Cement and Concrete Composites 65 (2016) 139e149

Contents lists available at ScienceDirect

Cement and Concrete Composites

journal homepage: www.elsevier.com/locate/cemconcomp

Fiber volume fraction and ductility index of concrete beams

Alessandro P. Fantilli*, Bernardino Chiaia, Andrea Gorino
Department of Structural, Building and Geotechnical Engineering, Politecnico di Torino, Corso Duca degli Abruzzi 24, 10129 Torino, Italy

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

Article history: The mechanical response of fiber-reinforced concrete (FRC) beams depends on the amount of fibers, and
Received 27 May 2015 the transition from brittle to ductile behavior in bending is related to a critical value of fiber volume
Received in revised form fraction. Such quantity, which is mechanically equivalent to the minimum amount of steel rebars in
2 October 2015
reinforced concrete beams, can be defined according to the new approach proposed herein. It derives
Accepted 9 October 2015
from the application of a general model and from the introduction of the so-called ductility index (DI).
Available online 19 October 2015
When FRC beams show a ductile behavior DI is positive, whereas DI is negative in the case of brittle
response. Both the theoretical and experimental results prove the existence of a general linear rela-
Keywords:
Fiber-reinforced concrete
tionship between DI and the fiber volume fraction. Accordingly, a new design-by-testing procedure can
Beams be used to determine the critical value of fiber volume fraction, which corresponds to a ductility index
Bending moment equal to zero.
Deflection-hardening © 2015 Elsevier Ltd. All rights reserved.
Ultimate limit state
Ductility index

1. Introduction whereas at ultimate load Pu (the second relative maximum) strain

Depending on the fiber volume fraction Vf used to reinforce the
cementitious matrix, fiber-reinforced concrete (FRC) ties behave
differently (Fantilli et al. [1], Naaman [2]). Specifically, in the case of
low values of Vf, the tensile force F after cracking (which occurs at
the cracking load Fcr), remains lower than Fcr as the elongation DL
increases (see Fig. 1a). Such ties fail in a brittle manner, and the so-
called strain-softening occurs in the presence of a single crack.
Conversely, with an higher amount of fibers, and after the softening
stage subsequent to the growth of the first crack, F increases and
reaches the ultimate load Fu > Fcr. This is the case of the ductile
response, in which the ties show strain-hardening behavior and
multiple cracking. At the transition from the brittle to the ductile
response, when Fu ¼ Fcr (Fig. 1a), the corresponding critical value of
Vf can be considered as the minimum amount of fibers that gua-
rantees the strain-hardening behavior of the FRC ties (Fantilli et al.
[1]).
Similarly, the behavior of FRC beams in three point bending, as
illustrated in Fig. 1b in terms of applied load P vs. midspan deflec-
tion d, is also a function of Vf (Naaman [2]). At the first relative
maximum (when P ¼ Pcr*), the effective cracking takes place,
* Corresponding author.
E-mail addresses: alessandro.fantilli@polito.it (A.P. Fantilli), bernardino.chiaia@
polito.it (B. Chiaia), andrea.gorino@polito.it (A. Gorino).
localization occurs in the tensile zone. The value of Pcr* is in turn
higher than the first cracking load (i.e., Pcr), which corresponds to
the attainment of the tensile strength in the bottom edge of the FRC
beam (Fig. 1b). For low amounts of fibers, Pu is always lower than
Pcr*, and the brittle response of the beams (called deflection-
softening) is evidenced by the presence of a single crack.
Conversely, both the deflection-hardening (i.e., Pu > Pcr* in Fig. 1b),
and the presence of more than one crack as well, are the results of
tests performed on FRC beams containing high amounts of fibers.
Consequently, at the transition from brittle to ductile behavior,
when Pu ¼ Pcr* in Fig. 1b, the minimum ductility is attained and a
critical amount of fibers can be defined. As this quantity of fibers
has the same mechanical role of the minimum area of steel rein-
forcing bars in reinforced concrete beams (Fantilli et al. [3]), it can
be defined as the minimum amount of fiber-reinforcement (i.e.,
Vf ¼ Vf,min in Fig. 1b).
A univocal and simple approach able to predict Vf,min in FRC
beams does not exist, despite the huge number of tests available in
the current literature. For instance, Naaman [2] proposed a formula
to compute Vf,min based on the equation Pcr* ¼ Pu, where the values
of both the loads are functions of the flexural and of the residual
tensile strengths of FRC, respectively. However, the relationship
between the strengths and the fiber volume fraction cannot be
easily and univocally defined, hence the analytical prediction of
Vf,min is not always effective. On the other hand, the approaches
suggested by code rules are even more complicated, due to the

