Engineering Structures 56 (2013) 372–383

Contents lists available at SciVerse ScienceDirect

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

Torsional behavior of ultra-high performance concrete squared beams

In-Hwan Yang a,1, Changbin Joh b,⇑, Jung Woo Lee b,2, Byung-Suk Kim c,3
a
Kunsan National University, Department of Civil Engineering, Kunsan, Jeonbuk 573-701, Republic of Korea
b
Korea Institute of Construction Technology, Structural Engineering Research Division, Goyang, Gyeonggi 411-712, Republic of Korea
c
Korea Institute of Construction Technology, SOC Research Institute, Goyang, Gyeonggi 411-712, Republic of Korea

a r t i c l e i n f o a b s t r a c t

Article history: This paper presents experimental results on the torsional behavior of ultra-high performance concrete
Received 23 October 2012 (UHPC) beams. Thirteen beam specimens with 300  300-mm cross section were cast from UHPC with
Revised 9 May 2013 the compressive strengths greater than 150 MPa. The experimental parameters were the specimens’ vol-
Accepted 14 May 2013
ume fraction of steel fibers, transverse reinforcement ratio, and longitudinal reinforcement ratio. The test
Available online 11 June 2013
results indicated that the beams’ initial cracking and ultimate torsional strength increased as the volume
fraction of steel fibers increased. The ultimate torsional strength and torsional stiffness after initial crack-
Keywords:
ing increased as the stirrup ratios increased, and ultimate torsional strength increased as the longitudinal
Ultra-high performance concrete (UHPC)
Steel fiber
rebar ratios increased. The effect of the quantity of transverse and longitudinal reinforcement on the
Crack angle of the diagonal compressive stresses was investigated. The results of this study provided valuable
Torsional strength data that could be used in future studies to develop computational models of the torsional behavior of
Diagonal compressive stresses UHPC beams and predict their ultimate strength.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction of steel fiber-reinforced concrete. The addition of steel fibers has

increased the ductile behavior [13–15], tensile strength [16], shear
Torsion can be a significant issue in concrete structural mem- strength [17–18], and torsional strength of concrete beams [19].
bers, such as eccentrically loaded beams, spandrel beams, and Gunneswara and Seshu [20] and Narayanan and Kareem-Palanjian
curved girders. These members must be able to resist torsion, [21] investigated the effects of steel fiber-reinforced concrete on
and design for torsion has become more important in concrete de- the torsional behavior of beams, while Mansur [22] and Chalioris
sign [1–7]. Certain tests of high-strength reinforced concrete and Karayannis [23] studied the torsional characteristics of fiber-
beams subjected to torsion have been reported in the literature reinforced concrete, the results of which have revealed the advan-
[8–11]. However, there are limitations to the use of conventional tages of fiber-reinforced concrete in torsion scenarios. Karayannis
and high-strength concrete in torsion scenarios, including its low [24] and Karayannis and Chalioris [25] also developed a nonlinear
tensile strength and ductility. According to the experimental work numerical method for studying the torsional behavior of steel fi-
by Bernardo and Lopes [12], the torsional ductility in high-strength ber-reinforced concrete beams.
concrete was low and the reinforcement ratio where ductility still Ultra-high strength concrete reinforced with steel fibers was
occurred was very narrow. In addition, the test results in the work developed more recently [26]. Richard and Cheyrezy [27] and Beh-
of Rasmussen and Baker [10] showed that the cracking of a high loul [28] studied the use of reactive powder concrete, which is a
strength concrete beam was more brittle than for a normal form of ultra-high performance concrete (UHPC). UHPC is an
strength concrete beam. To improve the low tensile strength and advanced cementitious composite that consists of a dense,
ductility of concrete steel fibers can be used. When steel fibers high-strength matrix and steel fibers. It is a promising material
are added to a concrete mix, they change the concrete from a brit- in construction and infrastructure rehabilitation because of its
tle material to a ductile one. remarkable properties, which include high tensile strength, high
Steel fiber-reinforced concrete has been increasingly applied in ductility [29], and low permeability [30]. Compared to conven-
this situation, and research efforts have been focused on the study tional concrete, UHPC possesses significantly increased tensile
strength [31] and post-cracking behavior [29,32]. It has also been
⇑ Corresponding author. Tel.: +82 31 910 0332. found that the high ductility of UHPC results from the bridging ef-
E-mail addresses: [email protected] (I.-H. Yang), [email protected] (C. Joh), fect of fibers across cracks and that the use of UHPC can limit the
[email protected] (J.W. Lee), [email protected] (B.-S. Kim). amount of rebar needed in a structure.
1
Tel.: +82 63 469 4752. Most studies on UHPC have focused on special concrete materi-
2
Tel.: +82 31 910 0581. als with characteristics that differ from those of conventional
3
Tel.: +82 31 910 0124.

0141-0296/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.engstruct.2013.05.027
 I.-H. Yang et al. / Engineering Structures 56 (2013) 372–383 373

concrete at the material level. However, few experimental test Table 1

results are available on the torsional behavior of UHPC beams. UHPC mix proportion by weight ratio.

