You are on page 1of 11

Engineering Structures 206 (2020) 110174

Contents lists available at ScienceDirect

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

Torsion in concrete beams reinforced with GFRP spirals T


a,b a a
Abdeldayem Hadhood , Mohammed Gamal Gouda , Mohamed H. Agamy ,

Hamdy M. Mohameda,c, , Alaa Sherifa
a
Department of Civil Engineering, Faculty of Engineering, Helwan University, Cairo, Egypt
b
Department of Civil Engineering, Queen’s University, Kingstone, ON, Canada
c
Department of Civil and Building Engineering, University of Sherbrooke, Sherbrooke, QC, Canada

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

Keywords: Current FRP guidelines do not have enough evidence on the torsional performance of beams reinforced with
Concrete glass fiber-reinforced polymer (GFRP) reinforcement due to the lack of research and complexity of the problem.
Beam A review of the past work was presented in the state-of-the-art section to stand on mile-stone prospective
GFRP considering the variety and development of GFRP manufacturing. This study examined the rectangular spirals
Spirals
and rectilinear stirrups to reinforce concrete beams in the transverse direction under pure torsion loading. The
Code
test parameters included four levels of transverse reinforcement ratio and the type of transverse reinforcement in
Design
Experiment terms of rectilinear stirrups vs. spirals. The test results demonstrated that rectangular GFRP spirals provided
Torsion superior tensile resistance required to develop high ultimate torsional moments after cracking and hence can be
Truss effectively used. The failure of the tested beams was attributed to the rupture of GFRP rectangular spirals at bent
Bars portions. The test results of the spiraled specimens were discussed and compared with predictions from a de-
veloped analytical model and based on CSA equations. Two values of failure plane angle and stress levels were
considered. The 45-degree inclination angle provided safe predictions with an average experimental-to-practiced
result of 1.16 compared to 0.99 using the 37.5-degree. Design equations were proposed and validated for the ACI
code for a probable future incorporation.

1. Background truss model.


Early studies showed that steel-RC beams could develop higher le-
Reinforced Concrete (RC) beams in buildings and bridges can be vels of torque, after cracking, based on the amount and strength of their
subjected to different levels of torsion, and they should, therefore, have tensile longitudinal and transverse reinforcement as well as their
adequate resistance to torsion. Common examples are the act of ec- compressed concrete struts between inclined cracks [9–12]. The failure
centric loading on the connected cantilever slabs and the continuity of mechanism, post-cracking stiffness, and failure plane angle are sub-
beams on rigid floors. Beams subjected to significant torsion must in- stantially affected by the longitudinal ρl and transverse ρT reinforcement
clude enough transverse reinforcement in terms of closed stirrups or ratios. Beams with high steel reinforcement ratios are called “over-re-
continues spiral and longitudinal reinforcement. The principles of inforced” and fail due to the crushing of concrete before the yielding of
equilibrium, compatibility conditions, and materials stress-strain re- steel in either direction occurs [13]. Others with low steel reinforce-
lationships formulated the primary foundation to develop the space truss ment are called “under-reinforced” and fail in a ductile manner due to
model and skew-bending theory for torsion problems during the 1990s the yielding of steel in both the longitudinal and transverse directions.
[1–5]. The design of RC beams reinforced with steel (steel-RC) pertains Procedures for constructing the failure boundaries can be approached
to the crushing of concrete struts or the yielding of steel reinforcement. elsewhere [13]. The relation between the failure plane angle θ, ρl, and
Torsion transfers by shear stress on a plane perpendicular to the long- ρT can be established based on the thin-tube and space truss model,
itudinal axis of beams. These stresses cause inclined cracks that spiral which will be explained later in this paper.
around the beam at which the torsion reinforcement can provide tensile
resisting forces. Design provisions have been well established for tor- ρl fl
cot2 θ =
sion problems in building codes [6–8] based on the thin-tube space ρT fT (1)


Corresponding author at: Department of Civil Engineering, Helwan University, Cairo, Egypt/ Department of Civil and Building Engineering, University of
Sherbrooke, Sherbrooke, QC, Canada.
E-mail address: Hamdy.Mohamed@Usherbrooke.ca (H.M. Mohamed).

https://doi.org/10.1016/j.engstruct.2020.110174
Received 14 October 2019; Received in revised form 29 December 2019; Accepted 2 January 2020
Available online 10 January 2020
0141-0296/ © 2020 Elsevier Ltd. All rights reserved.
A. Hadhood, et al. Engineering Structures 206 (2020) 110174

