You are on page 1of 15

Construction and Building Materials 106 (2016) 678–692

Contents lists available at ScienceDirect

Construction and Building Materials

journal homepage:

Flexural cracking behavior of normal strength, high strength and high

strength fiber concrete beams, using Digital Image Correlation technique
M. Hamrat a, B. Boulekbache a, M. Chemrouk b, S. Amziane c,⇑
Civil Engineering Department, University Hassiba Benbouali, Chlef, Algeria
University of Sciences and Technology Houari Boumediene, Algiers, Algeria
Institut Pascal, Polytech’Clermont-Ferrand, Clermont University, 63174 Aubière, France

h i g h l i g h t s

 Beam flexural behaviour of several concretes with and without fibers is investigated.
 Digital Image Correlation (DIC) technique is used to assess crack and strain variation.
 The DIC technique allows a high accuracy for the first crack detection.
 The addition of steel fibers increases the load at first crack.
 The flexural failure mode changes following the fibers content.

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

Article history: The paper presents the results of an experimental work on the flexural behavior of three types of con-
Received 3 July 2015 crete: normal strength concrete (NSC), high strength concrete (HSC) and high strength fiber concrete
Received in revised form 4 December 2015 (HSFC) in terms of crack detection, crack development, crack width measurements and strain compo-
Accepted 22 December 2015
nents, using the Digital Image Correlation (DIC) technique. The experimental results of the present work
Available online 2 January 2016
and from others published in the literature were used to assess the accuracy of the major codes provi-
sions (Eurocode 2, ACI 318 and the BS8110 code) for the crack width predictions.
The comparison between the classical measurement techniques and the DIC technique suggests that
High strength concrete
Fiber reinforced concrete
both of them are suitable for the analysis of the strain components (deformations) of structural members.
Digital Image Correlation The most important additional benefit of the DIC technique is that it allows the easy detection of the first
Crack width crack with a high precision, measures the crack opening and follows the progressive cracking process
Serviceability until failure of reinforced concrete members. Furthermore, the experimental results show that the addi-
Ductility tion of steel fibers increases the load at first crack and amplifies the number of cracks which leads to a
Brittleness remarkable decrease in both the crack spacing and crack width as well as an improvement of the ductil-
Toughness ity. The flexural failure of the beam specimens also changes from a brittle concrete crushing at the com-
pression zone to a ductile and smooth concrete compression failure.
Moreover, the present study shows that, at service load, the code’s provision on crack width are con-
servative for beams containing steel fibers since predicting excessively higher values than the experimen-
tal ones; the overestimation exceeded 300% in some cases and none of the present major design codes
takes into consideration the positive restraining effect of the fibers on the crack widening. The Rilem
technical document appears to be the only one which takes account the positive effect of steel fibers
on cracking, though still not accurate enough in predicting the crack width of fiber reinforced concrete.
Based on the present experimental work and on data from the literature, a modification of the Rilem
mathematical model is proposed with the aim of improving the crack width prediction of this challenging
concrete material.
Ó 2015 Elsevier Ltd. All rights reserved.

⇑ Corresponding author.
E-mail address: (S. Amziane).
0950-0618/Ó 2015 Elsevier Ltd. All rights reserved.
M. Hamrat et al. / Construction and Building Materials 106 (2016) 678–692 679

1. Introduction problems [9–15] owing to its numerous advantages such as non-

contact, full-field measurements, its simplicity in use and the con-
In recent years, the design of reinforced concrete members for tinuous measurements up to failure. By comparison, in the case of
serviceability has become increasingly important as the use of reinforced concrete structures, classic measurement techniques,
new materials has led to a decrease in both the member section such as strain gauges extensometers and linear variable differential
and the reinforcement ratio and consequently has resulted in transformer sensors (LVDTs), hardly allow for precise estimations
higher service stresses. In this sense, High Strength Concrete of strain fields or for an early crack detection. The strain gauges
(HSC) is emerging as a challenging concrete material; its use in may themselves break and get destroyed locally at critical sections
the construction industry is ever increasing in many regions of close to failure. In addition, strain gauges only measure strains at
the world. This is due to its improved physico-mechanical proper- the fixing points and in the direction of the gauges alignment,
ties such as the mechanical strength, the stiffness and the pro- and therefore they do not provide a full-field analysis.
jected long term durability. Economically, great benefits can be This paper presents an experimental investigation of nine rein-
achieved with the reductions in the geometric sections and the forced concrete beams, in order to examine the flexural behavior in
consequent gain in the architectural space to be exploited. How- terms of crack detection and development and to measure the
ever, with the increase in compressive strength of the material, crack widths using displacement fields derived from digital images
the brittleness of concrete can also be increased. Steel fibers may captured during the loading. The measured crack widths in rein-
reduce such brittleness, by rendering the material relatively more forced concrete flexural members with and without steel fibers
ductile, and improve its toughness and stress distribution. There- was compared with predictions from the major code models,
fore, High Strength Fiber Concrete (HSFC) appears to be a new namely Eurocode 2 [6], the ACI-318-95 [7] and the BS-8110 [8].
challenge in the recent development of concrete materials with a Finally, the suitability of the prediction models for fiber reinforced
high tensile strength and other improved physico-mechanical concrete is carefully analyzed.
properties [1,2] by comparison to ordinary concrete. Therefore,
due to the enhanced mechanical properties and improved long
term durability, HSFC and HSC are progressively gaining wider 2. Experimental program
acceptance in the construction of all types of buildings, bridges
and marine structures [3–5]. In the testing program, three concrete mixes were manufac-
The aim of this work, it was thought important to determine to tured: a normal strength concrete, (NSC), a high strength concrete
what extent the structural behavior of HSFC differs from that of (HSC) and a high strength fiber reinforced concrete (HSFC); the mix
HSC and NSC. In this context, a designer must satisfy not only portions are given in Table 1.
the strength requirements needed in these concretes (NSC, HSC The cement used in this study was a CEM I 52.5N, meeting the
and HSFC) but also the serviceability requirements, and therefore requirements of European Standards EN 197-1, having a fineness
the control of cracking and crack opening is becoming more impor- expressed through the specific surface of 3520 cm2/g and a density
tant. Indeed, one of the serviceability criteria is to ensure that the of 3160 kg/m3. Standard silica fume and limestone filler was used
crack width does not exceed a prescribed limit and hence needing as mineral additive; their specific surface was 23 m2/g and
an accurate prediction of the cracking of reinforced concrete struc- 3970 cm2/g respectively. The sand and aggregate had minimum/
tures under service loads. maximum sizes of 0/4 mm and 4/10 mm respectively. The high-
Cracking plays an important role for the structural behavior and range water-reducing admixture (Chrysofluid) was a polycarboxylate-
for the durability of concrete structures. Crack width should be based admixture used to achieve an adequate fiber dispersion and
limited to an extent that will not impair the proper function of a workability.
structure or cause its appearance to be unacceptable or even affect Nine reinforced concrete beams were tested under two point-
its long term durability [6]. In this sense, it is wise stating that the loads in this study. The beams were divided into three series:
cracking behavior of a reinforced concrete member is a complex
process which depends on a wide range of factors such as the ten-  The first series (N) of beams were made with normal strength
sile strength of concrete, the concrete cover to the reinforcement, concrete (NSC) (f c ¼ 44 MPa)
the area, the diameter and the spacing of the reinforcement and  The second series (F) of beams were made with high strength
the stress in the tension zone. fiber concrete (HSFC) (f c ¼ 78 MPa)
For the prediction of the crack width of reinforced concrete  The third series (H) of beams were made with high strength
members, equation models are generally obtained by multiplying 0
concrete (HSC) (f c ¼ 85 MPa)
the maximum crack spacing by the mean strain of the flexural steel
reinforcement. The design models in the major codes in use All the beams were singly reinforced. Shear reinforcements
throughout the world such as the Eurocode 2-04 [6], the ACI- were provided along the beam length except in the constant
318-95 [7] and the BS-8110 [8] are generally applicable to rein-
forced concrete members without steel fibers. The applications of
these models to steel fiber reinforced concrete members have not Table 1
yet been sufficiently reported in the literature and very little exper- Mix proportions of concretes.