Therefore, more information is needed to explore the structural Water- Cement Silica Filler Fine Water- Steel fiber by
behavior of UHPC beams subjected to torsion. binder fume aggregate reducing volume of
With this need in mind, the purpose of this study was to explore ratio admixture concrete

the torsional behavior of UHPC beams with compressive strengths 0.2 1.0 0.25 0.3 1.1 0.02 0%, 1.0%, and
greater than 150 MPa, and this paper describes the experiments 2.0%

performed on this topic. The experimental parameters included

steel fiber content, longitudinal steel reinforcement, and transverse
reinforcement. The experimental test results from the static load-
ing of the UHPC beams revealed the characteristics of the torsional
200
behavior of UHPC. The investigation of torsional behavior included
the consideration of cracking, failure patterns, and torsional capac- 180

Compressive Stress (MPa)

ity measurements. 160
140
2. Research significance 120
100
Experimental studies of the torsional characteristics of UHPC
beams have been extremely limited in the literature. This study 80 Specimen 1

investigated the characteristics of the structural behavior, cracking 60 Specimen 2

and failure patterns, cracking torsional strength, and ultimate tor- Specimen 3
40
sional strength of ultra high-performance concrete beams with Specimen 4
20 Specimen 5
squared solid sections when subjected to torsion. The experimental
data presented in this paper provide valuable information on 0
0 0.001 0.002 0.003 0.004 0.005
understanding the strength and structural behavior of ultra-high-
strength concrete squared beams reinforced with steel fibers. Strain
These data could also be useful for the development of design (a) UHPC containing a steel fiber volume fraction of 1%
equations that can predict the torsional strength of UHPC beams 200
in the future.
180
Compressive Stress (MPa)

160
3. Material properties
140
3.1. Materials and mix proportions 120
100
The UHPC investigated in this study was a type of reactive pow-
80 Specimen 1
der concrete [27]. Coarse aggregates were not included, and the
fine aggregates consisted of sand with a diameter of less than 60 Specimen 2
0.5 mm, which was the largest component of the UHPC. Portland 40 Specimen 3
cement was used as the binder, and the filler material was crushed
20 Specimen 4
quartz with an average diameter of 10 lm and a density of
2600 kg/m3. The silica fume, which was the smallest of the UHPC 0
0 0.001 0.002 0.003 0.004 0.005
components, had a diameter sufficiently small to fill the interstitial
Strain
voids between the cement and crushed quartz particles. The work-
(b) UHPC containing a steel fiber volume fraction of 2%
ability provided by the low water-to-cement ratio of the concrete
was maintained by the addition of a high-performance water- Fig. 1. UHPC compressive stress–strain.
reducing agent, a polycarboxylate superplasticiser with a density
of 1060 kg/m3. The steel fibers used for this concrete were straight
steel fibers with a diameter of 0.2 mm, and two different lengths of
16.5 and 19.5 mm were used together for each batch. The fibers
had a density of 7500 kg/m3 and a yield strength of 2500 MPa,
and they were added in volumes of 1% and 2% of the total mix vol-
ume. The proportions of the components in this UHPC mixture are
expressed in terms of their weight ratios in Table 1.

3.2. Mechanical properties of materials

The compressive strength of the UHPC was determined by the

compressive testing of cylindrical specimens with a diameter of
100-mm and height of 200 mm. The specimens were fabricated
simultaneously with each batch of test beams. A specially designed
axial deformation-measuring device was used for the compressive
strength test. Two parallel rings were rigidly attached to the cylin-
ders, leaving 100-mm spaces between the attachment points. The CMOD (mm)
upper ring held three linear variable displacement transducers
(LVDTs), the ends of which were supported by the lower ring. In Fig. 2. Load–CMOD curves (steel fiber content of 2%).
374 I.-H. Yang et al. / Engineering Structures 56 (2013) 372–383

this manner, the axial deformation of the cylinders was measured each batch and are presented in Fig. 2. These curves were subse-
accurately from the initiation of loading until failure. The loads and quently used for tensile constitutive material modeling by inverse
outputs from the three LVDTs were digitally recorded throughout analysis. The procedure of inverse analysis was explained in detail
testing. The loading was applied by a displacement control using in the work by Yang et al. [31]. The tensile behavior of the UHPC
a universal testing machine (UTM) with a 2000-kN capacity. was modeled with a linear elastic stress–strain relationship and a
The stress–strain curve of each batch of UHPC, shown in Fig. 1, softening relationship, as proposed by the AFGC design recommen-
was obtained based on the load–displacement relationship mea- dations [33]. As an example, the tensile stress–strain relationship
sured during the test, which was subsequently used to compute resulting from inverse analysis is shown in Fig. 3.
the sample’s compressive strength and elastic modulus. The FHWA Tensile stress–strain relationship for steel fiber-reinforced con-
[26] has suggested that the elastic modulus must be computed crete can be modeled by using direct tension test results. Karayan-
using the characteristic values that correspond to 10% and 30% of nis [24] approached the tensile response of steel fiber-reinforced
the ultimate compressive strength. concrete by two types of stress–strain curves resulting from direct
A series of tests was performed for each batch of UHPC prism tension test results. The first type was a tri-linear curve and was
specimens to determine their basic tensile behavior beyond the used for the case that the maximum post-cracking strength was
tensile strength. In the experimental program, a three-point bend- less than the cracking stress. The second type was an exponential
ing test was performed on notched prism specimens to investigate curve and was used for the case that the maximum post-cracking
the post-cracking behavior of the UHPC. This three-point bending strength was greater than the cracking stress.
test has been recommended by AFGC [33] for testing and modeling
the softening behavior of steel fiber-reinforced ultra-high perfor-
4. Test procedure
mance concrete.
The prism specimen had a height of 100 mm, width of 100 mm,
4.1. Test parameters, beam dimensions, and beam details
span of 300 mm, and length of 400 mm. A notch was cut into the
tension face of each beam to a depth of 10 mm before testing. Each
This experiment tested 13 UHPC beams with 300  300-mm
beam was rotated 90° from the casting surface and sawed through
cross sections, the details of which are listed in Table 2. The test
its width at midspan. The tests were controlled by the loading
parameters considered in the experimental program were the steel
point displacement method. The crack mouth opening displace-
fiber volume fraction, longitudinal reinforcement ratio, and trans-
ment (CMOD) was measured by a clip gauge with a 10-mm capac-
verse reinforcement ratio. The beams contained combinations of
ity attached to knife edges that had been epoxied to the bottom
two steel fiber volume contents (1% and 2%), while their longitudi-
face of the specimen on either side of the notch. In addition, the
nal reinforcement varied from 0% to 1.27% and their transverse
deflection of the prism was measured at midspan and the points
reinforcement varied from 0% to 0.70%. The ratios of steel rein-
measuring one third of the span’s distance from each support. In
forcement of the beams in Table 2 were calculated based on the
this manner, experimental load–CMOD curves were obtained for
following equations:
 
