ACI Materials Journal/March-April 2006 97

ACI MATERIALS JOURNAL TECHNICAL PAPER

ACI Materials Journal, V. 103, No. 2, March-April 2006.
MS No. 04-234 received July 21, 2004, and reviewed under Institute publication
policies. Copyright 2006, American Concrete Institute. All rights reserved, including
the making of copies unless permission is obtained from the copyright proprietors.
Pertinent discussion including authors closure, if any, will be published in the January-
February 2007 ACI Materials Journal if the discussion is received by October 1, 2006.
Rapid repair and retrofit of existing infrastructures demand
durable high-early-strength materials that not only deliver sufficient
strength within a few hours of placement but also significantly prolong
the maintenance interval. This paper reports a class of newly
developed polyvinyl alcohol (PVA) fiber-reinforced high-early-
strength engineered cementitious composites (ECC) materials
featuring extraordinary ductility. While micromechanics was applied in
many aspects of the material design process, emphasis of this paper is
put on the tailoring of preexisting flaw size distribution in matrix for
high tensile ductility. The resulting high-early-strength ECC materials
are capable of delivering a compressive strength of 21 MPa (3.0 ksi)
within 4 hours after placement and retaining long-term tensile
strain capacity above 2%.
Keywords: fibers; high early strength; strain.
INTRODUCTION
There is an increasing demand for durable high-early-
strength or rapid-hardening concrete materials in repair and
retrofit practices where minimum traffic disruption is
preferred. For instance, highway transportation authorities
often require the repair job to be completed in 6 to 8 hours at
night so that the lane can be opened to traffic the next morning.
In the past two decades, intensive experimental investigations
carried out by both academic and industrial groups have led to
successful formulation of concrete mixtures that can attain
sufficient compressive and flexural strengths at very early
ages. With various early strength gain rates, these concrete
mixtures obtain high early strength by using either proprietary
rapid hardening cements
1-5
or normal portland cement
together with accelerating admixtures.
6-8
Unfortunately, traditional concrete repairs often lack dura-
bility. It has been estimated that up to half of all concrete
repairs fail.
9
About 3/4 of the failures are attributed to the
lack of durability, with the remaining attributed to structural
failures. Premature deterioration is more common in repair
sites using high-early-strength concrete because many
proprietary binder systems often perform unpredictably
under various construction conditions. For example, reduced
freezing-and-thawing resistance was found in some very
high-early-strength concrete mixtures.
3
Meanwhile, early
age cracking, associated with autogeneous shrinkage and/or
thermal gradient under high temperature caused by rapid
hydration, also exacerbates the deterioration.
Engineered cementitious composite (ECC) is a special
type of high-performance fiber-reinforced cementitious
composite (HPFRCC) featuring significant tensile ductility
and moderate fiber volume fraction (typically 2%). The
design of ECC is guided by micromechanics models, which
provide quantitative links between composite mechanical
behavior and the properties of the individual phases, that is,
fiber, matrix, and interface.
10
Using the models, the desired
high tensile ductility, which is achieved by strain-hardening
and multiple-cracking, is converted to a set of constraints on
individual component properties. These components, that is,
the fiber, the matrix, and the interface, are then synergistically
tailored to meet the constraints.
ECC is expected to be an effective repair and retrofit material
due to its high ductility and tight crack width. Ductility is
being gradually recognized as the most critical property for
durable repair. As a repair material, ECC exhibits excep-
tional deformation compatibility with concrete substrate
structure. The high-fracture toughness and multiple-
cracking behavior enables ECC retrofit to effectively defuse
and arrest unstable crack propagation initiated from the
surrounding concrete or from the new/old concrete interface,
and hence eliminates spalling or delamination failures,
which are the common premature failure modes in repaired
structures.
11-13
Unlike conventional tension-softening FRC
materials, the crack width of ECC material in the strain-
hardening regime is an intrinsic material property and typically
below 100 m (0.004 in.) Because the flow rate in cracked
concrete scales as the third power of the crack width
14
and
approaches zero when the crack width is below 100 m
(0.004 in.),
15
the transport of water through cracks in ECC
cover through permeation is minimal
16
and hence prevents
the corrosion of steel reinforcement underneath. In
preloaded reinforced beam test under wet-dry cycles of a
chloride environment, Hiraishi et al.
17
demonstrated that the
steel reinforcement corrosion was significantly reduced
when concrete was replaced by ECC. Furthermore, even if
corrosion occurs, the strain-hardening capacity of ECC will
accommodate the expansion induced by corrosion and
prevent occurrence of spalling. Therefore, repair and retrofit
with ECC material may significantly extend the infrastructure
service life.
13,18
Conventional ECC mixtures use Type I ordinary portland
cement (OPC), which shows relatively slow strength develop-
ment. As a high-strength gain rate is desired, the selection of
a binder system has to take into consideration material cost,
