Construction and Building Materials 158 (2018) 217–227

Contents lists available at ScienceDirect

Construction and Building Materials

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

Development of ultra-high performance engineered cementitious

composites using polyethylene (PE) fibers
Ke-Quan Yu a,b, Jiang-Tao Yu a, Jian-Guo Dai b,⇑, Zhou-Dao Lu c, Surendra P. Shah d
a
Department of Civil Engineering, Tongji University, China
b
Department of Civil and Environmental Engineering, The Hong Kong Polytechnic University, China
c
Department of Civil and Environmental Engineering, Tongji University, China
d
Walter P. Murohy Professor of Civil Engineering (emeritus), Northwestern University, USA

h i g h l i g h t s

The tensile, compressive and flexural behaviors of UHP-ECC were systematically investigated.

UHP-ECC combines the strain-hardening and multiple crack characteristics and the high strength of mortar matrix.

Ultra-high-molecular-weight polyethylene (PE) fibers with a high aspect ratio were deployed.

The digital image correlation (DIC) technique was utilized to monitor the crack patterns during the tests.

Scanning electron microscope (SEM) analysis was conducted to understand the microstructure of UHP-ECC.

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

Article history: Ultra-high performance engineered cementitious composites (UHP-ECC), which combines the strain-
Received 26 May 2017 hardening and multiple crack characteristics and the high strength of mortar matrix, was investigated
Received in revised form 5 October 2017 in this study. The tensile strength and elongation of the UHP-ECC achieved were 20 MPa and 8.7%, respec-
Accepted 6 October 2017
tively. For the production of UHP-ECC, ultra-high-molecular-weight polyethylene (PE) fibers were
Available online 12 October 2017
deployed to reinforce the ultra-high strength mortar while special attention was paid to the mix process
to ensure satisfactory fiber dispersion. The tensile stress-strain curves, the compressive strength and elas-
Keywords:
tic modulus, and the flexural behavior of UHP-ECC were investigated to understand its mechanical per-
Strain hardening cementitious composites
Ultra-high strength mortar
formance. The digital image correlation (DIC) technique was utilized to monitor the crack patterns of
High strength UHP-ECC during the tensile and flexural tests. In addition, Scanning electron microscope (SEM) analysis
High ductility was conducted to achieve an in-depth understanding of the microstructure of UHP-ECC.
Multiple cracks Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction Ultra-high performance concrete (UHPC) with a high compres-

Strength, ductility and durability are crucial properties for
development of modern concrete technology. High-strength con-
crete facilitates the design of size-efficient structural members
and provides additional strength safety margins (particularly in
compression) for structures. High ductility concrete prevents
catastrophic structural collapse by absorbing massive amounts of
energy during extreme load/displacement events-such as earth-
quakes and blasts, while the use of highly durable concrete extends
the service life of concrete infrastructure, reduces the life-time
maintenance cost and improves the infrastructure sustainability.
⇑ Corresponding author.
E-mail address: cejgdai@polyu.edu.hk (J.-G. Dai).
sive strength up to 200–800 MPa [1–5] have been studied world-
wide. In general, the cement dosage of UHPC is over 800 kg/m3
and the water/binder ratio is lower than 0.20 while high-range
water-reducing admixture (HRWR) is required for use to achieve
the high compressive strength. Furthermore, the UHPC is com-
posed of very fine powders, such as crushed quartzite and silica
fume. The basic composition for the production of UHPC was well
explained in Richard and Cheyrezy (1995) [1]. In the report pub-
lished by Federal Highway Administration (FHWA 2013) [6], UHPC
is defined for those cementitious-based composites materials with
discontinuous fiber reinforcement, compressive strengths over
150 MPa, pre- and post-cracking tensile strengths above 5 MPa.
No particular requirement is imposed on the ductility of UHPC
although in the report an idealized uniaxial tensile mechanical
response of UHPC is recommended [6]. In contrast, strain

https://doi.org/10.1016/j.conbuildmat.2017.10.040
0950-0618/Ó 2017 Elsevier Ltd. All rights reserved.
218 K.-Q. Yu et al. / Construction and Building Materials 158 (2018) 217–227

hardening fiber cementitious composite (i.e., engineered cementi-
tious composites, ECC) featuring with excellent ductility and
multiple instances of micro-cracking with self-controlled widths
[7–13] with the addition of Polyvinyl Alcohol (PVA) fibers at a
volume fraction of 2%, can achieve a tensile strain capacity up to
3–5%. The micro-crack width characteristic is appreciated because
it improves the durability of concrete and entitles the fiber
composites with self-healing effects [7,14].
Recently researchers have spent significant efforts in develop-
ing ultra-high performance fiber-reinforced concrete (UHPFRC)
with strain hardening properties [15–25]. Fig. 1a and b presents
a summary of the compressive strength, tensile strength and ten-
sile strain capacity of the UHPFRCs found in existing literature. It
is seen that, in most cases, the achieved tensile strain capacity
was around 0.6% based on the tests of dumbbell specimens under
direct tension, while the tensile and compressive strengths range
from 10 to 18 MPa and 80 to 220 MPa, respectively. It should be
noted that the nature of materials and methods of reinforcement
may affect the compressive and tensile strength. Therefore the data
presented in Fig. 1 mainly reflect that there is a lack of good
combination of strength and ductility in existing UHPFRCs. It
may not provide an accurate quantitative comparison of the mate-
rial performance of these UHPFRCs developed by different research
groups. In particular, the tensile and flexural behaviors of ultra-
high-performance fiber reinforced cement composites, such as
UHPFRC, UHP-ECC. are strongly influenced by the fiber orientation
and dispersion [26–29] as well as the size effect [30–31]. Neverthe-
250
(Ductal 1999, 2000, 2004)
Compressive Strength/MPa
less, it is worth mentioning that Ranade (2013) developed success-
fully a UHPFRC with the 28 days compressive strength of 150 MPa,
tensile strength of around 15 MPa and a tensile strain capacity of
3.3% [32,33]. This development truly combined the high strength
of mortar matrix and high tensile ductility. However, saturated
cracks could not be developed in their direct tensile tests and the
observed residual crack width was much wider than often desir-
able for improved durability.
The present paper presents a further step development of the
UHPFRC, which aims to retain high tensile strain, strain hardening
and multiple microcracking characteristics, while achieving high
tensile and compressive strengths. This UHPFRC is termed as
UHP-ECC in this paper. The target values of its tensile strain capac-
ity (6–10%), tensile strength (16–20 MPa), and compressive
strength (>100 MPa) are indicated in Fig. 1. It has the intrinsic mul-
tiple microcracking behavior under tensile loading and its residual
crack width should be no more than 100 lm.
2. Theoretical background
The concept of ultra-high performance concrete (UHPC) was
utilized to realize a densely packed homogenous cementitious
matrix [2] to achieve a high compressive strength. While for
achieving high tensile ductility micromechanics-based strain hard-
ening criteria should be satisfied [8,32,33]. Hence, the design
approach of UHP-ECC entailed integrating these two approaches
into a single material by adapting the matrix features of UHPC in
combination with a high-performance fiber with an aspect ratio
and interfacial properties that satisfy the micromechanics-based
tensile strain-hardening criteria.
In order to attain the multiple cracking states, two conditions
should be satisfied [34,35]. The first condition is that the matrix

