Cement and Concrete Composites 88 (2018) 17e28

Contents lists available at ScienceDirect

Cement and Concrete Composites

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

Damage investigation of ultra high performance concrete under direct

tensile test using acoustic emission techniques
Jun-Yan Wang*, Jun-Yuan Guo
Key Laboratory of Advanced Civil Engineering Materials, Tongji University, Ministry of Education, Shanghai 201804, China

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

Article history: In this study, acoustic emission (AE) analysis method was applied to monitor the damage evolution
Received 11 August 2017 process of ultra high performance concrete (UHPC) under direct tensile test. Three types of UHPCs,
Received in revised form including high strain-hardening UHPC, low strain-hardening UHPC and strain-softening UHPC were
4 January 2018
investigated. Meanwhile, the crack width developments of UHPCs during the tensile test were measured.
Accepted 8 January 2018
Test results show that high strain-hardening UHPC exhibited high ductility by forming multiple
Available online 10 January 2018
microcracks invisible to naked eyes (typically below 0.05 mm) in the strain-hardening stage. The crack
width-strain curves indicate that increasing the ultimate tensile strain of UHPC can improve its crack
Keywords:
Ultra-high performance concrete
width control ability effectively. The AE analysis method could effectively detect the internal damages of
Strain-hardening the high strain-hardening UHPC at a strain of 600mε. At that time, the crack width was smaller than
Crack width control ability 0.01 mm that could not be detected by crack width measuring instrument in this study. For three types of
Acoustic emission UHPCs, damages were generated around the localized crack during the strain-softening stage. In a word,
Damage evolution mechanism the AE analysis method provides strong evidence to the multiple cracking behavior of UHPC during the
strain-hardening stage, and provides a clear explanation to the identical damage evolution mechanism
for three types of UHPCs during the strain-softening stage.
© 2018 Elsevier Ltd. All rights reserved.

1. Introduction hardening cementitious composite (SHCC). In the past decades,

Ultra high performance concrete (UHPC), which was developed
in the mid-1990s, has been considered as one of the most prom-
ising construction materials for architectural or civil structures
subject to extreme loads, such as earthquakes, impacts, and blasts
[1]. The UHPC is distinguished from high performance concrete and
fiber reinforced concrete by its own high tensile strength and
durability. Therefore, many UHPC design guidelines regard the
UHPC tensile properties as the most important parameters in the
structural design. In 2016, MCS-EPFL issued the latest recommen-
dation [2]. In this recommendation, UHPC is divided into three
types: UO (strain softening), UA (ultimate tensile hardening strain
is higher than 1500mε (micro strain) and UB (ultimate tensile
hardening strain is higher than 2000mε) based on the tensile be-
haviors after the matrix cracks. Typically, the elastic tensile strain of
UHPC is around 200mε [3].
The UA and UB types of UHPC belong to the family of the strain-
* Corresponding author.
E-mail address: wangjunyan@tongji.edu.cn (J.-Y. Wang).
SHCC has evolved with intensified research, characterized by its
unique macroscopic pseudo strain-hardening behavior after first
cracking under uniaxial tension. SHCCs, also referred to as high
performance fiber reinforced cementitious composites (HPFRCC)
[4], developed multiple cracks under tensile load. It shows a sharp
contrast to the single crack and tension softening behavior of or-
dinary concrete or conventional fiber reinforced concrete. The
multiple cracking accompanying pseudo strain-hardening implies
high ductility, large energy absorption capacity and high toughness
[5] at the material level and prevents catastrophic fracture at the
structural level, thus contributing to structural safety. Meanwhile,
material tensile strain-hardening (ductility) has been gradually
recognized as having a close connection with structural durability
by suppressing localized cracks with large width [6]. As a special-
ized version of SHCC, Engineered Cementitious Composites (ECC)
possesses tensile strain capacity from 2% to 5% [7e11], which is two
orders of magnitude higher than that of normal concrete. ECC
achieves such high ductility by forming multiple tight microcracks
with self-controlled crack width typically below 100 mm [9]. The
tight crack width allows ECC to maintain low water permeability
even under cracked state, which contributes to the excellent

https://doi.org/10.1016/j.cemconcomp.2018.01.007
0958-9465/© 2018 Elsevier Ltd. All rights reserved.
18 J.-Y. Wang, J.-Y. Guo / Cement and Concrete Composites 88 (2018) 17e28

durability of ECC under loading [12].