http://dx.doi.org/10.1016/j.cemconcomp.2015.10.019
0958-9465/© 2015 Elsevier Ltd. All rights reserved.
140 A.P. Fantilli et al. / Cement and Concrete Composites 65 (2016) 139e149
Fig. 1. The behavior of FRC structures as a function of the fiber content: (a) axial load vs. elongation diagram of a tie; (b) applied load vs. midspan deflection of a beam in three point
bending.
large experimental campaigns required for the definition of Vf,min.
In particular, Model Code 2010 (fib [4]) firstly recommends the
classification of FRC, and the evaluation of the residual strengths, by
means of three point bending tests on notched beams. Then, the
displacements measured in a second series of tests, performed on
full-scale FRC elements in bending, with different contents of fibers,
are needed. The ductility requirement in bending (and the corre-
sponding Vf  Vf,min) is satisfied when the ultimate or the peak
displacements are sufficiently large (Caratelli et al. [5], de la Fuente
et al. [6]).
With the aim of simplifying the evaluation of Vf,min, a new
design-by-testing procedure, capable of predicting the brittle/
ductile behavior of FRC beams in bending, is proposed in the
following. It is the result of both the theoretical and experimental
investigations described in the next sections.
2. General model
A multi-scale general model is introduced herein to predict the
behavior of the FRC beam depicted in Fig. 1b. The fiber-
reinforcement is modeled with an ideal tie (Fig. 2a), composed by
a straight fiber and the surrounding cementitious matrix, having a
single orthogonal crack in the midsection. The pullout mechanism
of this element provides the stress-strain relationship of the
cracked FRC. Only when this relationship is known, can the me-
chanical response of the FRC beams in bending be properly defined.
2.1. Modeling the fiber pullout
The ideal tie illustrated in Fig. 2a has a square cross-section, in
which the area of the cementitious matrix Ac is a function of the
amount of fibers used in the FRC beam:
Af p$f2
Ac ¼ ¼ (1)
Vf 4$Vf
where Af, f ¼ area and diameter of the fiber cross-section,
respectively.
The portion of the tie delimited by the cracked cross-section (in
the midspan) and the so-called Stage I cross-section (where the
perfect bond between steel and concrete is re-established) is
investigated. Within this block of length ltr (¼ transfer length), as
the horizontal coordinate z increases, stresses move from steel to
concrete in tension, due to the bond-slip mechanism acting at the
interface of the materials. Such slip s vanishes in the Stage I cross-
section (Fig. 2b), where stresses (of fiber sf,I and of concrete sc,I in
Fig. 2c) are computed with the well-known linear elastic formulae,
under the hypothesis of perfect bond between the materials:
N
sf;I ¼ n$ (2)
Ac þ n$Af
N
sc;I ¼ (3)
Ac þ n$Af
where n ¼ Ef/Ec ¼ ratio between the Young's moduli of the fiber and
of the cementitious matrix; N ¼ axial load applied to the ideal tie
(Fig. 2a).
Within the transfer length ltr, the interaction between fiber and
matrix is described by the following equilibrium and compatibility
 A.P. Fantilli et al. / Cement and Concrete Composites 65 (2016) 139e149 141

In Eq. (7), the value of the slip s0 in the cracked cross-section is

equal to the half of crack width w, whereas Eq. (8) states the
absence of slips at a distance ltr from the mid-section. Only when
ltr < Lf/2 (Lf ¼ length of the fiber), can this condition be considered
valid. Finally, Eq. (9) and Eq. (10) impose the Stage I conditions [Eq.
(2) and Eq. (3)] for the state of stress in fiber and matrix,
respectively.
According to Model Code 2010 (fib [4]), the mean tensile
strength of concrete fct can be estimated from the compressive
strength fc (expressed in MPa):

fct ¼ 0:3$ðfc  8 Þ2=3 for fc  58 MPa (11.a)

fct ¼ 2:12$lnð 1 þ 0:1$fc Þ for fc > 58 MPa (11.b)

The residual tensile stress on the crack surfaces of the ideal tie
(sc0 in Fig. 2c) can be defined by the “fictitious crack model” shown
in Fig. 3a. It consists of a bilinear stress-crack opening displacement
relationship, sc e w, as proposed by Model Code 2010 (fib [4]):
 
w
sc ðwÞ ¼ fct $ 1:0  0:8$ for 0 < w  w1 (12.a)
w1
 
w
sc ðwÞ ¼ fct $ 0:25  0:05$ for w1 < w  wc (12.b)
w1

where w1 ¼ GF/fct; wc ¼ 5$GF/fct; and GF ¼ 0:073$fc0:18 ¼ fracture

energy of concrete in tension (fc in MPa).
In the same way, the interaction between fiber and matrix
(Fig. 2b) needs the definition of a bond-slip t e s relationship. For
the sake of simplicity, the model proposed by Fantilli and Vallini [8],
originally developed for smooth steel fibers in a cementitious
Fig. 2. Modeling the fiber pullout: (a) the ideal tie composed by a straight fiber and the matrix, is adopted (see Fig. 3b):
surrounding cementitious matrix in presence of a single crack; (b) slip between fiber  a
and matrix; (c) stresses in the fiber and in the matrix. s
t ¼ tmax $ for 0  s < s1 (13.a)
s1
equations, according to Chiaia et al. [7]:  
t ¼ tf þ tmax  tf $eb$ðs1 sÞ for s1  s (13.b)
dsf p 4
¼  f $t½sðzÞ ¼  $t½sðzÞ (4)
dz Af f where tmax ¼ maximum bond stress; tf ¼ residual bond stress;
s1 ¼ 0.1 mm; a ¼ 0.5; and b ¼ 2/mm.
" #
ds sf ðzÞ sc ðzÞ The value of tf, which is a function of the compressive strength
¼  (5) (in MPa), and that of tmax, which also depends on the fiber diameter
dz Ef Ec
(in mm), can be evaluated with the following formulae (Fantilli and
Vallini [8]):
where sf, sc ¼ stress in fiber and matrix, respectively (Fig. 2c);
pf ¼ perimeter of the fiber cross-section; and t ¼ bond stress cor- pffiffiffiffi
tf ¼ 0:1$ fc (14)
responding to the slip s between the materials (Fig. 2b).
The resultant of axial stresses, acting in each cross-section of the pffiffiffiffi
1:572
tie, can be computed as: tmax ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi$ fc (15)
12:5 þ f
N ¼ sf $Af þ sc $Ac (6)
To solve the system composed by the Eqs. (4)e(6), the following
boundary conditions are needed:
2.2. Numerical evaluation of N (w)
w
s0 ¼ (7) The pullout model previously described, and referred to the
2
ideal tie of Fig. 2, consists of the so-called “tension-stiffening”
problem, which has to be solved within the one-dimensional
sðz ¼ ltr Þ ¼ 0 (8)
domain of length ltr (Fig. 2a). The proposed solution is based on
the following iterative procedure (see the flow-chart depicted in
sf ðz ¼ ltr Þ ¼ sf;I (9) Fig. 4):

sc ðz ¼ ltr Þ ¼ sc;I (10) 1. Assign a value to the crack width w in the midsection of the ideal
tie (Fig. 2a).
142 A.P. Fantilli et al. / Cement and Concrete Composites 65 (2016) 139e149

Fig. 3. Residual stress on the crack surface and bond stress at the interface of fiber and matrix: (a) fictitious crack model proposed by Model Code 2010 (fib [4]); (b) bond-slip model
proposed by Fantilli and Vallini [8].

