Cement & Concrete Composites 34 (2012) 342–348

Contents lists available at SciVerse ScienceDirect

Cement & Concrete Composites

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

Improved fiber distribution and mechanical properties of engineered

cementitious composites by adjusting the mixing sequence
Jian Zhou a,⇑, Shunzhi Qian b, Guang Ye c,d, Oguzhan Copuroglu c, Klaas van Breugel c, Victor C. Li e
a
Sinoma Research Institute, Sinoma International Engineering Co., Ltd., Beijing, China
b
School of Transportation, Southeast University, Nanjing, China
c
Microlab, Faculty of Civil Engineering and Geosciences, Delft University of Technology, Delft, The Netherlands
d
Magnel Laboratory for Concrete Research, Department of Structural Engineering, Ghent University, Technologiepark-Zwijnaarde 904, B-9052 Ghent (Zwijnaarde), Belgium
e
Department of Civil and Environmental Engineering, University of Michigan, USA

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

Article history: Engineered cementitious composites (ECC) is a class of ultra ductile fiber reinforced cementitious com-
Received 12 November 2010 posites, characterized by high ductility and tight crack width control. The polyvinyl alcohol (PVA) fiber
Received in revised form 26 November 2011 with a diameter of 39 lm and a length of 6–12 mm is often used. Unlike plain concrete and normal fiber
Accepted 28 November 2011
reinforced concrete, ECC shows a strain-hardening behavior under tensile load. Apart from the mix
Available online 6 December 2011
design, the fiber distribution is another crucial factor for the mechanical properties of ECC, especially
the ductility. In order to obtain a good fiber distribution, the plastic viscosity of the ECC mortar before
Keywords:
adding fibers needs to be controlled, for example, by adjusting water-to-powder ratio or chemical admix-
Engineered cementitious composites
Mixing sequence
tures. However, such adjustments have some limitations and may result in poor mechanical properties of
Fiber distribution ECC. This research explores an innovative approach to improve the fiber distribution by adjusting the
Tensile strain capacity mixing sequence. With the standard mixing sequence, fibers are added after all solid and liquid materials
High volume of sand are mixed. The undesirable plastic viscosity before the fiber addition may cause poor fiber distribution
and results in poor hardened properties. With the adjusted mixing sequence, the mix of solid materials
with the liquid material is divided into two steps and the addition of fibers is between the two steps. In
this paper, the influence of different water mixing sequences is investigated by comparing the experi-
mental results of the uniaxial tensile test and the fiber distribution analysis. Compared with the standard
mixing sequence, the adjusted mixing sequence increases the tensile strain capacity and ultimate tensile
strength of ECC and improves the fiber distribution. This concept is further applied in the development of
ECC with high volume of sand.
Ó 2011 Elsevier Ltd. All rights reserved.

1. Introduction durability [4–6] compared with standard concrete. The use of ECC
can prolong the service life of structures and reduce the mainte-
ECC, short for engineered cementitious composites, is a class of nance and repair cost. Nowadays ECC is emerging in broad applica-
ultra ductile fiber reinforced cementitious composites invented in tions, such as ECC link slab on bridge decks [7], ECC coupling beam
the early 1990s [1]. The polyvinyl alcohol (PVA) fiber with a diam- in high-rise buildings to enhance their seismic resistance, compos-
eter of 39 lm and a length of 6–12 mm is often used. This group of ite ECC/steel bridge deck and some concrete repair applications [8].
materials is characterized by high ductility and tight crack width. This unique tensile strain-hardening behavior results from an
Fig. 1 shows a typical tensile stress–strain curve of ECC and its tight elaborate ECC design methodology, which uses a micromechanics
crack width control [2]. Unlike plain concrete and normal fiber model. This micromechanics model takes into account the interac-
reinforced concrete, ECC shows a tensile strain-hardening behavior tions among fiber, matrix and fiber–matrix interface [9]. Apart
after first cracking and develops a high tensile ductility in the range from the mix design, the fiber distribution is another crucial factor
of 3–7%. The crack width of ECC is self-controlled to around 60 lm for the mechanical properties of ECC, especially the ductility.
without the presence of steel reinforcement, which is much smal- Torigoe et al. [10] found a strong correlation between the fiber dis-
ler than the typical crack width observed in the steel reinforced tribution and the tensile strain capacity of ECC. A poor fiber distri-
concrete. ECC thus shows lower water permeability [3] and better bution leads to a decrease in ductility and even absence of the
strain-hardening behavior. In order to improve the fiber distribu-
⇑ Corresponding author. tion, a high plastic viscosity of the fresh ECC mortar before the fiber
E-mail address: [email protected] (J. Zhou). addition is desirable [11]. For an adequate plastic viscosity, the