imental data exists on this aspect. It is thus important to provide Material Units NSC HSC HSFC
qualitative and quantitative information on cracking and crack Cement type CEMI-52.5N kg/m3 275 425 425
opening of fiber reinforced high strength concrete and assess the Silica fume kg/m3 – 42.50 42.50
applicability of the existing models on crack width predictions of Filler (limestone) kg/m3 44 – –
concrete in general and high strength concrete containing fibers Sand (0–4 mm) kg/m3 740 700 720
Aggregate (4–10 mm) kg/m3 990 980
in particular. Water L 170 144 150
In the aim of monitoring continuously the strain developments, Superplasticiser (Chrysofluid) L – 6.40 6.40
the cracking process and the crack openings, the Digital Image W/C – 0.62 0.34 0.35
Correlation technique is used. This technique has been applied suc- Steel fibers kg/m3 – – 39
Slump cm 17 15 13
cessfully to various mechanical and civil engineering structural
680 M. Hamrat et al. / Construction and Building Materials 106 (2016) 678–692

bending moment zone. The beams were 1.50 m long with which is about 160  400 mm2. DIC is an optical, non-contacting
10 cm  16 cm cross section. They were simply supported and sub- measuring technique, enabling to determine the displace-
jected to a symmetric loading composed of four points loading. ment and deformation field of a specimen’s surface under any type
Three flexural reinforcement ratios of 1.22% (2£10), 1.77% of loading condition. The positioning and components of the
(2£12) and 2.42% (2£14) were used for each series of beams. At DIC-based non-contact measurement system are shown in Fig. 3.
the top of the beams, two 6 mm diameter bars were placed to hold Different sections or Aramis gauges on the specimen surface are
the reinforcing cage. Steel fibers with hooked ends were added to selected to measure their axial displacements (strain components).
the high strength concrete in a proportion of 0.5% as a volume frac- The selected sections are distributed regularly along the axial
tion of concrete in series F beams. The fiber length (lf) was 40 mm direction and throughout the height of the beams, which are
and their diameter (df) was 0.5 mm, giving an aspect ratio (lf/df) of clearly marked in Fig. 2(b).
80, specific gravity of 7.80 and tensile strength of 1100 MPa. In order to validate the proposed procedure, the displacement
Table 2 presents the detailed testing program. Each beam is des- measured by DIC was used to determine the ‘‘mean” strain at the
ignated to indicate the concrete compressive strength and the ten- surface of concrete of the beam. This strain was then compared
sion reinforcement diameter. Thus B-H10 represents a beam with to the one obtained with the more classical method of measure-
high strength concrete and the number 10 indicates the tension ments, that is using strain gauges. In the case of the LVDT compara-
reinforcement diameter. Fig. 1 illustrates the reinforcement details tor, the average strain was obtained by dividing the measured
for all the beam specimens. vertical displacement by the specimen length. In the case of DIC,
All the test beams were equipped with a linear variable dis- it was achieved by averaging the strain field in each direction.
placement transducer (LVDT) for measuring vertical deflection at
mid-span and strain gauges to record deformation measurements 3. Test results and discussions
at the concrete surface (Fig. 2(a)) and at the surface of the
embedded steel reinforcement at mid-span (Fig. 2(a)). In addition, 3.1. Load–strain behavior
a Digital Image Correlation technique obtained by a digital record-
ing camera ‘‘Aramis” by GOM ( [16] (Fig. 3) The curves of load versus compressive strain at the top surface
was used to measure the deflections, monitor the development of the beams at mid span and load versus tensile strain on the lon-
of the cracks and measure their widths and detect strain compo- gitudinal reinforcement at the tension zone at mid-span section
nents. All the data resulting from the tests were recorded automat- are illustrated in Fig. 4 for the three series of beams N, F and H.
ically using data logger. These results are used to compare and Fig. 4 presents also a comparison between the strains measure-
calibrate the DIC technique. ments obtained by electrical resistance strain gauges and the
The DIC-based non-contact measurement system was placed on strains calculated using the DIC technique.
the side of the beam. A uniform/random speckle pattern is applied It is important to note that for beams B-H10, B-H12, B-F10, B-
on the surface of the beam as shown in Fig. 3. With system calibra- F12, the ultimate strain in the steel is higher than the conventional
tion, a field of view of DIC measurement system can be obtained, strain of 1% considered by the standard codes. For example, for
beam B-H10, a strain of 2% has been recorded prior to a tension
failure (Fig. 4). This is thought to be mainly due to the better qual-
Table 2 ity of the bond between concrete and the reinforcing steel,
Geometrical and mechanical properties of the test beams. enabling a better transmission of internal stresses and strains from
Series Beam 0
f t (MPa) As (mm2) q (%) concrete to the reinforcement. It is also important to note that for
f c (MPa)
B-N14, B-F14 and B-H14, the load–strain behavior shows a strain
N B-N10 44 3.37 157 1.22
hardening response. This allows for an increase in the load sup-
B-N12 226 1.77
B-N14 308 2.42 ported by the beam. The use of fibers increases the top concrete
strain at failure load, e.g. the concrete strain at the compression
F B-F10 78 4.85 157 1.22
B-F12 226 1.77 face varied from 2.42‰ to 3.86‰ for series B-F beams (high
B-F14 308 2.42 strength concrete containing steel fibers), from 2.30‰ to 3.24‰
H B-H10 85 4.20 157 1.22 for series B-H beams (high strength concrete) and from 2.36‰ to
B-H12 226 1.77 3.20‰ for series B-N beams (normal strength concrete). A close
B-H14 308 2.42 examination of Fig. 4 shows that the tensile strains measured by
electrical strain gauges and those measured with the DIC technique

P/2 P/2 2Ø6

Ø6 @11cm Zone without Ø6 @11cm


10 47.5 35 47.5 10

Dimensions in c
2∅6 2∅6

2Ø12 2Ø14

Fig. 1. Geometry and detailing of test beams.

M. Hamrat et al. / Construction and Building Materials 106 (2016) 678–692 681

Fig. 2. Strain measurement locations. (a) Electrical resistance strain gauges at the concrete surface and at the surface of the reinforcement. (b) DIC technique.