Eq. (1) can be simplified to Eq. (2) for the under-reinforced beams from a past study on rectangular sections [24] into their investigation
since the stresses in the steel longitudinal and transverse directions are on L-shaped beams which were reinforced with adhesively bonded
equal to the yield stress. GFRP bars or mild-steel in the transverse direction. The ultimate stress
ρl and elastic modulus of the GFRP bars were 400 MPa and 36.7 GPa,
cot2 θ = respectively, while the average compressive strength of cubes after
ρT (2)
28 days was 25 MPa. Their testing revealed that their three GFRP-RC L-
The failure plane angle for the under-reinforced beams can be, shaped beams failed due to concrete crushing, although different ulti-
therefore, estimated. For example, under-reinforced beams with equal mate torques were developed when changed the diameter of their
ρl and ρT should fail at 45 degrees, by substituting in Eq. (2). bonded stirrups. Based on their limited tests, these stirrups were,
Glass fiber-reinforced polymer (GFRP) reinforcement demonstrated therefore, recommended as an alternative transverse reinforcement to
a decent behavior compared to steel in different problems [14–18]. either mild steel stirrups or bent GFRP stirrups since they were intact.
Whereas steel is isotropic, susceptible to electrochemical corrosion, and
yields, GFRP is anisotropic, non-corroding material, and has superior 3.2. Bent-manufactured FRP stirrups in the transverse direction
performance particularly in tension. FRP has also been used, in strips
and sheets, to enhance shear and torsional performances for RC beams Landmark studies on FRP-RC beams under pure torsion were con-
on strengthening and retrofitting applications [19–23]. The effective- ducted by Mohamed et al. [27] and Mohamed and Benmokrane
ness of the urged technique can significantly vary from case to case as it [28,29]. These studies used sand-coated FRP closed stirrups manu-
depends on several parameters [22–23]. The FRP full wrapping factured by Pultral Inc. [30] in forms of rectilinear or connected “C”
strengthening technique was found to be the most effective technique shaped stirrups. These studies investigated vital parameters such as the
for T-beams subjected to combined torsion and shear although it is not type of reinforcement (steel, GFRP, CFRP) and spacing of stirrups.
practical due to the lack of access to all sides of the impaired beam [21]. Three comparable beams of 250 mm width and 600 mm height and
GFRP, as an internal reinforcement, can develop high tensile strains, in different reinforcement type (steel, CFRP and GFRP) were compared
torsion problems, to maintain the structural integrity of the section. [29]. The compared specimens achieved similar torque at cracking. The
Very limited research has, however, addressed the torsional behavior of maximum torques at the peak were 53.5 kN m, 62.9 kN m and 52.7 kN
RC beams with GFRP bars and stirrups (GFRP-RC) due to the com- m for the steel-, CFRP- and GFRP-RC specimens, respectively, which can
plexity of the analysis and experimental testing. be considered close (20% maximum difference). The GFRP-RC spe-
cimen had, however, a reduced post-cracking torsional stiffness com-
2. Research motivations and objectives pared to its steel and CFRP counterparts. Their test results revealed that
the failure of the steel-, CFRP- and GFRP-RC beams with no transverse
The effectiveness of GFRP reinforcement is an essential issue for stirrups was controlled by concrete splitting. The CFRP- and GFRP-RC
torsional problems which has not yet been considerably addressed. The beams with a transverse-reinforcement ratio of 0.54% failed due to the
performance of GFRP spirals as a torsional reinforcement in the trans- rupture of the FRP stirrups at the bent portion, whereas the torsional
verse direction has not however been put to the test. Few publications failure of the CFRP and GFRP beams with high transverse-reinforce-
have appeared in recent years utilizing various commercial types of FRP ment ratios of 1.07% resulted from the concrete crushing in the diag-
transverse reinforcement. This study aimed at presenting a state-of-the- onal strut. It can be inferred from their study on GFRP specimens [28]
art review on the past research and investigating the performance of that beams with ρT ≤ 0.54% can be considered as under-reinforced,
GFRP spirals and bars in beams subjected to significant pure torsion. where they fail due to rupture of stirrups at bent portions. In some cases
(ρT ≤ 0.27%), the rupture of stirrups initiated by the appearance of
3. State-of-the-art review on FRP significant wide diagonal cracks which was attributed to the large
spacing of GFRP stirrups and their low modulus [28]. This study
Limited research was conducted to investigate the performance of eventually revealed that closed GFRP stirrups could develop strain le-
FRP reinforcement as an alternative to steel in concrete beams sub- vels up to 0.0085 at the marked straight portions, which was equivalent
jected to torsion [24–29,31]. The failure of beams in torsion is mainly to the stress level of 381 MPa. The strength of the bent portion was very
contingent on the amount and strength of reinforcement, especially in close to the yielding stress of steel (around 480 MPa).
the transverse direction, as reflected by the design equations on North- The influence of polypropylene fibers on the torsional capacity of
America’s codes. Referring to limited studies, various types of FRP were beams was addressed by Zhou et al. [31]. The surface texture and
used in the transverse direction in terms of FRP type (glass and carbon); bending technique of the closed GFRP stirrups were not reported. Three
cross-sectional shape (strips and bars); surface texture (undefined, comparable beams of 150 mm width and 200 mm height with different
ribbed or sand coated); and most critical, the method of closing stirrups concrete characteristics (ordinary concrete, OC; fiber-reinforced con-
(heat-bent, epoxy-adhesively bonded, double pairs of C-shaped, closed crete, FRC; and engineered cementitious composite, ECC) were com-
rectilinear). The authors reviewed the results of the past studies in pared. The compressive strength of each was 42.5, 40.2 and 32.8 MPa,
terms of failure mechanism and performance of FRP as a transverse respectively, and the corresponding tensile strength was 2.41, 2.37 and
reinforcement. 2.27 MPa, respectively. Their test results showed that the torsional
moment of the FRC and ECC beams were 10% and 40% greater than OC
3.1. Bonded GFRP bars in the transverse direction beams, respectively, although their concrete tensile strengths were si-
milar. That might raise a question on the methodology of estimating the
Considerable efforts were exerted to fabricate closed stirrups by contribution of FRC and ECC in torsion on the ACI [32] and CSA [33]
heat-bending GFRP bars [24]; and adhesively-bonding GFRP bars [26]. codes where it has been deemed, in these codes, to zero to simplify
The heating-bending procedure [24] was not favorable since the outer design especially with the involvement of shear.
fibers at the bent portions were overstressed, and the inner buckled.
The bond between these stirrups and the surrounding concrete at the 3.3. Steel spirals in the transverse direction
corners was also weak, causing a significant loss in strength. The ef-
fectiveness of these stirrups was not adequately examined since the Steel spirals enhance the confinement of concrete and provide
failure initiated in the compressed struts. Closed stirrups fabricated adequate resistance to the diagonal cracks due to torsion action if they
from GFRP bars were not either effective due to the weak bond between are used in the locking direction to the applied torque, as reported by
GFRP and concrete [25]. Deifalla et al. [26] incorporated other results past studies [34,35]. Beams with rectangular steel spirals in the locked

2
A. Hadhood, et al. Engineering Structures 206 (2020) 110174

The GFRP longitudinal bars were fabricated by a local manufacturer,


while the stirrups and spirals were fabricated by Pultral inc [30]. No. 4
GFRP bars were used as longitudinal torsional reinforcement for all the
beams; No. 3 GFRP closed stirrups and continues spirals were used as
transverse torsional reinforcement. The rectilinear GFRP stirrups were
closed, cut from continuous rectangular spirals with adequate overlaps,
as shown in Fig. 2. The rectangular spiral in each specimen was fabri-
cated in one part without lap splices. The tensile properties of the
longitudinal GFRP bars and straight portions of stirrups/spirals were
determined according to ASTM D7205 [36] and ACI 440.3R [37],
whereas the nominal values are reported in Table 1. Normal-weight
concrete was cast using a tilting drum mixer. The targeted compressive
strength after 28 days was 35 MPa with a maximum aggregate size of
20 mm. The actual compressive strength of concrete was determined by
ASTM C39/C39M [38] by testing 150 mm × 300 mm cylinders that
were cured adjacent to the beams under similar curing conditions, as
shown in Table 2.

