1 Article

2 Piezoelectric Sensor-embedded Smart Rock for

3 Damage Monitoring in Prestressed Anchorage Zone
4 Quang-Quang Pham1, Ngoc-Loi Dang1, and Jeong-Tae Kim 1,*
5 Ocean Engineering Department, Pukyong National University, Busan 48723, Republic of Korea;
6 bkdn06x3a@gmail.com (Q.Q Pham); loi.ngocdang@gmail.com (N.L Dang)
7 *Correspondence: idis@pknu.ac.kr, Tel: +82-51-629-6585

8 Received: date; Accepted: date; Published: date

9 Abstract: In this paper, a piezoelectric sensor-embedded smart rock is proposed for

10 electromechanical impedance monitoring to detect internal concrete damage in a prestressed
11 anchorage zone. Firstly, a piezoelectric sensor-embedded smart rock is analyzed for impedance
12 monitoring in concrete structures. An impedance measurement model is analyzed for the PZT-
13 embedded smart rock under compression in a concrete member. Secondly, a prototype of the smart
14 rock embedded with a PZT (lead zirconate titanate) sensor is designed to sensitively catch variations
15 of impedance signatures induced by concrete damage in an anchorage zone. Thirdly, the
16 performance of smart rock is estimated from numerical analysis and experimental tests. Variations
17 in impedance signals under compressive test cases are analyzed to predetermine the sensitive
18 frequency band for the impedance monitoring. Lastly, an experiment on an anchorage zone
19 embedded with the smart rocks and surface-mounted PZT sensors is conducted for the impedance
20 measurement under a series of loading cases. The impedance variations are quantified to
21 comparatively evaluate the feasibility of the sensor-embedded smart rock for detecting internal
22 concrete damage in the anchorage zone. The result shows that the internal concrete damage was
23 successfully detected using the PZT-embedded smart rock, thus enabling the applicability of the
24 technique for the anchorage zone’s health monitoring.

25 Keywords: smart rock, electromechanical impedance, impedance-based technique, concrete

26 damage, anchorage zone.
27

28 1. Introduction
29 Prestressed concrete (PSC) has been widely used as critical members for bridge construction,
30 thanks to its cost-effectiveness and crack-resistance over original reinforced concrete structures. In
31 PSC bridges, anchorage zones have a vital function in carrying the designed prestress-force and
32 transferring it into structures. Considering time-dependent effects, the structural performance of PSC
33 members could gradually decline due to material deterioration, corrosion, and time-dependent
34 prestress loss, thus leading to long-term deformation. Incipient concrete damage, which is often in
35 the form of inner micro-cracks, could be produced and amalgamated to induce surface cracks. The
36 cracks often occur in tensile regions, and most cracks are invisible at an incipient state. When the
37 inner cracks propagate onto the PSC members’ surfaces, material degradations could be significant
38 due to environmental erosion. The severe concrete cracks in PSC bridges cause a rising concern for
39 designers, investigators, and managers [1-3]. Thus, the detection of concrete cracks in PSC structures
40 at the early stage is crucial to ensure structural integrity and reduce long-term maintenance costs.
41 Various SHM (structural health monitoring) methods have been proposed for the damage
42 detection of PSC members using vibration characteristics (i.e., natural frequencies, mode shapes, and
43 modal curvatures) [4, 5]. The methods utilize low modal parameters, which are conspicuous

