Composites Part B 90 (2016) 1e13

Contents lists available at ScienceDirect

Composites Part B
journal homepage: www.elsevier.com/locate/compositesb

Shear behaviour of new beams made of UHPC concrete and FRP rebar
E. Ferrier a, *, A. Confrere b, L. Michel a, G. Chanvillard b, S. Bernardi c
a
LGCIE Site Bohr, Universite Claude Bernard Lyon I, Domaine Scientifique de la DOUA, 82 Boulevard Niels Bohr, 69622 Villeurbanne Cedex, France
b
Lafarge, LCR, Saint Quentin Fallavier, France
c
Lafarge, Corporate, Paris, France

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

Article history: The primary objective of this study was to develop a new type of high-performance lightweight beam
Received 10 April 2015 with improved performance over conventional reinforced concrete (RC) by adding fibre-reinforced
Received in revised form polymer (FRP) reinforcing rebar to ultra-high-performance concrete with microfibre reinforcement
18 September 2015
(UHPC-SFR). This new type of beam was designed to be lightweight, to have high compressive and tensile
Accepted 4 November 2015
Available online 17 December 2015
strength, to be able to sustain large bending moments, and to be resistant to shear. The main objective
was to verify the mechanical behaviour of the beams and to compare it with the behaviour of typical RC
beams. An experimental program was designed to identify its failure modes and bending behaviour. The
Keywords:
A. Carbon fibre
results indicate that the behaviour of such RC beams is comparable to typical RC beam behaviour to a
B. Mechanical properties certain extent. An analytical model for validating this concept is presented here, which is based on the
B. Stress transfer typical material behaviour law hypotheses of nonlinear mechanical beam behaviour. The load
C. Analytical modelling edisplacement and momentecurvature relationships predicted using this model were compared to the
D. Mechanical testing experimental results obtained for several large-scale specimens. The comparisons revealed a good cor-
relation between the analytical and experimental results and illustrate the potential of these composite
beams in civil engineering structures.
© 2015 Elsevier Ltd. All rights reserved.

1. Introduction
Research and development on the application of high-
performance materials in the field of civil engineering has pro-
gressed significantly throughout Europe and the world in recent
years, which has included the study of advanced materials such as
ultra-high-performance micro-fibre-reinforced concrete (UHPC-
SFR) and fibre-reinforced polymer (FRP). UHPC-SFR is strong under
both compression and tension and can contribute to the mitigation
of the effects of environmental exposure due to its low perme-
ability. Thus, the use of this material is expected to increase,
especially when sustainable development principles are considered
[1]. In addition, the use of UHPC enables designers to create thinner
sections and longer spans that are lightweight, graceful and inno-
vative in both geometry and form, with low permeability and good
durability with respect to corrosion, abrasion and impact [2]. Shear
stirrups can be eliminated, which enhances safety, reduces weight
and allows for faster construction. Furthermore, its durability re-
duces the maintenance requirements and extends the service life.
* Corresponding author. Tel.: þ33 04 72 69 21 21; fax: þ33 04 78 94 69 06.
E-mail address: emmanuel.ferrier@univ-lyon1.fr (E. Ferrier).
All of these factors contribute to reductions in cost [3]. Finally, use
of UHPC in construction has increased significantly throughout the
world to such a degree that new ways to optimize its use are now
necessary [4,5].
Recently, the use of fibre-reinforced polymer (FRP) rebar to
replace steel rebar has emerged as one of the many techniques
proposed to enhance the corrosion resistance of reinforced con-
crete structures. It has already been used for strengthening, reha-
bilitation and new construction. In particular, FRP rebar offers great
potential for use in reinforced concrete construction under condi-
tions in which conventional steel-reinforced concrete has exhibited
unacceptable performance [6,7].
In emerging construction fields, research on hybrid structural
members that use FRP materials combined with other high per-
formance materials to yield new structural advantages has mark-
edly increased [7,8]. The objectives of these new designed hybrid
members are to enhance global structural performance and
corrosion resistance but also to save costs compared to conven-
tional steel construction. A study on the construction of an FRP-
reinforced concrete bridge deck in Wisconsin has revealed that
conventional construction technology and normal strength con-
crete can lead to a 57% savings in construction labour costs

http://dx.doi.org/10.1016/j.compositesb.2015.11.022
1359-8368/© 2015 Elsevier Ltd. All rights reserved.
2 E. Ferrier et al. / Composites Part B 90 (2016) 1e13

compared to the construction of a nominally identical steel rebar fibres. An elastic, multi-linear stressestrain relationship for UHPC-
reinforced deck [9]. SFR proposed by Habel et al. [19] and revised in AFGC 2013 rec-
Moreover, many studies have focused on the bond-slip behav- ommendations [2] was used to model the mechanical behaviour of
iour of FRP rebar embedded in concrete to prevent unexpected the UHPC-SFR concrete (Fig. 1). For concrete, the strain values
debonding [10e15]. It has been shown the increased concrete corresponding to critical levels on the stressestrain curve are
strength corresponds to the increased FRP rebar strength [16]. The shown in Table 1 and were determined by the equations below,
use of UHPC could allow the full composite action of the FRP rebar using the low strain hardening law. The tensile stress correspond-
to be developed. The bond behaviour between UHPC-SFR and FRP ing to the initial cracking is fctk,el/gcf, with a strain value given by
rebar is not developed in the present study, although this property 3 u,el ¼ 0.02%. The maximum point of the tensile stressestrain
represents another benefit of combining these two high- relationship is defined by the stress corresponding to fctk/(gcf K)
performance materials. with a strain value corresponding to 3 u,peak.
Here, we present the experimental results of an investigation of
a new type of RC beam composed of UHPC-SFR and reinforced with
w0:3 fctk;el
carbon fibre-reinforced polymer (CFRP). 3 u;peak ¼ þ (1)
To minimize the amount of UHPC and to improve the full use of lc gcf Ec;eff
the compressive strength, a specific cross section of the beam is
proposed. When the UHPC is associated with FRP rebars, it may
develop a higher tensile stress at failure. The level of material stress 0:01h fctk;el
3 1% ¼ 2h
þ (2)
at failure is improved by the combination of two high-performance 3
g cf Ec;eff
materials. The first cross section is a UHPC hollow section designed
to avoid shear failure. The second configuration is designed to only
include UHPC in areas where strength is essential, such as at the top
and bottom for compression and tension resistance. A layer of
UHPC is cast in the lower and upper beam parts on a few centi- σ
metres. In between, standard C25 concrete is used.
The experimental testing was conducted on beams with 2- and
4-m spans for which the high performance of this innovative RC fctk,el/γcf
structural configuration was demonstrated. Three types of sections
were considered. The objective was to develop different failure fctfk/(γcf.K)
modes. To accomplish this objective, the geometry of the section fctf,1%,k/(γcf.K)
was adapted for each case study. We sought to avoid was tensile
failure of the FRP reinforcements. εcud εc0d
ε
εu,el εu,peak εu,1% εu,lim
2. Design of FRP-UHPC SFR beams
Ecm
2.1. Material