Fig. 3. Positioning and components of the DIC technique.

are very similar and reveals that the new technique could be a reli- experimental results have been obtained between the two mea-
able measurement method free from any testing risk as are often surement techniques as clearly illustrated in Fig. 4, validating the
the strain gauges and the traditional crack measurement methods. DIC measuring method. However, the risk of damage of any mea-
The same Fig. 4 also shows that the compressive strains for the suring gauge, particularly when crossed by a crack, would make
top compression face, measured by electrical strain gauges and the traditional technique less efficient (about 2%). Moreover, once
those measured with the DIC technique are very close; this once a crack appears, the strain measurements of such traditional tech-
again reveals that the new technique could be a powerful measure- nique become less reliable. On the contrary, the DIC measuring
ment method since it is free from any testing risk and errors and technique gives strain fields up to the point of failure of the test
not requiring labor by comparison to the traditional measurement specimen and over a large area of it.
methods. The slight differences between the two measurement
techniques as noted in some of the beams in Fig. 4 are thought
to be mainly due to experimental errors as often occurs in any 3.2. Crack patterns and crack development
experimental work. To sum up, it can be deduced from the present
experimental work that the two techniques have proved effec- The Digital Image Correlation technique is used to investigate
tive to measure strain components. Tight similarities in the the cracking process and measure the crack width during the test.
682 M. Hamrat et al. / Construction and Building Materials 106 (2016) 678–692

100 affect the cracking process, as shown in Fig. 5(a and b), but a close
Serie N : fc = 44MPa B-N14-Aramis-Tens
examination of the results in Table 3 shows that the action of fibers
is rather expressed on the length and the number of cracks which
become shorter and numerous respectively. Beams of series B-F
(with steel fibers) present more cracks compared to those of series
Load (kN)

B-N and series B-H beams, both without fibers. In this sense, beam

B-F12 containing fibers presents 16 micro-cracks (Table 3) com-
pared to 10 cracks for beam B-N12 made of normal concrete and
to 14 for beam B-H12 made of high strength concrete. It was also
observed that he number of cracks increases as the strength of con-
crete increases, since going from 10 cracks for B-N12 beam to 14
Reinforcement B-N10-Gaug-Comp
cracks for B-H12 beam. This is thought to be due to the higher
0 quality bond between the reinforcement and concrete in the case
-5 0 5 10 15 20 25
of high strength concrete, leading to a better transmission of the
Strain (‰) internal stresses from concrete to steel and a more uniform distri-
bution of theses stresses.
Serie F: fc =78 MPa In addition, the lengths of cracks are slightly lower for beams
containing fibers compared to those in series B-H and series B-N
80 beams. Consequently, all the beams of series B-F had an increased
neutral axis depth (i.e. increased depth of the compression zone) at
failure compared to that in beams without steel fibers (B-N and B-
60 B-F12-Aramis-Tens
Load (kN)

H). Generally, the crack length decreases as the main longitudinal

steel ratio increases for the three types of concrete as in Table 3.
This is due to the fact that with the increase in the tension steel
ratio (q), the beams becomes less ductile and hence keeping the

B-F10- Aramis-Tens
tension zone lower; the length of a crack being in general propor-
B-F10- Gaug-Tens
20 tional to the depth of the tension zone.
Fig. 5(a) shows also a description of the crack pattern (horizon-
Concrete Reinforcement tal displacements) obtained from DIC technique and the visual
crack pattern obtained during the test. It can be clearly seen that
-5 0 5 10 15 20 25
the horizontal displacement profile of the DIC qualitatively agrees
Strain (‰)
with the crack pattern observed on the specimen.
100 B-H14-Aramis-Tens
It can be seen that all the beams failed in flexure through the
Serie H: fc =85 MPa crushing of the concrete in compression at the top face where

the bending moment is maximum (Table 3). However, the flexural
80 failure through the concrete crushing occurred after a marked ten-
sion behavior for beams having 1.22% and 1.77% of steel reinforce-
ment where the flexural crack opened very wide and extended up,
Load (kN)

60 B-H12-Gaug-Tens
reducing the compression zone into a very small area (Fig. 5(a and
b)). With further increase in load and hence in bending moment,
40 concrete in the very limited compression zone crushes. This is a
B-H10-Aramis-Tens typical tension failure of an under-reinforced beam. On the con-
B-H-10-Gaug-Tens trary, beams reinforced with 2.42% of tension steel did not exhibit
B-H10-Aramis-Comp full tension behavior and the cracks were not very wide at failure.
B-H10-gaug-Comp The neutral axis did not seem to have been pushed upward
Concrete Reinforcement
0 towards the compression zone as greatly as in beams with less ten-
-5 0 5 10 15 20 25 sion reinforcement. Such behavior could be more identified as a
Strain (‰) flexural compression failure through the crushing of concrete at
the compression zone at the maximum bending moment region
Fig. 4. Comparison of strains measurements collected from strain gauges and DIC
before exhibiting sufficient ductility. The crushing of concrete
appears to be more brittle as in beam BH-14 and BN-14 but softer
where fibers were used as in beam BF-14 (Fig. 6). The behavior of
The images captured at different levels of loading are processed in beams having 2.42% of longitudinal steel is similar to that of an
comparison with the reference image captured before loading. over-reinforced concrete beam where the tension reinforcement
is not strained at the maximum. The presence of steel fibers would
be beneficial in preventing the brittleness of the concrete in
3.2.1. Cracks patterns compression, particularly for high performance concrete (BH-14
Fig. 5(a and b) shows the developed crack patterns and the cor- in Fig. 6).
responding loads in kN. From this figure, the appearance of cracks
seems to be randomly distributed but at the same time, there 3.2.2. Cracking moment
seems to be a smaller distance between the cracks for beams with The cause of cracking of any concrete member is the weaker
steel fibers. A great difference, however, is noticed in the trajecto- tensile strength of concrete. From the experimental results shown
ries of the cracks which are straighter in beams made of HSC (B-H) in Fig. 5(a and b) and Table 3, the first crack always appeared at the
[10]. On the other hand, the trajectory of the cracks of the beams extreme tension fiber at the maximum bending moment region at
with steel fibers (B-F) seems to be disturbed by the presence of different loads for the beams, depending on the type of concrete
fibers (Fig. 5(a and b)). The insertion of the steel fibers does not making the test specimen. The first cracking load varied from 5%
M. Hamrat et al. / Construction and Building Materials 106 (2016) 678–692 683

B-N10 B-H10 B-F10

05 kN 07 kN 09 kN

18 kN 15 kN 13 kN

26 kN 20 kN 16 kN

40 kN 35 kN 31 kN

44 kN 47 kN 50 kN

Fig. 5(a). Morphology of cracking apparent on the face of the tested beams at different loads. (B-N10, B-H10 and B-F10).

to 17% of the ultimate load for beams made of NSC, from 7% to 25% [17], for the nine beams tested. The predicted cracking moment
of the ultimate load for beams made of HSC and from 8% to 30% of is calculated using the following equation:
the ultimate load for beams made HSFC. Such difference in the
f r  Ig
cracking loads of the three types of concrete is due to their Mcr ¼ ð1Þ
differences in the tensile strength and, to a lesser extent, in the
way the beams are reinforced. Hence, the first flexural cracking where: Ig is the moment of inertia of the gross concrete section;
and the cracking moment depend on the modulus of rupture of before first cracking, concrete in tension is not lost and the whole
the concrete which itself expresses the tensile strength of the section contributes to the stiffness,
material. Table 4 shows the experimental cracking moments, yt is the distance of the extreme tension fiber from the neutral
Mcr,exp, as worked out from the experimental data together with axis,fr is the modulus of rupture of concrete, given by:
the cracking moments predicted from the three different codes, qffiffiffiffi
namely the Eurocode 2-04 [6], the ACI318-95 [7] and the CSA-94 f r ¼ 0:62 f c for ACIð318-95Þ½7 ð2Þ
684 M. Hamrat et al. / Construction and Building Materials 106 (2016) 678–692

Fig. 5(b). Crack patterns at failure of tested beams (B-N12, B-H12 and B-F12).

