Composites: Part B 60 (2014) 359–370

Contents lists available at ScienceDirect

Composites: Part B
journal homepage: www.elsevier.com/locate/compositesb

Influence of matrix flowability, fiber mixing procedure, and curing

conditions on the mechanical performance of HTPP-ECC
Burak Felekoglu a, Kamile Tosun-Felekoglu a,⇑, Ravi Ranade b, Qian Zhang b, Victor C. Li b
a
Department of Civil Eng., Dokuz Eylul University, Izmir, Turkey
b
Department of Civil and Environmental Eng., University of Michigan, Ann Arbor, MI, USA

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

Article history: In this study, the influences of matrix flowability, fiber mixing procedure, and curing conditions on the
Received 8 October 2013 mechanical properties of Engineered Cementitious Composites (ECC) made with High Tenacity Polypro-
Received in revised form 10 December 2013 pylene (HTPP) fibers are investigated. While the HTPP-ECC examined in this study possesses moderate
Accepted 27 December 2013
compressive strengths (30–70 MPa), their tensile ductility (1.91–3.91%) is similar to that of ECC with
Available online 6 January 2014
Polyvinyl Alcohol (PVA) fibers. For the purpose of controlling matrix flowability, different dosages of
HRWR admixture were introduced to a matrix with fly ash/cement weight ratio of 2.8 and water/cemen-
Keywords:
titious material weight ratio of 0.23. Dogbone-shaped and 50 mm cube specimens were used to investi-
A. Fibers
B. Mechanical properties
gate uniaxial tensile and compressive properties of HTPP-ECC, respectively. Test results showed that
E. Cure control of flowability in a certain range is required to achieve robust tensile ductility. A further improve-
Engineered Cementitious Composites ment in tensile ductility and mechanical properties of HTPP-ECC was achieved through water-curing
instead of air curing typically used for PVA-ECC. The basic mechanisms that enhance tensile ductility
of HTPP-ECC through flowability control, mixing procedure modification, and water-curing are discussed
from the view point of micromechanics underlying ECC design, with supporting evidence from fiber
bridging stress–crack width (r–d) relations.
Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction hardening. The ductility of ECC is the sum total of distributed

deformation resulting from the diffused microcrack damage until
Over the last two decades, Engineered Cementitious Composites the tensile capacity of bridging fibers on one of these microcracks
(ECC) have been developed to address the shortcoming of brittle- is exhausted [6].
ness in conventional plain concrete. Unlike plain concrete and nor- Using the micromechanics-based conditions for multiple crack-
mal fiber reinforced concrete (FRC), ECC shows a tensile strain ing and the pseudo strain-hardening (PSH) indices described in
hardening behavior after first cracking and develops a high tensile Section 4.1, ECC with various fiber types have been designed. The
ductility in the range of 3–7% [1], while conventional concrete frac- first ECC reported in literature with significant strain-hardening
tures only at about 0.01% tensile strain with a localized crack of behavior was prepared with polyethylene (PE) SpectraÒ fiber. The
infinite width [2,3]. Due to this reason, successfully tailored ECC interface of PE fibers with the matrix was tailored using plasma
is also referred as strain hardening cementitious composites treatment to enhance the fiber/matrix bond [5,7]. However, the
(SHCC) [4]. Tensile ductility of ECC is engineered through a PE fiber is very expensive and alternative polymer fibers, particu-
micromechanics-based design framework developed by Li [5] larly Polyvinyl Alcohol (PVA), were explored for producing ECC
which is summarized in Section 4. In contrast to the design of ultra [8–10]. PVA fiber, unlike the PE fiber, forms a strong chemical bond
high strength concrete, ECC benefits from the pre-existing flaws by with the cementitious matrix, in addition to frictional bond. As a
managing their size range rather than by eliminating them. Typi- result, instead of plasma treatment used for PE fibers to increase
cally, the damage pattern under uniaxial tension is a large number the fiber/matrix bond, an oil coating is applied on the PVA fibers
of closely spaced microcracks of widths less than 100 lm, spaced a to reduce the fiber/matrix bond that may otherwise rupture the fi-
few millimeters apart. During multiple microcrack development, ber during pullout. Such interface tailoring ensures synergistic
the tensile load continues to increase. In this manner, localized interaction between fiber, matrix, and interface, which is the key
catastrophic brittle fracture is suppressed, in favor of strain- for achieving high tensile ductility [11].
Even though the PVA fibers are cheaper than PE fibers, they still
account for approximately 50% of the total material cost of ECC.
⇑ Corresponding author. Tel.: +90 232 3017059; fax: +90 232 3017253. Recycled plastic fibers and hybrid fiber reinforcement may also
E-mail address: [email protected] (K. Tosun-Felekoglu). be employed as cost effective fibers in ECC design, if fiber, matrix

1359-8368/$ - see front matter Ó 2014 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.compositesb.2013.12.076
360 B. Felekoglu et al. / Composites: Part B 60 (2014) 359–370