The UA and UB types of UHPC can achieve strain-hardening
behavior up to a strain around 0.2%e0.8%, which is smaller than
that of ECC, but higher than the yield strain of steel bar, so that it is
supposed to be enough for reinforced UHPC structures to have good
crack width control ability. Although many researchers focused on
the tensile strain-hardening behaviors of UHPC [13,14], the
insightful relationship between the multiple-cracking behavior and
the tensile strain-hardening capability of UHPC has not been well
understood so far. It is probably because the cracks of UHPC during
the strain-hardening stage are usually invisible to the naked eyes.
Nowadays, the Acoustic Emission (AE) technique [15] has been
extensively applied to detect internal damage in structures and
materials. As a non-destructive evaluation technique and an useful Fig. 1. Schematic diagram of acoustic emission analysis system.
tool for material research [16], AE is an efficient method to monitor
the damage evolution mechanism in concrete under various
loading cases such as flexural, cyclic, impact, freezing-thawing,
fatigue loading, and even under chemical influence like corrosion mechanical response of specimens. Source localisation is based on
[17e20]. Recently, a number of studies have been conducted to the arrival times of P-waves, recorded by several transducers
evaluate the internal damage of concrete by using AE techniques. (generally 4 in a 2D analysis, and 6 to 8 in 3D analyses) [23,24]. In
Qingli Dai [21] used the AE techniques with statistical analysis to 2D and 3D localisation schemes, the unknown parameters are the
study the damage evolution process of single-edge notched beam source location and the origin time at which the event is generated.
tests with normal strength concrete and UHPC specimens. The re- In order to obtain detailed information about the accuracy of the
sults indicated that the developed AE and statistical analysis event localisation algorithm, these unknown parameters need to be
techniques can be applied to study damage process under me- over-determined [24]. Classically, in a 3D analysis there are four
chanical loading for brittle materials such as concrete. Otsuka's unknown parameters: the origin time and the three coordinates of
work [22] applied the 3D AE techniques with X-ray computed to- the event location. Four transducers provide a unique solution.
mography to detect the fracture process zone in notched concrete When more than four transducers are used, the unknown param-
specimens. This study showed that the AE event concentrated at eters are over-determined. As such, AE source location in a great
the notch tip and the accumulation of microcracks were also majority cases is an over-determined problem and the key for the
observed at the same location. Overall, the AE technique is proved success of a method is that it can effectively provide the minimum
to be an effective method to investigate the damage evolution residual [25].
process of concrete.
In order to understand the relationship between the crack width
control ability and the different tensile behaviors of the UHPCs, AE
2.3. Exhaustion method of 3D AE source location
technique is applied to monitor the damage evolution process of
UHPC under direct tensile test in this study. Three types of UHPCs,
In the 3D AE source location test, the wave propagation velocity
including high strain-hardening UHPC (ultimate tensile hardening
vP was determined independently by pencil lead test, more details
strain is higher than 2000mε), low strain-hardening UHPC (ultimate
are given in section 3.4, the arrival time ti at i transducer can be
tensile hardening strain is lower than 2000mε) and strain-softening
monitored from the AE analysis system. So the observed difference
UHPC (without strain-hardening stage) were investigated. Mean-
between two transducers to the AE source distance d0i can be
while, the crack width of UHPC during the test was measured by the
calculated as:
crack width measuring instrument and camera with micro photo
lens. These tests aim to describe the damage evolution mechanism
doi ¼ ðti  tiþ1 Þ  vp (1)
of three types of UHPCs under direct tension.
And the calculated difference between two transducers to the
2. Mechanism of acoustic emission sources location AE source distance dci can be calculated as:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2.1. Acoustic emission analysis system
dci ¼ ðxae  xi Þ2 þ ðyae  yi Þ2 þ ðzae  zi Þ2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Acoustic Emission (AE) is a micro seismic (elastic) wave gener-  ðxae  xiþ1 Þ2 þ ðyae  yiþ1 Þ2 þ ðzae  ziþ1 Þ2 (2)
ated from dislocations, micro cracking and other irreversible
changes in a stressed material [23]. The schematic diagram of AE
analysis system is given Fig. 1. As shown in Fig. 1, the transmitted
waves are detected by transducers on the surface of a specimen.
 xae, yae, zae: The coordinate of the unknown AE source;
Through the preamplifier, all waveforms are recorded and stored by
 xi, yi, zi: The coordinate of the i transducer, i ¼ 1,2,3 … n,nþ1;
a windows-based AE data operation program and the program can
be used to extract the AE parameters for analysis, for example, the
The residual ri,iþ1 is the difference between the observed and
3D coordinates (xae, yae and zae) of the AE source and the origin time
calculated arrival distance as [23]:
tae when AE source was generated.
ri;iþ1 ¼ doi  dci (3)
2.2. Detecting method of AE source location
The least-squares method is used to be the criteria of minimum
AE source location is used to investigate the shape and extend of residual, and defines the total residuals as the sum of the squares of
the fracture process zone, and some correlation with the individual residuals [26]:
 J.-Y. Wang, J.-Y. Guo / Cement and Concrete Composites 88 (2018) 17e28
X
n  2 X
n
r¼ ri;iþ1 ¼ ðdoi  dci Þ2
i¼1 i¼1
For the objective function (4), the range of unknown AE source
coordinates is actually related to the dimension D of the specimen:
0  xae  Dx ; 0  yae  Dy ; 0  zae  Dz
By calculating the value of the objective function (4) on the node
of each grid and comparing the value, the xae, yae and zae which
satisfy the minimum value by function (4) can be regard as the
coordinates of the AE sources. The origin time tae can be calculated
as:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
tae ¼ ðxae  xi Þ2 þ ðyae  yi Þ2 þ ðzae  zi Þ2 v þ ti
Based on the coordinates (xae, yae and zae) and origin time tae of
AE source, the damage evolution process can be described.
3. Experimental program
3.1. UHPC mixture and specimen preparation
The UHPC materials used in this study were commercial prod-
ucts named TENACAL® provided by Shanghai Royang Innovative
Material Technologies Co., Ltd. TENACAL® consists of three com-
ponents, water, UHPC premixed powder and steel fibers.
Table 1 provides the mix proportion of the UHPC matrix. The
maximum aggregate size of the UHPC matrix is 1.18 mm. The vol-
ume fractions of steel fibers in high strain-hardening UHPC, low
strain-hardening UHPC and strain-softening UHPC are 2.5%, 2.0%
and 1.5% respectively. A type of straight steel fibers with length of
13 mm and diameter of 0.2 mm and with brass coating were used,
which properties are given in Table 2.
A laboratory mixer with 60-L capacity was used to prepare the
UHPC mixture. UHPC premixed powder was first dry-mixed for
about 1 min. Water was then added gradually and mixed for
another 3 min. When the mixture showed suitable workability and
