sensors

Article
Experimental Study on Damage Detection in
ECC-Concrete Composite Beams Using
Piezoelectric Transducers
Fengjiang Qin , Zhigang Zhang , Bo Xie and Rui Sun *
Key Laboratory of New Technology for Construction of Cities in Mountain Area, Ministry of Education,
School of Civil Engineering, Chongqing University, Chongqing 400045, China;
qinfengjiang@cqu.edu.cn (F.Q.); zhangzg@cqu.edu.cn (Z.Z.); m13678426488@163.com (B.X.)
* Correspondence: ruisun@cqu.edu.cn




Received: 29 May 2019; Accepted: 17 June 2019; Published: 22 June 2019


Abstract: The use of engineered cementitious composite (ECC) has attracted extensive attention
in recent years because of the highly enhanced ductility owing to its unique strain-hardening
behavior. In this paper, an electromechanical impedance-based technique is used to monitor the
structural damage of RC beams strengthened with an ECC layer at the tensile zone. To achieve
this purpose, three specimens are tested under bending loads to evaluate the proposed damage
detection methodology. Five externally bonded PZT transducers are uniformly distributed at the
surface of the ECC layer of the beams to measure the output conductance signatures in a healthy
state and in different damage scenarios induced by different load levels. Test results showed that
discrepancies exist between the signals measured in the intact state and each damage state, which can
be used to evaluate the structural integrity changes. To assess the damage of ECC-concrete composite
beams quantitatively, the statistical scalar index-root mean square deviation (RMSD) is used as the
index, which can be calculated from the variations of conductance measurements of PZT sensors.
The damage index values of the uniformly distributed PZT sensors provided cogent evidence of
damage and revealed the evolution of structural damage. The crack patterns of beams at different
damage levels are compared with the damage index values, and it shows the damage location can be
derived from the measured conductance signatures of an array of PZT transducers.

Keywords: engineered cementitious composite (ECC); flexural behavior; piezoelectric transducer;

damage detection

1. Introduction
Concrete is widely used as a construction material because of its high strength and durability
as well as the easy availability worldwide. Great attention has been paid in past decades to enhance
the mechanical performance of concrete due to its inherent weaknesses, such as low tensile strength,
brittle nature, and poor crack-resistance. In recent years, randomly-distributed short fibers with
specific properties (usually polyvinyl alcohol (PVA) or polyethylene (PE)) have been added to the
cement-based matrix to develop a new kind of high performance of polymer fiber reinforced concrete
(called engineered cementitious composite (ECC)) [1–5]. Although only a low fiber volume (typically
lower than 2%) is employed, the tensile strain capacity of material can be improved significantly due
to the interactions between the composite fibers and the cement matrix. The tensile strain of ECC
normally ranges between 3~8%, which is hundreds of times that of conventional concrete [6,7]. Unlike
the brittle nature of concrete, the fiber bridging effect of ECC within cracks allows the crack width to
be controlled below 100 µm, and thus prevents the appearance of millimeter-sized localized fracture
cracks which are usually observed in normal concrete [8,9].

Sensors 2019, 19, 2799; doi:10.3390/s19122799 www.mdpi.com/journal/sensors

Sensors 2019, 19, 2799 2 of 23

Numerous studies have been conducted to assess the mechanical behavior of steel reinforced
ECC members (R-ECC). Superior ductility is observed because of the interfacial interactions between
randomly-distributed fibers and cement matrix, which allow ECC to bear tensile, flexural and shear
loads [10–12]. Hence, the brittle failure modes such as shear failure and bond-splitting could be
well controlled by using ECC materials. The small-spacing micro-cracks of ECC enable it to resist
the aggressive environment and especially the corrosion of steel reinforcement [13–15]. Besides,
the unreacted cementitious particles at the location of cracks are likely to hydrate continually under
moisture environments and thus produce new hydration products to fill the micro-cracks, that is
a self-healing characteristic which can be used for rehabilitation of damaged structures [16–18].
Moreover, performance assessments of ECC members and structures subjected to monotonic, cyclic
and seismic motions, can be performed experimentally [19–21], which demonstrates that the stiffness,
strength, ductility and energy absorption of ECC structural members have a significant improvement
than that of conventional RC structures. In a word, due to its excellent mechanical properties, ECC is
a promising material to be applied in construction engineering [22–24] in order to enhance the structural
performance against cracks, impact, fatigue loading and earthquake motions.
With the application of ECC materials in construction engineering, the damage detection of
ECC structures has become an important issue to ensure their structural safety during their service
life. By monitoring structural damages in real-time, the operational conditions of structures can
be evaluated to prevent structural failures and reduce the risk of catastrophic accidents. In recent
years, an impedance-based structural health monitoring (SHM) technique by using piezoelectric
ceramic (PZT) sensors has attracted extensive attention due to its superior actuating/sensing properties,
high sensitivity, economical cost and free model [25–27]. The PZT transducer, that serves as both
actuator and sensor simultaneously, is usually internally embedded or bonded on the surface of a host
structure. It has been theoretically proven that the electrical impedance of PZT sensor can be expressed
in terms of the mechanical impedance of the host structure, which is directly related to the changes of
structural integrity [28], thus, indicators extracted from impedance signatures can be used to assess
structural damages. By applying PZT sensors, numerous experimental studies have been conducted to
monitor damages of pipe systems [29], bolt joints [30], RC structures [31], bridges [32] and composite
structures [33,34]. Numerical simulations are also carried out to study the sensitivity and accuracy of
the impedance analysis and to identify structural damages according to the variation of impedance
spectra [35–37].
Practical applications of the impedance-based methodology have demonstrated the feasibility
of PZT sensor for SHM. However, to the authors’ best knowledge, few works were concentrated on
the damage assessment of RC structures with ECC materials. Therefore, this study is focused on the
damage evaluation of ECC-concrete composite beams subjected to bending loads. To this end, three RC
beams were prepared for four-point bending load tests, and PZT transducers were externally bonded
at the surface of the ECC layer using an epoxy adhesive. The electrical conductance spectra of each PZT
sensor is measured under different structure damage scenarios. Thereafter, a statistical damage index
RMSD, which is derived from the variations of measured signatures, is deployed to quantitatively
assess the appearance and progression of structural damage as the applied load increases.

2. PZT Sensing Methodology for SHM

PZT sensors are made of piezoelectric material which is able to mutually transfer mechanical
vibration and alternating current due to the direct and inverse piezoelectric effect. Actuating and
sensing of PZT sensors would be simultaneously activated as long as an alternating voltage field is
applied across the poling direction on the sensor. When a PZT transducer is internally embedded or
externally bonded with a host structure, the mechanical vibration would interact with the PZT sensor
and the self-sensing property makes it able to measure the output electrical impedance in a particular
frequency range using an impedance analyzer (or LCR meter). The measured impedance signals are
coupled to both the electrical impedance of the PZT sensor and the mechanical impedance of the host
Sensors 2019, 19, 2799 3 of 23

structure, in which the second item is directly related to structural integrity. Therefore, any damages in
structures can be evaluated according to the variation of PZT output signatures.
Considering the constitutive equation and polarization direction along the thickness of a PZT
patch, Liang [28] developed the formulation of a one-dimensional model to represent the dynamic
interaction between PZT transducer and structure using a simplified single-degree of freedom system.
The electrical admittance signatures Y(ω) can be expressed as:

bPZT lPZT T Zs (ω) E

Y(ω) = G(ω) + jB(ω) = jω (ε33 − d2 Y ) (1)
ePZT Zs (ω) + Za (ω) 31 11

where G(ω) (real part) and B(ω) (imaginary part) represent the electrical conductance and susceptance,
E
respectively. ω is the angular frequency of the excitation voltage. Y11 denotes the complex Young’s
modulus in the length direction of a PZT patch with width, length and thickness designated by bPZT ,
lPZT and ePZT , respectively. εT33 and d31 are the electric permittivity in thickness direction and coupling
piezoelectric constant which are defined by the constitutive relationship of piezoelectric materials.
Za (ω) and Zs (ω) are the mechanical impedance of the PZT sensor and host structure.
The electrical admittance of PZT transducer Y(ω), which is composed of electrical conductance
G(ω) and susceptance B(ω), is in function of the mechanical impedance of the host structure as shown in
Equation (1). Hence, any variations of admittance spectra directly reflected in the structural properties,
such as stiffness, mass and damping, can be employed for damage detection. It is noted that the
electrical conductance G(ω) (real part of admittance) can reflect the changes of structural integrity,
while the susceptance B(ω) is more sensitive to the damage of PZT sensor itself. Therefore, in this work,
electrical conductance signature is used for structural health monitoring due to its better indication of
any structural property changes.
The selection of the frequency range would influence the effectiveness of the proposed
electromechanical admittance (EMA) methodology and the damage detection ability greatly depends
on the successful frequency range selected rather than the voltage of excitation. Usually, a hundreds of
kilohertz of frequency range is adopted due to the fact that the stress wavelength should be smaller than
the size of the damage present in the host structure, which is one of the major concerns. From various
studies of which the selected frequency ranges have been up to 1000 kHz [27], it is found that a lower
frequency range results in a larger sensing area and a higher frequency range results in a smaller
sensing area. However, it is also noticed that a higher frequency range is more sensitive to detect
incipient and minor damages at an early stage. Hence, the frequency range should be defined by
finding a balance between sensing area and ability of damage detection according to the requirements
in each real case. Furthermore, the EMA signature with a frequency range higher than 600 kHz
contains more information related to the physical properties of the PZT patch and the interfacial
bond layer rather than host structure itself [38–40]. Therefore, in this work the conductance signature
of a frequency band between 0–600 kHz is adopted for subsequent analysis and structural health
monitoring of ECC-concrete composite beams.