workability, practice restriction, and long-term durability. If
adequate early strength can be attained, a normal portland
cement-based mixture is always favored over proprietary
rapid hardening cement. Even though a large number of
accelerating admixtures are available on the market, very
few, if any, provide a compressive strength of 2 ksi (13.8 MPa)
3 hours after placement.
5
Therefore, the selection of a binder
system is highly performance orientated, that is, proprietary
rapid hardening cement may be the sole choice if a stringent
strength requirement has to be met. In this study, various
combinations of binder systems and accelerating admixtures
Title no. 103-M11
High-Early-Strength Engineered Cementitious Composites
by Shuxin Wang and Victor C. Li
ACI Materials Journal/March-April 2006 98
were investigated and some illustrative mixtures are
presented herein.
High-early-strength ECC for rapid repair applications
must meet minimum strength requirements before the structure
can be returned to service. However, there is no existing
standard for minimum compressive strength prior to opening
to traffic. Parker and Shoemaker
6
suggested that for road
patch repair 2 ksi (13.8 MPa) compressive strength was
sufficient to prevent raveling, abrasion, deformation, and
cracking when initially opened to traffic. The New Jersey
State Department of Transportation specified target
compressive and flexural strengths of 3000 and 350 psi (20.7
and 2.4 MPa), respectively, in 6 hours to the fast-tract mix
developed in the mid-90s.
7
The California State Department
of Transportation (Caltrans) requires a minimum flexural
strength of 400 psi (2.8 MPa) prior to opening to traffic for
full depth pavement repair.
19
This strength specification is
based on pavement design and the experience of Caltrans
engineers that if the slab is subjected to traffic prior to
obtaining this strength, the durability and life expectancy of
the repaired pavement may be jeopardized. For very high-
early-strength concrete, Caltrans specifies 400 psi (2.8 MPa)
flexural strength at 4 hours after placement. The FHWA
Manual of Practice: Materials and Procedures for Rapid
Repair of Partial-Depth Spalls in Concrete Pavements
20
recommends a compressive strength of 6.9 MPa (1.0 ksi) and
flexural strength of 3.1 MPa (450 psi) for rapid-setting
cementitious concrete. In this study, the minimum strength
requirement is set to be 3 ksi (20.7 MPa) in compression and
500 psi (3.5 MPa) in flexural.
This paper presents selected results from an extensive
experimental program, with emphasis on the aspect of material
design. High-early-strength ECC based on rapid hardening
cement and normal portland cement are reported. The experi-
mental program comprises three phases, namely matrix design
for high early strength, tailoring for tensile strain-hardening
behavior, and characterization of other mechanical properties.
In the first phase, the factors most pertinent to early strength
gain rate, including the water-binder ratio, accelerator type,
and dosage, were investigated. In the second phase, to accom-
modate the adverse impact on the matrix and interface micro-
mechanical properties due to the use of high-early-strength
composition, the microstructure tailoring approach is intro-
duced to recover the tensile ductility when necessary. For
those mixtures with satisfactory early strength and ductility,
other mechanical properties including flexural response and
Youngs modulus were then measured.
RESEARCH SIGNIFICANCE
Current state-of-the-art high-early-strength cementitious
materials for rapid repair, including various rapid hardening
cement-based mortar and polymer mortar, are all quasi-
brittle in nature. The incorporation of short reinforcing
fibers, most commonly steel fiber and polypropylene fiber,
does not eliminate the tension-softening behavior and low-
strain capacity despite improvement in fracture energy. This
paper reports a type of fiber-reinforced cementitious
composite delivering very high early strength and significant
tensile strain capacity. The effectiveness of micromechanics
design approach is highlighted through matrix microstructure
tailoring. Durability of repair or new construction using
these materials is expected to be fundamentally improved
due to superior deformability, limited crack width, and
compatibility to existing substrates.
MATERIAL DESIGN FRAMEWORK
Strain-hardening behavior of ECC materials is obtained
from careful tailoring of the constituent properties. Specifically,
these properties are chosen such that steady state cracking
prevails under tension. This requires the crack tip toughness
J
tip
to be less than the complementary energy J
b
calculated
from the bridging stress versus crack opening curve, as
illustrated in Fig. 1
21-22
(1)
J
tip
= (2)
where
0
is the maximum bridging stress corresponding to
the opening
0
; E
m
is the matrix elastic modulus; and K
m
is
the fracture toughness. Equation (1) is obtained by consid-
ering the balance of energy changes during extension of the
steady state flat crack. The stress-crack opening relationship
(), which can be viewed as the constitutive law of fiber
bridging behavior, is derived by using analytic tools of frac-
ture mechanics, micromechanics, and statistics. In particular,
the energetics of tunnel crack propagation along fiber/matrix
is used to quantify the debonding process and the bridging
force of a fiber with given embedment length;
23
statistics are
introduced to describe the random location and orientation of
fibers. The random orientation of fiber also necessitates the
accounting of the mechanics of interaction between an
inclined fiber and the matrix crack. Another condition for
J
tip

