Journal of Advanced Concrete Technology Vol. 4, No.

1, 123-132, February 2006 / Copyright © 2006 Japan Concrete Institute 123

Scientific paper

Shear Strength of Fiber Reinforced Reactive Powder Concrete

Prestressed Girders without Stirrups
Yen Lei Voo1, Stephen J. Foster2 and R. Ian Gilbert3

Received 11 August 2005, accepted 25 December 2005

Abstract
Experimental results from tests on seven 650 mm deep large-scale reactive powder concrete (RPC) I-section girders
failing in shear are reported herein. The girders were cast using 150-170 MPa steel fiber RPC and were designed to as-
sess the capacity to carry shear stresses in thin webbed prestressed beams without shear reinforcement. The tests
showed that the quantity and types of fibers in the concrete mix did not significantly affect the initial shear cracking
load but increasing the volume of fibers increased the failure load. A design model is developed to calculate the strength
of the RPC beams tested in this study. The model is based on crack sliding and uses plasticity theory combined with
observations from the variable engagement model for mode I failure of fiber reinforced concrete. The results of the
model are compared with test data and show a good correlation.

1. Introduction has higher durability and; it has higher fatigue, impact

Since the development of reactive powder concrete
(RPC) in the mid 1990’s by the Bouygues' laboratory in
France (Richard and Cheyrezy, 1994, 1995), extensive
research has been undertaken and is on-going into the
structural/construction use of the material, especially in
the fields of precast products and bridge engineering.
Reactive powder concrete has been used worldwide for a
number of structural applications including the single
span 60-meters Sherbrooke pedestrian bridge crossing
the river of Magog in Canada; the Seonyu Bridge in
South Korea with a single arch spanning 120-meters
supporting a 30 mm thick RPC deck; and the world’s
first RPC highway traffic bridge constructed by VSL at
Shepherd’s Gully Creek in NSW, Australia (Cavill and
Chirgwin, 2003).
In a fast moving century, the use of RPC as a con-
struction material comes to the fore when considering
seemingly conflicting goals such as providing construc-
tion and management efficiencies together with the pro-
duction of low maintenance, durable, structures. In this
regard, steel fiber reinforced RPC (SFR-RPC) provides
the bridge industry with a high tech alternative to con-
ventional technology in that SFR-RPC is a more ductile
material than normal/high strength concrete; it possesses
at least two order of magnitude higher fracture energy; it
has much higher compressive and tensile strengths; it
1 structures and structural components constructed with
Director, DURA Technology Sdn Bhd, Ipoh, Malaysia.
2 this new material is still in its infancy. This study makes
Associate Professor, School of Civil and
Environmental Engineering, The University of New
South Wales, Sydney, Australia.
E-mail: S.Foster@unsw.edu.au
3 high strength concrete.
Professor, School of Civil and Environmental
Engineering, The University of New South Wales,
Sydney, Australia.
and abrasion resistances. With these unique properties,
there is a potential for substantial reductions in cross-
section dimensions reducing the weight of members
which directly impact on substructure and erection costs
(Gilbert et al. 2000). Further, for SFR-RPC beams with-
out conventional shear reinforcement lower manpower
and supervision costs are also a consideration when con-
sidering total, long-term, costs. In terms of structural
performance, the inclusion of discrete fibers into the
concrete matrix can arrest cracks and thereby control
crack propagation.
This paper reports the experimental results of seven
SFR-RPC prestressed I-section girders (SB1 to SB7)
failing in diagonal tension (shear). The selection of vari-
ables, specimen fabrication procedures, test configura-
tion, instrumentation and gauging, experimental results
and comments on test results are presented. Also pre-
sented is a model for the design of SFR-RPC beams,
without web reinforcement, failing in shear. The results
of the model are compared with the experimental data.
2. Research significance
Concrete is the most widely used construction material
with the world cement production of about 1.6 billion
tonnes in 1998, compared to 10 million tonnes in 1900
(Aïtchin, 2000). Recently a new, high-tech, reactive
powder concrete has made its way into the marketplace.
Compared with “traditional” concrete, research into
a significant contribution to the understanding of the
strength behavior and failure modes in shear of
prestressed beams fabricated with fiber reinforced, ultra
124 Y. L. Voo, S. J. Foster and R. I. Gilbert / Journal of Advanced Concrete Technology Vol. 4, No. 1, 123-132, 2006

