Materials and Structures (2015) 48:2713–2726

DOI 10.1617/s11527-014-0348-x

ORIGINAL ARTICLE

Failure mechanism of fibre reinforced concrete

under splitting test using digital image correlation
Bensaid Boulekbache • Mostefa Hamrat •
Mohamed Chemrouk • Sofiane Amziane

Received: 13 June 2013 / Accepted: 29 May 2014 / Published online: 20 June 2014
Ó RILEM 2014

Abstract The study aims at analysing the fracture specimens after the main one. Moreover, correlations
mechanism of steel fibre-reinforced concrete in tensile between the main mechanical properties were devel-
splitting using digital image correlation. The work also oped. The results show that the fibres and the
aims to exam the influence of both the steel fibre compressive strength of concrete have a significant
addition and the compressive strength on the behav- effect on the behaviour of FRC in splitting. The fibres
iour of fibre-reinforced concrete (FRC) in splitting were more effective in high strength matrix by
tension test. Cylindrical concrete specimens having increasing both the ultimate load and the ductility.
110 mm diameter and 220 mm height were subjected This increase, which reached up to 70 %, is attributed
to splitting tests. The cylinders specimens were made to a better quality bond between the steel fibres and the
with different concrete types and reinforced with steel matrix.
fibres of two aspect ratios (65 and 80) using three
volume fractions (0, 0.5 and 1 %). The types of Keywords Fibre-reinforced concrete  Compressive
concretes with various compressive strengths and strength  Splitting tensile strength  Crack  Ductility 
rheology consisted of an ordinary concrete, a self- Aramis
compacting concrete and a high strength concrete. The
test results revealed the existence of two failure
mechanisms; the first one is associated with a central
unique crack, whereas the second is related to a 1 Introduction
secondary cracks initiated at the edges of the test
Fibre-reinforced concrete (FRC) has known many
developments and numerous studies have been carried
B. Boulekbache  M. Hamrat
Département de Génie Civil, Université Hassiba out on this material during the last three decades, in
Benbouali de Chlef, Chlef, Algeria order to overcome the tension and ductility weak-
e-mail: bboulekbache@yahoo.fr nesses common to all types of concretes.
An increase in the strength of concrete is associated
M. Chemrouk
Université des Sciences et de la Technologie Houari with an increase of both the compressive and the
Boumediene, Alger, Algeria tensile strengths [1] since both properties are strongly
dependent on the quality of the bond developed
S. Amziane (&)
between the aggregate and the cement paste. However,
Institut Pascal, Polytech’Clermont-Ferrand, Clermont
University, 63174 Aubière, France concrete with higher strengths is known to be more
e-mail: sofiane.amziane@univ-bpclermont.fr brittle than ordinary strength concrete and this
2714 Materials and Structures (2015) 48:2713–2726

drawback represents a significant limitation for their values between the splitting tensile strength and the
wide-range applications for innovative structural true tensile strength is due to the quasi-brittle material
design. behaviour of concrete and the related fracture mechan-
A particular purpose for adding fibres to a concrete ics size effect. Rocco et al. [13] showed that the split
matrix is to improve the ductility of the hardened cylinder test can yield in a true indication of the tensile
concrete material in order to make it more suitable for strength as long as the width of the loading strip does
use in structures subjected to impacts and earthquake not exceed 8 % of the diameter of the specimen.
loads. High-strength concrete containing steel fibres, The splitting test has several advantages such as the
therefore, seems to have an efficient process confer- easy specimen preparation; the simplicity of the test;
ring to the hardened material a relatively better the similarity in the specimen and testing equipment as
ductility, particularly that the steel fibres develop an for the compressive test; the splitting failure, which
improved bond with the dense and compact hardened initiates in the core of the specimen, is not affected by
cement matrix in this relatively new material [2]. the surface conditions. Moreover, in the splitting test,
During the past three decades, a number of research only external compressive loads are required. In this
works pertaining to experimental and analytical test, a cylindrical concrete specimen is compressed
methods for evaluating the strength characteristics of along its length (Fig. 1). According to the theory of
fibre-reinforced concrete have been published under elasticity, this loading generates almost uniform
varied specimen types, fibres types, fibres contents, tensile stress along the diameter, which causes the
curing time and testing methods [3]. Although the specimen to fail by splitting across a vertical plane
behaviour of concrete is governed significantly by its [13]. To prevent crushing at the loading points, the
compressive strength, its tensile strength is also load is distributed over two bearing strips whose
important with respect to the appearance and hence widths vary according to the standard used. The only
the serviceability and, at the longer term, the durability major disadvantage of the splitting test is that the
of this widely used material. Therefore, the compres- loading condition does not reflect a practical situation
sive strength, the tensile strength and the deformations for most applications.
ability (ductility) are important parameters for char- Yamaguchi and Peters are among the first to
acterizing FRC mechanical properties. In this sense, introduce the DIC method. The DIC technique has
while the compressive strength is necessarily required been successfully used to study the textile deforma-
in any structural design, the tensile strength is vitally bility, the cyclic/fatigue performance of polymer
required in some specific structural applications, such materials, the 3-D damage in concrete prisms, the
as earthquake resistant structures, liquid containing
structures, airfield runway, pavement slabs and so on.
There are three types of tests to measure the
strength in tension (i.e. direct tension, flexure and
splitting tension) [4]. It has been, however, well-
known that the simplest and the most reliable method
which generally provides a lower coefficient of
variation is the splitting tensile test of a cylindrical
specimen [5–7]. This testing method, also known as
the Brazilian test, was first proposed by Carneiro and
Barcellos [8] during the Fifth Conference of the
Brazilian Association for Standardization in 1953 and
has became a standardized testing method which is
adopted in various international concrete testing
standards such as ASTM C496, ISO 4105, BS 1881
[9–11]. The splitting tensile strength can be used to
estimate the direct tensile strength by multiplying with
a conversion factor of 0.9, as given in the CEB-FIB
Code [12] and by Hannant et al. [5]. The difference in Fig. 1 Load configuration in a splitting test
Materials and Structures (2015) 48:2713–2726 2715