Asl
ql ¼  100 ð%Þ ð1Þ
12 Ac
 
10 Ast  pt
qt ¼  100 ð%Þ ð2Þ
Ac  s
Tensile stress (MPa)

8
where ql = longitudinal rebar ratio (%), qt = stirrup ratio (%), Ac = -
6 gross area of the beam cross section, Asl = total area of longitudinal
steel, Ast = area of one stirrup leg, pt = stirrup perimeter, and s = stir-
4
rup spacing.
The suggestion that beams in pure torsion should have a mini-
2
mum volumetric ratio of reinforcement greater than 1.0% is used in
0
the current ACI 318 minimum torsion provisions [34]. In the paper,
0.000 0.002 0.004 0.006 0.008 the total ratio of the stirrups and rebars varied from less than 1% to
Strain more than 1%. Finally, the total ratio of reinforcement ranged from
(a) Steel fiber content of 1% 0.56% to1.97%.
20 The beams in Table 2 were named to indicate the shape of their
cross section, steel fiber volume fraction, longitudinal reinforce-
ment ratios, and transverse reinforcement ratios. For example,
15 SS–F1–L56–S35 indicates that the beam has a square cross section
Tensile stress (MPa)

(SS), a steel fiber volume fraction of 1.0% (F1), a longitudinal rein-

forcement ratio of 0.56% (L56), and a transverse reinforcement ra-
10 tio of 0.35% (S35). Beam SS–F2–L00–S00 is a reference beam
containing no longitudinal or transverse reinforcement. Four series
of test beams were studied, and each series contained three beams
5 with various test parameters. Table 2 lists the mean values of the
compressive strength and elastic modulus of each UHPC beam,
which resulted from the test method mentioned in the previous
0
section. Furthermore, the mean values of the tensile strength from
0.000 0.002 0.004 0.006 0.008
Strain
the inverse analysis of each specimen are listed in Table 2.
(b) Steel fiber content of 2% The overall length of each test beam was 3.0 m, and it had a
cross section of 300  300 mm. The test region was comprised of
Fig. 3. UHPC tensile stress–strain relationship. 1.6 m in the middle of the beam, which allowed spiraled diagonal
 I.-H. Yang et al. / Engineering Structures 56 (2013) 372–383 375

Table 2
Test beam material properties and experimental parameters.

Test beam Fiber volume Rebar volume Stirrup volume Compressive Elastic modulus Tensile strength
content (%) content (%) content (%) strength (MPa) (MPa) (MPa)
SS–F2–L00–S00 2.0 0.00 0.00 192 45,090 18.8
SS–F1–L56–S00 1.0 0.56 0.00 174 43,550 9.8
SS–F1–L56–S35 1.0 0.56 0.35 174 43,550 9.8
SS–F1–L56–S70 1.0 0.56 0.70 174 43,550 9.8
SS–F2–L56–S00 2.0 0.56 0.00 192 45,090 18.8
SS–F2–L56–S35 2.0 0.56 0.35 192 45,090 18.8
SS–F2–L56–S70 2.0 0.56 0.70 178 40,810 15.4
SS–F2–L88–S00 2.0 0.88 0.00 178 40,810 15.4
SS–F2–L88–S35 2.0 0.88 0.35 184 45,010 13.4
SS–F2–L88–S70 2.0 0.88 0.70 184 45,010 13.4
SS–F2–L127–S00 2.0 1.27 0.00 182 44,050 13.6
SS–F2–L127–S35 2.0 1.27 0.35 182 44,050 13.6
SS–F2–L127–S70 2.0 1.27 0.70 182 44,050 13.6

cracks to occur along the beams and accommodated the test-rig of the beam to allow beam failure to occur in its middle test region
length of 0.7 m at each end. Further details on the beam specimens of the beam.
are provided in Fig. 4. The transverse and longitudinal reinforce- The placement of the UHPC was performed vertically in steel
ments were arranged according to test variables. Stirrups with a forms when fabricating the beam specimens, and it was compacted
nominal diameter of 10 mm (D10) were used for transverse rein- using external vibrators. The beam specimens were covered with
forcement, and four rebars with a nominal diameter of 13 (D13), plastic sheets upon completion of the UHPC casting and then
16 (D16), or 19 mm (D19) were used for longitudinal reinforce- steam-cured for 3 days at 90 °C. After 4 days of setting, the beam
ment. The rebars used were ribbed bars identified as SD 400 specimens were demoulded and air-cured until testing.
(400 MPa). To obtain the actual mechanical properties of the yield
stress, tensile strength, yield strain, and elastic modulus, four rebar 4.2. Test setup and instrumentation
specimens of each diameter were tested. The average yield stress
was 445 MPa, and the average tensile strength was 508 MPa. The The beam specimens were subjected to the test setup shown in
average yield strain was 2210  106, and the average elastic mod- Figs. 5 and 6. Steel frames were mounted at both ends of the
ulus was 198,000 MPa. Additional stirrups were placed at each end beams. The steel frame gripping the west end of the beam was

(a) Stirrup content = 0%

(b) Stirrup content = 0.35%

(c) Stirrup content = 0.70%

Fig. 4. Details of the beam specimens.
376 I.-H. Yang et al. / Engineering Structures 56 (2013) 372–383

(Unit : mm)

Beam specimen

Steel frames

Arc-bearing

West support East support

(a) Elevation Fig. 6. Actual test set-up for beam specimens.

