86(b) 0.71 16.24 1.99 9.09
B9UR1 [12] 2.10 0.92 0.94 1.42 7.95 0.97 7.16 1.03 25.77 2.73 17.21
B12UR3 [12] 3.14 1.04 1.00 1.42 7.53 0.97 6.56(b) 1.03 24.28 2.74 16.07
B12UR4 [12] 2.81 1.10 1.10 1.47 7.40(b) 1.01 6.29(b) 1.08 23.85 2.84 16.12
B12UR5 [12] 2.51 1.41 0.90 1.45 8.15(b) 0.99 7.35(b) 1.06 26.46 2.81 17.84
H-12–12 [13] 3.40 1.22 0.90 2.35 13.10(b) 1.23 8.51(b) 1.69 39.89 4.54 29.01
H-14–10 [13] 3.80 1.38 0.98 1.80 11.76(b) 0.93 8.09(b) 1.28 35.21 3.47 24.32
N-12–12 [13] 3.11 1.22 0.90 1.58 8.81(b) 0.77 5.12(b) 1.05 22.36 3.05 13.12(b)
N-14–10 [13] 2.27 1.38 0.98 1.26 8.23(b) 0.61 5.00(b) 0.83 20.42 2.43 11.91(b)
NBS-43–44 [14] 4.59 0.44 1.03 1.61 8.85 0.78 5.07(b) 1.07 22.33 3.11 13.08
D3 [8] 1.55 1.17 1.02 1.28 6.09(b) 1.01 4.22(b) 0.81 11.15 2.38 6.90
D4 [8] 1.44 1.62 0.98 1.32 8.34(b) 1.10 6.97(b) 0.88 16.15 2.55 9.81(b)
T1 [9] 1.51 0.75 0.97 2.59 10.55 1.41 8.11(b) 1.46 21.48 5.00 13.32
T2 [9] 2.34 0.71 1.02 2.59 9.33 1.41 6.65(b) 1.46 18.89 4.99 11.78
VH1 [10] 1.21 0.33 0.99 1.25 5.82 0.65 3.10 0.44 8.93 1.80 4.87
VH2 [10] 1.94 0.67 0.99 1.25 11.64 0.65 7.73(b) 0.44 14.23(b) 1.80 9.73(b)
A1 [6] 3.45 0.19 1.05 2.04 8.12(b) 1.18 6.50 1.20 18.22 3.94(c) 12.01
A2 [6] 2.22 0.37 1.00 1.84 8.19 1.15 6.55 1.08 18.27 3.56 11.98
A3 [6] 1.47 0.49 0.96 1.77 8.28(b) 1.13 6.73(b) 1.04 18.33 3.42 11.95
A4 [6] 1.53 0.65 1.03 1.93 8.34(b) 1.20 6.81(b) 1.15 19.30 3.73 13.12
A5 [6] 1.45 0.83 1.11 2.02 8.37(b) 1.25 6.61(b) 1.20 19.19 3.90 12.95(b)
B2 [6] 1.73 0.40 0.99 2.24 10.16 1.46 8.59 1.36 24.58 4.32 18.03
B3 [6] 1.39 0.67 1.00 2.30 10.52(b) 1.53 9.01(b) 1.40 25.73 4.44 19.73
B4 [6] 1.32 0.89 1.09 2.48 11.17(b) 1.70 9.41(b) 1.51 27.32 4.78 21.19
C1 [6] 1.80 0.19 1.04 2.81 11.08 1.67 9.57 1.72 27.13 5.42(c) 22.54
(continued on next page)

767
M.M. Teixeira, L.F.A. Bernardo Engineering Structures 168 (2018) 759–769

Table 2 (continued)

ACI 318R-14 MC 10 EC 2 CAN3-A23.3–14

Beam μ θ,exp ρt mb ρt,min ρt,max ρt,min ρt,max ρt,min ρt,max ρt,min ρt,max
% % % % % % % % %

C2 [6] 1.55 0.37 1.01 2.61 10.89 1.61 9.51 1.61 26.62 5.04 22.53
C3 [6] 1.19 0.63 1.06 2.50 10.43(b) 1.57 8.92(b) 1.53 25.56 4.82 21.21
C4 [6] 1.21 0.86 1.07 2.65 11.03(b) 1.68 9.40(b) 1.63 27.00 5.12 22.39
C5 [6] 1.20 1.05 1.05 2.73 11.56(b) 1.74 10.03(b) 1.68 28.20 5.26 24.15
C6 [6] 1.30 1.34 1.08 2.41 10.43(b) 1.53 9.00(b) 1.48 25.60 4.65 20.74(b)
C100a [16] 2.31 0.79 1.04 3.17 20.25 2.97 20.39 2.30 49.64 6.12 40.95
D090a [16] 2.81 0.79 1.04 3.44 18.09 2.67 18.58 2.51 43.58 6.63 39.52

b
Brittle failure due to concrete crushing expected in these beams.
c
Brittle and premature failure due to insufficient reinforcement expected in these beams.
0.10 %
0.20 %