200 (Park 2012)

tensile cracking strength rc must not exceed the maximum fiber
(Wille 2011) bridging strength r0.
(Ranade 2013)
150 (Maeder 2004) rc < r0 ð1Þ
(Present research)
where rc is determined by the matrix fracture toughness Km and
100 (Kamal 2008)
preexisting internal flaw size a0.
(Sujiravorakul 2010) The second condition is that the crack tip toughness Jtip to be
less than the complementary energy J0 b, which can be calculated
50 from the bridging Stress r versus crack opening d curve, as illus-
trated in Fig. 2 [34,35].
Z d0
0
0 2 4 6 8 10 J tip 6 r0 d0  rðdÞdd  J0b ð2Þ
0
Tensile Strain/%
(a) Compressive strength vs. Tensile strain capacity K 2m
J tip ¼ ð3Þ
25 Em
(Present research)

20
Tensile Strength/MPa

(Park 2012)
σ
(Ranade 2013)
15 (Wille 2011)
(Ductal 1999, 2000, 2004)
σ0
(Kamal 2008)
(Sujiravorakul 2010) Jb
10
(Maeder 2004) σ ss
5 Jtip

0
0 2 4 6 8 10
Tensile Strain/%
(b) Tensile strength vs. Tensile strain capacity δ ss δ0 δ
Fig. 1. A summary of the strength and ductility of UHPFRC in existing literature. Fig. 2. Typical r (d) curve for tensile strain-hardening composite.
 K.-Q. Yu et al. / Construction and Building Materials 158 (2018) 217–227 219

While the strength-based criterion represented by Eq. (1) con- (DIC) technique, providing information on the number of cracks
trols the initiation of cracks, the energy criterion (Eq. (2)) governs and crack width distribution.
the crack propagation mode. Satisfaction of both Eqs. (1) and (2) is
necessary to achieve the strain-hardening behavior. The pseudo 3.1. Materials and mix proportions
strain-hardening (PSH) performance index has been used to quan-
titatively evaluate the margin and is defined as follows [36]. The mixture used in the manuscript was based in the mixture of
UHPC and the one of Ranade (2013a) with necessary revisions to fit
J 0b
PSH ¼ ð4Þ the local raw materials. The key aspect of the present paper is the
J tip
use of high aspect ratio polyester fibers to achieve the high
A focus was placed on ensuring a high complimentary energy as strength and high tensile rupture strain of UHP-ECC. Before the
dictated by the micromechanical model for pseudo strain- present paper, extensive trial and error tests were conducted in
hardening [8,32,33]. Ultra-high-molecular-weight polyethylene the laboratory to study the effects of fiber contents and GGBS/Silica
(UHMWPE, henceforth briefed as ‘‘PE”) fiber was selected for use fume contents on the tensile performance of ECC at three different
together with the UHPC matrix instead of the hooked steel fibers, strength grades and the results were reported in the authors’ pre-
which are popularly used in the conventional UHPC composite. vious paper (Yu et al. 2017). The present paper focuses on UHP-ECC
The PE fiber was selected due to its very high strength and only and represents a further step development towards a stable
hydrophobic nature. The high fiber strength is needed to transmit, mechanical performance.
without rupturing, the large interfacial frictional stress generated Ordinary Portland cement (OPC) 52.5R was used in this study.
by the densely packed UHPC matrix, and to ensure the fiber- Grade S105 Ground granulated blast furnace slag (GGBFS) accord-
bridging capacity relative to the high matrix cracking strength of ing to Chinese Code (GB/T 18046 2008) [39] was incorporated to
UHPC. The hydrophobic nature of PE fiber eliminates fiber/matrix replace cement and silica fume (SF) was used as the secondary
chemical bond, which significantly enhances the complimentary cementitious material. Silica fume is a highly reactive supplemen-
energy of fiber bridging [32]. tary cementitious material that can promote the formation of sec-
In previous research [32,33], the PE fiber with df = 28 lm and Lf ondary hydration products, thereby maximizing the calcium
= 12 mm was selected for the high-strength, high-ductility con- silicate hydrate (C-S-H) content. GGBFS is a beneficial mineral
crete mixture. A higher Lf/df was not used by them probably due admixture due to its pozzolanic nature at room temperature [40–
to the mixing difficulty. However, a higher Lf/df ratio results in a 42]. Its use can reduce the porosity and change in the mineralogy
larger fiber/matrix interfacial area for the same fiber-volume frac- of cement hydrates, leading to a reduction in mobility of chloride
tion [37,38] which increases the fiber-bridging capacity, tensile ions [43]. The silica flour used in conventional reactive power con-
strength and strain capacity of the composite. Therefore, a smaller crete was not used in the current mixture due to its low reactive
diameter of PE fiber with df = 20 lm and Lf = 18 mm was tried for property at room temperature [1,2]. Use of fine particles of silica
the mixture of UHP-ECC in the present study. This resulted in an fume (SF) (0.1 to 1 lm) and GGBFS (1 to 100 lm) was expected
increase of the Lf/df ratio from 428 in the previous study [32] to to increase the density of the matrix and aggregate-cement inter-
900 in the present UHP-ECC. face by filling the larger voids. A polycarboxylate-based high-
range water-reducing admixture (HRWRA) from BASF Co. Ltd
was used in UHP-ECC to maintain the flow ability of the mixture
3. Experimental program at the very low water/binder ratio. The aggregates or fillers used
in the UHP-ECC matrix were primarily fine silica sand with a max-
A comprehensive experimental program was conducted to eval- imum grain size of 180 lm and a mean size of 135 lm. Using such
uate the direct tensile property, compressive strength and Young’s a small aggregate size reduced the size of the weak interface
modulus, flexural strength and fracture energy of UHP-ECC. The between the aggregate and the cement. A smaller aggregate also
fresh properties and mixing process were also reported. The tensile reduced the fracture toughness of the matrix for crack initiation
crack pattern was investigated using the digital image correlation and work of fracture during the steady-state crack propagation,
both of which are desirable for composite ductility according to
micromechanics [32].
Table 1 presents the chemical compositions of GGBFS, SF and
Table 1
Chemical compositions of cement, silica fume and slag. ordinary Portland cement (OPC), which were obtained by the X-
ray Fluorescence (XRF) test. The typical physical and mechanical
OPC SF GGBFS
properties of OPC are provided in Table 2 according to the sup-
SiO2 20.10 92.26 39.66 plier’s information. The scanning electron micrograph (SEM) pho-
CaO 62.92 0.49 34.20
tographs of GGBFS and SF are shown in Fig. 3. Before scanning,
Al2O3 5.62 0.89 12.94
Fe2O3 2.17 1.97 1.58 the fine particles of GGBFS and SF were dispersed by an ultrasonic
MgO 1.14 0.96 6.94 dispersing apparatus to reduce agglomeration. Gradation curves of
Na2O 0.30 0.42 0.20 cement and GGBFS determined from particle size analysis (PSA) is
K2O 0.85 1.31 1.44 given in Fig. 4. The particle size of GGBFS is much finer than that of
SO3 2.92 0.33 0.72
cement to ensure its filling effect in the matrix. GGBFS had a speci-
L.O.I. 3.84 <6.0 1.2
fic surface area of 720 m2/kg according to PSA while SF had a value