3. Experimental program Loading 100

Support
3.1 Variables and specimen dimensions 2250
The main variables studied in the experimental program
were type and quantity of fibers and level of prestress. 650
The girders were 4.5 meters long, with a simply-
supported span of 4.0 meters and a total depth of 650
Reaction
mm. The webs of the girders were designed as a 50 mm Support 2000
thick membrane. The top flanges were 400 mm wide and CL
200 Elevation View
contained six 15.2 mm diameter high strength steel
prestressing wires. The bottom flanges were 250 mm
400
wide and contained twelve 15.2 mm diameter strands. 6-15.2 mm
50 300 dia. strands
The girder dimensions and details are shown in Fig. 1.
Each strand had a guaranteed ultimate tensile strength 125 100
(GUTS) of 1750 MPa and a nominal cross-sectional area
of 143 mm2. The specimens were designed to fail in di-
agonal shear in the web region.
50 400 650
3.2 Material, mix design and fabrication
Details of the RPC mixes for specimens SB1 to SB7 are
given in Table 1. The SFR-RPC used in this project is
produced using a high energy concrete mixer provided 125 100
by VSL Prestressing (Aust.). The cement used in this
project was Kandos Type 1 General Portland cement 70 30 200 12-15.2 mm
manufactured to AS3972 (1997); the undensified silica 250
dia. strands
fume used was produced in Western Australia and Syd-
ney sand was used with particle size range between 150 Cross-Section View
µm and 400 µm. The superplasticizer used in the mix
was Glenium 51, which is a polycarboxylic ether based Fig. 1 Fiber reinforced RPC prestressed girders.
superplasticizer. Type I fibers were straight 13 mm long
by 0.2 mm diameter and are fabricated from very high
strength steel with minimum tensile strength of 1800 Table 1 Fiber reinforced RPC mix designs (material
MPa. Type II fibers were end hooked fibers 30 mm long 3
quantities in kg per m of RPC).
by 0.5 mm diameter and are made from high strength
bright steel with a tensile strength of σfu = 1000 MPa. Specimen
Two fiber volume ratios were used 1.25 and 2.5 percent. Component
All the dry components (i.e. cement, silica fume and SB1 SB2 SB3 SB4 SB5 SB6 SB7
sand) were pre-batched into 0.5 tonne bags. The dry
components were then transported to the high energy GP Cement 928 920 911 939 937 911 920
mixer and mixed for about 10 minutes. Water and super-
plasticizer were added gradually until the materials were Sydney sand 928 920 911 939 937 911 920
uniformly mixed. The fibers were introduced last, dis-
persed uniformly and mixing continued for a further 10 Silica fume 223 221 219 225 225 219 221
minutes. Flow table tests as per ASTM C230-03 (2003)
were undertaken before casting of the specimen. All Superplasticizer 39 39 38 39 39 38 39
specimens were cast vertically in steel forms as shown in
Fig. 2. The forms were cleaned and greased to allow Steel fibers
176 175 173 89 107 0 131
smooth stripping. The fresh SFR-RPC was compacted (Type I)
using external vibrators which were attached to the steel
forms. Within one hour of casting, the specimens and Steel fibers
0 0 0 0 71 173 43
test control samples were covered under wet Hessian and (Type II)
plastic sheeting until the day of demoulding. After strip-
ping at age 2 days (5 days for specimen 2), the speci- Total water 186 193 200 188 178 200 193
mens were cured for 7 days at 80°C in a hot water bath.
At age 9 days (12 days for specimen SB2) the specimens W/B 0.16 0.17 0.18 0.16 0.15 0.18 0.17
were removed from the hot water bath and air cured until
the day of testing. Total Fiber
2.5 2.5 2.5 1.25 2.5 2.5 2.5
Ratio (%)
 Y. L. Voo, S. J. Foster and R. I. Gilbert / Journal of Advanced Concrete Technology Vol. 4, No. 1, 123-132, 2006 125

Table 2 Prestressing and fiber Details.

Prestressing
Beam Force (kN) σtop σbot σave
Fiber
No. MPa MPa MPa Volume (%)
Top Bottom