2. Assume a trial value to the axial load N (Fig. 2a).
3. Compute the slip s0 in the midsection (z ¼ 0 in Fig. 2b) with Eq.
(7).
4. Calculate the tensile stress of the matrix sc0 in the midsection
(z ¼ 0 in Fig. 2c) by means of Eq. (12).
5. According to Eq. (6), the tensile stress of the fiber in the
midsection (z ¼ 0 in Fig. 2c) can be evaluated with the following
equation:
N  sc0 $Ac
sf0 ¼
Af
according to Chiaia et al. [9], the crack opening displacement w can
be smeared into the equivalent strain εF as follows (Fig. 5b):
sc ðwÞ w
εF ¼ þ (20)
Ec Lf
In other words, Lf is assumed to be the characteristic length of
the FRC in tension, where the inelastic strains localize. Similarly, the
axial load N of Fig. 5a, applied to the cross-sectional area of the tie
(i.e., Af þ Ac in Fig. 2a), turns into the tensile stresses sF of the
(16)
cracked FRC (Fig. 5b):

In order to exclude the failure of the fiber, and the subsequent N

brittle response of the ideal tie, sf0 must be lower than fu (where sF ¼ (21)
Ac þ Af
fu ¼ tensile strength of the fiber).

6. In the Stage I cross-section (z ¼ ltr in Fig. 2c), compute sf,I with
Eq. (2) and sc,I with Eq. (3).
7. Consider Dl as a small part of the unknown ltr < Lf/2, and define
zi ¼ i ∙ Dl (where i ¼ 1, 2, 3, …).
8. For each i (or zi) calculate:
- The bond stress ti, related to the slip si1 [Eq. (13)];
- The stress sf,i in the fiber, by using Eq. (4) written in the finite
difference form:
4 moment Mcr* and the ultimate bending moment Mu, which occur in
sf;i ¼ sf ;i 1  $ti $Dl
f
- The stress sc,i in the matrix according to Eq. (6):
N  sf;i $Af
sc;i ¼
Ac
- The slip si by means of the finite difference form of Eq. (5):
! l ¼ parameter corresponding to the strain at the origin of the co-
sf ;i sc;i
si ¼ si1   $Dl
Ef Ec
9. When si ¼ 0 [Eq. (8)], if sf,i s sf,I or sc,i s sc,I [Eq. (9) and Eq.
(10), respectively], change N and go back to step 3.
For a given w, such procedure calculates the corresponding axial
load N. Therefore, the complete N - w curve illustrated in Fig. 5a can
be obtained by varying the assigned crack width. Then, the me-
chanical response of the cracked FRC in tension (when the strain in
FRC εF is higher than that at cracking εF1 ¼ fct/Ec in Fig. 5b) is defined
on the base of the pullout response previously computed. Indeed,
2.3. Modeling the beam response
The behavior of the FRC beam illustrated in Fig. 1b can also be
defined by the moment M vs. curvature m relationship of a single
cross-section. In fact, the cross-sectional M - m curves and the P -
d diagrams (of Fig. 1b) have similar shapes, both depending on the
amount of fibers Vf. As a consequence, the ductility requirement in
bending can be assessed by comparing the effective cracking
(17) the midspan cross-section of the FRC beam when P ¼ Pcr* and P ¼ Pu
(see Fig. 1b), respectively.
As depicted in Fig. 6, to evaluate the moment-curvature rela-
tionship of a beam cross-section, having a width B and a depth H
(see Fig. 6b), the classical hypothesis of linear strain profile is
adopted (Fig. 6c):
(18)
εF ¼ l þ m$y (22)
where εF ¼ strain in a generic zone of the cross-section;
(19) ordinate y (located at H/2 from the edges of the cross-section e
Fig. 6b).
In agreement with Chiaia et al. [9], in the absence of an external
axial load, the resultant R of the cross-sectional stresses becomes:
þH=2
Z
R ¼ B$ sF dy ¼ 0 (23)
H=2
where sF ¼ stress in a generic zone of the FRC cross-section.
The internal bending moment M, corresponding to a given state
of stress, can be computed as follows [9]:
 A.P. Fantilli et al. / Cement and Concrete Composites 65 (2016) 139e149 143

Fig. 5. The behavior of FRC in tension: (a) axial load vs. crack width diagram of an ideal
tie; (b) stress-strain relationship of the fiber-reinforced composite.

needs to be defined. It is composed by the ascending branch of the

Sargin's parabola (fib [4]) in compression, and by the linear elastic
law in tension (Fig. 7):
 
k$h  h2
sF ¼ fc $ for  εc1  εF < 0 (25.a)
1 þ ðk  2Þ$h

sF ¼ Ec $εF for 0  εF < εF1 (25.b)

where k ¼ Ec/Ec1 ¼ plasticity number (i.e., the ratio between the

tangent modulus of elasticity at the origin of the stress-strain dia-
gram Ec, and the secant modulus from the origin to the peak
compressive stress Ec1); h ¼ j εF/εc1 j ¼ compressive strain
normalized with respect to εc1 (¼ strain at the peak of stress in
compression).
The complete stress-strain sF e εF relationship of the FRC,
including both the uncracked and the cracked stages, is reported in
Fig. 7.