Sensors 2020, 20, x; doi: FOR PEER REVIEW www.mdpi.com/journal/sensors

 Sensors 2020, 20, x FOR PEER REVIEW 2 of 21

44 insensitive to incipient damage. Moreover, several local techniques, such as visual inspections and
45 X-rays, have been implemented for monitoring structural damage of PSC structures [6-8]. Although
46 the methods are recognized in several applications, their sensitivities are inadequate for developing
47 a reliable detection of the incipient damage (e.g., internal concrete damage) or material deteriorations,
48 in addition to their time-consuming, costly, and even unsafe for inspectors. Strain-based methods
49 have also been used to detect structural damage by utilizing the well-established relation between
50 stress-strain. Among various strain sensors, fiber optic sensors have been commonly used for damage
51 monitoring [9, 10] due to the high durability and sensitivity. Nonetheless, the fabrication of sensing
52 cable-embedded fiber optic sensors remains a challenge. Furthermore, the hybrid health monitoring
53 algorithms using the combination of the global dynamic characteristics (e.g., changes in natural
54 frequencies) and local responses (e.g., changes in acoustic emission features) yielded an effective
55 SHM system [11-13]. Lacidogna et al. [11] proposed multi-technique damage detection for a plain
56 notched concrete beam using modal frequencies for alarming damage occurrence and the combining
57 acoustic emission and digital image correlation for estimating damage evaluation. The result shows
58 the effectiveness for evaluating the surface crack development in notched beams.
59 In the past decades, various SHM applications have been successfully implemented using
60 functions of smart materials, such as piezoelectric ceramic. A PZT (e.g., lead zirconate titanate) patch
61 is attached to a monitored structure to acquire electromechanical impedance responses, and then
62 changes in the impedance signals are quantified and used as a damage indication [14-16]. Ai et al. [17]
63 and Narayanan and Subramaniam [18] attached PZT sensors on concrete members to monitor
64 concrete cracks-induced overloading. However, since impedance features of surface-mounted
65 sensors are quite sensitive to environmental changes [19], it requires a complex technique to
66 compensate for temperature effects [20, 21]. To overcome the issue, the PZT-embedded concrete block
67 (so-called smart aggregate [22]) has been employed for monitoring concrete structures. Several
68 applications have been carried out to evaluate the applicability of the smart aggregate to the SHM of
69 concrete members [22-24]. However, the predetermination of the frequency bands for the impedance
70 monitoring of the smart aggregate has not been conducted so far.
71 In PSC bridges, anchorage zones resist particularly high compressive force induced by
72 prestressing strands [25] during steel-strands’ installation and life-cycle structures. Under various
73 operation conditions (e.g., increments of traffic volume and material degradation) and environmental
74 effects (e.g., thermal expansions and natural disaster events), the anchorage zones could be
75 potentially damaged. Incipient damage often occurs in the form of micro-cracks before propagating
76 onto surfaces of the anchorages since anchorage systems are usually embedded in concrete structures
77 [26, 27]. To ensure structural integrity of the anchorage zones, some design methods and new
78 anchorage types have been proposed [28-30]. Moreover, many studies have been conducted on
79 accessing health conditions of prestressing strands using impedance features obtained from PZT
80 sensors [31, 32]. However, the implementing impedance-based technique for detecting internal
81 concrete damage of anchorage zones has not been significantly conducted. The major advantages of
82 the impedance-based method are as follow: (1) PZT sensors are embeddable, which directly receive
83 stress variation in an inspected structure because more stress variations yielded more changes in
84 impedance features [14, 33]; (2) it is a cost-effective method due to using low-cost, various forms and
85 lightweight of PZT sensors, thus enabling the damage localization using an array of PZT sensors [31];
86 and (3) it allows online and real-time monitoring capacity [34, 35].
87 This study presents an impedance-based damage detection method in a concrete anchorage zone
88 via the PZT-embedded smart rock. Firstly, an impedance measurement model is designed for the
89 impedance monitoring of a concrete member using the PZT-embedded smart rock. Secondly, a
90 prototype design of the smart rock embedded with a PZT sensor is proposed to sensitively catch
91 impedance variations induced by concrete damage in an anchorage zone. Thirdly, the performance
92 of smart rock is carried out using numerical analysis and experimental tests. Variations in impedance
93 signals under compressive loading cases are analyzed to predetermine the sensitive frequency band
94 for the impedance monitoring. Lastly, an experiment on an anchorage zone embedded with the smart
95 rocks and surface-mounted PZT sensors is conducted for the impedance measurement under a series
 Sensors 2020, 20, x FOR PEER REVIEW 3 of 21

96 of loading cases. PZT’s impedance variations are quantified to comparatively evaluate the feasibility
97 of smart rock for detecting internal concrete damage in the anchorage zone.

98 2. Theoretical Model of Piezoelectric Sensor-embedded Smart Rock

99 2.1 Impedance Measurement Model of PZT-embedded Smart Rock

100 Figure 1 shows an impedance measurement model proposed for concrete members using a PZT-
101 embedded smart rock. The smart rock is formed by embedding a protected PZT patch (e.g., epoxy)
102 into a small concrete block before casting. Then, the smart rock is positioned into a monitored
103 structure to acquire impedance responses of the coupling interactions between the PZT-smart rock
104 and the host structure (see Figure 1a). When the force F on the host structure changes, it leads to
105 changes in modal parameters (e.g., stiffness and damping ratio) of the protected layer, the concrete
106 block, and the host structure. As shown in Figure 1b, the 3-degrees of freedom (3-DOFs) impedance
107 model represents coupled motions of the protective layer, the concrete block, and the host structure.
108 The parameters m, k, and c symbolize the mass, the damping coefficient, and the spring stiffness,
109 respectively, and the subscripts p, r, and s denote the protective layer, the concrete block, and the
110 concrete structure, respectively.
111
F
PZT
xp

Protective layer
Structure Impedance mp
(ms, ks, cs) V() Analyzer
~ Smart rock
kp cp
PZT
Smart rock
xr
(mr, kr, cr)
mr kr cr
Impedance

Protective layer F F+DF xs

(mp, kp, cp)
Structure

z ms ks cs
x Frequency
112
113 a) Concrete structure embedded with smart rock b) 3-DOF impedance model
114 Figure 1. Concept of smart rock-based impedance measurement for concrete structure.

115 For the 3-DOF impedance model, the governing equation of motion under a harmonic excitation
116 Fp at the PZT-driving point can be formulated:
117
mp 0 0  xp   cp c p 0  xp   k p k p 0   x p   Fp 
0        
 mr 0   xr    c p c p  cr cr   xr   k p k p  kr kr   xr    0  (1)
 0 0 ms   xs   0 cr cr  cs   xs   0 k r kr  ks   xs   0 
118
119 where xp , xp , xp ; xr , xr , xr and xs , xs , xs are the accelerations, the velocities, and the
120 displacements corresponding to the motions of masses mp, mr, and ms, respectively. The coupling
121 structural-mechanical (SM) impedance, Zs of the protected layer, the concrete block, and the host
122 structure is determined [36]:
123
1  K122 
Zs   K11  
K11   K 22  K 23 / K 33  
(2)
j  2

124
125 where the terms Kmn (m,n = 1-3) are the dynamic stiffness coefficients that contain structural
126 parameters of the protective layer, the concrete block, and the monitored structure [36].
127 The electromechanical impedance, Z( ) is a function of the SM impedance, Zs of the smart rock-
128 host structure and that of PZT sensor, Za [37]:
129
 Sensors 2020, 20, x FOR PEER REVIEW 4 of 21