3. Experimental Program

3.1. Material Properties

The main ingredients of ECC used in this work consisted of Portland cement (CEMI-52.5), fly ash,
silica sand, PVA fibers, polycarboxylate-based high-range water reducing admixture (HRWRA),
and water. The average size of silica sand is 150 µm with the maximum size of 200 µm. A total volume
of 2% polyethylene (PE) fibers, that has a length of 12 mm and diameter of 26 µm, were added in the
mortar paste for mixing until disperse well. The mixture proportions of ECC materials are listed in
Table 1.
Sensors 2019, 19, 2799 4 of 23

Sensors 2019, 19, x 4 of 23

Table 1. Mixture Proportion of ECC Mixture (by weight).
Table 1. Mixture Proportion of ECC Mixture (by weight).
Cement Fly Ash Silica Sand Water HRWRA PE Fiber
Cement Fly Ash Silica Sand Water HRWRA PE Fiber
1 1 0.6 0.5 0.004 2% by volume
1 1 0.6 0.5 0.004 2% by volume

To investigate the
To investigate the tensile
tensilestress-strain
stress-strainrelationship
relationship of the
of the ECCECC material
material in thisin study,
this study,
three
three dogbone-shaped specimens were prepared and linear variable displacement
dogbone-shaped specimens were prepared and linear variable displacement transducers (LVDTs) transducers (LVDTs)
were
were used
used toto test
test the
the elongation
elongation of of specimens
specimens within
within thethe gauge
gauge length.
length. The
The detailed
detailed dimensions
dimensions of of
the
the specimens and the test set-up can be found in [41,42]. The loading rate of the tensile test
specimens and the test set-up can be found in [41,42]. The loading rate of the tensile test was
was
controlled
controlled asas 0.5
0.5 mm
mm perper minute,
minute, and
and the
the tensile
tensile strain
strain was
was calculated
calculated according
according to to the
the average
average value
value
of
of the LVDTs. The measured stress-stain curves of ECC from uniaxial tensile tests are presented in
the LVDTs. The measured stress-stain curves of ECC from uniaxial tensile tests are presented in
Figure
Figure 1. As shown,
1. As shown, all all tensile
tensile stress-strain curves of
stress-strain curves of ECC
ECC usedused inin this
this study
study exhibit
exhibit aa very
very evident
evident
strain-hardening
strain-hardening phenomenon
phenomenon and and high
high tensile
tensile strain
strain capacity.
capacity. TheThe average
average tensile
tensile strain
strain ofof ECC
ECC
reaches
reaches 7.27% with variance of 0.38%, which is nearly 700 times that of conventional concrete. On the
7.27% with variance of 0.38%, which is nearly 700 times that of conventional concrete. On the
other
other hand,
hand, due
due toto the
the fiber
fiber bridging
bridging effect,
effect, the
the ultimate
ultimate tensile
tensile strength
strength of of ECC
ECC is is 9.92
9.92 MPa
MPa with
with
variance
variance ofof 0.31
0.31 MPa. This value
MPa. This value isis rather
rather high
high as
as the
the tensile
tensile strength
strength ofof concrete
concrete with
with compressive
compressive
strength
strength of C80 grade is normally below 3 MPa. It is noted that the average compressive strength
of C80 grade is normally below 3 MPa. It is noted that the average compressive strength of
of
ECC used in this study is
ECC used in this study is 79 MPa.79 MPa.

12

10
Tensile stress (MPa)

0
0 2 4 6 8 10 12
Tensile strain (%)

Figure 1. Measured tensile stress-stain curves of ECC.

3.2. The Test

3.2. The Test Specimens
Specimens
AA total
total of
of three
three beams
beams were
were tested
tested inin this
this work. Two of
work. Two of them,
them, named
named BE1
BE1 and
and BE2,
BE2, respectively,
respectively,
were strengthened with a 6 cm and 9 cm ECC layer at the bottom. One
were strengthened with a 6 cm and 9 cm ECC layer at the bottom. One RC beam without RC beam without an ECCan layer
ECC
was tested as the reference beam. A timber mold with the dimensions of 2000
layer was tested as the reference beam. A timber mold with the dimensions of 2000 mm × 300 mm mm × 300 mm × 200×
mm was was
200 mm established to prepare
established the test
to prepare thespecimens. All beams
test specimens. were cast
All beams wereto cast
havetothe same
have thecross
same section
cross
size of 300 × 200 mm, differing only in the thicknesses of the different ECC layers that
section size of 300 × 200 mm, differing only in the thicknesses of the different ECC layers that were were 6 cm and
96 cm, respectively, as shown in Figure 2. The reinforcement ratio of the three specimens
cm and 9 cm, respectively, as shown in Figure 2. The reinforcement ratio of the three specimens was 1.18%,
in
waswhich four
1.18%, inlongitudinal
which four bars of 14 mmbars
longitudinal nominal
of 14 diameter
mm nominal were diameter
distributed in the
were tension area
distributed as
in the
steel reinforcement.
tension area as steel reinforcement.
Sensors 2019, 19, 2799 5 of 23
Sensors 2019, 19, x 5 of 23

2000
50 200 5×100 500 5×100 200 50

φ8 φ8

φ8 φ8
300

φ14

150 1700 150

200 200 200

210
240
300

90
60

(a) (b) (c)

Figure
Figure 2.
2. Longitudinal
Longitudinal and
and cross
cross section
section of
of the
the test
test beams:
beams: (a)
(a) RC
RC beam,
beam, (b)
(b) BE1,
BE1, (c)
(c) BE2
BE2 (unit:
(unit: mm).
mm).

3.3. Test
3.3. Test Description
Description
The schematic
The schematic test test setup
setup is is shown
shown in in Figure
Figure 3. A servo-hydraulic
3. A servo-hydraulic testing testing machine
machine was was used
used toto
perform four-point bending tests through a spreader steel beam. The
perform four-point bending tests through a spreader steel beam. The distance of the two loading distance of the two loading
points was
points was 500500mm, mm,andandthe length
the lengthbetween
between loading point
loading and and
point end support
end support was 600 was mm.
600Three
mm. LVDTs
Three
were placed
LVDTs wereatplaced
the middle
at theof middle
beam toof measure
beam to themeasure
deflection theof deflection
the midspan of and
the two loading
midspan and points.
two
Two LVDTs were installed to monitor the settlement of end supports
loading points. Two LVDTs were installed to monitor the settlement of end supports which werewhich were placed at a distance
of 150 mm
placed from theof
at a distance ends
150 ofmmthefrom
beam.the Five
endsPZT sensors
of the beam.withFive dimensions
PZT sensorsofwith 10 mm × 10 mm
dimensions of×101
mm were externally bonded on the bottom surface of each beam
mm × 10 mm × 1 mm were externally bonded on the bottom surface of each beam by using an by using an epoxy resin adhesive.
Two strain
epoxy resingauges,
adhesive.E1 and
TwoE2, weregauges,
strain placed E1 at the
andmiddle
E2, weresection
placedof the testmiddle
at the beams. section of the test
beams.It is noted that the PZT sensors bonded at the bottom of RC beams are easily broken with the
expansion of the
It is noted crack
that thewidth nearby.bonded
PZT sensors However, the bottom
at the micro-cracks
of RC of ECC are
beams reinforced concrete
easily broken beams
with the
with low crack
expansion of the width
crackwould
widthsteadily
nearby.progress
However, in the
a small spacing ofof3–6
micro-cracks ECC mm due to theconcrete
reinforced excellentbeams
crack
control
with low ability
crackofwidth
ECC materials,
would steadilywhich progress
makes it able
in a to maintain
small spacingthe of
integrity
3–6 mm of due
PZT tosensors bonded
the excellent
on ECC layers.
crack control ability of ECC materials, which makes it able to maintain the integrity of PZT sensors
bondedFigure
on ECC4 displays
layers.the typical distribution of micro-cracks of ECC layer at the bottom of the RC
beams in this study,
Figure 4 displays andthethe maximum
typical crack width
distribution at ultimateofload
of micro-cracks ECC is layer
less than
at the0.08 mm. Therefore,
bottom of the RC
in this study, the PZT patches were only bonded on the surface
beams in this study, and the maximum crack width at ultimate load is less than 0.08 mm. of ECC layers of two ECC-concrete
Therefore,
composite
in this study, beams BE1 patches
the PZT and BE2,were and only
the conductance
bonded on the signatures
surface of of these transducers
ECC layers of twowere measured
ECC-concrete
under several damage scenarios by using an LCR meter. The RC beam
composite beams BE1 and BE2, and the conductance signatures of these transducers were measured was only tested under bending
loads and
under useddamage
several as reference beamby
scenarios to compare
using an the LCR flexural
meter.behaviors
The RC of beam BE1wasand only
BE2. tested
Furthermore,
under
accordingloads
bending to previous
and used studies [25,27], abeam
as reference singletoPZT transducer
compare is more behaviors
the flexural effective toofdetect
BE1 structural
and BE2.
damages within
Furthermore, a sensing
according to range
previousof 400 mm in
studies the electromechanical
[25,27], a single PZT transducer conductance based
is more approach.
effective to
Hence,structural
detect five PZT patches
damages were located
within at the surface
a sensing range of of400
ECCmm layers with
in the a distance of 300 conductance
electromechanical mm between
each sensor
based approach.as shown
Hence,in five
Figure
PZT3.patches were located at the surface of ECC layers with a distance of
300 mm between each sensor as shown in Figure 3.
Sensors 2019, 19, 2799 6 of 23
Sensors 2019, 19, x 6 of 23
Sensors 2019, 19, x 6 of 23