2.4. Numerical evaluation of the moment-curvature relationship

Fig. 4. Flow-chart for the computation of the fiber pullout.
Due to the nonlinear response of the FRC (Fig. 7), the cross-
sectional M - m relationship of the beams in bending (Fig. 6a) can
be defined through the following iterative procedure (see the cor-
þH=2
Z responding flow-chart in Fig. 8):
M ¼ B$ sF $y dy (24)
1. Assign a value to the curvature m (Fig. 6c).
H=2
2. Assume a trial value for the strain parameter l (Fig. 6c).
To obtain the state of stress in the FRC cross-section, also the 3. Divide H in m stripes of depth Dh ¼ H/m (Fig. 6b), and define the
mechanical response of the uncracked FRC (when - εc1  εF < εF1) ordinate yj (where j ¼ 1, 2, … m):
144 A.P. Fantilli et al. / Cement and Concrete Composites 65 (2016) 139e149

Fig. 6. Modeling FRC beams in bending: (a)e(b) geometry of the three point bending
beam; (c) strain profile in a cross-section; (d) diagram of bending moment; (e) cross-
sectional moment-curvature diagram.

Fig. 8. Flow-chart for the computation of the moment-curvature relationship.

Fig. 7. Complete stress-strain relationship of FRC.

3. Definition of the ductility index

The brittle/ductile behavior of 54 ideal FRC beams in bending

(Fig. 1b) is herein assessed by means of the M - m relationship
H previously introduced. The FRC specimens, having B ¼ H/2 (Fig. 6),
yj ¼  þ j$Dh (26)
2 are divided into 18 groups of 3 beams, with the same geometrical
and material properties, but different fiber volume fractions Vf. Two
4. For each j (or yj) calculate depths (H ¼ 200 and 400 mm) and three classes of cementitious
- The strain εF,j with Eq. (22); matrix (fc ¼ 30, 45 and 60 MPa) are taken into account. The rein-
- The stress sF,j (corresponding to εF,j) by means of the nonlinear forcement of each group consists of steel fibers (Lf ¼ 60 mm,
relationship depicted in Fig. 7. fu ¼ 1000 MPa, and Ef ¼ 210,000 MPa) having one of the following
5. Compute the resultant R of the cross-sectional stresses [Eq. aspect ratios: Lf/f ¼ 40, 60, and 80.
(23)]. A single ideal FRC beam is identified by the alphanumeric label
6. If R s 0, then change l and go to step 4. SX_CYY_AZZ_W, where X is related to the beam depth (X ¼ 1 when
7. Calculate the internal bending moment M [Eq. (24)] H ¼ 200 mm, and X ¼ 2 when H ¼ 400 mm), YY is the compressive
strength (in MPa), ZZ is the fiber aspect ratio, and W is a number (1,
For a given m, such procedure provides the corresponding 2, or 3) associated to Vf. All the mechanical and geometrical prop-
bending moment M. Hence, the complete M - m diagram illustrated erties of the beams are reported in Table 1 (for the beams S1) and
in Fig. 6e can be obtained by varying the assigned curvature. Table 2 (for the beams S2).
 A.P. Fantilli et al. / Cement and Concrete Composites 65 (2016) 139e149 145

Table 1
Properties of the ideal beams S1.

Group Beam H (mm) fc (MPa) Lf/f Vf (%) Mcr* (kNm) Mu (kNm) DI Vf,min (%) v

1 S1_C30_A80_1 200 30 80 0.30 3.06 2.39 0.22 0.44 0.68

S1_C30_A80_2 0.50 3.29 3.61 0.10 1.13
S1_C30_A80_3 0.70 3.56 4.86 0.36 1.58
2 S1_C30_A60_1 200 30 60 0.40 3.10 2.43 0.22 0.59 0.67
S1_C30_A60_2 0.60 3.28 3.33 0.02 1.01
S1_C30_A60_3 0.80 3.49 4.27 0.22 1.35
3 S1_C30_A40_1 200 30 40 0.70 3.21 2.76 0.14 0.90 0.77
S1_C30_A40_2 1.00 3.44 3.69 0.07 1.11
S1_C30_A40_3 1.30 3.69 4.65 0.26 1.44
4 S1_C45_A80_1 200 45 80 0.30 3.87 2.74 0.29 0.48 0.63
S1_C45_A80_2 0.50 4.09 4.30 0.05 1.05
S1_C45_A80_3 0.70 4.31 5.85 0.36 1.47
5 S1_C45_A60_1 200 45 60 0.40 3.91 2.77 0.29 0.64 0.63
S1_C45_A60_2 0.60 4.10 3.93 0.04 0.94
S1_C45_A60_3 0.80 4.29 5.11 0.19 1.25
6 S1_C45_A40_1 200 45 40 0.70 4.05 3.18 0.21 0.98 0.72
S1_C45_A40_2 1.00 4.26 4.36 0.02 1.02
S1_C45_A40_3 1.30 4.47 5.55 0.24 1.33
7 S1_C60_A80_1 200 60 80 0.30 4.54 3.08 0.32 0.48 0.62
S1_C60_A80_2 0.50 4.71 4.89 0.04 1.03
S1_C60_A80_3 0.70 4.88 6.69 0.37 1.45
8 S1_C60_A60_1 200 60 60 0.40 4.58 3.10 0.32 0.65 0.62
S1_C60_A60_2 0.60 4.72 4.47 0.05 0.93
S1_C60_A60_3 0.80 4.87 5.83 0.20 1.24
9 S1_C60_A40_1 200 60 40 0.70 4.69 3.57 0.24 0.99 0.71
S1_C60_A40_2 1.00 4.87 4.94 0.02 1.01
S1_C60_A40_3 1.30 5.05 6.32 0.25 1.32

Table 2
Properties of the ideal beams S2.

