Construction and Building Materials 174 (2018) 253–262

Contents lists available at ScienceDirect

Construction and Building Materials

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

Self-sensing properties of Engineered Cementitious Composites

Yi Huang a, Hongliang Li a,b, Shunzhi Qian a,⇑
a
School of Civil and Environmental Engineering, Nanyang Technological University, 639798, Singapore
b
Departmentof Civil Engineering, Tianjin University, Tianjin 300006, PR China

h i g h l i g h t s

High self-sensing sensitivity and superb ductility were acquired simultaneously.

High ductility of CB-ECC achieved by adding AEA.

Gauge Factor (GF) was calculated to analyze self-sensing behavior quantitatively.

Crack patterns of CB-ECC, HFA-ECC and ordinary ECC were compared.

GF shows different trend with tensile strain for different ECC crack pattern.

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

Article history: The piezoresistivity of cement-based material has already been investigated, but mainly restricted to
Received 14 November 2017 compressive stress sensing due to brittleness of concrete. Conversely, Engineered Cementitious
Received in revised form 2 April 2018 Composites (ECC) presents superb tensile ductility and pseudo strain-hardening property, which offers
Accepted 13 April 2018
unique opportunity for exploring sensing of tensile stress/strain. In this paper, Carbon Black (CB) and sup-
Available online 19 April 2018
plementary cementitious materials (SCM) were incorporated into ECC to decrease the bulk resistivity and
simultaneously acquire high tensile ductility. The resistance of ECC, high fly ash (HFA)-ECC and CB-ECC
Keywords:
was measured through a two-probe method under uniaxial tension test. All specimens exhibited increase
Carbon Black
Engineered Cementitious Composites
of resistivity once cracks occurred between two electrodes. The fractional Gauge Factor (GF) in strain-
Electric resistivity hardening segment was calculated and the relationship of GF and tensile strain was experimentally
Crack pattern investigated, which could be potentially utilized in the field of structural health monitoring to enhance
Tensile ductility the safety of concrete infrastructures.
Structural health monitoring Ó 2018 Elsevier Ltd. All rights reserved.

1. Introduction Recent researches have focused on searching for materials of

Electrical resistivity of cement based materials, has been exten-
sively studied and applied in investigating hydration of cement
paste and transport properties of concrete. The change in resistiv-
ity of concrete in response to external loading, i.e. piezoresistivity
can also be used to sense the strain/stress and damage, which is of
great significance for engineers to monitor and maintain structures
timely and thereby extend the service life. McCarter et al. [1–3]
first analyzed the microstructure and hydration of cementitious
materials by studying the electrical characteristics in the 1980s.
It was realized that the electrical resistivity of concrete is directly
related to the rate of hydration of the cement. Chung et al.
increased the conductivity of concrete by adding steel fiber and
carbon fiber to make the concrete a sensor to detect strain [4].
⇑ Corresponding author.
E-mail address: szqian@ntu.edu.sg (S. Qian).
higher piezoresistive coefficient to achieve sensors with better sen-
sibility, higher accuracy, and wider range mainly by adding con-
ductive additives. Han and Ou et al. found that cementitious
composites filled with nano graphite platelets (NGPs) possess sen-
sitive piezoresistive effect and stable repeatability under different
loading conditions [5,6]. Nevertheless, the research related to
self-sensing concrete is mainly restricted to compressive strain
and stress, which may be attributed to the brittleness of concrete
that makes it not feasible to study self-sensing properties under
flexural and tensile loading.
Engineered Cementitious Composites (ECC) exhibited superb
ductility and deformation capability after incorporating a small
volume percent of Polyvinyl Alcohol (PVA) fiber. The tensile ductil-
ity is obtained by producing multiple micro cracks under loading,
and the crack width is controlled by PVA fiber bridging. Hence, it
is valuable to investigate the self-sensing properties of ECC due
to its excellent mechanical performance especially in tension and