crack variations in masonry walls, the cracking Table 1 Properties of steel fibres used
process in RC slabs, the behaviour of prestressed Type 1 Type 2
concrete structures [14]. This technique was later on
extended by others to measure the 3-D displacements Length lf (mm) 35 60
[15, 16] and the tensile strength, to monitor the crack Diameter df (mm) 0.55 0.75
development in concrete beams [17–19], to study the Aspect ratio (lf/df) 65 80
corrosion of steel reinforcement [20] and to assess the Cross-section area Af (mm2) 0.24 0.44
strengthening of RC beams with FRP [21]. Tensile strength (MPa) 1,100 1,100
However, no work has been reported on the use of Number of fibres per kg 14,500 4,600
the DIC technique for characterizing the behaviour of
FRC specimens under splitting test. Therefore, it was
thought necessary to study this behaviour using such Two types of hooked-end steel fibres with different
an advanced imaging technique. aspect ratios were used in this investigation. Table 1
The main purpose of the present work is to provides the aspect ratio, the length and the diameter
investigate experimentally the effect of some param- for the different steel fibres used. These fibers, which
eters on the splitting strength of steel fibre-reinforced had specific gravity of 7.80, are glued together in
concrete with various levels of rheology and com- bundles with a water soluble glue, and hence can be
pressive strength of the material and various fibre easily dispersed in the fresh mix. The fibres were
volume fractions and aspect ratios (l/d: length–diam- added in small amounts to avoid fibre balling and to
eter ratios). The failure mechanisms of FRC in the produce a concrete with uniform material consistency
splitting tests are also studied. This work provides an and good workability. Two fibres dosages 0.5 and 1 %
important information about the first crack initiation by volume of concrete (i.e. 39 and 78 kg/m3 of
and crack propagation using a measuring system based concrete) were made along with a concrete without
on image analysis, obtained with a digital recording fibres for reference.
camera, which allows to detect the first crack with a In the testing program, three concrete mixes were
great precision, measures the crack opening and manufactured: an ordinary fluid concrete (OC), a self-
follows the progressive cracking process until failure. compacting concrete (SCC) and a plastic high strength
This measurement method leads to a precise determi- concrete (HSC), with the mix portions given in
nation of the specimen’s ductility. Table 2. For each of the three mixes, six concrete
cylinders (110 9 220 mm) were cast to determine
both the compressive strength and the splitting tensile
2 Experimental work strength of the three types of concrete. Compaction
was achieved by means of vibrating table. The
2.1 Materials and fabrication of the test specimens samples were demoulded after 24 h and cured in a
moist room maintained at 23 °C and 95 % relative
The cement (C) used in this study was a CEM I 52.5 N, humidity until the day of testing.
meeting the requirements of European Standards EN
197-1, having a fineness expressed through the specific 2.2 Splitting test setup
surface of 3,520 cm2/g and a density of 3,160 kg/m3;
the sizes for the cement particles ranged from 0 to Prior to the testing execution, the faces of the
100 lm. Standard silica fume (SF) and limestone filler specimen were polished. The tests were conducted
(LF) was used as mineral additive; their specific surface on a servo-hydraulic universal testing machine. The
was 23 m2/g, 3,970 cm2/g respectively. The fine and tests were carried out under displacement-controlled
coarse aggregates had minimum/maximum sizes of conditions with a loading rate of 0.01 mm/s through
0/4 mm and 4/10 mm respectively. The high-range two loading strips of 9 mm width (8 % of the 110 mm
water-reducing admixture (HRWRA) was a polycar- diameter) and 4 mm thickness in accordance with the
boxylate-based admixture used to achieve an adequate recommendations by Rocco et al. [13] for reduction of
fibre dispersion and workability. The specific gravity of size effect. These loading strips are glued to the
the admixture was 1.05. specimen and were not reused.
2716 Materials and Structures (2015) 48:2713–2726

Table 2 Mix proportions of concretes in kg/m3

Mixture OC FROC FROC SCC FRSCC FRSCC HSC FRHSC FRHSC
35-0.5 35-1 35-0.5 35-1 35-0.5 35-1
FROC FROC FRSCC FRSCC FRHSC FRHSC
65-0.5 65-1 65-0.5 65-1 65-0.5 65-1

Cement 275 275 275 425 425 425 425 425 425
Silica fume 0 0 0 0 0 0 42.5 42.5 42.5
Limestone filler 90 90 90 200 200 200 90 90 90
Gravel 4/10 910 900 900 825 814 814 825 814 814
Sand 0/3.15 830 820 820 750 740 740 750 740 740
Water 178 178 178 192 192 192 161 161 161
HRWRA 0 1.95 1.95 5.10 7.22 7.22 4.25 6.75 6.75
Steel fibres – 39 78 – 39 78 – 39 78
W/Pa 0.49 0.49 0.49 0.32 0.32 0.32 0.30 0.30 0.30
The first number following the type of concrete is the length of the fibre (35 or 60 mm)
The number after hyphen gives the dosage of the fibres by volume (0.5 or 1 %)
For example, FROC 35-0.5: ‘‘35’’ is the length of the fibre in mm, ‘‘0.5’’ is the dosage of the fibres by volume
FROC fibre reinforced ordinary concrete, FRSCC fibre reinforced self-compacting concrete, FRHSC fibre reinforced high strength
concrete
a
P (Powder) = C ? SF ? LF