Group Beam H (mm) fc (MPa) Lf/f Vf (%) Mcr* (kNm) Mu (kNm) DI Vf,min (%) v

10 S2_C30_A80_1 400 30 80 0.30 24.47 19.12 0.22 0.44 0.68

S2_C30_A80_2 0.50 26.30 28.88 0.10 1.13
S2_C30_A80_3 0.70 28.52 38.89 0.36 1.58
11 S2_C30_A60_1 400 30 60 0.40 24.77 19.40 0.22 0.59 0.67
S2_C30_A60_2 0.60 26.24 26.67 0.02 1.01
S2_C30_A60_3 0.80 27.92 34.20 0.22 1.35
12 S2_C30_A40_1 400 30 40 0.70 25.71 22.07 0.14 0.90 0.77
S2_C30_A40_2 1.00 27.51 29.51 0.07 1.11
S2_C30_A40_3 1.30 29.53 37.19 0.26 1.44
13 S2_C45_A80_1 400 45 80 0.30 30.94 21.91 0.29 0.48 0.63
S2_C45_A80_2 0.50 32.76 34.38 0.05 1.05
S2_C45_A80_3 0.70 34.50 46.82 0.36 1.47
14 S2_C45_A60_1 400 45 60 0.40 31.29 22.13 0.29 0.64 0.63
S2_C45_A60_2 0.60 32.76 31.47 0.04 0.94
S2_C45_A60_3 0.80 34.30 40.90 0.19 1.25
15 S2_C45_A40_1 400 45 40 0.70 32.36 25.42 0.21 0.98 0.72
S2_C45_A40_2 1.00 34.09 34.84 0.02 1.02
S2_C45_A40_3 1.30 35.73 44.41 0.24 1.33
16 S2_C60_A80_1 400 60 80 0.30 36.28 24.62 0.32 0.48 0.62
S2_C60_A80_2 0.50 37.70 39.14 0.04 1.03
S2_C60_A80_3 0.70 39.08 53.50 0.37 1.45
17 S2_C60_A60_1 400 60 60 0.40 36.61 24.81 0.32 0.65 0.62
S2_C60_A60_2 0.60 37.76 35.72 0.05 0.93
S2_C60_A60_3 0.80 38.97 46.62 0.20 1.24
18 S2_C60_A40_1 400 60 40 0.70 37.52 28.57 0.24 0.99 0.71
S2_C60_A40_2 1.00 38.95 39.55 0.02 1.01
S2_C60_A40_3 1.30 40.36 50.56 0.25 1.32

As an example, the M - m diagrams of the beams S1_C45_A60_1
and S1_C45_A60_3 are reported in Fig. 9a and Fig. 9b, respectively.
Two stationary points, corresponding to the effective cracking
moment (Mcr*) and the ultimate bending moment (Mu), are clearly
evident in both the figures. Fig. 9a shows the so-called deflection-
softening response (i.e., Mu < Mcr* and the FRC beam is under-
reinforced), whereas the M - m diagram illustrated in Fig. 9b rep-
resents the typical deflection-hardening behavior (i.e., Mu > Mcr*
and the fiber-reinforcement is higher than Vf,min). The values of Mcr*
and Mu, taken on the M - m diagrams of the 54 ideal FRC beams, are
collected in Tables 1 and 2.
According to Fantilli et al. [10], the ductile behavior of FRC beams
corresponds to a positive value of the following ductility index (DI):
Mu  Mcr Pu  Pcr
DI ¼ ¼ (27)
Mcr Pcr
As DI < 0 in beams showing a deflection-softening, the mini-
mum amount of fibers Vf,min (or, equivalently, the minimum
146 A.P. Fantilli et al. / Cement and Concrete Composites 65 (2016) 139e149

Fig. 9. Application of the general model to beams of group 5 (see Table 1): (a) M - m diagram of the beam S1_C45_A60_1; (b) M - m diagram of the beam S1_C45_A60_3; (c) DI - Vf
relationship and definition of Vf,min.

required ductility) can be computed by imposing DI ¼ 0. attained. Thus, the values of Vf,min, detected for each group with the
Tables 1 and 2 report the values of DI calculated for the ideal intersection between the line DI - Vf and the horizontal axis (i.e.,
beams investigated herein. For each group of beams (e.g., those of DI ¼ 0), are reported in Table 1 (for the beams S1) and Table 2 (for
group 5 in Fig. 9c), a linear relationship between DI and Vf is the beams S2).

Table 3
Properties of FRC beams tested in some experimental campaigns.

Group Beam B (mm) H (mm) fc (MPa) Lf/f Vf (%) Fiber References

I N2 100 100 97.4 50 0.50 F_S Banthia and Gupta [15]