https://doi.org/10.1016/j.conbuildmat.2018.04.129
0950-0618/Ó 2018 Elsevier Ltd. All rights reserved.
254 Y. Huang et al. / Construction and Building Materials 174 (2018) 253–262
flexure. ECC, as a sensor and simultaneously a structural material,
has many advantages over embedded and attached strain/stress
sensors, for instance, (1) it could detect a wider strain range,
including elastic, strain-hardening and tension-softening stage,
(2) it does not require installation of external sensors, which could
be quite costly and difficult to maintain, (3) it is possible to mon-
itor the entire structure, not just its specific parts over the whole
service life [7].
The mechanism in electric conduction of concrete was investi-
gated and it was discovered that the movement of positive and
negative ions, like K+, Na+, Ca+ and OH, in continuously connected
micro-pores contributes to electrical conduction of concrete [8,9].
The bulk resistivity of concrete is influenced by many factors,
including type of cement, the ratio of water to binder, degree of
hydration, porosity and moisture content. Previous study demon-
strated that the resistivity of wet concrete ranged from 103 to
104 ohm-cm, while dry concrete could increase up to 108–109
ohm-cm[10], as the resistivity of hydration product is a few orders
of magnitude higher than that of the pore solution[9]. Huge varia-
tion in resistivity makes it not feasible to utilize electrical proper-
ties of cement-based material in structural health monitoring.
Some highly conductive materials, such as steel fiber, graphite,
and carbon nanotube, are added into cement-based material to
enhance its conductive properties and simultaneously reduce its
dependence on ions inside pore solution. Carbon Black (CB), charac-
terized by desirable electrical conductivity and readily dispersible
ability in cement paste, was used to improve electrical conductivity
of cement-based composites. CB particles incorporated into the
matrix form a spatial network and the resistivity of composites
expresses an exponential function of potential barrier width (the
distance between two adjacent CB particles) based on tunneling
theory [11,12]. Both volume fraction of CB and its dispersion degree
determine the potential barrier width, which in turn influences
resistivity of the composites. It was found that the CB distributed
in the polymer in an aggregate form because of the agglomeration
of CB [13–15]. However, there is no research concerning the disper-
sion and agglomeration of CB in cement paste.
Based on previous research four-probe method with above 1
kHz frequency alternating current (AC) supply was recommended
to minimize the error produced by material polarization and con-
tact impedance [16]. Modified two-probe method under direct cur-
rent was used in this paper due to simplicity and availability of the
instruments. Drying induced resistivity was neglected, since rela-
tive humidity in lab is nearly stable. The gauge factor [16–18] (frac-
tional change in resistivity normalized by corresponding change in
strain) was proposed to evaluate the sensitivity of resistivity to
tensile strain and crack patterns. The self-sensing properties of
CB-ECC was initially investigated by Li et al. [17], and it was found
that increasing CB content reduced the resistivity of composites as
expected. However, the tensile strain capacity also decreased due
to poor fiber dispersion caused by addition of CB, which has very
high surface area. Ranade et al. [18] investigated experimentally
and analytically the influence of the microcracking on the compos-
ite resistivity using normal ECC and high fly ash ECC. It was found
that normal ECC with larger crack width resulted in higher gauge
factors, despite of its lesser number of cracks at a given strain when
Table 1
ECC mixture proportion.
Mix Cement Fly Ash Silica Sand Water
ECC 1.0 1.2 0.36 0.32
HFA-ECC 1.0 3.0 0.36 0.32
CB-ECC 1.0 1.2 0.36 0.38
compared with that of HFA-ECC. Nevertheless, they did not include
any conductive additive to lower down the bulk resistivity and the
crack width ranges of these two mixtures are relatively close. In
this paper, the mix design of ECC is manipulated by reducing
amount of silica sand, adding more fly ash, incorporation of Carbon
Black and other supplementary materials for instance air entrain-
ing agent to achieve ECCs with a wide range of bulk resistivity
and crack width while still maintaining decent tensile strain capac-
ity. The influence of tensile strain and bulk resistivity on the rela-
tive change of resistivity is studied experimentally. Furthermore,
the influence of crack width and crack number on gauge factor is
also discussed.
2. Experimental investigation
2.1. Material and mix proportions
The mix proportions are presented in Table 1, where water, silica sand and CB
are the ratio to binder (cement and fly ash), while fly ash is the ratio to cement by
weight. Carbon BlackVXC72R from Cabot Corporation was chosen to develop CB-
ECC in this paper. More water and superplasticizer were added for admixture with
CB in order to acquire good workability as CB has huge nitrogen-specific surface of
254 m2/kg [19]. Air Entraining Agent (AEA, Megachem Limited) with main compo-
nent of Sodium Lauryl Sulfate, was added into the mixture to form air bubbles. Type
I Portland Ordinary Cement was partially replaced by fly ash (Class F, Jaycee
Resources Private Limited) to develop high tensile ductility of ECC. The size of silica
sand used in this paper ranged from 160 to 180 mesh (82–93 mm). PVA fiber sup-
plied by Kuraray Company was added, whose length and oil coating by weight
are 8 mm and 1.2% respectively.
2.2. Experimental procedures
All dry ingredients including CB and AEA were mixed in a Hobart mixer with
20L capacity for 1 min at low speed, and then water and superplasticizer were
added and mixed until homogeneous slurry was formed. Finally, PVA fiber was