150 600 500 600 150

150 600 500 600 150
F 1 F
F 1 F

240
300

6060 240
300

PZT1 PZT2 PZT3 PZT4 PZT5 E1 PZT3 E2

PZT1 PZT2 PZT3 PZT4 PZT5 E1 PZT3 E2
1 Section 1-1
1 Section 1-1
750 250 250 750
750 250 250 750
400 300 300 300 300 400
400 300 300 300 300 400

PZT transducer LVDT Strain gauge

PZT transducer LVDT Strain gauge

Figure 3. Schematic test setup and layout of LVDTs and PZT sensors on the beams (unit: mm).
Figure 3. Schematic
Schematic test
test setup
setup and layout of LVDTs and PZT
LVDTs and PZT sensors
sensors on
on the
the beams
beams (unit:
(unit: mm).
mm).

Figure 4. Maximum crack width on ECC layer during loading test.

Figure 4.
Figure 4. Maximum
Maximum crack
crack width
width on
on ECC
ECC layer during loading
layer during loading test.
test.

A
A preloading
preloading procedure
procedure was
was firstly conducted
conducted for
firstly conducted
firstly the
for the
for specimens.
specimens. All
the specimens. All of
All of the
the test
test beams
beams were
were
subjected
subjected to to bending
bending loads
loads with
with an
an increment
increment of of 33 kN
kN up up toto 66 kN,
kN, and
and the
the bending
bending loads
loads were
were
released
released toto check
check the structural
the structural
the condition
condition of
structural condition specimens.
specimens. Then,
of specimens.
of Then, the
the applied
applied loads
loads onon the
the specimens
specimens
specimens
were gradually increased up to the ultimate bearing capacity of beams in order
were gradually increased up to the ultimate bearing capacity of beams in order to generate different
to generate different
levels of
levels of
structural
of structural damages.
structuraldamages. After
damages.After that,
Afterthat, the
that,the bending
thebending
bending loads
loads
loads
were
were
were
released
released
released
gradually
gradually
gradually
by controlling
by controlling
by controlling the
the
the servo-hydraulic
servo-hydraulic testing
testing machine.
machine. During
During the
the loading
loading process,
process, the
the deflection
deflection at
at midspan
midspan of
of three
three
servo-hydraulic testing machine. During the loading process, the deflection at midspan of three test
test
test beamswere
beamswere measured
measured by
by LVDTs
LVDTs as
as shown
shown in Figure
in 3,
Figure 3,
3, and
and they
they were
were continuously
continuously read and
beamswere measured by LVDTs as shown in Figure and they were continuously read and read and
recorded
recorded
recorded by using a data acquisition system.
by using aby using
data a data acquisition
acquisition system. system.
Internal
Internal embedded
embedded PZT PZT patches,
patches, also
also known
known as as “smart
“smart aggregate”,
aggregate”, werewere usedused to
to detect
detect
structural
structural damages
damages in in previous
previous studies
studies [43–46],
[43–46], but
but they
they must
must be be deployed
deployed prior
prior toto concrete
concrete casting
casting
Sensors 2019, 19, 2799 7 of 23

Internal embedded PZT patches, also known as “smart aggregate”, were used to detect structural
Sensors 2019, 19, x 7 of 23
damages in previous studies [43–46], but they must be deployed prior to concrete casting and are not
applicable
and are not forapplicable
existing structures.
for existing Therefore, in this
structures. work PZTintransducers
Therefore, this work were PZT externally
transducers bonded
were
on the bottom surface of beams in order to monitor the appearance and
externally bonded on the bottom surface of beams in order to monitor the appearance and evolution evolution of structural
damages
of structuralduring its service
damages duringtime. As shown
its service time. As in Figure
shown 5, in an LCR5,meter
Figure an LCR (model
meterE4980A, Agilent
(model E4980A,
Technologies,
Agilent city, statecity,
Technologies, abbrev
stateif abbrev
USA, country)
if USA, was employed
country) to measuretothe
was employed conductance
measure signatures
the conductance
signatures of five uniformlydistributed PZT patches of each test specimen under differentscenarios.
of five uniformlydistributed PZT patches of each test specimen under different damage damage
To achieve the purpose, a sinusoidal voltage signal with 1 volt amplitude was
scenarios. To achieve the purpose, a sinusoidal voltage signal with 1 volt amplitude was generated generated by the LCR
meter to apply on the surface of PZT patches to sweep over a defined frequency
by the LCR meter to apply on the surface of PZT patches to sweep over a defined frequency range. range. The output
electrical
The output conductance of PZT sensors
electrical conductance were
of PZT measured
sensors wereto detect the
measured todamages
detect theinduced
damages byinduced
changesby in
the mechanical
changes impedance impedance
in the mechanical of host structure
of hostasstructure
shown inasEquation
shown in(1). The alternations
Equation in the output
(1). The alternations in
signals become indicative of the structural damages when the host structure
the output signals become indicative of the structural damages when the host structure is is monitored bymonitored
extracting
theextracting
by electrical conductance
the electricalsignals of PZTsignals
conductance patches. of Therefore,
PZT patches. notTherefore,
only the appearance
not only the of appearance
damages in
ECC-concrete
of damages incomposite beamscomposite
ECC-concrete can be detected
beamsbut canalso
bethe propagation
detected but alsocan the
be assessed according
propagation can beto
the measured electrical conductance signatures during the loading test.
assessed according to the measured electrical conductance signatures during the loading test.

Figure 5. Schematic
Schematic diagram
diagram of
of the
the experimental setup for electrical conductance acquisition.