1
 w l  T 1  
Z ( )   j a a ˆ33  d312 Yˆ11E   (3)
 ta  Z a ( ) Z s ( )  1  
130
131 where wa, la and ta are the width, length, and thickness of the piezoelectric patch, respectively; ˆ33
T
is
132 the complex dielectric constant of at zero stress; Z a    Yˆ11E wata /  jla  is the SM impedance of the PZT
133 patch; d31 is the piezoelectric constant in 1-direction at zero stress; Yˆ11E denotes the complex Young's
134 modulus of the PZT patch at the zero electric field.
135 As noted in Eq. (3), the real-part impedance responses Z( ) contains both the SM impedance of
136 the PZT patch and the smart rock and the host structure. The 3-DOF impedance model represents the
137 coupled vibrations of the smart rock and the host structure. When the electric and mechanical
138 properties of the PZT sensor are constant, any external effects (e.g., concrete damage, force change)
139 would exert an effect on its impedance responses.

140 2.2 Mechanism of PZT-embedded Smart Rock under Compression

141 Figure 2 shows stress responses and corresponding impedance signatures of a PZT-embedded
142 smart rock during the strength development of concrete. The coated PZT has no stress acting on the
143 cover layer of PZT, and it is under compression o1 due to the strength development of concrete (see
144 Figure 2a). The smart rock is embedded in a concrete structure before casting, so the smart rock is
145 under compression o2, while the coated PZT is under o1+o2 (see Figure 2b). Previous studies show
146 that the hydration process of a cubic concrete sample embedded with PZT sensors [38, 39] caused
147 rightward shifts in frequencies and decreased in impedance magnitudes. Thus, impedance responses
148 of the PZT-embedded smart rock before applying external loads are illustrated in Figure 2c.
149
Structure
Coated PZT
o2
Impedance responses

Smart rock
Coated PZT o1 Smart rock, SR

SR in structure

Protective layer o1o2

Concrete
Frequency
150
a) PZT-embedded smart rock b) Smart rock embedded in structure c) Impedance responses
151 Figure 2. Behaviors of smart rock under developing strength of concrete.
152
153 Figure 3 shows stress responses and corresponding impedance signals of the PZT-embedded
154 smart rock in the concrete structure under compressive force F. The smart rock’s surfaces are under
155 compression, F along with the longitudinal direction (i.e., z-axis) and under tension F (due to
156 Poisson’s effect, in which  is Poisson’s ratio) for the other surfaces, as seen in Figure 3a. As a result,
157 the compressive stresses on the smart rock (o2+F) and the coated PZT (o1+o2+F) are increased for
158 the z-direction, but they are reduced for the other directions. For concrete members under
159 compression, inner concrete damage (often in the form of cracks) could occur before surface cracks
160 [40, 41]. The behavior of a concrete member can be classified into two states: an intact state (no
161 damage occurrence) and a damage state (inner concrete damage, crack propagation, and failure). In
162 the intact state, impedance signals had rare changes due to the pre-stress acting on the coated PZT
163 (i.e., o1+o2) [24]. When inner concrete cracks occur locally close to the coated PZT sensor, the pre-
164 stress is partially released. It reveals that the boundary conditions surrounding the coated PZT are
165 changed, resulting in rushed changes in the impedance responses of the embedded PZT sensors [24,
166 40]. Figure 3b illustrates impedance responses of the PZT-embedded smart rock in a concrete member
167 under compression.
168
 Sensors 2020, 20, x FOR PEER REVIEW 5 of 21

F
Structure

Damage occurrence
F +o2 F -o2

Impedance responses
force F+dF
o1o2F
under force F

z
Cracks x
169 Frequency
170 a) PZT-embedded smart rock under compression b) Variation of impedance signals
171 Figure 3. Behaviors of smart rock under compression.

172 2.3 Statistical Quantification of Impedance Responses

173 To quantify the changes in impedance signatures via the PZT-embedded smart rock, the root
174 mean square deviation (RMSD) damage metric is commonly [42] used:
175
 N 2
RMSD  Z , Z *      Z * (i )  Z (i )   /  Z (i )
N
2
(4)
 i 1  i 1
176
177 where Z( i) and Z*( i) are the real part of impedance signatures obtained from the intact and loading
178 states of the smart rock at ith frequency, respectively, and N denotes the number of frequency
179 sampling points in the sweep. The RMSD metric, which represents the intensity of random spectra,
180 is useful to quantify the variation of impedance signals in a frequency band.
181 Ideally, the RMSD value is larger than zero if the compression forces increase (i.e., the stress
182 variation); otherwise, it should be zero. However, the statistical index RMSD can be larger than zero
183 due to experimental and computational errors; even there are no changes in the applied forces. The
184 upper control limit (UCL) [43] is established for decision making on the stress variations, as follows:
185
UCL    3 (5)
186
187 where  and  are the mean and the standard deviation of impedance signatures at the intact
188 condition, respectively. The UCL was calculated by 3 corresponds to a confidence level of 99.7%.
189 The calculation of UCL is outlined in three steps: firstly, N impedance signals are measured at the
190 intact condition; secondly, RMSD values for N impedance signals are computed by Eq. (4), then a set
191 of RMSD values is obtained; and finally, UCL of the RMSD values is calculated using Eq. (5). The
192 occurrence of damage (e.g., the change of stress) is indicated when the RMSD values are larger than
193 the UCL. Otherwise, there is no indication of damage occurrence.