added into the mixer and then changed to intermediate speed mixing for another
5 min to ensure good dispersion of PVA fiber. CB-ECC required higher w/b, super-
plasticizer content and more mixing time to make sure there are no fiber bundles.
The fresh mixture was cast into oiled moulds, and then covered with plastic sheet.
The specimens were demolded after 24 h. These specimens were cured in a sealed
plastic container for 6 days and then cured in lab room environment until testing at
the age of 28 days.
Copper tape (3M Company), serving as electrodes, was tailored in 4 mm width
and 200 mm length, submerged into butanone solution to dissolve the backing
adhesive, and then washed by tap water. Dry and clean copper tape was glued with
silver conductive epoxy around the surface of dogbone specimens in the position
shown in Fig. 1.
All dogbone-shaped specimens prepared for uniaxial tension test were divided
into three sections by electrodes and LVDT (Fig. 1). Section I was between the inter-
nal boundary of two electrodes and its span was 108 mm, where resistance would
be measured. Section II was between outer boundary of electrodes and the LVDTs
(linear variable differential transformer), and the length of Section II was 11 mm
each. Cross-sectional area in Sections I and II remained unchanged, measuring 36
mm in width and 16 mm in thickness. Section III was outside the gauge length,
where cross-sectional area was non-uniform.
The resistance was measured by two-probe method per second using multime-
ter (GWINSTEK GDM-8261) with its range up to 100Mohm. The bulk resistivity
could be calculated based on Ohm’s law, shown in Eq. (1).
RS
q¼ ð1Þ
L
where q and R are the bulk resistivity and resistance measured by multimeter
respectively, and S and L represent area of cross section and the distance between
electrodes respectively.
Super Plasticizer (g/L) AEA Carbon Black PVA fiber
(g/L) (vol%)
3 0.1 0 2
3 0.1 0 2
6 0.1 0.01 2
 Y. Huang et al. / Construction and Building Materials 174 (2018) 253–262
Fig. 1. Dog bone-shaped specimen pasted with copper tape electrodes.
Four to five dogbone specimens were prepared for each mixture to conduct uni-
axial tension test using the Instron 5569 after polarization was stabilized. The two
ends of dogbone specimens were wedged into fixture, and the loading rate was 0.2
mm/min under displacement control. Two external LVDTs were fixed on both sides
to measure the tensile strain of gauge length (Fig. 2). The test stopped after the soft-
ening point appeared. The number of cracks occurred in Section I, II, III was counted
separately and the residual crack width except for final localized crack was mea-
sured by optical microscope with a magnification of 30 times. Typical crack width
measurements under optical microscope are shown in Fig. 3.
3. Results and discussion
3.1. Material polarization
The resistance of all dogbone-shaped specimens of ECC, HFA-
ECC and CB-ECC after 28 days curing was measured by two-
probe method for the first 1200 s. Table 2 illustrated the resistivity
evolution of each mixture from zero to 1200 s, whereby the mean
Fig. 2. Uniaxial tension test setup.
255
bulk resistivity and corresponding standard deviation of ECC, HFA-
ECC and CB-ECC are summarized. It can be observed from Table 2
that the resistivity of all mixtures rose steeply, especially for the
first 100 s, known as material polarization. At 1200 s, the resistivity
of all mixtures stabilized since material polarization effect gradu-
ally diminished.
The bulk resistivity of ECC, HFA-ECC rose to 0.924 and 2.087
Mohm-cm respectively at 1200 s, while the resistivity of CB-ECC
was 0.571 Mohm-cm. High percentage of fly ash would likely block
the continuous pores inside the matrix, that is, the current path
allowing for the movement of ions, eventually resulting in higher
resistivity. In addition, the standard deviation of resistivity of
ECC and HFA-ECC maintained around 0.4 and 0.5 Mohm-cm after
1000 s, approximately 4 and 5 times of CB-ECC. In general, the
addition of CB contributed to the reduced bulk resistivity of ECC
and lower standard deviation, as the network of CB inside the
matrix played a significant role in electrical conduction. Besides
lowered bulk resistivity, this is yet another advantage by introduc-
ing CB into ECC, which suggests that CB-ECC may be more reliable
as self-sensing material due to small variation in resistivity. Never-
theless lower resistivity may accelerate the electrochemical pro-
cess of chloride-induced or carbonation-induced corrosion which
need to be further studied.
The relative increase percentage from 1100 to 1200 s for ECC,
HFA-ECC and CB-ECC slowed down to 1.54%, 2.25% and 1.06%. It
is desirable to obtain more or less constant resistivity before the
specimens were loaded. Consequently, change in resistivity will
correspond well with tensile stress/strain instead of material
polarization. Hence, tension test was performed after 1200 s when
the resistivity was stabilized.
3.2. Tensile properties of three ECC mixtures
All specimens in Fig. 4 exhibited pseudo strain-hardening and
multiple cracking properties under uniaxial tension test. The ten-
sile properties were summarized in Table 3. Typical crack pattern
of three mixtures were shown in Fig. 5.
Gauge length contained Sections I and II of specimens, never-
theless, only cracks appeared in Section I were shown in Table 3
for the purpose of investigating how crack patterns (crack width
and crack number) influenced resistivity. The crack width distribu-
tions (Fig. 6) could be obtained by measuring the crack width in
Section I along the dashed centerline (Fig. 5).
The average tensile strain capacity for ECC was 2.61%, while
HFA-ECC and CB-ECC arrived at 3.84% and 3.44% respectively.
The respective ultimate tensile strength achieved 3.47 MPa, 4.52
MPa and 5.08 MPa. Moreover, saturation of micro cracks was
256 Y. Huang et al. / Construction and Building Materials 174 (2018) 253–262