Load
of the beam was allowed to slide in the longitudinal direction. The
vertical load was applied using a 1000-kN actuator with a torsional
arm to produce the torque. The torsional arm extended 800 mm
from the centroidal axis of the beam. All specimens were tested
under ram displacement control.
The instrumentation used for the beam specimens is shown in
Shim plates
Fig. 7 and consisted of linear variable differential transformers
Arc-bearing (LVDTs), strain gauges, and Demec gauges. An LVDT was used to
measure the deflection of the torsional arm at the east support
Note : and to calculate the rotation of the cross section of the beam spec-
Not to scale
imens. An aluminum frame was attached to the beam 525 mm
from the east support with an LVDT to estimate the rotation of
West support East support the cross section of the specimen.
(b) Section Electrical resistance strain gauges were used to measure the
concrete strain and were mounted on the surface of the rear side
Fig. 5. Details of the test set-up for the beam specimens. of the specimen at 45° and 135° to the beam axis, as shown in
Fig. 7. The length of the strain gauges mounted on the concrete
connected to the laboratory’s strong floor, and the frame gripping was 60 mm. These strain gauges were bonded with epoxy after
the east end of the beam was supported by an arc-bearing support. the concrete surface had been ground for planeness. Individual
This arc-bearing support allowed the beam to rotate while the gauges were used until their readings became unreliable due to
beam was loaded. To avoid longitudinal restraint, the east support cracking in the underlying concrete and were capable of detecting

East Support
(Unit : mm) West Support

Note : Not to scale

Section A-A Section B-B

Fig. 7. Instrumentation system.

 I.-H. Yang et al. / Engineering Structures 56 (2013) 372–383 377

strain due to small crack openings at each location. However, the 140
readings of the concrete strain gauges became unreliable when a
localised macrocrack developed. 120
C3
The test was performed until beam failure, and a remarkable
loss of load-carrying capacity and increased deformation were ob- 100
C4 C6

Torque (kN-m)
served. The applied load, angle of twist of the beam axis, strain in
C1 80
the concrete surface, development and propagation of cracks, and C2
width of each crack were recorded until beam failure. C5 60

40
5. Test results
20
5.1. Cracking behavior
0
The propagation of cracks and failure patterns were observed as -10000 -5000 0 5000 10000 15000 20000 25000
part of this study, and the test results indicated tensile cracking Microstrain
behavior and tensile fiber pullout behavior around the UHPC. No
cracks were observed when the load increased linearly at the Fig. 8. Torque–strain curve (beam SS–F2–L127–S70).

beginning of the test. Until the point of cracking, the behavior of

the beam was essentially elastic and torsion was mainly resisted spirally around the beam. When the beam reached its peak torque,
by the concrete. The measured diagonal surface concrete strains one specific crack grew significantly wider than any other cracks in
at 45° and 135° in beam SS–F2–L127–S70 are shown in Fig. 8. In the beam. Mechanically, the fibers bridging the cracks might expe-
this figure, tension occurred along the direction 45° to the beam rience more stress than ones where cracks did not happen. The
axis (C2, C4, and C6) and compression occurred along the direction steel fibers bridging the major cracks were pulled out of the UHPC
135° to the beam axis (C1, C3, and C5). Prior to cracking, the sur- matrix, and a major diagonal crack resulted in the tensile failure of
face concrete strains were of similar magnitude but with opposite the beam.
signs. As an example, the crack behavior of beam SS–F2–L56–S70 was
The first cracks occurred diagonally and spirally around the test examined, as shown in Fig. 9. The first crack was observed diago-
region because the principal stress distribution in a beam sub- nally at a torque of 79 kN m. New diagonal cracks formed as the
jected to pure torsion consisted of diagonal tension and compres- torque was increased, and cracks were observed spirally around
sion. According to elasticity theory, the cracks of a beam the span. The width of crack also increased as the load increased
subjected to pure torsion occur on the larger faces of a rectangular and measured 0.2 mm at a torque of 92 kN m. As the torque neared
cross section because these faces have the largest shear stresses. In its peak, the crack width increased significantly and a major diag-
this test, the first cracks were observed on all four sides almost onal crack occurred. After the peak torque of 110 kN m was
simultaneously because the cross sections of the specimens were reached, the torque decreased gradually and the width of the major
square. The first cracks were barely visible, and their presence crack increased suddenly. The width of this crack measured
was clearly indicated by an audible signal during the test. 11.0 mm at a torque of 86.4 kN m on the descending path of the
New diagonal cracks occurred as the load was increased, and torque–twist curve. The failure mode indicated a diagonal tensile
the number of cracks parallel to the initial cracks increased until failure in the beam. The steel fibers were pulled out along the
the ultimate state was reached. In this manner, cracks developed length of the crack, as shown in Fig. 9d.

Fig. 9. Typical crack pattern (beam SS–F2–L56–S70).

378 I.-H. Yang et al. / Engineering Structures 56 (2013) 372–383

Table 3
Test results for cracking parameters.

Test beam ql (%) qt (%) fly (MPa) fty (MPa) (ql/qt)  (fly/fty) wcr (mm) scr (mm) h (deg)
SS–F2–L00–S00 0.00 0.00 – – – – – 44
SS–F1–L56–S00 0.56 0.00 442 – – 0.53 20.6 50
SS–F1–L56–S35 0.56 0.35 442 445 1.59 0.49 19.2 38
SS–F1–L56–S70 0.56 0.70 442 445 0.79 0.47 15.3 49
SS–F2–L56–S00 0.56 0.00 442 – – 0.43 15.9 27
SS–F2–L56–S35 0.56 0.35 442 445 1.59 0.38 15.1 38
SS–F2–L56–S70 0.56 0.70 442 445 0.79 0.37 13.0 39
SS–F2–L88–S00 0.88 0.00 442 – – 0.35 14.9 34
SS–F2–L88–S35 0.88 0.35 442 445 2.50 0.33 14.5 52
SS–F2–L88–S70 0.88 0.70 442 445 1.25 0.25 13.8 49
SS–F2–L127–S00 1.27 0.00 442 – – 0.27 14.7 35
SS–F2–L127–S35 1.27 0.35 442 445 3.60 0.25 13.2 46
SS–F2–L127–S70 1.27 0.70 442 445 1.80 0.24 12.2 52