and interface properties are modified according to micromechani- tures in order to achieve two different matrix flowability. Flowabil-
cal theory proposed by Li [4]. There are many studies regarding the ity was indirectly measured with a modified Marsh cone flow test
utilization of recycled plastic fibers in concrete for the purpose of reported in Li [21]. The orifice diameter of Marsh cone was in-
ductility improvement and cracking control [12–15]. However, creased to 15 mm. The targeted flow times were 15–20 s for high
most of these studies reported tension and/or deflection softening flowability (HF) and 25–35 s for low flowability (LF).
behavior at low fiber dosages (<2% by volume) mostly due to the
incompatible matrix-interface properties to provide robust multi- 2.2. Mixing procedures
ple cracking behavior. There is no doubt that steel fibers signifi-
cantly improve the ductility, post cracking strength and crack Two different cement–mortar mixers with varying mixing
opening behavior of plain concrete up to a certain dosage (usually capacities were used to observe the effects of mixing speed and
between 20 and 80 kg/m3 for workability reason) [16,17]. How- bowl volume on the mix quality. The first mixer is a 1/6 hp coun-
ever, this dosage range is not enough to provide multiple cracking tertop mixer with a 4.7 L bowl volume operating at three different
and strain hardening behavior if a sufficient physical modification mixing speeds (Table 2). The second mixer is a floor mixer
on steel fibers’ geometry (such as twisting) is not performed [18]. equipped with a high torque motor (3/4 hp) and more voluminous
In order to further decrease the cost of ECC, High Tenacity Poly- bowl capacity (28.4 L) than the first. Three different mixing proce-
propylene (HTPP) fibers have been studied in the past as alternatives dures were investigated. Countertop mixer was used for the first
to other fibers by Yang [3]. Recently, HTPP fibers with tensile and second procedures (P1 & P2) with batch volume of 1.5 L. The
strength of 800–900 MPa and diameters of 10–12 lm diameter floor mixer was used for the third mixing procedure (P3) with
are commercially available. Yang [3] developed HTPP-ECC incorpo- batch volume of 5 L. The mixing procedures are shown schemati-
rating 2% HTPP fiber (by total mix volume) using the micromechan- cally in Fig. 1. Mixing sequence was similar for all procedures prior
ics-based design approach, which exhibited tensile strain capacity to fiber addition. After achieving the desired matrix flowability, as
as high as 4% with 2–2.5 MPa ultimate tensile strength. determined by Marsh cone flow test, fibers were gradually added
A major drawback of HTPP-ECC is its difficult processing due to and mixed for 2 min at the same rotational speed in mixing proce-
very high aspect ratio of HTPP fibers. Micromechanical analysis dure P1 (Fig. 1a). In the procedure P2, an additional 1 min of mix-
suggests that a high aspect ratio (length/diameter ratio) is needed ing at the highest rotational speed was applied (Fig. 1b). However,
for fibers, such as HTPP, with low interfacial bond with the matrix countertop mixer overheated in this case which is an indication of
to effectively bridge the micro-cracks in ECC. However, due to their inadequate motor torque for effective mixing. In procedure P3, an
high aspect ratio, HTPP fibers are susceptible to clumping and poor additional 3 min of mixing at the highest rotational speed was ap-
dispersion in a cementitious matrix. At equivalent fiber dosage plied by using the floor mixer with high torque capacity without
(2%) the number of HTPP fibers (12 lm diameter, 10 mm length) any problem (Fig. 1c), suggesting the adequacy of the floor mixer
per cubic meter of ECC is nearly 12.7 times higher than that of torque. After the completion of mixing sequence, the flow-spread
PVA fibers (39 lm diameter, 12 mm length) in PVA-ECC. Due to diameters of fresh HTPP-ECCs were measured with an ASTM C
this reason, proper mixing of HTPP fibers in a cementitious matrix 230 [22] flow-table.
even at comparatively low fiber dosages is a challenging task.
The objective of this study is to enhance the magnitude and 2.3. Specimens preparation, curing and testing
consistency of the composite mechanical properties of HTPP-ECC
developed by Yang [3] by investigating the influences of matrix After the completion of fresh tests, all mixtures were cast into
flowability, fiber mixing procedure, and curing conditions. The ba- dogbone-shaped and 50 mm cube molds moderately vibrated on
sic mechanisms that enhance tensile ductility of HTPP-ECC through a vibration table. Four dogbone-shaped specimens and three cubes
flowability control, mixing procedure modification, and water-cur- were prepared for each mixture. The geometry of dogbone-shaped
ing are discussed from the view point of micromechanics underly- specimens used in this study is shown in Fig. 2, which is in accor-
ing ECC design, with supporting evidence from fiber bridging dance with the JSCE recommendations for HPFRCC specimens [23].
stress–crack width (r–d) relations. Specimens were demolded one day after mixing and subjected to
the following curing regimes:
2. Experimental studies
1. Air curing (A): Specimens stored in laboratory conditions
2.1. Materials and mixture proportions (23 ± 3 °C, 45 ± 5% RH) for 28 days, which is the usual practice
for PVA-ECC specimens.
Type I ordinary Portland cement (OPC) and class F fly ash con- 2. Mixed curing (WA): Specimens kept in water (23 ± 3 °C) for
forming to the requirements of ASTM C150 [19] and ASTM C618 7 days and then stored at room temperature until the age of
[20], respectively, were used in the ECC matrix. The specific gravi- 28 day, at a RH of around 45 ± 5%.
ties of OPC and fly ash are 3.15 and 2.45, respectively. The chemical 3. Water curing (W): Specimens kept in water (23 ± 3 °C) for
analysis of fly ash indicates that it is mainly composed of SiO2 28 days.
(44%), Al2O3 (23%), Fe2O3 (8%), and CaO (14%), with other com-
pounds in trace amounts. 83% by weight of fly ash is finer than Uniaxial tensile tests were performed with a servo-hydraulic
44 lm which indicates high pozzolanic reactivity with the hydra- test system under displacement control. The rate of displacement
tion products of OPC. A polycarboxylate-based high range water was 0.5 mm/min, chosen to simulate a quasi-static loading condi-
reducing (HRWR) admixture was used to achieve flowability. HTPP tion. Two external linear variable displacement transducers
fiber with 12 lm diameter and 10 mm length was used in all mix- (LVDT), with a gage length of approximately 100 mm, were at-
tures. The density, nominal tensile strength, Young’s modulus, and tached to the specimen (Fig. 2). Stress–strain curves were then re-
elongation at rupture of HTPP fibers were 0.91 g/cm3, 850 MPa, corded to determine the behavior of specimens under direct
6 GPa, and 21%, respectively. tension. First crack strength is determined from the point where
The mixture proportion of ECC adapted from Yang [3] is pre- the stress–strain relation becomes non-linear [23]. Peak stress
sented in Table 1. The fly ash/cement ratio is 2.8. At the end of ma- and the corresponding strain value are accepted as ‘‘tensile
trix mixing procedure (prior to fiber addition) a highly flowable strength’’ and ‘‘tensile ductility’’, respectively. In the cases where
matrix was obtained. HRWRA content was varied between mix- slight stress decrease after peak stress is observed, tensile ductility
 B. Felekoglu et al. / Composites: Part B 60 (2014) 359–370 361

Table 1
The mixture proportions of HTPP-ECCs (kg/m3).