Fig. 3. Typical crack width measurements of (a) ECC, (b) HFA-ECC and (c) CB-ECC.

Table 2
First 1200-s mean resistivity for ECC, HFA-ECC and CB-ECC and corresponding standard deviation.

Time Mean resistivity (106 ohm-cm) Standard deviation of resistivity (106 ohm-cm)
ECC HFA-ECC CB-ECC ECC HFA-ECC CB-ECC
1 0.280 0.656 0.138 0.133 0.113 0.072
100 0.636 1.441 0.400 0.378 0.270 0.099
200 0.701 1.546 0.451 0.381 0.282 0.100
300 0.743 1.620 0.479 0.381 0.298 0.100
400 0.775 1.683 0.499 0.381 0.317 0.099
500 0.802 1.740 0.514 0.383 0.336 0.099
600 0.824 1.794 0.526 0.385 0.357 0.099
700 0.844 1.847 0.536 0.387 0.381 0.098
800 0.863 1.897 0.545 0.391 0.404 0.097
900 0.880 1.947 0.553 0.394 0.429 0.097
1000 0.895 1.994 0.559 0.397 0.453 0.096
1100 0.910 2.041 0.565 0.400 0.477 0.097
1200 0.924 2.087 0.571 0.404 0.500 0.096

observed, especially for HFA-ECC and CB-ECC, and their respective
crack spacing was 1.08 mm and 1.48 mm. Hence it was more
microcracks rather than wider crack width that contributed to
higher tensile ductility of these two materials.
The mean residual crack width for ECC was 65 mm, much wider
than HFA-ECC and CB-ECC whose average residual crack width
were 7 and 15 mm respectively. Single PVA fiber was taken out of
the fresh matrix after mixing and was rinsed thoroughly in running
water. Then PVA fibers from ECC, and CB-ECC (1% CB) and another
CB-ECC (0.5%CB) were observed by optical microscope at a magni-
fication of 10 (Fig. 7). The PVA fiber from ECC was nearly transpar-
ent, while fibers from 0.5%CB-ECC and 1.0%CB-ECC were
significantly darkened, and the latter was darker presumably due
to more CB particle absorption on the PVA fiber surface. Narrowed
 Y. Huang et al. / Construction and Building Materials 174 (2018) 253–262 257

7
ECC HFA-ECC 1.0%CB-ECC

5
Tensile stress/MPa

0
0 1 2 3 4 5
Tensile strain/%

Fig. 4. Tensile stress-strain curves for ECC, HFA-ECC and CB-ECC.

Table 3
Tensile properties of ECC, HFA-ECC and CB-ECC.

Mix No First cracking Elastic strain (%) Ultimate tensile Tensile Location of final Crack number* Mean residual
strength (MPa) strain (%) strength (MPa) failure crack width (microns)*
ECC 1 2.67 0.031 2.44 3.48 Sec I 18 60
2 3.08 0.035 2.27 3.32 Sec I 16 92
3 2.98 0.036 3.07 3.64 Sec I 21 50
4 2.84 0.030 2.66 3.45 Sec II 21 56
Mean 2.89 0.033 2.61 3.47 — 19 65
HFA-ECC 1 3.58 0.027 3.99 4.50 Sec I 94 9
2 3.58 0.031 4.00 4.59 Sec III 90 6
3 3.03 0.026 3.92 4.72 Sec III 107 6
4 2.98 0.025 4.14 4.70 Sec III 109 7
5 3.36 0.025 3.15 4.11 Sec II 99 9
Mean 3.31 0.027 3.84 4.52 — 100 7
CB-ECC 1 3.64 0.020 2.87 4.83 Sec I 59 13
2 3.89 0.023 3.88 5.03 Sec III 99 14
3 2.92 0.018 4.01 5.22 Sec I 103 12
4 2.98 0.019 4.55 5.26 Sec I 76 19
5 2.96 0.024 1.87 5.06 Sec I 30 18
Mean 3.28 0.021 3.44 5.08 — 73 15
*
Cracks occurred in Sec I, excluding Sec II and Sec III.