4.2. Specimen design, fabrication and details

The beam specimens were designed to examine the influence of the


configuration of transverse torsional GFRP reinforcement on the gen-
eral behavior and torsional resistance. Six full-scale RC beams re-
inforced with GFRP bars were designed to provide different levels of
torsional resistance. Four were entirely reinforced with GFRP bars and
continuous GFRP spirals; one was entirely reinforced with GFRP bars
and stirrups, and the last beam had not any transverse reinforcement
and marked as the reference specimen. The test parameters included
the configuration of transverse torsional reinforcement (GFRP spirals
versus GFRP stirrups) and transverse torsional reinforcement ratio ρT
(in terms of spiral pitch). Four levels of ρT were considered in this study
where they varied between 0.36% and 0.89% (represented by spiral
spacings, s varied between 250 mm and 100 mm), as calculated by Eqs.
(3) and (4). The longitudinal reinforcement ratio ρl was maintained
constant at 0.72%, as calculated by Eq. (5).
At × ρo
ρT =
Acp × s (3)

ho bo
ρo = 2 × ( + )
sin φfront sin φtop (4)

Al
ρl =
b×h (5)

where At and Al are the nominal area of GFRP stirrups/spirals and


GFRP longitudinal bars, respectively; ρo , bo and ho are the perimeter,
width, and height of closed stirrups or continues spirals at centerlines
Fig. 1. GFRP reinforcement: (a) individual rectangular GFRP rectilinear closed
stirrups; (b) rectangular GFRP spiral; (c) cages of spirally-specimens. considering members’ inclinations as given in Eq. (4); Acp is the area
enclosed by the outside perimeter of the section which can simply be
calculated as the multiplication of the width, b and the height, h; φfront
direction achieved 14–18% higher ultimate torque over their counter- and φtop are inclination angles of a rectangular spiral in the vertical and
parts with stirrups [35]. However, using the spirals in the unlock di- horizontal directions, respectively. All specimens had a rectangular
rection of the external twist resulted in a decrease in the ultimate cross-section with dimensions of 3,000 mm long, 200 mm wide, and
torque by 19–23% compared to their counterparts with stirrups. To the 400 mm deep, as shown in Fig. 3. The concrete clear cover was main-
best knowledge of the authors, the behavior of GFRP spirals as trans- tained constant at 25 mm around all specimens. The details of re-
verse torsional reinforcement in beams has never been investigated. inforcement, dimensions of test specimens, and different configurations
are shown in Table 2. Each specimen was identified by a code denoting
4. Experimental program the tested parameters, where the characters B, G, GST, and GW refer to
beam, reinforced with GFRP spirals, GFRP stirrups, and without
4.1. Properties of materials transverse reinforcement, respectively; the number refers to the spiral
pitch/stirrups’ spacing in mm.
Sand-coated GFRP bars, rectilinear rectangular stirrups, and rec-
tangular spirals were used to reinforce the beams in the longitudinal 4.3. Test setup
and transverse directions, respectively, as shown in Fig. 1. The sand-
coated surface is known to be prominent in enhancing the bond per- The test setup was designed to allow for the prerequisite degrees of
formance between the bars/stirrups/spirals and surrounding concrete. freedom at ending supports and the applied torsional loading. Beam

3
A. Hadhood, et al. Engineering Structures 206 (2020) 110174

Fig. 2. Details of the tested specimens (dimensions in mm).

specimens were typically held in place on prefabricated bi-directional 4.4. Testing procedure and instrumentations
roller supports to release the in- and out-of-plane rotations, as shown in
Fig. 4a and b. The center-to-center beam spans were maintained at The beam specimens were tested at a two-point torsional loading
2,090 mm. The beam specimens were fastened and clamped to rigid setup, as previously described. The torsional loading was applied using
steel box frames close to ending supports in two opposite directions. a 1000 kN universal testing machine through a spreader beam on tor-
These frames were welded to 450 mm cantilever steel I-beams, which sional arms to produce the required torque. Two load cells were in-
transferred the torsional load through secondary load cells to the beam stalled between the spreader beam and the steel arms to equalize the
specimens, as shown in Fig. 4b. The torque was calculated by multi- applied loads on both sides. Electrical resistance strain gauges were
plying the load P at each side and the structural arm (500 mm) from the fixed on the torsional reinforcement (longitudinal bars, rectangular
load cell to the beam centerline. Overhang portions of 460 mm on each spirals, and closed stirrups) at critical locations within the test regions.
side were also created to provide adequate anchorage length and pre- Concrete gauges were also attached to surfaces at 45-degrees to record
clude slippage of reinforcement throughout the testing. concrete strains. The relative twist angles were determined at cracking
φcr and at the peak φu by capturing the lateral displacements at beam
ends, which were recorded by linear variable displacement transfor-
mers (LVDTs). The load cells, strain gauges, and LVDTs were connected

4
A. Hadhood, et al. Engineering Structures 206 (2020) 110174

Table 1
Mechanical properties of the GFRP reinforcement.
Bar Size Diameter Area Elastic Tensile Tensile Tensile
(mm) (mm2) Modulus Strength (MPa) Strain (%)
(GPa)

GFRP rectangular spiral/rectilinear stirrups


#3 9.5 71 44.8 f fu = 950 1.89
* f ftb = 474

f ftb = 485
GFRP bars
#4 13.5 143 35.2 f fu = 595 1.71

* f ftb is the ultimate tensile bend strength obtained from the B.5 test method
according to ACI 440 [37].

f ftb is the estimated tensile bend strength using the ACI 440.1R [32] equation
(† f ftb = (0.05(rb/ db) + 0.30) f fu ≤ f fu ), rb is the bend radius, and db is the stirrup
diameter.

over channels to the data-acquisition system.