The experimental values of the mean crack width (wcr) and crack spacing. These results showed that the fiber contents in the
crack spacing (scr) at the ultimate state are given in Table 3. Both beams were effective at promoting multiple cracking behaviors
the mean crack width and crack spacing were measured in the and stress redistribution after the initial cracking.
middle region of each beam. The UHPC beams consistently exhib- A comparison of the cracking and failure patterns in beams with
ited small crack widths for the ultimate torque level. It was shown different stirrup ratios of 0.0%, 0.35%, and 0.70% is shown in Fig. 10.
that the higher fiber content corresponded to a smaller crack The figure indicates that more tightly spaced cracks occurred as the
width. This result was because the steel fibers across the cracks re- stirrup ratio increased. These results suggest that stirrups in beams
duced the stress in the tensile steel reinforcement. It was also were effective at promoting multiple cracking behavior and the
shown that the higher fiber content corresponded to the smaller redistribution of stresses after initial cracking.

Fig. 10. Comparison of crack patterns.

 I.-H. Yang et al. / Engineering Structures 56 (2013) 372–383 379

100 72.0 kN m when it contained 2% steel fiber content, while beam

90 with 1% steel fiber content had a cracking torsional strength of
65.6 kN m. In comparison, the ultimate torsional strength of the
80
beam with 2% steel fiber content was 85.6 kN m, while that of
70 beam with 1% steel fiber content was 75.3 kN m. Beam with a stir-
Torque (kN-m)

60 rup ratio of 0.70% and longitudinal rebar ratio of 0.56% had a crack-
ing torsional strength of 79.2 kN m when it contained 2% steel fiber
50
content, while beam with 1% steel fiber content had a cracking tor-
40 sional strength of 66.4 kN m. In comparison, the ultimate torsional
30 strength of beam with 2% steel fiber content was 109.8 kN m, while
20 that of beam with 1% steel fiber content was 86.7 kN m. However,
the cracking and ultimate torsional strength of beam SS–F2–L56–
10
S00 were less than those of beam SS–F1–L56–S00. The torque–
0 twist relationship for beam SS–F2–L56–S00 is shown in Fig 12b.
0.00 0.01 0.02 0.03 0.04
The torque of this beam suddenly decreased after the peak torque,
Angle of twist (rad/m) whereas the other beams that included steel fibers did not show
Fig. 11. Torque–twist curve (beam SS–F2–L00–S00).

80

70
5.2. Contribution of steel fibers to the post-cracking behavior
60
The torque–twist relationship for beam SS–F2–L00–S00 is

Torque (kN-m)
50
shown in Fig. 11. The beam did not contain conventional torsional
40
reinforcement, such as stirrups and longitudinal rebars. As ex-
pected, the torque–twist curve was linear before the beam cracked. 30
Its cracking torsional strength was 79.1 kN m. The initial cracking
20
torsional strength was calculated with the torque at the end of
the initial linear zone in the torque–twist curve. After the initial 10
cracking, the torsional capacity increased to the ultimate torsional 0
strength of 88.5 kN m, which was the peak torque on the curve. 0.00 0.01 0.02 0.03 0.04 0.05 0.06
The test results indicate that the UHPC beam showed significant Angle of twist (rad/m)
improvement in its post-cracking behavior, particularly after the (a) Beams with ρl = 0.56% and ρ s = 0%
peak torque. The post-cracking behavior of the UHPC beam was
100
different from that of the conventional concrete beam without
steel reinforcement, which failed abruptly by crushing. The addi- 90

tion of steel fibers was essential to the improvement of the post- 80

cracking behavior in the UHPC beam without conventional steel 70
Torque (kN-m)

reinforcements because steel fibers bridged the cracks and pro- 60

vided enhanced torsional capacity. 50
40
5.3. Effect of steel–fiber content on torsional behavior
30
20
Table 4 shows the initial cracking torsional strength and ulti-
mate torsional strength for each beam. The cracking and ultimate 10

torsional strength increased as the steel fiber content increased. 0

0.00 0.02 0.04 0.06 0.08 0.10
For example, beam with a stirrup ratio of 0.35% and longitudinal
rebar ratio of 0.56% had a cracking torsional strength of Angle of twist (rad/m)
(b) Beams with ρl = 0.56% and ρ s = 0.35%

120
Table 4
Test results for cracking and ultimate torque.
100
Test beam Initial cracking Ultimate state
Torque, Tcr Twist, hcr Torque, Tu Twist, hu 80
Torque (kN-m)

(kN m) (rad/m) (kN m) (rad/m)

SS–F2–L00–S00 79.1 0.0153 88.5 0.0183 60
SS–F1–L56–S00 68.0 0.0091 73.3 0.0111
SS–F1–L56–S35 65.6 0.0095 75.3 0.0355 40
SS–F1–L56–S70 66.4 0.0095 86.7 0.0448
SS–F2–L56–S00 61.6 0.0090 66.1 0.0113
20
SS–F2–L56–S35 72.0 0.0143 85.6 0.0363
SS–F2–L56–S70 79.2 0.0125 109.8 0.0515
SS–F2–L88–S00 87.2 0.0137 95.1 0.0218 0
SS–F2–L88–S35 88.0 0.0162 114.7 0.0461 0.00 0.02 0.04 0.06 0.08 0.10
SS–F2–L88–S70 82.4 0.0132 115.2 0.0438 Angle of twist (rad/m)
SS–F2–L127–S00 74.4 0.0167 85.2 0.0314
(c) Beams with ρl = 0.56% and ρ s = 0.70%
SS–F2–L127–S35 73.6 0.0114 109.6 0.0557
SS–F2–L127–S70 64.0 0.0118 119.3 0.0766
Fig. 12. Torque–twist curves of beams with different steel fiber contents.
380 I.-H. Yang et al. / Engineering Structures 56 (2013) 372–383