0

0
( ) J
b
d
0

K
m
2
E
m
-------
ACI member Shuxin Wang is manager of MSeals Co. Ltd., Ningbo, China. He
received his BS and MS from Tsinghua University at Beijing, Beijing, China, and his
PhD from the Department of Civil and Environmental Engineering at the University of
Michigan, Ann Arbor, Mich. His research interests include the development of high-per-
formance fiber-reinforced cementitious composites.
ACI member Victor C. Li is a professor of civil and environmental engineering at the
University of Michigan. He is a member of ACI Committee 544, Fiber Reinforced
Concrete. His research interests include micromechanics-based composite materials
design and engineering, innovative structures design based on advanced materials
technology, and sustainable infrastructure engineering.
Fig. 1Typical () curve for strain-hardening composite.
Hatched area represents complimentary energy J
b
. Shaded
area represents crack tip toughness J
tip
.
ACI Materials Journal/March-April 2006 99
pseudo strain-hardening is that the tensile first crack strength

fc
must not exceed the maximum bridging stress
0
,

fc
<
0
(3)
where
fc
is determined by the maximum preexisting flaw size
max[a
0
] and the matrix fracture toughness K
m
. Details of these
micromechanical analyses can be found in References 23 and
24. Satisfaction of Eq. (1) and (3) is necessary to achieve
ECC behavior. Otherwise, normal tensile softening FRC
behavior results.
The shape of the () curve, which determines the value
of complementary energy J
b
, is further related to a number
of fiber/matrix interaction mechanisms. For hydrophilic
polyvinyl alcohol (PVA) fiber, which has been used for the
reinforcement of ECC materials, the fiber/matrix interaction
is characterized by interfacial frictional stress
0
, chemical
bond G
d
, and slip-hardening coefficient accounting for the
slip-hardening behavior during fiber pullout. In addition,
snubbing coefficient f and strength reduction factor f are
introduced to account for the interaction between fiber and
matrix as well as the reduction of fiber strength when pulled
at an inclined angle. Besides the interfacial properties, the
() curve is also governed by the fiber content V
f
, fiber
diameter d
f
, length L
f
, and Youngs Modulus E
f
.
The aforementioned steady state cracking criteria calls for
high margin between J
b
and J
tip
, for example, the moderate
matrix toughness and high complementary energy. The latter
further requires judicious control of fiber geometry and
interface properties such that fiber could be pulled out under
sufficient resistance with limited amount of rupture. For the
PVA fiber used in ECC, special surface treatment is applied
to the fiber during production to reduce the excessive bond
to cementitious matrix.
25
Additionally, the regular ECC
mixtures contain high content of water and fly ash for
controlling matrix toughness and interface properties.
26
In
the design of high early strength ECC, however, fly ash
content has to be reduced and a low water-cement ratio (w/c)
is preferred for quick strength gain at early ages. The use of
high-early-strength cement may also alter the interface
properties, mostly unfavorable to high complementary energy.
When the margin between J
b
and J
tip
is small, it becomes
critical to control pre-existing flaw distribution in the matrix
to retain high strain capacity. While Eq. (1) and (3) guarantee
the occurrence of multiple-cracking, the number of the
cracks that could be developed (the multiple-crack saturation
level) is governed by the flaw size and their spatial distribution.
Limited by the peak bridging stress, a lower bound of critical
flaw size c
mc
is set such that only those flaws larger than c
mc
can be activated prior to reaching
0
and contribute to
multiple cracking. Therefore, to achieve saturated multiple
cracking, a sufficient number of such large flaws must exist
in the matrix. In regular ECC without explicit control of flaw
size distribution, low toughness matrix is often used such
that the reduced c
mc
would produce adequate margin to
activate a large number of cracks. Considering the random
nature of pre-existing flaws inherent to processing, one
approach that can be easily adopted in practice is to introduce
artificial flaws with prescribed sizes just above c
mc
as a
superimposition to the natural flaw system. A variety of
particles can be used as the artificial flaws in practice,
provided that the presence of these particles can cause local
stress concentration in the brittle matrix under loading. A
lower bound of the content of artificial flaws can be estimated
from the minimum cracking spacing constrained by fiber
interface properties and matrix cracking stress, while the
upper bound is limited by processing and disturbance of the
relatively large particles to fiber distribution. Details
regarding this approach can be found in Reference 27.
EXPERIMENTAL PROGRAM
Extensive experiments were conducted to examine the early
strength gain rate of various combinations of binder system
and accelerating admixture under approximate proportion of
ECC materials. In addition to Type I OPC and Type III high
early strength portland cement, calcium aluminate cement,
gypsum cement, and blended calcium sulfo-aluminate were
also evaluated. Commercially available accelerators, which
effective components include calcium chloride, calcium
nitride, and ammonium calcium nitrate were used in the
investigation together with a high-range water-reducing
admixture that is necessary for proper fiber dispersion. A
high-range water-reducing admixture, depending on chemical
species, exhibits considerable influence on early strength
development. High-range water-reducing admixture based on
naphthalene sulfonate, melamine formaldehyde sulfonate, and
polycarboxylate were tested. In addition, the influences of the
water-binder ratio and sand content were also investigated.
Five illustrative mixture proportions are listed in Table 1.
Mixtures SC01 and SC19 both use a rapid-hardening
blended portland cement (Blaine surface area 4730 cm
2
/g)
with calcium sulfo-aluminate. Mixtures HP08 and HP09
both use Type III portland cement (surface area 5000 cm
2
/g).
Mixture OP08 is a regular ECC mixture using Type I OPC
(surface area 3300 cm
2
/g) and the results are presented
herein for comparison. Mixtures SC01 and SC09 have same
mixture proportions except that SC09 contains 4.6% by
volume polypropylene (PP) particles for creating artificial
flaws. These disk-shape particles with a diameter of
approximately 4 mm and a thickness of approximately 2 mm
have little bond to surrounding cementitious species. Similarly,
Mixtures HP08 and HP09 have same mixture proportions
Table 1Mixture proportions
Mix-
ture
ID
Cement,
kg/m
3

(lb/yd
3
)
Sand,
kg/m
3

(lb/yd
3
)
Other
aggregates,
kg/m
3
(lb/yd
3
)
Water,
kg/m
3

(lb/yd
3
)
Fine fly
ash,
kg/m
3
(lb/yd
3
)
PVA
fiber,
kg/m
3

(lb/yd
3
)
Admixture,
*
L/m
3
(fl oz/yd
3
)
SC01
906

(1526)
724
(1221)