Mixture codea Cement (kg/m3) Fly ash (kg/m3) Water (kg/m3) HTPP fiber (kg/m3) HRWR (kg/m3) Target Marsh cone flow time (s)
HF 412 1150 362 18 10.7 15–20
LF 7.4 25–35
a
HF: matrix with high flowability, LF: matrix with low flowability.

Table 2
Properties of mixers used in this study.

Mixer type Motor torque (hp) Mixer bowl volume (L) Speed-1 slow (rpm) Speed-2 medium (rpm) Speed-3 high (rpm)
Hobart countertop mixer 1/6 4.7 60 124 255
Hobart floor mixer 3/4 28.4 54 100 183

Fig. 1. Mixing sequence of procedures applied in this study: (a) P1, (b) P2, (c) P3 (see Table 2 for mixer speed in rpm).

is defined as strain at 95% of the peak stress, instead of ‘‘at peak the specimen width, it is counted as a separate crack. Crack widths
stress’’. Compressive strength tests were performed at a loading (CW) were calculated using the following formula:
rate of 0.28 MPa/s (40 psi/s).
As multiple cracking is the indicator of tensile strain hardening ðL  15Þ cpixel
CWðmmÞ ¼ ð1Þ
capacity of ECC specimens, crack patterns were investigated in this ypixel
study. A high resolution camera was used for photo documentation
of damage evolution during uniaxial tensile testing. Photos were where L is gage length in mm, cpixel is the average crack width of
captured at the beginning of the test and at 1–2–3–4% strain values each crack in pixels; ypixel is the height of captured area between
if specimen reaches these strain values. As shown in Fig. 3, the LVDT attachments in pixels. cpixel is measured by counting the ver-
crack pattern of the focused region (gage length L – 15 mm) was tical pixel number of each crack image. The cracks formed out of
used in analysis. 15 mm is the width of a holder bar in the LVDT gage length were not taken into consideration. This method is only
fixture. Pixel resolution varied between 20 and 45 lm depending applicable to specimens that exhibit crack widths larger than 20 lm
on the gage length (L). The number of cracks is counted with the due to resolution restraints. Crack widths smaller than 20 lm may
following assumption: If a crack is visible for more than half of not be detected with this method.
362 B. Felekoglu et al. / Composites: Part B 60 (2014) 359–370

showed that HF mixtures are more susceptible to bleeding and fi-

ber clumping. With P1 mixing procedure, significant amount of fi-
ber clumping was observed in case of HF mixtures. The degree of
fiber clumping was reduced (although not completely eliminated)
in case of LF mixture, when the same mixing procedure (P1) was
employed. With the application of additional 1 min of high speed
mixing in procedure P2, the degree of fiber clumping was further
reduced for both HF and LF mixtures. In spite of the improvements
in fiber dispersion in procedure P2, the high speed mixing was not
enough to fully breakdown the fiber clumps as the torque capacity
of countertop mixer is limited. Thus, the proper matrix cohesion
obtained in LF mixtures along with fast mixing speed seems to re-
sult in the best fiber dispersion among the above.
Due to above observations, only LF mixtures are prepared fol-
lowing the third mixing procedure (P3). A more homogeneous LF
mixture was obtained by using a floor mixer with high torque
capacity and mixing speed. A spread-flow value as high as
170 mm was recorded with a LF matrix despite its low HRWR
admixture dosage (Fig. 4 P3-LF). Thus, effective mixing provided
by procedure 3 enhanced the flow-diameter of LF mixture and
eliminated fiber clumping with an additional advantage of HRWR
admixture saving.

3.2. Uniaxial tensile performance

3.2.1. Specimens prepared with mixing procedure P1

The tensile stress–strain curves of specimens prepared with P1-
HF and P1-LF mixtures subjected to mixed curing (WA) and water
curing (W) are presented in Fig. 5. Due to significant fiber clumping
Fig. 2. Dimensions and loading configuration of dogbone-shaped specimen for in all mixtures produced with procedure P1, air curing (A) was not
uniaxial tensile testing.
considered in this part of the study as it tends to worsen the
mechanical performance compared to WA and W curing regimes
(discussed below). Average crack widths (CW) along with standard
deviations, and average crack number (CN) of 4 specimens at 1%,
2% and 3% strains are shown at the top of all figures (Fig. 5). In gen-
eral, all specimens showed more than one crack. However, the
crack density (number of cracks per unit length) is very low. The
number of cracks is limited to 5 (HF) and 7 (LF) in case of mixed
curing, and 6 (HF) and 12 (LF) in the case of water curing. Average
crack widths at 1% strain are also large (230 lm and 150 lm for
mixed curing) compared to typical PVA-ECC, which typically
exhibits average crack widths smaller than 60 lm [3].
Test results in terms of first crack strength, ultimate tensile
strength, tensile ductility and average crack width at 1% strain
are shown in Table 3. First crack strength values of HF mixtures
are lower than LF mixtures, which points towards the possibility
of larger maximum flaw size in the HF mixtures compared to LF
mixtures. The ultimate tensile strength values of HF mixtures
are lower than that of LF mixtures due to non-homogenous fiber
dispersion (results with fiber clumping) in HF mixtures. High ma-
Fig. 3. Methodology of average crack width determination from crack pattern
analysis. trix cohesion was found to be advantageous for improving the
tensile ductility in the case of LF mixtures. For LF specimens,
entrapment of air bubbles due to low flowability may possibly in-
3. Results and discussion crease the number of flaws for crack propagation in the right size
range.
3.1. Fresh state tests Water cured specimens also performed better than the speci-
mens with mixed curing in terms of tensile ductility. The highest
The spread-flow diameters of fiber reinforced HF and LF mix- average ductility value (2.64%) among this set of specimens is ob-
tures prepared with mixing procedure P1 and P2 were photo- tained from water cured specimens prepared with a cohesive ma-
graphed and presented in Fig. 4. In general, spread flow trix. The coefficients of variation in tensile ductility are very high
diameters of fiber reinforced mixtures are proportional to the ma- for all specimens (15–53%), which can be attributed to the fiber
trix flowability. This is observable in Fig. 4 as the HF mixtures show clumping as a result of insufficient mixing (Table 3). Overall, LF
larger spread flow diameters than LF mixtures for both mixing mixtures with water curing had better tensile performance, partic-
procedures P1 and P2. Although better flowability is generally pref- ularly tensile ductility, compared to HF mixtures with mixed
erable for moldability and practical use, visual observations water–air curing.
 B. Felekoglu et al. / Composites: Part B 60 (2014) 359–370 363

Fig. 4. The spread-flow diameters of fiber reinforced HF and LF mixtures prepared with different mixing procedures (see Fig. 1 for P1, P2 and P3 procedures).