sudden decrease after the peak torque. The post-peak behavior of corresponding torsional stiffnesses of 618 and 784 kN m2, respec-
the UHPC beams is affected by the tensile resistance of the fiber tively, whereas beams SS–F1–L56–S35 and SS–F1–L56–S70, de-
crossing the crack plane, and the tensile resistance is also affected signed with a steel fiber content of 1%, had corresponding
by the fiber distribution during the mixing process and the place- torsional stiffnesses 373 and 575 kN m2, respectively. These results
ment of the UHPC. The non-uniform fiber distribution of beam SS– show that the cracked torsional stiffness increased as the steel fiber
F2–L56–S00 might decrease the cracking and ultimate torsional content increased. However, the torsional stiffness of beam SS–F2–
strength of the beam. L56–S00 was 1956 kN m2, whereas the torsional stiffness of beam
The torque–twist curves of the beams with different steel fiber SS–F1–L56–S00 was 2650 kN m2. This result shows that the
content (1% and 2%) are shown in Fig. 12. The torque–twist rela- stiffness of beam SS–F2–L56–S00 was less than that of beam SS–
tionship was linear until initial cracking occurred. These linear F1–L56–S00. As mentioned previously in this section, it might be
characteristics may have resulted from the linear compressive due to the non-uniform fiber distribution of beam SS–F2–L56–S00.
stress–strain relationship of the UHPC beam, as shown in Fig. 1,
and the linear tensile stress–strain relationship of the UHPC beam,
as shown in Fig. 3. The curves even showed an almost linear rela-
5.4. Effect of stirrup ratio on torsional behavior
tionship between the torque and twist after initial cracking had oc-
curred. Furthermore, the linear curve between the initial cracking
The torque–twist curves for beams with different stirrup ratios
and the maximum torque may have resulted from the linear tensile
are shown in Fig. 13. A comparison of the torque–twist curves of
stress–strain relationship after cracking, as shown in Fig. 3. The
beams with steel fiber content of 1% and a longitudinal rebar ratio
twist of the beam with 2% steel fiber content at failure was not sig-
of 0.56% is shown in Fig. 13a, which indicates that the ultimate tor-
nificantly different from that of the beam with 1% steel fiber con-
sional strength increased as the stirrup ratio increased. The in-
tent. It resulted from the similarity between the stress–strain
crease in the ultimate torsional strength of beams with stirrup
relationship of UHPC with 2% steel fiber content and that with
ratios of 0.35% (SS–F1–L56–S35) and 0.70% (SS–F1–L56–S70) was
1% steel fiber content.
3% and 18% of the ultimate torsional strength of a beam without
After cracking, the cracked torsional stiffness (kcr) is defined as
stirrups (SS–F1–L56–S00), respectively. A comparison of the tor-
follows.
que–twist curves of beams with steel fiber content of 2% and a lon-
 
T u  T cr gitudinal rebar ratio of 0.56% is shown in Fig. 13b, which also
kcr ¼ ð3Þ indicates that the ultimate torsional strength increased as the stir-
uu  ucr
rup ratio increased. The increase in the ultimate torsional strength
where Tcr is the measured cracking torque of beam specimen, Tu the of beams with stirrup ratios of 0.35% (SS–F2–L56–S35) and 0.70%
measured maximum torque of the beam specimen, ucr is the (SS–F2–L56–S70) was 29% and 66% of the ultimate torsional
measured angle of twist at initial cracking, and uu is measured strength of a beam without stirrups (SS–F2–L56–S00), respectively.
the angle of twist at maximum torque. Beams SS–F2–L56–S35, These results indicate that the increase of the ultimate torque of
and SS–F2–L56–S70, designed with a steel fiber content of 2%, had beams with steel fiber content of 2% at various stirrup ratios was

100 140
90
120
80
70 100
Torque (kN-m)
Torque (kN-m)

60 80
50
60
40
30 40
20
20
10
0 0
0.00 0.02 0.04 0.06 0.08 0.00 0.02 0.04 0.06 0.08 0.10
0.10
Angle of twist (rad/m) Angle of twist (rad/m)
(a) Beams with F = 1.0% and ρl = 0.56% (c) Beams with F = 2.0% and ρl = 0.88%

120 140

100 120

100
Torque (kN-m)
Torque (kN-m)

80
80
60
60
40
40

20 20

0 0
0.00 0.02 0.04 0.06 0.08 0.10 0.00 0.02 0.04 0.06 0.08 0.10 0.12

Angle of twist (rad/m) Angle of twist (rad/m)

(b) Beams with F = 2.0% and ρl = 0.56% (d) Beams with F = 2.0% and ρl = 1.27%

Fig. 13. Torque–twist curves of beams with different stirrup ratios.

 I.-H. Yang et al. / Engineering Structures 56 (2013) 372–383 381

greater than that of the ultimate torque of beams with steel fiber 100
content of 1%. 90
A similar phenomenon was observed in the test results for
80
beams with a steel fiber content of 2% and longitudinal rebar ratios
70
of 0.88%, and 1.27%, as shown in Fig. 13c and d. These figures indi-

Torque (kN-m)
cate that the ultimate torsional strength of the UHPC beams in- 60
creased as the stirrup ratio increased. In addition, the torsional 50
stiffness of the beams after initial cracking increased as the stirrup 40
ratio increased. As an example, the increase in torsional stiffness
30
after initial cracking of a beam without stirrups (SS–F2–L88–S00)
20
was insignificant. However, the increase in torsional stiffness after
initial cracking of beams with stirrup ratios of 0.35% (SS–F2–L88– 10
S35) and 0.70% (SS–F2–L88–S70) was more significant than that 0
0.00 0.02 0.04 0.06 0.08
of beam without stirrups.
Angle of twist (rad/m)
(a) Beams with stirrup ratio of 0% ( ρs = 0%)
5.5. Effect of longitudinal rebar ratio on torsional behavior
140
The torque–twist curves used to compare the characteristics for
beams with different longitudinal rebar ratios are shown in Fig. 14. 120
Beams SS–F2–L00–S00, SS–F2–L88–S00, and SS–F2–L127–S00 con-
tained longitudinal rebar ratios of 0%, 0.88%, and 1.27%, respec- 100