194 3. PZT-embedded Smart Rock for Impedance-based Concrete Damage Detection

195 3.1 Prototype Design of PZT-embedded Smart Rock

196 Since most piezoelectric materials (e.g., PZT patch) are fragile, it is commonly covered in the
197 form of the smart aggregate, which is a coated PZT embedded into a small concrete block [22, 44]
198 before positioning into a concrete member to monitor structural performance during concrete curing
199 [45] and service-time [16]. For the impedance-based SHM methods, the monitoring results mainly
200 depend on the selections of effective frequency bands [46, 47]. Meanwhile, the sensitive frequency
201 bands are dependent on material properties and geometric sizes of the smart aggregate’s components
202 [24, 48]. In general, the sensitive frequency band relies on the orientation of sensors in target
203 structures. The predetermination of sensitive frequency ranges could help to minimize the time for
204 the impedance-based monitoring technique in practices and to detect incipient damage in a
205 monitored structure [13, 31].
 Sensors 2020, 20, x FOR PEER REVIEW 6 of 21

206 As shown in Figure 4, a PZT-embedded smart rock was designed for structural health
207 monitoring of concrete structures. A PZT 5A patch had 10101 mm in size, and it was welded with
208 an electric wire for acquiring impedance responses. Also, the PZT patch was covered by 0.5 mm thick
209 epoxy to waterproof and insulate electricity (see Figure 4a). Then, the protected PZT was placed in a
210 small concrete block ( 26 mm and H = 26 mm) before casting (see Figure 4b). The concrete mixture
211 for an anchorage zone was listed in Table 1. As noted in the table, the size of the smart rock is
212 simplified as Dmax of the aggregate. A concrete mixture including cement, sand, and water (without
213 aggregate) was used for the fabrication of smart rocks. Figure 4c shows the samples of PZT-
214 embedded smart rock at 24h after removing frameworks. The smart rocks were used to analyze the
215 realistic performance under compression, as described in Sections 3.2 and 3.3. They were embedded
216 into a concrete anchorage zone (detailed in Section 4) for impedance-based damage monitoring.
217
D = 26 mm
PZT patch Concrete
(10x10x1.0 mm) block
H = 26 mm

Epoxy layer
(0.5 mm)
Electric wire
218
219 a) PZT coated by epoxy b) Smart rock geometry c) Smart rock samples
220 Figure 4. PZT-embedded smart rock for impedance measurement.

221 Table 1. Concrete mixture for anchorage zone*.

Material Mass (kg)
Cement 346
Sand 800
Aggregate (Dmax 25) 997
Water 165

222 (*) A mixture of cement, sand, and water was used for smart rock’s fabrication
223
224 Figure 5 shows the impedance spectra of a naked PZT, a coated PZT (covered by epoxy resin at
225 12h and 20h), and a PZT embedded in concrete (smart rock). This figure shows the information on
226 how the process affects the PZT’s impedance signals. Peak 1’s impedance of the naked PZT was 191.5
227 kHz, and it had the highest real impedance magnitudes. The impedance frequency and its magnitude
228 were reduced when the PZT was coated with the epoxy. This observation was consistent with the
229 previous study [40]. During developing the strength of epoxy and concrete, Peak 1’s impedance
230 reduced in magnitude and increased in frequency. Conversely, impedance signals of the smart rock
231 would be increased when the concrete is damaged, and the constrain condition of the PZT patch is
232 released, thus promising the PZT-embedded smart rock to disclose the occurrence of damage in
233 concrete.
234

Naked PZT
Smart rock

Coated PZT

235
236 Figure 5. Impedance spectra of naked PZT, coated PZT, and PZT embedded in concrete (smart rock)
 Sensors 2020, 20, x FOR PEER REVIEW 7 of 21