2.1.1. UHPC-SFR
The ultra-high-performance concrete used was a Ductal® G2
premix. To evaluate its mechanical properties, mechanical tests
were performed on each concrete batch. The beams were cast in fcd
four different batches. Six UHPC prisms are subjected to four-point
bending tests to measure their modulus of elasticity Ec and flexural
(a) UHPC behaviour law in both traction and compression at ULS
strength (Table 1). Six cylindrical concrete specimens were
compression tested 28 days after casting, in accordance with the
specifications of the UHPC standard from JSCE [17]. A mean
compressive strength fc of 171 ± 1.8 MPa was obtained. fFRP,r
UHPC-SFR has been thoroughly studied regarding tension, and
several mechanical laws are available. Won et al. [18] confirmed the
properties of UHPC-SFR internally reinforced with micrometallic

Table 1
Parameters of the mechanical behaviour laws (mean values).

Material Property Parameter Value

UHPC Compression fcd (MPa) 171

Efrp
a
3 cud (%) 0.27
Tension fctk,el (MPa) 13.4
a
3 u,el (%) 0.02
fct,fka (MPa) 25.9
Young's modulus Ec (GPa) 53.9 εre
CFRP rebar Tension ffrpa (MPa) 1890
a
3 rc (%) 1.35
(b) FRP behaviour law in traction
Young's modulus Efrp (GPa) 130
a
Mean values given by manufacturers. Fig. 1. Stressestrain relationship of the materials.
 E. Ferrier et al. / Composites Part B 90 (2016) 1e13
l Geometry and details of the reinforcements.
3 u;lim ¼ f
4lc
In these equations, h is the depth of the section, fctk,el is the char-
acteristic elastic tensile strength, fctk is the characteristic maximal
post cracking stress, and Ec,eff is the effective elastic modulus of
UHPC. lc is a characteristic length equal to 2h/3, and lf is the length
of a fibre. gcf is a partial safety factor of the fibres. Failure is assumed
to occur in tension in the UHPC when the higher section strains in
the tensile region reach the value of 3 u,lim. When the concrete strain
is greater than 3 u,lim, the tensile stress is equal to zero.
This assumption can be made for the cracking behaviour of
concrete internally reinforced by micrometallic fibres [2].
Regarding concrete under compression, the maximum compressive
strain at ULS, 3 cud, can be taken equal to 2.7  103, a value rec-
ommended for preliminary designs by AFGC [2].
2.1.2. Usual concrete
In the case of beams made of three layers of concrete, a standard
concrete compressive strength of 25 MPa is used in the least
stressed area, which is the central part of the section between the
two layers of UHPC. This concrete was made according to the
method of Dreux-Gorisse, and the ratio W/C of 0.6 is determined by
the formula of Bolomey. A standard Portland cement is used, and
the sand and gravel granular materials are rolled types. The particle
size analysis shows a granular type class 0/4 for the sand and 5/10
for the gravel. The concrete was produced in the laboratory of the
LYON 1 university. The composition of the concrete with a target
value of 25 MPa is given in Table 2.
For each batch, three control samples (cylinder 160  320 mm)
were cast according to the NF EN 12390-1 standard. The control
samples were then compression tested according to NF EN 12390-4
to determine the compressive strength of each concrete. The
average compressive strength obtained was 25.2 MPa. Tensile tests
and measurements for determining the Young's modulus were not
carried out. The values of EC2 [20] are considered for the
calculations.
2.1.3. FRP rebar
The material properties of the FRP rebar were not tested in the
laboratory. Instead, the property values provided by the suppliers
were used in the calculations. The CFRP rebar used was made with
continuous carbon fibres impregnated in a vinylester resin. The
surface is helically wrapped and sand coated. Additional mechan-
ical properties of the rebar (tensile strength ffu and Ef the Young's
modulus) are given in Table 1, and the geometries of the beams and
ratios of reinforcement are summarized in Table 3.
2.2. Section description
The first section is a typical RC beam with longitudinal and
transverse steel rebar.
The objective of the configuration of the second section was to
reduce the weight of the beam. Two plastic hollow tubes were
placed in the sections such that the minimum vertical layer of the
Table 2
Composition of the designed 25 MPa concrete (kg/m3).
Material Quantity (kg/m3)
Cement CEM I 32.5 N 370
Sand 0/4 704
Aggregates 5/10 992.5
Water 213.3
3
Table 3
(3)
Beam designation L (mm) h/b Rebar section (mm2) R2 (%)
S1-ST-2d14 2000 1.7 310 0.83
S1-L-2d15-hfl30-hf260 2000 1.7 350 0.93
S1-La-2d15-hfl30-hf260 2000 1.7 350 0.93
S1-H-2d15-Rl75-R2100 2000 1.7 350 0.93
S1-H-2d15-Rl100-R2100 2000 1.7 350 0.93
S1-H-4d15-Rl100-R2100 4000 1.7 710 1.9
a
UHPC 1% volume fibres.
concrete did not match the maximum shear stress along the height
of the section.
Because the tensile stresses in the concrete will be high, UHPC-
SFR is poured into all of the sections around the hollow tubes
placed in a plywood mould. Three FRPeconcrete composites beams
were cast using different sizes of hollow plastic tubes, and the
height and width of the section are given in Table 3. The detailed
material and geometrical properties of these three beams are given
in Table 4.
The objective of the third configuration section was to reduce
the cost of the hybrid beam composed using UHPC-SFR and FRP
rebar. The UHPC-SFR was poured only at the bottom and top por-
tions of the beams where the tensile and compressive stresses are
the highest. Between these two layers, a normal strength concrete
is poured. In the case of type 2 (Fig. 2), the thickness of the bottom
layer of UHPC in the beam that is reinforced with FRP is denoted hf2.
The thickness of the top layer of UHPC in the beam is denoted hf1.
Experimentally, the UHPC-SFR layers are poured on the same day as
the normal strength concrete, and thus the interface between the
two concretes results from a fresh-to-fresh pouring. Two beams of
type 2 were designed, and the characteristics are also reported in
Table 4.
Several parameters were selected for evaluating the efficiency of
the beams. Geometric and material parameters, such as the beam
depth and span, the depth-to-width ratio, the volume percentage of
the concrete versus micro steel fibres, the volume ratio of tensile
rebar and rebar mechanical properties, including axial stiffness,
and the mechanical properties of the concrete and the FRP, are all
significant. In this study, the parameters that were investigated are
beam span, depth-to-width ratio, and tensile strength of the rebar
in the lower part (Fig. 2). The two beams are identical except that
one beam has a 2% volume percentage of micrometallic fibres,
whereas the second beam has a 1% volume percentage.
To develop the highest possible strength in the CFRP rebar,
UHPC-SFR was poured in the bottom part of the beams as the
surrounding concrete along the rebar, as shown in Fig. 2. This
combination allows the global tensile strength of the bottom
portion of the RC beam to increase. A reference beam was first
designed to compare the performances of the new hybrid beams
studied here. Then, two section configurations were chosen.
The designed beams are 150 mm wide by 250 mm high. Two
rebars of 15-mm-diameter CFRP always reinforce the section for
comparing the beams. The beam is designed to incorporate a top 3-
cm layer of UHPC and a lower 6-cm layer of UHPC that encases the
FRP reinforcements.
To compare the standard solution with the sandwich beam, the
density of each material is summarized in Table 5. The weight of
each component is then calculated, and the results are shown in
Table 6.
The beam weight of these two solutions is equivalent, so there is
no weight gain for the configuration of the sandwich beam. How-
ever, there is a UHPC economy relative to the full UHPC beams or to
the hollow beams.
4 E. Ferrier et al. / Composites Part B 90 (2016) 1e13