crack width of CB-ECC when compared with ECC could probably be Fig. 8 shows the development of tensile stress and bulk resistiv-
attributed to the absorption of CB particles on the surface of PVA ity with time. The first part of the figure before about 1200 s is
fiber, which subsequently resulted in stronger frictional bond actually the same as Table 2, during which no tensile loading
strength between PVA fiber and matrix [17]. was applied and bulk resistivity gradually stabilized after initial
rapid rise. After around 1200 s, all samples, especially for ECC
and CB-ECC, exhibited a steep increase once tensile loading was
3.3. Self-sensing properties of three ECCs
applied. To investigate the influence of first cracking upon resistiv-
ity, Fig. 9 was presented to reveal change in resistivity before and
Gauge Factor (GF, defined in Eq. (2)) was normally used to eval-
after first cracking.
uate the self-sensing properties [10–12].
Tensile stress continued to rise after first cracking instead of
sudden load drop like normal concrete or fiber reinforced concrete.
Dq=qo
GF ¼ ð2Þ The first cracking point and the corresponding point in resistivity-
De
time curve were highlighted in Fig. 9. There seemed little variation
where Dq is the change in resistivity during the corresponding in resistivity between stable polarization and elastic stage. It was
change in tensile strain De and q0 is the bulk resistivity when ten- easy for CB-ECC to differentiate between elastic and inelastic seg-
sile strain is zero. ment as the resistivity curve displayed an apparent turning point
258 Y. Huang et al. / Construction and Building Materials 174 (2018) 253–262
Fig. 5. Typical crack patterns of (a) ECC, (b) HFA-ECC and (c) CB-ECC under uniaxial tension test.
before and after first cracking, while the resistivity of other two
mixtures did not vary greatly at this time scale.
The relative change in resistivity (Dq/q0, the numerator of GF)
with measuring time was calculated and listed in Table 4. The
relative change of resistivity respectively in elastic and early
strain-hardening segment was displayed in Fig. 10. The latter seg-
ment referred to the period from first cracking time to 1600 s.
From elastic stage to early strain-hardening stage, the relative
change in resistivity per second (4q/q0/t) for ECC, HFA-ECC and
CB-ECC increased by 6.79, 3.38, and 12.16 (104/s) respectively
(Fig. 10). CB-ECC was around 2 times of ECC and 4 times of HFA-
ECC. It would be desirable if resistivity could grow fast once the
crack formed as ECC could be a more sensitive self-sensing unit
to detect the occurrence of crack.
Among three mixtures, CB-ECC seems to be the most sensitive
one to detect cracking so as to distinguish between elastic and
strain-hardening stage through the measurement of electrical
properties. Based on previous discussion, it was already shown that
the addition of CB could result in reduced bulk resistivity and
lower variability. Hence it could be an ideal self-sensing material
in structural health monitoring.
Fig. 11 shows an inverse relationship between bulk resistivity
(e = 0%) and relative change in resistivity per second (4q/q0/t) at
early strain-hardening stage. With lower bulk resistivity, such as
CB-ECC, it will facilitate more sensitive response to cracking, i.e.
higher relative change in resistivity per second. In this figure, we
can approximately let time be equivalent to strain (hence the
vertical axis become GF), which will be more rigorously evaluated
in the next section. Tensile stress and relative change in resistivity
(Dq/q0) was plotted in Fig. 12 with tensile strain, where the slope
of relative change in resistivity (Dq/q0)-tensile strain curve was
GF. Segmented GF values in strain-hardening stage at every 0.5%
strain were calculated and plotted in Fig. 13.
GF values at 0.5% tensile strain were mainly influenced by the
bulk resistivity (e = 0%), which was consistent with the previous
conclusion that relative change in resistivity per second (4q/q0/
t) at early strain hardening stage was inversely proportional to
bulk resistivity (e = 0%). GF values from elastic strain (ee) to 0.5%
tensile strain of ECC specimens except for ECC-2 varied from 24.6
to 33.5, and ECC-2 was as high as 97.1 due to much lower bulk
resistivity at zero strain in this specimen. Bulk resistivity (e = 0%)
of all HFA-ECC specimens was the highest among these three mix-
tures, therefore their corresponding GF (ee < e < 0.5%) ranged from
16.7 to 24.8, much lower than the other two groups. GF values (ee
< e < 0.5%) from four out of five CB-ECC specimens were over 70,
with one outlier as low as 33.7 due to much higher resistivity at
zero strain in this sample.
GF of ECC showed a rising trend with tensile strain, although
one of them was much higher than the rest. GF grew around 2–4
times from the first to the last interval within inelastic stage. GF
values of HFA-ECC maintained almost constant at around 25 prior
to 3.5% tensile strain, and went up to approximately 50 at 4% ten-
sile strain. The GF value for CB-ECC showed significant fluctuation,
without revealing a clear trend as in the case of ECC.
 Y. Huang et al. / Construction and Building Materials 174 (2018) 253–262 259

6 4 5.0E+6

5 ECC
fitting curve 4.0E+6

Resistivity/ohm-cm
3

Tensile stress/MPa
4
Crack number

3.0E+6
2
3
ECC-1 2.0E+6
ECC-2
2
1 ECC-3
1.0E+6
1 ECC-4

0 0.0E+0
0 0 500 1000 1500 2000 2500 3000 3500 4000
10 20 30 40 50 60 70 80 90 100 110 120 130 140 Time/sec
Crack width/microns (a)
(a)
6 8.0E+6
HFA-ECC-1 HFA-ECC-2 HFA-ECC-3
30
HFA-ECC-4 HFA-ECC-5 7.0E+6
5
25 HFA-ECC 6.0E+6