350
(590)
90.6
(153)
26.0
(43.8)
13.9 (359)
[PT20]
SC19
863

(1455)
690
(1163)
41.4

(69.8)
334
(563)
86.0
(145)
26.0
(43.8)
13.3 (343)
[ML330]
HP08
893

(1506)
893
(1506)

292
(492)

26.0
(43.8)
6.3 (163)
[GL3200]
25.7 (664)
[NC534]
HP09
848

(1430)
848
(1430)
54.0
||
(91.0)
278
(469)

26.0
(43.8)
5.9 (153)
[GL3200]
24.4 (631)
[NC534]
OP08
583
#
(983)
467
(787)

298
(502)
700
(1180)
26.0
(43.8)
15.0 (388)
[ML330]
*
PT20: accelerator and high-range water-reducing admixture containing ammonium
calcium nitrate and naphthalene sulfonate salt; ML330: melamine formaldehyde sulfonate-
based high-range water-reducing admixture; GL3200: polycarboxylate-based high-range
water-reducing admixture; and NC534: calcium nitrate-based accelerator.

Rapid-hardening cement (Type S-30, Korea).

Polypropylene (PP) beads used as artificial flaws.

Type III cement.

||
Polystyrene (PS) beads used as artificial flaws.
#
Type I ordinary portland cement.
100 ACI Materials Journal/March-April 2006
except that 5.0% by volume of polystyrene (PS) particles are
introduced as artificial flaws in Mixture HP09. These PS
beads have a skewed cylinder shape and sharp edges. The
length of longest dimension of the PS particles is approximately
4 mm. A small amount of ultra fine fly ash
28
(surface area
8000 cm
2
/g) with high reactivity is incorporated into SC01
and SC19 for rheology control. The sand used is a type of
fine silica sand with size distribution from 50 to 250 m
(0.002 to 0.010 in.) and a mean size of 110 m (0.004 in.)
(8% < 53 m, 33% < 75 m, 77% < 100 m, 93% < 150 m).
A special PVA fiber designed particularly for ECC applications
is used at a moderate volume fraction of 2% in all mixtures.
The fiber has a length of 12 mm (0.47 in.), a diameter of
39 m (0.0015 in.), and a nominal tensile strength of
1620 MPa (235 ksi). The density of the fiber is 1300 kg/m
3
(2191 lb/yd
3
). For the rapid-hardening cement-based
mixtures, a powerful dispersant is sufficient to achieve rapid
strength gain, and the effect of accelerator was found to be
minimal regardless of its type. In SC01, a multiple purpose
admixture acting as both accelerator and high-range water-
reducing admixture was used, while in SC19 only melamine
formaldehyde sulfonate-based high-range water-reducing
admixture was used. For the Type III cement-based system,
however, a combination of polycarboxylate-based high-
range water-reducing admixture and calcium nitrate-based
accelerator exhibits most significant accelerating effect
when calcium chloride is excluded, as found in Reference 8
for high-early-strength concrete.
The ECC mixture was prepared in a Hobart-type mixer
with 10L (0.35 ft
3
) capacity. Solid ingredients, including
cement, sand, and fly ash and particles for artificial flaws if
applicable were first mixed for approximately 1 minute, then
water was added and mixed for another 3 minutes. Next, a
high-range water-reducing admixture was added into the
mixer. Once a consistent mixture was reached, fiber was
slowly added. Accelerating admixture, if used, was added
before casting. The whole mixing procedure typically takes
8 to 10 minutes. The mixture was then cast into molds with
moderate vibration applied. Due to small specimen dimensions,
the molds were covered with plastic sheets and stored in a
container with an insulation wall to reduce the loss of hydration
heat. Specimens were demolded after 24 hours and then cured
in air at room temperature (16 to 20 C) before testing. For those
tested before 24 hours, the specimens were demolded just
before testing. The room relative humidity is 55 5%.
All compressive tests use cylinder specimens 75 mm (3 in.)
in diameter by 150 mm (6 in.) in height. The ends of the
specimens were capped with sulfur compound. Testing
began at 3 hours after casting or when adequate strength had
developed. The age of the specimen was recorded as the time
elapse from finish of casting to test. The mixing and casting
procedure took approximately 20 to 30 minutes.
Direct uniaxial tensile test was conducted to characterize
the tensile behavior of the ECC material.
26
The coupon
specimen used here measures 304.8 x 76.2 x 12.7 mm (12 x
3 x 0.5 in.). Aluminum plates were glued at the ends of the
coupon specimen to facilitate gripping. Tests were
conducted under displacement control at a loading rate of
0.005 mm/s (0.0002 in./s). Two external linear variable
displacement transducers were attached to the specimen
surface with a gauge length of approximately 180 mm (7.1 in.)
to measure the displacement. Youngs modules was calculated
from the linear elastic portion of the uniaxial stress-strain curve
prior to cracking.
Flexural behavior was measured using beam specimen
with dimensions of 304.8 mm (12 in.) in length by 76.2 mm
(3 in.) in width by 25.4 mm (1 in.) in depth under four point
bending. The span between supports was 254 mm (10 in.)
and the constant moment length was 76.2 mm (3 in.). The
loading rate was 0.05 mm/s (0.002 in./s), and the displacement
at the loading point was recorded. It should be noted that a
flexural test is not a reliable test for strain-hardening but is
carried out here mainly because flexural strength are specified in
some repair applications.
A single fiber pullout test was conducted to measure the
fiber/matrix interfacial properties, including frictional stress