Table 2
Physical and mechanical properties of cement.

Property Specific gravity Cement Blaine Initial setting Final setting Volume Compressive
(g/cm3) (m2/kg) times (min) times (min) expansion (mm) strength (MPa)
7d 28d
Value 3.13 380 130 210 1.00 53.2 61.9
220 K.-Q. Yu et al. / Construction and Building Materials 158 (2018) 217–227

(a) GGBFS (b) silica fume

Fig. 3. SEM photographs of GGBFS and silica fume.

Table 4
100 PE fiber properties.

Fiber properties Value

90
Diameter df, lm 20
80 Length Lf, mm 18
Volume fraction Vf, % 2
70
Pass percentage/%

Nominal strength, MPa 3000

60 Nominal Young’s modulus, GPa 100
Elongation at break, % 2–3
50 Specific gravity, g/cm3 0.97
Melting temperature, °C 150
40

30
85 40 80 40 85
20 GGBFS

10
Cement
Plan View Thickness
0 30 60
0.1 1 10 100 1000 10000 Unit: mm 13mm
Particle size/μm Displacement
Measurement
Zone
Fig. 4. Particle size distributions of GGBFS.

Fig. 5. Dimension of dog bone specimen.

Table 3
Mixture properties of UHP-ECC.

Cement (C) (kg/m3) 800 Loading point

3
Silica fume (SF) (kg/m ) 150
Ground granulated blast furnace slag (GGBFS) (kg/m3) 750
Silica sand (kg/m3) 500
Water (W) (kg/m3) 230
150mm 150mm 150mm
PE fiber (kg/m3) 20
HRWRA (kg/m3) 25
W/(C + SF + GGBFS) 0.135
W*/(C + SF + GGBFS) 0.14

Note: W* included the water from HRWRA.

Span=450mm

Fig. 6. Dimension of beam specimens for flexural tests.

of 20,000 m2/kg according to the manufacturer’s data. It was diffi-

cult to obtain the particle size distribution of SF due to the agglom-
eration problem. The binder composition of the matrix was
designed using the factorial design method [44]. The mix propor-
tions are given in Table 3. Table 4 presents the physical/mechanical
properties and the geometry of polyethylene (PE) fibers.
3.2. Specimen preparation
Dumbbell-shaped specimens (Fig. 5), were used in this study to
measure the complete stress-strain behavior of UHP-ECC under
direct uniaxial tension (JSCE 2008) [45]. Four UHP-ECC dumbbell
specimens were cast and tested after 28 days curing. The Dumbbell
geometry forces most of the cracks to occur in the gauge region
that has a smaller cross-sectional area, thus allowing more reliable
measurement of the tensile strains.
The uniaxial compression compressive strength of UHP-ECC
was measured using cubes with two different edge lengths of
70.7 mm and 100 mm [32,33,45], both of which have been widely
used in existing research. The compressive modulus of UHP-ECC
was measured by six prisms with a length of 200 mm and a cross
section of 100  100 mm. Four beams with length of 500 mm
(span length of 450 mm), width of 100 mm, and depth of 100
mm were cast for four-point flexural tests (Fig. 6) [46].
 K.-Q. Yu et al. / Construction and Building Materials 158 (2018) 217–227 221