The transversal displacement of the specimen (i.e. camera ‘‘Aramis’’ by GOM (http://www.gom.com) [25]
displacement perpendicular to the loading plane) was (Fig. 2a, b) enabled us to detect the first cracking with a
measured by averaging signals from the linear variable great precision, to measure the crack opening and to
displacement transducers (LVDTs) mounted on datum follow the progression of the cracking process through-
points which were glued to the second side of the out the splitting test. It has thus allowed correcting any
specimen as shown in Fig. 2c. Photos of both test possible errors in calculating the ductility indexes.
configurations are shown in Fig. 2. Aramis is a non-contact optical 3D deformation
The applied load and deflection data were elec- measuring system. The setup comprised two cameras
tronically acquired at an acquisition frequency of with a resolution of 4 megapixel (2048 9 2048
1 Hz. All measurements were made on three speci- pixels), with a maximum frame rate of 7 Hz, posi-
mens. The splitting tensile strength, fsp (MPa), can be tioned 535 mm away from the specimen and 400 mm
calculated using the following equation [22]: apart from each other. Two lenses with focal length of
2P 50 mm were used. Images were captured during
fsp ¼ ðMPaÞ ð1Þ testing with time intervals of 0.20 s and coupled to a
pDL
data acquisition card to store images on the hard drive
where P is the load. L and D are the length and the of a high-performance computer. A uniform/random
diameter of the cylinder specimen respectively. speckle pattern is applied on the entire side of the
specimen as shown in Fig. 2d. An external light source
2.3 Methodology of detecting crack propagation was also provided in order to maintain a constant
illumination on the measurement surface. The size and
The first cracking is defined in the load–deflection curve location of measuring volume with respect to the
as the point at which the first deviation from linearity cameras, and the position of the two cameras with
occurs. However, locating the point of first crack is respect to each other are first determined by taking a
subject to human error [23, 24]. In the present study, the set of pictures of the so-called ‘‘calibrating objects’’
use of a measuring system based on a digital image which have targets mounted on them with known sizes
correlation technique obtained by a digital recording and distances to each other.
Materials and Structures (2015) 48:2713–2726 2717

LVDT
Specimen

Digital
cameras

(a) Camera ARAMIS system (b) Side 1 of the sample

LVDT
Bearing
strips

(c) Side 2 of the sample (d) Uniform/random speckle pattern

applied on the measuring surface

Fig. 2 Tensile splitting test

The idea of the camera is to compute, for the initial of DIC technique for measuring the deformations.
state (before loading), the positions of points repre- However, the major limitation of using DIC is the loss
sented by a ‘‘speckle’’ on a well-defined zone of the of data in the regions where excessively large cracks
specimen. During the test, with a regular time interval t, and/or spalling of concrete is observed.
photos are taken, each one making it possible to measure
a new position of the points. A calculation algorithm
then analyzes these consecutive measurements and 3 Test results and discussions
calculates relative displacement compared to the initial
state (field measurement technique). This allows the 3.1 Fresh concrete properties
construction of the displacement fields of different
points depending on the load and the time. The first The measurement of the slump flow indicates the
crack, the crack profile and the crack opening and the workability of the mixture. The compactness of the
crack tip position can be evaluated for each load step. SCC matrix, due to the greater amount of fine and
This system allows a varied use, in particular in extra-fine particles, may improve interface zone
terms of three-dimensional visualization of the fields properties [26]. Consequently, it may also improve
of deformations, and enables the location of the zones the fibre-matrix bond, leading to an enhanced post-
of deformation concentration and the detection of the cracking toughness and energy absorption capacity. In
birth of the cracks which remain difficult to follow. most applications, the combination of fibres with sand
The correlation obtained by comparing DIC-Aramis and gravels affects the packing density of both fibres
and LVDT results for one sample (FRSCC35-1) is and aggregates [27].
satisfactory (Fig. 3). A good correlation was observed Slump and slump flow tests were carried out to
between the two approaches, demonstrating the validity determine the consistency of fresh concrete using
2718 Materials and Structures (2015) 48:2713–2726

Abrams cone (cone size 100–200 mm diameter, around 7, 6, and 10 % for FROC, FRSCC and FRHSC
height 300 mm). The ‘slump flow’ is the average respectively. In fact, the fibres are more efficient when
diameter of the horizontal flow measured in two subjected to tension; they then can prevent the
perpendicular directions after lifting Abrams’ cone. propagation of cracking. In compressive crushing,
The results of the rheological properties of the the role of fibres is reduced and could even have a
concretes tested are given in Table 3. The flowability negative effect through local stress concentration, pre-
rate decreases slightly with the addition of fibres, initiating crushing in the hardened cement paste by
although the rheological properties of fibre-reinforced punching it. The reduction in the compressive
concretes (slump and slump flow) are not dramatically strengths has been widely studied and documented in
changed in this study. In the case of SCC, no sign of the literature. It is thought that the addition of fibres
segregation was detected, the mixture showed good into a concrete mix substantially disturbs the matrix by
homogeneity and cohesion. inducing voids and hence additional defects in the
matrix through insufficient compactness leaving
3.2 Compressive strength empty places for an increased porosity of FRC [28–
30].
The compression test was performed on 110 9 The role of fibres in the pre-peak zone is negligible.
220 mm cylindrical concrete specimens following The mechanical role of fibres is activated only after the
EN 12390-3. The axial displacements were measured appearance of cracking; they greatly improve the post-
using three LVDTs arranged in a horizontal circular peak behaviour and considerably reduce the descend-
arrangement. An electronic data acquisition system ing slope of the stress–strain curve by softening the
was used to record the displacements and the corre- material when approaching failure as can be seen in
sponding loads. Fig. 4. This observation is even more pronounced in
Typical compressive stress–strain curves are shown the case of high-strength concrete. A slight increase in
in Fig. 4. In the present study, the presence of fibres the strain at the peak stress can be noted for the fibre-
has slightly reduced the compression strength by reinforced concrete.
Splitting tensile strength (MPa)