3.2.2. Specimens prepared with mixing procedure P2 curing have positive effects on the tensile performance of HTPP-
As explained above, mixing procedure P2 is essentially P1 plus ECC.
additional 1 min of mixing at high speed (Fig. 1), which signifi-
cantly improved the fiber dispersion in both high and low flowable 3.2.3. Specimens prepared with mixing procedure P3
mixtures. Two curing regimes, air curing (A) and water curing (W) Unlike the aforementioned procedures (P1 and P2), in which the
(explained in Section 2.3), were applied to the mixtures cured with batch volume was approximately 1.5 L and the mixer torque
procedure P2. Similar to P1, both mixtures with high (HF) and low capacity was limited, a batch volume of 5 L with additional mixing
(LF) flowability were produced for each curing regime, resulting in time and greater torque were used in Procedure P3 (Fig. 1). In this
a set of four mixtures. procedure, only high cohesion matrix (low flowability) was pre-
The tensile stress–strain curves and results of crack pattern pared. 12 dog-bone specimens were prepared and subjected to
analysis for mixtures produced with procedure P2 are presented three different curing conditions: air curing (A), mixed curing
in Fig. 6. Compared to the P1 series, improvements in the ultimate (WA) and water curing (W) (explained in Section 2.3).
tensile strength values of all P2 series specimens are observed, The tensile stress–strain curves and crack pattern analysis of
which indicates better fiber dispersion in P2 specimens than P1 the P3 series of specimens are presented in Fig. 7. In addition to
specimens according to the principles of micromechanics. Another the stress–strain curves, crack pattern photograph of P3-LF-W-4
important difference between the tensile performances arises from specimen at 4% strain is also shown in the same figure. In this pho-
different curing regimes. Water curing significantly improved the tograph, 28 cracks are counted on this specimen with an average
multiple cracking performances and tensile ductility of specimens crack spacing of about 3 mm under 3.5 MPa tensile stress. This
compared to air-cured ones. The average number of cracks in- specimen represents the highest crack density achieved in this
creased from 4 (Fig. 7b) to 25 (Fig. 7d), the average crack widths study. According to crack pattern analysis, average number of
at 1% strain decreased from 258 to 74 lm in LF mixtures when cracks successively increase from 7 to 13 to 23 and the average
the curing regime was changed from air to water. Similar improve- crack width at 1% strain decreases from 95 lm to 88 lm to
ments are observable in HF mixtures. Overall, additional mixing 62 lm as the curing regime is changed from A to WA to W. In most
time at fast speed in P2 compared to P1 and water curing instead cases, crack width variability at any given strain is also reduced
of air curing improved the multiple cracking performance of with water curing. Tables 3 and 4 clearly show the decrease in
HTPP-ECC. coefficient of variations of crack width by water curing. These re-
Tensile properties of HTPP-ECCs (Fig. 6) produced with proce- sults reaffirm the conclusion from P1 and P2 series of specimens
dure P2 are summarized in Table 4. Higher first crack strengths that extending the water curing stage enhances the robust multi-
and ultimate tensile strengths (up to about 4.25 MPa for LF speci- ple cracking performance of HTPP-ECC.
mens) were recorded for the P2 series in comparison to the P1 ser- Tensile properties of HTPP-ECC prepared with P3 procedure are
ies. From Table 4, it can be observed that curing causes no summarized in Table 5. Tensile ductility tends to increase with
significant difference between ultimate tensile strength values of longer water curing; however, tensile strength remains nearly
ECC. On the other hand, the average tensile ductility of water- the same. Tensile strain capacity as high as 3.91% was recorded
cured specimens (1.96% and 3.23%) is significantly higher than that from specimens (P3-LF-W) cured in water, while average tensile
of air-cured specimens (0.80% and 0.81%). While high matrix cohe- strength of these specimens was 3.6 MPa. A slight decrease in ulti-
sion (comparatively low flowability) enhanced the tensile ductility mate tensile strength of P3 specimens compared to P2 specimens
of water-cured specimens (from 1.96% to 3.23%), no such improve- was observed, which may be attributed to air entraining effect of
ment was observed in the case of air-cured specimens. The benefi- fibers and HRWR due to prolonged mixing. Coefficients of variation
cial effect of curing on tensile ductility can be attributed to the in tensile ductility of P3 specimens were equal or relatively lower
improved bond between HTPP fiber and matrix due to the forma- than that for the specimens prepared with P2 mixing procedure
tion of additional hydration products in the vicinity of HTPP fibers. (Table 5). Overall, P3 procedure enhanced the tensile performance
Loss of internal curing water at low humidity in the case of air cur- of HTPP-ECC particularly in terms of tensile ductility and consis-
ing may interrupt the formation of hydration products at the tran- tency of properties.
sition zone between matrix and HTPP fibers. According to Table 4,
coefficients of variation in tensile ductility are lower for specimens 3.3. Compressive strength performance
prepared with low flowable matrix compared to ECC with high
flowable matrix. Thus, longer mixing time (at higher speed) in pro- Average compressive strength values of 50 mm cube specimens
cedure P2 (than P1), relatively high matrix cohesion, and water prepared with various matrix flowability, mixing procedure, and
364 B. Felekoglu et al. / Composites: Part B 60 (2014) 359–370

Fig. 5. Tensile stress–strain curves of specimens prepared with P1-HF and P1-LF mixtures subjected to mixed curing (WA) and water curing (W) (CW: average crack width
and CN: average crack number at designated strain).

curing conditions are plotted in Fig. 8. LF mixtures showed slightly cured in air (between 30 and 40 MPa). The degree of cement hydra-
lower compressive strengths than HF mixtures (for P1 and P2 ser- tion is limited in the absence of water curing, as the mixtures have
ies), which is attributable to air entrainment during mixing of a rel- a low water/cementitious material (w/cm) ratio of only 0.23. In
atively cohesive mixture, similar to the observations in a previous addition, micro-cracking induced by drying shrinkage negatively
research [21] on PVA-ECC. affects the compressive strength values. It is interesting to note
Among various curing conditions, the lowest compressive that, in spite of a decrease in compressive strengths of air-cured
strength values were recorded for the specimens (P2 and P3 series) mixtures, their tensile strength was approximately similar to that
 B. Felekoglu et al. / Composites: Part B 60 (2014) 359–370 365

Table 3
Tensile properties of specimens prepared with mixing procedure P1 and subjected to mixed curing (WA) and water curing (W) conditions.