Table 3 and normally should not be applied for high strength concrete. In
First crack, length and number cracks, measured with DIC. general, the overestimation is related to the fact that the modulus
Beams First crack Length of crack1 Number of Failure of rupture used in the ACI code, and indeed in all the design codes,
(kN) (mm) cracks2 mode does not take into consideration the possible existence of invisible
B-N10 5.1 130.2 08 Bending micro-cracks in concrete induced by shrinkage even before the
B-N12 7.1 123.2 10 Bending loading starts when predicting the cracking moment; these
B-N14 6.2 111.5 07 Bending micro-cracks soon develop into visible macro-cracks on loading.
B-H10 7.0 146.7 10 Bending
B-H12 9.4 136.8 14 Bending
B-H14 8.6 117.1 09 Bending 3.2.3. Crack spacing
B-F10 9.1 124.4 13 Bending It can be seen that the increase in the number of cracks in
B-F12 11.1 118.4 16 Bending beams (series B-H and B-F of Figs. 5 and 6) reduces crack spacing.
B-F14 12.3 97.9 14 Bending
The average crack spacing values obtained in the test results are
At failure. presented in Table 6. This average crack spacing varied from
Number of cracks along the span. 40 mm to 60 mm in series B-F beams (high strength concrete
beams containing fibers), from 65 mm to 80 mm for series B-H
    beams (high strength concrete beams) and from 75 mm to
f r ¼ max f ctm ; 1:6  f ctm for Eurocode 2-04½6 ð3Þ 100 mm for series B-N beams made of ordinary concrete.
The crack spacing is an important parameter which is directly
related to the crack opening; its correct evaluation is of a major
with f ctm ¼ 2:12½1 þ ð10 Þ considered as the mean axial tensile importance for a better prediction of the crack widths. In this
strength of concrete, and h is the beam height in mm sense, the experimental average crack spacings are compared with
qffiffiffiffi those predicted by theoretical models presented in different tech-
0 nical documents such as the Eurocode 2 (EC2-92 [18] and EC2-04
f r ¼ 0:30 f c for CSA-94 ½17 ð4Þ
[6]), the Model Code 90, MC-90 [19] and Rilem TC 162-TDF [20];
Table 4 shows that the CSA-94 method greatly underestimates these different theoretical models are summarized in Table 5.
the cracking moment, with an average ratio of the measured crack- Table 6 shows that EC2-92 and MC-90 models give a good pre-
ing moment to the predicted one approaching a value of 2. This is diction of the crack spacing Srm for beams without steel fibers (ser-
mainly due to the underestimated modulus of rupture of concrete ies B-N and B-H). However, for beams containing steel fibers such
used in this design code. The ACI and, to a certain extent, the Euro- as in series B-F beams, the crack spacing Srm is seriously overesti-
code 2 predictions of the cracking moments are fairly acceptable, mated by both theoretical models since none of the two models
though slightly overestimating the cracking capacity of the three takes into consideration the efficient restraining effect of fibers
types of concrete. Of the three design codes, the American ACI on the crack widening as observed in the testing and expressed
seems to be more accurate in predicting the cracking moment of in the experimental crack spacing in Table 6. The EC2-04 crack
concrete. When considering the ACI predictions for ordinary con- spacing model, though relatively more representative, since it
crete, the predictions, though still overestimated, seem to be better takes account of more parameters, is found to overestimate the
than those for high strength concrete. This is due to the fact that crack spacing for all the tested beams and should be refined. The
the mathematical model used in the ACI to calculate modulus of Rilem TC 162-TDF model is the only one of the four models consid-
rupture of concrete is that developed for normal strength concrete ered in this work to take account of the steel fibers in restraining
M. Hamrat et al. / Construction and Building Materials 106 (2016) 678–692 685




Fig. 6. Final crack patterns for beams: B-N14, B-H14 and B-F14 (damage of compression zone).

Table 4
Experimental and theoretical moments at first cracking.

Beams Experimental values Theoretical values Ratio

Mcr,Exp (kN.m) Mcr,ACI (kN.m) Mcr,EC2 (kN.m) Mcr,CSA (kN.m) M cr;Exp M cr;Exp M cr;Exp
Mcr;ACI Mcr;EC2 M cr;CSA

B-N10 1.21 1.67 2.18 0.81 0.72 0.55 1.50

B-N12 1.69 1.67 2.18 0.81 1.01 0.77 2.08
B-N14 1.47 1.67 2.18 0.81 0.88 0.67 1.82
B-H10 1.66 2.44 3.04 1.18 0.68 0.55 1.41
B-H12 2.23 2.44 3.04 1.18 0.92 0.73 1.89
B-H14 2.04 2.44 3.04 1.18 0.84 0.67 1.73
B-F10 2.16 2.29 2.95 1.11 0.94 0.73 1.95
B-F12 2.64 2.29 2.95 1.11 1.15 0.89 2.38
B-F14 2.92 2.29 2.95 1.11 1.28 0.99 2.64
Mean 0.94 0.73 1.93
SD 0.18 0.14 0.37

Mcr,Exp: moment at first observed crack.

Table 5
Formulas for the prediction of the crack spacing as given by the different codes.
3.2.4. Crack width
The cracks were more closely spaced in all B-F beams as shown
Technical document Formula in Fig. 5(a and b) and consequently the crack widths are smaller in
Eurocode 2-92 [18] Srm ¼ 50 þ 0:25k1 k2 /=qeff these beams compared to those in series B-N and B-H beams
Eurocode 2-04 [6] Srm ¼ 3:4c þ 0:425k1 k2 /=qeff (Fig. 7). It can be noticed that the evolution of the opening of cracks
MC-90 [19] Lsm ¼ 3:6/q is similar for the three types concrete (i.e. NSC, HSFC and HSC)

Rilem TC 162-TDF [20] srm ¼ ð50 þ 0:25:K 1 :K 2 : q/ Þ: L 50

throughout the loading (Fig. 7). In the phase of a stabilized cracking
eff f f
and at a given load level, the increase in the strength of concrete
Notation: /: bar diameter; qeff: effective reinforcement ratio; c: concrete cover, k1: leads to a reduction in the crack width by up to 30%. This is mainly
0.8 for high bond bar or 1.6 for smooth bars; k2: 0.5 for bending; Lf: fiber length; df:
due to the better steel–concrete bond in the case of B-H and B-F
fiber diameter.
beams, which leads to an increase in the participation of the tensile
concrete between the cracks, known as the tension stiffening
and stitching the cracks in reinforced concrete. It takes account of effect. It is also observed that for a specified load level (service
the steel fibers explicitly through their aspect ratio (Lf/df) and was load), B-F beams exhibit smaller crack widths than those observed
found fairly accurate in predicting the crack spacing. From Table 6, for B-H and B-N beams. This can be attributed to the steel fibers
the average measured crack spacing to the predicted one is 1.16 for crossing the cracks and hence sewing them and preventing the
the three beams containing steel fibers. Provided a correct strain in crack faces from widening.
tension is used, this model could predict fairly accurately crack The crack width is an important design aspect for the service-
widths for concrete members containing steel fibers. ability as well as the durability of reinforced concrete construction.
686 M. Hamrat et al. / Construction and Building Materials 106 (2016) 678–692

Table 6
Comparison of experimental and theoretical crack spacing (Srm).