237 3.2 Numerical Analysis of Smart Rock-based Impedance Monitoring

238 3.2.1 FE Model of PZT-embedded Smart Rock

239 The sensitivity of impedance responses obtained from a PZT-embedded smart rock could be
240 affected by the orientation of PZT sensors in an inspected structure. To deal with the issue, two finite
241 element (FE) models (Models 1-2) of the smart rock under effects of the axial and lateral compressive
242 force were established in Comsol Multiphysics, as shown in Figure 6a-b, respectively. For the effect
243 of the axial force, a force P was applied on the smart rock’s surface along with the axial direction (i.e.,
244 z-axis) (see Figure 6a). For the effect of the lateral force, the compressive force P was acted on 2.626
245 mm (0.1DH [49]) of the smart rock’s surface (see Figure 6b).
246 The FE models consist of the PZT 5A patch (namely PZT-A for the axial force effect and PZT-L
247 for the lateral force effect), the epoxy layer, and the concrete block. The geometric parameters of each
248 component were previously described in Section 3.1. The material properties of the PZT sensor were
249 listed in Table 2 [14]. Material properties of the concrete block and the epoxy layer [50] were listed in
250 Table 3. The compressive strength of concrete was 23.3 MPa for concrete mixture Table 1 (tested on 
251 15 cm & H = 30 cm). Noticing most concrete has a compressive strength of more than 28 MPa, the
252 compressive strength of the smart rock (see Figure 4c) is a little lower. Since the concrete mix of smart
253 rock was fabricated using the same ratio of cement, sand, and water as the concrete mixture of the
254 anchorage zone, its material properties were assumed to have the same properties of the main
255 structure (see Table 3). For impedance analysis, also, the material properties of the smart rock’s
256 components were assumed as linear elastic materials.
257 For Model 1, the complete meshed FE model of the smart rock consists of 10,972 elements,
258 including 9,964 elements for the concrete block, 752 elements for the epoxy layer, and 256 elements
259 for the PZT. The fixed boundary condition was assigned for the bottom surface of the concrete block
260 (see Figure 6a). For Model 2, the FE model consists of 10,634 elements. The fixed boundary condition
261 was assigned on the opposite surface, as shown in Figure 6b. The quadratic tetrahedral elements were
262 used for the FE models.
263 Seven cases (P1–P7) of the compressive forces were simulated in a series to obtain impedance
264 responses of the PZT-embedded smart rock, as follows: P1 (1.0 kN), P2 (1.5 kN), P3 (2.0 kN), P4 (2.5
265 kN), P5 (3.0 kN), P6 (3.5 kN), and P7 (4.0 kN), corresponding to axial stress change from 0-5.65 MPa
266 (i.e., P1-P7). The forces P1-P7 could cause stress lateral stress changes from 0-3.8 MPa, in which the
267 lateral stress could be computed as 2P/(HD) [49] (with D = 26 mm and H = 26 mm in Figure 6). The
268 axial stress variation was slightly different from the lateral stress under the same applied forces.
269 However, when the smart rock is embedded in the structure, the concrete of the smart rock-host
270 structure is unified. The force P on the host structure causes the same stress variations for a local
271 region close to the PZT sensor, and it is not affected by the shape of smart rocks. To acquire
272 impedance signatures, a 1V harmonic voltage was applied to the top surface of the PZT patch while
273 the bottom surface was assigned as the ground electrode.
274
Uniform load, P
Fixed
Uniform load, P
0.1D
Concrete
13

Uniform load, P

PZT-A PZT-L

Epoxy
(0.5mm thick)
Fixed
z
13

Fixed y
x
275 D = 26 mm D = 26 mm
276 a) Model 1: effect of axial force b) Model 2: effect of lateral force
277 Figure 6. FE models of PZT-embedded smart rock under compression.
 Sensors 2020, 20, x FOR PEER REVIEW 8 of 21

278 Table 2. Material properties of PZT 5A.

Mass Dielectric Coupling constant Young's modulus Damping loss Dielectric loss
density,  constant T33 d31 (m/V) E (GPa) factor factor
(kg/m3) (Farad/m)  d
7750 1.53x10-8 -1.71x10-10 62.1 0.0125 0.015

279 Table 3. Material properties of concrete and epoxy layer.

Properties Concrete Epoxy

Young’s modulus (GPa) 23.6 0.74
Poisson’s ratio 0.2 0.3
Mass density (kg/m ) 3 2300 1090
Tensile strength (MPa) 2.3 20.1
Compressive strength (MPa) 23.3 32.3

280 3.2.2 Numerical Impedance Responses of PZT-embedded Smart Rock

281 Figures 7a-b show impedance responses of PZT-A (under axial force) and PZT-L (under lateral
282 force) in the frequency range of 100-600 kHz under the compressive forces P1-P7. There are three
283 main resonant frequencies, which were about 200 kHz for Peak 1, 245 kHz for Peak 2, and 460 kHz
284 for Peak 3. The resonant impedance frequency peaks were slightly different due to dissimilarity in
285 boundary conditions of the smart rocks in the two FE models. Among the three peaks, Peak 1’s
286 impedance shows the most variation, but changes in impedance peaks were insignificant.
287 It is known that the resonant impedance frequency ranges represent meaningful structural
288 information [14, 51]. The frequency range of 100-250 kHz (including Peak 1’s impedance) was selected
289 to quantify variations in impedance signals induced by the compressive forces. The RMSD indices of
290 PZT-A and PZT-L were computed for the selected frequency range under the forces P1-P7, as
291 depicted in Figure 8. The RMSD indices were linearly increased under increasing the compressive
292 forces, and the PZT-A’s RMSD indices show slightly larger than those of PZT-L.
293
Selected range
Peak 1 (200.3 kHz) Peak 3 (457.2 kHz)

Peak 2 (244.7 kHz)

294
295 a) PZT-A under axial compressive force
Selected range
Peak 1 (201.3 kHz) Peak 3 (458.8 kHz)

Peak 2 (246.9 kHz)

296
297 b) PZT-L under lateral compressive force
298 Figure 7. Numerical impedance responses of PZT-embedded smart rock under compression.
 Sensors 2020, 20, x FOR PEER REVIEW 9 of 21

299
300 Figure 8. RMSD indices of numerical impedance responses of PZT sensors under compression.

