Engineering Fracture Mechanics 74 (2007) 210–222

www.elsevier.com/locate/engfracmech

Controlling fracture toughness of matrix with mica flake

inclusions to design pseudo-ductile fibre
reinforced cementitious composites
a,*
H. Mihashi , João P. de Barros Leite a, S. Yamakoshi a, A. Kawamata b

a
Department of Architecture & Building Science, Tohoku University, Aoba 6, Sendai 980-8579, Japan
b
Institute of Technology, Tekken Corporation, Japan

Available online 9 March 2006

Abstract

High-performance fibre reinforced cementitious composites (FRCC) are usually designed by using high-performance
fibres and matrix of high bond strength to enhance fibre debonding mechanism. The crack formation in high-strength
matrices results in considerable strain-energy release. In such a case, fibres may not be able to effectively bridge the initial
crack and failure due to propagation of the crack is relatively brittle. An approach is proposed in which artificial micro-
cracks are smeared in the matrix to optimize fracture toughness while preserving sufficient bond strength. Mica flakes were
introduced into the composite mixture as source of microcracks, and showed promising capabilities for controlling duc-
tility of the composite material. Key parameters to control ductility were identified and procedures for mixture optimiza-
tion are under development.
Ó 2006 Elsevier Ltd. All rights reserved.

Keywords: Fracture toughness; Ductility; Material optimization; Fibre–matrix bond; Micro-mechanics

1. Introduction

Constant increases in size of concrete members lead to advances in concrete technology particularly focus-
ing on developing high-performance concrete, which possesses high compressive strength and durability. Yet,
concrete has two main detrimental properties as structural material, namely low tensile strength and poor frac-
ture toughness. Gains in compressive strength have usually resulted in even lower strain capacity and resis-
tance to crack propagation.
Recent concerns in concrete design are shifting towards safety and durability of concrete structures. Con-
sequently, concrete technologies have branched into design of pseudo-ductile materials. The use of fibres in the
composite matrix emerged as a strategy to partially offset the low strain capacity of concrete. Though fibres

*
Corresponding author. Tel.: +81 22 2177864; fax: +81 22 2177886.
E-mail address: mihashi@timos.str.archi.tohoku.ac.jp (H. Mihashi).

0013-7944/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engfracmech.2006.01.016
 H. Mihashi et al. / Engineering Fracture Mechanics 74 (2007) 210–222 211

improve general post-peak mechanical behaviour, they were introduced mainly to arrest propagation of
cracks. The presence of fibres results in minimal increase in stiffness prior to cracking and holds the matrix
together after cracking. During initial stage of loading, the fibre–matrix interaction is elastic, with stress trans-
fer occurring through shear at the interface. As the load increases, shear may gradually destroy the chemical
bond along fibre and matrix interface. Debonding activates a combination of elastic and frictional stress trans-
fer mechanisms. Hence, the fibre keeps bridging the crack until one of its ends is completely extracted from the
matrix or the fibre breaks. Pullout of the fibre is preferred as a toughening mechanism, since it consumes much
more energy and prevents catastrophic failure. The role of toughening mechanisms is to consume energy, thus
increasing the total energy required for fracture. Evans [1] distinguished toughening mechanisms for ceramic
materials according to overall mechanical behaviour into two categories: process zone formation and bridging.
Fibre reinforcement falls into the latter category. Yet for performance reasons, due to current advances in
cement and fibre technology resulting in considerable differences in matrix–fibre interactions, fibre pullout
may be better regarded as a complex two-phase toughening mechanism. The first phase may be described
as the work done in destroying the initial (chemical) bond between the matrix and fibre and the second phase
as the work done by sliding friction (frictional bond) while the fibre is extracted from the matrix. Several
approaches have been proposed to improve the toughening mechanism of fibre reinforcement, in which the
key element was to increase the bond strength and friction at fibre–matrix interface.
Developing other energy consuming mechanisms in the composite matrix may further increase the material
toughness. A third mechanism to be considered is called crack deflection, in which toughness is elevated by
increasing the area of the fracture surface. Some recently proposed approaches to optimize efficiency of tough-
ening mechanisms include tailoring of the matrix composition and particle size [2,3]. Complexity arises in the
tailoring of cost-effective composite, when one takes into consideration the wide range of mechanical, physical
and chemical properties of fibres, as well as properties of different particulate matrices. The resulting fibre rein-
forced cementitious composites (FRCC) may display somehow distinct mechanical behaviour and failure
mechanisms. Appropriate choice will depend on application but in any case it will result in an individual opti-
mization problem, in which the volume fraction, size and distribution of each selected components in the mix-
ture, are the design parameters.
This study investigates an alternative toughening mechanism based on ‘‘microcrack formation’’. The basic
principle is the homogenization of the matrix through the uniform distribution of mica flakes as source of fic-
titious microcracks. Such uniform distribution of microcracks encourages the development of multiple con-
current cracks, which will delay the localization of failure. Hence by controlling size and content of the
artificial microcracks in order to maintain stable crack propagation, additional energy may be consumed in
developing a number of concurrent crack surfaces. Note that the intended toughening mechanism here is
not the microcrack toughening with formation of a fracture process zone (FPZ), as usually described in the
literature [4,5]. Although the development of a field of disconnected microcracks or microvoids may relax
crack-tip triaxiality and diffuse the intensity of crack-tip stress singularity in brittle matrices, significant energy
consumption in FRCC may only be obtained with activation of pullout mechanisms in larger number of
fibres, and to produce pseudo-ductile behaviour.