o
, chemical bond G
d
, and slip-hardening coefficient . The
specimen preparation and test method can be found in
Reference 25. The data processing and calculation of the
parameters follow Reference 29.
The matrix fracture toughness K
m
was determined by
three-point bending test compliant to ASTM E 399. The
beam specimen of the matrix mixture without fiber measured
304.8 mm (12 in.) in length by 76.2 mm (3 in.) in width by
38.1 mm (1.5 in.) in depth; the loading support spanned
254.0 mm (10 in.). The notch depth-to-height ratio was 0.4.
RESULTS AND DISCUSSION
The compressive strength development of high-early-
strength ECC mixtures up to 28 days is illustrated in Fig. 2,
along with that of regular ECC Mixture OP08. The fast-
strength gain is recorded after the first 5 hours in both the
rapid-hardening cement-based mixtures and the Type III
portland cement-based mixtures, in which a compressive
strength of 21 MPa can be easily achieved at 3 to 4 hours
after placement, sufficient for most emergency repair appli-
cations. In contrast, regular ECC OP08 develops same
strength only after approximately 24 hours. Compared with
rapid-hardening cement, which sets and hardens within one
hour, Type III cement-based mixtures exhibit little strength
at the first two hours even with the aid of an accelerator.
Higher early strength can be readily achieved by further
reducing water to binder ratio; however, the consequent
increase in interface bond properties and matrix toughness
diminishes the potential of developing multiple-cracking to
an extent that only tension-softening behavior prevails.
The introduction of PP and PS beads at small volume frac-
tion has demonstrated negligible effect on the early age
compressive strength. Compared with the mixtures without
Fig. 2Compressive strength development of rapid-hardening
cement-based mixtures.
ACI Materials Journal/March-April 2006 101
artificial flaws (for example, SC01 and HP08), the corre-
sponding mixes containing the plastic beads (for example,
SC19 and HP09) show little difference in compressive
strength prior to approximately 30 MPa. With maturing of the
matrix, however, the weakening effect of the soft PP beads
prevails due to the loss of deformation compatibility between
the bead and the surrounding mortar. At 28 days, a reduction
of approximately 20% in strength was observed in those
mixtures with plastic beads. Nevertheless, the compressive
strength of SC19 and HP09 at 28 days is considered adequate
for most applications.
Figure 3 shows the tensile behavior of SC01 at different
ages. A rapid decrease in strain capacity with age was
observed, however, accompanied by increase of first
cracking strength and ultimate tensile strength. At 3 hours
after casting, the material exhibits satisfactory strain-hard-
ening behavior with a strain capacity above 3%. The
ductility is further improved at the age of 7 hours, where an
ultimate strain of 6% is demonstrated. However, the strain
capacity quickly decreases to about 1.0% after 24 hours and
retains only 0.7% after 3 days.
The rapid decrease of strain capacity can be attributed to
the continuing increase of both interfacial bond strength and
matrix toughness. Figure 4 presents the age dependency of
the interface properties, for example frictional stress
o
,
chemical bond G
d
, and slip-hardening coefficient ,
measured from single fiber pullout tests, as well as K
m
deter-
mined from notched beam bending test. Despite scattering of
the data, the trend clearly shows that the development of
bond between fiber and matrix remains relatively slow at the
first 24 hours and then accelerates considerably before
finally saturating after 7 days. In particular, the gain rate of