Table 4
Detailed geometrical properties for the designed beams.

Beam designation hf1 (mm) hf2 (mm) R1 (mm) R2 (mm) Weight (kg/lm)

S1-ST-2d14 e e e e 102
S1-L-2d15-hfl30-hf260 30 60 e e 103
S1-La-2d15-hfl30-hf260 30 60 e e 103
S1-H-2d15-Rl75-R2100 e e 75 100 80.5
S1-H-2d15-Rl100-R2100 e e 100 100 69.5
S1-H-4d15-Rl100-R2100 e e 75 200 69.5
a
UHPC 1% volume fibres.

Fig. 2. Section configurations and the main geometrical parameters.

Table 5
Densities and weights of materials.

Material Type Density (kg/m3) Weight for linear metre (kg/lm)

Steel Ø14 (longitudinal) 7850 e

Ø8 (stirrup)

CFRP Ø15 (longitudinal) e 0.13

GFRP Ø12 (stirrup) e 0.29

Concrete Standard C25a 2279 e

UHPC* 2496 e
a
Asterisks are directly related to the experimental values obtained for the formulation of both concretes.

The quantities of UHPC for each of the 2-m beams mentioned conventional concrete should be added when considering former,
above are summarized in Table 7. whereas for the latter, the remaining section is empty. The section
UHPC is less critical for casting beams relative to sandwich that is the most profitable in terms of cost can be calculated. In
beams. However, from an economics point of view, the cost of addition, adhesion problems between the different concrete layers

Table 6
Weight comparison between standard beams and multi-layered beams reinforced with CFRP.

Specimen Material Number Diameter (mm) Length (m) Area (cm2) Volume (cm3) Weight (kg)

S1-ST-2d14 (steel) Longitudinal 2 14 2.3 3.08 708.11 5.56

Stirrup 20 9 0.6 10.05 603.19 4.74
Concrete C25 e e e e 8.5  104 193.58

Total weight of the standard beam 203.87

S1-3C-2d15 (FRP) Long. CFRP 2 15 2.3 3.53 812.89 1.196

Steel stirrup 20 8 0.6 10.05 603.19 4.74
UHPC e e e e 3.1  104 77.5
Concrete C25 e e e e 5.4  104 122.57

Total weight of the multi-layered beam 206

 E. Ferrier et al. / Composites Part B 90 (2016) 1e13 5

Table 7
UHPC volume comparison between hollow sections and sandwich sections.