2. Experimental procedure

Series of three-point bending tests were performed on HCP (hardened cement paste) or FRCC beam spec-
imens as shown in Fig. 1. Tests were carried out using an INSTRON-5567 Universal Materials Testing
Machine, at a loading speed of 0.2 mm/min, with measurement of load–CMOD (crack mouth opening dis-
placement) by means of a clip-gauge.
A first round of tests was performed to evaluate the influence of different parameters into the fracture
toughness of HCP specimens, whose components also constitute the matrix of the FRCC specimens. The frac-
ture toughness parameter (JIC) was estimated for quantitative evaluation of the fracture toughness. In the
tests, different contents of mica flakes of three distinct Fuller distributions were added to the HCP composition
to investigate influence of mica flakes on properties of the matrix. In the second round of tests, the water–
binder ratio (W/B) as well as the content and size of mica flakes were varied in the composition of FRCC spec-
imens, to investigate the effects of these changes on overall material properties.
212 H. Mihashi et al. / Engineering Fracture Mechanics 74 (2007) 210–222

Fig. 1. Setup and detailing of the three-point bending tests.

2.1. Materials and mix proportions

The mix proportions for HCP and FRCC specimens are presented in Table 1. The composite binder (B)
constitutes of early strength Portland cement (C) and silica fume (SF). Since a superplasticizer (SP) is required
to attain workability of the FRCC, the same amount of SP was added to the HCP specimens in order to match
the composition of the matrix of FRCC specimens.
Three different particle size grading of fine mica flakes (MF) were employed alternatively in the composi-
tions to investigate the effect of different sizes of microcracks. Mica is the generic name for a group of complex
aluminosilicate minerals having laminate structure with different chemical compositions and physical proper-
ties. It has nearly perfect basal cleavage in the direction of the large surfaces, which permits easy splitting into
optically flat films, as thin as one micron in thickness. After splitting, the thin films remain tough and elastic
even at high temperature. Since the mica flakes were not intended to reinforce the matrix in any sense, a phlog-
opite mica flake was preferred, because its tensile strength (95 MPa) is about half of the strength of the other
type (muscovite). Mica exhibits a high elastic modulus ranging from 135 GPa to 210 GPa and density around
2.8 g/cm3. The Fuller distributions for the three MF grading employed in the experiments, identified by their
respective average sizes (55, 160 and 620 lm), are presented in Table 2.
As fibre reinforcement, a special type of polyethylene fibre (DyneemaÒ SK-60) developed by Toyobo Co.,
was employed. The fibre is produced out of ultra-high-molecular weight polyethylene (UHMWPE) and exhib-
its high tensile strength (2.77 GPa), high elastic modulus (88 GPa), high abrasion resistance and low density
(0.97 g/cm3). Geometrical properties of the fibres are 12 lm of diameter and 30 mm of length. Preliminary
pullout tests [6] show that the chemical bond between this fibre and matrix is apparently very low or ineffec-
tive. Frictional resistance at interfaces between fibres and matrices with normal to high W/B seems signifi-
cantly strong. However, for low W/B the frictional bond seems substantially reduced (see Fig. 2).