0958-9465/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.cemconcomp.2011.11.019
 J. Zhou et al. / Cement & Concrete Composites 34 (2012) 342–348 343

(1) Mix proper amount of (2) Proper amount liquid

powder materials materials is added and mixed
Fig. 1. Tensile stress–strain curve and tight crack width control of ECC [2].

water-to-powder (w/p) ratio of ECC mortar before mixing fibers

needs to be controlled. However, the low w/p ratio could cause
the increase in the yield stress of the fresh ECC and thus the loss
of workability. Moreover, the low w/p ratio results in a strong
ECC matrix, causing a decrease in ductility [11]. The ECC mix de-
sign is, therefore, limited in a small range of w/p ratio, mainly from
0.2 to 0.3 [11]. This is similar for solid materials. For instance, since
the addition of aggregates can cause balling of fibers at mixing and
poor fiber distribution [12].
This research explores an innovative approach, focusing on
adjusting the ECC mixing sequence to improve the fiber distribution.
As shown in Fig. 2, the standard mixing sequence is carried out as fol-
lows: (1) all solid and liquid materials are first mixed for several
minutes; (2) fibers are added and mixed. The plastic viscosity of (4) The rest powder and (3) Fibers are added and
ECC mortar before the fiber addition determines the fiber distribu- liquid materials are added mixed
tion in matrix. Consequently, if the fresh properties do not reach
Fig. 3. Adjusted ECC mixing sequence.
the condition for a good fiber distribution, although the mix design
is appropriate, ECC will not show good hardened properties [11].
With the adjusted mixing sequence, the mix of solid materials with studied as an example to demonstrate the advantages of the ad-
the liquid material is divided into two steps and the addition of fibers justed mixing sequence. The influence of different mixing se-
is between the two steps. As shown in Fig. 3, proper amounts of pow- quences is investigated by comparing the experimental results of
der and liquid materials are first mixed, aiming at a desirable plastic uniaxial tensile test and fiber distribution analysis. Then, the appli-
viscosity for the good fiber distribution [11]. Then, fibers are added. cation of the adjusted mixing sequence in the development of
After fibers are mixed homogenously, the rest powder and/or liquid high-volume sand ECC (HVS-ECC) is presented.
materials are added to tailor the workability and the hardened prop-
erties of ECC. As a result, the fiber distribution is improved and ECC 2. Experimental program
shows robust fresh and hardened properties.
This paper aims to investigate the improved fiber distribution 2.1. Materials
by adjusting the mixing sequence. Yang et al. [11] have reported
that w/p ratio has a significant effect on the fiber distribution in The mix compositions of ECC are given in Table 1. All mixtures
ECC. Therefore, the different water mixing sequences are first had the same mix composition of solid materials and the different

(1) Mix all (2) All liquid materials (3) Fibers are
powder materials are added and mixed added and mixed
Fig. 2. Standard ECC mixing sequence.
344 J. Zhou et al. / Cement & Concrete Composites 34 (2012) 342–348

Table 1
Mix compositions of ECC.

Mixture LP#/PC* BFS/PC w/p ratio PC (kg/m3) LP (kg/m3) BFS (kg/m3) Water (kg/m3) SP$ (kg/m3) PVA fiber (kg/m3)
M1&M1A 2 1.2 0.3 356 711 427 448 7.1 26
M2&M2A 2 1.2 0.35 330 661 397 486 6.6 26
#
LP represents limestone powder.
*
PC represents Portland cement CEM I 42.5N.
$
SP represents superplasticizer.