301 3.3 Experimental Analysis of Smart Rock-based Impedance Monitoring

302 3.3.1 Experimental Set-up on Compressive Tester

303 To analyze the realistic performance of the PZT-embedded smart rock (in Section 3.2), an
304 experimental test on smart rocks was carried out, as illustrated in Figure 9. A smart rock was placed
305 between two plates of a compressive machine (see Figure 9a). A load cell was used to measure real
306 applied forces. Loading speed was set as 0.1 mm per minute to introduce compressive forces to the
307 smart rock. Four tests (Tests 1-4) on the three smart rocks, namely SR-1, SR-2, and SR-3, were
308 conducted for impedance measurement, as listed in Table 4. The smart rocks SR-1 and SR-2 were
309 tested under axial compressive forces (Test 1) and lateral compressive forces (Test 2). For the smart
310 rock SR-3, first, effects of the axial forces (Test 3) on the smart rock’s impedance responses were
311 conducted. After Test 3, SR-3 was removed and put as free boundary conditions for about two hours.
312 Then, effects of the lateral forces (Test 4) were tested. The geometric parameters and material
313 properties of the smart rocks were described in Section 3.1.
314
Load cell Impedance measurement
system
Smart rock

Temp. measurement system

Impedance analyzer
Kyowa EDX- HIOKI 3532
100A
Connected to
HIOKI 3532

Load Disp.
315
316 a) Set-up of compressive tester b) Impedance and temperature measurement systems
317 Figure 9. Test set-up and instrumentation for impedance measurement of smart rock under
318 compression.

319 For each test (Tests 1-4), the compressive force was increased from 1.0-4.0 kN (P1-P7) with an
320 interval of 0.5 kN, as noted in Table 4. To set the same boundary conditions for the smart rocks, the
321 force was set as 1.0 kN (a baseline). An impedance analyzer (HIOKI 3532) was used to excite (using
322 1V harmonic excitation) and record impedance responses in the range of 100-600 kHz (500 intervals)
323 from the smart rocks. Temperature variation measured via KYOWA (EDX-100A) was less than 1oC
324 during the impedance measurement. So, temperature effects on impedance responses could be
325 ignored in this analysis. Also, to minimize effects of electromagnetic noises on impedance signals
326 induced by flexing, twisting, or transient impacts on a coaxial cable, the cable was placed in such a
327 way that the cable was not affected by anything dynamic during the impedance measurements. Four
328 ensembles of impedance signals were recorded for each case of the compressive force to compute the
329 control threshold (UCL shown in Eq. 5) for making a decision on damage occurrence and to estimate
330 the effects of noises on impedance responses.
 Sensors 2020, 20, x FOR PEER REVIEW 10 of 21

331 Table 4. List of tests for smart rock’s performance under compression.

Force direction
Test Name Applied load
Axial Lateral
1 SR-1 o -
P1 (1.0 kN)  P7 (4.0 kN)
2 SR-2 - o
0.5 kN increment
3 o -
SR-3 in a row.
4 - o
332

333 3.3.2 Experimental Impedance Responses of PZT-embedded Smart Rock under Compression
334 Figures 10a-b, respectively, show measured impedance spectra (in the frequency range of 100-
335 600 kHz) of SR-1 and SR-2 under seven loading cases (i.e., P1-P7). The three resonant impedance
336 signals (i.e., three peaks in the figure) were measured as follows: about 198 kHz for Peak 1, 250 kHz
337 for Peak 2, and 465 kHz for Peak 3. The impedance responses of SR-3 under the axial and lateral
338 compressive forces were zoomed in the frequency range of 150-250 kHz, as seen in Figure 10c. In
339 intact case (P1), Peak 1’s impedance of smart rock SR-3 was 182.5 kHz in Test 3 (under the effect of
340 axial load) and 182.7 kHz in Test 4 (under the effect of lateral force). The impedance frequency had
341 ignorable change, thus indicating that the smart rock would not be damaged before Test 4. Peak 1’s
342 impedance signatures show high real-impedance responses and the most variation under
343 compressive force among the three impedance peaks. Thus, the frequency range of 150-250 kHz was
344 chosen to quantify impedance responses induced by the force variations.
345

Selected range

Peak 1
(194.5 kHz)

Peak 2
Peak 3
(242.1 kHz)
(470.3 kHz)

346
347 a) SR-1 under axial compressive force

Selected range

Peak 1
(199.2 kHz)

Peak 2 Peak 3
(252.4 kHz) (465.6 kHz)

348
349 b) SR-2 under lateral compressive force
 Sensors 2020, 20, x FOR PEER REVIEW 11 of 21

Axial force Lateral force

Peak 1 Peak 1
(182.5 kHz) (182.7 kHz)

350
351 c) SR-3 under axial- and lateral compressive forces
352 Figure 10. Experimental impedance responses of smart rocks under compression.

353 Figure 11 shows the RMSD indices of impedance signals of SR-1, SR-2, and SR-3 (computed for
354 the frequency range 150-250 kHz) under loading cases P1-P7. The control threshold UCLs were
355 established using measured signals at the force P1 (the baseline signal). The error bars calculated
356 from 4 ensembles of impedance signals were also depicted. As observed in the figure, the magnitudes
357 of RMSD were ignorable in the intact case, but they are beyond the UCLs for the loading cases P2-P7,
358 thus suggesting that the variation of compressive forces were successfully alarmed. Standard
359 deviations of impedance signals were comparatively small (see error bars in the figures), thus
360 suggesting a low dispersion of the data. In other words, the electromagnetic effects on impedance
361 features could be ignored. Moreover, the RMSD indices of SR-1 versus SR-3 (under axial force) and
362 SR-2 versus SR-3 (under lateral force) were linearly increased and comparable to each other. The
363 result demonstrated the feasibility of the PZT-embedded smart rock by consistently measuring
364 impedance features.
365

UCL UCL

366
367 a) SR-1 under axial compressive force b) SR-2 under lateral compressive force
Axial force Lateral force

RMSD s mean
Error bar

UCL UCL

368
369 c) SR-3 under axial and lateral compressive force
370 Figure 11. RMSD indices of experimental impedance responses from smart rocks under
371 compression.