Beams Srm, Exp (mm) Srm EC2-92 (mm) Srm EC2-04 (mm) Srm MC-90 (mm) Srm Rilem (mm) Srm; Exp Srm; Exp Srm; Exp Srm; Exp
Srm;EC292 Srm;EC204 Srm;MC90 Srm;Rilem

B-N10 101.2 75.5 131.8 70.1 – 1.34 0.77 1.44 –

B-N12 90.4 70.2 122.7 55.4 – 1.29 0.74 1.63 –
B-N14 75.3 66.5 116.4 45.1 – 1.13 0.65 1.67 –
B-H10 80.6 76.2 132.9 72.2 – 1.06 0.61 1.12 –
B-H12 73.2 70.8 123.8 57.4 – 1.03 0.59 1.28 –
B-H14 66.5 67.0 117.3 46.9 – 0.99 0.57 1.42 –
B-F10 61.3 75.8 132.2 70.9 45.8 0.81 0.46 0.86 1.34
B-F12 42.6 70.5 123.3 56.5 40.5 0.60 0.35 0.75 1.05
B-F14 39.7 66.8 117.0 46.2 36.8 0.59 0.34 0.86 1.08
Mean 0.98 0.56 1.23 1.16
SD 0.25 0.14 0.32 0.13


B-F 10 B-H 10 B-N 10

Applied load (kN)





0,00 0,20 0,40 0,60 0,80 1,00 1,20 1,40
Crack width (mm)

90 90

B-F 12 B-H 12 B-N 12 75

75 B-F 14
Applied load (kN)
Applied load (kN)

60 60
B-H 14
45 45

30 B-N 14

0,00 0,20 0,40 0,60 0,80 1,00 1,20 1,40
0,00 0,20 0,40 0,60 0,80 1,00 1,20 1,40
Crack width (mm)
Crack width (mm)

Fig. 7. Applied load versus Crack width measured using DIC.

It was then thought important to record the load at which the ser- the ultimate for both B-H12 and B-H14 as compared to a load in
viceability limit of crack width is reached for the three types of the order of 65% of the ultimate for B-N12 and B-N14 both made
concrete since this is structurally very significant. Indeed, service- of normal strength concrete.
ability pathologies have become the main problems to deal with in To study the cracking pathology, a number of research works
reinforced concrete construction today. From Fig. 8 (a, b), the available in the literature have put forward theoretical empirical
0.3 mm serviceability limit of cracking was reached at a load of models for predicting the crack width of flexural members. The
20 kN, 25 kN and 33 kN respectively for B-N10, B-H10 and BF-10 models that have gained wider acceptance were adopted in the
beams; an improvement of 30% in load was recorded for high major design codes in use throughout the world as design tools
strength concrete when fibers were used. From normal concrete for crack width predictions. However, these models have in general
to high strength concrete, an improvement of 65% in load was been developed on the basis of experimental works carried on nor-
recorded in the presence of steel fibers. For beams with higher lon- mal strength concrete members. It is then important to study their
gitudinal steel reinforcement, the 0.3 mm serviceability limit of applicability for high strength concrete and for high strength con-
cracking was reached at the ultimate load when fibers were used, crete with fibers, two challenging concrete materials bound to be
stipulating that serviceability may not be a problem when using of a wider use in the coming future. The crack width measure-
high strength concrete containing steel fiber’s reinforcement. The ments from the present study together with other data available
graphs also show that the serviceability problem of cracking for in the literature are used to assess the crack width models recom-
high strength concrete may not be as acute as is in normal strength mended in the American code ACI 318-95 [7], in the Eurocode 2 [6],
concrete provided that the structural member is not too ductile, in the British Standard BS8110 [8] and in Rilem TC 162-TDF tech-
since the 0.3 mm limit was reached at a load exceeding 80% of nical document [20]. The formulas used in these different codes
M. Hamrat et al. / Construction and Building Materials 106 (2016) 678–692 687

90 90

75 75
Applied load (kN)

Applied load (kN)

60 60

45 45
B-H 10
B-N 10
30 30
Eurocode 2
Eurocode 2

BS8110 Service load BS8110

15 Service load 15

ACI318 ACI 318

0 0
0,0 0,2 0,4 0,6 0,8 1,0 1,2 1,4 0,0 0,2 0,4 0,6 0,8 1,0 1,2 1,4

Crack width (mm) Crack width (mm)

90 90

75 75
Applied load (kN)

Applied load (kN)

60 60

45 B-H 12
B-N 12
Eurocode 2
30 Eurocode 2 30

Service load BS8110

Service load 15
ACI 318 ACI 318

0 0
0,0 0,2 0,4 0,6 0,8 1,0 1,2 1,4 0,0 0,2 0,4 0,6 0,8 1,0 1,2 1,4

Crack width (mm) Crack width (mm)

90 90

75 75
Applied load (kN)

Applied load (kN)

60 B-N 14 60
B-H 14
45 Eurocode 2
Eurocode 2
30 BS8110
Service load 30
Service load
ACI 318
15 ACI318

0,0 0,2 0,4 0,6 0,8 1,0 1,2 1,4 0
0,0 0,2 0,4 0,6 0,8 1,0 1,2 1,4
Crack width (mm) Crack width (mm)

Fig. 8(a). Crack width: comparison of test results with the code’s models -beams without steel fibers-.

and technical document are summarized in Table 7, while Tables 8 unpredictable, BS8110 predictions could be considered as fairly
and 9 give the predicted values of the crack width at service loads, satisfactory (Table 8). When analyzing the results according to
taken as equal to 0.60 times the ultimate loads as suggested by the type of concrete, it can be clearly deduced that BS8110 model
Rashid and Mansur [21]. does not take into consideration the presence of fibers which effec-
A close examination of Table 8 reveals that the values of crack tively retrain the cracking and limit their width. Indeed, the results
the width predicted by the code provisions show a large scatter show that the values of the ratio WExp/WBS8110 varied from 0.20 to
among the different code formulas. In general, BS8110-97 model 0.70 for the whole data presented in Table 9 with a mean value of
gives the best prediction of the crack width with a mean value of 0.44 when fibers were used. When no fibers were used in concrete,
1.02 for the ratio of WExp/WBS8110 and a standard deviation of the ratio of WExp/WBS8110 varied from 0.7 to 1.5 with a mean value
0.19; this latter parameter expressing that the results were rela- of 1.02, expressing relatively more realistic predictions (Table 8).
tively scattered as clearly expressed by the variance coefficient of From the present experimental work and from the data taken from
19%. However, regarding the nature of cracking, hazardous and the literature, it appears that BS8110 crack width model, though
688 M. Hamrat et al. / Construction and Building Materials 106 (2016) 678–692

90 90

75 75
Applied load (kN)

Applied load (kN)

60 60 B- F 12

Modified Rilem model

45 B- F 10 45
Modified Rilem model
30 Rilem 30 Eurocode 2

Eurocode 2 Service load BS 8110

Service load
15 BS 8110 15
ACI 318
0 0
0,0 0,2 0,4 0,6 0,8 1,0 1,2 1,4 0,0 0,2 0,4 0,6 0,8 1,0 1,2 1,4
Crack width (mm) Crack width (mm)


B- F 14

60 Modified Rilem model

Applied load (kN)

Eurocode 2

BS 8110
Service load
ACI 318


0,0 0,2 0,4 0,6 0,8 1,0 1,2 1,4
Crack width (mm)

Fig. 8(b). Crack width: comparison of test results with the code’s models -beams with steel fibers.