Flange Flange

SB1 0 0 0 0 0 Type I- 2.5

SB2 450 900 -4.72 -27.2 -14.3 Type I- 2.5

SB3 225 450 -2.36 -13.6 -7.15 Type I- 2.5

SB4 225 450 -2.36 -13.6 -7.15 Type I- 1.25

Type I- 1.5
SB5 225 450 -2.36 -13.6 -7.15
Type II- 1.0
Fig.2 Pouring of RPC in the steel forms.
SB6 225 450 -2.36 -13.6 -7.15 Type II- 2.5
3.3 Test setup and instrumentation
Details of the specimens, experimental variables, includ- Type I- 1.88
SB7 225 450 -2.36 -13.6 -7.15
ing levels of prestress, and the test setup and instrumen- Type II- 0.62
tation are given in Table 2 and Fig. 3, respectively.
Notes: σtop and σbot are the extreme fiber stresses of
Specimen SB3 was set as the reference specimen with
concrete at transfer and σave is the average prestress on
2.5 percent of Type I fibers and with 15 percent prestress
the section.
(i.e. the strands were stressed to a load corresponding to
15 percent of the guaranteed ultimate tensile strength of
the strand, GUTS). Specimens SB3 to SB7 were
prestressed to 15 percent of the GUTS of the strands to ensure a shear failure under load. While the prestress
giving an average prestress in the section of 7.15 MPa. per strand is relatively low, the total prestress is high
Each strand in specimens SB1 and SB2 was prestressed compared to that used in common practice for conven-
to 0 and 30 percent of the GUTS, respectively. Specimen tional prestressed concrete girders.
SB4 was similar to specimen SB3 except it had half the All the specimens had a similar experimental setup
amount of Type I fiber in its mix. Specimen SB5 and and instrumental gauging (see Fig. 3). Each specimen
SB7 contained varying quantities of Type I and II fibers was loaded by a single concentrated load applied at the
and beam SB6 contained only 2.5 percent of Type II mid-span of the specimens. All specimens were tested in
fibers. The relatively low prestress adopted, per strand, a 5000 kN capacity stiff testing frame and tested under
was so the beams maintained sufficient tensile capacity ram displacement control.

hydraulic jack
load cell
spreader beam
spherical seat
East 200 West
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32
200
Demec gauge no. pin and roller
(gauge length = 250 mm) support
LVDT
pin support

2000 2000

Fig. 3 Experimental setup.

126 Y. L. Voo, S. J. Foster and R. I. Gilbert / Journal of Advanced Concrete Technology Vol. 4, No. 1, 123-132, 2006

One end of each specimen was a pinned support and Table 3 Mechanical properties.
the other end had a pin and roller support. The pins and
rollers were greased to minimize friction and to give free Specimen SB1 SB2 SB3 SB4 SB5 SB6 SB7
rotation and horizontal translation, as required.
Eo, GPa 44 45 43 43 49 40 46
The instrumentation used for each specimen is shown
in Fig. 3 and includes 99 sets of Demec gauges and a
ν 0.15 0.15 0.14 0.14 0.14 0.13 0.13
linear variable displacement transducer (LVDT). Demec
gauges 1 to 16 were located at the top (compressive) fcm (MPa) 161 160 149 164 171 157 169
flange of the specimen with Demec gauges 17 to 32 lo-
cated at the bottom (tensile) flange. Demec gauges 33 to fcu (MPa) 176 178 166 180 187 168 185
99 (not shown) were located in the web regions of the
specimens to measure longitudinal and shear strains in fsp (MPa) 19.2 20.9 21.9 18.0 22.4 18.3 23.5
the web (refer Voo et al., 2003).
The LVDT was used to measure the midspan dis- fdp (MPa) 11.9 11.2 10.6 10.3 13.6 10.2 11.1
placement of the beam. For each specimen, the test was
paused at preset load points (for the increasing load part fcf (MPa) 29.8 26.4 23.2 14.8 26.3 25.2 23.8
of the test) and the displacement held constant while
Demec gauge data were recorded and the crack pattern Gf (N/mm) 27.7 24.7 21.0 14.3 15.5 12.4 18.6
traced.
Flow (mm) 170 180 210 170 150 210 180