Resistivity/ohm-cm
Tensile stress/MPa
4
5.0E+6
Crack number

20
3 4.0E+6
15
3.0E+6
2
10 2.0E+6
1
5 1.0E+6
0 0.0E+0
0 0 500 1000 1500 2000 2500 3000 3500 4000 4500
3 6 9 12 15 18 21 24 27 30 Time/sec
Crack width/microns (b)
(b)
7 3.0E+6
1.0%CB-ECC-1 1.0%CB-ECC-2
25 1.0%CB-ECC-3 1.0%CB-ECC-4
6
1.0%CB-ECC-5
2.5E+6
1.0%CB-ECC
Tensile stress/MPa

Resistivity/ohm-cm
20 5
fitting curve 2.0E+6
Crack number

4
15 1.5E+6
3

10 1.0E+6
2

1 5.0E+5
5
0 0.0E+0
0 500 1000 1500 2000 2500 3000 3500 4000
0
5 10 15 20 25 30 35 40 Time/sec
(c)
Crack width/microns
(c) Fig. 8. Tensile stress and resistivity with time for (a) ECC, (b) HFA-ECC and (c) CB-
ECC.
Fig. 6. Residual crack width distributions of (a) ECC, (b) HFA-ECC and (c) CB-ECC.

In the process of tension testing, more cracks were observed for

Fig. 7. Microscopic view of PVA fibers from ECC, 0.5%CB-ECC and 1%CB-ECC.
HFA-ECC and CB-ECC, and the mean crack number for ECC was only
one fifth of HFA-ECC and one fourth of CB-ECC (Table 3). Neverthe-
less, GF of HFA-ECC exhibited the lowest due to high bulk resistiv-
ity. Residual crack width distribution for all three mixtures was
displayed in Fig. 14. The residual crack width for ECC distributed
over very wide range of around 10–120 mm, while HFA-ECC and
CB-ECC showed a relatively narrow range centered around 7 and
15 mm. By comparing Figs. 13 and 14, it may be suggested that
crack width range and crack number both had significant influence
on the trend of GF. With wide crack width range, ECC showed an
apparent upward trend, presumably due to nonlinear relationship
between increase of resistivity and tensile strain (potentially as a
result of crack width increase with tensile strain [20], therefore
crack width increase contributes more to the tensile strain), as
revealed in Figs. 8(a) and 12(a). With many more microcracks
within a narrow range, HFA-ECC revealed an almost constant GF
260 Y. Huang et al. / Construction and Building Materials 174 (2018) 253–262

4 2.5E+6 Table 4
ECC-1 ECC-2 ECC-3 ECC-4 Self-sensing properties of ECC, HFA-ECC and CB-ECC.

2.0E+6 Mix No Starting Resistivity Relative change

Resistivity/ohm-cm
Tensile stress/MPa

3 time for (e=0%) in resistivity per

tension (106 ohm-cm) second (104/s)
1.5E+6
test (s)
2 elastic strain-
stage hardening*
1.0E+6
ECC 1 1335 1.462 0.52 3.88
1 2 1210 0.459 1.11 14.96
5.0E+5 3 1227 0.967 1.19 6.69
4 1204 0.846 1.75 6.18
0 0.0E+0 Mean 1244 0.934 1.14 7.93
1200 1300 1400 1500 1600
HFA-ECC 1 1211 1.565 0.61 3.54
Time/sec 2 1340 2.734 1.53 3.90
(a) 3 1215 1.925 0.99 2.91
4 1204 1.719 0.54 4.91
5 5 1208 2.607 1.55 4.90
5.0E+6
HFA-ECC-1 HFA-ECC-2 HFA-ECC-3 Mean 1236 2.110 0.65 4.03
HFA-ECC-4 HFA-ECC-5
CB-ECC 1 1214 0.518 0.79 24.94
4 2 1262 0.565 0.47 8.11

Resistivity/ohm-cm
4.0E+6
Tensile stress/MPa

3 1208 0.509 0.45 10.18

4 1203 0.527 0.33 12.38
3
5 1209 0.739 0.24 4.76
3.0E+6 Mean 1219 0.572 0.09 12.07
2 *
From first cracking to 1600 s.