8 90
Compressive strength (MPa)

HSC
7 80

6 70 FRHSC
SCC
60
5
50
4
40 FRSCC
3 LVDTs
DIC-ARAMIS 30
2 20
OC FROC
1 10
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 0 1 2 3 4 5 6 7 8 9 10
Transversal displacement (mm) Axial strain (‰)

Fig. 3 Comparison between LVDT and DIC-Aramis transver- Fig. 4 Strength–strain curve in compression for concretes with
sal displacement and without fibres

Table 3 Rheological properties of concretes tested

OC FROC 35-0.5 FROC 35-1 SCC FRSCC 35-0.5 FRSCC 35-1 HSC FRHSC 35-0.5 FRHSC 35-1
FROC 65-0.5 FROC 65-1 FRSCC 65-0.5 FRSCC 65-1 FRHSC 65-0.5 FRHSC 65-1

Slump (cm) 24 21 20 – – – 16 13 12
Slump flow (cm) – – – 70 65 63 – – –
Materials and Structures (2015) 48:2713–2726 2719

Table 4 Compressive strength and splitting tensile strength of concretes tested

Compressive First crack Splitting tensile Increase Standard
strength fc (MPa) strength (MPa) strength fspu (MPa) in fspu (%) deviation

OC 29.3 – 2.82 – 0.33

FROC 35-0.5 27.0 2.09 3.09 9 0.13
FROC 35-1 28.4 2.14 3.50 43 0.22
FROC 60-0.5 28.3 2.11 3.13 11 0.47
FROC 60-1 28.5 2.37 3.24 40 0.11
SCC 61.3 – 4.50 – 0.24
FRSCC 35-0.5 57.1 3.55 6.70 15 0.39
FRSCC 35-1 53.6 4.10 7.43 41 0.40
FRSCC 60-0.5 56.7 3.71 6.38 9 0.27
FRSCC 60-1 54.6 3.95 6.90 31 0.47
HSC 82.6 – 4.57 – 0.45
FRHSC 35-0.5 79.1 4.05 6.75 41 0.36
FRHSC 35-1 78.8 4.80 7.73 59 0.32
FRHSC 60-0.5 73.9 4.34 6.42 34 1.10
FRHSC 60-1 72.2 4.75 6.99 45 0.28

3.3 Splitting tensile strength tensile strength which is greatly influenced. Records of
increases ranging from 15 to 40 % for a fibre volume
The results concerning the average values of the fraction of 0.5 % and from 40 to 59 % for 1 % were
measured cracking strength (fcr); the ultimate splitting obtained. The aspect ratio of fibres has a slight
strength (fspu) and the percent increase in ultimate influence on the splitting tensile response of the
strength of FRC specimens are given in Table 4. A specimens; the results revealed an improvement of
typical split-cylinder test set-up after failure used in about 10 % in the splitting tensile strength for an
this study is shown in Fig. 5. increase of their aspect ratio from 65 to 80.
The split tensile strength testing results of the fibre-
reinforced concrete specimen with different fibre 3.4 Failure pattern of specimens
volume fractions and aspect ratios are presented in
Table 4. The standard deviations of the tests vary from As stated above, the morphology of cracks is carried out
0.11 to 1.10 %. The results show that all types of by the digital image camera; the images in Figs. 6, 7, 8
concrete specimens provided higher splitting strength and 9 show the cracking evolution of the specimens
than those of their respective plain concrete. with the type of concrete. The more the colour evolves
In general, the splitting tensile strength gain was to red, the higher the strains become. It can be noted that
higher for high strength level specimens with higher the cracks obtained from the non fibred concrete are
percentages of fibre volume and larger fibre aspect ratio. straight (see Fig. 6). Usually, once the tensile strength
The enhanced performance of fibres in concrete with of the material is reached, a crack initiates at the centre
high strength levels is attributed to the improved bond of the specimen and propagates along the diameter
characteristics between the fibre and the matrix [31]. loaded toward the loading points. The crack opening
Concrete of higher strength reinforced with fibre and its length increase when the transversal displace-
provided greater splitting tensile strength increases than ment increases. The failure was brittle and fragile;
those of ordinary concrete. For FRHSC60-1 specimens especially for high strength concrete (Fig. 6).
with 1 % fibre volume and aspect ratio of 80, the Figures 7 and 8 show a typical sequence of the
increase was 115 % in comparison with FROC60-1. evolution of the splitting failure of fibre-reinforced
The first crack strength however, is not affected by concrete. For steel fibre-reinforced concrete, the main
the presence of fibres by comparison to the ultimate crack starts in the centre of the specimen and
2720 Materials and Structures (2015) 48:2713–2726

FRHSC), related oblique cracks appear on both sides

of the feed belts, but their lengths and widths are small
(Fig. 9). The origin of these cracks is the localization
of maximum tensile stresses at the loading regions.
These cracks develop in depth, parallel to the main
crack. When approaching failure, the development of
the related cracks become asymmetric, the width of
the main crack becomes very important and then the
sample would split into two half-cylinders.