Section UHPC volume (m3) % of the total section

Hollow section Section 1 0.02179 58%

Section 2 0.025523 67%

Sandwich section 0.0135 36%

could occur in the multi-layered beam, whereas the hollow beam is section corresponds to 157 cm2. The values are reported in Table 8.
monolithic. The second section further reduces the shear stress in the beam
section. Furthermore, the maximum shear stresses reached are
divided by more than two in the case of the two holes. Regarding
2.3. Shear design the expected maximum shear stresses, the use of UHPC is justified.
Preference is therefore given to the inclusion of two rather than one
2.3.1. Hollow beams cylinder in the recessed sections of the experimental design.
Using Autodesk Robot Structural Analysis software, two sections
are evaluated to determine the final configuration of the tested Table 8
beams. The first section (Fig. 3) consists of one hollow cylinder of Maximum shear stress of the two hollow section configurations.
200-mm diameter. To evaluate the influence of the arrangement
Section Empty surface (cm2) txz,max (MPa)
and the number of holes, the second configuration (Fig. 3) in-
corporates two hollow sections of 100-mm diameter holes. The first Section 1 (Fig. 3a) 122.7 22.06
Section 2 (Fig. 3b) 157 9.89
section has a surface of 122.7 cm2 vacuum, whereas the second

Fig. 3. Maximum shear stresses for the (a) two possible hollow beam configurations and (b) the three-layered beams.
6 E. Ferrier et al. / Composites Part B 90 (2016) 1e13

 
2.3.2. Multi-layered beams 3 cu
3f ¼ d  xf (11)
As described above, the multi-layered beam has a top 3-cm xf
UHPC layer, then a middle layer of typical concrete (C25) and a 6-
cm bottom layer. The top and bottom shear strengths are neglec- The integral is calculated discretely using the Simpson approxi-
ted in the design. The beam is strengthened by 8-mm diameter mation. In numerical analysis, the Simpson method, named after
transverse stirrups, and a stirrup is placed every 150 mm. The Thomas Simpson, is a method used to obtain a result of the inte-
calculation of the shear strength of the beam is provided by the gration of functions on a finite domain. The obtained result is
Eurocode 2 standard. generally easily approximated. In our case, b(z) depends on the
equation of a circle and varies as a function of the depth of the
VRd ¼ VRd;s (4) beam. The objective is to obtain the value of the integral given by
Eq. (5) at each step of the calculation.
Shear failure load is calculated according to
Zziþ1
Frupt;cis ¼ 2VRd (5) bðzÞdz (12)
zi1
The concrete strength contribution is given by equation (6).
The steel contribution is calculated according to equation (6). This method utilizes a second order approximation of the function
using a polynomial equation that takes the same value for each ziþ1,
Asw zi1 and m, where m is given by
VRd;s ¼ zfywk (6)
s
ziþ1 þ zi1
m¼ (13)
2
z ¼ 0:9d (7)
Eq. (7) can then be calculated.
where Asw is the transverse stirrups area, s is the spacing between
the stirrups, fywk is the steel strength, and z is 0.9d. ðz  mÞ$ðz  ziþ1 Þ ðz  zi1 Þ$ðz  ziþ1 Þ
bðzÞ ¼ bðzi1 Þ bðmÞ
The stirrups contribution is 64 kN. Finally, the calculated shear ðzi1  mÞ$ðzi1  ziþ1 Þ ðm  zi1 Þ$ðm  ziþ1 Þ
strength is 86.6 kN with a total failure load of 173 kN. ðz  zi1 Þ$ðz  mÞ
þ bðziþ1 Þ
ðziþ1  zi1 Þ$ðziþ1  mÞ
(14)
2.4. Flexural design of hollow beams and multi layered beams
It then becomes easy to integrate the function, such as Eq. (15)
The flexural capacity of the beams is calculated. For the flexural
strength, the compression of the concrete is limited, and it is Zziþ1    
ziþ1  zi1 z þ ziþ1
assumed that the deformation reaches 0.0027 m/m when failure of bðzÞdz ¼ $ bðzi1 Þ þ 4b i1 þ bðziþ1 Þ
the concrete occurs. The calculation is carried out using an incre- 6 2
zi1
mental iterative calculation. The neutral axis depth xf is unknown
(15)
due to geometrical properties of the section, nonlinear behaviour of
the UHPC, and the use of multi layered cross sections. The process
As the UHPC exhibits considerable tensile strength, the calculation
for calculating the resultant compression force is given in Fig. 4. The
of the tensile force takes into account the contribution of the UHPC.
beam depth is divided into 1000 layers along its height. First, it is
The same integration model used is for the compressive force. Due
assumed that xf ¼ h/2, and the stress distribution is calculated.
to the failure strain limit of the FRP, the maximal strain that is
Then, simplifying the mechanical behaviour law of the concrete to a
considered in the calculation for the UHPC is 1%. For each level of
bilinear law and knowing the constitutive law of reinforcement, the
strain and for each layer of the calculation, the stress is considered
stress of each layer corresponding to the assumed strain is
based on the tensile mechanical behaviour of the UHPC. Solving the
calculated.
system yields an approximate value of xf. The internal moment is
The resultant compression force is expressed for each layer by
then calculated using Eq. (16):
Eq. (8):
 
x
Zziþ1 Mr ¼ Fc d  f (16)
2
Fi ¼ sðzÞbðzÞdz (8)
zi1 The beams were tested in 4-point bending, and hence the failure
force during bending is calculated using Eq. (17)
ðsðzÞbðzÞÞi1 þ ðsðzÞbðzÞÞiþ1 h 6Mr
Fi ¼ Dz with Dz ¼ (9) Pu;fl ¼ (17)
2 1000 L
Based on the section equilibrium, the compressive and tensile
forces are equal, allowing the calculation of the neutral axis 3. Experimental results
position.
3.1. Specimen preparation and mechanical testing
X Z
ziþ1