2.3. Casting and curing

Table 2
Chemical compositions of CEM I 42.5N, limestone powder and BFS (by mass). The data The fresh ECC was cast into coupon specimens with the dimen-
of CEM I 42.5N and limestone powder were from the producer and that of BFS was
sion of 240 mm  60 mm  10 mm. The specimens were covered
measured by energy dispersive X-ray analysis.
with plastic paper and cured for 1 day at a temperature of 20 °C.
Compound CEM I 42.5N (%) Limestone powder (%) BFS (%) Then, the specimens were demoulded and cured under sealed con-
CaO 64.1 – 40.77 dition at a temperature of 20 °C for another 27 days.
SiO2 20.1 0.26 35.44
Al2O3 4.8 0.08 12.98
2.4. Uniaxial tensile test
Fe2O3 3.2 0.06 0.53
MgO – 0.21 7.99
K2O 0.52 – 0.49 The uniaxial tensile test was conducted on the coupon speci-
Na2O 0.28 – 0.21 mens at the age of 28 days as shown in Fig. 4. Four measurements
SO3 2.7 – 0.1 were done for each mixture. Before testing, coupon specimens
CaCO3 – 98.8 –
were sanded for a flat surface and four aluminum plates were then
glued on two ends of each specimen. After 1-day curing of the glue,
w/p ratios. M1 and M1A had a w/p ratio of 0.3, and M2 and M2A the specimen was fixed on the test set-up by clamping the alumi-
had a w/p ratio of 0.35. M1 and M2 were mixed following the stan- num plate-glued ends with two pairs of steel plates, which were
dard mixing sequence, while M1A and M2A were mixed following connected to the loading device. Then, two LVDTs were attached
the adjusted mixing sequence. The mixing sequence is described in on both sides of the specimen to measure the specimen deforma-
detail in Section 2.2. tion. The tests were conducted with displacement control at a
Table 2 lists the chemical compositions of CEM I 42.5N, lime- loading rate of 0.005 mm/s. The testing gauge length was 70 mm.
stone powder and blast furnace slag (BFS). The densities of CEM I The testing procedure was described in detail in [14].
42.5N, limestone powder and BFS were 3150 kg/m3, 2700 kg/m3
and 2850 kg/m3, respectively. The Blaine values of CEM I 42.5N, 2.5. Fiber distribution analysis
limestone powder and BFS were 310 m2/kg, 480 m2/kg and
320 m2/kg, respectively. In all mixtures, the fiber content was 2% After the uniaxial tensile tests, four thin sections, perpendicular
by volume. Fiber used in this study was the PVA fiber with a length to the loading direction, were taken from each specimen. The
of 8 mm and a diameter of 39 lm, and the fiber aspect ratio was dimension of the thin sections was 30  10 mm2. The samples
205. The tensile strength of the PVA fiber is 1600 MPa and the den- were glued on a glass and ground. The ground surface was glued
sity is 1300 kg/m3. The surface of the fiber is coated with 1.2% oil to an objective glass. They were sawn to a thickness of around
by weight to optimize the properties of the fiber–matrix interface, 1 mm. Then, the sawn surface was ground until the thickness of
aiming at achieving the strain-hardening behavior and high ductil- the thin section reached 100 lm.
ity [13]. Images were captured with a magnification of 100 from the
samples by optical microscope with plane polarization mode. Sixty
2.2. Mixing images were taken from all of the four pieces from each specimen.
In the images, the fibers appear white and the matrix appears black
M1 and M2 were mixed following the standard mixing se- as shown in Fig. 5 left. Then, the images were converted into binary
quence as follows: images to segment the fibers. In the binary images as shown in
Fig. 5 right, black dots indicate the fibers and the rest indicates
(1) All water were mixed with solid materials and superplasti- the matrix. The number of the fibers in the binary images was
cizer at low speed for 1 min and then at high speed for counted digitally. With the variation coefficient of the number of
2 min. the fibers in sixty images, the fiber distribution coefficient was cal-
(2) PVA fibers were added and mixed at high speed for 4 min. culated as follows [15]:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
P
M1A and M2A were mixed following the adjusted mixing se- ðX i =X average  1Þ2
quence with two-step water addition as follows: a ¼ expð Þ ð1Þ
n

(1) Part of water, with the w/p ratio of 0.27 in the reasonable where Xi is the number of the fibers in the ith image, Xaverage is the
range [11], was mixed with solid materials and superplasti- average number of the fibers in all images, and n is the number of
cizer at low speed for 1 min and then at high speed for images.
2 min.
(2) PVA fibers were added and mixed at high speed for 2 min. 3. Results and discussion
(3) The rest water was added and mixed at high speed for
another 2 min. 3.1. Uniaxial tensile test

In total, the standard and adjusted mixing sequences hold the The uniaxial tensile stress–strain curves of all mixtures at the
same mixing time of 7 min. age of 28 days are plotted in Fig. 6. Under the uniaxial tensile load-
 J. Zhou et al. / Cement & Concrete Composites 34 (2012) 342–348 345

3 mm

Steel bar

Steel
14 mm
Specimen plate

LVDT
Bolt

Steel
plate Aluminum plate
glued on the
Specimen specimen

Fig. 4. Uniaxial tensile test set-up.