Table 1
Mix proportions of testing specimens
Proportions HCP FRCC
C:SF 79.3:20.7 (vol.%)
SP/B 2.0 (wt.%)
W/B 30, 40, 50 (wt.%)
MF/B 0.0, 1.0, 2.5, 5.0 (wt.%)
PE 0.0 1.5 (vol.%)
 H. Mihashi et al. / Engineering Fracture Mechanics 74 (2007) 210–222 213

Table 2
Fuller distribution of mica flakes
Mesh (lm) AS-55 AS-160 AS-620
850 – 0.2 5.0
425 – 32.0 75.0
212 0.8 77.0 97.0
106 10.0 95.0 #
63 40.0 #
45 60.0
Accumulated 100.0 100.0 100.0

4
Frictional Bond Strength (MPa)

3.5

2.5

1.5

1
PVA
0.5 Polyethylene
Aramid
0
20 30 40 50 60 70
W/B (%)

Fig. 2. Comparison between frictional bonds of three different fibers (PVA, Aramid and the SK-60 polyethylene fibers).

2.2. Experimental analysis method

In the first series of three-point bending tests on HPC (matrix) specimens, the fracture toughness KIC was
determined from the maximum bending load (Pmax) using Eq. (1) proposed by Srawley [7]
P max s
K IC ¼ fb ; ð1Þ
bd 3=2
where s is the span between support reactions; b and d are the thickness and depth of the beam, respectively;
and fb is the geometry correction for bending load, given as function of the ratio (a) between notch size and
depth of the beam
3a1=2 ½1:99  að1  aÞð2:15  3:93a þ 2:7a2 Þ
fb ¼ . ð2Þ
2ð1 þ 2aÞð1  aÞ3=2
Since HCP displays considerably brittle behaviour, linear elastic fracture mechanics (LEFM) may be assumed,
i.e. the critical value of the J integral is roughly the value of the critical energy release rate for linear elastic
material (JIC  GIC). Hence, by introducing the value of KIC obtained from Eq. (1) into Eq. (3), where E is
the elastic modulus of the HCP specimen, JIC may be readily calculated as follows:
K 2IC
J IC ¼  GIC . ð3Þ
E
For comparative assessment of the ductile behaviour of HCP specimens, the value of JIC estimated by Eq. (3)
is practical and sufficiently accurate. For the FRCC specimens, however, LEFM is no longer applicable and
another parameter for assessment needs to be introduced. Conventional ductility index such as du/dy (defor-
mation at failure/deformation at yielding), used for evaluating the ductile behaviour of metals and/or RC
214 H. Mihashi et al. / Engineering Fracture Mechanics 74 (2007) 210–222

members, may be somehow cumbersome to employ for designing ductile FRCC. The load–displacement
curves of FRCC usually present two peaks of similar magnitude—one at very low deformation (due to strong
chemical bond and/or high-strength matrix) other at substantially larger deformation (due to strong frictional
bond)—often separated by deep valleys as shown in Fig. 3. The apparent ‘‘yielding point’’ on the pseudo-
ductile behaviour of FRCC is not so clearly defined as in metals. Unlike in metals, the plastic deformation
in FRCC is strongly localised and mainly concentrated in crack openings. Therefore, gains in ductility imply
certain cost in terms of crack development. The pseudo-ductile behaviour should be attained through distrib-
uted fine cracking and attention should be paid to enforce limits of crack opening in design. Consequently, it
seems more appropriate to relate performance measurements with crack opening displacements rather than
with global deformations. Difficulties also arise to employ existent energy ductility indices such as Uu/Up
(energy at failure/energy at peak load) or Ucp/Uci (energy to crack propagation/energy to crack initiation).
For evaluating ductile behaviour of materials, deformation energy obtained from simple measurements of
areas is relatively insensitive to the shape of the load–displacement curves, as shown in Fig. 4. Hence, the
present study suggests an alternative index of ductility (ID), which is related to the shape of the load–CMOD
diagram for the target design and is relatively easy to be calculated.
The target design used as reference for calculations of the index of ductility (ID) was idealised as rigid–
perfectly-plastic material, with ID defined as the ratio between the area below the curve of the load–CMOD
diagram of the FRCC and the area of an idealised rigid–perfectly plastic diagram with same displacement and
peak load. Note that for FRCC an elastic–perfect-plastic material is a more realistic design to be expected.
However, since the elastic deformation may be expected to be very small compared to plastic deformation,
the rigid–perfect-plastic diagram suits better for the design, in the sense that it penalises design approaches,
which result in great loss of stiffness. Note also that due to limitations on the clip-gauge capacity,

High CB CB = Chemical Bond

High FB FB = Frictional Bond
Load

CMOD
CB FB

Fig. 3. Typical load–CMOD curves of FRCC with different matrix and bond strength characteristics.