372 3.4 Sensitivity of Impedance Responses via PZT-embedded Smart Rock to Stress Variation
373 Figure 12 shows a comparison of numerical and experimental impedance frequencies for Peaks
374 1-3 under axial and lateral compressive forces. The differences in impedance frequencies relatively
 Sensors 2020, 20, x FOR PEER REVIEW 12 of 21

375 small (less than 3% for SR-1 and SR-2). Meanwhile, Peak 1’s frequency of SR-3 was about 7% different
376 from those of SR-1, SR-2, or the numerical result. Also, the impedance spectra of SR-3 had a broader
377 bandwidth than that of the other smart rocks. The differences could come from an inaccurate
378 thickness of the epoxy layer among the coated PZT sensors. The previous study shows that the
379 thickness of the adhesive layer had significant impedance responses [48, 52]. Despite that, the FE tool
380 was applicable to analyze impedance responses of the PZT-embedded smart rock.
381 Figure 13 shows a comparison of the rates of the RMSD’s numerical (see Figure 8) and
382 experimental (see Figure 11) impedance for the effects of axial and lateral forces. The RMSD index at
383 the force P7 was selected as the reference for the computation. The relations between the applied
384 forces and RMSD indices were well-matched for the smart rocks under the axial forces (see Figure
385 13a) and the lateral forces (Figure 13b) except for some errors at SR-2 (under the lateral force).
386 Furthermore, there were inconsistencies in magnitudes of RSMD indices between the numerical (see
387 Figure 8) and experimental analyses (see Figure 13). The reasons could be: (1) the effect of the pre-
388 stress o1 due to the hydration process of concrete acting on the coated PZT (see Figure 2a) was not
389 be considered in the FE analysis, and (2) the nature of concrete and the epoxy layer could not be linear
390 materials under compressive force.
391 From the analyses, the following observations were made: (1) there were slight differences in
392 impedance features under the loading-direction effects; (2) the impedance feature had repeatability
393 and was gradually increased under compressive forces; and (3) the predetermined frequency band
394 for the smart rock was 100-600 kHz, in which the sensitive frequency band was 150-250 kHz.
395

396
397 a) Under axial compressive force b) Under lateral compressive force
398 Figure 12. Comparison of numerical and experiment impedance frequencies of smart rock.

399
400 a) Under axial compressive force b) Under lateral compressive force
401 Figure 13. Comparison of rate of RMSD indices of PZT-embedded smart rock.

402 4. Feasibility Evaluation of PZT-embedded Smart Rock for Concrete Damage Detection in
403 Prestressed Anchorage Zone

404 4.1 Test Set-up and Test Scenarios

405 4.1.1 Experimental Set-up on Anchorage Zone

406 A lab-scale experiment on a concrete anchorage zone equipped with smart rocks was conducted
407 to evaluate the feasibility of the PZT-embedded smart rock for the health monitoring of the anchorage
408 zone, as schematized in Figure 14. The anchorage zone includes a concrete block having a size of
 Sensors 2020, 20, x FOR PEER REVIEW 13 of 21

409 252532 cm, and four steel-blocks (3.83.83.8 cm), which play a role as a bearing plate [53] to
410 transfer compressive forces to the structure. The steel blocks were evenly distributed on the concrete
411 surface, and its distance to the center of the anchorage was 6.5 cm (see Figure 14b). A cylinder (110
412 mm made by PVC, Polyvinyl chloride) plays a role as a duct for cables (see Figure 14a). For the
413 feasibility evaluation of the smart rock, the anchorage zone was designed without reinforcement. The
414 material properties of concrete were listed in Table 3 with the compressive strength c = 23.3 MPa.
415 Two smart rocks, namely SR-A (on the left side) and SR-B (on the right side), were embedded in
416 the concrete anchorage (3.5 cm from the concrete surface) for the impedance monitoring, detailed in
417 Figure 14a-c. The geometries and material properties of the smart rocks were presented in Section 3.
418 Also, two surface-mounted PZT sensors, namely PZT-S1 (close to SR-A) and PZT-S2 (close to SR-B)
419 were utilized for impedance measurement to compare their sensitivities with the embedded sensors.
420 The surface-mounted PZT sensors had a size of 10x10x0.5 mm, and they were attached via aluminum
421 interfaces (10x10x0.5 mm) before mounted onto the anchorage’s surface. The use of the aluminum
422 interface could enhance the sensitivity of impedance responses [54]. PZT5A also was selected for the
423 surface-mounted PZTs sensors with materials properties, as shown in Table 2.
424 Figure 14c shows the set-up of the anchorage zone on a supporting steel frame. The steel frame
425 with a capacity of 300 kN was designed to resist tension forces introduced by the anchorage zone.
426 Two thick steel-plates (steel plates 1-2) were connected by four steel-tubes using bolted connections.
427 The compressive force was introduced to the anchorage zone using a hydraulic jack, and a load cell
428 was used to record real applied forces.
429
P
Structure Orientation of PZT
A A Aluminum D
3.5

PZT PZT
Load cell

H
Steel block Steel plate
3.8x3.8x3.8 cm 3.5
Right surface
PZT-S1 PZT-S2
3.5

Smart rock A Smart rock B PZT-S2

(SR-A) (SR-B)
6.5

11 PZT-S1 3.5
PVC duct
SR-A SR-B
Concrete block
25x25x32 cm 25 Left surface
z
x Steel plate
430 Unit: cm 25
431 a) Side view b) View A-A
432
Bolts