3.5 Fracture mechanism in tensile splitting test

Fig. 5 Typical split-cylinder of FRC Typical load-transversal displacement curves obtained

for the cylinder splitting tests performed as part of this
propagates along the plane of symmetry to the feed study are shown in Fig. 10. A curve showing that there is a
belts. This crack is still very often the main crack with low discrepancy between the tests (Fig. 10a). The
an opening greater at the centre, although often other observed material behaviour shows two peak load
secondary cracks appear. Once the main crack was conditions, one at the end of the cement matrix contri-
extended along the plane of symmetry of the specimen bution and the second when the fibres reach their ultimate
(transversal displacement (d) greater than 55 lm for capacity. The results indicate that the material behaves

Fig. 6 Morphology of
cracking apparent on the OC SCC HSC
face of a specimen of
concrete without fibres

fsp= 2.20 MPa, δ = 14 μm fsp= 3.27 MPa, δ = 14 μm fsp= 4.52 MPa, δ = 15 μm

failure failure failure

fsp= 2.39 MPa, δ = 74 μm fsp= 4.06 MPa, δ = 88 μm fsp= 4.57 MPa, δ = 90 μm

Fig. 7 Sequence of splitting failure of fibre-reinforced concrete

Materials and Structures (2015) 48:2713–2726 2721

Propagation of the
First cracking related oblique Asymmetric related Development of
main crack to the Failure (6)
(1) cracks (3) cracks (4) related cracks (5)
feed belts (2)

image 1 image 2 image 3 image 4 image 5 image 6

8
FRSCC 35-1
7 image 5
Splitting tensile strength (MPa)

image 4 image 6
6 x x
x x
5 x
image 3

4 image 2
x

3 image 1

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2
Transversal displacement (mm)

Fig. 8 Cracking patterns of a tested specimen FRC

elastically in a linear manner until the first peak is reached. 3.6 Relationship between the splitting tensile
The first crack appears once the cracking has been strength and the compressive strength
initiated in the central zone of the specimen. The load
stabilizes at this level before increasing again with an The compressive strength (fc) and the splitting tensile
increased displacement. Eventually, because crack prop- strength (fsp) are two important parameters for char-
agation is restrained by fibres, a second and higher peak acterizing the mechanical properties of concrete.
load is reached. In conventional splitting tests, this second Figure 11 shows the relationship between fsp and fc
ultimate peak load is the only value that would be for plain concretes. It can be observed that the splitting
recorded. The results of this study however, indicate that tensile strength increases with an increase in the
the ultimate peak load is related to a secondary cracking compressive strength. The measured 28-day splitting
mechanism in the tests and not to linear elastic tensile tensile strength is in the range of 2.40 to 5.10 MPa.
stress distribution in the cylinder. Fibres with an aspect The traditional empirical approach to estimate the
ratio of 65 are more effective in restraining small openings splitting tensile strength fsp from the compressive
of cracks, whereas those with an aspect ratio of 80 are strength fc is given in the following mathematical form
effective in restraining larger crack openings as shown in [32, 33]:
Fig. 10, this result was also found by Pereira et al. [17].
fsp ¼ mðfc Þn ð2Þ
At high matrix strengths (FRHSC in Fig. 10d), there
is usually an abrupt and sudden drop in the load-carrying where m and n are coefficients that can be obtained
capacity after the occurrence of peak load. However, in from regression analysis. The n value is generally
an efficient fibre-reinforced concrete (FROC and between 1/2 and 3/4.
FRSCC in Fig. 10b, c), the sudden release of energy The ultimate splitting tensile strength does not vary
will be offset by an effective bridging of fibres. in the same proportion with the compressive strength.
2722 Materials and Structures (2015) 48:2713–2726

Step FROC FRSCC FRHSC

1
first cracking

fsp= 2.26 MPa, δ = 19 μm fsp= 4.14 MPa, δ = 15 μm fsp= 4.29 MPa, δ = 16 μm

2
Propagation of
the main crack
to the feed belts

fsp= 2.34 MPa, δ = 36 μm fsp= 4.25 MPa, δ = 31 μm fsp= 4.47 MPa, δ = 30 μm

3
Propagation of
the main crack
to the feed belts

fsp= 2.39 MPa, δ = 86 μm fsp= 4.49 MPa, δ = 65 μm fsp= 4.61 MPa, δ = 55 μm

4
Appearance of
related oblique
cracks

fsp= 2.82 MPa, δ = 351 μm fsp= 5.66 MPa, δ = 300 μm fsp= 5.84 MPa, δ= 361 μm

5
Asymmetric
related cracks

fsp= 2.90 MPa, δ = 441 μm fsp= 5.82 MPa, δ = 462 μm fsp= 6.00 MPa, δ = 460 μm
Fig. 9 Morphology of cracking apparent on the face of a specimen of FRC

In effect, as the compressive strength fc increases from shown in Fig. 11. According to the obtained results in
ordinary concrete to high strength concrete, the tensile this study, the ratio (fsp/fc) varies from 1/15 to 1/17 for
one fsp also increases but at a very decreasing rate, as HSC and from 1/10 to 1/12 for FRHSC.
Materials and Structures (2015) 48:2713–2726 2723

Splitting tensile strength (MPa)