5. Experimental results

5.1. Torsional behavior until cracking

This section discusses the behavior of the tested specimens before


Fig. 3. Cross-sections and transverse reinforcement details of the test specimens
cracking. The torque-twist relationships were developed based on the
(dimensions in mm).
twist at edges and the corresponding torque, as shown in Fig. 5a and b.
The torque was approached as the load times the lever arm concerning
the beam axis. The lever arm in the current experiments was 450 mm Figs. 6–8. As the load and strains were simultaneously increasing, in-
from the point of loading to the beam axis. The tested specimens de- clined cracks propagated along the section periphery. These cracks
monstrated almost identical initial stiffness kcr , with linear responses, started at long faces and rapidly spiraled around specimens. Funda-
until the cracking occurred where each specimen developed a different mental and modern theories for torsion presumed that the beam model
response. The cracking torques for all specimens varied between 11.6 is made up of 45-degree diagonal concrete struts, longitudinal and
kN m and 12.9 kN m which were higher (12–25%) than the ACI 318 transverse reinforcing truss elements. Beams can sustain the applied
(2014) cracking torque of 10.34 kN m, as per Eq. (6). This equation was torque as long as the resistance provided by these elements con-
developed based on the Thin-walled tube space truss analogy, in which tinuously maintained.
the concrete torsional resistance was neglected in calculations after the
1995 ACI version.
5.2.1. Beam with no transverse torsional reinforcement
Acp2 The absence of transverse reinforcement was fundamentally ex-
Tcr = 0.33 fc′ ( )
ρcp (6) amined in specimen BGW that was reinforced only in the longitudinal
direction. As the cracking torque approached, inclined cracks initiated
where Acp and ρcp are the area enclosed by the outside perimeter and on the long faces at 30 degrees, measured on the longitudinal axis, as
the outside perimeter of the concrete cross-section, respectively. shown in Fig. 9, and extended to the short faces. These cracks rapidly
propagated and widened, which undermined the concrete integrity to
5.2. Post-cracking behavior and failure mechanisms withstand compression in the perpendicular direction. The beam BGW
went, accordingly, in a softening response. The failure was controlled
Cracking typically occurs, as previously described, when the prin- by concrete splitting due to the lack of torsional reinforcement. A si-
cipal tensile stress, acting on any face, equals the tensile strength of milar response was reported in a past study ([28]; here will be referred
concrete. After cracking, strains on concrete surfaces and on long- to as “MB study”) where the specimen with only longitudinal GFRP
itudinal and transverse reinforcement gradually increased, as shown in reinforcement also failed by concrete splitting in a brittle manner with a

Table 2
Test matrix and specimen details.
Beam ID fc′ Bar Type Longitudinal Torsional Reinforcement Transverse Torsional
(MPa) Reinforcement

Bottom Top ρl Configuration db s ρT


(%) mm mm (%)

BG100 36.6 GFRP 2 No.4 2 No.4 0.72 RSP 9.5 100 0.89
BG150 36.6 GFRP 2 No.4 2 No.4 RSP 9.5 150 0.59
BG200 36.6 GFRP 2 No.4 2 No.4 RSP 9.5 200 0.44
BG250 36.6 GFRP 2 No.4 2 No.4 RSP 9.5 250 0.36
BGST-200 36.6 GFRP 2 No.4 2 No.4 RST 9.5 200 0.43
BGW 36.6 GFRP 2 No.4 2 No.4 — — — —

“RSP” stands for rectangular spirals and “RST” stands for rectilinear stirrups.

5
A. Hadhood, et al. Engineering Structures 206 (2020) 110174

Steel arm 500 mm


length
Fig. 5. Torque–twist relationships of specimens: (a) specimens with spirals; (b)
specimens with spiral/stirrup spaced at 200 mm.
Fig. 4. Test setup components for pure torsional loading: (a) Load application;
final installation of the specimen (Elevation View); (b) Fixation of the steel arm
to generate the applied torsional moment (Side View).

failure plane angle of 27 degrees. Their specimen exhibited, however, a


slight hardening response after cracking. That was attributed to the
high reinforcement ratio provided in the longitudinal direction
(ρl = 1.5%). The maximum torque of a beam with no transverse re-
inforcement should, therefore, be conservatively approached as the
cracking torque which can be estimated as per the elastic theory.

5.2.2. Beams with transverse torsional reinforcement


Unlike BGW, the transverse torsional reinforcement intersected the
initiated cracks at long and short faces and provided additional re-
sistance to resist the induced torque. These specimens developed,
therefore, hardening responses after cracking. The post-cracking rigid-
ities ku significantly decreased compared to their initial, where ku
varied between 83 kN m2 to 184 kN m2 (4–10% of their initial rigid-
ities), as shown in Table 3. This reduction was much pronounced at Fig. 6. Torque-diagonal compressive concrete strain curves of specimens.
specimens with larger spiral pitches. Similarly, high decreases in post-
cracking torsional rigidities were also reported for MB study [28], 6. Discussions on the experimental results
where their specimens (with GFRP stirrups) lost 97–99% of the initial
torsional rigidities. The previous sections described the failure modes of beams with
The measured strains on concrete surfaces varied between −0.0012 and without transverse torsional reinforcement where ρl = 0.72% and
and −0.0024, which were less than the ultimate compressive strain as ρT varied between 0.36% and 0.89%. The failure of specimens with
per the ACI 440.1R [32] and the CSA S806 [33]. The resistance pro- transverse reinforcement was dominant by GFRP spiral/stirrups rupture
vided by the torsional reinforcement continuously increased until the at the bent portions, as shown in Fig. 10, and not controlled by concrete
failure occurred, which was initiated by the rupture of the GFRP crushing. The maximum usable stress for GFRP spirals or stirrups
transverse reinforcement. The GFRP spirals/stirrups failed in the bent should not, therefore, exceed the strength of their bent portions. This
portions, which hold lower strength compared to the straight parts.

6
A. Hadhood, et al. Engineering Structures 206 (2020) 110174

section discusses the influence of test parameters and the efficiency of


GFRP transverse reinforcement on the torsional behavior and code
design requirements.