4. Test
4. Test Results
Results and
and Discussion
Discussion
4.1. Flexural Behavior
4.1. Flexural Behavior
As shown in Figure 6a–c, the relationships of load and midspan displacement of three test
As shown in Figure 6a–c, the relationships of load and midspan displacement of three test
specimens are obtained by using LVDTs installed at the center of the beams. The first cracks of the RC
specimens are obtained by using LVDTs installed at the center of the beams. The first cracks of the
beam, BE1 and BE2 appeared at loads of 45 kN, 51 kN and 56 kN, respectively. The micro-cracks of BE1
RC beam, BE1 and BE2 appeared at loads of 45 kN, 51 kN and 56 kN, respectively. The micro-cracks
and BE2 perpendicular to the direction of maximum principal stress were spaced tightly along the ECC
of BE1 and BE2 perpendicular to the direction of maximum principal stress were spaced tightly
layer of the beams and crack width were smaller than that of the reference RC beam. Then, the tensile
along the ECC layer of the beams and crack width were smaller than that of the reference RC beam.
longitudinal steel of BE1 and BE2 yielded at 205 kN and 210 kN with related midspan deflection of
Then, the tensile longitudinal steel of BE1 and BE2 yielded at 205 kN and 210 kN with related
6.2 mm and 6.7 mm, respectively, while the yielding load of RC beam was 191 kN with a midspan
midspan deflection of 6.2 mm and 6.7 mm, respectively, while the yielding load of RC beam was 191
deflection of 5.5 mm. Afterwards, the midspan deflection increased quickly while the applied load
kN with a midspan deflection of 5.5 mm. Afterwards, the midspan deflection increased quickly
increased slowly until the final failure of the three test specimens, and the number of cracks of BE1
while the applied load increased slowly until the final failure of the three test specimens, and the
and BE2 increased with the increase in the applied load while the crack width nearly performed to be
number of cracks of BE1 and BE2 increased with the increase in the applied load while the crack
constant, owing to the excellent capability to resist crack growth of ECC materials.
width nearly performed to be constant, owing to the excellent capability to resist crack growth of
ECC materials.
Sensors 2019, 19, 2799 8 of 23

Sensors 2019, 19, x 8 of 23

250

200

Load (kN) 150

100

50

0
0 5 10 15 20 25 30 35
Midspan deflection (mm)
(a)
: Admittance measurement
250 : Yield point
μδ = 1.0
200 μδ = 2.0 μ = 3.0
D2 δ
D3
Load (kN)

150

μδ = 0.57
100 D1

50
Healthy state
0
0 5 10 15 20 25 30 35
Midspan deflection (mm)
(b)
: Admittance measurement
250 : Yield point
μδ = 1.0 μδ = 3.0
200 μδ = 2.0 D3
D2
Load (kN)

150
μδ = 0.51
100 D1

50
Healthy state
0
0 5 10 15 20 25 30 35
Midspan deflection (mm)
(c)
Figure6.6.Loadmidspan
Figure Loadmidspandeflection
deflectionof
ofthe
the three
three test
test specimens:
specimens: (a)
(a) RC
RCbeam;
beam;(b)
(b)BE1;
BE1;(c)
(c)BE2.
BE2.
Sensors 2019, 19, 2799 9 of 23

Sensors 2019, 19, x 9 of 23

By contrast, the crack spaces and crack width of RC beam are larger than those with ECC layers at
bottom, and the crack
By contrast, width
the crack would
spaces andincrease as the of
crack width cracks propagated
RC beam towards
are larger the compression
than those zone
with ECC layers
during the loading
at bottom, and thetest.
crackAccording
width wouldto theincrease
test results, there
as the is no obvious
cracks propagated improvement
towards theof load-carrying
compression
capacity of RCthe
zone during beams by using
loading an ECC layer
test. According to as
thestrengthening. Nevertheless,
test results, there the midspan
is no obvious deflection
improvement of
of BE1 and BE2 related to ultimate load are 23.1 mm and 28.8 mm, respectively,
load-carrying capacity of RC beams by using an ECC layer as strengthening. Nevertheless, the which represent
amidspan
significant increaseofofBE1
deflection 26.2%
andandBE257.4%
relatedcompared
to ultimate with 18.3
load mm
are ofmm
23.1 the reference RC beam.
and 28.8 mm, This is
respectively,
mainly attributed to the fiber bridging and excellent crack control ability of the
which represent a significant increase of 26.2% and 57.4% compared with 18.3 mm of the reference PVA-ECC matrix in the
ECC layer according
RC beam. to its strain-hardening
This is mainly attributed to the response. The load
fiber bridging andand midspan
excellent deflection
crack control correspond to
ability of the
yielding
PVA-ECC and ultimate
matrix in thepoints
ECC are summarized
layer according to inits
Table 2.
strain-hardening response. The load and midspan
deflection correspond to yielding and ultimate points are summarized in Table 2.
Table 2. Yield and ultimate load, and related midspan deflection of test beams.
Table 2. Yield Yielding
and ultimate load, and related midspan deflection of test beams.
Ultimate
Beam Type Failure Mode
Load (kN) Midspan
Yielding Deflection (mm) Load (kN) Midspan
UltimateDeflection (mm)
Beam Type Failure Mode
RC beam Load 191
(kN) Midspan Deflection
5.5 (mm) Load211
(kN) Midspan Deflection
18.3 (mm) Concrete crushing
RCBE1
beam 191
205 5.5
6.2 211
212 18.3
23.1 Concrete crushing
Concrete crushing
BE2
BE1 210
205 6.7
6.2 224
212 28.8
23.1 Concrete crushing
Concrete crushing
BE2 210 6.7 224 28.8 Concrete crushing

Failure modes of the test specimens under bending loads are displayed in Figure 7a–c. Similar
Failure modes of the test specimens under bending loads are displayed in Figure 7a–c. Similar
failure modes of three specimens were observed that the concrete in compressive zone crushed at the
failure modes of three specimens were observed that the concrete in compressive zone crushed at
ultimate load after the yielding of steel reinforcement resulting in the final failure of beams.
the ultimate load after the yielding of steel reinforcement resulting in the final failure of beams.

(a)

(b)

(c)
Figure 7.
Figure 7. Failure modes of
Failure modes of three
three test
test specimens:
specimens: (a)
(a) RC beam; (b)
RC beam; (b) BE1;
BE1; (c)
(c) BE2.
BE2.

Furthermore, it is clear that multiple micro-cracks

Furthermore, micro-cracks were
were distributed
distributed inin a tightly space in the ECC
at the
layer at thebottom
bottomofofBE1
BE1 and
and BE2
BE2 as as shown
shown in Figure
in Figure 7b,c,7b,c,
whichwhich is distinctively
is distinctively different
different with
with those
those
of the of the conventional
conventional RC beam.RC beam.
Owning to the micro-cracks and small crack width of the ECC layer, the two strain gauges (E1
and E2) installed at the middle section of BE1 and BE2 are able to be used to measure the ECC strain
at midspan as shown in Figure 3. The load-strain relationships of the ECC layers of BE1 and BE2 are
Sensors 2019, 19, 2799 10 of 23

Owning to the micro-cracks and small crack width of the ECC layer, the two strain gauges (E1
and E2) installed at the middle section of BE1 and BE2 are able to be used to measure the ECC strain at
midspan
Sensorsas shown
2019, 19, x in Figure 3. The load-strain relationships of the ECC layers of BE1 and 10 ofBE2
23 are
displayed in Figure 8. The ECC strain at the bottom of two beams increases as the bending loads
displayed
increase until thein Figure 8. The of
final failure ECCthestrain at the bottom
test beams. of two ultimate
The average beams increases as the of
ECC strains bending
BE1 andloads
BE2 are
increase until the final failure of the test beams. The average ultimate ECC strains of BE1 and BE2
4400 and 4100 µε respectively, which demonstrates the excellent ductility of ECC materials.
are 4400 and 4100 με respectively, which demonstrates the excellent ductility of ECC materials.

240

200 E1
E2
160
Load (kN)

120

80

40

0
0 1000 2000 3000 4000 5000
Strain (με)
(a)
240

200 E1
E2
160
Load (kN)

120

80

40

0
0 1000 2000 3000 4000 5000
Strain (με)
(b)
Figure
Figure 8. Load-strain
8. Load-strain relationshipofofECC
relationship ECClayer
layer at
at the
themidspan
midspanofofbeams: (a)BE1,
beams: (b)BE2.
(a) BE1, (b) BE2.

4.2. Measured
4.2. Measured Conductance
Conductance Signatures
Signatures
To measure
To measure the conductance
the conductance spectra
spectra of five
of the the five
PZTPZT patches
patches of BE1
of BE1 andand
BE2,BE2,
fourfour different
different damage
damage levels of ECC reinforced concrete beams aredefined in Table 3 and described
levels of ECC reinforced concrete beams aredefined in Table 3 and described as follows: as follows:
(1) The intact beam before applying bending loads is defined as “healthy state (D0)”;
(1) The intact beam before applying bending loads is defined as “healthy state (D0)”;
(2) At midspan deflection δ = 3 mm before the yielding point of steel reinforcement, which relate to
(2) At midspan
a deflection ductility δμδ==30.48
deflection mm before
and the for
μδ = 0.45 yielding point
BE1 and BE2,of steel reinforcement,
respectively which
(μδ = δ/δy, the relate to
midspan
a deflection ductility µ = 0.48 and µ = 0.45 for BE1 and BE2, respectively (µ
deflections related toδyielding point δδy are 6.2 mm and 6.7 mm for BE1 and BE2 as shown
δ = δ/δ y , the midspan
in Table 2).
deflections
This staterelated to yielding
is defined as “damage levelδ1y (D1)”.
point are 6.2 mm and 6.7 mm for BE1 and BE2 as shown in
Table 2). This state is defined as “damage level 1 (D1)”.
Sensors 2019, 19, 2799 11 of 23
Sensors 2019, 19, x 11 of 23

(3)(3) AtAtmidspan
midspan deflection
deflection δ=12.4δ=12.4
and and 13.4after
13.4 mm mmsteel
after steel yielding
yielding for BE1 andfor BE2,
BE1that
andcorresponded
BE2, that
µδ = 2.0 are to
tocorresponded μδ = 2.0asare
defined defined level
“damage as “damage level
2 (D2)”, and 2 (D2)”,
at the and
same at way, δ = way,
the same δ = 18.6
18.6 and 20.1and
mm
20.1 mm corresponded to μ
corresponded to µδ = 3.0 for BE1 and BE2, respectively, are defined as “damage level 3 (D3)”.3
δ = 3.0 for BE1 and BE2, respectively, are defined as “damage level
(D3)”.