0.60 %
0.85 %
0.90 %
0.45 %

1.15 %

0.60 %

1.15 %

1.85 %
0.20 %
0.45 %

1.25 %

1.75 %

2.25 %

2.50 %
2.40 %
6.0 6.0

5.0 5.0

4.0 4.0
μθ,exp

μθ,exp
0.60% - 1.15%
3.0 0.45% - 0.60% 3.0
1.25% - 1.75%
0.60% - 0.85%
2.0 2.0 1.85% - 2.25%
0.90% - 1.15%
1.0 2.40% - 2.50%
1.0

0.0 0.0
0.0 1.0 2.0 3.0 4.0 0.0 1.0 2.0 3.0 4.0
ρt [%] ρt [%]

Fig. 7. Transverse reinforcement ratios versus ductility (ACI 318R-14 [17]). Fig. 10. Transverse reinforcement ratios versus ductility (CAN3-A23.3-14
[23]).

don’t comply with ρt,min computed from the code. However, as referred
0.05 %
0.15 %

1.10 %
0.35 %

0.60 %
0.90 %
0.95 %

6.0
before, these beams did not show a brittle and premature failure due to
5.0 insufficient reinforcement.
For the studied codes, it was observed that they all failed to predict
4.0
brittle failure due to concrete crushing for several reference beams,
μθ,exp

3.0 0.35% - 0.60% mainly for ACI 318R-14, MC 10 and CAN3-A23.3-14. This can be ex-
0.60% - 0.90% plained because some of these beams, mainly beams with higher ρt ,
2.0
0.95% - 1.10% show a ductile behavior prior to the yielding of the torsional re-
1.0 inforcement. This is probably due to the nonlinear behavior of the
compressive concrete before the peak load, due to the initiation of the
0.0
0.0 1.0 2.0 3.0 4.0 crushing phenomena (as discussed in Section 4). This phenomenon
ρt [%] seems not be properly incorporated by the codes through the upper
limit for the torsional reinforcement, in order to avoid brittle failure
Fig. 8. Transverse reinforcement ratios versus ductility (MC 10 [22]). due to excessive reinforcement. As previously stated, this limit is
computed indirectly from the maximum stress allowed for the com-
pressive concrete in the struts. For most of the studied codes, this limit
seems to be too conservative because it don’t cover the full range of RC
0.05 %
0.15 %

1.05 %

1.55 %

3.10 %
3.35 %
1.60 %

2.35 %
2.50 %

3.70 %

6.0
beams with torsional ductility.
5.0

4.0 7. Conclusions
μθ,exp

1.05% - 1.55%
3.0 In this article the torsional ductility of RC beams with rectangular
1.60% - 2.35%
2.0 cross section was studied. From the results obtained through this study,
2.50% - 3.10%
the following conclusions are drawn:
1.0 3.35% - 3.70%

0.0 – The torsional ductility index used in this study proved to be ade-
0.0 1.0 2.0 3.0 4.0 quate to characterize the torsional ductility of RC beams and to
ρt [%] perform comparative analysis between the beams;
– For plain RC beams, torsional ductility increases as the compressive
Fig. 9. Transverse reinforcement ratios versus ductility (EC 2 [18]).
strength increases, while the opposite tendency is observed for
hollow RC beams;
computed from the code, but in a small number when compared to ACI – The torsional ductility decreases notably as the total torsional re-
318R-14 and MC 10. When compared with all the other codes, CAN3- inforcement ratio increases. This is true until a limit is reached and
A23.3-14 is more restrictive as far as ρt,min is concerned. Few beams