Fc ¼ Ft ¼ Af Ef 3 f þ sðzÞbðzÞdz (10) The FRPeconcrete composite beams used in this study were
i zi1 fabricated via casting, as described above. The FRP rebar were fixed
in the mould due to its low density. The beams were cast hori-
where zontally from bottom to top. The FRP bars are straight at their end,
 E. Ferrier et al. / Composites Part B 90 (2016) 1e13 7

Fig. 4. Calculation methodology for the compression force.

and no specific anchorage is placed. The tensile force in FRP is
obtained only regarding the bonding issue.
The tests conducted to determine the bending stiffness and load
bearing capacities of the composite beams and their
momentecurvature relations are illustrated in Fig. 5. The beam
specimens, constructed as described above, were subjected to four-
point loading tests. The distance between the supports and applied
load was greater than twice the depth of the beam. For the 2-m
beams, this distance was 0.7 m (denoted a in Fig. 5), whereas for
the 4-m beams, it was 1.3 m (denoted a on Fig. 5). The distances
between the forces (b in Fig. 5) were 0.6 m and 1.4 m for the 2-m and
4-m beams, respectively. The loading was displacement-controlled.
The loads and displacements were measured at 2-s intervals using
both load cells and LVDT transducers and recorded using a data
logger. To obtain the strain distribution in the mid-span section,
three strain gauges were bonded to the lateral faces of the beams.
The rebar strains were also measured during loading using gauges
bonded to the rebar before the beams were cast. Before the tests, the
beams were conditioned for at least 28 days to mean values of
approximately 12% for moisture content and 20  C for temperature.
3.2. Loadedisplacement curve
Table 9 summarizes the cracking load of the beam, the final
rupture force, the state of strains in the concrete and the re-
inforcements, and the maximal displacement at failure.
An LVDT measures the deflection at mid-range, and the force is
measured using a sensor located at the point of the hydraulic servo
actuator loading. The analysis of the loadedisplacement curves
indicated that there were two stages of distinct behaviour during
the tests, corresponding to progressive damage in the constituent
materials (concrete and rebar). The loadedisplacement curves for
the five 2-m beams are shown in Fig. 6, and the curve for the 4-m
beam S2-H-4d15-R1100-R2100 is shown in Fig. 7. For all beams
composed of UHPC reinforced with CFRP, the first stage of the
behaviour corresponds to the uncracked section. During this stage,
the beams exhibited high bending stiffnesses. The second stage of
behaviour began when the load reached approximately 50 kN for
the 2-m beams and approximately 18 kN for the 4-m beams. In this
stage, the concrete in the bottom flange of the beams began to
crack, and the bending stiffness reduced. This behaviour is consis-
tent with the typical reinforced concrete beam mechanical
response. The behaviour of the composite beam remains linear
until failure. After the cracking load, the behaviour remained elastic
with constant stiffness until failure. This behaviour highlights the
advantage of using FRP rebar to increase the tensile capacity of the
beam due to the ultimate strength of the FRP, resulting in an in-
crease in the ultimate capacity of the beam (Table 10).
This first section is the S1-L*-2d15-hf130-hf260 beam, the three-
layered sandwich beam. The asterisk indicates that the formula
contains only 1% UHPC fibres. The cracking load at which the strain
in the rebar begins to increase matches the increasing strength of

Fig. 5. Four-point bending flexural tests.