Table3.3.Three
Table Threedifferent
different damage
damage levels of BE1
levels of BE1 and
andBE2.
BE2.

Damage Level
Damage Level μµδ δ Damage
DamageDescription
Description
D0
D0 00 Healthy
Healthystate
state
D1
D1 0.48(BE1),
0.48(BE1),0.45(BE2)
0.45(BE2) Flexural Cracking,
Flexural Cracking,Multiple
Multiplemicro-cracks
micro-cracks in
in ECC layer
ECC layer
D2
D2 2.0
2.0 Steel
Steel yielding,
yielding, Multiplemicro-cracks
Multiple micro-cracksin
in ECC
ECC layer
layer
D3
D3 3.0
3.0 Steel
Steel yielding,
yielding, Multiplemicro-cracks
Multiple micro-cracksin
in ECC
ECC layer
layer

Electricalconductance
Electrical conductancespectra
spectraof
ofthe
the externally
externally bonded
bonded PZT
PZTpatches
patcheswere
werefirst
firstmeasured
measured atat
the
the
beginning of the test for BE1 and BE2, which were under healthy state D0. After that,
beginning of the test for BE1 and BE2, which were under healthy state D0. After that, when the when the three
damage
three levels
damage D1, D2
levels D1,and
D2D3
andwere
D3 achieved with the
were achieved increase
with of applied
the increase load as illustrated
of applied in Figurein
load as illustrated
6b,c during the test, the conductance signatures were also captured by using the LCR meter. Figures
Figure 6b,c during the test, the conductance signatures were also captured by using the LCR meter.
9 and 10 show the conductance signatures measured by the PZT transducers of BE1 and BE2 under
Figures 9 and 10 show the conductance signatures measured by the PZT transducers of BE1 and BE2
different damage scenarios, from D0 to D3. The main resonant peaks around 100–200 kHz and
under different damage scenarios, from D0 to D3. The main resonant peaks around 100–200 kHz and
400–500 kHz can be explained by the electromechanical coupling effects.
400–500 kHz can be explained by the electromechanical coupling effects.

7 PZT1 D0
6 D1
D2
5 D3
Conductance (mS)

0
0 100 200 300 400 500 600
Frequency (kHz)

6 PZT2 D0
D1
5 D2
D3
Conductance (mS)

0
0 100 200 300 400 500 600
Frequency (kHz)

Figure 9. Cont.
Sensors 2019, 19, 2799 12 of 23
Sensors 2019, 19, x 12 of 23

6 PZT3 D0
D1
5 D2
D3

Conductance (mS)
4

0
0 100 200 300 400 500 600
Frequency (kHz)

6
PZT4 D0
D1
5
D2
D3
Conductance (mS)

0
0 100 200 300 400 500 600
Frequency (kHz)

5 D0
PZT5
D1
4 D2
D3
Conductance (mS)

0
0 100 200 300 400 500 600
Frequency (kHz)

Figure
Figure9. Comparisons
9. Comparisons of electrical conductance
of electrical signature
conductance of each
signature PZT sensor
of each under under
PZT sensor different damage
different
levels of BE1.
damage levels of BE1.
Sensors 2019, 19, 2799 13 of 23
Sensors 2019, 19, x 13 of 23

12 PZT1 D0
D1
10 D2
D3

Conductance (mS)
8

0
0 100 200 300 400 500 600
Frequency (kHz)

10 D0
PZT2
D1
D2
8
D3
Conductance (mS)

0
0 100 200 300 400 500 600
Frequency (kHz)

10 PZT3 D0
D1
8 D2
D3
Conductance (mS)

0
0 100 200 300 400 500 600
Frequency (kHz)

Figure 10. Cont.

Sensors 2019, 19, 2799 14 of 23
Sensors 2019, 19, x 14 of 23

12 PZT4 D0
D1
10 D2
D3

Conductance (mS)
8

0
0 100 200 300 400 500 600
Frequency (kHz)

10 D0
PZT5
D1
8 D2
D3
Conductance (mS)

0
0 100 200 300 400 500 600
Frequency (kHz)

Figure 10. Comparisons

Figure 10. Comparisonsofof
electrical conductance
electrical signature
conductance of each
signature of PZT
eachsensor under different
PZT sensor damage
under different
levels
damage of levels
BE2. of BE2.

It should be noted that variations of temperature and humidity would affect the measured
It should be noted that variations of temperature and humidity would affect the measured
conductance signatures of PZT patches and result in erroneous detection results regarding the
conductance signatures of PZT patches and result in erroneous detection results regarding the
structural integrity. Considering the experimental tests in this work were conducted at almost constant
structural integrity. Considering the experimental tests in this work were conducted at almost
temperature and humidity, thus, the influences of environment were neglected in the proposed work.
constant temperature and humidity, thus, the influences of environment were neglected in the
Discrepancies among different damage scenarios can be observed which reflect the variations of
proposed work.
structural integrity under different loading levels. It can be observed that the variations of conductance
Discrepancies among different damage scenarios can be observed which reflect the variations of
spectra at a higher frequency range are greater than those in lower frequency range. This is mainly
structural integrity under different loading levels. It can be observed that the variations of
due to the high sensitivity of conductance spectra in high frequency band to detect structural changes,
conductance spectra at a higher frequency range are greater than those in lower frequency range.
which have been confirmed in previous studies [29]. Furthermore, it is also found that by only
This is mainly due to the high sensitivity of conductance spectra in high frequency band to detect
observing the variations in conductance signatures it is difficult to conduct damage assessment to
structural changes, which have been confirmed in previous studies [29]. Furthermore, it is also
different damage levels.
found that by only observing the variations in conductance signatures it is difficult to conduct
According to the EMA technique adopted in this work, the measured conductance before
damage assessment to different damage levels.
damage (initial state D0) and after damage (state D1, D2 and D3) will change as the applied load
According to the EMA technique adopted in this work, the measured conductance before
increases. Root mean square deviation (RMSD), which is directly proportional to the severity of
damage (initial state D0) and after damage (state D1, D2 and D3) will change as the applied load
structural damages with a maximum value of 1 indicating severe damage, is typically used to obtain
increases. Root mean square deviation (RMSD), which is directly proportional to the severity of
structural damages with a maximum value of 1 indicating severe damage, is typically used to obtain
Sensors 2019, 19, 2799 15 of 23
Sensors 2019, 19, x 15 of 23

aa quantitative
quantitative evaluation
evaluation ofofstructural
structuraldamages.
damages.The Thegreater
greaterthe
thedamage,
damage, the greater
the greaterthethe
variation in
variation
the conductance spectra and, based on its formulation in Equation (2), the greater
in the conductance spectra and, based on its formulation in Equation (2), the greater the RMSD the RMSD value.
By introducing
value. the statistical
By introducing scalar
the statistical index,
scalar quantification
index, forfor
quantification conductance
conductancevariations
variationsinduced
induced byby
damage can be identified. Thus, RMSD index is deployed to assess the evolution of damage
damage can be identified. Thus, RMSD index is deployed to assess the evolution of damage severity severity
of
of ECC-concrete
ECC-concrete composite
composite beams
beams in this
in this work:
work:
N