Fig. 5. Image captured on the thin section (left) and corresponding binary image (right).

4 4
Tensile strain capacity (%)

3 M1A 3
Tensile stress (MPa)

2 2
M2 M2A
M1
1 1

0 0
0 1 2 3 4 M1 M1A M2 M2A
Strain (%)
Fig. 7. Tensile strain capacity of the four mixtures at the age of 28 days. M1 and M2
were mixed following the standard mixing sequence, and M1A and M2A were
Fig. 6. Typical uniaxial tensile stress–strain curves of the four mixtures at the age of
mixed following the adjusted mixing sequence.
28 days. M1 and M2 were mixed following the standard mixing sequence, and M1A
and M2A were mixed following the adjusted mixing sequence.
beginning and then drops. The first stress drop corresponds to the
ing, M2 shows a single-cracking behavior, while the other mixtures first cracking in the specimen. The peak stress is called the first
all show a multiple-cracking behavior. The uniaxial tensile stress– cracking strength, which is related to the tensile strength of the
strain curve of M2 is featured by a prominent peak at very little matrix. Subsequently, as the tensile stress increases slowly, the
strain corresponding to the only cracking. For the mixtures M1, specimens exhibit strain-hardening and multiple-cracking
M1A and M2A, the tensile stress increases dramatically at the very behavior.
346 J. Zhou et al. / Cement & Concrete Composites 34 (2012) 342–348

4 1
Ultimate tensile strength (MPa)

Fiber distribution coefficient

0.8
3

0.6
2

0.4

1
0.2

0
M1 M1A M2 M2A 0
M1 M1A M2 M2A
Fig. 8. Ultimate tensile strength of the four mixtures at the age of 28 days. M1 and
M2 were mixed following the standard mixing sequence, and M1A and M2A were Fig. 10. Fiber distribution coefficient of the four mixtures. M1 and M2 were mixed
mixed following the adjusted mixing sequence. following the standard mixing sequence, and M1A and M2A were mixed following
the adjusted mixing sequence.

4 3
First cracking strength (MPa)

Tensile strain capacity (%)

R = 0.98
3
2

1
1

0
0
M1 M1A M2 M2A 0.55 0.6 0.65 0.7 0.75 0.8
Fiber distribution coefficient
Fig. 9. First cracking strength of the four mixtures at the age of 28 days. M1 and M2
were mixed following the standard mixing sequence, and M1A and M2A were Fig. 11. Correlation between the fiber distribution coefficient and the tensile strain
mixed following the adjusted mixing sequence. capacity.

Figs. 7–9 show the tensile strain capacity, the ultimate tensile 3.2. Fiber distribution coefficient
strength and the first cracking strength of the four mixtures, cal-
culated by averaging the results of the four measurements for A higher fiber distribution coefficient indicates a more uniform
each mixture. The ultimate tensile strength is defined as the max- fiber distribution. The maximum value of the fiber distribution
imum stress in the tensile stress–strain curve, and the corre- coefficient is 1, which means that the number of the fibers in every
sponding strain is called the tensile strain capacity. As the w/p image is the same. Fig. 10 shows the fiber distribution coefficient.
ratio increases, for both the mixtures mixed following the stan- The fiber distribution coefficient of M1A is 8% higher than that of
dard mixing sequence and those with the adjusted mixing se- M1, and the distribution coefficient of M2A is 24% higher than that
quence, the tensile strain capacity and the ultimate tensile of M2. The increased fiber distribution coefficients indicate that
strength decrease. For the same w/p ratio, the mixtures mixed adjusting the mixing sequence indeed improves the fiber distribu-
following the adjusted mixing sequence (M1A and M2A) show a tion in ECC. Besides, the improved fiber distribution is also con-
greater strain capacity and a higher ultimate tensile strength than firmed by the observation of many fiber bundles in the mixing
the mixtures mixed following the standard mixing sequence (M1 with the standard sequence rather than with the adjusted
and M2), respectively. The increases in the strain capacity and of sequence.
the ultimate tensile strength appear more prominent in the mix- The correlation between the tensile strain capacity and the fiber
tures with a higher w/p ratio. The mixtures with the w/p ratio of distribution coefficient is plotted in Fig. 11. The tensile strain
0.3 show a 66% increase in the tensile strain capacity and a 5% in- capacity shows a strong correlation with the fiber distribution
crease in the ultimate tensile strength. For the mixtures with the coefficient. The experimental results imply that a more uniform fi-
w/p ratio of 0.35, M2A retains a tensile strain capacity of 2%, ber distribution is desirable for higher tensile ductility. The im-
which is more than 50 times that of M2. The ultimate tensile proved fiber distribution explains the better tensile properties of
strength of M2A is 14% higher than that of M2. As shown in the mixtures mixed following the adjusted mixing sequence.
Fig. 9, M1A and M2A have a slightly higher first cracking strength
than M1 and M2, respectively. The first cracking strength mainly 4. Discussion
relates to the tensile strength of ECC matrix. It means that the dif-
ferent mixing sequences do not have significant effect on the The PVA fibers are provided in the form of fiber bundles. In or-
properties of the ECC matrix. der to separate the fiber bundles and to mix fibers uniformly, high
 J. Zhou et al. / Cement & Concrete Composites 34 (2012) 342–348 347