Mixture code First crack strength (MPa)a Ultimate tensile strength (MPa)a Tensile ductility (%)a Average crack width at 1% strain (lm)a
P1-HF-WA 1.47 ± 0.39 (27%) 2.30 ± 0.23 (10%) 0.80 ± 0.42 (53%) 222 ± 80 (35%)
P1-LF-WA 2.12 ± 0.60 (28%) 2.41 ± 0.25 (10%) 1.61 ± 0.25 (15%) 231 ± 74 (32%)
P1-HF-W 1.84 ± 0.46 (25%) 2.49 ± 0.18 (7%) 1.60 ± 0.33 (20%) 179 ± 29 (16%)
P1-LF-W 2.02 ± 0.06 (3%) 2.58 ± 0.19 (7%) 2.64 ± 1.04 (39%) 112 ± 20 (18%)
a
Average ± standard deviation (coefficient of variation).

Fig. 6. Tensile stress–strain curves of specimens prepared with P2-HF and P2-LF mixtures subjected to air curing (A) and water curing (W), (CW: average crack width and CN:
average crack number at designated strain).
366 B. Felekoglu et al. / Composites: Part B 60 (2014) 359–370

of the water-cured mixtures (Tables 3–5); however the tensile duc- the fiber bridging capacity r0 should be greater than the matrix
tility of air-cured mixtures was significantly lower than that of cracking strength rcr on any given crack plane. The second
water-cured mixtures for the reasons discussed in detail in the condition (Energy Criterion – Eq (2)) requires that the crack tip
numerical micromechanics-based analysis proposed in Section 4. toughness Jtip should be less than the complementary energy J 0b
The influence of mixing procedure on compressive strength val- of fiber bridging (Fig. 9):
ues can be compared under constant matrix flowability and curing Z d0
conditions. For instance, improvement of mixing effectiveness J tip 6 r0 d0  rðdÞdd  J0b ð2Þ
(from P1 to P3) successively increased the compressive strength 0
values from about 40 MPa up to 70 MPa in case of LF-W specimens
Due to the random nature of preexisting flaw size and fiber distri-
(Fig. 8). Mix homogeneity through more effective mixing and im-
bution in ECC, a large margin between J 0b and Jtip is preferred for high
proved bond between HTPP fiber and matrix by water curing
tensile ductility. Additionally, from the view point of strength crite-
(which stabilizes the axial splitting crack) were found to be the
rion a larger r0/rcr(fc) ratio is required for saturated multiple crack-
most dominant factors influencing the compressive strengths.
ing. rcr(fc) is the cracking stress required for the initiation of the first
crack in the ECC matrix. As a result, pseudo strain-hardening (PSH)
4. Micromechanical modeling and application of fiber bridging J0
model to HTPP-ECCs cured at different conditions performance indices (defined as PSHenergy ¼ Jtipb and PSHstrength ¼ rr0c )
have been used to quantitatively evaluate the two margins
4.1. Basics of micromechanical modeling [26,27]. Materials with larger PSH indices have better chance of sat-
urated multiple cracking.
According to the micromechanics based design theory of ECC The PSHstrength can be improved by increasing the fiber bridging
two necessary conditions must be satisfied to ensure strain-hard- capacity r0 through the use of a higher fiber dosage; however, it
ening [24,25]. The first condition (Strength Criterion) requires that leads to higher cost, potentially poor workability, and inhomoge-

(a)

(b)

(c) (d)
Fig. 7. Tensile stress–strain curves of specimens prepared with P3-LF mixtures subjected to (a) air curing (A), (b) mixed curing, (c) water curing (W), (d) Crack pattern of P3-
LF-W-1 specimen at 4% strain.
 B. Felekoglu et al. / Composites: Part B 60 (2014) 359–370 367

Table 4
Tensile properties of specimens prepared with mixing procedure P2 and subjected to air curing (A) and water curing (W) conditions.

Mixture code First crack strength (MPa)a Ultimate tensile strength (MPa)a Tensile ductility (%)a Average crack width at 1% strain (lm)a
P2-HF (A) 3.26 ± 0.52 (16%) 4.05 ± 0.30 (7%) 0.81 ± 0.52 (64%) 116 ± 33 (28%)
P2-LF (A) 2.74 ± 0.58 (21%) 4.29 ± 0.33 (8%) 0.80 ± 0.31 (39%) 258 ± 146 (57%)
P2-HF (W) 1.98 ± 0.43 (22%) 3.79 ± 0.15 (4%) 1.96 ± 1.05 (53%) 95 ± 24 (25%)
P2-LF (W) 2.67 ± 0.62 (23%) 4.25 ± 0.13 (3%) 3.23 ± 0.53 (16%) 74 ± 14 (19%)
a
Average ± standard deviation (coefficient of variation).

Table 5
Tensile properties of P3-LF specimens exposed to air, air–water and water curing.