N3 92.6 0.75
N4 86.6 1.00
II F80/60_Cf10 150 125a 33.0b 80 0.13 H_S Barros et al. [12]
F80/60_Cf20 0.25
F80/60_Cf30 0.38
III Mix_1_1 1000 100 21.5 113 0.36 S_P Fantilli et al. [10]
Mix_1_2
Mix_1_3
Mix_2_1 23.4 0.74
Mix_2_2
Mix_2_3
Mix_3_1 22.9 1.10
Mix_3_2
Mix_3_3
IV 50C(40) 100 50 54.0b 60 0.51 H_S Jones et al. [13]
50C(80) 1.02
V 75C(40) 100 75 54.0b 60 0.51 H_S
75C(80) 1.02
a b
VI C25/30(25) 150 125 33.0 65 0.32 H_S Lee and Barr [14]
C25/30(75) 0.96
VII 0.20(A) 150 150 50.2b 60 0.20 U_S Michels et al. [16]
0.52(A) 48.7b 0.52
0.65(A) 51.7b 0.65
0.91(A) 52.7b 0.91
VIII S13-HL-28d 150 150 28 167 0.17 H_S Mobasher et al. [17]
S26-HL-28d 0.33
S39-HL-28d 0.50
IX 4P-LN-20 150 150 42.7b 80 0.25 H_S Soetens and Matthys [18]
4P-LN-40 48.8b 0.51
X DWP_0.2 100 100 e 150 0.20 S_P Wu [19]
DWP_0.5 0.50
DWP_1 1.00
DWP_1.5 1.50
XI FP_0.2 100 100 e 89 0.20 S_P
FP_0.5 0.50
FP_1 1.00
FP_1.5 1.50
a
Depth of the ligament of a notched beam.
b
Estimated mean cylindrical compressive strength.
 A.P. Fantilli et al. / Cement and Concrete Composites 65 (2016) 139e149
If the normalized reinforcement ratio v ¼ Vf/Vf,min is introduced
(Fig. 10a), the existence of a linear function DI - v can be argued and
the results can be extended to all the FRC beams. This line passes
through the point corresponding to Vf,min (i.e., v ¼ 1, DI ¼ 0) and has
a well-defined slope. By applying the least square approximation to
all the couples of DI and v values, previously computed for the 54
ideal FRC beams, this slope is computed and results equal to 0.7 (see
Fig. 10a). In other words, for all the groups of FRC beams, the
evaluation of the ductility index can be performed by means of the
following general equation:
DI ¼ 0:7$ðv  1Þ
4. Experimental results compared with the predictions of the
ductility index
To verify the accuracy of the proposed linear model, the pre-
dicted values of DI [i.e., Eq. (28)] are compared with those measured
in 11 groups of beams in bending (whose B, H, fc and the fiber aspect
ratio are reported in the first columns of Table 3). In a single group,
at least two FRC beams, having the same geometrical and material
properties and without steel reinforcing bars, are taken into
consideration. The beams have different amounts of fiber-
reinforcement Vf (see column #7 of Table 3), here indicated in
terms of volume fraction. The selected amounts of fibers are limited
to those that produce, unequivocally, the failure in tension. Indeed,
when high fiber volume fractions are used, the strains detected in
the compressed zones of a beam cross-section (Fig. 6c) can localize
(Uchida et al. [11]). In such situations, the ascending branch of the
Sargin's parabola depicted in Fig. 7 (fib [4]), and used to obtain Eq.
(28), is no longer representative of the FRC behavior. Furthermore,
column #8 of Table 3 reports two letters that define, respectively,
the geometry (S ¼ straight, H ¼ with hooked ends, F ¼ with flat
ends, U ¼ undulated) and the material (S ¼ steel, P ¼ plastic) of the
fibers used to reinforce the beams.
The 11 groups considered in Table 3, in which the FRC beams are
labeled with the original names given by the Authors, comprise
both three point bending (Barros et al. [12], Fantilli et al. [10], Jones
et al. [13], Lee and Barr [14]) and four point bending tests (Banthia
and Gupta [15], Michels et al. [16], Mobasher et al. [17], Soetens and
Matthys [18], Wu [19]).
The values of Pcr* and Pu, experimentally measured for all the
Fig. 10. The computation of DI with the proposed relationship [Eq. (28)]: (a) comparison with the results of the general model (Tables 1 and 2); (b) comparison with the
experimental data (Tables 3 and 4).
147
beams, are collected in Table 4. According to Eq. (27), these loads
provide the ductility index, which is reported in the same
Table together with the values of Vf,min. The latter is evaluated for each
group as indicated in Fig. 9c. The values of DI and of the normalized
reinforcement ratio v ¼ Vf/Vf,min, both referred to a single beam, define
a point in the diagram of Fig. 10b. In the same picture, the proposed
linear function, as defined by Eq. (28), is also reported and compared
with the experimental data. Although the unavoidable dispersion of
the results, the linear relationship between DI and v is significantly
confirmed by the tests. Indeed, Fig. 10b shows a good agreement
between the computation of DI [with Eq. (28)] and the experimental
(28) data collected in Tables 3 and 4.
Moreover, if the non-dimensional parameters DI and v are
introduced, the hypothesis used to calculate the pullout response
(i.e., fiber symmetrically and orthogonally positioned with respect
to the crack, Lf assumed as the characteristic length of the FRC in
tension, and a single bond-slip relationship for all the types of the
fiber) seem to be irrelevant to assess the brittle/ductile behavior of
FRC beams. Specifically, this is particularly true for the fiber
orientation, which affects in the same manner both the values of
Pcr* and Pu in Eq. (27). For this reason, it does not influence the
values of DI, as confirmed by the tests performed on different FRC
beams (see Table 4) and, therefore, with different fiber orientations.
From a practical point of view, a simple-to-apply procedure,
requiring the use of Eq. (28) and a single test on a full-scale spec-
imen (i.e., a single FRC beam), can provide Vf,min. Indeed, from the
ductility index measured in the test (i.e., DI1 in Fig. 10b), the cor-
responding value of the normalized reinforcement ratio v1 can be
obtained through Eq. (28) (or graphically in Fig. 10b). Then the
calculation of the minimum content of fibers is possible by using
the inverse formula Vf,min ¼ Vf/v1 (where Vf is the amount of fibers
in the tested beam). Thus, the proposed approach, which is a sort of
compromise between (complex and not generalizable) theoretical
methods and (onerous) experimental campaigns, introduces a
more practical and user-friendly tool for the evaluation of the
minimum content of fibers, in order to have a deflection-hardening
behavior of FRC beams.
5. Conclusions
According to the analyses previously described, the following
conclusions can be drawn:
148 A.P. Fantilli et al. / Cement and Concrete Composites 65 (2016) 139e149

Table 4
Evaluation of the ductility index, and of the minimum amount of fiber, in the FRC beams tested in some experimental campaigns.