Six notched prism specimens of dimension 305  76  38 mm

with a notch-depth ratio of 0.4 were prepared to measure the Dry powder and
silica sand
matrix toughness. The bridging stress (r)-crack opening (d) rela-
tion (r-d curve) of UHP-ECC was experimentally determined from Add water and mix for
five single-crack tests with notched rectangular coupons to check if 10 minutes
the theoretical criteria for designing the UHP-ECC are satisfied. The
detailed dimensions are shown in Fig. 7. Flowable matrix
Add adding half fiber, Add adding all the fiber,
3.3. Mixing procedure mix rapidly for 1 min mix rapidly for 2 mins

The mixing procedure was purposely designed to ensure good UHP-ECC

fiber dispersion leading to good material performance. The mix-
Fig. 8. Mix process of UHP-ECC.
tures were prepared in a planetary type, vertical axis, speed adjus-
table mixer (two speed grades). The speed of the mixer could be
adjusted up to 420 rpm. Dry powders and aggregates were mixed
with the agitating speed of 140 rpm for about 1 min. After the addi-
tion of water (with HRWRA), the mixtures were mixed for about 10
mins with the agitating speed of 140 rpm and then 470 rpm for
another 2 mins. PE fibers were evenly added in four batches. After
half fibers were added, the mixing speed turned to 420 rpm for 1
min for better fiber dispersion. After all the fibers were added, the
agitating speed was maintained at 420 rpm for another 2 mins.
Finally, obvious particle agglomerations on the mixing blade were
removed using hand operation. Fig. 8 illustrates the mixing proce-
dures. Because of the high binder volume and very low water/bin-
der ratio, consistency of mixtures was low. Thus, casting and
compaction were realized by mechanical vibration. The specimens
were kept in the moulds for 24 h at room temperature. After
demolding, all the specimens were cured in room temperature for
28 days. No heat curing was adopted to avoid thermal damage to
PE fibers according to the suggestion of the material supplier.

3.4. Test setup and procedures

The slump of the matrix (without fibers) and composite (with

fibers) was measured according to CECS (2009) [46] using flow
table test. The tensile tests were conducted at a fix loading rate
of 0.5 mm/min according to JSCE (2008) [45]. The test set-up is Fig. 9. Test setup of direct tensile test.

shown in Fig. 9. Two linear variable differential transformers

(LVDTs) were mounted on both sides of the test specimens to mea-
sure the elongation in the gauge region with the length of 100 mm
during the whole test.
In addition, crack pattern analysis was conducted by using the
DIC technique [47–49]. DIC is a non-contact measurement method
to analyze the surface displacement/strain within a calculated area
of an object. The most significant advantage of the DIC technique is
that it can provide full field surface displacement/strain with high
resolution, which is difficult to be accomplished by conventional
strain gauge measurement [49]. For the DIC analysis, a random
110
2
15 15 2 closed loop displacement-controlled testing machine with the
Unit:mm
60 13
Fig. 7. Dimension of single-crack specimen.
speckle pattern was created on the specimen surface using white
and black spray painting. A digital camera was mounted on a tri-
pod in front of the testing machine, and digital images were taken
at a 20-s interval using remote control. Then 2-D image correlation
analyses were conducted to measure the in-plane deformations
using a commercial software VIC 2DTM.
The four-point bending tests were performed on UHP-ECC
beams following the Chinese standard test specification (CECS
2009) [46]. A constant loading rate of 1 mm/min was used in these
bending tests. The deflections at support, loading points and mid-
span point were traced by five LVDTs (Fig. 10). As a result, the mid-
span deflection was calculated by subtracting the value of LVDT at
the mid-span point by the average displacement of the two sup-
ports. The DIC method was also utilized to monitor the deforma-
tion/strain field during the whole load process. The load and
displacement readings and the DIC image could be synchronized
using the time interval as a control parameter.
Fig. 11 is the test setup used for cube compression tests. A
capacity of 3000 kN was used to load the cubes. The compressive
loading rate for the cube specimens (70.7 mm and 100 mm) was
1 MPa/s according to CECS (2009) [46]. However, for prism speci-
mens, the loading rate was fixed at 0.04 mm/s to ensure a stable
load-displacement curve. The compressive displacement was mea-
sured using two high-precision potentiometers spanned 100 mm
on the two sides of the prisms (Fig. 11).
222 K.-Q. Yu et al. / Construction and Building Materials 158 (2018) 217–227

Fig. 10. Test setup of four-point bending.

(a) Cube specimen (b) Prism specimen

Fig. 11. Set-up of compression tests.

4. Results and discussion 20

4.1. Flow ability and density

16
The UHP-ECC matrix is well flowable with a slump of approxi-
mately 220 mm, while the composite is moderately flowable with
Stress/MPa

a slump of approximately 160 mm. As a result, a moderate level of 12

vibration may be needed to place it properly in the molds or form-
work. The density was determined by measuring the weights and
8
actual dimensions of cube and prism specimens. The average den-
sity was 2405 kg/m3 with a coefficient of variation (C.o.V) of 2%. In
spite of the absence of coarse aggregates, the bulk density of UHP- Test curve
4 Test curve
ECC is similar to that of normal concrete (2300 to 2400 kg/m3), Test curve
which is attributed to the dense particle packing within the UHP- Test curve
Average curve
ECC matrix. 0
0 2 4 6 8 10
4.2. Tensile properties Strain/%

Fig. 12. Tensile stress-strain curves of UHP-ECC.