Mixture code First crack strength (MPa)a Tensile strength (MPa)a Tensile ductility (%)a Average crack width at 1% strain (lm)a
P3-LF-A 2.37 ± 0.32 (13%) 3.64 ± 0.26 (7%) 1.91 ± 0.47 (25%) 95 ± 22 (23%)
P3-LF-WA 1.52 ± 0.49 (32%) 3.81 ± 0.30 (8%) 2.95 ± 0.35 (12%) 85 ± 15 (18%)
P3-LF-W 2.19 ± 0.24 (11%) 3.64 ± 0.07 (2%) 3.91 ± 0.67 (17%) 63 ± 13 (21%)
a
Average ± standard deviation (coefficient of variation).

neous fiber dispersion. Using a fiber with high tensile strength can strength reduction factor (f0 ) are closely related to fiber/matrix fric-
also improve r0 but it increases the cost of the composite. Control- tional bond, which is low for HTPP fibers. The values proposed by
0
ling the fiber/matrix interfacial bond properties in a manner that Yang [3] as f = 0.39 and f = 0.1 are used in this study. The Cook-
utilizes the fibers most effectively seems to be a more appropriate Gordon parameter was determined by using the assumption of
solution [4]. a = 15.df (15  12 lm = 180 lm), which is appropriate for polypro-
The PSHenergy can be improved by decreasing Jtip through the pylene fiber-reinforced concrete [30]. Matrix parameters Em and
reduction of matrix fracture toughness Km or by using a larger flaw rm were determined from tensile tests of dogbone shaped speci-
size c. However, excessive lowering of Km or large flaw size c may mens. The initial slope of the tensile stress–strain curve of the
lead to low compressive strength and low first crack strength un- composite was used as an approximate estimate for Em. Matrix
der tension [4]. spalling coefficient k = 500 is chosen similar to that used in [3],
as the observed micro-spall size was under 30 lm during the direct
4.2. Application of fiber bridging model to HTPP-ECCs cured at tension tests. These parameters were used as inputs in the micro-
different conditions

The fiber-bridging constitutive law r(d) describes the relation-

ship between the bridging stress r transferred across a crack by
the fibers and the opening of this crack d. The construction of the
fiber-bridging constitutive law starts from micromechanics-based
modeling a single fiber pull-out behavior against the surrounding
matrix. The r(d) relationship can then be obtained by averaging
the contributions from fibers with different embedment lengths
and orientations across the crack plane [7,28]. This relationship re-
lates the material microstructure (micro-scale) to the composite
tensile strain hardening behavior (macro-scale). Therefore, con-
trolling the r(d) curve through tailoring material microstructure
is the key to successfully design ECC material properties in general,
and tensile properties (tensile strain capacity, ultimate tensile
Fig. 8. Average compressive strength values of 50 mm cube specimens exposed to
strength, and steady-state crack width) in particular [3].
different curing conditions (Fig. 1 and Section 2.3 for notation details).
As observed in the experimental results in the previous section,
different curing conditions can modify the composite mechanical
behavior of HTPP-ECC which is further dependent on the fiber
bridging performance (r(d)). In this section, the influence of curing
conditions on the r(d) relationship of HTPP-ECC is analytically
investigated.
A MATLAB code using the numerical procedure for computing
the r(d) relationship of ECC developed by Yang [3] was used for
this micromechanics-based analysis. The micromechanical param-
eters used as model input for HTPP-ECCs cured at different condi-
tions are listed in Table 6. Fiber properties; df, lf, Ef and rf are
provided by the manufacturer. Interface parameters; s0, Gd and b
were measured by single fiber pull-out tests [29]. In that investiga-
tion, P3-LF matrix and three curing regimes (A, WA and W) were
used. The coefficients of variations of s0 were 34%, 31% and 26%,
for air, mixed and water cured specimens respectively. Gd and b Fig. 9. Typical r(d) curve for strain-hardening material (dark shaded area repre-
values were very low compared to PVA fibers, however, they were sents right-hand side of Eq. (1), light hatched area represents complementary
also introduced to the model. The snubbing coefficient (f) and energy of composite) (adapted from Wang [2]).
368 B. Felekoglu et al. / Composites: Part B 60 (2014) 359–370

Table 6
The micromechanics parameters used as model input for HTPP-ECCs cured at different conditions.

Fiber parameters
df(lm) lf(mm) Ef(GPa) rf(MPa)
12a 10a 6a 850a
Curing condition Interface parameters Matrix parameters
0
s0(MPa) Gd(J/m2) b f f a(lm) Em(GPa) rm(MPa) k
Air curing 0.48 b 0.0070 b 0.0043b 0.39c 0.1c 180c 12.7 b 1.0 ± 0.3b 500c
Mixed curing 0.99b 0.0051b 0.0020b 13.0b 1.5 ± 0.3b
Water curing 1.27b 0.0032 b 0.0044 b 13.3b 2.3 ± 0.7b
a
Test results provided by manufacturer.
b
Test result values.
c
Assumed values.