8 E. Ferrier et al. / Composites Part B 90 (2016) 1e13
Table 9
Comparison of the experimental and theoretical results at failure.
Beam designation Mexp (kN m) Mr,calc (kN m)
S1-L-2d15-hfl30-hf260 60.3 98
S1-La-2d15-hfl30-hf260 66 84
S1-H-2d15-Rl75-R2100 67.1 125
S1-H-2d15-Rl100-R2100 69.2 104.5
S1-H-4d15-Rl100-R2100 100.7 117.6
a
S denotes shear failure, and F denotes flexural failure.
Fig. 6. Experimental loademid span deflection of the 2-m-long tested beams.
Fig. 7. Experimental loadedeflection of the 4-m-long beam.
the concrete during compression. This value is clearly illustrated in
Fig. 8. Before 22.8 kN, the section is not cracked, which corresponds
to Phase 1. During Phase 2, the cracks are opened and elastically
develop until force Fnl, at which nonlinear behaviour starts. In Phase
3, the deformations in the reinforcement and concrete do not
continue to increase due to the appearance of a large shear crack.
From 179 kN, the strain related to the shear takes precedence over
bending. This observation confirms the final shear failure.
The second beam (S1-L-2d15-hf130-hf260) is a three-layered
beam containing 2% UHPC fibre. The value of the cracking force is
approximately doubled. The passage of the uncracked section to the
opening crack phase is abrupt. In phase 3, the strain gauges were
destroyed before end of the experiment, where the sudden increase
in the value of the strain corresponds to the end of the test (Fig. 9).
The beam (S1-H-2d15-R175-R2100) section includes a 75-mm
top hole and a 100-mm bottom hole. A non-linear change in force
as a function of displacement appears during phase 2, and then a
Mexp
Mr;calc tmax (MPa) Mode of failurea
0.62 5.2 S
0.78 5.69 S
0.53 10.8 S
0.66 9.7 S
0.86 8.12 S and F
change in the slope of the rebar strain curve occurs after a load of
175 kN. Moreover, the CFRP tensile reinforcements are less stressed
than for the case of three layers (Fig. 10).
Next, the S1-H-2d15-R1100-R2100 beam is lighter in weight, as it
contains two hollow tubes of 100 mm in diameter. The crack
opening appears for a load of 9.4 kN, and the slope changes to
127.5 kN (Fig. 11). During Phase 3, the cracks open faster.
The final beam, S2-H-4d15-R1100-R2100, incorporates two hol-
low tubes that are 100 mm in diameter for a total beam length of
4 m. The cracking force of 12.7 kN is slightly greater than the same
2-m-long beam section. The beginning of phase 3 is similar to that
of the 2-m beam, corresponding to 131 kN vs. 127.5 kN for the 2-m
and final beams, respectively. However, the failure load is 151 kN,
which is 25% less than that of the 2-m beam. Indeed, the resistance
to shear force is reduced when the distance a/d is increased, where
a is the distance between the support and the load point, and d is
the height between the upper longitudinal reinforcement and top
concrete fibre. For a failure load of 151 kN, the failure mode is
borderline between bending failure and shear failure (Fig. 12).
The tensile strengths of the beams composed of type 2 and type 3
sections are approximately two times the failure strength of the
reference beam reinforced with steel because the PRF carbon bars
tensile strength is more than twice that of steel. For the reference
beam, the steel yielding strength is 540 MPa, corresponding to a
force approximately 90 kN. The other beams exhibit a quasi-elastic
behaviour until failure, influenced by the purely elastic behaviour of
the FRP reinforcement. Furthermore, before crack opening, the
bending stiffness of the reference steel rebar reinforced beam and
the hollow beams are equivalent, whereas the beams with three
layers exhibit a slightly lower bending stiffness. Thereafter, during
crack opening and the propagation phase, the hollow beams and
three-layered beams have higher bending stiffnesses than the
reference beam. This can be explained because UHPC corresponds to
a Young's modulus that is almost double in compression and tensile
behaviour that is superior to typical concrete. Finally, between the
two special configurations, the type 3 hollow section is 35% stiffer
than the type 2 three-layered beam for 22% UHPC (Table 7).
Loadedisplacement curve of the hollow 4-m-long beam is given
in Fig. 7. The hollow 4-m beam behaves similarly to the hollow 2-m
beams. After the cracking load, the behaviour of the beam remains
elastic, whereas only the stiffness slightly differs. From a certain
force, the opening of a larger crack leads to the shear failure of the
UHPC concrete, and the general behaviour is nonlinear. The values
of the cracking force given above are accurately defined using force-
mid span material strain graphs, as shown in Figs. 13e17.
In the following part, section at the mid-span of the beams is
studied to evaluate if the mechanical behaviour follows the
NaviereBernoulli hypothesis, i.e., whether the cross sections
remain planar during bending.
3.3. Strain distribution in the section at the mid-span
Based on the experimental strain measurements, the strain as a
function of the measurement location (depth of the beam) is
 E. Ferrier et al. / Composites Part B 90 (2016) 1e13 9

Table 10
Summarized experimental results.

Beam designation Fcr (kN) Ffail (kN) 3 c,max (mm/m) 3 frp,max mm/m) dmax (mm)
S1-ST-2d14 10 93 e e 23.2
S1-L-2d15-hfl30-hf260 51.4 177.4 1873 9997 23.6
S1-La-2d15-hfl30-hf260 22.8 194 2218 8759 27.3
S1-H-2d15-Rl75-R2100 18.3 197.4 1763 6251 17.7
S1-H-2d15-Rl100-R2100 9.4 203.6 1750 6984 21.3
S1-H-4d15-Rl100-R2100 12.7 151 2567 11,444 101
a
UHPC 1% volume fibres.

Fig. 8. Comparison of deflection and rebar strain vs. load in the S1-L*-2d15-hf130-hf260 Fig. 10. Comparison of deflection and rebar strain vs. load in the S1-H-2d15-R175-
beam. R2100 beam.

Fig. 11. Comparison of deflection and rebar strain vs. load in the S1-H-2d15-R1100-
Fig. 9. Comparison of deflection and rebar strain vs. load in the S1-L-2d15-hf130-hf260
R2100 beam.
beam.

the strain profiles confirms that no slip occurred at the interfaces

plotted for several loading levels to obtain the strain distribution in
the middle section of the beam. The tensile strain on the rebar
provides the minimum curve values. Fig. 13 shows the typical
evolution of strain in the central section of a beam at various load
levels. The lower strain values on the FRP bars are obtained using
strain gauges. The results for the RC beams containing carbon FRP
rebar are shown in Figs. 18e22. The strain distributions are planar
and remain planar throughout loading in the compressive part, and
a small change in slope occurs in the bottom part. The linearity of
between the FRP bars and the UHPC in this section. This observa-
tion also confirms the results previously obtained by Davalos et al.
[21], Won et al. [18], Yang et al. [22] and Baena [16], who conducted
pull-out tests and observed that use of high strength concrete with
FRP results in high shear strength and that perfect bonding can be
assumed. For example, according to Mazaheripour et al. [23], using
10-mm FRP bars and an anchorage length ten times the diameter,
approximately 20 MPa concrete shear strength was achieved in
UHPC.
10 E. Ferrier et al. / Composites Part B 90 (2016) 1e13
Fig. 12. Comparison of deflection and rebar strain vs. load in the S2-H-4d15-R1100-
R2100 beam.
Fig. 13. Comparison of concrete strain and rebar strain vs. load in the S1-L*-2d15-
hf130-hf260 beam.
Splitting failure never occurred because the fibre reinforcement
mechanisms prevented the degeneration of microcracks into
macrocracks [24].
The planar sections remain planar to failure. Some slip may
occur at the end of the FRP bars near the support. However, the
effect on the global behaviour of the beam and section equilibrium
can be neglected. The bonding between the UHPC and the FRP rebar
maintains strain continuity throughout the full height of the sec-
tion, which confirms the reliability of this composite beam
configuration. This was observed for all of the tested specimens.
This result confirms that the Navier and Bernoulli hypotheses
regarding the mechanical behaviour of the beams are satisfied and
are useful for modelling.
Strain diagrams can be used to calculate the beam curvature for
every load value and to measure the displacement of the neutral
axis from its original position. Before cracking, the neutral axis is
located at the mid-depth of the beam (Figs. 18e21). As cracking
progresses, the neutral axis shifts upward and stabilises at a height
of approximately 70% of the total beam height, as measured from
the bottom of the beam.
Fig. 14. Comparison of concrete strain and rebar strain vs. load in the S1-L-2d15-hf130-
hf260 beam.
Fig. 15. Comparison of concrete strain and rebar strain vs. load in the S1-H-2d15-R175-
R2100 beam.
Navier diagrams are drawn with the experimental values and
the positions of the gauges with respect to the beam height. Only
the Navier pattern of the three-layered S1-L-2d15-hf130-hf260
beam reinforced with 2% fibres does not have a clear intersection
point.
In this section, the vertical displacements of beams were
compared to the materials strains to identify the cracking and the
failure load.
3.4. Behaviour at failure
The 2-m beams all failed in shear, due to their short span, as
expected. Photographs of the rupture are shown in Figs. 23e26.
Note that the shear cracks of the hollow beams are more vertical
than those of the sandwich beams. The final failure of the 4-m-long
beam is shear, as shown in Fig. 27, but it was very close to the
tension failure in the lower part of the beam because the moment is
 E. Ferrier et al. / Composites Part B 90 (2016) 1e13 11