Fig. 12 shows the typical tensile stress-strain curves of UHP-ECC
at the curing age of 28 days. It is seen that all the specimens exhib-
ited tensile strain-hardening behavior. The fluctuations in the
curves results are normal in cementitious materials like ECC, which
are caused by the cracking of matrix and the propagation of cracks
in the loading process. The average tensile strength of the Dumb-
bell specimens approached 17.42 MPa with a C.o.V of 6%. The cor-
responding average tensile strain capacity was 8.17% with the C.o.V
of 5%. The average initial cracking stress was 10.2 MPa with a C.o.V
of 5%. The average tensile elastic modulus of UHP-ECC was com-
 K.-Q. Yu et al. / Construction and Building Materials 158 (2018) 217–227 223

puted from the slope of tensile stress-strain curve and the value 35
was 41.2 GPa with a C.o.V of 3%. The specific fracture energy under
direct tension was greater than 1500 kJ/m3. 30
Fig. 13 shows the DIC pictures of tensile specimens at different
strain levels. It is very clear that multiple cracks saturated over the
25
calculated area as the strain level increases. Most importantly,
cracks were not localized even under a high imposed strain level.

Stress/MPa
When unloading, the residual crack width of tensile specimens 20
can be less than 100 lm, indicating that UHP-ECC can have low
permeability compared to conventional fiber reinforced concrete. 15

10
4.3. Flexural properties

The four-point bending test results of two UHP-ECC beams with 5 Test curves
the dimensions of 100  100  500 mm are shown in Fig. 14. In Average
this figure, the flexural stress is plotted against the mid-span net 0
0 2 4 6 8 10 12 14
deflection of the beam. The flexural stress was computed from
the applied load and the dimensions (100  100500 mm) of the Displacement/mm
UHP-ECC beams. The average initial cracking stress (the flexural
Fig. 14. Flexural stress vs. mid-span displacement curves of UHP-ECC beams.
stress corresponding to the first crack) and modulus of rupture
(MOR) of the four beams were 20.40 MPa and 27.68 MPa, respec-
tively. After cracking, the flexural beam experienced a deformation Beam span/mm
hardening process till the peak stress. In addition, UHP-ECC beams
0 100 200 300 400
exhibited extremely high ductility along with the high MOR. The 0
average mid-span deflection at the MOR reached 10.1 mm, which
was about 1/50 of the span length. The combination of high -2
strength and high ductility ensured the excellent energy absorp-
tion capacity of UHP-ECC, which is a very important characteristic
Deflection/mm

-4
needed for seismic applications.
The deflections of one specific flexural beam along the span at
-6
different load stages are shown in Fig. 15. The mid-span net deflec-
tion and loading point net deflection were computed as the aver-
-8 9.55kN
age measurement of the LVDT at one side and the DIC result at 23.74kN
the other side of the UHP-ECC beam. Before cracking, the deflection 31.86kN
-10 42.02kN
is small due to the high elastic modulus of UHP-ECC. Afterwards, 46.05kN
52.70kN
the deflection increased continuously during the loading process 53.30kN(Peak)
till to the peak stress, and eventually a localized crack occurred -12
in the flexural beam. The deflection corresponding to the peak
Fig. 15. Flexural deflection along tested beam at different loading stages.
stress reached 10.7 mm (i.e., about 1/50 of beam span).
Fig. 16(a)–(d) show the crack pattern and development of one of
the flexural beams monitored by DIC method during the whole tiple cracks developed over the constant moment zone, but were
loading process. At the load level of 31 kN, a first crack occurred. not saturated even at the deflection to span ratio of 1/40 as shown
Afterwards, the crack number increased with the load while the in Fig. 16(d).
crack width kept almost constant at 42 kN. At the peak load, mul-
4.4. Compressive properties
25
The uniaxial compression test results from six 70.7 mm cubes
and six 100 mm cubes are summarized in Table 5. The average
compressive strength of these two types of cubes are 121.5 MPa
20
and 111.4 MPa for 70.7 and 100 mm cubes respectively. The com-
pressive strength decreased slightly with size increase. The ratio of
compressive strengths of 100 mm cube to 70.7 mm cube is 0.92,
Stress/MPa

15
which is coincident with the previous knowledge [50,51]. The com-
pressive Young’s modulus was determined from the compressive
stress-strain curve of 100  100  200 mm prisms. The average
10
compressive Young’s modulus of UHP-ECC was 40.9 GPa with the
C.o.V of 9.2%. The stress-strain curves of 100 mm prisms are shown
in Fig. 17. The curves are linearly elastic up to peak and then
5
become nonlinear due to the occurrence of microcracks in the
prisms. These cracks were stabilized through the fiber-bridging
Cracking ε=3% ε=5% ε=7.2% effect, resulting in a relatively smooth descending part as com-
0 pared to the sharp decrease that was typically observed in case
0 2 4 6 8 10
Strain/% of high-strength concrete [52,53]. After the peak load, the load
drop stopped at about 100 MPa and then decreased gradually to
Fig. 13. The DIC pictures of UHP-ECC. zero with the increasing compressive strain.
224 K.-Q. Yu et al. / Construction and Building Materials 158 (2018) 217–227

(a) First crack (31kN) (b) Fifth crack (42kN)

(c) Peak load (54kN) (d) 1/40 deflection to span ratio (52kN)
Fig. 16. Crack pattern of flexural beam at different loading stages.

Table 5
Compressive properties of UHP-ECC.

Cube Prism
70.7 mm strength (MPa) 100 mm strength (MPa) Strength (MPa) Strain at the peak stress (le) Young’s Modulus (GPa)
125.00 108.46 123.70 3115 44.28
110.51 104.77 121.06 2848 43.76
122.54 113.93 117.23 2790 44.61
126.36 112.72 116.39 3065 44.39
125.78 113.63 — — —
118.84 114.72 — — —
Average 121.50 112.69 119.60 2954 44.26
STD 6.05 3.91 9.90 160 0.31
C.o.V 0.050 0.035 0.085 0.054 0.007

140 4.5. Single-crack test results

The bridging stress (r)-crack opening (d) relation (r-d curve) of

120
UHP-ECC was experimentally determined from five single-crack
tests of notched rectangular coupons (Fig. 7) under direct tension.
100 All the five measured curves are shown in Fig. 18. Two distinct
phases are observed in the measured curves. Initially, when the
Stress/MPa