viscosity for uniform distribution of fibers, the steel fibers were
dispersed carefully by hand into the mixture and mixed for another
3 min. Finally, the UHPC mixtures were then placed in the molds
without vibration due to the self-compacting properties of the
UHPC mixtures. The specimen dimension is shown in Fig. 2(a).
Since the width (50 mm) of the specimens is longer than three
times of the length of the fibers (13 mm), it could reduce the in-
fluence of the fiber orientation distribution. For each mixture,
specimens were prepared as follows: 1) nine specimens for direct
tensile test; 2) three 100 mm cubic specimens for 28-day
compressive strength test according to Chinese standard GB/T
31387e2015; 3) three 100  100  400 mm prisms for flexural
strength test according to ASTM C1609; 4) three
100  100  300 mm prisms for elastic modulus test according to
Chinese standard GB/T 31387e2015. After casting, these specimens
were covered with plastic sheets and stored at room temperature
for 24 h, then cured in the water at a temperature of 20  C until 28
days. The UHPC tensile specimen after curing is shown in Fig. 2(b).
Table 1
Mix proportions of UHPC matrix.
Cement Silica fume
733 kg/m3 220 kg/m3
19
3.2. Test method for stress-strain curve
(4)
The direct tensile test setup is described in Fig. 3(a). Both ends of
the specimen have hinge systems to maintain pure tensile load
without moment. A universal testing machine (WDW-300 servo-
controlled testing system) running in displacement control
manner was used to conduct the tensile tests. Four linear variable
(5) differential transformers (LVDTs) were attached to the specimen
over a gauge length L ¼ 150 mm to measure the average elongation.
The load from the load cell and the average elongation from the
four LVDTs were used to determine the average tensile stress,
average tensile strain according to equation (7), and then the ten-
sile stress-strain curve of each specimen.
ε ¼ (Dl1þDl2þDl3þDl4)/4L  100% (7)
(6)
p
Where,
 Dl1, Dl2, Dl3, Dl4: The displacement value of the four LVDTs;
3.3. Test method for crack width-strain curve
As shown in Fig. 3 (b) and (c), during the UHPC direct tensile
test, the positions of the cracks generated at different tensile strain
were firstly located by the Canon EOS 5DSR camera (50.6 MP) with
zoom lens (EF17-40 mm f/4 L USM). The camera's picture could be
displayed on the computer screen in real time. After locating the
crack, the width of the crack was immediately measured by the
crack width measuring instrument with 0.01 mm resolution. In
order to take the crack image more clearly and measure the crack
width more accurately, the loading rate of the tensile testing was
set as 0.1 mm/min. The load for the start and end of the test were
0.5 kN and 20% of the peak stress, respectively. The whole test
program lasted about 1 h.
3.4. Test method for localisation of AE sources
The AE analysis system described in section 2 was applied to
monitor the damage evolution process of the UHPCs. The setup of
the direct tensile test with AE analysis system is given in Fig. 3. As
shown in Fig. 3 (a) and (d), in order to cover the expected zone of
the process, 8 AE transducers were placed in a rectangular array
just above the surface of double sides of the specimens. The AE
transducers were coupled to the surface of UHPC specimens with
Vaseline locating at the positions shown in Fig. 3 (d).
In this study, pencil lead test and localisation test of AE sources
were conducted. The pencil lead tests were conducted with UHPC
specimens by following the ASTM E976-99 standard [27], which
used HB or 2B pencil lead with 0.5 mm diameter as a manual AE
source. During the test, the pencil lead elongation was set to be
about 2.5 mm and the angle between the pencil lead and the UHPC
specimen surface was about 30 . The pencil lead was broken within
30 mm to transducers employing approximately equal force at each
time to get almost the same AE sources. The purpose of pencil lead
test was to determine the AE wave propagation velocity through
UHPC materials and the noise level of the test environment so as to
Ground quartz Quartz sand Water Superplasticizer
220 kg/m3 982 kg/m3 147 kg/m3 3.67 kg/m3
20 J.-Y. Wang, J.-Y. Guo / Cement and Concrete Composites 88 (2018) 17e28
Table 2
Properties of steel fibers.
Fiber shape Tensile strength Elastic modulus
/MPa /GPa
Straight smooth 2500 200
Fig. 2. Dimension and photo of UHPC tensile specimen.
set the corresponding threshold and amplification. As shown in
Fig. 3 (d), the pencil lead was broken 3 times in each location, there
had two different locations. Through the pencil lead test, the
threshold of the AE system was set as 35 dB. The stress level of AE
activities above this threshold are detected by transducers, and
recorded by the windows-based AE data operation program. Value
for gain of preamplifier was set as 40 dB, sampling frequency was
set as 3 MHz and band characteristic was set as 20 kHze200 kHz.
The windows-based AE data operation program was carried out
in parallel with the UHPC direct tensile test as shown in Fig. 3 (e).
The coordinates (xae, yae and zae) and origin time tae of AE source
monitored by this program can be plotted for description of the
damage evolution process of the UHPC specimen under tensile
loading test. An example is given in Fig. 3 (e).
4. Test results and discussion
4.1. Compressive strength, flexural strength and elastic modulus of
three types of UHPCs
The mechanical properties of three types of UHPCs are sum-
marized in Table 3, including 28-day compressive strength
(100 mm*100 mm*100 mm cubic specimen), flexural strength
(100 mm*100 mm*400 mm prismatic specimen) and elastic
modulus (100 mm*100 mm*300 mm prismatic specimen). Table 3
shows that with an increase in fiber volume fraction from 1.5% to
2.5% results in an increase of compressive strength and flexural
Length/mm Diameter/mm Aspect ratio Density/kg$m3
13 200 65 7850
strength of the UHPCs, while the elastic modulus remains almost
the same.
4.2. Tensile properties of three types of UHPCs
Fig. 4 shows the axial tensile stress-strain curves of three types
of UHPCs, each type of UHPCs contains three specimens. The stress-
strain curves of high strain-hardening UHPC (Fig. 4(c)) start with a
steep initial ascending portion up to first cracking, followed by a
strain-hardening branch where a considerable reduction in
modulus of elongation, i.e. the ratio of stress to strain. The point
where first cracking occurs is mainly influenced by the tensile
strength of the matrix, which is called the elastic tensile strength
(fUte). The peak point at the end of the strain-hardening branch is
determined by the fiber bridging capacity, which is defined as the
ultimate tensile strength (fUtu) [3,28]. Following the peak point,
there is generally a descending branch which is strain-softening
stage. Generally, the tensile behavior of high strain-hardening
UHPC can be divided into three stages, including strain based
elastic stage, strain based strain-hardening stage and strain based
strain-softening stage. The low strain-hardening UHPC (Fig. 4(b))
also has three stages based on the tensile behavior, while the strain-
softening UHPC (Fig. 4(a)) exhibits strain softening immediately
following the first cracking.
Detailed information about the tensile properties including fUte,
εUte, fUtu, εUtu is summarized in Table 4. According to the literature
[13], the method of detecting the fUte is illustrated in Fig. 4. If the
J.-Y. Wang, J.-Y. Guo / Cement and Concrete Composites 88 (2018) 17e28 21