Table 3
Mix compositions of ECC and HVS-ECC (kg/m3).

Mixture CEM I 42.5N Limestone powder BFS Silica fume River sand Water W/p ratio Super-plasticizer PVA fiber
S1 235 782 547 78 0 410 0.25 15.6 26
S2&S2A 145 482 337 48 1012 253 0.25 9.6 16.1

plastic viscosity of fresh mortar before fiber addition is desirable

[11]. The adequate plastic viscosity can be achieved by properly 5
adjusting the w/p ratio and the dosage of superplasticizer. A rea-
S1
sonable w/p ratio should be in the range of 0.25 ± 0.05 for the stan- 4

Tensile stress (MPa)

dard mixing sequence [11]. Beyond this range, the inappropriate
fresh properties may lead to non-uniform fiber distribution and
3
thus decrease in hardened properties. The poor tensile properties
S2A
and poor fiber distribution of M2 with the w/p ratio of 0.35 support
this conclusion. 2
The experimental results in this study reveals that the ad-
justed mixing sequence can improve the fiber distribution. For 1 S2
the adjusted mixing sequence, a proper amount of water, instead
of all, are mixed with the solid materials and superplasticizer
0
aiming at required rheological properties. When fibers are added, 0 1 2 3 4 5
the adequate plastic viscosity leads to uniform fiber distribution.
Strain (%)
Then, the addition of the rest water does not cause significant
effect on the fiber distribution. Although the final w/p ratio ex- Fig. 12. Typical uniaxial tensile stress–strain curves of ECC and HVS-ECC at the age
ceeds the desirable value, a good fiber distribution in ECC can of 28 days. S1 and S2 were mixed following the standard mixing sequence, and S2A
was mixed following the adjusted mixing sequence.
be achieved by properly adjusting the mixing sequence. As
shown in the experimental results, although M2A has the same
mix composition as M2, M2A, mixed following the adjusted se-
Table 4
quence, shows a much more uniform fiber distribution and bet-
Tensile strain capacity and ultimate tensile strength of ECC and HVS-ECC.
ter hardened properties than M2, mixed following the standard
sequence. Mixture Tensile strain capacity (%) Ultimate tensile strength (MPa)
Extending this concept, the fiber distribution is possible to be S1 3.7 3.8
improved by adjusting the mixing sequence of the other ingredi- S2 0.5 2.9
ents, such as solid materials and chemical admixtures. ECC is first S2A 2.2 3.5