P P P
2 2 1

Aδe = 2
δ
1 2 3

.5
Aδe = 2
Aδe = 2
δ
1 2 1 2 3 4 5

Fig. 4. Hypothetical load–CMOD curves exhibiting identical value of fracture energy but different shapes.
 H. Mihashi et al. / Engineering Fracture Mechanics 74 (2007) 210–222 215

Aδe(5mm) – Real Material

2.5
Aδe(5mm) – Target Design
2.0
ID = (Aδe R.M.) / (Aδe T.D.)

Applied Load (kN)

Pmax
1.5

1.0

0.5

0.0
0 1 2 3 4 5
CMOD (mm)

Fig. 5. Method for determination of the index of ductility ID of FRCC specimens.

load–CMOD curves as well as measurements were stopped at the limit value of 5 mm CMOD. Nevertheless,
ductility beyond the 5 mm CMOD would not have much practical significance since the material integrity
would be severely compromised by cracking.
Fig. 5 shows an example of computation of the index of ductility ID, as the ratio between the area below the
load–CMOD curve of a tested FRCC specimen, up to the limit displacement of 5 mm, and the area of the
diagram of a target design, i.e. the maximum peak load times the limit displacement of 5 mm.

3. Results and discussion

Fig. 6 presents curves showing the influence of fine mica flakes on values of GIC of matrices (HCP). In the
different mix compositions of the matrices, the water–binder ratio (30%, 35% and 40%) was varied, as well as
the content (0%, 1%, 2.5% and 5%) of mica flakes. All curves show a clear inflection point, which suggests that
mica flake at low contents may induce localised failure by reducing the fracture toughness of the material.

12

10

8
JIC (N/m)

2 30c
35c
40c
0
0 1 2 3 4 5 6
Mica Flake Content (wt. %)

Fig. 6. Curves describing the behaviour of JIC vs. mica flake content for HPC specimens with three different W/B.
216 H. Mihashi et al. / Engineering Fracture Mechanics 74 (2007) 210–222

However, as the content increases they may also act somehow as reinforcement or as other type of crack resis-
tant mechanism. The results also show that the W/B exerts strong influence in the fracture toughness of the
HCP. This may perhaps be explained by the growth of the porous structure as a result of the increase in
W/B, and associated reduction of the matrix strength. Yet such increase in porous media apparently does
not produce any detrimental effect on the frictional bond strength of the polyethylene fibre, confirming results
from fibre pullout tests by Kiyota et al. [6].
For the experimental results presented in Fig. 7, the W/B was fixed at 35% and the influence of the mica
flake size on the fracture toughness of the matrix (HCP) was investigated. The curves show that the finer
flakes, at low contents, seem to act as localised imperfections while their short lengths allow them to neither
provide any sort of reinforcement to the matrix nor promote substantial crack deflection. The coarser flakes
seem capable to provide some sort of strengthening and perform better at low contents. Yet at high contents,
the finer flakes seem to perform considerably more effectively in toughening the matrix. One explanation for
this may be that different toughening mechanisms result from finer and coarse flakes in the mix, e.g. the for-
mation of FPZ by very fine flakes and the crack bridging or deflection by larger flakes.
Fig. 8 shows the effect of two different mica flake size distributions, types ‘‘a’’ (AS-620) and ‘‘b’’ (AS-160),
on load–CMOD curves from bending tests of FRCC specimens. The use of mica flakes in the matrix of the
FRCC, in most cases, seems to produce certain improvement in ductility, when comparing the index of duc-
tility to that for the FRCC without mica flakes (ID = 0.591). Yet the mica flake type ‘‘b’’ produced very little
effect (average ID = 0.611). The values of ID, which are displayed in Fig. 8, were given in general by the aver-
age of three curves. However, for coarser size distributions of mica flake, the values of ID for the three curves
may vary significantly from the average value.
When finer mica flakes (type ‘‘c’’) were added to the FRCC matrix composition the indices of ductility are
similar in order to those of type a, but there are substantial differences with regard to the effect of the volume
content of mica flake. The coarser mica flakes (type ‘‘a’’) were capable of imparting positive gain in ductility to
the FRCC only when employed in low volume contents, yet with unstable post-peak behaviour. The finer mica
flakes showed higher capabilities for controlling steady post-peak behaviour and consequently higher poten-
tial for obtaining optimal ductility levels in different mix compositions.
The poor performance of the intermediate size (type ‘‘b’’) suggests that at the two extremes, coarse mica
flake (type ‘‘a’’) and fine mica flakes (type ‘‘c’’), the improved performance is due to completely different
toughening mechanisms for both cases. The toughening mechanism in coarse mica flake (type ‘‘a’’) seems
more complex, and consequently more difficult to be identified. The highly diffuse cracking pattern reveals
substantial branching and deflection. The failure mechanism exhibits often a cracking band along which