2.0E+6
1

0 1.0E+6
1200 1300 1400 1500 1600 21
Time/sec
Relative change in resistivity per second

18 elastic
(b)
strain-hardening at early stage
15
1.0E+6
1.0%CB-ECC-1 1.0%CB-ECC-2 1.0%CB-ECC-3
( ρ/ρ0/t, 10-4/s)

1.0%CB-ECC-4 1.0%CB-ECC-5 12
Resistivity/ohm-cm

9
Tensile stress/MPa

8.0E+5
6

6.0E+5 0
ECC HFA-ECC 1.0%CB-ECC
-3

4.0E+5 Fig. 10. Relative change in resistivity per second for ECC, HFA-ECC and CB-ECC in
1200 1300 1400 1500 1600 elastic stage and early strain-hardening stage.
Time/sec
(c)
Fig. 9. Tensile stress and resistivity with time for (a) ECC, (b) HFA-ECC and (c) CB-
ECC before and after first cracking.
Relative change in resistivity per second

ECC
due to constant crack width development with tensile strain (i.e.
HFA-ECC
increase of crack number contributes more to the tensile strain). CB-ECC
It was not very clear why CB-ECC showed such large variation
( ρ/ρ0/t, 10-2 %/s)

which deserved further in-depth investigation, nevertheless it still

roughly revealed a constant trend, which may be attributed to its
many microcracks within a narrow range, similar to that of HFA-
ECC. The above discussion may be confirmed in a further study
where the evolution of crack width with tensile strain will be clo-
sely monitored, which is beyond the scope of this study.

4. Conclusions
Resistivity(ε=0%, 10Mohm-cm)
This paper investigates self-sensing properties of ECC, HFA-ECC
and CB-ECC using a two-probe method under static current field. Fig. 11. Relative change in resistivity per second and bulk resistivity (e = 0%)
Some conclusions could be drawn as below. relation of ECC, HFA-ECC and CB-ECC.
 Y. Huang et al. / Construction and Building Materials 174 (2018) 253–262 261

500 250

Relative change in resistivity/%

ECC-1 ECC-2 ECC-3 ECC-4 ECC-1
400 200 ECC-2
Tensile stress/MPa

ECC-3
300 150 ECC-4

GF
200
100

100
50

0
0 0.5 1 1.5 2 2.5 3 3.5 4 0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Tensile strain/%
Tensile strain/%
(a) (a)
6 150 250
HFA-ECC-1 HFA-ECC-2
HFA-ECC-1

Relative change in resistivity/%

HFA-ECC-3 HFA-ECC-4
5 HFA-ECC-2
HFA-ECC-5 120 200
HFA-ECC-3
Tensile stress/MPa

4 HFA-ECC-4
90 150 HFA-ECC-5

GF
3
60 100
2
50
30
1

0
0 0 0.0 1.0 2.0 3.0 4.0 5.0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Tensile strain/%
Tensile strain/%
(b)
(b)
250
7 500 1.0%CB-ECC-1 1.0%CB-ECC-2 1.0%CB-ECC-3
1.0%CB-ECC-1 1.0%CB-ECC-2
Relative change in resisitivty/%

6 1.0%CB-ECC-3 1.0%CB-ECC-4 1.0%CB-ECC-4 1.0%CB-ECC-5

200
1.0%CB-ECC-5 400
Tensile strss/MPa

5
300 150
4
GF

3 200 100
2
100 50
1

0 0
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
Tensile strain/% Tensile strain/%
(c) (c)
Fig. 12. Tensile stress and relative change in resistivity with tensile strain relation Fig. 13. Fractional GF in strain-hardening segment at 0.5% strain intervals for (a)
for (a) ECC, (b) HFA-ECC and (c) CB-ECC. ECC, (b) HFA-ECC and (c) CB-ECC.

(1) Addition of CB helped lower resistivity, approximately half (3) At early inelastic stage, both GF values and relative change in
of ECC and one fourth of HFA-ECC. Furthermore, resistivity resistivity per second (4q/q0/t) were inversely proportional
of CB-ECC displayed the lowest variability among three mix- to bulk resistivity (e = 0%). CB-ECC showed the most sensi-
tures as CB provided alternative path for electrical conduc- tive response to first cracking, which potentially could be
tion of composites. utilized to detect micro cracking in structures by means of
(2) Average tensile strain capacity of ECC, HFA-ECC and CB-ECC variation in resistivity.
reached 2.61%, 3.84% and 3.44%, respectively, and their (4) With wide crack width range, ECC showed an apparent
respective ultimate tensile strength arrived at 3.47 MPa, upward trend, presumably due to nonlinear relationship
4.52 MPa and 5.08 MPa. Crack spacing of these three mix- between increase of resistivity and tensile strain (potentially
tures arrived at 5.68 mm, 1.08 mm and 1.48 mm, respec- as a result of crack width increase with tensile strain, there-
tively. In addition, mean residual crack width of HFA-ECC fore crack width increase contributes more to the tensile
and CB-ECC narrowed to 7 and 15 mm. Addition of CB and strain). With many microcracks within a narrow range,
AEA into ECC seems help enhance tensile ductility and HFA-ECC revealed an almost constant GF due to constant
tensile strength. crack width development with tensile strain (i.e. increase
262 Y. Huang et al. / Construction and Building Materials 174 (2018) 253–262

45

40 ECC

HFA-ECC
35
1.0%CB-ECC
30
Crack number

25

20

15

10

0
0 10 20 30 40 50 60 70 80 90 100 110 120
Crack width/microns

Fig. 14. Residual crack width distributions of ECC, HFA-ECC and CB-ECC.