N (Zk − Z0 )
2
v
u
u
u
2
u
u P
(Zk − Z0 )
u
×100, k = 1, 2,3
u
RMSD(%)= uu 11 N
u
(2)
u
u
RMSD(%) = × 100, k = 1, 2, 3 (2)

t
P (Z )2
N 2
(Z00)
11

where Z0 represents the conductance of the PZT transducer at healthy state (D0), which is used as
where Z0 represents the conductance of the PZT transducer at healthy state (D0), which is used as
baseline signature. Zk is the measurement at different damage states, where the subscript k=1,2,3 is
baseline signature. Zk is the measurement at different damage states, where the subscript k = 1,2,3
related to the three damage levels D1, D2 and D3 defined in Table 3. N is the sampling frequency
is related to the three damage levels D1, D2 and D3 defined in Table 3. N is the sampling frequency
points of measurements.
points of measurements.
To comprehensively assess the structural damage of BE1 and BE2 monitored by PZT sensors
To comprehensively assess the structural damage of BE1 and BE2 monitored by PZT sensors
that are distributed along the length of beam, RMSD value of each five sensors by using electrical
that are distributed along the length of beam, RMSD value of each five sensors by using electrical
conductance measured with a sweeping frequency range between 0 and 600 kHz are displayed in
conductance measured with a sweeping frequency range between 0 and 600 kHz are displayed in
Figures 11 and 12. It is clear that, more or less, with the increase of damage level, the RMSD values
Figures 11 and 12. It is clear that, more or less, with the increase of damage level, the RMSD values
evaluated from the conductance measurements in most of the PZT transducers are ascending from
evaluated from the conductance measurements in most of the PZT transducers are ascending from D1
D1 to D3.The induced structural damages, such as concrete cracking and steel yielding, would
to D3.The induced structural damages, such as concrete cracking and steel yielding, would enlarge
enlarge with the increase of bending loads during the experimental test. Therefore, the progressive
with the increase of bending loads during the experimental test. Therefore, the progressive growth
growth of damage levels is confirmed by the increase of damage indicators calculated from
of damage levels is confirmed by the increase of damage indicators calculated from conductance
conductance measurements, which corresponds quite well with the real damage case during the
measurements, which corresponds quite well with the real damage case during the bending test.
bending test. Furthermore, it is noted that slight decreases are observed for PZT 3 of BE1 from D2 to
Furthermore, it is noted that slight decreases are observed for PZT 3 of BE1 from D2 to D3 and PZT 1
D3 and PZT 1 of BE2 from D2 to D3, this might be attributed to the disturbance during the test, e.g.,
of BE2 from D2 to D3, this might be attributed to the disturbance during the test, e.g., measurement
measurement noise and temperature variation. In general, the damage indicator predicts a
noise and temperature variation. In general, the damage indicator predicts a reasonable quantitative
reasonable quantitative trend that is compatible with the progressive growth of structural damages.
trend that is compatible with the progressive growth of structural damages.

12 12
PZT 1 PZT 2

10 10

8 8
RMSD (%)
RMSD (%)

6 6

4 4

2 2

0 0
D1 D2 D3 D1 D2 D3

Figure 11. Cont.

Sensors 2019, 19, 2799
x 16 of 23
Sensors 2019, 19, x 16 of 23

12 12
12 PZT 3 12 PZT 4
PZT 3 PZT 4
10 10
10 10

8 8
8 8
(%)

(%)
RMSD(%)

RMSD(%)
6 6
6 6

RMSD
RMSD

4 4
4 4

2 2
2 2

0 0
0 0
D1 D2 D3 D1 D2 D3
D1 D2 D3 D1 D2 D3

12
12 PZT 5
PZT 5
10
10

8
8
(%)
RMSD(%)

6
6
RMSD

4
4

2
2

0
0
D1 D2 D3
D1 D2 D3

Figure 11.
Figure 11.
Figure Experimental RMSD
Experimental RMSD
11. Experimental values
RMSD values of
values of each
of each PZT
each PZT sensor
PZT sensor for
sensor for BE1
for BE1 under different
BE1 under different damage
damage levels.
levels.

16 16
16 PZT 1 16 PZT 2
PZT 1 PZT 2
14 14
14 14
12 12
12 12
10 10
10 10
(%)

(%)
RMSD(%)

RMSD(%)

8 8
8 8
RMSD

RMSD

6 6
6 6
4 4
4 4
2 2
2 2
0 0
0 0
D1 D2 D3 D1 D2 D3
D1 D2 D3 D1 D2 D3
16 16
16 PZT 3 16 PZT 4
PZT 3 PZT 4
14 14
14 14
12 12
12 12
10 10
10 10
(%)
(%)

RMSD(%)
RMSD(%)

8 8
8 8
RMSD
RMSD

6 6
6 6
4 4
4 4
2 2
2 2
0 0
0 0
D1 D2 D3 D1 D2 D3
D1 D2 D3 D1 D2 D3

Figure 12. Cont.

Sensors 2019, 19, 2799 17 of 23
Sensors 2019, 19, x 17 of 23
Sensors 2019, 19, x 17 of 23
16
PZT 5
16
14 PZT 5
14
12
12
10

(%)(%)
10
8

RMSD
8
6

RMSD
6
4
4
2
2
0
0 D1 D2 D3
D1 D2 D3
Figure 12. Experimental RMSD values of each PZT sensor for BE2 under different damage levels.
Figure 12. Experimental RMSD values of each PZT sensor for BE2 under different
different damage
damage levels.
levels.
In order to evaluate the influence of the location and severity of damage on conductance
In
In order
orderto evaluate thethe
influence of theoflocation and severity of damageof on conductance signatures
signatures andtotheevaluate
ability of influence
local sensing thePZT
of location and severity
transducers, the RSMD damage
values on conductance
calculated from
and the ability of local sensing of PZT transducers, the RSMD values calculated from experimental
experimental measurements of each five PZT sensor uniformly located along the length of BE1from
signatures and the ability of local sensing of PZT transducers, the RSMD values calculated and
measurements
experimental of each five PZTofsensor
measurements uniformly located along thelocated
length of BE1 and BE2 areofcompared
BE2 are compared with the realeach five
crack PZT sensor
patterns underuniformly
different damage along
levels,the
as length
shown inBE1 and
Figures
with the real crack patterns under different damage levels, as shown in Figures 13–18.
BE2 are compared with the real crack patterns under different damage levels, as shown in Figures
13–18.
13–18.
14
14 D1
D1
12
12
10
10
8
(%)

8
RMSD(%)

6
RMSD

6
4
4
2
2
0
0
PZT 1 PZT 2 PZT 3 PZT 4 PZT 5
PZT 1 PZT 2 PZT 3 PZT 4 PZT 5

Figure 13. RMSD of each PZT sensor at D1 and the related cracking pattern of BE1.
Figure 13. RMSD of each PZT sensor at D1 and the related cracking pattern of BE1.
Figure 13. RMSD of each PZT sensor at D1 and the related cracking pattern of BE1.
Sensors 2019, 19, 2799 18 of 23
Sensors 2019,
Sensors 19, 19,
2019, x x 18 of
1823
of 23

1414
D2D2
1212

1010

8 8
RMSD (%)
RMSD (%)

6 6

4 4

2 2

0 0
PZT 1 1 PZT
PZT 2 2 PZT
PZT 3 3 PZT
PZT 4 4 PZT
PZT 5 5
PZT

Figure
Figure 14.14.
Figure
14. RMSD
RMSD of
of each
RMSD each PZT
of each
PZT sensor
PZT at
at D2
sensor
sensor and
at D2
D2 and the
and related
thethe cracking
related
related pattern
cracking
cracking of BE1.
pattern
pattern of BE1.
of BE1.

1414
D3D3
1212

1010

8 8
RMSD (%)
RMSD (%)

6 6

4 4

2 2

0 0
PZT 1 1 PZT
PZT 2 2 PZT
PZT 3 3 PZT
PZT 4 4 PZT
PZT 5 5
PZT

Figure 15.15.
Figure RMSD of each
RMSD PZT
of each sensor
PZT at D3
sensor andand
at D3 thethe
related cracking
related pattern
cracking of BE1.
pattern of BE1.
Figure 15. RMSD of each PZT sensor at D3 and the related cracking pattern of BE1.
Sensors 2019, 19, 2799 19 of 23
Sensors 2019,
Sensors 2019, 19,
19, xx 19 of
19 of 23
23

16
16 D1
D1
14
14
12
12
10
10
(%)
RMSD(%)
88
RMSD

66
44
22
00
PZT 11
PZT PZT 22
PZT PZT 33
PZT PZT 44
PZT PZT 55
PZT

Figure
Figure
Figure 16.
16.16. RMSD
RMSD
RMSD of each
of of each
each PZTPZT
PZT sensor
sensor
sensor at D1
at
at D1 D1
andand
and the
thethe related
related
related cracking
cracking
cracking pattern
pattern
pattern of BE2.
of BE2.
of BE2.

16
16 D2
D2
14
14
12
12
10
10
(%)
RMSD(%)

88
RMSD

66
44
22
00
PZT 11
PZT PZT 22
PZT PZT 33
PZT PZT 44
PZT PZT 55
PZT

Figure 17.
Figure 17. RMSD
RMSD of
of each
each PZT
PZT sensor
sensor at
at D2
D2 and
and the
the related
related cracking
cracking pattern
pattern of
of BE2.
BE2.
Figure 17. RMSD of each PZT sensor at D2 and the related cracking pattern of BE2.
Sensors 2019, 19, 2799 20 of 23

Sensors 2019, 19, x 20 of 23

16 D3

14
12
10
RMSD (%)
8
6
4
2
0
PZT 1 PZT 2 PZT 3 PZT 4 PZT 5

Figure
Figure 18.18. RMSD
RMSD of each
of each PZTPZT sensor
sensor at D3
at D3 andand
thethe related
related cracking
cracking pattern
pattern of BE2.
of BE2.