12

10

8
JIC (N/m)

2 35a
35b
35c
0
0 1 2 3 4 5 6
Mica Flake Content (wt. %)

Fig. 7. Curves describing the behaviour of JIC vs. content mica flake of different size distributions: ‘‘a’’ (AS-620), ‘‘b’’ (AS-160) and ‘‘c’’
(AS-55).
 H. Mihashi et al. / Engineering Fracture Mechanics 74 (2007) 210–222 217

2.5
W/B = 35% W/B = 35%
2.0 MF(a) = 1% MF(b) = 1%

1.5

P (kN)
1.0

0.5
I D = 0.693 I D = 0.595
0
0 1 2 3 4 5 0 1 2 3 4 5
CMOD (mm) CMOD (mm)

2.5
W/B = 35% W/B = 35%
2.0 MF(a) = 2.5% MF(b) = 2.5%

1.5
P (kN)

1.0

0.5
I D = 0.699 I D = 0.623
0
0 1 2 3 4 5 0 1 2 3 4 5
CMOD (mm) CMOD (mm)

2.5
W/B = 35% W/B = 35%
2.0 MF(a) = 5% MF(b) = 5%

1.5
P (kN)

1.0

0.5
I D = 0.657 I D = 0.615
0
0 1 2 3 4 5 0 1 2 3 4 5
CMOD (mm) CMOD (mm)

Fig. 8. Comparison between loads–CMOD curves of FRCC specimens with different content (1%, 2.5%, and 5%) and size distribution
(‘‘a’’ and ‘‘b’’) of mica flake.

the composite material disintegrates. Hence, it seems that the coarse mica flakes act as localised flaws inducing
cracking in alternative directions. Crack deflection may increase fracture path and surface, branching may
reduce stress level in debonding of competing cracks, rotations of the disaggregated material may increase
snubbing friction [8], and there is also a possibility that the mica flakes provide some slight reinforcement
to the matrix. However, none of these hypotheses could yet be quantified or proved as mechanisms responsible
for the extra toughness ductility imparted by the mica flakes.
The fine mica flake (type ‘‘c’’), on the other hand, seems not fit for crack bridging or crack deflection. The
anchorage length in most cases would be insufficient to provide effective reinforcement. Since the great major-
ity of these flakes are very fine, there may be also a possibility of formation of a fracture process zone in the
vicinity of the crack-tip, resulting in certain stress redistribution. That is more likely to occur at high contents
of mica flakes. Another hypothesis, which is under investigation through numerical modelling and simulation,
is the occurrence of a segmented debonding of the fibre due to the existence of microcracks (mica flakes) in the
vicinity of the fibre. Yet the major increase in toughness ductility in the case of fine mica flakes is believed to
218 H. Mihashi et al. / Engineering Fracture Mechanics 74 (2007) 210–222