mixed with a certain mix composition having good fiber distribu-

tion. After fibers are mix homogenously, the rheological and the same as the total amount of the other solid materials. The fiber
mechanical properties of fresh and hardened ECC are tailored by content in S1 was 2% by volume and it was 1.2% by volume in S2
adjusting water, solid materials and/or chemical admixtures and S2A. S1 and S2 were mixed following the standard mixing se-
according to the requirement of rheological properties and the quence. S2A was mixed following the adjusted mixing sequence as
ECC design theory. This concept resulted in the development of follows: (1) all solid materials except sand were first mixed with
ECC mixtures with the good fresh and hardened properties, such water and superplasticizer at low speed for 1 min and at high
as high volume sand ECC (HVS-ECC) [16]. speed for 2 min; (2) PVA fibers were added and mixed at high
speed for 2 min; (3) sand was then added and mixed at high speed
for another 2 min. The casting and curing procedures were the
5. Applications of the concept of adjusting the mixing sequence same as described in Section 2.3.
in the development of HVS-ECC The uniaxial tensile test, as described in Section 2.4, was used to
study the tensile behavior of HVS-ECC. The typical uniaxial tensile
With the presence of fibers, the addition of aggregates can cause
balling at mixing and poor fiber distribution [12]. The effect be- 3000
comes more pronounced when the maximum aggregate size in-
creases. This is one of the reasons why the size and amount of 2500
S1
aggregates are limited in ECC. The aggregate in standard ECC is fine
silica sand with a maximum grain size of 250 lm and a mean size
Shrinkage (×10-6)

2000
of 110 lm. The sand content of ECC is normally less than 30% of the
weight of the solid materials. The low volume of sand and thus S2A
1500
high volume of cementitious materials in ECC results in large dry-
ing shrinkage, an increase in material cost and a decrease in mate-
1000
rial eco-efficiency.
The adjusted mixing sequence was applied to design a version
of HVS-ECC [16]. River sand with the particle size ranging from 500
0.25 to 0.5 mm was used in ECC. The addition of sand might cause
poor particle packing on the surface of fibers and thus a poor fiber– 0
0 20 40 60 80 100 120
matrix interface. Silica fume was added in order to improve the
Age (days)
fraction-based bond of the fiber–matrix interface [17]. Table 3
gives the mix compositions of HVS-ECC. Sand was mixed in S2 Fig. 13. Drying shrinkage of ECC and HVS-ECC at a temperature of 20 °C and RH of
and S2A but not in S1. In S2 and S2A, the amount of sand was 50%.
348 J. Zhou et al. / Cement & Concrete Composites 34 (2012) 342–348