6.1. Influence of test parameters on torsional behavior

The ACI 440.1R [32] and the CSA S806 [33] limit the maximum
spacing of GFRP stirrups to approximately half of the flexural depth d to
ensure that at least one stirrup can intercept shear cracks. Four levels of
GFRP spiral spacings were investigated: 0.7d (ρT = 0.36%), 0.55d
(ρT = 0.44%), 0.4d (ρT = 0.59%), and 0.3d (ρT = 0.89%); while
maintaining the longitudinal reinforcement ratio at 0.72%. The heavily
reinforced specimens outperformed their counterparts in terms of
strength and post-cracking stiffness. The spacing between spiral turns
(s) controlled the post-cracking responses and the ultimate torque of the
tested specimens. The bar-chart presented in Fig. 11 shows the increase
Fig. 7. Torque-longitudinal reinforcement strain response. in the ultimate torque for each specimen after the cracking (△T ), as
given in Eq. (7). The GFRP spirals could enhance the ultimate torque of
the tested specimens, after cracking, up to 148% (at s = 0.3d).
Tu − Tcr
△T = × 100
Tcr (7)

where Tu and Tcr are the experimental ultimate and cracking torques
for each beam, respectively. Similar results were reported in [28],
where their beams achieved ultimate torques up to 200% over their
cracking torques.
Specimen BG200 exhibited higher post-cracking stiffness over
BGST200, as shown in Fig. 5b, and could, therefore, achieve higher
ultimate torque than BGST200. That was attributed to the superior
performance of spirals over stirrups in terms of confining concrete and
providing better distribution of spiral legs inside the cross-section. That
helped, consequently, to intersect and capture more cracks over a larger
distance. The failure of these specimens was similar and controlled by
the strength of the bent portions. The stirrups could, however, achieved
Fig. 8. Torque-spiral/stirrup strain response.
slightly higher strains over the spirals at the marked positions.

6.2. Effectiveness of GFRP torsional reinforcement

The level of ultimate torque for an RC beam depends on the effec-


tiveness of its entire torsional reinforcement and concrete core. The
design codes in North-America [6,7,33] followed the analogy of thin-
walled tube space truss to establish nominal torsional strength in which
the concrete core is ignored. The strain results demonstrated that
longitudinal and spirals were not activated until the cracking, after
which, the strains gradually increased until failure. The longitudinal
GFRP bars exhibited a variable level of stresses, as shown in Fig. 7, in
which higher levels achieved at closer pitches of spirals. The maximum
recorded strains (and estimated stresses) were 0.0059 (208 MPa),
0.0054 (190 MPa), 0.0048 (169 MPa) and 0.0044 (155 MPa) for
BG100, BG150, BG200, and BG250, respectively, where the stresses
were calculated by multiplying the measured strains by the GFRP ten-
sile modulus. These levels did not exceed 35% of the ultimate tensile
strength of the used GFRP bars. The results revealed that the GFRP bars
were intact during the tests and could develop the required forces to
maintain the equilibrium until the end of the tests. The strain responses
of spirals at mid-span are shown in Fig. 8. The GFRP spirals had a po-
sitive impact on the torsional behavior of all specimens and dominated
their failure. At the peak, the strains were 0.0084, 0.0077, 0.0072, and
0.0060 for BG100, BG150, BG200, and BG250, respectively, which
correspond to stress levels of 376 MPa, 345 MPa, 322 MPa, and
269 MPa, respectively. The experiments conducted by the authors and
Fig. 9. Failure plane angles of the tested specimens.
other studies revealed that the failure of beams subjected to torsion is
governed by the stress of rectangular spirals/rectilinear stirrups at the
bent portions. The maximum stress in transverse reinforcement should
not, therefore, exceed the ultimate tensile bend strength.

7
A. Hadhood, et al. Engineering Structures 206 (2020) 110174

Table 3
Experimental test results.
Beam ID At cracking At the peak Failure

Tcr φcr k cr Tu φu ku θexp Type


(kN m) (rad/m) (kN m2) (kN m) (rad/m) (kN m2) (Deg)

BG100 12.6 0.0066 2083 31.2 0.1079 184 50 Tension


BG150 12.7 0.0062 2048 21.2 0.0741 167 49 Tension
BG200 12.0 0.0061 1966 16.5 0.0652 125 45 Tension
BG250 11.6 0.0064 1818 12.9 0.0537 83 42 Tension
BGST-200 12.9 0.0060 2145 14.2 0.0811 125 43 Tension
BGW 11.9 0.0069 1724 — 0.0069 — 30 Splitting

q s = FfT , t cot θ (9)

The shear flow can be substituted by the toque multiplied by twice


the area enclosed by the shear flow path [39], which was formulated in
Eq. (10).

T
q =
2Ao (10)

From Eqs. (8) and (10), the ultimate toque of RC beams with GFRP
Fig. 10. Close-up views (rupture of GFRP spirals at failure). bars and spirals was established in Eq. (11).

AfL f fL + AfT f fT cos φfront


TL = 2Ao tan θ
po (11)

Likewise, the ultimate toque was established in Eq. (12) based on


the GFRP spirals in transverse reinforcement:

AfT f fT sin φfront


TT = 2Ao cot θ
s (12)

Also, the failure plane angle θ was finally developed by combining


Eqs. (11) and (12).