o
at early age is faster than G
d
and . Meanwhile, the devel-
opment of K
m
at early age is even faster than the interface
properties, and it saturates after approximately 3 days. The
different development rate may be explained by the fact that the
accelerated hydration is mainly taking place among the cement
species and the reactivity of fiber surface is less affected.
The evolution of micromechanical properties results in the
quick change of J
b
/J
tip
ratio that indicates the margin for strain-
hardening. PVA fiber features high bond strength both frictional
and chemical to hydrated cement. In addition, PVA fiber
exhibits strong slip-hardening behavior during pullout. There-
fore, the fiber often ruptures instead of being pulled out from the
matrix. Reflecting on the micromechanical property, the
premature failure of fiber leads to a diminishing complementary
energy J
b
. For rapid-hardening cement-based mixture,
Fig. 3Tensile behavior of rapid-hardening cement-based
Mixture SC01.
Fig. 4Age dependency of microchemical properties of
Mixture SC01.
102 ACI Materials Journal/March-April 2006
significant increases of
o
, G
d
, and after 24 hours cause
severe fiber rupture. As illustrated in Fig. 5(a), J
b
rapidly
decreases in first 48 hours, while J
tip
increases with the
maturing of the matrix. Consequently, the J
b
/J
tip
ratio drops
from approximately 15 to approximately 1 after 3 days and
flattens after 7 days. Despite the simplifying assumptions
made in the model that tends to underestimate J
b
, the model
precisely captures the nature of the deterioration of strain-
hardening behavior with age. When plotting the J
b
/J
tip
ratio
and the strain capacity together against age, it is evident that
both follow the same trend (Fig. 5(b)) except for the very early
age at 3 hours where premature failure at the grips were
encountered due to difficulty in applying epoxy glue at very
early age.
For SC01, the condition for strain-hardening (J
b
/J
tip
> 1)
is easily satisfied at the first 24 hours, as demonstrated by the
experimentally determined high strain capacities. The gap
between J
b
and J
tip
becomes rather small after 3 days.
According to the criterion, strain-hardening should still
occur though the diminishing margin may restrict multiple-
cracking saturation. In fact, although the crack number is
very limited, multiple-cracking is still observed after 3 days.
Furthermore, the model indicates the possibility to achieve
high strain-hardening after 7 days because J
tip
is still within
the range of J
b
.
The potential of multiple-cracking is confirmed in SC19,
which has same mixture proportion as SC01 except that PP
beads are introduced as artificial flaws at volume fraction of
4.6%. The size of the PP beads was chosen to be comparable
to the largest flaws found in section examination of the
uniaxial test specimens. Figure 6 shows the tensile stress-
strain curves at ages of 1, 3, and 21 days, and the strain
capacity at age after 24 hours is significant increased. The
strain capacity remains approximately 2.5% after 21 days,
whereas the saturation of micromechanical properties has
been reached at approximately 14 days (Fig. 4), suggesting
that the strain capacity can be retained in the long term. Well-
distributed and near-saturated multiple cracking was observed
with average crack spacing of 3 to 4 mm (0.12 to 0.16 in.).
A similar trend was also seen in the Type III cement-based
high-early-strength Mixture HP08, where the fine grain size
of grounded portland cement along with low w/c necessi-
tated for rapid strength gain cause a rapid increase in K
m
and
hence reduce the margin for multiple-cracking. Figure 7
shows the tensile stress-strain curve of HP08 at 28 days,
where multiple-cracking behavior is evident, but strain
capacity is barely above 1.0%. In contrast, with the introduction
of 5.0% by volume of PS beads as artificial flaws, HP09
demonstrates strain capacity of 3.5% at 50 days, retaining
Fig. 5Evolution of complementary J
b
and crack tip tough-
ness J
tip
with age in Mixture SC01.
Fig. 6Tensile behavior rapid-hardening cement-based
Mixture SC19 at different ages.
Fig. 7Tensile behavior of Type III cement-based mixture
without artificial flaws (Mixture HP08) at 28 days.
ACI Materials Journal/March-April 2006 103
most ductility exhibited at 5 hours in spite of 100% increase
of cracking strength (Fig. 8). It was also noticed that the
presence of artificial flaws only slightly reduces the first
cracking strength due to its well controlled size distribution.
The high ductility of HP09 therefore originates from a more
homogeneous defect system that imparts the potential of
saturated multiple-cracking, as self-evident in comparison
between the cracking patterns of HP08 and HP09 (Fig. 9). It
should be mentioned that microcrack width in ECC materials
remains almost constant during strain-hardening regime, and
increase of strain capacity for a particular system mainly relies
on increase of multiple-cracking density. For HP09, the average
crack width is 65 m (0.0026 in.), and the average crack
spacing is 2.5 mm (0.1 in.).
Although the effectiveness of flaw size tailoring on
improving ductility is highlighted, this approach should be
viewed as supplemental to interface and matrix toughness
control. Figure 10 shows typical stress-strain curves of the
OPC-based ECC Mixture OP08 tested at the age of 24 hours
when the 21 MPa (3.0 ksi) minimum strength criterion is
reached and at the age of 90 days when hydration has been
stabilized. In both cases, strain capacity about 2.5% is
recorded, even though the flaw system is not deliberately
tailored. The selection of binder composition, water-binder
ratio, sand content, and fiber properties in OP08 has ensured
relatively large margin between J
b
and J
tip
such