In In Figure
Figure 13,13,thethe
RMSDRMSD show
show low low values
values forfor
allall sensors,
sensors, which
which represents
represents thatthat
BE1 BE1hashas a low
a low
damage
damage severity
severity (D1).(D1). It be
It can cannoted
be noted
that the that the damage
damage indexofvalues
index values PZT 3 and of PZTPZT34and at thePZT 4 at the
midspan
of beam are higher than those of PZT 1, PZT 2 and PZT5, which indicates the presence of crack damages of
midspan of beam are higher than those of PZT 1, PZT 2 and PZT5, which indicates the presence
crack
at the damages
midspan at the midspan
as cracking as cracking
load is achieved. load is achieved.
Afterwards, a significantAfterwards,
increase ofa RSMDsignificant
values increase
can be of
RSMD for
observed valueseachcan be observed
sensor for each
as the applied loadsensor as the
increases applied load
as illustrated increases
in Figures 14as andillustrated in Figures
15, especially for
PZT 2, PZT 3 and PZT 4 which are located in the vicinity of the midspan of BE1. This phenomenon can of
14 and 15, especially for PZT 2, PZT 3 and PZT 4 which are located in the vicinity of the midspan
be BE1. This phenomenon
explained by the increasing cannumber
be explained by the
of multiple increasing number
micro-cracks in the ECC of layers
multiple andmicro-cracks
the propagation in the
ECC layers
of concrete andand
cracks, the apropagation
high RMSD value of concrete cracks,
of the PZT andindicates
sensor a high RMSD value ofseverity
a high damage the PZTofsensorthe
host structure and vice versa. General speaking, by comparing the damage index RMSD values of theby
indicates a high damage severity of the host structure and vice versa. General speaking,
comparing
distributed PZT thesensors
damage onindex
the beam,RMSD thevalues
growthofofthe distributed
damage severity PZT sensors
would cause onathe beam,
relative the growth
increase of
of damage severity would cause a relative
the RMSD value of the transducer nearest to the damage. increase of the RMSD value of the transducer nearest to
theSimilar
damage. phenomena are shown in Figures 16–18 for the damage detection of BE2. PZT 3, that is
located Similar phenomena
at the middle aredisplays
of beam, shown ina Figures
higher RMSD 16–18 value
for thethandamage detection
the other sensors of BE2. PZT 3,level
at damage that is
located at the middle of beam, displays a higher RMSD value than
D1after initial cracking. Then, the expansion of induced damages caused by the applied load would the other sensors at damage level
D1afterthe
influence initial
outputcracking. Then, theofexpansion
measurements PZT sensors of induced
and result damages
in higher caused
valuesby ofthe
RMSD applied
metric.loadInwould
the
sameinfluence
way, the theRMSD
outputvalues
measurements of PZTwhich
of PZT patches sensorsare and result inbonded
externally higher values
in the of RMSD metric.
neighborhood of In
middle section of BE2, such as PZT 2, PZT 3 and PZT 4, are higher than PZT 1 and PZT 5, which are of
the same way, the RMSD values of PZT patches which are externally bonded in the neighborhood
middle
placed nearsection
the two of supports.
BE2, such as PZT 2, PZT 3 and PZT 4, are higher than PZT 1 and PZT 5, which are
placed near the two supports. between damage prediction and cracking patterns in a real case,
According to the comparison
it can be concluded that forcomparison
According to the an array of PZT between damageinprediction
transducers ECC-concrete andcomposite
cracking patterns
beams the instructural
a real case,
it can alters
damage be concluded that for measurements
the conductance an array of PZT of transducers
a particular PZT in ECC-concrete
sensor in inverse composite beamstothe
proportion
the distance from the damage location, which implies the nearest PZT is most influenced in
structural damage alters the conductance measurements of a particular PZT sensor andinverse
the
proportion
farther ones less. to Taking
the distance from the
into account damage
the large scalelocation, which and
of the problem implies the nearest during
the uncertainties PZT is the most
influenced
tests, RMSD value and obtained
the farther fromones
theless. Taking
variations of into account
measured the large signatures
conductance scale of theusing problem andofthe
an array
uncertainties
externally bonded during the tests,
PZT sensors seemRMSD value obtained
to provide satisfactory from the variations
reliability of measured of
for the identification conductance
damage
signatures
severity using an
at different array levels.
damage of externally bonded PZT sensors seem to provide satisfactory reliability
for the identification of damage severity at different damage levels.

5. Conclusions
Sensors 2019, 19, 2799 21 of 23

5. Conclusions
An impedance-based approach was used to monitor the evolution of structural damage of
ECC-concrete composite beams subjected to bending loads. To achieve this goal, three specimens
were prepared for four-point loading tests, and their flexural behaviors were recorded using a data
acquisition system. Meanwhile, three damage scenarios were selected to represent the different
damage severities of the specimens during the test, and electrical conductance signatures in the range
of 0–600 kHz were measured under different damage levels by using LCR meter. RMSD index is
successfully used to conduct quantitative damage assessment for ECC-concrete composite beams.
Following conclusions can be drawn from the experimental test:

(1) The ECC layer in the RC beams significantly increases the ultimate deflection due to its unique
high ductility, which is beneficial to the structural safety.
(2) The maximum crack width is less than 0.08 mm and a small crack spacing of 3-6 mm is observed
in this work, which demonstrates that the development of cracks in ECC-concrete composite
beams is well controlled due to the fiber bridging effect of ECC materials. This is very helpful to
prevent corrosion of steel reinforcements, in particular for aggressive environments.
(3) Damage detection has been successfully implemented using the statistical scalar index RMSD,
which is calculated from the variations of PZT conductance measurements. The upwards trending
of RMSD value confirms the increase of damage severity in the host concrete beam.
(4) The location of the occurred damages are also estimated by using an array of PZT sensors
externally bonded on the ECC layer, since the RMSD value of PZT transducer placed near the
region of higher damage severity is higher than those for other PZT sensors.

Author Contributions: F.Q. and Z.Z. took part in the experiments and paper edition. B.X. helped collect and
analyze the data. R.S. conceived this study, and wrote most part of the paper.
Funding: This research was funded by the National Natural Science Foundation of China (Project No. 51708062
and 51608069), 111 Project of China (Grant No. B18062).
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Li, V.C.; Mishra, D.K.; Wu, C. Matrix design for pseudo strain-hardening fiber reinforced cementitious
composites. Mater. Struct. 1995, 28, 586–595. [CrossRef]
2. Li, V.C.; Wang, S.; Wu, C. Tensile strain-hardening behavior of polyvinyl alcohol engineered cementitious
composite (PVA-ECC). ACI Mater. J. 2001, 98, 483–492.
3. Fischer, G.; Li, V.C. Design of engineered cementitious composites (ECC) for processing and workability
requirement. Brittle Matrix Compos. 2003, 7, 29–36.
4. Wang, S.; Li, V.C. Polyvinyl alcohol fiber reinforced engineered cementitious composites: Material design
and performances. Available online: http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.67.4523&
rep=rep1&type=pdf (accessed on 22 June 2019).
5. Zhang, Z.G.; Yuvaraj, A.; Di, J.; Qian, S.Z. Matrix design of light weight, high strength, high ductility ECC.
Constr. Build. Mater. 2019, 210, 188–197. [CrossRef]
6. Kim, D.J.; Naaman, A.E.; El-Tawil, S. Comparative flexural behavior of four fiber reinforced cementitious
composites. Cem. Concr. Compos. 2008, 30, 917–928. [CrossRef]
7. Blunt, J.D.; Ostertag, C.P. Deflection hardening and workability of hybrid fiber composites. ACI Mater. J.
2009, 106, 265–272.
8. Zhang, J.; Leung, C.K.Y.; Gao, Y. Simulation of crack propagation of fiber reinforced cementitious composite
under direct tension. Eng. Fract. Mech. 2011, 78, 2439–2454. [CrossRef]
9. Yuan, F.; Pan, J.L.; Wu, Y.F. Numerical study on flexural behaviors of steel reinforced engineered cementitious
composite (ECC) and ECC concrete composite beams. Sci. China Technol. Sci. 2014, 57, 637–645. [CrossRef]
Sensors 2019, 19, 2799 22 of 23