AfL f fL + AfT f fT cos φfront s


cot2 θ =
AfT f fT sin φfront po (13)

where Ao is the area enclosed by the centerline of the shear flow, as


given in Eq. (14); AfL is the area of GFRP longitudinal top and bottom
bars; AfT is the branch area of a GFRP spiral; s is the GFRP spiral pitch;
Fig. 11. Influence of GFRP transverse torsional reinforcement ratio on the ffL and ffT are the stresses on GFRP bars and spirals, respectively.
torque increase after cracking.
ho bo
Ao = 0.85 × ( × )
sin φfront sin φtop (14)
7. Theoretical investigation
ho and bo are the height and width of the plane section that is en-
7.1. Analytical model for GFRP-RC beams with spirals closed by the centerlines of GFRP spirals; φfront and φtop are the acute
angles of inclination of GFRP rectangular spirals, as shown in Figs. 2
The ultimate torque of GFRP-RC beams can be approached con- and 12.
sidering the inclination of GFRP spirals based on the thin-tube and
space truss model developed by Rausch [1]. The model consists of
7.2. Ultimate torsion in FRP North-America’ design guidelines
longitudinal and spiral reinforcement-resisting tension forces and con-
crete struts-resisting compression. The shear flow path follows the
The ACI 440.1R [32] and CSA S806 [33] limit the tensile strain in
centerline of the hoop reinforcement, as shown in Fig. 12. In this model,
GFRP shear reinforcement to 0.004 and 0.005, respectively, but not
the torsional moment T was assumed to be resisted by an internal
exceeding the strength of the bent portion. These values were adopted
torque resulting from the shear flow q, which developed in the center of
to prevent degradation of aggregate interlock and concrete in shear.
a shear flow zone with an effective wall thickness td. The equilibrium
Nevertheless, the CSA S806 [33] replaced the 0.005Ef by 0.4 times the
equations were then established in the longitudinal and transverse di-
ultimate tensile strength (ffu) of the straight portion for sections sub-
rections considering a failure plane angle θ, as given by Eqs. (8) and (9),
jected to shear and torsion which allowed for increasing the usable
respectively, with the aid of Fig. 12c. The force equilibrium along the
stress level. The experimental evidence of the current study indicated
longitudinal direction:
that the GFRP spirals achieved strain and stress levels up to 0.0084 and
q po = (FfL + FfT , l ) tan θ 376 MPa (0.4 ffu), respectively, which accurately hits the boundary
(8)
limit of the CSA S806 [33]. The following section summarizes the de-
Force equilibrium along the transverse direction: sign equations provided by the CSA S806 [33] and the proposed
equations for the ACI 440.1R [32].

8
A. Hadhood, et al. Engineering Structures 206 (2020) 110174

Fig. 12. Simplified torsional analysis based on the space truss theory.

7.2.1. CSA S806 [33] ⎧


A f
t yt

The CSA S806 [33] includes a design approach for sections sub-
⎪ 2Ao s cot θ (a)
Tnsteel =
jected to combined shear and torsion where the transverse reinforce- ⎨ 2A Al fy cot θ (b)
⎪ o po (18)
ment shall be equal to the sum of that required for shear and that re- ⎩
quired for the coexisting torsion. The factored torsional resistance Tr is The angle θ is permitted to take equal to 45 degrees for non-pre-
calculated as given in Eq. (15) such that to exceed the factored torsional stressed members. The yield stress in transverse and longitudinal steel
moment Tf . can be replaced by other stress levels adequate to the GFRP stirrups and
longitudinal bars as proposed by MB study [28]. The nominal torsional
ϕf AfT f fT strength of GFRP-RC beams can be then established.
Tr = 2Ao cot θ
s (15)
AfT f fT
TnGFRP = ⎧2Ao
where Ao = 0.85 Aoh and f fT is the smaller of 0.4 f fu and 1200 MPa. ⎨
⎩ s (19)
Additionally, the longitudinal reinforcement in tension/compression
shall be designed to resist the additional tension forces caused by shear Afl f fl
and torsion. In pure torsion problems, the longitudinal reinforcement in 2Ao
po (20)
each side of any beam must satisfy Eq. (16). Tr can now be formulated in
Eq. (17) in terms of longitudinal bars by rearranging Eq. (16) and The stress level in the GFRP transverse torsional reinforcement
considering AfL on the both sides. should be limited to control diagonal crack widths at service loads and
to avoid failure at the bent portion of the GFRP stirrup [28], similar to
Tf ph what is required for shear. It is, therefore, substantial to correlate the
AfL / side ⩾ 0.45 cot θ (at tension or compression sides)
2Ao f fL (16) stress value to the experimental results.

7.3. Comparison with experimental results


AfL f fl
Tr = 2Ao tan θ
0.9po (17) In this part, a comparison was made between the torque results
obtained from the experimental testing and those from the analytical
model, CSA S806 [33], and the proposed ACI 440.1R. In the analytical
model: the torque was calculated as the minimum of Eqs. (11) and (12);
7.2.2. ACI 440.1R [32] ffL was taken as the ultimate bar strength, as given in Table 1; θ was
The current version of the ACI 440.1R [32] does not include any obtained from experimental results. Two levels of spiral stress were
provision to beams subjected to torsion. However, the design procedure examined: (i) equal to the B.5 ultimate tensile bend stress fftb given in
in the ACI 318 [6] used for steel‐RC beams can be adapted for GFRP‐RC Table 1 and (ii) 0.4 ffu as specified in the CSA S806 [33] for torsion
beams and validated with the current tests. The torsional design problems. The prediction using CSA S806 [33] included the use of two
strength ϕ Tn must equal or exceed the factored torsional moment Tu . values for θ: (i) the general value of 45 degrees and (ii) 37.5 degrees as

9
A. Hadhood, et al. Engineering Structures 206 (2020) 110174

Table 4
Experimental-to-predicted ultimate torque assessment.
Beam ID Exp. Analytical Model S806-12 Proposed ACI 440.1R

ffT = fftb ffT = 0.4 ffu θ = 37.5 deg θ = 45 deg ffT = 0.004Ef ffT = 0.005Ef

Tu Tn Tu/Tn Tn Tu/Tn Tn Tu/Tn Tn Tu/Tn Tn Tu/Tn Tn Tu/Tn


(kN.m) (kN.m) (kN.m) (kN m) (kN m) (kN m) (kN m)