Fig. 9. Load–CMOD curves of FRCC for different W/B and different mica flake contents in the FRCC mixture.
 H. Mihashi et al. / Engineering Fracture Mechanics 74 (2007) 210–222 219

arise from a certain homogenization of the matrix, which promotes the development of multiple and compet-
itive cracking. This could be clearly observed in the cracking pattern. Hence type ‘‘c’’ was preferred in this
initial investigation.
Fig. 9 presents load–CMOD curves from experiments using mica flake type ‘‘c’’, varying the content of
mica flake and the water–binder ratio, in the composition of the FRCC matrix. The load–CMOD curves
of FRCC without mica flake show clearly that the increase of W/B value results in gain in ductility and frac-
ture energy (GF). This could be naturally expected since the pullout tests in Fig. 2 shows substantial increase in
frictional bond with the increase of W/B within the tested range (30–40%). However, this may also favour the
hypothesis of microvoid or microcrack toughening in the case of FRCC. Note that such toughening mecha-
nism can only be effective in materials presenting stable microcrack growth. Pre-existing flaws in brittle mate-
rials remain stationary only shortly until they satisfy the Griffith criterion. Stable crack advance requires either
a rising fracture work with increases in crack extension, or physical barriers/restraints in the material that inhi-
bit crack growth. While stationary cracking is not observed in HCP specimens, it was clearly demonstrated by
the fine multiple cracking in FRCC. Apart from bridging cracks, networks of fibres may be effective to restrain
microcrack advances in small confined regions of the material.
The addition of mica flake into the matrix composition, while the W/B was kept constant, resulted in fur-
ther increase of ductility and fracture energy. The optimum value for the mica flake content may be dependent
on value of W/B, though a clear trend was not yet identified. It seems that a key element for the design of
ductile FRCC is to attain maximum energy consumption through fibre debonding. The reduction of matrix
strength allows stresses being drifted away from the initial crack by promoting concurrent fine cracks. Hence,
if rise in stresses can be relieved early in its development, the localised failure may be delayed. However, a
straightforward method to determine how weak the matrix should be, i.e. an optimum value of the matrix
fracture toughness, cannot be derived from the present experimental results.
Fig. 10 presents curves describing the ductility as function of the mica flake content for three W/B values.
Then for lower W/B values a point of inflection was present in the curves between 2.5% and 3.5% content of
mica flake, while for a W/B value of 40% there was no sign of inflection even with 5% content. It may be pos-
sible that with low W/B values the flowability of the mixture is affected and consequently the fibres may not be
uniformly distributed. This speculation remains to be experimentally investigated.
Fig. 11 shows good performance for mica flake types ‘‘a’’ and ‘‘c’’. This may indicate that few relative large
flaws (fictitious cracks), as well as large number of very fine micro-defects distributed in the matrix, may pro-
duce positive effects in the ductility of the FRCC. The crack pattern in Fig. 12 (left) shows some large flakes of
mica at the crack surface indicating that cracks have developed from the mica. In such case, for mica flake type

1.0

0.9
Index of Ductility (ID)

0.8

0.7

0.6

0.5
30c
35c
40c
0.4
0 1 2 3 4 5 6
Mica Flake Content (wt. %)

Fig. 10. Ductile behaviour of FRCC as a function of the content of mica flake and value of W/B.
220 H. Mihashi et al. / Engineering Fracture Mechanics 74 (2007) 210–222

1.0

0.9

Index of Ductility (ID)

0.8

0.7

0.6

0.5
35a
35b
35c
0.4
0 1 2 3 4 5 6
Mica Flake Content (wt. %)

Fig. 11. Ductile behaviour of FRCC as a function of the size and content of mica flake and fixed value of W/B.

‘‘a’’, considerable curving cracks were observed. The crack pattern on the right, for mica flake type ‘‘c’’, shows
a development of parallel cracks along the fracture surface. In both cases crack blunting can be expected. In
the case of mica flake type ‘‘b’’ at the centre of Fig. 12, little may be said but it seems that in this case there is a
higher trend for development of crack in areas of low fibre reinforcement content.