80 ligament was uncracked, the tensile load increased linearly with

the applied displacement as the cementitious matrix carried the
60 majority of the load. This was accompanied by a proportional
increase in the stress intensity at the notch tip. Once the stress
intensity exceeded the fracture toughness of the matrix, a sudden
40
crack propagation occurred, resulting in a loss of the tensile stress
previously carried by the matrix (Fig. 18). After this point, the
20 Test curves applied tensile load was equal to the bridging stress transferred
Average by the fibers across the crack. The tensile load increased again with
0 the crack opening until the collective bridging capacity of the fibers
0 1000 2000 3000 4000 5000 6000
was exhausted. The bridging stress decreased gradually after this
Strain/με
point as an increasing number of fibers were either pulled out or
Fig. 17. Compressive stress-strain curves of UHP-ECC prisms. broken.
 K.-Q. Yu et al. / Construction and Building Materials 158 (2018) 217–227 225

18 the r-d curve was 2086.08 N/m with a C.o.V. of 6%. The Jtip could be
calculated by Eq. (3) with Km = 1.02 MPa m1/2 for the UHP-ECC
16
matrix, which was determined by the three point bending test.
14 Em = 41.2 GPa was assumed equal to the composite tensile modu-
lus. The average Jtip calculated from the three point bending tests
12
of the matrix was 24.75 N/m with a C.o.V. of 10%. Therefore, the
Stress/MPa

10 PSH value, (i.e., J0 b/Jtip) was equal to 83.0, which satisfied fully the
energy criterion as described in Eq. (4) and thus ensured the mul-
8 tiple cracking property of UHP-ECC. The high Lf/df PE fiber used in
6 the present research increased largely the bond strength between
the fiber and the matrix, which increased significantly the comple-
4 mentary energy Jtip.
2
4.6. Microstructure of UHP-ECC
0
0.0 0.5 1.0 1.5 2.0
Crack opening/mm The microstructure of the tested specimen was investigated by
Hitachi S3400 N scanned electron microscope (SEM) analysis. The
Fig. 18. Bridging stress (r)-crack opening (d) relation. samples were prepared by taking small pieces from the dumbbell
specimens on the fractured surface. The microstructure and mor-
Only the ascending portion of the r-d curve up to the bridging phology of the UHP-ECC samples were observed on fractured sur-
capacity was utilized (shown in Fig. 18) is needed to check the faces using the secondary electron imaging. Fig. 19(a) and (b) show
multiple cracking criteria [8]. The average bridging capacity (i.e., the images of the PE fiber and PE fiber/matrix interface. Due to the
the peak of r-d curve) measured at the notched sections of the five high strength of the matrix and the large Lf/df of PE fiber, the bond
specimens was 14.12 MPa, which was higher than the cracking strength between the PE fiber and the matrix was strong enough to
strength of matrix (i.e., 10.23 MPa). The average J0 b calculated from fracture the PE fiber. Several fracture surfaces of PE fibers were

(a) Fiber fracture and surface damage (b) Fiber/matrix interface

Fig. 19. SEM images of PE fiber and fiber/matrix interface.