stress–strain curves of S1-3 at the age of 28 days are plotted in and the decreased drying shrinkage. Considering the low
Fig. 12. The tensile strain capacity and the ultimate tensile strength, cementitious material and fiber contents, the HVS-ECC is
calculated by averaging the results of the four measurements, are gi- expected to have increased material eco-efficiency and
ven in Table 4. For ECCs mixed following the standard mixing se- reduced material cost, since the PVA fiber dominates the
quence, the addition of sand caused 85% decrease in the tensile material cost.
strain capacity and 26% decrease in the ultimate tensile strength.
The adjusted mixing sequence leads to an improvement of the
mechanical properties of ECC. S2A shows a tensile strain capacity Acknowledgements
of 2.2% and ultimate tensile strength of 3.5 MPa, which were 4.1
times the tensile strain capacity and 1.2 times the ultimate tensile This research is financially supported by the Delft Clusters and
strength of S2. The better tensile performance can be attributed to Heijmans Infrastructure B.V. Their support is gratefully acknowl-
the improved fiber distribution by adjusting the mixing sequence. edged. We would like to thank BAS B.V. for their help in measuring
S2A also shows advantage in the dimensional stability. The dry- the particle size distribution of powder materials. V.C. Li would like
ing shrinkage of HVS-ECC was measured in accordance to the Euro- to acknowledge the US National Science Foundation CI-Team Grant
pean standard EN 480-3 [18]. The tests were carried out on prism OCI 0636300 for supporting international research collaboration.
specimens with the dimension of 40  40  160 mm3 exposed to a
temperature of 20 °C and RH of 50% after 24-h sealed curing. Fig. 13 References
shows the drying shrinkage of S1 and S2A measured in the first
100 days. Each value represents the average drying shrinkage of [1] Li VC. From micromechanics to structural engineering – the design of
cementitious composites for civil engineering applications. J Struct Mech
three measurements. At the age of 100 days, S1 shows a drying Earthquake Eng-JSCE 1993;10(2):37–48.
shrinkage of 2730 lstrain, while S2A shows a drying shrinkage of [2] Weimann MB, Li VC. Hygral behavior of engineered cementitious composite
1480 lstrain. The addition of sand results in 45% decrease in the (ECC). Int J Restoration Build Monument 2003;9(5):513–34.
[3] Lepech MD, Li VC. Water permeability of cracked cementitious composites. In:
drying shrinkage. Carpinteri A, editor. Proceedings 11th international conference on fracture;
The adjusted mixing sequence method was successfully applied 2005 [paper 4539].
in the development of HVS-ECC. The adjusted mixing sequence can [4] Li VC. Engineered cementitious composites (ECC) – material, structural and
durability performance. In: Nawy E, editor. Concrete construction engineering
recover a large part of the loss of tensile strain capacity and ulti- handbook. Boca Raton: CRC Press; 2007 [chapter 24].
mate tensile strength due to the addition of sand. Moreover, the [5] Sahmaran M, Li VC, Andrade C. Corrosion resistance performance of steel-
high sand content and the low fiber content also reduce the mate- reinforced engineered cementitious composite beams. ACI Mater J
2008;105(3):243–50.
rial cost and improve the eco-efficiency of ECC.
[6] Sahmaran M, Li VC. Durability of mechanically loaded engineered cementitious
composites under highly alkaline environment. Cem Concr Comp
6. Conclusions 2008;30(2):72–8.
[7] Michigan Department of Transportation. Bridge decks going jointless:
cementitious composites improve durability of link slabs. Construct Technol
A new ECC mixing sequence is developed, aimed at improving Res Rec 2005;100:1–4.
the fiber distribution in ECC. In this paper, the influence of the dif- [8] Kunieda M, Rokugo K. Recent progress on HPFRCC in Japan. J Adv Concr Techol
2006;4(1):19–33.
ferent mixing sequences of water is first investigated by comparing [9] Li VC, Leung CKY. Theory of steady state and multiple cracking of random
the experimental results of the uniaxial tensile test and the fiber discontinuous fiber reinforced brittle matrix composites. J Eng Mech-ASCE
distribution analysis. Extending this concept, a new version of 1992;118(11):2246–64.
[10] Torigoe S, Horikoshi T, Ogawa A, Saito T, Hamada T. Study on evaluation
HVS-ECC is developed by adjusting the sand mixing sequence. method for PVA fiber distribution in engineered cementitious composite. J Adv
Based on these results, the following conclusions are drawn: Concr Technol 2003;1(3):265–8.
[11] Yang EH, Sahmaran M, Yang Y, Li VC. Rheological control in the production of
engineered cementitious composites. ACI Mater J 2009;106(4):357–66.
(1) The adjusted mixing sequence results in an increase in the
[12] De Koker D, van Zijl G. Extrusion of engineered cement-based composite
tensile strain capacity and an increase in the ultimate tensile material. In: di Prisco M, et al., editors. Proceedings 6th RILEM symposium on
strength of ECC. The increases are more prominent in the FRC; 2004. pp. 1301–10.
mixtures with a higher w/p ratio. The adjusted mixing [13] Li VC, Wu C, Wang S, Ogawa A, Saito T. Interface tailoring for strain-hardening
PVA-ECC. ACI Mater J 2002;99(5):463–72.
sequence does not have a significant effect on the properties [14] Zhou J, Qian S, Guadalupe Sierra Beltran MGM, Ye G, van Breugel K, Li VC.
of ECC matrix, such as first cracking tensile strength. Development of engineered cementitious composites with limestone powder
(2) Compared with the standard mixing sequence, the adjusted and blast furnace slag. Mater Struct 2010;43(6):803–14.
[15] Kobayashi K. Fiber reinforced concrete. Tokyo: Ohm-Sha; 1981.
mixing sequence improves the fiber distribution in ECC. [16] Zhou J, Qian S, Ye G, van Breugel K. Engineered cementitious composites with
(3) A good correlation between the fiber distribution coefficient low volume of cementitious materials. In: Oh BH, editor. Proceedings
and the tensile ductility is found. The improved fiber distri- FramCos-7; 2010. p. 1551–6.
[17] Yan H, Sun W, Chen H. Effects of silica fume and steel fiber on the dynamic
bution is responsible for the enhanced mechanical proper- mechanical performance of high-strength concrete. Cem Concr Res
ties of hardened ECC. 1999;29(3):423–6.
(4) The adjusted mixing sequence was applied in the develop- [18] EN 480-3. Admixtures for concrete, mortar and grout – test methods – part 3:
determination of shrinkage and expansion. European Standard; 1991.
ment of HVS-ECC. HVS-ECC mixed following the adjusted
mixing sequence showed the improved tensile properties