Fig. 3. Setup of the direct tensile test with AE analysis system.

22 J.-Y. Wang, J.-Y. Guo / Cement and Concrete Composites 88 (2018) 17e28

Table 3
Mechanical properties of three types of UHPCs.

Specimen types Compressive strength Flexural strength Elastic modulus

/MPa(CoV) /MPa(CoV) /GPa(CoV)

Strain-softening UHPC 125.7 (0.031) 21.37(0.080) 47.4(0.026)

Low strain-hardening UHPC 137.6 (0.051) 24.78(0.129) 48.5(0.037)
High strain-hardening UHPC 155.4 (0.042) 32.33(0.073) 48.9(0.021)

there is no clear first peak point, two lines are drawn from both the
linear elastic and hardening stages, and the intersection point be-
tween the two lines is selected as fUte in this research.
The average elastic tensile strength fUte of high strain-hardening
UHPC, low strain-hardening UHPC and strain-softening UHPC are
10.4 MPa, 8.7 MPa and 7.7 MPa, respectively, and the corresponding
strain εUte are 0.024%, 0.023% and 0.021%, respectively. The average
ultimate tensile strength fUtu in high strain-hardening UHPC and
low strain-hardening UHPC is 12.6 MP and 9.4 MPa, respectively,
and the corresponding strain εUtu is 0.412% and 0.116%, respectively.
It can be concluded that elastic tensile strength, ultimate tensile
strength and their corresponding strains of high strain-hardening
UHPC are the highest. The ultimate tensile strain of high strain-
hardening UHPC is higher than the yield strain of steel bar (0.2%),
and achieves strain-hardening with high ductility under direct
tensile loading, which would contribute to the ultimate tensile
loading capacity of the reinforced ultra high performance concrete
(R-UHPC) structures.

4.3. Crack control ability of three types of UHPCs

In this paper, crack with a minimum width of 0.01 mm is defined

as the detectable crack that can be measured by the crack width
measuring instrument during the direct tensile test. Detectable
crack with a width above 0.05 mm is defined as a crack visible to
the naked eye [28]. The 0.05 mm crack width is usually regards as
the critical value for concrete durability. For general industrial and
civil buildings, they can be supposed as structures without cracks if
crack width is less than 0.05 mm [29]. Fig. 5 presents the repre-
sentative results of the crack width-strain curves and stress-strain
curves of three types of UHPCs. As shown in Fig. 5, point A and point
B represent the crack width at 0.05 mm and the strain at 2000mε
(yield strain of steel bar), respectively. A photo of 0.05 mm crack
was shown in Fig. 6, which was taken by the Canon camera with
micro photo lens MP-E65 mm f/2.8 1e5x. Table 5 summarizes the
tensile properties of three types of UHPCs at these two points. The
analysis of the results is summarized as below.