that sufficient
natural flaws can be activated for multiple cracking. In the
design of high-early-strength ECC, these micromechanics
criteria were also followed in the selection of mixture proportion,
which enables the further tailoring of flaw size distribution.
Figure 11 illustrates the deflection-hardening behavior of
SC19, HP09, and OP08 at different ages. The cracking
strength and ultimate flexural strength of SC09 are 4.5 MPa
(0.65 ksi) and 6.7 MPa (0.97 ksi), respectively, at 3 hours
after placement, which are considerably higher than conven-
tional high early strength repair mortar. At later age, the flexural
strength stabilizes at about 15 MPa (2.17 ksi). Compared with
SC09, flexural ductility of HP09 is noticeably higher at the
same age, while similar ultimate flexural strength was
observed. At 4 hours after placement, the material delivers
10 MPa (1.45 ksi) flexural strength. It was noticed that the
presence of artificial flaws had a less significant effect on the
flexural behavior than on the uniaxial tensile behavior. For
instance, the bending response of OPC-based OP08 without
artificial flaws demonstrated higher ductility, though OP08
showed less tensile strain capacity in uniaxial tension tests
compared with HP09. One plausible explanation is that the
cracking initiation effect of the artificial flaws, which is
dominant under uniaxial tension, is masked by the stress
gradient effect (cracks propagate from tensile into compression
Fig. 8Tensile behavior of Type III cement-based Mixture
HP09 with artificial flaws (PS beads 5.0 vol. %) at ages
of 5 hours and 50 days.
Fig. 9Cracking patterns of Type III cement-based mixtures:
(a) unsaturated multiple-cracking in Mixture HP08 at 28 days;
and (b) near-saturated multiple-cracking in Mixture HP09
with artificial flaws at 50 days.
Fig. 10Typical stress-strain curves of OPC-based regular
ECC Mixture OP08 at 24 hours and 90 days.
104 ACI Materials Journal/March-April 2006
zone of beam) under flexural bending that stabilizes crack
propagation and allows more cracks to be developed. As a
result, a composite hardly showing strain-hardening under
uniaxial tension may exhibit unambiguous flexural deflection-
hardening as observed in many FRC materials. This
behavior, however, is dependent on beam height.
CONCLUSIONS
High-early-strength ECC materials based on various binder
systems were developed under the guidance of micromechan-
ical models. Table 2 summarizes the typical mechanical
properties of two exemplary ECC mixtures based on rapid-
hardening cement and Type III portland cement respectively,
along with that of the regular ECC mixture which uses ordi-
nary portland cement for comparison. In particular, the
following conclusions can be drawn:
1. Proprietary rapid-hardening cement needs to be used
when stringent high early strength is required, for
example, 21 MPa (3.0 ksi) within 3 hours. The exemplary
mixture delivers 24 MPa (3.5 ksi) compressive strength at
3 hours and retains tensile strain capacity above 2.0% in
the long term;
2. The Type III portland cement-based mixture containing
a polycarboxylate-based high-range water-reducing
admixture and calcium nitrate-based accelerator is capable
of attaining 21 MPa (3.0 ksi) at 4 hours and retain strain
capacity 3.5% at 50 days;
3. The development of interface frictional stress, chemical
bond and slip-hardening coefficient of PVA fiber in rapid-
hardening cement-based mixture remains slow at first 24 hours,
then accelerates and finally stabilizes after 14 days, while the
matrix fracture toughness evolves at a faster pace and saturates
after about 3 days. Micromechanics model reveals that the
quick deterioration in strain capacity can be attributed to
rapid drop of complementary energy and continuous rise of
crack tip toughness;
4. In particular, the matrix microstructure tailoring flaw-size
distribution control is necessary to promote multiple-cracking
in high-early-strength ECC. Introduction of artificial flaws
with prescribed size distribution has been demonstrated to be
an effective approach to retain high strain capacity at late age.
The presence of a small volume fraction (for example, 5%) of
graded plastic particles with weak bond-to-cement hydrates
results in nearly saturated multiple cracking; and
5. Although the bending response benefits less from the
artificial flaw tailoring approach, high-early-strength ECC
materials in general show significant deflection-hardening
behavior. The flexural strength reaches 10 MPa (1.4 ksi)
in 4 hours and 16 MPa (2.3 ksi) at a late age.
ACKNOWLEDGMENTS
The authors would like to thank Wonha Construction Co. Ltd., Korea, and
the Michigan Department of Transportation for supporting this research.
REFERENCES
1. Seehra, S. S.; Gupta, S.; and Kumar, S., Rapid Setting Magnesium
Phosphate Cement for Quick Repair of Concrete PavementsCharacter-
ization and Durability Aspects, Cement and Concrete Research, V. 23,
No. 2, 1993, pp. 254-266.
2. Knofel, D., and Wang, J. F., Properties of Three Newly Developed Quick
Fig. 11Typical bending behaviors of: (a) Mixture SC19;
(b) Mixture HP09; and (c) Mixture OP08 at different ages.
Table 2Mechanical properties of examples
Mixture ID SC19 HP09 OP08
Compressive
strength,
MPa (ksi)
24 (3.5) at 3 h
37 (5.4) at 24 h
60 (8.7) at 28 days
26 (3.8) at 4 h
42 (6.1) at 24 h
53 (7.7) at 28 days
18 (2.6) at 16 h
24 (3.5) at 24 h
65 (9.4) at 28 days
Flexural
strength,
MPa (ksi)
7.0 (1.0) at 3 h
14 (2.0) at 24 h
16 (2.3) at 28 days
10 (1.4) at 4 h
12 (1.7) at 24 h
16 (2.3) at 28 days
11 (1.6) at 24 h