Splitting tensile strength (MPa)
6 6
FROC 35-0.5 (a) FROC
5 5

4 4

3 FROC 60-1
3

FROC 35-1
2 2
FROC 60-0.5
1 1 FROC 35-0.5

0 0
0 0.25 0.5 0.75 1 1.25 1.5 1.75 2 2.25 2.5 0 0.25 0.5 0.75 1 1.25 1.5 1.75 2 2.25 2.5
Latéral displacement (mm) Transversal displacement (mm)
(a) strength-displacement typical curve (b) FROC
Splitting tensile strength (MPa)

Splitting tensile strength (MPa)

8 8
(b) FRSCC (c) FRHSC
7 7
6 6 FRHSC 60-1
FRHSC 35-1
5 5
4 4
FRSCC 60-1
3 FRSCC 35-1 3
FRSCC 60-0.5
2
2 FRHSC 35-0.5
1 FRSCC 35-0.5 FRHSC 60-0.5
1
0
0 0.25 0.5 0.75 1 1.25 1.5 1.75 0
0 0.25 0.5 0.75 1 1.25 1.5 1.75 2 2.25 2.5 2.75 3
Transversal displacement (mm) Transversal displacement (mm)
(c) FRSCC (d) FRHSC

Fig. 10 Splitting tensile stress-transversal displacement curves of concretes tested

Regression analysis was carried out on these exper- The IAE measures the relative deviations of data
imental data concerning the splitting tensile strength and from the regression equation. When the IAE is zero,
the compressive strength. A correlation between the two the predicted values from the regression equation are
mechanical properties for plain concretes is proposed equal to the observed values; this situation rarely
bellow. The coefficient of determination (R2) of this occurs. IAE values of different empirical relations are
proposed relation is 0.97, suggesting a strong correlation presented in Table 5. When comparing the different
between these two mechanical properties. equations proposed, the regression equation giving the
smallest value of the IAE can be judged as the most
fsp ¼ 0:33 ðfc Þ0:61 ðMPaÞ ð3Þ reliable. A range of the IAE from 0 to 10 % may be
Previously published empirical relations between considered as the limits for an acceptable regression
splitting tensile strength and compressive strength are equation [34].
presented in Table 5. In this study, the reliability of the It can be seen that the IAE values of the empirical
relationships derived from the regression analysis was relations proposed by Akazawa, Oluokun and Carnei-
assessed on the basis of the integral absolute error IAE. ro are within 10 % and were in a close agreement,
This index has been used by others [6, 34] to assess the translating a better accuracy for predicting the splitting
precision of proposed relationships; it is computed tensile strength.
from Eq. (4) In the literature, for fibre-reinforced concrete, there
X j f i  Pi j is still a lack of work concerning the fibre contribution
IAE ¼ P  100 ð4Þ to the tensile strength of concrete. In this sense, the
fi
work of Ashour and Wafa [40] is among the very rare
where fi is the measured value, and Pi is the predicted published in the literature; he has reported a relation-
value from the regression equation. ship for the ultimate splitting tensile strength as a
2724 Materials and Structures (2015) 48:2713–2726

Splitting tensile strength (MPa)

7 Table 5 Results of several empirical relations used in
6
y = 0.33x0.61 regression analyzes
R2 = 0.97
5 Sources Empirical relations IAE (%)

4 CEB-FIB [12] ftf ¼ 0:3ðfc Þ2=3 14.79

3 Arioglu [34] ftf ¼ 0:231ðfc Þ0:661 20.15
ACI 318-99 [35] 0:5 12.76
2 ftf ¼ 0:56ðfc Þ
Carino [36] 0:71 23.18
1 ftf ¼ 0:272ðfc Þ
Ahmad [37] 0:55 12.67
0 ftf ¼ 0:462ðfc Þ
0 20 40 60 80 100 0:667
Rapahel [38] ftf ¼ 0:313ðfc Þ 20.16
Compressive strength (MPa)
Akazawa [39] 0:73 5.80
ftf ¼ 0:209ðfc Þ
Fig. 11 Correlation between the splitting and the compressive Oluokun [6] 0:7 8.54
ftf ¼ 0:21ðfc Þ
strengths for plain concretes
Carneiro [8] 0:735 7.65
ftf ¼ 0:185ðfc Þ

function of its fibre volume fraction for an aspect ratio

of fibre of 75 as follows: values of displacement, calculated as the ratio of the
fspf ¼ fsp þ 3:02Vf ðMPaÞ ð5Þ area generated by the load–displacement curve
corresponding to a displacement of 3, 5.5, 10.5, and
In the present study, the splitting tensile strength 15.5 times the displacement corresponding to the first
was found to be linearly dependent on the volume crack. The witness specimens did not contain fibres
fraction of fibres Vf. With the help of a regression and hence failed immediately after first cracking.
analysis, the ultimate splitting tensile strength of a Consequently, the ductility index for these concretes
high strength concrete containing fibres (fspf) can be is equal to 1. The ductility indexes are shown in
expressed as a function of its splitting tensile strength Table 6.
of the plain concrete (fsp) and of its fibre volume Ductility indexes I5 and I10 do not accurately
fraction Vf (%), as in the expression below : describe the behaviour in splitting strength in the post-
fspf ¼ fsp þ 3:35Vf ¼ 0:33ðfc Þ0:61 þ 3:35Vf ðMPaÞ ð6Þ peak branch; they are often located in the instable
zone. It is better to consider higher ductility indexes
2
The corresponding R of this proposed relation is corresponding to the higher deflections, such as I20 and
0.90. I30, for a complete description of the structural
splitting behaviour. In general, the addition of fibres
3.7 Ductility of concretes resulted in an improvement of the ductility for all
cases. The specimens with higher fibre volume
For the behaviour in splitting tension, the issue when fraction and the larger aspect ratio were more ductile.
assessing the usefulness of fibres in concrete is not The ductility of the three types of concrete was
only the increase in the final splitting strength, but also significantly improved by the presence of fibres.
the ductile behaviour which can make the use of this Moreover, the ductility was in general higher for
concrete material worthwhile and structurally inter- concrete with good workability (FRSCC and FROC)
esting. A cracked concrete continues to support further as shown in Table 6. This shows the significant
increases in loading without worsening the crack effect of fibre distribution in the concrete matrix. It
width through the ability of the fibres to restrain cracks was concluded that the dispersion and alignment of
and through their deformations. At the final stage, the fibres have effects on the mechanical properties
failure is very ductile and soft since most of the energy of fluid concretes (i.e. FRSCC) [31, 42]. The results
is absorbed by the deformed fibres. show that even when higher volume of fibres are
To compare the energy absorbed by FRC, their used in the self-compacting concrete, due to its
ductility indexes have been calculated according to better workability, they can be dispersed homoge-
ASTM C1018 [41]. The ductility indexes I5, I10, I20 neously without clumping, resulting in an enhanced
and I30 represent the energy absorbed for different toughness of concrete.
Materials and Structures (2015) 48:2713–2726 2725