BG100 31.2 26.0 1.20 20.8 1.50 25.1 1.24 27.6 1.13 13.0 2.40 16.3 1.92
BG150 21.2 20.0 1.06 16.0 1.32 24.0 0.88 18.4 1.15 8.7 2.44 10.9 1.95
BG200 16.5 17.2 0.96 13.8 1.20 18.0 0.92 13.8 1.20 6.5 2.54 8.1 2.03
BG250 12.9 15.3 0.84 12.3 1.05 14.4 0.90 11.0 1.17 5.2 2.48 6.5 1.99

inferred in the substitutions in clause 8.4.7. Three values of ffT were were still intact and did not crush.
examined for the proposed model of the ACI 440.1R to estimate the 7. Using a 45-degree to account for the inclination angle of compres-
level of conservatism of each: (i) 0.004Ef as given for shear problems sive struts in the CSA S806-12 is recommended based on the safe
and (ii) 0.005Ef as specified in the CSA S806 [33] for shear problems. predictions to the experimental results. The 37.5-degree angle
The results of this comparison indicated that the considered models overestimated the torque of some specimens.
could predict the ultimate torque with different levels of conservatism. 8. An analytical model was proposed to the ACI 440.1R to predict the
The analytical model could not safely predict the experimental ultimate torsional capacity of beams. This model had a similar form as ACI
torque, particularly for specimens with low torsional reinforcement, 318-14, and it recommends using a maximum strain level in the
while using the B.5 ultimate tensile bend stress fftb. The experimental- transverse reinforcement of 0.5% to provide safe predictions with a
to-predicted ultimate torque ratios ranged from 0.84 to 1.20. Better reasonable level of conservatism.
predictions were obtained when the stress in spirals was taken as 0.4 ffu,
where the experimental-to-predicted ultimate torque ratios ranged from CRediT authorship contribution statement
1.05 to 1.50. That revealed that the effective stress for spirals in torsion
is less than the B.5 bend stress fftb. The CSA-Model had the best safe Abdeldayem Hadhood: Writing - original draft. Mohammed
predictions with θ = 45 degrees, which was closer to the experimental Gamal Gouda: . Mohamed H. Agamy: Methodology. Hamdy M.
inclination angles. Using θ = 37.5 degrees underestimated the ultimate Mohamed: Supervision, Methodology, Writing - review & editing. Alaa
torque for three specimens. Mohamed and Benmokrane [28] also in- Sherif: Supervision, Writing - review & editing.
dicated that the CSA-Model with θ = 45 degrees had reasonable but
rather conservative results where the experimental-to-predicted ulti- Declaration of Competing Interest
mate torque ratios ranged from 1.25 to 1.78 compared to 1.13 to 1.20
in this study. The proposed ACI-Model had the highest overestimations We declare that we have no conflict of interest.
among all approaches. That was returned to the low-stress level con-
sidered for this model. Using a spiral strain of 0.005Ef could safely Acknowledgment
predict the experimental ultimate torque with a high level of con-
servatism of all specimens. Table 4 summarizes the results of this The authors thank the technical staff of the structural laboratory in
comparison. the Department of Civil Engineering at Helwan University. The authors
would like to express their special thanks for Pultrall Inc. for the do-
8. Conclusions nation of GFRP materials. The first author would like to acknowledge
the funding received from the Fonds de recherche du Québec - Nature et
The current study investigated the performance of GFRP beams with technologies (FRQNT) on his postdoctoral research fellowship.
GFRP spirals as a transverse torsional reinforcement. Summing up the
results, the following remarks were drawn. References

1. The rectangular-spiraled GFRP beam (spirals in the locking direction [1] Rausch E. Design of Reinforced Concrete in Torsion and Shear, (Berechnung des
to the applied torque) outperformed its counterpart with stirrups in Eisenbetons gegen Verdrehung [Torsion] und Abscheren), Technische Hochschule,
Berlin, Germany, 53 pp. (in German); second edition in 1938; third edition in 1953,
terms of post-cracking stiffness and ultimate torque. This specimen “Drillung (Torsion) Schub und Scheren in Stahlbelonbau, Deutcher Ingenieur-Verlag
provided, however, a reduced angle of twist. GmbH,” Dusseldorf, Germany; 1929. p. 168.
2. The GFRP reinforcement provided in the longitudinal and transverse [2] Lampert P, Thürlimann B. Ultimate strength and design of reinforced concrete
beams in torsion and bending, publications. Int Assoc Bridge Struct Eng (Zurich)
directions could provide high tensile stresses, which allowed for the 1971;31-I:107–31.
development of ultimate torques up to 148% of those at cracking. [3] Lampert P, Collins MP. Torsion, bending, and confusion—an attempt to establish
3. The rectangular GFRP spirals developed a significant tensile re- the facts. ACI J 1972;69(8): p. 500–4.
[4] Collins MP, Mitchell D. Shear and torsion design of prestressed and non-prestressed
sistance up to the ultimate B.5 tensile bend strength, which was concrete beams. PCI J 1980;25(5):32–100.
approximately 474 MPa (0.5ffu). [5] Hsu TTC, Mo YL. Softening of Concrete in Torsional Members, Research Report No.
4. All beams failed in tension due to the rupture of GFRP transverse ST-TH-001-83, Department of Civil Engineering. Houston, TX: University of
Houston; 1983. p. 107 pp..
reinforcement at bent portions where ρT varied between 0.36% and
[6] American Concrete Institute (ACI). Building code requirements for structural con-
0.89%. crete. ACI 318-14 and commentary, ACI 318R-14. Farmington Hills, Mich; 2014.
5. Reducing the pitches of the rectangular spirals, in this study, en- [7] Canadian Standards Association (CSA), 2014. Design of concrete structures. CSA
hanced the post-cracking stiffness and ultimate torsional capacity up Standard A23.3-14, Rexdale, Ontario, Canada.
[8] Eurocode 2 (2005): Design of concrete structures EN1992-1-1. In Symposium
to the failure of spirals at bent portions. Eurocodes: Background and Applications, Brussels.
6. Based on point 4, using GFRP spirals as a transverse torsional re- [9] Lessig NN. Determination of Load-Carrying Capacity of Rectangular Reinforced
inforcement up to 0.89% for beams with 36 MPa concrete strength Concrete Elements Subjected to Flexure and Torsion, Trudy No. 5, Institut Betona i
Zhelezobetona (Concrete and Reinforced Concrete Institute), Moscow, pp. 5-28 (in
can be considered as under-reinforcement, where concrete struts Russian). Translated by Portland Cement Association, Foreign Literature Study No.