3.4 Mechanical properties

The results of the material control tests are summarized
in Table 3 and full details of the strength tests can be the specimens and were loaded at 1 MPa per minute. The
obtained from Voo et al. (2003) and Voo (2004). The equation as proposed by Chen and Yuan (1980) was used
mean compressive strength (fcm) was determined from to evaluate the double punch tensile strength.
six 200 mm high by 100 mm diameter cylinders stressed
under load control at a rate of 20 MPa per minute. The 4. Test results and observations
ends of the cylinders were ground flat. The cube com-
pressive strength (fcu) was determined from 70 mm cubes The experimental results of the shear tests are summa-
stressed under load control at a rate of 20 MPa per min- rized in Table 4 where Pcr is the load at which the first
ute. shear cracking was detected visually on the specimen
The modulus of elasticity (Eo) and the Poisson’s ratio and Pu is the maximum load recorded during the ex-
(ν ) were obtained from stress-strain tests on 200 mm periments. Specimen SB3 is the reference specimen and
high by 100 mm diameter cylinders tested under circum- it is used to compare the ultimate shear strength of the
ferential displacement control at a rate of between 25 µε other six specimens.
per minute and 150 µε per minute over a period of ap- Comparing the mean cylinder compression strengths
proximately two hours. with the double punch tensile strength (refer Table 3) for
The tensile strength of the material was evaluated us-
ing split cylinder (Brazil) tests, notched beam flexural
tests and double punch tensile tests. The split cylinder Table 4 Experimental results.
tensile strength (fsp) was obtained from tests on six 200
mm high by 100 mm diameter cylinders loaded at 1.0 Pcr Pu Pu Pu fcm,SB3
MPa per minute via a 10 mm wide loading strip.
Specimen ⋅
(kN) (kN) Pcr Pu ,SB3 f cm
The notched three point flexural tension strength (fcf)
was obtained from 100 mm square prisms spanning 400 SB1 300 860 2.9 0.96
mm with a notch depth of 25 mm. The specimens were
counter balanced to eliminate the effect of the self- SB2 400 994 2.5 1.13
weight on the fracture measurement and the specimens
were controlled under crack mouth opening displace- SB3 300 856 2.9 1.00
ment (CMOD) at a rate of 500 µε per minute. The frac-
ture energy (Gf) was obtained from notched three point SB4 300 673 2.2 0.75
bending tests with either the total area under the load
versus displacement or the load versus CMOD curves SB5 400 880 2.2 0.87
divided by the net cross-sectional area of the specimens.
The double punch tensile strength (fdp) tests were un- SB6 250 660 2.6 0.75
dertaken on 200 mm high by 100 mm diameter cylinders
SB7 350 800 2.3 0.88
using a pair of 25 mm high by 25 mm diameter rigid
steel circular punches on the top and bottom surface of
 Y. L. Voo, S. J. Foster and R. I. Gilbert / Journal of Advanced Concrete Technology Vol. 4, No. 1, 123-132, 2006 127

specimens SB1, SB2 and SB3 (the specimens with 2.5
percent of 13 mm, straight fiber), the ratios of tensile
strength to compressive strength are similar. A similar
variation is also seen in the flexural tension strengths
and fracture energies but, curiously, not in the split cyl-
inder tension results. It is thought that the split cylinder
test does not give a reliable measure of the tensile
strength due to the effect of the fibres providing a resid-
ual tensile capacity across the crack sufficient to control
the splitting and induce a bearing failure beneath the
loading strips (Voo, 2004). In Table 4 the experimental
results are compared to that for specimens SB3 corrected
for the variation in compressive strength.
In terms of the prestressing levels, the comparison
presented in Table 4 indicates a 15 percent variation in
strength due to the effect of prestressing. In terms of the

WEST EAST
700 700
500 400 500 700 600 600 600 600 600 500 700 600 600 650
300 500 400 500 500
400 400 500 600
600 400
500 400
300 400 600 600 600
600
400 500 500 300 400 400
300 500 300

SB1
WEST EAST
800 700 900 800 800
900 700 900 900800 750 900 900
700 600 600 750 800 900
800 750
800 700 900 600 900
700 600 600
600 700 500 600 500 750
750 700 700
900 600 700 900 600

SB2
WEST EAST
600 700 760 600 600 500 500
700 500 600
600 600 600 500 600 500 600 700 600 500 400 400 600
500
700 400 300 300 500 600 500 500
600 400 400 400
500 300 300 300 400 500 600
300
400 400 400

SB3
WEST EAST
600 500 500 500
600 600 600 400500 500 500 400 400400
400 400400 300 400
450 400 400
400 500 500
500 400 300 500
600 300
500 500500 400
400
500 400400 400
400
600 600 300
400 300 300

SB4
WEST EAST
700
800 800 800 700 700 700 500 700 700 700 700 700 700 800 800 700
700 700 700 600
600 700
500
600 500 400
600 600 450
600 600 500

SB5
WEST EAST
600 450 450500500 400 400 500 500 500 500
500 400 400 450 600 400 400 400 400 400 400
400 400 450400 200 300 300 400 450 500
400 500 450 450
450
200 450 400
600 500 600
400 300 300 600 450 450400
300
450 400 300 300

SB6
WEST EAST
790 790 700 700 600 700700 700 700 700 700 700 700 700 700 700 700
700
790
600 600 600600600 600 600 600 600 600 700 600 600 600 600
600 600 600 500
500 600
500 500 500 500 400
500 400 500 500 500
500 400 400 400 500
400 400
400 400 500 500

SB7

Fig. 4 Crack patterns for specimens SB1 to SB7.