the most realistic of the three considered models, needs more Even without the presence of fibers, the ACI model overesti-
refinement to take up the crack restraining effect of fibers. The mates the crack width of concrete members; the values of WExp/
model, originally developed on the basis of ordinary concrete, WACI vary from 0.6 to 1.40 with an average ratio of 0.84 as com-
can be used for high strength concrete, even though the better pared to 1.02 and 1.11 for BS8110 and EC2 respectively (Table 8).
quality bond that this concrete material can develop with the rein- When fibers are used to a concrete mix, the ACI model predict
forcing steel needs to be reassessed in the model. crack widths which are very high by comparison to the real ones.
According to the results in Table 8 for beams without fibers, the The WExp/WACI ratios for fiber reinforced concrete varied from 0.2
crack width predictions of Eurocode 2 seem to agree with the mea- to 0.6 with an average value of 0.34 as compared to average values
sured ones with a mean value of WExp/WEC2 of 1.11 and a standard of 0.44 and 0.46 respectively for BS8110 and EC2 (Table 9). The
deviation of 0.19. The variance coefficient of 18% indicates that the results of Tables 8 and 9 give the crack widths of concrete without
results, which were comparable to those predicted by BS8110, fibers and with fibers. They illustrate clearly the conservative
were once again scattered owing to the hazardous nature of crack- aspect of the American model for crack width prediction of con-
ing in concrete. Table 9 shows the crack width predictions for crete in general and for fiber reinforced concrete in particular
beams containing fibers; the ratio WExp/WEC2 varied from 0.20 to and stress on the need for more work on the cracking aspect of this
0.70, with a mean value of 0.46 expressing an overestimation of structural material in the aim of refining the existing design mod-
the predicted crack width in the same manner as for BS8110. For els. The ACI crack width model is directly proportional to the stress
concrete without fibers, the ratio WExp/WEC2 ranged from 0.8 to acting on the tension reinforcement and does not take into account
1.5 almost in the same interval as those of BS8110, with a mean the quality of bond between concrete and steel which tends to pre-
value of 1.11 (Table 8). This indicates that the EC2 model is fairly vent or at least delay the opening up of concrete in the form of a
satisfactory for predicting the crack width of concrete without crack. In the presence of higher steel stresses as a result of better
fibers, even though the improved bond quality of high strength a transmission of internal forces from concrete to the reinforce-
concrete need to be better expressed in the EC2 model in the same ment, higher crack widths will be predicted by this model without
manner as suggested for BS8110 model. For fiber reinforced con- taking into consideration the effective adherence between the two
crete however, both models, the EC2 and the BS8110, need modifi- materials which limits wider opening of the cracks. According to
cations and amendments to allow for the sewing effect of fibers the three crack width models, a higher concrete cover to the rein-
crossing the cracks. forcement leads to higher crack widths as observed experimentally
M. Hamrat et al. / Construction and Building Materials 106 (2016) 678–692 689

Table 7
Formulas for the crack width prediction given by the major design codes.

Model Expression Comments

ACI 318-95 [7] W k ¼ 0:011:b:rs : 3 dc :A  103 A = Ae/nb: effective tensile area/number of bars
nb: the number of tension reinforcing bars
dc: the distance measured from the centroid of tensile steel to the extreme tensioned fiber
b = h  x/d  x
rs: stress in tensile reinforcement calculated on basis of cracked section
Eurocode 2-04 [6] W k ¼ Srm ðesm  ecm Þ Kt = 0.6:for short term loading
Srm ¼ 3:4c þ 0:425k1 k2 /=qeff K1 = 0.8: for high bond reinforcing bars
qeff Þ K2 = 0.5: for sections subjected to pure bending
esm  ecm ¼ rEss  kt f ctmEð1þn
s qeff
c: concrete clear cover
fctm: mean value of tensile strength of the concrete
qeff = As/Aceff
(esm  ecm): mean tensile strain
Srm: crack spacing
/: bar diameter
cr em
BS8110-97 [8] W k ¼ 1þ2ðacr3a
cmin Þ=ðhxÞ
cmin: minimum cover to the tension steel
2 h: overall depth of the member
em ¼ hx
dx s
e  bðhxÞ
3Es As ðdxÞ x: depth of neutral axis, b: width of the section
acr: distance from the point considered to the surface of the nearest longitudinal bar
es: strain in the reinforcement
em: average strain at the level where the cracking is being considered;
Rilem TDF [20] W k ¼ srm :esm rs: stress in tensile reinforcement calculated on basis of cracked section;
esm ¼ rEss ð1  b1 :b2 ðrrsrs Þ2 Þ rsr: stress in the tension longitudinal reinforcement under the first crack
  Es: modulus of elasticity of the reinforcement
srm ¼ 50 þ 0:25:K 1 :K 2 : q/ : L 50
=d Lf: fiber length; df: fiber diameter
eff f f

b1: coefficient taking into account the bond properties of bar

b2: coefficient taking into account the duration of the load
(b1 = b2 = 1.0)
k1 = 0.8: for high bond bar
k2 = 0.5: for bending
£: bar diameter
Srm: average final crack spacing
esm: mean strain in the tension reinforcement

" ! #" #
by a number of researchers. However, such dependency is / 50 rs kt :f ctm ð1 þ nqeff Þ
W Modified Rilem model ¼ 50 þ 0:25:k1  k2  
expressed differently in the three models and the ACI is the only qeff Lf =df Es Es qeff
one of the three in which the crack width is directly proportional ð5Þ
to the concrete cover. The two arguments justify the conservative
aspect of the ACI model for crack with prediction. When readjusting the bond coefficient k1 to 0.6 instead of 0.8 as
To sum up, the three code models for predicting the crack width suggested in Rilem model (Tables 5 and 7), the predictions are
do not consider the local stitching actions of fibers on the cracks further improved with an average WExp/WModified Rilem model of
and the relatively reduced crack spacing in their predictions and 0.98 as presented in Table 10.
hence these models do not seem to be directly applicable to fiber The results of Tables 8–10 are drawn in Fig. 8(a and b) for
reinforced concrete. In this sense, the technical document RILEM clearer illustrations. The curves load versus crack width show that
TC-162-TDF [20] appears to be the only model which clearly takes some existing models (EC2 and BS8110) are relatively more realis-
into account the presence of fibers when predicting the crack tic in predicting the crack widths of concrete members without
width of fiber reinforced concrete; the model given in Tables 5 steel fibers, though requiring more refinement in the case of high
and 7 displays explicitly the fiber parameter Lf/df. In this model, strength concrete. In effect, the high quality bond between high
the crack width is expressed proportionally to the mean strain strength concrete and the reinforcement allows for a better
esm acting in the tension zone. However, such a strain seems to restraint of the cracking, leading to closely spaced finer cracks in
be overestimated (Table 10) and hence not realistically assessed structural members made of this concrete material. The existing
in the model since not expressed as a function of the fiber param- design models are mostly based on data taken from experimental
eters. Indeed, the experimental values of the mean strain given in tests carried out on ordinary concrete and hence not correctly cov-
Table 10 clearly reveal that this parameter is overestimated by the ering high strength concrete. However, when predicting crack
Rilem model. Consequently, this model predicts crack widths widths, one should bear in mind the unpredictable and hazardous
which are conservative, with an average ratio WExp/Wrilem of 0.76, nature of cracking. Fig. 8(b) shows clearly that none of the existing
and hence relatively unrealistic by comparison to the measured major design codes considered in this study, namely the American
ones as given in Table 9. Comparatively, the EC2 mean strain ACI, the European EC2 and the British BS 8110, predicts correctly
expressed as (esm  ecm) seems relatively closer to the experimental the crack widths of fiber reinforced concrete and all of the three
values (Table 10). When replacing the Rilem mean strain by the models completely omit the beneficial fiber stitching and restraint
EC2 mean strain given by the corresponding model in Table 7 in of the cracks in concrete members. The crack width of fiber rein-
the Rilem crack width model as in Eq. (5) below, improved predic- forced concrete could be overestimated by as much as 350‰ when
tions were obtained for the crack widths of fiber reinforced con- using these design codes. In contrast, the Rilem model is the only
crete, with an average values WExp/WModified Rilem model of 0.88 as one which explicitly takes this fiber effect into consideration and
given in Table 10. The modified Rilem model, which certainly yields relatively better predictions. Moreover, Fig. 8(b) shows
needs more refinement, is rewritten as follows: that the modifications proposed in this work for the Rilem model
690 M. Hamrat et al. / Construction and Building Materials 106 (2016) 678–692

Table 8
Predictions of the crack width for beams without steel fibers at service loads.