Group Beam Pcr*(kN) Pu (kN) DI Vf,min (%) v References

I N2 25.0 17.0 0.32 1.13 0.44 Banthia and Gupta [15]

N3 27.5 22.8 0.17 0.66
N4 21.1 19.5 0.07 0.88
II F80/60_Cf10 14.4 5.4 0.62 0.46 0.28 Barros et al. [12]
F80/60_Cf20 15.9 8.3 0.48 0.55
F80/60_Cf30 13.6 12.0 0.12 0.83
III Mix_1_1 18.2 11.5 0.37 1.02 0.36 Fantilli et al. [10]
Mix_1_2 18.6 12.4 0.33
Mix_1_3 20.6 13.2 0.36
Mix_2_1 21.3 18.8 0.12 0.72
Mix_2_2 22.5 15.6 0.31
Mix_2_3 23.1 19.5 0.16
Mix_3_1 22.8 24.0 0.05 1.08
Mix_3_2 21.5 23.2 0.08
Mix_3_3 20.4 21.6 0.06
IV 50C(40) 2.3a 1.9a 0.20 0.73 0.70 Jones et al. [13]
50C(80) 2.9a 3.6a 0.26 1.39
V 75C(40) 6.5a 5.0a 0.24 0.74 0.69
75C(80) 6.9a 8.9a 0.29 1.38
VI C25/30(25) 13.8 7.4 0.46 0.73 0.44 Lee and Barr [14]
C25/30(75) 16.2 20.4 0.26 1.31
VII 0.20(A) 25.8 12.1 0.53 0.64 0.31 Michels et al. [16]
0.52(A) 28.2 29.2 0.03 0.81
0.65(A) 29.0 29.8 0.03 1.01
0.91(A) 31.6 37.3 0.18 1.42
VIII S13-HL-28d 23.3 5.8 0.75 0.56 0.30 Mobasher et al. [17]
S26-HL-28d 26.1 15.2 0.42 0.59
S39-HL-28d 25.1 21.9 0.13 0.89
IX 4P-LN-20 3.6a 3.3a 0.06 0.31 0.83 Soetens and Matthys [18]
4P-LN-40 4.8a 5.9a 0.23 1.66
X DWP_0.2 12.0a 3.9 0.67 2.05 0.10 Wu [19]
DWP_0.5 5.3 0.56 0.24
DWP_1 8.1 0.33 0.49
DWP_1.5 9.3 0.23 0.73
XI FP_0.2 13.6a 4.0 0.71 2.89 0.07
FP_0.5 5.7 0.58 0.17
FP_1 6.9 0.49 0.35
FP_1.5 8.8 0.36 0.52
a
Value estimated from the experimental curves.