128 Y. L. Voo, S. J. Foster and R. I. Gilbert / Journal of Advanced Concrete Technology Vol. 4, No. 1, 123-132, 2006

quantity of the same type of fiber, Table 4 shows the

comparative strength of specimen SB4 (with 1.25 per-
cent fibers) was 30 percent lower that specimen SB3
(with 2.5 percent fibers).
For the test using different fiber types, Table 4 shows
that the failure loads were lower with increasing quanti-
ties of the 30 mm end hooked fibers and reducing quan-
tities of 13 mm straight fibers. This indicates that fiber
fracture may have had a significantly greater influence
for the longer end hooked fiber than for the shorter
straight fiber. This is further evidenced by the lower
fracture energies measured in the control specimens for
increasing end hooked fiber to straight fiber ratios (refer
Table 3).
A comparison of the crack patterns (Fig. 4) show all
the girders behaved in a similar manner with diagonal
Fig. 5 SB3: major diagonal shear crack at P = 850 kN.
shear cracking initiated in the web regions of each of the
shear spans. The diagonal cracks then multiplied and
propagated toward the top flange and smeared across the
spans with increasing load. Finally, failure resulted from
tensile fracture across a single, dominant crack or from a
coalescence of cracks leading to the formation of a
dominant crack.
The response of specimen SB3 is typical of all speci-
mens. In SB3 the first diagonal shear cracking was ob-
served at a load of 300 kN. Fig. 5 shows as the loading
was increased a number of diagonal shear cracks formed
and the cracks were smeared across the span. At 850 kN
a major diagonal crack formed in the east span. The peak
load was 856 kN. After the peak load, the diagonal crack
opened considerably and the load reduced. Fig. 6 was
taken at a post peak load of 700 kN with all fibers hav-
ing been pulled out along the crack plane. Testing of
specimen SB3 was concluded when the post peak load
had reduced to 320 kN.
The raw data for the Demec strain readings for all of Fig. 6 SB3: fiber pullout at the critical diagonal crack at
the seven shear beams tested are given in Voo et al. post peak load of 700 kN.
(2003). The load-deflection curves of specimens SB1 to
SB7 are given in Fig. 7. shear strength increased by approximately 16 percent.
In terms of varying the ratio of fiber quantities in the
5. Analysis of results cocktail mixes (ie. the mixes containing both types of
fibres), Fig. 7b compares the load versus mid-span dis-
Comparison of the crack patterns in the shear beams placement of specimens SB3, SB5, SB6 and SB7. The
show that the quantity of fibers, type of fiber and ultimate shear strength and the stiffness of the specimens
prestressing level only marginally affected the shear decreased as the quantity of the straight steel fibers was
cracking load (Pcr) but had a significant influence on the reduced.
rate of crack growth. After the initial cracking of the In the calculations of the strains in the beams, pre-
specimens, the cracks propagated at a faster rate in sented below, the strains due to the initial prestressing
specimens SB2 and SB4, the specimens with the lowest forces applied to the beams were assessed using linear
prestress and lowest quantity of steel fibers, respectively. elastic finite element modeling (Voo et al., 2003) and the
Figure 7a compares the load versus mid-span deflec- initial prestrains added to the measured data. For speci-
tion of specimens SB1, SB2, SB3 and SB4. The figure mens SB3 to SB7 the prestrain in the top flange was -71
shows that at different levels of prestressing, not only µε and in the bottom flange was -232 µε (measured at
did the ultimate shear strength of the specimens increase the level of the tendons). For specimen SB2 the pre-
with increasing prestress but the prestress also influ- strains were -142 µε and -464 µε in the top and bottom
enced the stiffness of the girders. Comparison of speci- flanges, respectively. At peak loads, the compressive
men SB1 (i.e. with no prestressing) and SB2 (with 14.3 strains were of the order of 600 µε to 800 µε, which are
MPa average prestress in the web) shows the ultimate equivalent to compressive stresses of 27 MPa to 36 MPa.
 Y. L. Voo, S. J. Foster and R. I. Gilbert / Journal of Advanced Concrete Technology Vol. 4, No. 1, 123-132, 2006 129

1200
Distance from Centre Line (mm)
SB 2 -1875 -1125 -375 375 1125 1875
1000 -800
SB 1
SB 3
800 Load(kN)
CL 0
-600
200

Strain (µε)
600 SB 4 400

)
500
-400 600
Load (kN)
L

400 700
800
d (kN)

Strain (
850
200 -200

0 0
0 20 40 60 80 100 120 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Midspan Displacement (mm) Demec Position
(a) (a)
1200 Distance from Centre Line (mm)
-1875 -1125 -375 375 1125 1875
1000 4000
SB 3 SB 5
SB 7 3000 Load (kN)
800 CL
0
Load (kN)

SB 6 200
Strain (µε)

600 2000 400

)

500
600
400 1000 700
800
Strain (

850
200 0

0 -1000
0 20 40 60 80 100 120 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32
Midspan Displacement (mm) Demec Position

(b) (b)
Fig. 8 Strains in beam SB3 (including prestrain): (a) top
Fig. 7 Load versus midspan deflection of (a) SB1 to SB4 flange (Demecs 1 to 16); (b) strain in bottom flange
and (b) SB3, and SB5 to SB7. (Demecs 17 to 32).