fiber distribution and the assumptions of uniform 2D distribution

and a uniform fiber volume fraction of 2% used in the model.
Crack opening at bridging capacity decreased with water curing
according to r(d) curves presented in Fig. 10. The main reason of
this improvement can be attributed to the fiber/matrix interfacial
frictional bond (s0). From Table 6, it can be observed that
s0 = 0.48, 0.99 and 1.27 MPa for A, WA, and W curing regimes,
respectively. In general, similar decreasing trend for crack widths
was experimentally observed (Figs. 6 and 7).
In order to calculate the crack tip toughness (J tip ¼ K 2m =Em ) for
evaluating the PSHenergy index for multiple cracking, matrix frac-
ture toughnesses (Km) of P3-LF matrices cured at different condi-
tions were determined experimentally according to ASTM E399
[34] standard. Although this standard based on Linear Elastic Frac-
ture Mechanics (LEFM) is proposed for metal testing, it is also
applicable for testing brittle mortar matrices due to small scale
yielding near the crack tip [35]. Matrix phase of P3-LF mixture
Fig. 10. Model predicted r(d) curves of HTPP-ECCs cured at different conditions. without fiber was prepared by using the same procedure described
in previous sections. Four beam specimens with dimensions
mechanical model to determine the r(d) relation of HTPP-ECC 305  76  38 mm3 were prepared for each curing regime. Beam
cured under various regimes. specimens were notched with notch-depth to total beam height ra-
Relevant results of this numerical analysis consist of r(d) curves tio of 0.60 prior to flexural loading and the span length was
between 0-600 lm crack opening (Fig. 10), computed complemen- 254 mm. The matrix fracture toughness of air, mixed, and water
tary energy of fiber bridging: J 0b (J/m2), crack opening at bridging cured specimens were determined equal to 0.44 ± 0.02,
capacity: d0 (lm), and fiber bridging capacity: r0 (MPa) (Table 7). 0.36 ± 0.03, and 0.32 ± 0.05 MPa m0.5 respectively, using the ASTM
According to these analytical results, r0 ranges between E399 calculations (Table 7).
4.1 MPa and 5.2 MPa, which is larger than the experimental ulti- PSH indices of specimens subjected to different curing condi-
mate tensile strength values (Table 4: P2 series: and 3.8–4.3 MPa, tions were calculated with the model output data and the fracture
Table 5: P3 series: 3.6–3.8 MPa). Due to the inherent inhomogene- toughness test data, and the results are shown in Table 7. The com-
ity of fiber dispersion in the ECC and other fiber reinforced mix- puted PSH indices are plotted against observed tensile ductility in
tures, various specimen cross sections are bridged by different Fig. 11. Kanda [27] proposed threshold PSH index values for
number of fibers [31,32]. As a result, the final failure occurs at achieving robust strain hardening accounting for the inhomogene-
the crack with least number of effective fibers. One of the condi- ity in the material. Based on experimental data, Kanda [27] pro-
tions of multiple cracking, therefore, requires the stresses at which posed these practical thresholds for PVA and PP fibers:
cracks initiate to be smaller than the maximum bridging capacity PSHstrength = 1.45 (PVA), and 2 (PP), PSHenergy = 3 (for both). Accord-
of the least bridged crack [33]. Thus, the difference between the ing to Table 7 all specimens pass the PSHenergy threshold (Table 7 –
values of r0 and experimental ultimate tensile strengths of marked in bold) which means that they have the potential of mul-
HTPP-ECC may be attributed to the difference between the actual tiple cracking. However, air cured specimens failed to pass the PSH

Table 7
Calculation of energy and strength criterion from model output and relevant test results.

Specimen Model output Jtip(J/ First crack PSH Index PSH index Tensile
curing m2) strengtha rfc (energy) (strength) r0/ ductilitya
Fiber bridging Crack opening at Fiber bridging
condition (MPa) J0b =Jtip rfc (%)
strength: r0 bridging capacity: d0 complementary energy J0b
(MPa) (lm) (J/m2)
Air curing 4.16 216 217 15.2 2.37 14.3 1.8 1.90
Mixed curing 5.00 115 145 10.0 1.52 14.5 3.3 2.95
Water curing 5.22 93 121 7.8 2.19 15.5 2.4 3.91
a
From Table 5.
 B. Felekoglu et al. / Composites: Part B 60 (2014) 359–370 369

Energy criterion Strength criterion Tensile ductility

20 4

(energy & strength)

15.5

Tensile ductility (%)

14.3 14.5
15 3

PSH indices
10 2

5 3.3 1
1.8 2.4

0 0
air curing mixed curing water curing

Fig. 11. Relationship between PSH indices and tensile ductility of HTPP-ECCs cured at different conditions.