4. Numerical modelling

Although the experiments show clear evidence of the positive effect of incorporating mica flakes in the
design of ductile FRCC, the inherent mechanism is far from being understood and controlled. Changes in
W/B may affect the flowability, and consequently fibre distribution, orientation and bonding in the FRCC,
as well as the microstructure and strength of the matrix. Addition of mica flakes to the composition may sim-
ilarly be reflected by changes in various parameters governing fracture and also introduce additional tough-
ening mechanisms, such as crack arresting and blunting. For most cases, the parameters governing the fracture
behaviour cannot be isolated in experimental analyses. Numerical modelling and analyses are essential for the
material optimization. A discrete model for simulation fibre pullout has been developed to investigate the
influence of different mica flake size and content in debonding mechanics, in particular the hypothesis of a
segmented debonding. Thereafter, the results may be introduced into a mesolevel model for concrete fracture
simulation [9,10], whose generation mechanism is being adapted to generate a coarse porous media and mica
flakes into the HCP matrix. This second stage of numerical modelling aims to investigate the differences in the
mechanism of coarse and fine mica flakes; effects of fibre distribution and orientation; as well as optimum val-
ues of mica flake content for the design of ductile FRCC. Results of the numerical analyses are to be presented
in subsequent publications.

5. Concluding remarks

Crack formation is instinctively regarded as leading to failure because the remaining area of sound material
undergoes higher stress. However, this analysis applies mainly to a single crack; concurrent cracking may
relieve localised stresses and slow the rate of crack propagation. In similar way, assuming that stronger matrix
will result in stronger material may be proved wrong in the case of FRCC. The weakening of the matrix allows
easier development of concurrent cracks, and consequently, more energy consumption. The results presented
here show that the weakening of the matrix resulted in certain cases, even a little increase in the peak-load
value, as well as considerably higher fracture energy and ductility. It may of course depend on how ‘‘strength’’
of the matrix is measured; hence in this study strength was assessed in terms of resistance to fracture, or frac-
 H. Mihashi et al. / Engineering Fracture Mechanics 74 (2007) 210–222 221

Fig. 12. Typical crack opening in FRCC specimens with 5% content of mica flake of types a, b and c, respectively.

1.0

0.9
Index of Ductility (ID)

0.8

0.7

0.6

0.5

0.4
0 2 4 6 8 10 12
JIC (N/m)

Fig. 13. Relationship between ductility of FRCC and matrix fracture parameter JIC.

ture toughness. Fig. 13 shows strong evidence that the ductility of the FRCC increase with the decrease of frac-
ture toughness of the matrix. The fracture toughness of the matrix may depend considerably on the W/B ratio,
222 H. Mihashi et al. / Engineering Fracture Mechanics 74 (2007) 210–222

however the use of mica flake in the composition proved to be a promising approach to control fracture tough-
ness of different matrices. Optimum values of mica flakes or fracture toughness may depend on fibre
debonding characteristics and other parameters that can only be investigated with assistance of numerical
analyses and simulations.

References

[1] Evans AG. The new high toughness ceramics. ASTM STP 907. Philadelphia: American Society for Testing and Materials; 1989.
p. 267–91.
[2] Li VC. Engineered cementitious composites—Tailored composites through micromechanical modelling. In: Banthia N, Bentur A,
Mufti A, editors. Fibre reinforced concrete: present and future. Montreal: Canadian Society for Civil Engineering; 1998. p. 64–97.
[3] Li VC. On engineered cementitious composites (ECC): a review of the material and its applications. J Adv Concr Technol 2003;1(3):
215–30.
[4] Ritchie RO. Mechanisms of fatigue crack propagation in metals, ceramics and composites: role of crack-tip shielding. Mater Sci
Engng 1988;A103:15–28.
[5] Flinn RA, Trojan PK. Engineering materials and their applications. John Wiley and Sons; 1995.
[6] Kiyota M, Mihashi H, Kanda T, Kawamata A. Pull-out behavior of single fiber embedded in matrix. Cem Sci Concr Technol 2002;
56:625–32 (in Japanese).
[7] Srawley JE. Wide range stress intensity factor expressions for ASTM E-399 standard fracture toughness specimens. Int J Fracture
1976;12:475–6.
[8] Li VC, Wang Y, Backer S. Effect of inclining angle, bundling and surface treatment on synthetic fibre pull-out from a cement matrix.
Composites 1990;21(2):132–40.
[9] Slowik V, Leite JPB. Modellierung des mechanischen Verhaltens von Betonen auf der Ebene des Mesogefüges (in German). DFG
Research Report (German Research Institute), 1999, Germany.
[10] Leite JPB, Slowik V, Mihashi H. Computer simulation of fracture processes of concrete using mesolevel models of lattice structures.
Cem Concr Res 2004;34(6):1025–33.