10. Xu, S.; Hou, L.; Zhang, X. Flexural and shear behaviors of reinforced ultrahigh toughness cementitious
composite beams without web reinforcement under concentrated load. Eng. Struct. 2012, 39, 176–186.
[CrossRef]
11. Khan, M.I.; Abbass, W. Flexural behavior of high-strength concrete beams reinforced with a strain hardening
cement-based composite layer. Constr. Build. Mater. 2016, 125, 927–935. [CrossRef]
12. Meng, D.; Lee, C.K.; Zhang, Y.X. Flexural and shear behaviours of plain and reinforced
polyvinylalcohol-engineered cementitious composite beams. Eng. Struct. 2017, 151, 261–272. [CrossRef]
13. Zhang, Z.G.; Qian, S.Z.; Ma, H. Investigating mechanical properties and self-healing behavior of micro-cracked
ECC with different volume of fly Ash. Constr. Build. Mater. 2014, 52, 17–23. [CrossRef]
14. Zhang, Z.G.; Zhang, Q. Self-healing ability of engineered cementitious composites (ECC) under different
exposure environments. Constr. Build. Mater. 2017, 156, 142–151. [CrossRef]
15. Zhang, Z.G.; Hu, J.; Ma, H. Feasibility study of ECC with self-healing capacity applied on long-span steel
bridge deck overlay. Int. J. Pavement Eng. 2019, 20, 884–893. [CrossRef]
16. Herbert, E.M.; Li, V.C. Self-healing of microcracks in engineered cementitious composites (ECC) under
a natural environment. Materials 2013, 6, 2831–2845. [CrossRef] [PubMed]
17. Zhang, Z.; Qian, S.; Liu, H.; Li, V.C. Ductile concrete material with self-healing capacity for jointless concrete
pavement use. Transport Res Rec. 2017, 2640, 78–83. [CrossRef]
18. Zhang, Z.G.; Zhang, Q.; Li, V.C. Multiple-scale Investigations on Self-healing Induced Mechanical Property
Recovery of ECC. Cem. Concr. Compos. 2019, 103, 293–302. [CrossRef]
19. Gencturk, B.; Elnashai, A.; Lepech, M.; Billington, S. Behavior of concrete and ECC structures under simulated
earthquake motion. J. Struct. Eng. 2013, 139, 389–399. [CrossRef]
20. Qudah, S.; Maalej, M. Application of engineered cementitious composites (ECC) ininterior beam–column
connections for enhanced seismic resistance. Eng. Struct. 2014, 69, 235–245. [CrossRef]
21. Hosseini, F.; Gencturk, B.; Aryan, H.; Cadaval, G. Seismic behavior of 3-D ECC beam-column connections
subjected to bidirectional bending and torsion. Eng. Struct. 2018, 172, 751–763. [CrossRef]
22. Kim, Y.Y.; Fischer, G.; Li, V.C. Performance of bridge deck link slabs designed with ductile ECC. ACI Struct. J.
2004, 101, 792–801.
23. Lepech, M.D.; Li, V.C. Application of ECC for bridge deck link slabs. RILEM J. Mater. Struct. 2009, 42,
1185–1195. [CrossRef]
24. Lepech, M.D.; Li, V.C. Sustainable pavement overlays using engineered cementitious composites. J. Pavement
Res. Tech. 2010, 3, 241–250.
25. Park, G.; Sohn, H.; Farrar, C.R.; Inman, D.J. Overview of piezoelectric impedance-based health monitoring
and path forward. Shock Vib. Dig. 2003, 35, 451–463. [CrossRef]
26. Annamdas, V.G.; Radhika, M.A. Electromechanical impedance of piezoelectric transducers for monitoring
metallic and non-metallic structures: A review of wired, wireless and energy-harvesting methods. J. Intell.
Mater. Syst. Struct. 2013, 24, 1021–1042. [CrossRef]
27. Wongi, S.; Jongdae, B. A review of the piezoelectric electromechanical impedance based structural health
monitoring technique for engineering structures. Sensors 2018, 18, 1307.
28. Liang, C.; Sun, F.P.; Rogers, C.A. Coupled electro-mechanical analysis of adaptive material
system-determination of the actuator power consumption and system energy transfer. J. Intell. Mater.
Syst. Struct. 1994, 5, 12–20. [CrossRef]
29. Min, J.; Park, S.; Yun, C.B.; Lee, C.G.; Lee, C. Impedance-based structural health monitoring incorporating
neural network technique for identification of damage type and severity. Eng. Struct. 2012, 39, 210–220.
[CrossRef]
30. Sevillano, E.; Sun, R.; Gil, A.; Perera, R. Interfacial crack-induced debonding identification in FRP-strengthened
RC beams from PZT signatures using hierarchical clustering analysis. Compos. Part B Eng. 2016, 87, 322–335.
[CrossRef]
31. Wang, D.S.; Song, H.Y.; Zhu, H.P. Numerical and experimental studies on damage detection of a concrete
beam based on PZT admittances and correlation coefficient. Constr. Build. Mater. 2013, 49, 564–574.
[CrossRef]
32. Kim, J.T.; Park, J.H.; Hong, D.S.; Cho, H.M.; Na, W.B.; Yi, J.H. Vibration and impedance monitoring for
prestress-loss prediction in PSC girder bridges. Smart Struct. Syst. 2009, 5, 81–94. [CrossRef]
Sensors 2019, 19, 2799 23 of 23

33. Sun, R.; Sevillano, E.; Perera, R. Identification of intermediate debonding damage in FRP-strengthened RC
beams based on a multi-objective updating approach and PZT sensors. Compos. Part B Eng. 2017, 109,
248–258. [CrossRef]
34. Wang, Y.; Li, X.; Li, J.; Wang, Q.; Xu, B.; Deng, J. Debonding damage detection of the CFRP-concrete interface
based on piezoelectric ceramics by the wave-based method. Constr. Build. Mater. 2019, 210, 514–524.
[CrossRef]
35. Tseng, K.K.; Wang, L. Structural damage identification for thin plates using smart piezoelectric transducers.
Comput. Methods Appl. Mech. Eng. 2005, 194, 3192–3209. [CrossRef]
36. Yan, W.; Cai, J.B.; Chen, W.Q. An electromechanical impedance model of a cracked composite beam with
adhesively bonded piezoelectric patches. J. Sound Vib. 2011, 330, 287–307. [CrossRef]
37. Sun, R.; Sevillano, E.; Perera, R. Debonding detection of FRP strengthened concrete beams by using impedance
measurements and an ensemble PSO adaptive spectral model. Compos. Struct. 2015, 125, 374–387. [CrossRef]
38. Yang, Y.; Divsholi, B.S. Sub-frequency interval approach in electromechanical impedance technique for
concrete structure health monitoring. Sensors 2010, 10, 11644–11661. [CrossRef]
39. Ai, D.; Zhu, H.; Luo, H.; Wang, C. Mechanical impedance based embedded piezoelectric transducer for
reinforced concrete structural impact damage detection: A comparative study. Construct. Build. Mater. 2018,
165, 472–483. [CrossRef]
40. Ghafari, E.; Yuan, Y.; Wu, C.; Nantung, T.; Lu, N. Evaluation the compressive strength of the cement paste
blended with supplementary cementitious materials using a piezoelectric-based sensor. Construct. Build.
Mater. 2018, 171, 504–510. [CrossRef]
41. Zhang, Z.G.; Ding, Y.Z.; Qian, S.Z. Influence of bacterial incorporation on mechanical properties of engineered
cementitious composites (ECC). Construct. Build Mater. 2019, 196, 195–203. [CrossRef]
42. Zhang, Z.G.; Zhang, Q. Matrix tailoring of engineered cementitious composites (ECC) with non-oil-coated,
low tensile strength PVA fiber. Construct. Build Mater. 2018, 161, 420–431. [CrossRef]
43. Song, G.; Gu, H.; Mo, Y.L. Smart aggregates: Multi-functional sensors for concrete structures-a tutorial and
a review. Smart Mater. Struct. 2008, 17, 033001. [CrossRef]
44. Yan, S.; Sun, W.; Song, G.; Gu, H.; Huo, L.S.; Liu, B.; Zhang, Y.G. Health monitoring of reinforced concrete
shear walls using smart aggregates. Smart Mater. Struct. 2009, 18, 1–6.
45. Yan, S.; Ma, H.; Li, P.; Song, G.; Wu, J. Development and application of a structural health monitoring system
based on wireless smart aggregates. Sensors 2017, 17, 1641. [CrossRef] [PubMed]
46. Wang, Z.; Chen, D.; Zheng, L.; Huo, L.; Song, G. Influence of axial load on electromechanical impedance
(EMI)of embedded piezoceramic transducers in steel fiber concrete. Sensors 2018, 18, 1782. [CrossRef]

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).