strength index threshold due to their comparatively high first crack possible to conduct this research at the University of Michigan.
strength. Low tensile ductility of air-cured specimens from direct Materials supply from Lafarge (cement), Headwaters (fly ash),
tension tests confirmed the applicability of these index limits. WR Grace (HRWRA), and Saint-Gobain Brasil (HTPP fiber) is also
Results of experimental and analytical studies showed that gratefully acknowledged.
water curing, compared to air curing, increased the fiber bridging
capacity and decreased the matrix toughness. The larger gap be-
tween first crack strength and ultimate strength, in the case of References
water curing, enables the activation of more smaller flaws within
the matrix and results in an improved multiple cracking behavior [1] Zhou J, Qian S, Ye Q, Copuroglu O, Breugel K, Li VC. Improved fiber distribution
and tensile ductility. and mechanical properties of engineered cementitious composites by
adjusting the mixing sequence. Cement Concr Compos 2012;34:342–8.
[2] Wang S. Micromechanics based matrix design for engineered cementitious
composites. The University of Michigan, PhD Thesis (Civil Engineerin g); 2005.
5. Conclusions 222p.
[3] Yang E-H. Designing added functions in engineered cementitious composites.
 This study showed that the processing details of ECC are as The University of Michigan, PhD Thesis (Civil Engineering); 2007. 276p.
[4] Li VC. Tailoring ECC for special attributes: a review. Int J Concr Struct Mater
important as the selection of proper ingredients and mix pro- 2012;6(3):135–44.
portions. Control of matrix flowability, efficient mixing, and [5] Li VC. From micromechanics to structural engineering – the design of
water curing are key factors for achieving robust strain harden- cementitious composites for civil engineering applications. JSCE J Struct
Mech Earthquake Eng 1993;10(2):37–48.
ing and multiple cracking in HTPP-ECC.
[6] Li VC. Can concrete be bendable? J Am Sci 2012;100(6):484–93. November–
 According to the test results, matrix cohesion should be high December.
enough to facilitate fiber dispersion. Modified Marsh cone flow [7] Lin Z, Li VC. Crack bridging in fiber reinforced cementitious composites with
slip-hardening interfaces. J Mech Phys Solids 1997;45(5):763–87.
time of 25–35s is found more appropriate. Matrix with high
[8] Redon C, Li VC, Wu C, Hoshiro H, Saito T, Ogawa A. Measuring and modifying
flowability (15–20s) causes fiber clumping and resulted in poor interface properties of PVA fibers in ECC matrix. ASCE J Mater Civil Eng
strain hardening performance. However, cohesive matrix itself 2001;13(6):399–406. November/December.
does not guarantee homogenous material production and ten- [9] Li VC, Wang S, Wu C. Tensile Strain-hardening Behavior of PVA-ECC. ACI Mater
J 2001;98(6):483–92. November–December.
sile ductility, as inefficient mixing and improper curing condi- [10] Li VC, Wu C, Wang S, Ogawa A, Saito T. Interface tailoring for strain-hardening
tions may negatively influence the performance of HTPP-ECC. PVA-ECC. ACI Mater J 2002;99(5):463–72. September–October.
 Ultimate tensile strength and ductility values of HTPP-ECC [11] Li VC. Engineered Cementitious Composites (ECC) – tailored composites
through micromechanical modeling’’. In: Banthia N, Bentur A, Mufti A, editors.
improved in magnitude and consistency with changes in mixing Fiber reinforced concrete: present and the future. Canadian Society of Civil
procedure. Fiber clumping was significantly reduced with Eng, Montreal; 1998. p. 64–97.
longer mixing time, higher mixing speed, and a mixer with [12] Ochi T, Okubo S, Fukui K. Development of recycled PET fiber and its application
as concrete-reinforcing fiber. Cement Concr Compos 2007;29:448–55.
higher torque. [13] Kim SB, Yi NH, Kim HY, Kim J-HJ, Song Y-C. Material and structural
performance evaluation of recycled PET fiber reinforced concrete. Cement
Unlike PVA-ECC, water curing is beneficial for tensile ductility Concr Compos 2010;32:232–40.
[14] Fraternali F, Ciancia V, Chechile R, Rizzano G, Feo L, Incarnato L. Experimental
in HTPP-ECC due to the improved HTPP fiber–matrix interfacial
study of the thermo-mechanical properties of recycled PET fiber-reinforced
frictional bond strength (s0). Micromechanical analysis showed concrete. Compos Struct 2011;93:2368–74.
that higher s0 causes an increase in fiber bridging strength (r0). [15] Fraternalli F, Farina I, Polzone C, Pagliuca E, Feo L. On the use of R-PET strips for
Although the increase in s0 also reduces the complementary en- the reinforcement of cement mortars. Composites: Part B 2013;46:207–10.
[16] Michels J, Waldmann D, Maas S, Zürbes A. Steel fibers as only reinforcement
ergy (J 0b ), the PSHenergy is still high enough (due to low Jtip) for for flat slab construction – experimental investigation and design. Constr Build
ensuring strain hardening. Mater 2012;26:145–55.
[17] Caggiano A, Cremona M, Faella C, Lima C, Martinelli E. Fracture behavior of
concrete beams reinforced with mixed long/short steel fibers. Constr Build
Acknowledgements Mater 2012;37:832–40.
[18] Naaman AE. Engineered steel fibers with optimal properties for reinforcement
of cement composites. J Adv Concr Technol 2003;1(3):241–52.
This study was conducted at Advanced Civil Engineering – [19] ASTM C150. Standard specification for Portland cement. Annual Book of ASTM
Materials Research Laboratory (ACE-MRL: http://ace-mrl.engin.u- Standards, PA, USA; 2012.
[20] ASTM C618. Standard specification for coal fly ash and raw or calcined natural
mich.edu/), University of Michigan, Ann Arbor, MI. Financial sup-
Pozzolan for use in concrete. Annual Book of ASTM Standards, PA, USA; 2012.
port for the first author provided by TUBITAK (The Scientific and [21] Li M. Multi-scale design for durable repair of concrete structures. The
Technological Research Council of Turkey) under the Grant No. University of Michigan, PhD Thesis (Civil Engineering); 2009. 425p.
2219 (International post-doctoral research fellowship programme) [22] ASTM C230. Standard specification for flow table for use in tests of hydraulic
cement. Annual Book of ASTM Standards, PA, USA; 2008.
is acknowledged. The second author is grateful to YOK (The Council [23] JSCE. Recommendations for design and construction of high performance fiber
of Higher Education, Turkey) for the fellowship, which made it reinforced cement composites with multiple fine cracks (HPFRCC). Concrete
370 B. Felekoglu et al. / Composites: Part B 60 (2014) 359–370

Committee, Rokugo K. (chair), Japan Society of Civil Engineers, March; 2008. [30] Li VC, Stang H, Krenchel H. Micromechanics of crack bridging in fiber
212p. reinforced concrete. J Mater Struct 1993;26:486–94.
[24] Li VC, Leung CKY. Steady state and multiple cracking of short random fiber [31] Caggiano A, Etse G, Martinelli E. Zero-thickness interface model formulation
composites. ASCE J Eng Mech 1992;188(11):2246–64. for failure behavior of fiber-reinforced cementitious composites. Comput
[25] Li VC. Damage tolerance of engineered cementitious composites. In: Karihaloo Struct 2012;98–99:23–32.
BL, Mai YW, Ripley MI, Ritchie RO, editors. Advances in fracture research, proc, [32] Tosun-Felekoğlu K, Felekoğlu B, Ranade R, Lee BY, Li VC. The role of flaw size
9th ICF conference on fracture, Sydney, Australia, Pub. Pergamon, UK, p. 619– and fiber distribution on tensile ductility of PVA-ECC. Compos B Eng
30. 2014;56:536–45.
[26] Kanda T, Li VC. Practical design criteria for saturated pseudo strain hardening [33] Ranade R, Stults MD, Lee B, Li VC. Effects of fiber dispersion and flaw size
behavior in ECC. J Adv Concr Technol 2006;4(1):59–72. distribution on the composite properties of PVA-ECC. In: Parra-Montesinos GJ,
[27] Kanda T. Design of engineered cementitious composites for ductile seismic Reinhardt HW, Naaman AE, editors. Proc, sixth int’l workshop on high
resistant elements. PhD dissertation, department of civil and environmental performance fiber reinforced cement composites (HPFRCC6), Ann Arbor, MI;
engineering, University of Michigan, Ann Arbor; 1998. 329 p. 2011. p. 106–113.
[28] Li VC, Wang Y, Backer S. A micromechanical model of tension-softening and [34] ASTM E399. Standard test method for plane-strain fracture toughness of
bridging toughening of short random fiber reinforced brittle matrix metallic materials. Annual Book of ASTM Standards, PA, USA; 1997.
composites. J Mech Phys Solids 1991;39(5):607–25. [35] Li VC, Mishra DK, Wu HC. Matrix design for pseudo strain-hardening fiber
[29] Felekoglu B. Development of cost-effective strain hardening engineered reinforced cementitious composites. RILEM J Mater Struct 1995;28(183):5.
cementitious composites. Tubitak-2219: international post doctoral research
fellowship programme, 2. Progress, Report; 2013. 41p.