Authors Beam fc MPa As mm2 A’s mm2 P* kN WExp WACI WEC2 WBS8110 WExp/ WExp/ WExp/
(mm) (mm) (mm) (mm) WACI WEC2 WBS8110
Tan et al. [22] AF 40.8# 157 56.5 7.0 0.21 0.20 0.14 0.17 1.03 1.48 1.26
Purkiss [23] No fibers 66.9* 100 0 13.0 0.19 0.13 0.13 0.16 1.42 1.48 1.20
Vandewalle [24] Beam1 42# 628 0 60.0 0.13 0.23 0.15 0.20 0.57 0.83 0.63
Rachid [21] A111 42.8 1119 265 101 0.22 0.30 0.24 0.20 0.73 0.91 1.09
A211 42.8 1964 265 136 0.17 0.21 0.17 0.17 0.82 1.00 1.03
B211 74.6 1964 265 146 0.19 0.22 0.18 0.17 0.87 1.06 1.09
B211a 73.6 1964 265 147 0.18 0.22 0.18 0.18 0.82 1.00 1.02
B311 72.8 2946 265 221 0.18 0.28 0.18 0.20 0.64 1.00 0.90
B312 72.8 2946 265 215 0.19 0.27 0.18 0.20 0.69 1.08 0.97
B313 72.8 2946 265 218 0.15 0.28 0.15 0.20 0.54 0.97 0.76
B321 77 2946 531 225 0.18 0.28 0.19 0.20 0.63 0.97 0.89
B331 72.8 2946 804 227 0.18 0.29 0.19 0.20 0.63 0.96 0.88
C211 85.6 2366 265 191 0.24 0.27 0.19 0.20 0.91 1.25 1.20
C311 88.1 2768 265 215 0.24 0.27 0.18 0.20 0.88 1.33 1.20
C411 85.6 3574 265 265 0.23 0.29 0.18 0.21 0.79 1.30 1.11
C511 88.1 4330 265 259 0.22 0.27 0.14 0.18 0.82 1.55 1.21
D211 115 1964 265 178 0.24 0.26 0.22 0.21 0.92 1.08 1.14
E211 126 1964 265 175 0.22 0.25 0.22 0.21 0.86 1.02 1.06
Kumar [25] C1 74# 402 99.0 21.4 0.22 0.26 0.21 0.22 0.85 1.07 0.99
C2 74# 628 98.5 32.9 0.21 0.27 0.20 0.23 0.77 1.06 0.91
C3 74# 942 99.0 46.2 0.22 0.23 0.17 0.21 0.97 1.33 1.06
CS1 78# 402 99.0 21.2 0.17 0.25 0.20 0.22 0.67 0.84 0.77
CS1 78# 628 98.5 32.9 0.23 0.27 0.20 0.23 0.84 1.17 1.00
CS1 78# 942 99.0 46.8 0.18 0.23 0.17 0.21 0.79 1.08 0.86
Maghsoud [26] BC6 73.7 2462 1231 609 0.30 0.41 0.30 0.32 0.73 0.98 0.93
B6 71.0 2462 0 561 0.29 0.39 0.28 0.31 0.74 1.03 0.93
BC7 66.8 2864 1448 723 0.40 0.31 0.30 0.26 1.27 1.34 1.51
B7 70.5 2864 0 608 0.32 0.27 0.25 0.23 1.19 1.30 1.41
BC8 77.7 3511 1740 767 0.38 0.32 0.26 0.29 1.17 1.46 1.31
B8 71.8 3511 0 737 0.31 0.32 0.25 0.29 0.96 1.25 1.08
Present study B-N10 40 157 0 12.4 0.21 0.22 0.17 0.18 0.97 1.23 1.19
B-N12 40 226 0 15.6 0.17 0.20 0.15 0.17 0.83 1.13 1.03
B-N14 40 308 0 23.9 0.16 0.25 0.17 0.20 0.64 0.92 0.80
B-H10 85 157 0 13.9 0.19 0.23 0.18 0.19 0.81 1.06 0.98
B-H12 85 226 0 18.5 0.15 0.23 0.17 0.19 0.65 0.88 0.79
B-H14 85 308 0 25.4 0.14 0.25 0.18 0.20 0.56 0.79 0.69
Mean 0.84 1.11 1.02
STD 0.20 0.19 0.19
COV 0.24 0.18 0.19
Cube compressive strength.
Service load.

Table 9
Predictions of the crack width for beams with steel fibers at service loads.

Authors Beam fc MPa P* kN WExp mm WACI mm WEC2 mm WBS mm WRilem mm WExp/WACI WExp/WEC2 WExp/WBS WExp/WRilem
Tan [22] BF 41 9.0 0.10 0.23 0.16 0.14 0.09 0.43 0.63 0.71 1.08
CF 42# 10.9 0.08 0.27 0.19 0.15 0.11 0.30 0.42 0.55 0.74
DF 42# 10.0 0.04 0.24 0.17 0.09 0.09 0.17 0.24 0.45 0.43
EF 41# 10.0 0.02 0.25 0.17 0.05 0.10 0.08 0.12 0.38 0.21
Purkiss [23] Fibers 69* 22.5 0.10 0.17 0.18 0.21 0.18 0.58 0.57 0.48 0.55
Vandewalle [24] Beam2 42# 77.8 0.12 0.24 0.20 0.21 0.10 0.49 0.60 0.57 1.20
Beam3 42# 78.2 0.10 0.23 0.19 0.20 0.10 0.43 0.52 0.49 1.04
Beam4 42# 78.0 0.08 0.24 0.17 0.20 0.07 0.34 0.46 0.40 1.14
Beam5 42# 78.3 0.10 0.23 0.21 0.20 0.08 0.43 0.48 0.50 1.19
Viktor[27] S3-1-F05 56 28.0 0.09 0.20 0.16 0.20 0.13 0.45 0.58 0.46 0.70
S3-2-F05 56 32.1 0.08 0.25 0.22 0.25 0.18 0.32 0.37 0.32 0.45
S3-1-F10 48 35.1 0.07 0.23 0.20 0.23 0.19 0.30 0.35 0.31 0.37
S3-2-F10 48 29.8 0.07 0.21 0.13 0.18 0.15 0.33 0.52 0.38 0.46
S3-1-F15 52 28.2 0.02 0.14 0.04 0.11 0.10 0.14 0.46 0.18 0.21
S3-2-F15 52 30.4 0.04 0.20 0.09 0.17 0.14 0.20 0.44 0.24 0.29
Present study B-F10 78 15.0 0.13 0.23 0.18 0.19 0.08 0.56 0.73 0.68 1.63
B-F12 78 19.3 0.09 0.23 0.17 0.19 0.07 0.39 0.53 0.48 1.31
B-F14 78 26.3 0.05 0.25 0.18 0.20 0.07 0.20 0.28 0.25 0.71
Mean 0.34 0.46 0.44 0.76
STD 0.14 0.14 0.14 0.41
Cube compressive strength.
Service load.
M. Hamrat et al. / Construction and Building Materials 106 (2016) 678–692 691

(Eq. (5) with k1 = 0.6) improve further the crack width predictions
of fiber reinforced concrete. More work is however needed to
assess the Rilem model and the modifications brought to it.
K1 = 0.6







4. Conclusions

The following conclusions can be drawn from the present

experimental work using the DIC measurement technique:

1. The comparison between the classical measurement techniques

(strain gauges, LVDT sensors) and the DIC technique suggests
K1 = 0.8

that the two techniques are suitable for the analysis of strain







components. The rather good agreement between the two mea-
surement methods indicates that the DIC technique is an effi-
cient measuring tool for obtaining displacement and strain
WMod. Rilem Eq. (5)

fields during all the loading process from the start until failure.
K1 = 0.6 (mm)

Measurements of strains and displacements at or close to fail-

ure are usually not possible with the classical methods due to
Mean the risk involved in terms of safety to the personnel and damage




SD to the equipment.
2. The experimental results confirm that fibers, when efficiently
used, are able to increase the first-cracking load by as much
as 10–25% of the ultimate load and reduce the strain of longitu-
WMod. Rilem Eq. (5)

dinal reinforcement. As a consequence, they reduce both the

K1 = 0.8 (mm)

crack spacing and the crack width by as much as 35–70%.