This indicates that crushing of the top flanges was not an of the cement-fine aggregate matrix with high tensile
issue. The maximum tensile strains at the bottom flange strengths maintained for large crack openings. Thus,
were 2500 µε to 4500 µε indicating that the strands re- plasticity is an appropriate approach for the design of
mained in the elastic range and yielding of the tension such structural elements.
reinforcement was also not an issue (strand yield strain, In the crack sliding model, concrete is treated as a
εpy = 8970 µε). The strains in the top and bottom flanges Mohr-Coulomb material with a zero tension cut-off. A
of SB3 are given in Fig. 8 and are typical of the data more comprehensive description of the theory can be
obtained. found in Johansen (1958), Sandbye (1965) and Nielsen
(1967). According to the crack sliding model, the crack-
6. Shear strength – plasticity model ing of concrete introduces a potential yield-slip line
which, due to a reduced sliding resistance, forms the
Plasticity approaches came to the forefront of reinforced critical failure mechanism.
concrete design with work published by Nielsen (1963, For both non-prestressed and prestressed simply sup-
1967). Based on the upper bound theory of plasticity, ported rectangular beams with rectangular cross-sections
Zhang (1994) developed the crack sliding model to cal- and loaded with two symmetrically located point loads,
culate the shear strength of flexurally over-reinforced the ultimate shear strength of the section can be deter-
rectangular beams without shear reinforcement. In the mined by
development of the model that follows the flange out-
stands of the I-girders are ignored for the purpose of ⎛ 2 ⎞
1 * ⎜ ⎛x⎞ x⎟
assessing the shear capacity. The crack sliding model is Vu = fc b h ⎜ 1 + ⎜ ⎟ − ⎟ (1)
selected in this study as the use of high quantities of fi- 2 ⎜ ⎝h⎠ h⎟
⎝ ⎠
bers leads to a relatively plastic response after cracking
130 Y. L. Voo, S. J. Foster and R. I. Gilbert / Journal of Advanced Concrete Technology Vol. 4, No. 1, 123-132, 2006

where fc* is the effective concrete strength, b and h are Pu cL

the width and depth of the section, respectively, and x is a b
the horizontal projection of the yield line as shown in
Fig. 9 (Zhang, 1994). dp1
h d P
To determine the starting position of the critical crack, e
the ultimate shear (Vu) is taken as equal to the shear at
diagonal cracking (Vcr). For definition purposes, the di- Vu x
agonal cracking load (Vcr) is defined as the load capacity (a)(a)
of the beam when the diagonal crack is fully developed Pcr cL
a b
with a uniform effective tensile stress bridging the crack. A
For a simply supported beam under point loading, the
dp1
diagonal cracking load is determined by taking the mo- h d P
e
ment about the crack tip (marked A in Fig. 9b) and is f t*
written as
Vcr
x

Vcr =
1 * h +x
ft b +
2 2 ∑ Pe d p
i (2)
(b)(b)
2 a a
Fig. 9 Simply supported beam with critical diagonal
where dpi is the distance of the effective prestressing crack: (a) yield line; (b) cracking load.
force (Pe) at the ith level from the top surface of the
beam, a is the shear span and ft* is the effective tensile
strength of the concrete.
The solution for x is obtained by equating Vu given by V
Eq. 1 to Vcr given in Eq. 2. That is, the solution of x is
Vu (Eq. 1) Prestresssed
the intersection of Eqs. 1 and 2 (Fig. 10) giving Beams

⎛ 2 ⎞
⎜ ⎛ x⎞ x⎟
f c* ⎜ 1 + ⎜ ⎟ − ⎟ =
⎜ ⎝h⎠ h⎟
⎝ ⎠ Non-Prestresssed
Beams

(x2 + h2 ) + 2∑ Pi d pi
Pe d pi / a

ft* (3) Vcr (Eq. 2)

ah abh
x
....... for 0 ≤ x ≤ a
Fig. 10 Relationship between Vu and Vcr with x.
In the analyses that follow, the effective tensile
strength of the fiber reinforced composite is taken as