(a) Specimen cross section (b) Tobermorite in one pore

Fig. 20. SEM images of specimen matrix.
226 K.-Q. Yu et al. / Construction and Building Materials 158 (2018) 217–227
Fig. 21. EDS analysis at point A of tobermorite.
observed in Fig. 19(a). The lateral surface of PE fiber also suffered
severe damage during the pull out or the fracture process.
Fig. 20(b) and (a) also revealed that the UHP-ECC has very dense
microstructures. In general, the matrix phase of UHP-ECC is signif-
icantly denser and more homogeneous than that of normal con-
crete, leading to very low porosity. A few entrapped air pores
were also found in the UHP-ECC matrix [Fig. 20(a)]. Most of these
spherical pores with the diameter ranging between 10–150 lm
were formed possibly due to the side effect of high amount of
superplasticizer. These pores were partially filled with tobermorite
[Fig. 20(b)]. Energy dispersive spectroscopy (EDS) analysis in
Fig. 21 showed that the Ca/Si, S/Ca and Al/Ca ratios of this type
of tobermorite were 1.78, 0.027 and 0.17, respectively.
5. Conclusions
Ultra high performance engineered cementitious composites
(UHP-ECC) that combines the high strength and high ductility
has been developed in this research. Ultra-high-molecular-weight
polyethylene fibers were used and special attention has been paid
to the mix process to ensure a satisfactory material property.
Through a comprehensive test program on the mechanical proper-
ties the following findings have been obtained:
1. The average tensile strength of the Dumbbell specimens
approached 17.42 MPa with a C.o.V of 6%. The corresponding
average rupture strains was 8.17% with the COVs of 5%. The
average initial cracking stress was 10.3 MPa with a C.o.V of
5%. The specific fracture energy of UHP-ECC under direct ten-
sion was greater than 1500 kJ/m3.
2. The average modulus of rupture (MOR) of UHP-ECC beams
could achieve 27.68 MPa with a C.o.V of 4%. Along with such
high MOR, the beams exhibited extremely high ductility, as
the average of the mid-span net deflection corresponding to
the MOR achieved about 2.5% of the span length.
3. The average compressive strength values were 121.5 MPa and
112.69 MPa for 70.7 mm and 100 mm cubes respectively. The
compressive strength decreased slightly with the increasing
size due to the size effect. The average compressive Young’s
modulus was 44.26 GPa with a C.o.V of 0.7%. The strain corre-
sponding to the peak stress reached 2954 le.
4. Due to the high strength of the UHP-ECC matrix and the large Lf/
df, the bond strength between the PE fiber and matrix was
strong enough to fracture the PE fiber, which was also validated
by SEM observation. The lateral surface of PE fiber suffered sev-
ere damage during the pull out or the fracture process.
Acknowledgements
The authors are grateful for the financial support provided by
the National Key Research Program of China (Grant No:
2017YFC0703403), The Research Grants Council of the Hong Kong
SAR (Project Code: PolyU 5145/13E), the National Natural Science
Foundation of China (Project codes: 51478406 and 51278441).
The authors also appreciate the DSM Co. Ltd for providing the PE
fibers, the BASF chemical company for providing the HRWR and
the help from Mr Houming ZHU during the experiments.
References
[1] P. Richard, M. Cheyrezy, Composition of reactive powder concretes, Cem.
Concr. Res. 25 (7) (1995) 1501–1511.
[2] M. Cheyrezy, V. Maret, L. Frouin, Microstructural analysis of RPC (reactive
powder concrete), Cem. Concr. Res. 25 (7) (1995) 1491–1500.
[3] Y. Zhang, W. Sun, S. Liu, C. Jiao, J. Lai, Preparation of C200 green reactive
powder concrete and its static-dynamic behaviors, Cem. Concr. Compos. 30 (9)
(2008) 831–838.
[4] H. Yiğiter, S. Aydın, H. Yazici, A.S. Karabulut, Mechanical performance of low
cement reactive powder 521 concrete (LCRPC), Compos. Part B: Eng. 43 (8)
(2012) 2907–2914.
[5] W. Zheng, B. Luo, Y. Wang, Compressive and tensile properties of reactive
powder concrete with steel fibres at elevated temperatures, Constr. Build.
Mater. 41 (2013) 844–851.
[6] FHWA-HRT-13-060, Ultra-High Performance Concrete: A State-of-the-Art
Report for the Bridge Community, June 2013.
[7] Y. Shao, S.P. Shah, Mechanical properties of PVA fiber reinforced cement
composites fabricated by extrusion processing, ACI Mater. J. 94 (6) (1997) 555–
564.
 K.-Q. Yu et al. / Construction and Building Materials 158 (2018) 217–227
[8] V.C. Li, S. Wang, C. Wu, Tensile strain-hardening behavior of polyvinyl alcohol
engineered cementitious composite (PVA-ECC), ACI Mater. J. 98 (6) (2001)
483–492.
[9] P. Jun, V. Mechtcherine, Behaviour of strain-hardening cement-based
composites (SHCC) under monotonic and cyclic tensile loading: part 1–
experimental investigations, Cem. Concr. Compos. 32 (10) (2010) 801–809.
[10] G.P.A.G. Van Zijl, Improved mechanical performance: shear behaviour of
strain-hardening cement-based composites (SHCC), Cem. Concr. Res. 37 (8)
(2007) 1241–1247.
[11] G.P.A.G. Van Zijl, F.H. Wittmann, B.H. Oh, Durability of strain-hardening
cement-based composites (SHCC), Mater. Struct. 45 (10) (2012) 1447–1463.
[12] K.Q. Yu, J.G. Dai, Z.D. Lu, et al., Mechanical properties of engineered
cementitious composites subjected to elevated temperatures, J. Mater. Civ.
Eng. 27 (10) (2015) 04014268.
[13] K.Q. Yu, Z.D. Lu, J. Yu, Residual compressive properties of strain-hardening
cementitious composite with different curing ages exposed to high
temperature, Constr. Build. Mater. 98 (2015) 146–155.
[14] L.L. Kan, H.S. Shi, A.R. Sakulich, V.C. Li, Self-healing characterization of
engineered cementitious composite materials, ACI Mater. J. 107(6) (2010).
[15] P. Acker, M. Behloul, DuctalÒ technology: A large spectrum of properties, a
wide range of applications, in: Proc. of the Int. Symp. on UHPC Kassel,
Germany, 2004, pp. 11–23.
[16] B.A. Graybeal, Material property characterization of ultra-high performance
concrete. US department of transportation, Federal Highway Administration,
2006, pp. 1–188.
[17] A. Kamal, M. Kunieda, N. Ueda, H. Nakamura, Evaluation of crack opening
performance of a repair material with strain hardening behavior, Cem. Concr.
Compos. 30 (10) (2008) 863–871.
[18] M. Kunieda, M. Hussein, N. Ueda, H. Nakamura, Enhancement of crack
distribution of UHP-SHCC under axial tension using steel reinforcement, J. Adv.