Fig. 16. Comparison of concrete strain and rebar strain vs. load in the S1-H-2d15-
R1100-R2100 beam.

Fig. 19. Strain along the midspan depth in the S1-L-2d15-hf130-hf260 beam.

Fig. 17. Comparison of concrete strain and rebar strain vs. load in the S2-H-4d15-
R1100-R2100 beam.

Fig. 20. Strain along the midspan depth in the S1-H-2d15-R175-R2100 beam.

Fig. 18. Strain along the midspan depth in the S1-L*-2d15-hf130-hf260 beam. Fig. 21. Strain along the midspan depth in the S1-H-2d15-R1100-R2100 beam.
12 E. Ferrier et al. / Composites Part B 90 (2016) 1e13

Fig. 24. Failure of beam S1-L*-2d15-hf130-hf260.

Fig. 22. Strain along the midspan depth in the S2-H-4d15-R1100-R2100 beam.

close to the maximum resistant moment of the section. The resis-

tant moments for all of the FRP reinforced beams were calculated
(Table 9). For the 2-m-long beams, the calculated moments are far
from the experimental moments, as the shear failure appeared
prematurely. It should be noted that experimental moment is ob-
tained by dividing the failure force by 2 and multiplying the lever
arm by L/3. The maximum shear stress is also calculated using
Autodesk Robot Structural Analysis software, and maximum shear
stresses of approximately 5e6 MPa for the normal strength con-
crete in case of the three-layered beams or approximately
8e10 MPa for UHPC in case of the monolithic hollow beams are
coherent values.
Using the mid-section strain measurements (Table 10), the
mechanical behaviour of the composite beams and the efficiency of
the RC solution at failure were evaluated. For the upper UHPC
section under compression, the strain rate at failure was greater
than 2500 mm/m for the S2-H-4d15-R1100-R2100 beam (for the 4-m
hollow beam) and nearly 2000 mm/m for the S1-L-2d15-hf130-hf260
and S1-L*-2d15-hf130-hf260 beams (3-layered beams). For the FRP
rebar, the maximum strength was attained in for the S2-H-4d15- Fig. 25. Failure of beam S1-H-2d15-R175-R2100.
R1100-R2100 beam and the S1-L-2d15-hf130-hf260 beam