f t * = ν t f tf (4) 8 σ (MPa)
Range of COD at the
f approximate peak load
where νt is the tensile effectiveness factor and ftf is the tf for FR-RPC girders
6
maximum tensile stress of the fiber component and can in this study
be calculated using a model such as variable engagement
model (VEM) of Voo and Foster (2003, 2004), for ex- 4
ample.
From the experimental data it was observed that at the
peak load the crack widths for the SFR-RPC experimen- 2
tal girders tested in this study were in the range of 0.5 to
1.5 mm. From the VEM the effective tensile strengths w (mm)
0
(ft*) in this range for the 13 mm long by 0.2 mm diame- 0.5 1.5 6.5
ter fibers used were in the range of 70 to 90 percent of
the peak fiber contribution to the tensile strength (Fig. Fig. 11 Range of tensile stress across failure surface for
11). In the analyses that follow, the effective tensile SB1 to SB3.
strength of the fiber reinforced composite is calculated
using Eq. 4 with νt = 0.8.
The effective compressive strength (fc*) takes into ac-
count the cohesion and friction along the crack during f c * = ν c f cm (5)
sliding (refer Zhang, 1994) and is taken as where fcm is the mean cylinder compressive strength and
 Y. L. Voo, S. J. Foster and R. I. Gilbert / Journal of Advanced Concrete Technology Vol. 4, No. 1, 123-132, 2006 131

the compression effectiveness factor is taken as νc = 0.8.
Further justification of this value, however, is needed for
concretes containing relatively high fiber volumes and
this is a matter for further research.
In the calculation of x the limits are 0 ≤ x ≤ a. The so-
lution to Eq. 3 is not readily obtained analytically. In-
stead, a numerical method is used to find the critical
value of x/h. In this way the starting position of the criti-
cal diagonal crack for a given a/h is found and, conse-
quently, the ultimate load and the shear capacity can be
determined from Eq. 1. Finally, the ultimate shear
strength is written as
τ u = Vu (bh )
The shear strengths calculated using the plasticity
model described above for the SFR-RPC shear beams
tested in this study are compared with the experimental
data and the results are summarized in Table 5. The
prestressing force used in the calculations is that given in
Table 2 without losses. Although the level of prestress,
in absolute terms was relatively high, the prestress in
each strand was low and below the threshold for relaxa-
tion of the tendons. Also, while the average compressive
stresses on the section were high relative to current de-
sign practice, they are not high relative to the compres-
sive strength of the RPC. For these reasons the losses are
small and not significant to the strength calculations. The
mean model to experimental shear capacity ratio is 0.89
and the coefficient of variation is 0.095.
While it is shown that the rectangular section model is
capable of calculating the shear strength of the SFR-RPC
prestressed girders tested, the angle of the failure crack
is, generally, flatter than that of the experiments. Voo et
al. (2003) show that a crack sliding model refined to
include the flanges can improve the calculation of the
crack sliding angle. The model is, however, substantially
more complex and gives only small differences in the
calculated shear strengths, as shown in Table 5.
7. Conclusions
(6) Seven reactive powder concrete prestressed girders
without stirrups were tested to study the capacity of fiber
reinforced RPC beams in shear. The test variables were
the quantity and type of fibers and the prestress. The
steel fibers used in the tests consisted of either 13 mm
straight fibers and/or 30 mm end-hooked fibers. All the
tested specimens had the same cross-section and were
subjected to mid-point loading over a shear span of 2
meters. The shear span to effective depth ratio for the
beams was 3.33.
From the experimental study the following conclu-
sions are drawn:
1. The quantity of fibers and type of fibers used in the
concrete mix do not significantly affect the cracking
load but have a significant influence on the rate of

Table 5 Comparison of the model shear strength calculations with the experimental data.

Rectangular Beam I-Beam Model

# Model
Vu,exp θexp ftf ft* fc *

Beam
kN deg. MPa MPa MPa Vu,theo θtheo Model Vu,theo θtheo Model
kN deg. Exp. kN deg. Exp.