1. When the ultimate bending moment equates the effective References

cracking moment, which corresponds to the transition from the
brittle to the ductile behavior, the minimum amount of fibers
(Vf,min) can be defined for FRC beams. Such content of fibers,
having the same mechanical function of the minimum area of
steel rebars in reinforced concrete beams, can also be obtained
by imposing the ductility index DI equal to zero.
2. The values of DI linearly increase with the normalized rein-
forcement ratio v ¼ Vf/Vf,min, regardless of the geometrical and
mechanical properties of materials. Both numerical approaches
and experimental results seem to confirm the existence of the
proposed linear relationship [Eq. (28)].
3. With respect to the current approaches, Eq. (28), accompanied
by a single test on a full-scale FRC beam in bending, defines a
more effective and user-friendly tool for the evaluation of Vf,min.
Finally, further studies should be developed to introduce DI in
FRC beams containing also steel rebars, in order to identify a
possible relationship between the ductility index and the amount
of reinforcement (fibers and rebars) in hybrid structures.
Acknowledgment
The grant given by the Italian Laboratories University Network
of seismic engineering (ReLUIS), and used to develop this research
work, is gratefully acknowledged.
[1] A.P. Fantilli, H. Mihashi, T. Nishiwaki, Tailoring hybrid strain-hardening
cementitious composites, ACI Mater. J. 111 (2) (2014) 211e218.
[2] A.E. Naaman, Strain hardening and deflection hardening fiber reinforced
cement composites, in: Proceedings of the International Workshop High
Performance Fiber Reinforced Cement Composites, RILEM Publications, 2003,
pp. 95e113. Pro. 30.
[3] A.P. Fantilli, D. Ferretti, I. Iori, P. Vallini, Behaviour of R/C elements in bending
and tension: the problem of minimum reinforcement ratio, Eur. Struct. Integr.
Soc. 24 (1999) 99e125.
de
[4] fib - Fe ration internationale du be ton, Model Code 2010-Final Draft, Volume
1, Switzerland, Lausanne, 2012 fib Bulletin, 65.
[5] A. Caratelli, A. Meda, Z. Rinaldi, Design according to MC2010 of a fibre-
reinforced concrete tunnel in Monte Lirio, Panama, Struct. Concr. 13 (2012)
166e173.
[6] A. de la Fuente, P. Pujadas, A. Blanco, A. Aguado, Experiences in Barcelona with
the use of fibres in segmental linings, Tunn. Undergr. Space Technol. 27 (2012)
60e71.
[7] B. Chiaia, A.P. Fantilli, P. Vallini, Evaluation of crack width in FRC structures
and application to tunnel linings, Mater. Struct. 42 (2009) 339e351.
[8] A.P. Fantilli, P. Vallini, Bond-slip relationship for smooth steel reinforcement,
in: N. Bicanic, R. de Borst, H. Mang, G. Meschke (Eds.), Computational
Modelling of Concrete Structures (EURO-C 2003), St. Johann Im Pongau, March
17e20th, 2003, A.A. Balkema Publishers, Lisse, 2003, pp. 215e224.
[9] B. Chiaia, A.P. Fantilli, P. Vallini, Evaluation of minimum reinforcement ratio in
FRC members and application to tunnel linings, Mater. Struct. 40 (2007)
593e604.
[10] A.P. Fantilli, A. Gorino, B. Chiaia, Precast plates made with lightweight fiber-
reinforced concrete, in: Proceedings of the FRC 2014 Joint ACI-fib Interna-
tional Workshop, Montreal, July 24-25th, 2014, 2014, pp. 224e234.
[11] Y. Uchida, M. Kawai, K. Rokugo, Back analysis of tensile stress-strain rela-
tionship of HPFRCC, in: Proceedings of the Fifth International RILEM Work-
shop on High Performance Fiber Reinforced Cement Composites, Mainz, July
10-13th, 2007, 2007, pp. 49e56.
 A.P. Fantilli et al. / Cement and Concrete Composites 65 (2016) 139e149
[12] J.A.O. Barros, V.M.C.F. Cunha, A.F. Ribeiro, J.A.B. Antunes, Post-cracking
behaviour of steel fibre reinforced concrete, Mater. Struct. 38 (2005) 47e56.
[13] P.A. Jones, S.A. Austin, P.J. Robins, Predicting the flexural load-deflection
response of steel fibre reinforced concrete from strain, crack-width, fibre
pull-out and distribution data, Mater. Struct. 41 (2008) 449e463.
[14] M.K. Lee, B.I.G. Barr, A four-exponential model to describe the behaviour of
fibre reinforced concrete, Mater. Struct. 37 (2004) 464e471.
[15] N. Banthia, R. Gupta, Hybrid fiber reinforced concrete (HyFRC): fiber synergy
in high strength matrices, Mater. Struct. 37 (2004) 707e716.
[16] J. Michels, R. Christen, D. Waldmann, Experimental and numerical investiga-
tion on postcracking behavior of steel fiber reinforced concrete, Eng. Fract.
Mech. 98 (2013) 326e349.
[17] B. Mobasher, M. Bakhshi, C. Barsby, Backcalculation of residual tensile
strength of regular and high performance fiber reinforced concrete from
flexural tests, Constr. Build. Mater. 70 (2014) 243e253.
[18] T. Soetens, S. Matthys, Different methods to model the post-cracking behav-
iour of hooked-end steel fibre reinforced concrete, Constr. Build. Mater. 73
(2014) 458e471.
[19] Y. Wu, Flexural strength and behavior of polypropylene fiber reinforced
concrete beams, J. Wuhan Univ. Technol. 17 (2) (2002) 54e57.
Nomenclature
Ac: area of the cementitious matrix of an ideal tie (Fig. 2a)
Af: area of the fiber cross-section
B: width of a FRC beam cross-section
DI: ductility index
Ec: Young's modulus of concrete in tension and compression
Ec1: secant modulus of elasticity from the origin of the stress-strain diagram to the
peak of compressive stress of concrete
Ef: Young's modulus of fiber
F: axial load applied to a FRC tie (Fig. 1a)
fc: compressive strength of concrete
Fcr: cracking load of a FRC tie (Fig. 1a)
fct: tensile strength of concrete
Fu: axial load at the strain localization in a FRC tie (Fig. 1a)
fu: tensile strength of fiber
GF: fracture energy of concrete in tension
H: depth of a FRC beam cross-section
i: subscript referred to the generic cross-section of an ideal tie (Fig. 2a)
j: subscript referred to the generic strip of a FRC beam cross-section (Fig. 6b)
k: Ec / Ec1 ¼ plasticity number of FRC in compression [Eq. (25.a)]
Lf: length of the fiber
ltr: transfer length of an ideal tie (Fig. 2a)
M: bending moment associated to the generic curvature m;
149
m: number of strips in a FRC beam cross-section
Mcr: bending moment at first cracking
Mcr*: bending moment at the effective cracking
Mu: bending moment at the strain localization
N: axial load associated to the crack width w of an ideal tie (Fig. 2a)
n: Ef /Ec ¼ ratio between the Young's moduli of the fiber and of the cementitious
matrix
P: load applied to a FRC beam in three point bending (Fig. 1b)
Pcr: load corresponding to the first cracking of a FRC beam in three point bending
(Fig. 1b)
Pcr*: load corresponding to the effective cracking of a FRC beam in three point
bending (Fig. 1b)
pf: perimeter of a fiber cross-section
Pu: load corresponding to the strain localization of a FRC beam in three point
bending (Fig. 1b)
R: resultant of stresses in a FRC cross-section
s, s1: slip between fiber and surrounding concrete, and the value at the peak in the
bond-slip model [Eq. (13)]
s0: value of s in the cracked cross-section of an ideal tie (Fig. 2b)
v: normalized reinforcement ratio of a FRC beam
Vf, Vf,min: fiber volume fraction of a FRC beam and its minimum value
w, w1, wc: crack width, and limit values of w in the fictitious crack model [Eq. (12)]
y: vertical coordinate of a FRC cross-section (Fig. 6b)
z: horizontal coordinate of an ideal tie (Fig. 2a)
a: exponent in the bond-slip model [Eq. (13.a)]
b: coefficient in the bond-slip model [Eq. (13.b)]
d: midspan deflection of a FRC beam in bending (Fig. 1b)
Dh: H / m ¼ depth of the strips of a FRC beam cross-section (Fig. 6b)
DL: elongation of a FRC tie (Fig. 1a)
Dl: part of the transfer length
εc1: FRC strain at the peak of compressive stress (Fig. 7)
εF: compressive or tensile strain of FRC (Fig. 7)
εF1: FRC strain at the peak of tensile stress (Fig. 7)
h: j εF / εc1 j ¼ normalized strain of the FRC in compression [Eq. (25.a)]
l: parameter of a FRC cross-sectional strain profile
m: curvature of a FRC cross-section
sc: concrete stress of an ideal tie (Fig. 2c)
sc,I: concrete stress in the Stage I cross-section of an ideal tie (Fig. 2c)
sc0: concrete stress in the cracked cross-section of an ideal tie (Fig. 2c)
sF: compressive or tensile stress of FRC (Fig. 7)
sf: stress in the fiber of an ideal tie (Fig. 2c)
sf,I: stress of the fiber in the Stage I cross-section of an ideal tie (Fig. 2c)
sf0: stress of the fiber in the cracked cross-section of an ideal tie (Fig. 2c)
t: bond stress
tmax, tf ¼ maximum and residual values of t in the bond-slip model [Eq. (13)]:
f: diameter of the fiber cross-section.