Torque (kN-m)
tively, but no stirrups. The ultimate torsional strengths of these
beams were 88.5, 95.1, and 85.2 kN m, respectively, as shown in 80
Table 4, which indicates that the ultimate torsional strength did
60
not increase significantly as the longitudinal steel ratio increased.
The increase of the ultimate torsional strength of beam SS–F2–
40
L127–S35 was 28% of the ultimate strength of beam SS–F2–L56–
S35, and the increase of the ultimate torsional strength of beam 20
SS–F2–L127–S70 was 9% of the ultimate strength of beam SS–F2–
L127–S70. The increase of the ultimate torsional strength ranged 0
0.00 0.02 0.04 0.06 0.08 0.10
between 9% and 28% as the longitudinal rebar ratio increased from
0.56% to 1.27%, while the increase of the ultimate torsional Angle of twist (rad/m)
strength ranged between 18% and 66% as the stirrup ratio in- (b) Beams with stirrup ratio of 0.35% ( ρs = 0.35%)
creased from 0% to 0.70%, as mentioned in the previous section.
140
A comparison between these values indicates that the influence
of the stirrups ratio on the ultimate torsional strength was greater
120
than that of the longitudinal rebar ratio.
The post-cracking behavior of the beams was affected by the 100
Torque (kN-m)

longitudinal rebar ratios, as shown in Fig. 14a. The twist at the ulti-
mate state increased as the longitudinal rebar ratio increased, and 80
the increase in twisting after cracking was more significant in
beams SS–F2–L88–S00 and SS–F2–L127–S00 than in beam SS– 60
F2–L00–S00. A similar phenomenon was observed in the beams
with stirrup ratios of 0.35% and 0.70%. Considering beams with a 40
stirrup ratio of 0.35%, the twist at the ultimate state of the beam
20
with a longitudinal rebar ratio of 1.27% (SS–F2–L127–S35) was
much greater than that of the beam with a longitudinal rebar ratio
0
of 0.56% (SS–F2–L56–S35), as shown in Fig. 14b. Considering 0.00 0.02 0.04 0.06 0.08 0.10 0.12
beams with a stirrup ratio of 0.70%, the twist at the ultimate state Angle of twist (rad/m)
of the beam with a longitudinal rebar ratio of 1.27% (SS–F2–L127– (c) Beams with stirrup ratio of 0.70% ( ρs = 0.70%)
S70) was much greater than that of the beam with a longitudinal
rebar ratio of 0.56% (SS–F2–L56–S70), as shown in Fig. 14c. Fig. 14. Torque–twist curves of beams with different longitudinal rebar ratios.

5.6. Angle of diagonal compressive stresses

reinforcements. The angle of the major diagonal crack versus the
The effects of the amount of torsional reinforcement on the an- parameter (ql/qt)  (fly/fty) for each beam is shown in Table 3.
gle of the diagonal compressive stresses were also investigated in Fig. 15a and b shows the angles of the major diagonal cracks at
this study. Multiple diagonal cracks were observed after the initial the ultimate states in beams SS–F2–L88–S00 and SS–F2–L88–S70.
cracking of each UHPC beam, and the number of cracks increased The major diagonal crack angle observed in beam SS–F2–L56–S00
as the applied torque increased. A major diagonal crack was ob- was 34°, and that in beam SS–F2–L56–S70 was 49°. Fig. 15c and
served at an angle of approximately 27–53° from the beam axis d shows the angles of the major diagonal cracks corresponding to
as the torque approached the ultimate torsional strength. The an- the ultimate states in beams SS–F2–L127–S00 and SS–F2–L127–
gle of the major diagonal crack depended on the orientation of S70. The major diagonal crack angle observed in beam SS–F2–
the principal stresses and also the angle of the diagonal L127–S00 was 35°, and that in beam SS–F2–L127–S70 was 52°.
compressive stresses depended on the transverse and longitudinal These test results indicate that the angle of the diagonal
382 I.-H. Yang et al. / Engineering Structures 56 (2013) 372–383

(a) SS-F2-L88-S00

(b) SS-F2-L88-S70

(c) SS-F2-L127-S00

(d) SS-F2-L127-S70
Fig. 15. Angle of major diagonal cracks.

compressive stresses from the beam axis increased as the quantity longitudinal steel ratio increased, the twist at the ultimate state
of stirrups increased. increased and the ultimate torsional strength did not increase
significantly.
6. Conclusions 5. The angle of the diagonal compressive stresses to the beam axis
was affected by transverse and longitudinal reinforcement.
This paper presents an experimental study on the torsional Major diagonal cracks were observed at angles measuring 27–
characteristics of UHPC beams reinforced with steel fibers. The fol- 53° from the beam axis.
lowing conclusions were drawn from the test results of the beam
specimens. Acknowledgements