Concr. Technol. 8 (1) (2010) 49–57.
[19] U. Maeder, I. Lallemant-Gamboa, J. Chaignon, J.P. Lombard, Ceracem, a new
high performance concrete: Characterizations and applications, in: Schmidt,
et al. (Eds.), Ultra high Performance Concrete, Kassel, Germany, 2004, pp 59–
68.
[20] G. Orange, P. Acker, C. Vernet, A new generation of UHP concrete: DuctalÒ.
Damage resistance and micromechanical analysis, in: Third International
Workshop on High Performance Fiber Reinforced Cement Composites
(HPFRCC3), Mainz, Germany, 1999, pp. 101–111.
[21] Orange G, Dugat J, Acker P. DuctalÒ: New ultra high performance concretes.
Damage resistance and micromechanical analysis, in: RILEM Symposium on
Fibre-Reinforced Concretes, 2000, pp. 781–790.
[22] S.H. Park, D.J. Kim, G.S. Ryu, K.T. Koh, Tensile behavior of ultra high
performance hybrid fiber reinforced concrete, Cem. Concr. Compos. 34 (2)
(2012) 172–184.
[23] C. Sujiravorakul, Development of high performance fiber reinforced cement
composites using twisted polygonal steel fibers [Ph.D. thesis], University of
Michigan, Ann Arbor, 2010, pp. 1–230.
[24] K. Wille, D.J. Kim, A.E. Naaman, Strain-hardening UHP-FRC with low fiber
contents, Mater. Struct. 44 (3) (2011) 583–598.
[25] K. Yu, Y. Wang, J. Yu, et al., A strain-hardening cementitious composites with
the tensile capacity up to 8%, Constr. Build. Mater. 137 (2017) 410–419.
[26] L. Ferrara, N. Ozyurt, M. Di Prisco, High mechanical performance of fibre
reinforced cementitious composites: the role of ‘‘casting-flow induced” fibre
orientation, Mater. Struct. 44 (1) (2011) 109–128.
[27] K. Tosun-Felekoğlu, B. Felekoğlu, R. Ranade, B.Y. Lee, V.C. Li, The role of flaw
size and fiber distribution on tensile ductility of PVA-ECC, Compos. Part B: Eng.
56 (2014) 536–545.
[28] D.Y. Yoo, N. Banthia, S.T. Kang, Y.S. Yoon, Effect of fiber orientation on the rate-
dependent flexural behavior of ultra-high-performance fiber-reinforced
concrete, Compos. Struct. 157 (6) (2016) 62–70.
[29] D.Y. Yoo, S.T. Kang, Y.S. Yoon, Effect of fiber length and placement method on
flexural behavior, tension-softening curve, and fiber distribution
characteristics of UHPFRC, Constr. Build. Mater. 64 (5) (2014) 67–81.
227
[30] M.D. Lepech, V.C. Li, Preliminary findings on size effect in ECC structural
members in flexure, in: Proceedings of the Seventh International Symposium
on Brittle Matrix Composites, Warsaw, Poland, 2003, pp. 57–66.
[31] D.Y. Yoo, N. Banthia, S.T. Kang, Y.S. Yoon, Size effect in ultra-high-performance
concrete beams, Eng. Fract. Mech. 157 (3) (2016) 86–106.
[32] R. Ranade, V.C. Li, M.D. Stults, F.H. William, S.R. Todd, Composite properties of
high-strength, high-ductility concrete, ACI Mater. J. 110 (4) (2013) 413–422.
[33] R. Ranade, V.C. Li, M.D. Stults, F.H. William, S.R. Todd, Micromechanics of high-
strength, high-ductility concrete, ACI Mater. J. 110 (4) (2013) 375–384.
[34] D.B. Marshall, B.N. Cox, A J-integral method for calculating steady-state matrix
cracking stresses in composites, Mech. Mater. 7 (2) (1988) 127–133.
[35] V.C. Li, C.K.Y. Leung, Steady-state and multiple cracking of short random fiber
composites, J. Eng. Mech. 118 (11) (1992) 2246–2264.
[36] T. Kanda, V.C. Li, Multiple cracking sequence and saturation in fiber reinforced
cementitious composites, Concr. Res. Technol. 9 (2) (1998) 19–33.
[37] A.E. Naaman, H. Najm, Bond-slip mechanisms of steel fibers in concrete, ACI
Mater. J. 88 (2) (1991) 135–145.
[38] Z. Lin, T. Kanda, V.C. Li, On interface property characterization and
performance of fiber reinforced cementitious composites, Concr. Sci. Eng. 1
(3) (1999) 173–184.
[39] GB/T 18046 Ground granulated blast furnace slag used for cement and
concrete [S], National Standardization Management Committee, 2008, pp. 1–
10.
[40] S.C. Pal, A. Mukherjee, S.R. Pathak, Investigation of hydraulic activity of ground
granulated blast furnace slag in concrete, Cem. Concr. Res. 33 (9) (2003) 1481–
1486.
[41] S. Kumar, A. Bandopadhyay, V. Rajinikanth, Improved processing of blended
slag cement through mechanical activation, J. Mater. Sci. 39 (10) (2004) 3449–
3452.
[42] S. Kumar, R. Kumar, A. Bandopadhyay, T.C. Alex, B.R. Kumar, S.K. Das, S.P.
Mehrotra, Mechanical activation of granulated blast furnace slag and its effect
on the properties and structure of portland slag cement, Cem. Concr. Compos.
30 (8) (2008) 679–685.
[43] H. Yazici, M.Y. Yardımcı, S. Aydin, A.S. Karabulut, Mechanical properties of
reactive powder concrete containing mineral admixtures under different
curing regimes, Constr. Build. Mater. 23 (3) (2009) 1223–1231.
[44] C. Shi, D. Wang, Wu. L. and Wu, Z., The hydration and microstructure of ultra
high-strength concrete with cement-silica fume-slag binder, Cem. Concr.
Compos. 61 (2015) 44–52.
[45] CECS. Standard test methods for fiber reinforced concrete [S]. China
Engineering and Construction Society, Beijing, China, 2009, pp. 44–48.
[46] JSCE. Recommendations for Design and Construction of High Performance
Fiber Reinforced Cement Composites with Multiple Fine Cracks. Japan Society
of Civil Engineers, Tokyo, Japan, 2008, pp. 1–16.
[47] J. Kozicki, J. Tejchman, Experimental investigations of strain localization in
concrete using Digital Image Correlation (DIC) technique, Arch. Hydro-Eng.
Environ. Mech. 54 (1) (2007) 3–24.
[48] S.G. Shah, J.M.C. Kishen, Fracture properties of concrete-concrete interfaces
using digital image correlation, Exp. Mech. 51 (3) (2011) 303–313.
[49] S. Choi, S.P. Shah, Measurement of deformations on concrete subjected to
compression using image correlation, Exp. Mech. 37 (3) (1997) 307–313.
[50] B. Graybeal, M. Davis, Cylinder or cube: strength testing of 80 to 200 MPa (11.6
to 29 ksi) ultra-high-performance fiber-reinforced concrete, ACI Mater. J. 105
(6) (2008) 603–609.
[51] J.R. Del Viso, J.R. Carmona, G. Ruiz, Shape and size effects on the compressive
strength of high-strength concrete, Cem. Concr. Res. 38 (3) (2008) 386–395.
[52] B.A. Graybeal, Compressive behavior of ultra-high-performance fiber-
reinforced concrete, ACI Mater. J. 104 (2) (2007) 146–152.
[53] K. Wille, A.E. Naaman, G.J. Parra-Montesinos, Ultra-high performance concrete
with compressive strength exceeding 150 MPa (22 ksi): a simpler way, ACI
Mater. J. 108 (1) (2011) 46–54.