768
M.M. Teixeira, L.F.A. Bernardo Engineering Structures 168 (2018) 759–769

from which a residual ductile behavior seems to be observed (which [4] Pang XB, Hsu TTC. Behavior of reinforced concrete membrane elements in shear.
is higher for plain beams). This limit seems to increase as the con- ACI Struct J 1995;92(6):665–79.
[5] Zhang LX, Hsu TTC. Behavior and analysis of 100 MPa concrete membrane ele-
crete strength increases and has a value in the range 2.0–3.0% for ments. J Struct Eng 1998;124(1):24–34.
plain beams and 1.0–1.5% for hollow beams; [6] Bernardo LFA, Lopes SMR. Torsion in HSC hollow beams: strength and ductility
– Similar RC beams with plain and hollow cross sections show dif- analysis. ACI Struct J 2009;106(1):39–48.
[7] Lopes SMR, Bernardo LFA. Twist behavior of high-strength concrete hollow beams –
ferent level of torsional ductility due to the influence of the concrete formation of plastic hinges along the length. Eng Struct 2009;31(1):138–49.
core. For a given reinforcement ratio, plain beams are more ductile [8] Hsu TTC. Torsion of structural concrete – behavior of reinforced concrete rectan-
than hollow beams. Plain beams also show a ductile behavior for a gular members. Torsion of structural concrete, SP-18. American Concrete Institute;
1968. p. 261–306.
larger range of the torsional reinforcement ratio; [9] Lampert P, Thürlimann B. Essais de Poutre en Béton Armé sous Torsion Simple et
– The total reinforcement ratio shows higher influence on the tor- Flexion Combinées (Torsionsversuche und Torsions-Biege-Versuche an
sional ductility when compared to the concrete compressive Stahlbetonbalken). Comitée Européen du Béton, B.I. no 71; 1969. p. 177–207.
[10] Leonhardt F, Schelling G. Torsionsversuche na Stahlbetonbalken. Bulletin No. 239.
strength;
Deutscher Ausschuss für Stahlbeton; 1974. 122 pp.
– For ACI 318R-14, MC 10 and EC2, it was observed that the rule to [11] McMullen AE, Rangan BV. Pure torsion in rectangular sections – a re-examination. J
define the minimum torsional transverse reinforcement for RC Am Concrete Inst 1978;75(10):511–9.
beams with approximately balanced reinforcement, in order to [12] Koutchoukali N-E, Belarbi A. Torsion of high-strength reinforced concrete beams
and minimum reinforcement requirement. ACI Struct J 2001;98(4):462–9.
avoid brittle and premature failure due to insufficient reinforce- [13] Fang IK, Shiau JK. Torsional behavior of normal- and high-strength concrete beams.
ment, is adequate. For CAN3-A23.3–14, this limit seems to be ACI Struct J 2004;101(3):304–13.
slightly excessive; [14] Chiu H-J, Fang I-K, Young W-T, Shiau J-K. Behavior of reinforced concrete beams
with minimum torsional reinforcement. Eng Struct 2007;29:2193–205.
– ACI 318R-14 and MC 10, as well as CAN3-A23.3–14 although to a [15] Peng X-N, Wong Y-L. Behavior of reinforced concrete walls subjected to monotonic
less lesser extent, are excessively restrictive with respect to the pure torsion – an experimental study. Eng Struct 2011;33:2495–508.
maximum torsional transverse reinforcement and lead to the un- [16] Jeng CH. Unified softened membrane model for torsion in hollow and solid re-
inforced concrete members modeling the entire pre- and post-cracking behavior. J
acceptance of RC beams with ductile behavior under torsion. This is Struct Eng 2015;141(10).
mainly true for plain beams, again with approximately balanced [17] ACI Committee 318. Building code requirements for reinforced concrete (ACI 318-
reinforcement. From the studied codes, EC 2 is the one that has 14) and commentary (ACI 318R-14). Detroit (MI): American Concrete Institute;
2014.
shown to be the better with regard to the maximum torsional [18] NP EN 1992-1-1. Eurocode 2: design of concrete structures – Part 1: general rules
transverse reinforcement. For the other studied codes, the value for and rules for buildings; 2010.
the maximum stress allowed for the compressive concrete in the [19] Bernardo LFA, Andrade JMA, Nunes NCG. Generalized softened variable angle
truss-model for reinforcement concrete beams under torsion. Mater Struct
struts seems to be too conservative and should be reviewed.
2015;48:2169–93.
[20] Pilakoutas K, Neocleous K, Guadagnini M. Design philosophy issues of fiber re-
References inforced polymer reinforced concrete structures. J Compos Constr
2002;6(3):154–61.
[21] Teixeira MM. Ductility of reinforced concrete beams under torsion. Master thesis.
[1] Bernardo LFA, Lopes SMR. Plastic analysis of HSC beams in flexure. Mater Struct Covilhã (Portugal): Department of Civil Engineering and Architecture, University of
2009;42(1):51–69. Beira Interior; 2015 [in Portuguese].
[2] Carmo RNF, Lopes SMR. Ductility and linear analysis with moment redistribution in [22] CEB-FIP MODEL CODE. Suisse: Comité Euro-International du Béton; 2010.
reinforced high strength concrete beams. Can J Civ Eng 2005;32(1):194–203. [23] CSA Standard. Design of concrete structures – A23.3-14. Missisauga: Canadian
[3] Pecce M, Fabbrocino G. Plastic rotation capacity of beams in normal and high- Standards Association; 2014.
performance concrete. ACI Struct J 1999;96(2):290–6.

769