Table 6 Ductility indexes of fibre reinforced concretes tested

Specimens designation FROC FRSCC FRHSC
I5 I10 I20 I30 I5 I10 I20 I30 I5 I10 I20 I30

35-0.5 5.2 9.9 16.6 19.0 4.6 7.5 12.6 23.6 4.2 6.4 10.4 12.3
35-1 6.0 11.4 17.9 24.0 5.7 11.4 19.5 28.0 5.6 9.3 14.5 17.0
60-0.5 4.9 8.5 15.7 21.0 5.2 9.4 17.2 25.4 4.8 7.7 11.5 12.9
60-1 5.0 9.2 16.3 25.0 6.1 13.4 22.3 30.8 5.1 9.5 13.6 18.0

4 Conclusions volume of 0.5 % fibre and from 40 to 60 % for a

dosage of 1 %.
From the test results, the following conclusions can be The ductility is the main property which compen-
drawn: sates for the brittle behaviour of concrete. The results
This paper reports on the benefits of using digital showed that the post peak behaviour was considerably
image correlation based non-contact measurement improved, exhibiting a marked softening of the mate-
approach. The most important added benefit of the rial. This behaviour is particularly useful in seismic
technique was found to be the ability to clearly regions where in case of failures, the energy is not
identify the fracture mechanism in the splitting. It was liberated instantaneously but gradually in a soft manner.
observed that the DIC technique could provide very
accurate and detailed information.
References
The fracture mechanism in the splitting test for
SFRC is associated with related cracks; appearing 1. Swamy RN (1987) High-strength concrete-material
after the main central crack as a consequence of the properties and structural behavior. ACI Maert J SP-
redistribution of stresses at the edges of specimen. The 87:119–146
fibres has allowed for such stress redistribution. 2. Hsu TC, Mau ST, Chen B (1987) Theory of shear transfer
strength of reinforced concrete. ACI Struct J 84(2):149–160
Increasing the concrete strength does increase the 3. Ramesh K, Seshu DR, Prabhakar M (2003) Constitutive
splitting tensile strength but not in the same proportion behavior of confined fibre reinforced concrete under axial
as the compressive strength. The use of steel fibres in compression. Cem Concr Compos 25(3):343–350
concrete slightly improved the splitting tensile 4. Neville AM (1995) Properties of concrete, 4th edn. Long-
man Group Ltd, Essex, p 844
strength, particularly in high strength concrete. Ratios 5. Hannant DJ, Buckley KJ, Croft J (1973) The effect of
of splitting tensile strengths to compressive strengths aggregate size on the use of the cylinder splitting test as a
varied from 1/15 to 1/17 when no fibres were used; measure of tensile strength. Mater Struct 6(31):15–21
with the presence of fibres, these ratios were increased 6. Oluokun FA, Burdette EG, Deatherage JH (1991) Splitting
tensile strength and compressive strength relationships at
to range from 1/10 to 1/12 for high strength concrete. early ages. ACI Mater J 88(2):115–121
However, it is common practice to relate the tensile 7. Kadlecek V, Modry S, Kadlecek VJ (2002) Size effect of
strength with the compressive one, the latter being the test specimens on tensile splitting strength of concrete:
most important property of concrete and relatively general relation. Mater Struct 35(1):28–34
8. Carneiro FL, Barcellos A (1953) Tensile strength of con-
easy to determine experimentally, particularly with cretes. RILEM Bull (Paris) 13:97–123
the development of the non destructive testing applied 9. ASTM C496-90 (1991) Standard test method for splitting
to real structures. In this sense, a formula is proposed tensile strength of cylindrical concrete specimens, Annual
to relate the splitting tensile strength and the com- Book ASTM Standards 4(04.02):266–269
10. BS 1881 Part 117 (1983) Testing concrete part 117: Method
pressive strength of steel fibre-reinforced concrete. for determination of tensile splitting strength. British Stan-
The fibres were more effective in high-strength dards Institution
matrix, increasing both the ultimate load and the 11. ISO 4108-1980 (E) (1980). Concrete determination of ten-
ductility, due to the good adhesion between the fibres sile splitting strength of test specimen. International Orga-
nization for Standardization
and the matrix of concrete. The increase in the tensile 12. CEB-FIP Model Code for Concrete Structures (1990)
splitting strength varies from 15 to 40 % for a dosage Evaluation of the Time Dependent Behaviour of Concrete.
2726 Materials and Structures (2015) 48:2713–2726