of crack number contributes more to the tensile strain). It
was not very clear why CB-ECC showed such large variation,
nevertheless it still roughly revealed a constant trend, which
may be attributed to constant crack width develop with
strain, similar to that of HFA-ECC. The above discussion
may be confirmed in a further study where the evolution
of crack width with tensile strain will be closely monitored.
Conflict of interest
There is no conflict of interest.
Acknowledgement
The authors would like to thank for the financial support by the
Startup Grant from Nanyang Technological University, Singapore,
under Grant No. M4081208.
References
[1] W.J. Mccarter, A.B. Afshar, Some aspects of the electrical properties of cement
paste, J. Mater. Sci. Lett. 3 (12) (1984) 1083–1086.
[2] W.J. Mccarter, A.B. Afshar, A Study of the early hydration of Portland cement,
Proc. Inst. Civil Eng, (1985).
[3] M.C. Forde, W.J. Mccarter, H.W. Whittington, Resistivity characteristics of
concrete, Proc. Inst. Civil Eng. Res. Theory 73 (1) (1981) 223–224.
[4] S. Wen, D.D.L. Chung, A comparative study of steel- and carbon-fibre cement as
piezoresistive strain sensors, Adv. Cem. Res. 15 (3) (2003) 119–128.
[5] S. Sun, B. Han, S. Jiang, X. Yu, Y. Wang, H. Li, et al., Nano graphite platelets-
enabled piezoresistive cementitious composites for structural health
monitoring, Constr. Build. Mater. 136 (2017) 314–328.
[6] B. Han, X. Yu, J. Ou, Challenges of self-sensing concrete – self-sensing concrete
in smart structures (Chapter 11), in: Self-Sensing Concrete in Smart Structures,
2014, pp. 361–376.
[7] J.M. Lopez-Higuera, L.R. Cobo, A.Q. Incera, A. Cobo, Fiber optic sensors in
structural health monitoring, J. Lightwave Technol. 29 (4) (2011) 587–608.
[8] D.A. Koleva, O. Copuroglu, K. van Breugel, G. Ye, J.H.W. de Wit, Electrical
resistivity and microstructural properties of concrete materials in conditions of
current flow, Cem. Concr. Comp. 30 (8) (2008) 731–744.
[9] G.L. Song, Equivalent circuit model for AC electrochemical impedance
spectroscopy of concrete, Cem. Concr. Res. 30 (11) (2000) 1723–1730.
[10] W. Elkey, E.J. Sellevold, Electrical resistivity of concrete, Concr. Int. 37 (5)
(1995) 41–46.
[11] H. Li, H.G. Xiao, J.P. Ou, Effect of compressive strain on electrical resistivity of
carbon black-filled cement-based composites, Cem. Concr. Comp. 28 (9)
(2006) 824–828.
[12] H.G. Xiao, H. Li, J.P. Ou, Modeling of piezoresistivity of carbon black filled
cement-based composites under multi-axial strain, Sensor Actuat. A-Phys. 160
(1–2) (2010) 87–93.
[13] J. Liang, Q.Q. Yang, Aggregate structure and percolation behavior in
polymer/carbon black conductive composites, J. Appl. Phys. 102 (8) (2007).
[14] K. Dai, X.B. Xu, Z.M. Li, Electrically conductive carbon black (CB) filled in situ
microfibrillar poly(ethylene terephthalate) (PET)/polyethylene (PE) composite
with a selective CB distribution, Polymer 48 (3) (2007) 849–859.
[15] L. Adriaanse, J. Reedijk, P. Teunissen, H. Brom, M. Michels, J. Brokken-Zijp,
High-Dilution Carbon-Black/polymer composites: hierarchical percolating
network derived from Hz to THz ac conductivity, Phys. Rev. Lett. 78 (9)
(1997) 1755–1758.
[16] T.C. Hou, Wireless and electromechanical approaches for strain sensing and
crack detection in fiber reinforced cementitious materials (PhD Dissertations),
University of Mechanical, 2008.
[17] M. Li, J. Lin, J.P. Lynch, V.C. Li, Carbon black engineered cementitious
composites-mechanical and electrical characterization, ACI Spec. Pub. 292
(2013) 1–16.
[18] R. Ranade, J. Zhang, J.P. Lynch, V.C. Li, Influence of micro-cracking on the
composite resistivity of Engineered Cementitious Composites, Cem. Concr. Res.
58 (2014) 1–12.
[19] C.K. Leong, D.D.L. Chung, Pressure electrical contact improved by carbon black
paste, J. Electron. Mater. 33 (3) (2004) 203–206.
[20] E.B. Pereira, G. Fischer, J.A.O. Barros, Direct assessment of tensile stress-crack
opening behavior of Strain Hardening Cementitious Composites (SHCC), Cem.
Concr. Res. 42 (6) (2012) 834–846.