10
A. Hadhood, et al. Engineering Structures 206 (2020) 110174

371 (Available from S. L. A. Translation Center, John Crerar Library Translation 2019;201:109795.
Center, Chicago, IL); 1959. [24] Shehab HK, El-Awady HS, Husain M, Mandour S. Behavior of concrete beams re-
[10] Yudin VK. Determination of the load-carrying capacity of rectangular reinforced inforced by FRP bars under torsion. Proceedings of the 13th ICSGE. 2009. p. 6.
concrete elements subjected to combined torsion and Bending, Beton i [25] Ragab KS, Eisa AS. Torsion behaviour of steel fibered high strength self compacting
Zhelezobeton (Concrete and Reinforced Concrete), Moscow; 1962, No. 6, p. 265–9. concrete beams reinforced by GFRP bars. Int J Civil Sci Eng 2013;7(9):218–28.
[11] Collins MP, Walsh PF, Archer FE, Hall AS. Ultimate strength of reinforced concrete [26] Deifalla A, Hamed M, Saleh A, Ali T. Exploring GFRP bars as reinforcement for
beams subjected to combined torsion and bending, torsion of structural concrete, rectangular and L-shaped beams subjected to significant torsion: an experimental
SP-18. Farmington Hills, MI: American Concrete Institute; 1968. p. 279–402. study. Eng Struct 2014;59:776–86.
[12] Elfgren L, Karlsson I, Losberg A. Torsion-bending-shear interaction for concrete [27] Mohamed HM, Chaallal O, Benmokrane B. Torsional moment capacity and failure
beams. J Struct Divis 1974;100(8):1657–76. mode mechanisms of concrete beams reinforced with carbon FRP bars and stirrups.
[13] Leu LJ, Lee YS. Torsion design charts for reinforced concrete rectangular members. J Compos Constr 2015;19(2):04014049.
J Struct Eng 2000;126(2). pp. 210–218. H.K. [28] Mohamed HM, Benmokrane B. Torsion behavior of concrete beams reinforced with
[14] Mohamed HM, Benmokrane B. Design and performance of reinforced concrete glass fiber-reinforced polymer bars and stirrups. ACI Struct J 2015;112(5):543.
water chlorination tank totally reinforced with GFRP bars: case study. J Compos [29] Mohamed HM, Benmokrane B. Reinforced Concrete Beams with and without FRP
Constr 2014;18(1):05013001. Web Reinforcement under Pure Torsion. J Bridge Eng 2016;21(3):04015070.
[15] Hadhood A, Mohamed HM, Ghrib F, Benmokrane B. Efficiency of glass-fiber re- [30] Pultrall Inc., 2012. Composite Reinforcing Rods Technical Data Sheet. Thetford
inforced-polymer (GFRP) discrete hoops and bars in concrete columns under Mines, Canada, www.pultrall.com.
combined axial and flexural loads. Compos Part B Eng 2017;114:223–36. [31] Zhou J, Shen W, Wang S. Experimental study on torsional behavior of FRC and ECC
[16] Hadhood A, Mohamed HM, Benmokrane B, Nanni A, Shield CK. Assessment of beams reinforced with GFRP bars. Constr Build Mater 2017;152:74–81.
design guidelines of concrete columns reinforced with glass fiber-reinforced [32] American Concrete Institute (ACI). Guide for the Design and Construction of
polymer bars. ACI Struct J 2019;116(4):193–207. Structural Concrete Reinforced with FRP Bars. ACI 440.1R-15, Farmington Hills,
[17] Eladawy M, Hassan M, Benmokrane B, Ferrier E. Lateral cyclic behavior of interior MI; 2015. p. 44.
two-way concrete slab–column connections reinforced with GFRP bars. Eng Struct [33] Canadian Standards Association (CSA). Design and construction of building struc-
2019:109978. tures with fibre-reinforced polymers. CAN/CSA S806-12, Mississauga, Ontario,
[18] Ali AH, Mohamed HM, Chaallal O, Benmokrane B, Ghrib F. Shear resistance of RC Canada; 2012. p. 206.
circular members with FRP discrete hoops versus spirals. Eng Struct [34] Prakash S, Belarbi A, You YM. Seismic performance of circular RC columns sub-
2018;174:688–700. jected to axial force, bending, and torsion with low and moderate shear. Eng Struct
[19] Salom PR, Gergely J, Young DT. Torsional strengthening of spandrel beams with 2010;32(1):46–59.
fiber-reinforced polymer laminates. J Compos Constr 2004;8(2):157–62. [35] Chalioris CE, Karayannis CG. Experimental investigation of RC beams with rec-
[20] Chalioris CE. Torsional strengthening of rectangular and flanged beams using tangular spiral reinforcement in torsion. Eng Struct 2013;56:286–97.
carbon fibre-reinforced-polymers–experimental study. Constr Build Mater [36] ASTM. Standard test method for tensile properties of fiber reinforced polymer
2008;22(1):21–9. matrix composite bars. ASTM D7205-11, West Conshohocken, PA; 2011.
[21] Deifalla A, Ghobarah A. Strengthening RC T-beams subjected to combined torsion [37] American Concrete Institute (ACI). Guide Test Methods for Fiber-Reinforced
and shear using FRP fabrics: experimental study. J Compos Constr Polymers (FRPs) for Reinforcing or Strengthening Concrete Structures. ACI 440.3R-
2010;14(3):301–11. 04, Farmington Hills, MI; 2004. p. 40.
[22] Deifalla A, Awad A, Elgarhy M. Effectiveness of externally bonded CFRP strips for [38] ASTM. Standard test method for compressive strength of cylindrical concrete spe-
strengthening flanged beams under torsion: an experimental study. Eng Struct cimens. ASTM C39/C39M-12, West Conshohocken, PA; 2012.
2013;56:2065–75. [39] Bredt R. Kritische Bemerkungen zur Drehungselastizität, Zeitschrift des Vereines
[23] Hadhood A, Agamy MH, Abdelsalam MM, Mohamed HM, El-Sayed TA. Shear Deutscher Ingenieure, Band 40, No. 28, July 11, pp. 785-790; No. 29, July 18; 1896.
strengthening of hybrid externally-bonded mechanically-fastened concrete beams p. 813–7 [in German].
using short CFRP strips: experiments and theoretical evaluation. Eng Struct

11

You might also like