3. The BS8110 and Eurocode 2 formulas for the crack width pre-
dictions were found in good agreement with the measured






crack widths for reinforced concrete beams without steel fibers;

the average ratios of WExp/WBS8110 and WExp/WEC2 were 1.02 and
1.11, respectively. The ACI code formula, however, was found to

overestimate the values of the crack widths (with an average





ratio of WExp/WACI of 0.84). When applied to fiber reinforced

concrete beams, the three design codes (ACI318, Eurocode 2
and BS8110) were found to seriously overestimate the mea-
esm  ecm  104

sured crack widths; the overestimations vary from 30% to

350%. The stitching and restraining effects of fibers on the
cracks do not seem to be considered in the models used in these

major design documents. The application of these major design





codes of crack’s design in fiber reinforced concrete appears to be

Comparison of measured and predicted crack width at service load -Modified Rilem model-.

limited and even more so for high strength fiber reinforced con-
crete. The improved cracking behavior induced by the presence
esm  ecm  104

of fibers is not considered in these design documents.

4. The RILEM TC-162-TDF crack width model is the only one
which considers the efficient restraining and stitching effects
on the cracks and gives relatively better predictions for the




crack width of steel fiber reinforced concrete beams. However,

some parameters thought to be of importance in the cracking
process are not correctly represented in the model. In this
sense, the average tension strain (esm) and the bond coefficient
esm ecm  104

of the tension reinforcement k1 seem both to be overestimated.

Based on the present experimental results and on data from
others, modifications are proposed in the present study for

the Rilem model. The modifications put forward concern mainly

the average tension strain and the bond coefficient. The modi-
fied Rilem model was used to predict the crack widths of the

steel fiber reinforced concrete beams of the present study and



those of other data taken from the literature. A substantial



improvement of the crack width prediction is obtained. The

modified Rilem model needs however more experimental work
to be further assessed and refined.
Vandewalle [24]

Present study
Purkiss [23]

Viktor [27]
Tan [22]

Table 10

[1] L.R. Taerwe, Influence of steel fibers on strain-softening of high strength

concrete, ACI Mater. J. 88 (6) (1992) 54–60.
692 M. Hamrat et al. / Construction and Building Materials 106 (2016) 678–692

[2] B. Boulekbache, M. Hamrat, M. Chemrouk, S. Amziane, Flowability of fibre- [16] GOM Aramis user manual v6 (2007) GOM mbH, Germany, (http://
reinforced concrete and its effect on the mechanical properties of the material,
Constr. Build. Mater. 24 (9) (2010) 1664–1671. [17] CSA94:CSA Technical Committee, Design of concrete structure for buildings,
[3] M. Di Prisco, FRC: structural applications and standards, Mater. Struct. 42 (9) CAN3-A23.3-M94, Canadian Standards Association, Rexdale, Ontario, 1994.
(2009) 1169–1171. [18] Eurocode 2: design of concrete structures-part 1: general rules and rules for
[4] State-of-the-Art Report on high strength concrete reported by ACI Committee buildings, DD ENV 1992-1-1:1992. CEN (Comité Européen de Normalisation).
363–1997, ACI 363R–92, 1997. Brussels, 1992.
[5] B.W. Russell, Impact of high strength concrete on the design and construction [19] CEB-FIP model code 1990: design code, Comité Euro International du Béton-
of pretensioned girder bridges, PCI J. 39 (4) (1994) 76–89. Fédération International de la Précontraint, London, Thomas Telford Service
[6] Eurocode 2: design of concrete structures-part 1: general rules and rules for Ltd, 1993.
buildings, BS EN 1992-1-1:2004, CEN (Comité Européen de Normalisation), [20] T.C. Rilem, 162-TDF. Final recommendation of RILEM TC 162-TDF: test and
Brussels, 2004. design methods for steel fibre reinforced concrete, Mater. Struct. 36 (262)
[7] ACI Committee 318, Building code requirements for structural concrete, ACI (2003) 560–567.
318-95 and commentary, Michigan: ACI, Farmington Hills, 1995. [21] M.A. Rashid, M.A. Mansur, Reinforced high strength concrete beams in flexure,
[8] Structural Use of Concrete. Part 2: Code of Practice for Special Circumstances ACI Struct. J. 102 (5) (2005) 462–471.
BS 8110: Part 2: British Standard Institution, London, 1997. [22] K.H. Tan, P. Paramasivam, K.C. Tan, Cracking characteristics of reinforced steel
[9] B. Boulekbache, M. Hamrat, M. Chemrouk, S. Amziane, Failure mechanism of fiber concrete beams under short and long term loadings, Adv. Cem. Based
fibre reinforced concrete under splitting test using digital image correlation, Mater. 2 (4) (1995) 127–137.
Mater. Struct. 48 (2015) 2713–2726. [23] J.A. Purkiss, P.J. Wilson, P. Blagojević, Determination of the load-carrying
[10] M. Hamrat, B. Boulekbache, M. Chemrouk, S. Amziane, Shear behaviour of RC capacity of steel fibre reinforced concrete beams, Compos. Struct. 38 (1) (1997)
beams without stirrups made of normal strength and high strength concretes, 111–117.
Adv. Struct. Eng. 13 (1) (2010) 29–41. [24] L. Vandewalle, Cracking behaviour of concrete beams reinforced with a
[11] E.B. Pereira, G. Fischer, J.A.O. Barros, Direct assessment of tensile stress-crack combination of ordinary reinforcement and steel fibers, Mater. Struct. (1999)
opening behavior of strain hardening cementitious composites (SHCC), Cem. 164–170.
Concr. Res. 42 (6) (2012) 834–846. [25] P.S. Kumar, M.A. Mannan, V.J. Kurian, H. Achuytha, Investigation on the
[12] M. Hamrat, B. Boulekbache, M. Chemrouk, S. Amziane, Effects of the transverse flexural behaviour of high performance reinforced concrete beams using
reinforcement on the shear behaviour of high strength concrete beams, Adv. sandstone aggregates, Build. Environ. 42 (7) (2007) 2622–2629.
Struct. Eng. 15 (8) (2012) 1291–1306. [26] A.A. Maghsoudi, Y. Sharifi, The serviceability considerations of HSC heavily
[13] N. Attari, S. Amziane, M. Chemrouk, Flexural strengthening of concrete beams steel reinforced members under bending, Am. J. Appl. Sci. 5 (9) (2008) 1135–
using CFRP, GFRP and hybrid FRP sheets, Constr. Build. Mater. (2012) 746–757. 1140.
[14] Y.A. Syed, S. Jacqueline, A. Loukili, Fracture examination in concrete through [27] G. Viktor, K. Gintaris, K.H.K. Albert, B. Darius, U. Darius, Deriving stress–strain
combined digital image correlation and acoustic emission techniques, Constr. relationships for steel fibre concrete in tension from tests of beams with
Build. Mater. (2014) 232–242. ordinary reinforcement, Eng. Struct. 42 (2012) 387–395.
[15] S.G. Shah, K.J.M. Chandra, Fracture properties of concrete-concrete interfaces
using digital Image correlation, Exp. Mech. 51 (3) (2011) 303–313.