Fig. 23. Failure of beam S1-L-2d15-hf130-hf260. Fig. 26. Failure of beam S1-H-2d15-R1100-R2100.
 E. Ferrier et al. / Composites Part B 90 (2016) 1e13
Fig. 27. Failure of beam S2-H-4d15-R1100-R2100.
containing the CFRP bars, whereas approximately 80% of the CFRP
strength was reached at failure for the S2-H-4d15-R1100-R2100
beam. For the 2-m beams, the tensile stress in the CFRP varied from
40 to 70% of the maximum FRP strength due to the premature shear
failure of the beam. The beam design allows each material in the
concrete to reach a high strain at failure during compression,
confirming the benefits of this approach.
At the maximum load level, various failure modes were
observed, depending on the span and characteristics of the beam.
The 2-m beams failed in shear, whereas the 4-m beams exhibited
shear, almost flexural, failure (Fig. 27). This confirms that for a Lw/
hw ratio of approximately 8, the design for shear should be carefully
investigated, and the width of the section should be increased to
prevent shear failure. When the Lw/hw value is approximately 16,
the tensile or crushing failure is critical design criteria if the width
of the beam is large. The failure of the RC beams occurred between
loads of 90 kN (for the RC S1-ST-2d14 beam) and 204 kN (for the S1-
H-2d15-R1100-R2100 beam) (Table 10). There was no debonding of
the FRP rebar from the UHPC in any of the beams, unlike the FRP
bars embedded in standard concrete [25,26]. From these results, it
was concluded that these new innovative RC beams allow UHPC to
support a higher stress at failure and that the compressive strength
of UHPC tensile failure of the FRP can be obtained depending on the
design, which improves the high performance material used.
4. Conclusions
This study describes the development and testing of an inno-
vative RC beam constructed of ultra-high-performance concrete
and reinforced with micro fibres and FRP reinforcing bars. The re-
sults demonstrate that the beam exhibits typical RC beam behav-
iour in terms of concrete cracking and shear failure that is nearly
flexural failure. The results also show that reinforcing concrete with
carbon FRP rebar increases its bending stiffness, even when using
small rebar diameter, by increasing its Young's modulus.
To analyse the cracking behaviour of the concrete under tension,
an iterative calculation method was developed. The maximum
13
bending capacity can be estimated. The ultimate shear load is an
issue that was investigated here, and the shear failure can be
correctly calculated for both multi-layered beams and hollow
beams.
This study confirms that the composite action of high-
performance materials in an RC beam can result in a lightweight
structure that is mechanically efficient.
The results of this study should be confirmed by a more
extensive experimental program incorporating large-scale beams
and a larger number of specimens. The fatigue and creep behav-
iour of the RC beam developed in this study should also be
investigated.
References
[1] Behloul M, et al. The Sherbrooke footbridge: the first reactive powder concrete
structure. Struct Eng Int 1998:140e4.
[2] AFGC. Ultra-high performance fiber reinforced concrete [Interim recommen-
dations, revised edition]. France: AFGC publication; 2013.
[3] Acker P, Behloul M. Ductal technology: a large spectrum of properties, a wide
range of applications. 2004 [FIB symposium, Avignon, France].
[4] Perry V, Royce M. Innovative field-cast UHPC joints for precast bridge decks
(side-by-side deck bulb-tees), Village of Lyons, NY e design, prototyping,
testing and construction. In: Concrete bridge conference, USA; 2007.
[5] Akhnoukh AK, Xie H. Welded wire reinforcement versus random steel fibers
in precast/prestressed ultra-high performance concrete I-girders. Constr Build
Mater 2010;24(11):2200e7.
[6] Chaallal O, Benmokrane B. Fiber-reinforced plastic rebars for concrete appli-
cations. Composites Part B: Eng 1996;27(3e4):245e52.
[7] El-Hacha R, Chen D. Behavior of hybrid FRP-UHPC beams subjected to static
flexural loading. Composites Part B: Eng 2012;43(2):582e93.
[8] Yinghao L, Yong Y. Arrangement of hybrid rebars on flexural behavior of HSC
beams. Composites Part B: Eng 2013;45(8):22e31.
[9] Berg AC, Bank LC, Oliva MG, Russell JS. Construction and cost analysis of an
FRP reinforced concrete bridge deck. Constr Build Mater 2006;20(2):515e26.
[10] Hao Q, Wang Y, He Z, Ou J. Bond strength of glass fiber reinforced polymer
ribbed rebars in normal strength concrete. Constr Build Mater 2009;23(2):
865e71.
[11] Achillides Z, Pilakoutas K. Bond behavior of fiber reinforced polymer bars
under direct pullout conditions. J Compos Constr 2004;8(2):173e81.
[12] Cosenza R, Manfredi E, Realfonzo G. Behavior and modeling of bond of FRP
rebars to concrete. J Compos Constr 1997;1(2):40e51.
[13] Ahmad F, Foret G, Roy RL. Bond between carbon fibre-reinforced polymer
(CFRP) and ultra high performance fibre reinforced concrete (UHPFRC):
experimental study. Constr Build Mater 2011;25(2):479e85.
[14] Lin X, Zhang Y. Bond-slip behaviour of FRP-reinforced concrete beams. Constr
Build Mater 2013;44(3):110e7.
[15] Malvar LJ, Cox JV, Cochran KB. Bond between carbon fiber reinforced polymer
bars and concrete: experimental study. J Compos Constr 2003;7(2):154e63.
[16] Baena M. Study of bond behaviour between FRP reinforcement and concrete
[PhD thesis]. Girona, Spain: The University of Girona; 2010.
[17] JSCE. Recommendations for design and construction of ultra-high strength
fiber reinforced concrete structures. Tokyo, Japan: Japan Society of Civil En-
gineering (JSCE); 2006.
[18] Won Jong-Pil, Park Chan-Gi, Kim Hwang-Hee, Lee Sang-Woo, Jang Chang-Il.
Effect of fibers on the bonds between FRP reinforcing bars and high-strength
concrete. Composites Part B: Eng July 2008;39(5):747e55.
[19] Habel K, Viviani M, Denarie  E, Brühwiler E. Development of the mechanical
properties of an ultra-high performance fiber reinforced concrete (UHPFRC).
Cem Concr Res July 2006;36(7):1362e70.
[20] CEN. Eurocode2: design of concrete structures e part 1.1: general rules and
rules for buildings (EN 1992-1-1:2004). Brussels, Belgium: Comite  Europe
en
de Normalisation; 2004.
[21] Davalos JF, Chen Y, Ray I. Effect of FRP bar degradation on interface bond with
high strength concrete. Cem Concr Compos 2008;30:722e30.
[22] Yang Jun-Mo, Min Kyung-Hwan, Shin Hyun-Oh, Yoon Young-Soo. Effect of
steel and synthetic fibers on flexural behavior of high-strength concrete
beams reinforced with FRP bars. Composites Part B: Eng April 2012;43(3):
1077e86.
[23] Mazaheripour H, Barros JAO, Sena-Cruz JM, Pepe M, Martinelli E. Experimental
study on bond performance of GFRP bars in self-compacting steel fiber rein-
forced. Compos Struct January 2013;95:202e12.
[24] Pepe M, Mazaheripour H, Barros J, Sena-Cruz J, Martinelli E. Numerical cali-
bration of bond law for GFRP bars embedded in steel fibre-reinforced self-
compacting. Composites Part B: Eng July 2013;50:403e12.
[25] Wu ZJ, Ye JQ. Strength and fracture resistance of FRP reinforced concrete
flexural members. Cem Concr Compos 2003;25:253e61.
[26] Bakis CE, Uppuluri VS, Nanni A, Boothby TE. Analysis of bonding mechanisms
of smooth and lugged FRP rods embedded in concrete. Compos Sci Technol
August 1998;58(8):1307e19.