4.3.1. Strain-softening UHPC

As shown in Fig. 5(a), strain-softening UHPC exhibits the first
cracking when it reached the strain-softening stage. The first crack
kept opening so that it reached a crack width of 0.05 mm imme-
diately when the tensile strain was only 360mε (point A). After that,
the crack width-strain of strain-softening UHPC started with a
steep initial ascending portion up to point B. Comparing with the
point A, the crack width of point B increased by 10 times, while the
strain only reached 2000mε. In addition, during the crack width
measurement, no more cracks was observed, and only the first
crack kept opening together with the increasing tensile strain. The
results indicate that the elongation growth during the strain-
Fig. 4. Stress-strain curves of three types of UHPCs.
softening stage is caused by the single-crack opening (crack
localisation).

tensile stress-strain curve shows a clear load drop right after first 4.3.2. High strain-hardening UHPC
cracking, as shown in Fig. 4(a), the first peak stress in tensile stress- As shown in Fig. 5 (c), high strain-hardening UHPC exhibits
strain curve is selected as fUte. However, as shown in Fig. 4 (b)-(c), if strain-hardening behavior after the first cracking, accompanying by
 J.-Y. Wang, J.-Y. Guo / Cement and Concrete Composites 88 (2018) 17e28 23

Table 4
Tensile properties of three types of UHPCs.

Specimen types fUte fUte,average CoV εUte εUte,average CoV fUtu fUtu,average CoV εUtu εUtu,average CoV
(fUte) (εUte) (fUtu) (εUtu)
MPa MPa % % MPa MPa % %

Strain-softening UHPC 1 8.2 7.7 0.08 0.021 0.021 0.16 / / / / / /

2 8.0 0.018 / /
3 7.0 0.025 / /

Low strain-hardening UHPC 1 9.1 8.7 0.05 0.020 0.023 0.19 9.8 9.4 0.04 0.114 0.116 0.26
2 8.6 0.021 9.4 0.147
3 8.3 0.028 9.0 0.087

High strain-hardening UHPC 1 10.0 10.4 0.04 0.024 0.024 0.06 13.0 12.6 0.03 0.404 0.412 0.03
2 10.3 0.025 12.2 0.426
3 10.8 0.022 12.6 0.405