15 (2.2) at 28 days
Tensile
strength,
MPa (ksi)
3.5 (0.5) at 3 h
3.5 (0.5) at 24 h
5.2 (0.8) at 28 days
3.5 (0.5) at 3 h
3.2 (0.5) at 5 h
5.5 (0.8) at 28 days
3.5 (0.5) at 3 h
3.0 (0.4) at 24 h
5.5 (0.8) at 28 days
Tensile strain
capacity, %
>2.0 >3.0 >3.0
Youngs
modulus,
GPa (ksi)
20.8 (3014)
at 28 days
22.8 (3304)
at 28 days
21.0 (3043)
at 28 days
ACI Materials Journal/March-April 2006 105
Cements, Cement and Concrete Research, V. 24, No. 5, 1994, pp. 801-812.
3. Whiting, D., and Nagi, M., Strength and Durability of Rapid Highway
Repair Concretes, Concrete International, V. 16, No. 9, Sept. 1994, pp. 36-41.
4. Sprinkel, M. M., Very-Early-Strength Latex-Modified Concrete
Overlay, Report No. VTRC99-TAR3, Virginia Department of Transportation,
Richmond, Va., 1998, 11 pp.
5. Baraguru, P. D., and Bhatt, D., Rapid Hardening Concrete, Report
No. FHWA NJ 2001-3, New Jersey Department of Transportation, Trenton,
N.J., 2000, 22 pp.
6. Parker, F., and Shoemaker, M. L., PCC Pavement Patching Materials
and Procedures, Journal of Materials in Civil Engineering, ASCE, V. 3,
No. 1, 1991, pp. 29-47.
7. Kurtz, S.; Balaguru, P.; Consolazio, G.; and Maher, A., Fast Track
Concrete for Construction Repair, Report No. FHWA 2001-015, New
Jersey Department of Transportation, Trenton, N.J., 1997, 67 pp.
8. Anderson, J.; Daczko, J.; and Luciano, J., Producing and Evaluating
Portland Cement-Based Rapid Strength Concrete, Concrete International,
V. 25, No. 8, Aug. 2003, pp. 77-82.
9. Mather, B., and Warner, J., Why do Concrete Repairs Fail, interview
held at University of Wisconsin, Department of Engineering Professional
Development, Wis., <http://aec.engr.wisc.edu/resources/rsrc07.html>,
accessed Dec. 2004.
10. Li, V. C., Engineered Cementitious CompositesTailored Composites
through Micromechanical Modeling, Fiber Reinforced Concrete: Present
and the Future, N. Banthia, A. Bentur, and A. Mufti, eds., Canadian Society for
Civil Engineering, Montreal, Quebec, Canada, 1998, pp. 64-97.
11. Lim, Y. M., and Li, V. C., Durable Repair of Aged Infrastructures
Using Trapping Mechanism of Engineered Cementitious Composites, Journal
of Cement and Concrete Composites, V. 19, No. 4, 1997, pp. 373-385.
12. Li, V. C.; Horii, H.; Kabele, P.; Kanda, T.; and Lim, Y. M., Repair
and Retrofit with Engineered Cementitious Composites, Engineering
Fracture Mechanics, V. 65, 2000, pp. 317-334.
13. Li, V. C., High Performance Fiber Reinforced Cementitious
Composites as Durable Material for Concrete Structure Repair, International
Journal for Restoration of Buildings and Monuments, V. 10, No. 2, 2004,
pp. 163-180.
14. Tsukamoto, M., Tightness of Fiber Concrete, Darmstadt Concrete,
V. 5, 1990, pp. 215-225.
15. Wang, K.; Jansen, D. C.; Shah, S.; and Karr, A. F., Permeability
Study of Cracked Concrete, Cement and Concrete Research, V. 27, No. 3,
1997, pp. 381-393.
16. Lepech, M., and Li, V. C., Water Permeability of Cracked Cementitious
Composites, Proceedings of ICF11, Paper 4539, Turin, Italy, Mar. 2005.
(CD-ROM)
17. Hiraishi, Y.; Honma, T.; Hakoyama, M.; and Miyazato, S., Steel
Corrosion at Bending Cracks in Ductile Fiber Reinforced Cementitious
Composites, Proceedings of the JCI Symposium on Ductile Fiber Reinforced
Cementitious Composites (DFRCC), Tokyo, Japan, 2003. (in Japanese)
18. Li, V. C., and Lepech, M., Crack Resistant Concrete Material for
Transportation Construction, Transportation Research Board 83rd Annual
Meeting, Washington, D.C., Compendium of Papers, Paper 04-4680, 2004.
(CD-ROM)
19. Paving Repair Finds a Four-Hour Champion, Concrete Construction,
V. 46, No. 12, 2001, pp. 69-70.
20. Federal Highway Administration (FHWA), Manual of Practice:
Materials and Procedures for Rapid Repair of Partial-Depth Spalls in
Concrete Pavements, 1999, 135 pp.
21. Marshall, D. B., and Cox, B. N., A J-Integral Method for Calculating
Steady-State Matrix Cracking Stresses in Composites, Mechanics of
Materials, No. 8, 1988, pp. 127-133.
22. Li, V. C., and Leung, C. K. Y., Theory of Steady State and Multiple
Cracking of Random Discontinuous Fiber Reinforced Brittle Matrix
Composites, Journal of Engineering Mechanics, ASCE, V. 118, No. 11,
1992, pp. 2246-2264.
23. Lin, Z.; Kanda, T.; and Li, V. C., On Interface Property Characterization
and Performance of Fiber-Reinforced Cementitious Composites, Journal of
Concrete Science and Engineering, RILEM, No. 1, 1999, pp. 173-184.
24. Li, V. C., Post-Crack Scaling Relations for Fiber-Reinforced
Cementitious Composites, Journal of Materials in Civil Engineering,
ASCE, V. 4, No. 1, 1992, pp. 41-57.
25. Li, V. C.; Wu, C.; Wang, S.; Ogawa, A.; and Saito, T., Interface
Tailoring for Strain-Hardening Polyvinyl Alcohol-Engineered Cementitious
Composite (PVA-ECC), ACI Materials Journal, V. 99, No. 5, Sept.-Oct.
2002, pp. 463-472.
26. Li, V. C.; Wang, S.; and Wu, C., Tensile Strain-Hardening Behavior of
PVA-ECC, ACI Materials Journals, V. 98, No. 6, Nov.-Dec. 2001,
pp. 483-492.
27. Wang, S., and Li, V. C., Tailoring of Pre-Existing Flaws in ECC
Matrix for Saturated Strain Hardening, Proceedings of FRAMCOS-5, Vail,
Colo., Apr. 2004, pp. 1005-1012.
28. Obla, K. H.; Hill, R. L.; Thomas, M. D. A.; Shashiprakash, S. G.;
and Perebatova, O., Properties of Concrete Containing Ultra-Fine Fly
Ash, ACI Materials Journal, V. 100, No. 5, Sept.-Oct. 2003, pp. 426-433.
29. Redon, C.; Li, V. C.; Wu, C.; Hoshiro, H.; Saito, T.; and Ogawa, A.,
Measuring and Modifying Interface Properties of PVA Fibers in ECC
Matrix, Journal Materials in Civil Engineering, ASCE, V. 13, No. 6,
2001, pp. 399-406.