1. Initial cracking was observed to be diagonal and spiral along the This research was supported by a grant from a Strategic Re-
beam spans. Additional diagonal cracks propagated as the search Project (Design Technology for Ultra-High Performance
applied torque increased. One major crack became significantly Concrete) funded by the Korea Institute of Construction
wider than any other cracks in the beam when the torque Technology.
approached its ultimate value. The addition of steel fibers was
essential to the improvement of the post-cracking behavior of References
the UHPC beams because the steel fibers bridged the cracks.
2. Increasing the volume fraction of steel fibers improved the [1] Hsu TTC. Torsion of reinforced concrete. Van Nostrand Reinhold Company;
cracking and ultimate torsional strengths of the beams. The 1984.
[2] MacGregor JG, Ghoneim MG. Design for torsion. ACI Struct J 1995;92(2):211–8.
cracking and ultimate torsional strength of beam with a stirrup [3] Nanni A. Design for torsion using steel fiber reinforced concrete. ACI Struct J
ratio of 0.70% and longitudinal rebar ratio of 0.56% increased by 1990;87(6):556–64.
19% and 27%, respectively, when it contained 2% steel fiber [4] Lopes SMR, Bernardo LFA. Twist behavior of high-strength hollow beams –
formation of plastic hinges along the length. Eng Struct 2009;31(1):138–49.
compared to beam containing 1% steel fiber. [5] Ali MA, White RN. Toward a rational approach for design of minimum torsion
3. The ultimate torsional strength increased as the stirrup ratio reinforcement. ACI Struct J 1999;96(1):40–5.
increased. Beam with a steel fiber content of 1%, a longitudinal [6] Algorafi MA, Ali AAA, Othman I, Jaafar MS, Anwar MP. Experimental study of
externally prestressed segmental beam under torsion. Eng Struct
rebar ratio of 0.56%, and a stirrup content of 0.70% showed 2010;32(11):3528–38.
improved ultimate torsional strength of 18% compared to that [7] Chiu HJ, Fang IK, Young WT, Shiau JK. Behavior of reinforced concrete beams
containing no stirrups. Beams with a steel fiber content of 2%, with minimum torsional reinforcement. Eng Struct 2007;29(9):2193–205.
[8] Fang IK, Shiau JK. Torsional behavior of normal-and high-strength concrete
a longitudinal rebar ratio of 0.56%, and stirrup contents of
beams. ACI Struct J 2004;101(3):304–13.
0.35% and 0.70% showed improved ultimate torsional strength [9] Koutchoukali NE, Belarbi A. Torsion of high-strength reinforced concrete
of 29% and 66%, respectively, compared to that containing no beams and minimum reinforcement requirement. ACI Struct J
stirrups. 2001;98(4):462–9.
[10] Rasmussen LJ, Baker G. Torsion in reinforced normal and high-strength
4. The addition of longitudinal rebars was effective at improving concrete beams – part 1: experimental test series. ACI Struct J
the post-cracking behavior of the UHPC beams. As the 1995;92(1):56–62.
 I.-H. Yang et al. / Engineering Structures 56 (2013) 372–383 383

[11] Rasmussen LJ, Baker G. Torsion in reinforced normal and high-strength [24] Karayannis CG. Nonlinear analysis and tests of steel–fiber concrete beams in
concrete beams – part 2: theory and design. ACI Struct J 1995;92(2):149–56. torsion. Struct Eng Mech 2000;9(4):323–38.
[12] Bernardo LFA, Lopes SMR. Torsion in high-strength concrete hollow beams: [25] Karayannis CG, Chalioris CE. Experimental validation of smeared analysis for
strength and ductility analysis. ACI Struct J 2009;106(1):39–48. plain concrete in torsion. J Struct Eng 2000;126(6):646–53.
[13] Oh BH. Flexural analysis of reinforced concrete beams containing steel fibers. J [26] FHWA (US Federal Highway Administration). Material property
Struct Eng 1992;118(10):2821–36. characterization of ultra high performance concrete. US Department of
[14] Ashour SA, Wafa FF. Flexural behavior of high-strength fiber reinforced Transportation, Federal Highway Administration; 2006.
concrete beams. ACI Struct J 1993;90(3):279–87. [27] Richard P, Cheyrezy M. Composition of reactive powder concretes. Cem Concr
[15] Issa MS, Metwally IM, Elzeiny SM. Influence of fibers on flexural behavior and Res 1995;25(7):1501–11.
ductility of concrete beas reinforced with GFRP rebars. Eng Struct [28] Behloul M. Tensile behavior of reactive powder concrete (RPC). In: Proceedings
2011;33(5):1753–63. of the 4th international symposium utilization of high strength/high
[16] Campione G, Mangiavillano ML. Fibrous reinforced concrete beams in flexure: performance concrete; 1996. p. 1375–81.
experimental investigation, analytical modeling and design considerations. [29] Yang IH, Joh C, Kim BS. Structural behavior of ultrahigh performance concrete
Eng Struct 2008;30(11):2970–80. beams subjected to bending. Eng Struct 2010;32(11):3478–87.
[17] Mansur MA, Ong KCG, Paramasivam P. Shear strength of fibrous concrete [30] Schydt J, Herold G, Müller HS. Long term behavior of ultra high performance
beams without stirrups. J Struct Eng 1986;112(9):2066–79. concrete under the attack of chlorides and aggressive waters. In: Proceedings
[18] Meda A, Minelli F, Plizzari GA, Riva P. Shear behavior of steel fibre reinforced of the 2nd international symposium on ultra high performance concrete; 2008.
concrete beams. Mater Struct 2005;38(277):343–51. p. 231–8.
[19] EI-Niema EI. Fiber reinforced concrete beams under torsion. ACI Struct J [31] Yang IH, Joh C, Kim BS. Flexural response predictions for ultra high
1993;90(5):489–95. performance fiber-reinforced concrete beams. Mag Concr Res
[20] Gunneswara TD, Seshu RR. Torsion of steel fiber reinforced concrete members. 2012;64(2):113–27.
Cem Concr Res 2003;33(11):1783–8. [32] Yang IH, Joh C, Kim BS. Flexural strength of large scale ultra high performance
[21] Narayanan R, Kareem-Palanjian AS. Torsion in beams reinforced with bars and concrete prestressed T-beams. Can J Civil Eng 2011;38(11):1185–95.
fibers. J Struct Eng 1985;112(1):53–66. [33] AFGC (Association Française du Génil Civil). Bétons Fibrés à Ultra-hautes
[22] Mansur MA. Bending–torsion interaction for concrete beams reinforced with Performances. AFGC-SETRA; 2002.
steel fibres. Mag Concr Res 1982;34(121):182–90. [34] ACI Committee 318-11. Building code requirements for structural concrete
[23] Chalioris CE, Karayannis CG. Effectiveness of the use of steel fibres on the and commentary. American Concrete Institute; 2011.
torsional behaviour of flanged concrete beams. Cem Concr Compos
2009;31(1):331–41.