multiple microcracks indicated by the decreasing modulus of
elongation. During this stage, these microcracks were bridged and
controlled by steel fibers, but they were invisible to the naked eyes.
Photos of the high strain-hardening UHPC specimen at different
strains in the tensile test are shown in Fig. 7. It shows that although
the tensile strain reached 3500mε, the cracks of the specimen still
could not be observed by the naked eyes (crack width smaller than
0.05 mm).
As shown in Fig. 5(c), in the high strain-hardening UHPC spec-
imen, the crack with crack width larger than 0.01 mm cannot be
detected until the tensile strain reached 673mε, because the crack
width measuring instrument has only a resolution of 0.01 mm.
Since 673mε exceeds the elastic tensile strain of high strain-
hardening UHPC (around 250mε), the high strain-hardening UHPC
with strain range from 250mε to 673mε is supposed to be with
cracks. These cracks cannot be detected so that they all have crack
width below 0.01 mm.
The ultimate tensile strain εUtu of high strain-hardening UHPC is
around 4260mε. To better describe the crack pattern, the distribu-
tion of the crack width is plotted in Fig. 8. The crack width ranges
from 0.01 mm to 0.05 mm, and average crack width is 0.03 mm,
which will benefit the impermeability of the high strain-hardening
UHPC with tensile strain below 4000mε. The results indicate that
high strain-hardening UHPC achieves high ductility by forming
multiple microcracks invisible to naked eyes (typically below
0.05 mm) in the strain-hardening stage.
4.3.3. Comparison of three types of UHPCs
As shown in Table 5, it is observed that when strain at point B
(strain at 2000mε), the crack width of high strain-hardening UHPC,
low strain-hardening UHPC and strain-softening UHPC were
0.02 mm, 0.1 mm and 0.5 mm, respectively. The results indicate
that strain-hardening behavior can improve the crack width control
ability of UHPC. In the other hand, when crack width reached
0.05 mm (point A), the corresponding strain of high strain-
hardening UHPC and low strain-hardening UHPC (0.431% and
0.093%, respectively) are very close to their ultimate tensile strain
(0.426% and 0.115%, respectively). Meanwhile, the slope of crack
width-strain curves in the curves of two strain-hardening UHPCs
become significantly steep at point A as shown in Fig. 5(b)-(c),
which is a strong evidence to the crack localisation when UHPC
reached the ultimate tensile strain and turns into strain-softening
stage. In other word, the crack width control ability of UHPC can
be improved by increasing its ultimate tensile strain, which is
meaningful for the durability or structural safety issues of UHPC in
the serviceability limit state.
4.4. Damage evolution mechanism of three types of UHPCs
Figs. 9e11 show the evolution of AE source locations during the
direct tensile test, where the corresponding stress-strain curves of
three types of UHPCs are also illustrated. As shown in Figs. 9e11, six
points are chosen to describe different stages of the evolution of AE
source locations: Point A and C are chosen at 100mε and 600mε,
which represent the elastic stage and strain-hardening stages
(strain-hardening UHPCs) of stress-strain curves, respectively;
Point B, D, E are the points responding to the ultimate tensile
strength in the stress-strain curves of strain-softening UHPC, low
strain-hardening UHPC and high strain-hardening UHPC, respec-
tively; Point F is the point 4000mε later after the peak point of each
type of UHPCs, which represent the strain-softening stage of the
stress-strain curves.
4.4.1. Damage evolution process at different states
It is shown that no AE sources were detected till the point A in
three types of UHPCs, which indicates that the defects of UHPC
generated in the elastic stage cannot be detected by this AE analysis
system. At the ultimate tensile strain of strain-softening UHPC
(point B), 13 AE sources generated and almost all of them lay in a
plane perpendicular to the direction of tension. By contrast, there
was no AE sources in the low strain-hardening UHPC, and only 4 AE
sources were detected in high strain-hardening UHPC. The AE
sources in this stage might be motivated by the first UHPC matrix
cracking and fiber-matrix debonding.
At point C, more AE sources at different height were detected in
strain-hardening UHPCs, indicating that multiple cracks occurred,
while the new AE sources generated in the same plane for strain-
softening UHPC. It is interesting to see that although 53 AE sour-
ces were detected in high strain-hardening UHPC at point C, where
none of the detectable cracks with minimum width of 0.01 mm
were detected. This is a strong evidence to the fact that microcracks
with width smaller than 0.01 mm were generated in the high
strain-hardening UHPC when it has a strain below 600mε.
During the point D to point F, AE source locations of strain-
softening UHPC show a tendency to concentrate around the thin
plane at the same height, while that of strain-hardening UHPCs
show a wider distribution range. Especially for high strain-
hardening UHPC, the AE source locations distribute a widely-
spread over the whole specimen.
To better represent the AE sources distribution, Fig. 12 shows the
distribution of the AE source locations till point E. The specimen is
divided into ten equal parts along the height of 500 mm. In strain-
softening UHPC, AE sources were detected in three parts, where
24 J.-Y. Wang, J.-Y. Guo / Cement and Concrete Composites 88 (2018) 17e28
Fig. 5. Crack width-strain curves and stress-strain curves of three types of UHPCs.
89.61% were concentrated in one part with height of 200e250 mm.
In low strain-hardening UHPC, AE sources were detected in six
parts and the highest density part lay in the height of 100e150 mm,
accounting for 37.27%. In high strain-hardening UHPC, the AE
sources were detected in each part and the highest density part lay
in the height of 200e250 mm, only accounting for 21.48%, which
was the lowest compared with the other two UHPCs. It is observed
that in the same condition of elongation (4050mε), the high strain-
hardening UHPC shows a more homogeneous distribution of AE
sources than the other two UHPCs. Equivalent elongation is
balanced in the form of more cracks, so that the width of each crack
Fig. 6. Micro photo picture of 0.05 mm crack.
of high strain-hardening UHPC can be controlled at a low level. The
AE analysis method provides strong evidence to the multiple
cracking behavior of strain-hardening UHPCs.
4.4.2. Damage evolution process during the strain-softening stage
For three types of UHPCs in this study, following the peak point
of the stress-strain curves, there is generally a descending branch
(the strain-softening stage), which indicates the crack localisation
(single crack opening). According to the MCS-EPFL Recommenda-
tion [2], stress-crack opening law can describe the softening
behavior of UHPC. Crack opening will be converted to strain ac-
cording to the following relationship:
wUt
εUt;soft ¼ (8)
lU
Where,
 εUt,soft: The softening strain of UHPFRC;
 wUt: Opening of discrete softening cracks in UHPFRC
 lU: Reference length used to compute the strain from the crack
opening
This equation indicates that during strain-softening stage, the
tensile elongation of the UHPC specimen will convert to the in-
crease of localized crack width.
In order to investigate the damage evolution process of three
types of UHPCs during the strain-softening stage, the AE source
generated from the peak stress to the point 4000mε later after the
peak point (point F) are plotted in Fig. 13, where the photos of
UHPC specimens after direct tensile test are also shown together.
As shown in Fig. 13(a), 64 AE sources of strain-softening UHPC
were detected in the height of 200e250 mm, which located at
same plane that perpendicular to the direction of tension.
Meanwhile, 39 and 53 AE sources were detected in the low strain-
hardening UHPC and high strain-hardening UHPC, respectively,
and most of them lie in the height of 150e300 mm and
100e250 mm, respectively, which formed inclined planes. By
comparison with the photos of three types of UHPC specimens
after direct tensile test, it can be observed that almost all AE
source locations lie around the localized crack in three types of
UHPCs. This indicates that during the strain-softening stage, the
damage generated around the localized crack. For high strain-
hardening UHPC, 35 AE sources were detected in the height of
300e400 mm during strain-hardening stage, while no AE sources
generated in this area during the strain-softening stage, and only a
 J.-Y. Wang, J.-Y. Guo / Cement and Concrete Composites 88 (2018) 17e28 25

Table 5
Tensile properties of three types of UHPCs at point A and B illustrated in Fig. 5.

Specimen types fUte εUte fUtu εUtu A (w* ¼ 0.05 mm) B (ε* ¼ 0.02%)
/MPa /% /MPa /%
s/MPa ε/% s/MPa w/mm)

Strain-softening UHPC 8.0 0.018 / / 7.8 0.039 7.1 0.50

Low strain-hardening UHPC 9.1 0.020 9.8 0.115 9.6 0.093 8.9 0.10
High Strain-hardening UHPC 10.3 0.025 12.2 0.426 12.0 0.431 11.0 0.50

Note: w ¼ crack width; ε ¼ tensile strain.

Fig. 7. Photos of the high strain-hardening UHPC specimen at different tensile strains.

few of AE sources generated in the position far away from the

localized crack. The result indicates that even though multiple
microcraks generated in the strain-hardening stage, most of them
kept stable during the strain-softening stage, while only the
localized crack kept opening. In a word, during the strain-
softening stage, the damage evolution mechanism that damages
were generated around the localized crack is identical for three
types of UHPCs.