SB1 430 37 6.5 5.2 129 332 18 0.77 338 30 0.79

SB2 497 34 6.5 5.2 128 445 24 0.90 430 42 0.86

SB3 428 32 6.5 5.2 119 357 21 0.83 336 37 0.79

SB4 337 26 3.4 2.7 131 338 18 1.00 312 33 0.93

SB5 440 29 5.4 4.3 137 364 19 0.83 340 34 0.77

SB6 330 24 3.9 3.1 126 323 18 0.98 299 33 0.93

SB7 400 21 5.8 4.7 135 370 19 0.92 345 35 0.86

Average 0.89 - - 0.85

COV 0.095 - - 0.079

Notes: # fiber contribution to the peak tensile strength (ftf ) is calculated using the variable engagement model of Voo and
Foster (2003, 2004).
132 Y. L. Voo, S. J. Foster and R. I. Gilbert / Journal of Advanced Concrete Technology Vol. 4, No. 1, 123-132, 2006
crack propagation and on the failure loads.
2. At the peak load, many fine cracks had formed in the
web, with the cracks well distributed through the
shear spans. The failure loads were more than twice
the cracking loads.
A design model is also developed to calculate the
strength of steel fiber reinforced reactive powder con-
crete beams failing in shear. The model is based on the
crack sliding model of Zhang (1994) combined with
observations from the variable engagement model of
Voo and Foster (2004) for mode I fracture of fiber rein-
forced concrete. The results of the model are compared
with the tests reported herein and a good correlation was
observed for the strengths calculated by the design
model when compared to the test data. However, the
model, generally, calculates a flatter crack sliding plane
compared to that observed in the tests. To better deter-
mine the crack sliding angle a more refined model is
required that includes the flange effects.
Acknowledgements
This project was undertaken in the Heavy Structural
Laboratory of University of New South Wales and it was
supported by an Australian Research Council Strategic
Partnerships with Industry Grant 2001-2003 in collabo-
ration with VSL (Aust). This support is acknowledged
with appreciation.
References
AS3972 (1997). Portland and Blended Cements.
Standards Australia.
ASTM C230-03 (2003). “Standard Specification for
Flow Table for use in Tests of Hydraulic Cement.”
Aїtchin, P. C. (2000). “Cement of yesterday and today
concrete of tomorrow.” Cement and Concrete
Research, 30 (9), 1349-1359.
Cavill, B. and Chirgwin, G. (2003). “The worlds first
RPC road bridge at Shepherds Gully Creek, NSW.”
21st Biennial Conference of the Concrete Institute of
Australia (CIA), July, 89-98.
Chen, W. F. and Yuan, R.L. (1980). “Tensile strength of
cconcrete: Double-punch test.” Journal of the
Structural Division, ASCE, 106, 1673-1693.
Gilbert, R. I., Gowripalan, N. and Cavill, B. (2000). “On
the design of precast, prestressed reactive powder
concrete (Ductal) girders.” Proceedings of 4th
Austroads Bridge Engineering Conference, Adelaide,
Australia, Nov 30 - Dec 1, 3, 313-324.
Johansen, K.W. (1958). “Brudbetingelse for sten og
beton (Failure criteria for rock and concrete).”
Bygningsstatiske Meddelelser, 29 (2), 25-44.
Nielsen, M. P. (1963). “Yield condition for reinfroced
concrete shells in the membrane state.” Non-classical
Shell Problems (ed. Olszak and Sawczuk), Proc. IASS
Symposium, Warsaw, Amsterdam, 1030-1040.
Nielsen, M. P. (1967). “Om forskydningsarmering i
jernbetonbjælker- On shear reinforcement in
reinforced concrete beams.” Bygninsstatiske
Meddelelser, 38 (2), November, 33-58.
Richard, P. and Cheyrezy, M. (1994). “Reactive powder
concretes with high ductility and 200-800 MPa
compressive strength.” ACI, SP-144(24), San
Francisco, USA, 507-518.
Richard, P. and Cheyrezy, M. (1995). “Composition of
reactive powder concretes.” Cement and Concrete
Research, 25 (7), 1501-1511.
Sandbye, P. (1965). “A plastic theory for plain concrete.”
Bygningsstatiske Meddelelser, 36 (2), 41-62.
Voo, J. Y. L. and Foster, S. J. (2003). “Variable
engagement model for fibre reinforced concrete in
tension.” UNICIV Report R-420, School of Civil and
Environmental Engineering, The University of New
South Wales, June, 86 pp.
Voo, J. Y. L. and Foster, S. J. (2004). “Tensile fracture of
fibre-reinforced concrete: Variable engagement
model.” 6th RILEM Symposium on Fibre-Reinforced
Concretes (FRC) – BEFIB, 20-22 September, Varenna,
Italy, 875-884.
Voo, J. Y. L, Foster S. J. and Gilbert R. I. (2003). “Shear
Strength of Fibre Reinforced Reactive Powder
Concrete Girders without Stirrups.” UNICIV Report
R-421, School of Civil and Environmental
Engineering, The University of New South Wales,
November, 131 pp.
Voo, J. Y L. (2004). “An investigation into the behaviour
of prestressed reactive powder concrete girders
subjects to non-flexural actions.” PhD Thesis, School
of Civil and Environmental Engineering, The
University of New South Wales, Sydney, Australia.
Zhang, J. P. (1994). “Strength of cracked concrete: Part 1
– Shear strength of conventional reinforced concrete
beams, deep beams, corbels, and prestressed
reinforced concrete beams without shear
reinforcement.” Report No. 311, Technical University
of Denmark, Department of Structural Engineering,
Lyngby, 106 pp