Bulletin d’Information n° 199, Comite European du Béton/ 28. Boulekbache B, Hamrat M, Chemrouk M, Amziane S
Fédération Internationale de la Precontrainte, Lausanne (2010) Flowability of fibre-reinforced concrete and its effect
13. Rocco C, Guinea G, Planas J, Elices M (1999) Size effect on the mechanical properties of the material. Constr Build
and boundary conditions in the Brazilian test: experimental Mater 24(9):1664–1671
verification. Mater Struct 32(3):210–217 29. Boulekbache B, Hamrat M, Chemrouk M, Amziane S
14. Gencturk B, Hossain K, Kapadia A, Labib E, Mo YL (2014) (2009) Influence de la rhéologie des bétons renforcés de
Use of digital image correlation technique in full-scale fibres métalliques sur leurs propriétés mécaniques. Eur J
testing of prestressed concrete structures. Measurement Env Civil Eng 13(4):473–488
47:505–515 30. Rossi P (1994) Steel fibre reinforced concretes (SFRC): an
15. Shokouhi P, Zoëga A, Wiggenhauser H, Fischer G (2012) example of French research. ACI Mater J 91(3):273–279
Surface wave velocity–stress relationship in uniaxially 31. Boulekbache B, Hamrat M, Chemrouk M, Amziane S
loaded concrete. ACI Mater J 109(2):141–148 (2012) Influence of yield stress and compressive strength on
16. Skocek J, Stang H (2010) Application of optical deforma- direct shear behaviour of steel fibre-reinforced concrete.
tion analysis system on wedge splitting test and its inverse Constr Build Mater 27(1):6–14
analysis. Mater Struct 43:63–72 32. Zain MFM, Mahmud HB, Ilham A, Faizal M (2002) Pre-
17. Pereira EB, Fischer G, Barros JAO (2012) Direct assess- diction of splitting tensile of high-performance concrete.
ment of tensile stress-crack opening behavior of strain Cem Concr Res 32(8):1251–1258
hardening cementitious composites (SHCC). Cem Concr 33. Xu BW, Shi HS (2009) Correlations among mechanical
Res 42(6):834–846 properties of steel fibre reinforced concrete. Constr Build
18. Hamrat M, Boulekbache B, Chemrouk M, Amziane S Mater 23(12):3468–3474
(2010) Shear behaviour of RC beams without stirrups made 34. Arioglu N, Girgin ZC, Arioglu E (2006) Evaluation of ratio
of normal strength and high strength concretes. Adv Struct between splitting tensile strength and compressive strength
Eng 13(1):29–41 for concrete up to 120 MPa and its application in strength
19. Hamrat M, Boulekbache B, Chemrouk M, Amziane S criterion. ACI Struct J 103(1):18–24
(2012) Effects of the transverse reinforcement on the shear 35. ACI Committee 318 (1999). Building code requirements for
behaviour of high strength concrete beams. Adv Struct Eng structural concrete (ACI 318-99) and commentary (318R-
15(8):1291–1306 99). Farmington Hills, MI
20. Michel A, Solgaard AOS, Pease BJ, Geiker MR, Stang H, 36. Carino NJ, Lew HS (1972) Re-examination of the relation
Olesen JF (2013) Experimental investigation of the relation between splitting tensile strength and compressive strength
between damage at the concrete-steel interface and initia- of normal weight concrete. ACI J Proc 79(3):136–147
tion of reinforcement corrosion in plain and fibre reinforced 37. Ahmad SH, Shah SP (1985) Structural properties of high
concrete. Corros Sci 77:308–321 strength concrete and its applications for precast prestressed
21. Attari N, Amziane S, Chemrouk M (2012) Flexural concrete. PCI J 30(6):97–123
strengthening of concrete beams using CFRP, GFRP and 38. Raphael JM (1984) Tensile strength of concrete. ACI J Proc
hybrid FRP sheets. Constr Build Mater 37:746–757 81(2):158–165
22. Timoshenko SP, Goodier JN (1991) Theory of elasticity. 39. Akazawa T (1953) Tension test methods for concretes. Ri-
McGraw-Hill, New York lem Bull (Paris) 16:11–23
23. Johnston CD (1982) Steel fibre reinforced and plain con- 40. Ashour SA, Wafa FF (1992) Mechanical properties of high-
crete: factors influencing flexural strength measurement. strength fibre reinforced concrete. ACI Struct J 89(5):449–455
ACI Mater J 79(2):131–138 41. ASTM C 1018 (1997) Standard test method for flexural
24. Banthia N, Trottier JF (1995) Concrete reinforced with toughness and first-crack strength of fibre reinforced con-
deformed steel fibres part II: toughness characterization. crete (using beam with third-point loading. ASTM Inter-
ACI Mater J 92(2):146–154 national, West Conshohocken
25. GOM Aramis user manual v6 (2007) GOM mbH, Germany, 42. Akcay B, Tasdemir MA (2012) Mechanical behaviour and
(http://www.gom.com). Accessed 7 May 2014 fibre dispersion of hybrid steel fibre reinforced self-com-
26. Corinaldesi V, Moriconi G (2004) Durable fibre reinforced pacting concrete. Constr Build Mater 28:287–293
self-compacting concrete. Cem Concr Res 34(2):249–254
27. Ziad Bayasi M, Soroushian P (1992) Effect of steel fibre
reinforcement on fresh mix properties of concrete. ACI
Mater J 89(4):369–374