4.4.3. Effect of different dosages of steel fibers on AE source number

The accumulated AE source numbers at different strains are also
presented in Figs. 9e11. As shown in Figs. 9e11, the accumulated AE
sources number at E point (4050mε) of strain-softening UHPC, 72
points, is much less than both of the other two strain-hardening
UHPCs, 220 points for low strain hardening type and 149 points
for high strain hardening type, respectively. The main reason for
Fig. 8. Distribution for crack width of high strain-hardening UHPC specimen at 4000mε
strain.
this phenomenon may be that strain-softening UHPC has only a
26 J.-Y. Wang, J.-Y. Guo / Cement and Concrete Composites 88 (2018) 17e28

Fig. 9. Damage evolution process of the strain-softening UHPC.

Fig. 10. Damage evolution process of the low strain-hardening UHPC.

Fig. 11. Damage evolution process of the high strain-hardening UHPC.

 J.-Y. Wang, J.-Y. Guo / Cement and Concrete Composites 88 (2018) 17e28 27

sources were detected in the height of 100e150 mm, where is the

arc section of the specimen contacting with the grip system.
Further research should be conducted on more specimens to give a
reasonable explanation to the relationship between fiber dosage
and AE source number for strain hardening UHPC and strain soft-
ening UHPC.

5. Conclusion

In this study, acoustic emission (AE) analysis method was

applied to monitor the damage evolution process of three types of
UHPCs under the direct tensile test, including strain-softening
UHPC, low strain-hardening UHPC and high strain-hardening
UHPC with 1.5%, 2% and 2.5% of steel fiber, respectively. Mean-
while, the crack widths of UHPCs during the tensile test were
measured by the crack width measuring instrument and camera
with micro photo lens. The following conclusions can be drawn
from this study:

(1) The ultimate tensile strain of high strain-hardening UHPC is

Fig. 12. AE sources distribution of three types of UHPCs.
single crack opening. It is hard to draw a conclusion on the rela-
tionship between fiber dosage and the AE source number at this
point.
The AE sources number of low strain-hardening UHPC is more
than that of the high strain-hardening UHPC. The reason for this
phenomenon may be the stress concentration issue for the low
strain-hardening UHPC specimen. As shown in Fig. 12, about 37% AE
higher than the yield strain of steel bar (0.2%), exhibiting
high ductility by forming multiple microcracks invisible to
naked eyes (typically below 0.05 mm) in the strain-
hardening stage, which would contribute to the ultimate
tensile loading capacity and durability in the serviceability
limit state of the R-UHPC structures. The crack width-strain
curves of different UHPCs indicates that increasing the ulti-
mate tensile strain of UHPC can improve its crack width
control ability by multiple cracking.
(2) The average elastic tensile strength fUte of high strain-
hardening UHPC, low strain-hardening UHPC and strain-
softening UHPC are 10.4 MPa, 8.7 MPa and 7.7 MPa, respec-
tively, and the corresponding strain εUte are 0.024%, 0.023%
and 0.021%, respectively. The defects of UHPCs generated in
the elastic stage are hardly detected by the AE analysis sys-
tem. Therefore, the detectable AE sources might be moti-
vated by the UHPC matrix cracking and fiber-matrix
debonding.
(3) The AE analysis method could effectively detect the internal
damages of the high strain-hardening UHPC at a strain of
600mε (the crack width less than 0.01 mm), which could not
be detected by crack width measuring instrument. This is a
strong evidence to the fact that microcracks with width
smaller than 0.01 mm were generated in the high strain-
hardening UHPC when it had a strain below 600mε.
(4) It is observed that in the same condition of elongation
(4050mε), the high strain-hardening UHPC showed a more
homogeneous distribution of AE sources than the other two
UHPCs. Equivalent elongation is balanced in the form of more
cracks, so that the width of each crack of high strain-
hardening UHPC could be controlled at a low level that is
invisible to the naked eyes. The AE analysis method provides
strong evidence to the multiple cracking behavior of strain-
hardening UHPCs.
(5) The AE analysis result indicates that even though multiple
microcraks generated in the strain-hardening stage, most of
them kept stable during the strain-softening stage, while
only the localized crack kept opening. In a word, during the
strain-softening stage, the damage evolution mechanism
that damages were generated around the localized crack is
identical for three types of UHPCs.
28 J.-Y. Wang, J.-Y. Guo / Cement and Concrete Composites 88 (2018) 17e28

Fig. 13. AE sources distribution after the peak of stress-strain curves.

Acknowledgments
This work was supported by the Fundamental Research Funds
for the Central Universities [grant number 0500219248], the Na-
tional Nature Science Foundation of China [grant number
51609172], the Shanghai Pujiang Program [grant number
16PJ1409900] and the Shanghai Municipal Science and Technology
Project [grant number 17DZ1204200]. The financial supports are
greatly appreciated.
References
[1] D.Y. Yoo, N. Banthia, Size-dependent impact resistance of ultra-high-
performance fiber-reinforced concrete beams, Construct. Build. Mater. 142
(1) (2017) 363e375.
[2] M.C.S.-E.P.F.L. Lausanne, Recommendation: Ultra-High Performance Fibre
Reinforced Cement-based composites (UHPFRC) Construction material,
dimensioning und application, Zurich, Switzerland, 2016.
[3] A.E. Naaman, H.W. Reinhardt, Proposed classification of HPFRC composites
based on their tensile response, Mater. Struct. 39 (5) (2006) 547e555.
[4] A.E. Naaman, H.W. Reinhart, Characterization of high performance fiber
reinforced cement composites (HPFRCC). In high performance fiber reinforced
cementitious composites, in: RILEM Proceedings, vol. 31, 1996, pp. 1e23.
[5] S. Wang, C. Wu, V.C. Li, Tensile strain-hardening behavior of polyvinyl alcohol
engineered cementitious composite (PVA-ECC), ACI Mater. J. 98 (6) (2001)
483e492.
[6] V.C. Li, Strategies for high performance fiber reinforced cementitious com-
posites development, in: Proceedings of International Workshop on Advances
in Fiber Reinforced Concrete, Bergamo, Italy, 2004, pp. 93e98.
[7] V.C. Li, S. Wang, C. Wu, Tensile strain-hardening behavior or polyvinyl alcohol
engineered cementitious composite (PVA-ECC), ACI Mater. J. 98 (2001)
483e492.
[8] V.C. Li, Advances in ECC research, ACI special publication on concrete, Mater.
Sci. Appl. SP 206-223 (2002) 373e400.
[9] M.B. Weimann, V.C. Li, Drying Shrinkage and Crack Width of ECC, Brittle
Matrix Composites 7, 2003, pp. 37e46. Warsaw, Poland, October.
[10] E.B. Pereira, G. Fischer, J.A. Barros, Direct assessment of tensile stress-crack
opening behavior of Strain Hardening Cementitious Composites (SHCC),
Cem. Concr. Res. 42 (6) (2012) 834e846.
[11] Z. Zhang, H. Ma, S. Qian, Investigation on properties of ECC incorporating
crumb rubber of different sizes, J. Adv. Concr. Technol. 13 (5) (2015) 241e251.
[12] M.D. Lepech, V.C. Li, Water permeability of engineered cementitious com-
posites, Cement Concr. Compos. 31 (10) (2009) 744e753.
[13] S.H. Park, J.K. Dong, G.S. Ryu, Tensile behavior of ultra high performance
hybrid fiber reinforced concrete, Cement Concr. Compos. 34 (2) (2010)
172e184.
[14] K. Wille, J.K. Dong, A.E. Naaman, Strain-hardening UHP-FRC with low fiber
contents, Mater. Struct. 44 (3) (2011) 583e598.
[15] S. Huguet, N. Godin, R. Gaertner, L. Salmon, D. Villard, Use of acoustic emission
to identify damage modes in glass fibre reinforced polyester, Compos. Sci.
Technol. 62 (10e11) (2002) 1433e1444.
[16] A.G. Beattie, Acoustic emission, principles and instrumentation, J. Acoust.
Emiss. 2 (1983) 95e128.
[17] A.A. Pollock, Acoustic emission inspection, Metals handbook, ASM Int. 17
(1989) 278e294.
[18] J.F. Labuz, S. Cattaneo, L.H. Chen, Acoustic emission at failure in quasi-brittle
materials, Construct. Build. Mater. 15 (2001) 225e233.
[19] T.M. Morton, R.M. Harrington, J.G. Bjeletich, Acoustic emissions of fatigue
crack growth, Eng. Fract. Mech. 5 (3) (1973) 691e692. IN17-IN18, 693-697.
[20] A. Berkovits, D. Fang, Study of fatigue crack characteristics by acoustic emis-
sion, Eng. Fract. Mech. 51 (1995) 401e409.
[21] Q. Dai, K. Ng, J. Zhou, E.L. Kreiger, T.M. Ahlborn, Damage investigation of
single-edge notched beam tests with normal strength concrete and ultra high
performance concrete specimens using acoustic emission techniques,
Construct. Build. Mater. 31 (6) (2012) 231e242.
[22] Otsuka, H. Koji, Date, Fracture process zone in concrete tension specimen, Eng.
Ngfract. Mech. 65 (2) (2000) 111e131.
[23] S. Granger, A. Loukili, G. Pijaudier-Cabot, G. Chanvillard, Experimental char-
acterization of the self-healing of cracks in an ultra high performance
cementitious material: mechanical tests and acoustic emission analysis, Cem.
Concr. Res. 37 (4) (2007) 519e527.
[24] B. Schechinger, T. Vogel, Acoustic emission for monitoring a reinforced con-
crete beam subject to four-point-bending, Construct. Build. Mater. 21 (3)
(2007) 483e490.
[25] M. Ge, Comparison of least squares and absolute value methods in ae/ms
source location: a case study, Int. J. Rock. Mech. Min. 34 (3e4) (1997) 93.
[26] Y. Wang, S.X. Wu, J.K. Zhou, D.J. Shen, 3D acoustic emission source location
based on exhaustion method, Nondestruct. Test. 30 (6) (2008) 348e352.
[27] ASTM E976-99, Standard Guide for Determining the Reproducibility of
Acoustic Emission Sensor Response, ASTM International, West Conshohocken,
PA, 2000.
[28] T. Makita, E. Brühwiler, Tensile fatigue behaviour of ultra-high performance
fibre reinforced concrete (UHPFRC), Mater. Struct. 47 (2014) 475e491.
[29] T.M. Wang, Control of Cracking in Engineering Structure, China Architecture
and Building Press, Beijing, 1997.