Steel plate 1
Hydraulic jack

Tube Embedded PZT Load cell

(SR-A)
Bearing plate 20x20 cm
Steel block 3.8x3.8 cm

Right surface

Surface-mounted PZT
(PZT-S1)

Left surface
Steel plate 2
433 Concrete: 25x25x32 cm
434 c) Set-up anchorage on supporting steel frame
435 Figure 14. Experimental set-up in anchorage zone for impedance measurement.
436
 Sensors 2020, 20, x FOR PEER REVIEW 14 of 21

437 4.1.2 Test Scenarios for Impedance Measurement on Anchorage Zone

438 Table 5 shows 11 loading cases (F1-F11) simulated for the impedance measurement in the
439 anchorage zone. At first, the compressive force was set by a force F1 = 100 kN, as a baseline. For the
440 first six loading cases (i.e., F1-F6), the applied forces were gradually increased from 100-200 kN with
441 an increment of 20 kN for each case. For the next three loading cases (i.e., F6-F9), the applied forces
442 were progressively raised from 200-230 kN with an interval of 10 kN. Following that, the force was
443 increased to F10 (235 kN). The concrete was partially collapsed under the force F10, and then the
444 compressive force was reduced to F11 (102 kN).
445 As noted in Table 5, the bearing stress b was varied from 17.3-40.7 MPa (i.e., 0.74-1.74c) under
446 the forces F1-F11. It is well-known that the bearing stress in the anchorage zone (with reinforcement)
447 allows from 1.25-3.0ci [41, 55, 56], in which ci (0.85c) is the compressive strength of concrete at
448 steel strand’s installation. In the experiment, the concrete surface crack (near SR-A) occurred under
449 the bearing stress b = 1.71c (i.e., under the force F9), and the anchorage failure occurred under b =
450 1.74c (under the force F10). The observation of the concrete crack development during the loading
451 procedure was summarized in Table 5, and also illustrated in Figure 15. There was no crack when
452 the force was lower than 220 kN (i.e., the force F8), as seen in Figures 15a1-15b1. When the force was
453 increased to F9, an incipient crack was found only on the left surface of the anchorage zone (close to
454 SR-A), seen in Figures 15a2-15b2. When the force was increased to F10, cracks occurred and
455 propagated (mostly surrounding SR-A), and then the concrete was failed first in a region close to SR-
456 A (see Figure 15a3). As a result, the larger cracks occurred (close to SR-B, see Figure 15b3), and the
457 concrete was failed after the other side damaged. A Dazor 8MC-100 magnifier was used to check the
458 crack occurrence during the loading cases.

459 Table 5. Simulation cases of lab-scale anchorage zone for impedance monitoring.

Applied force
Concrete surface’s
Case Bearing stress
F (kN) crack development
b (MPa)**

F1 F6 100200 17.3 34.6

- Surface near SR-A: no crack (see Figure 15a1).
F7 210 36.4
- Surface near SR-B: no crack (see Figure 15b1).
F8 220 38.1

- Surface near SR-A: surface crack initiation,

F9 230 39.8 crack size 0.230 mm (see Figure 15a2).
- Surface near SR-B: no crack (see Figure 15b2).

F10 235* 40.7 - Surface near SR-A: cracks propagation,

concrete failure occurred (see Figure 15a3).
F11 102 17.7 - Surface near SR-B: large crack occurred,
failure after the other side (see Figure 15b3).
460 (*) Impedance signals of SR-A, SR-B, and PZT-S1 were measured during crack propagation on the left surface of concrete
461 (**) Stress ahead of steel blocks (bearing plate) is so-called bearing stress.
462
463 An impedance analyzer HIOKI 3532 was used to generate (1V harmonic voltage) and measure
464 impedance signatures of PZT sensors in the frequency range of 100-600 kHz (501 points). For each
465 test case, four ensembles of impedance signals were recorded. As noted in Table 5, impedance signals
466 of SR-A, SR-B, and PZT-S1 were measured during crack propagation under the force F10. Meanwhile,
467 PZT-S2 was not measured due to the sudden collapse of the anchorage zone, which came from the
468 lack of reinforcement. During the impedance measurement, the room temperature was controlled at
469 about 22oC by utilizing air conditioners to minimize the influence of temperature variations on the
470 impedance features.
471
 Sensors 2020, 20, x FOR PEER REVIEW 15 of 21
a1) No surface crack at F8 a2) Surface crack at F9 a3) Concrete failure at F10F11

a1) No surface crack at F8 a2) Surface crack at F9 a3) Concrete failure at F10F11

Crack
(~0.2x30 mm)
PZT-S1 Crack Concrete damage
(~0.2x30 mm)
PZT-S1 Concrete damage
472
473 a) Left side of anchorage zone (near SR-A)
b1) No surface crack at F8 b2) No surface crack at F9 b3) Concrete sudden failure at F10F11

b1) No surface crack at F8 b2) No surface crack at F9 b3) Concrete sudden failure at F10F11

PZT-S2 Concrete damage

474
475 b) Right side of anchorage zone (near SR-B)
PZT-S2 No crack occurrence Concrete damage
476 Figure 15. Crack development on concrete surface of anchorage zone during loading process.

477 4.2 Damage Monitoring Result on Anchorage Zone using Smart Rock

478 4.2.1 Impedance Responses of PZT-embedded Smart Rocks and Surface-mounted PZTs
479 Figure 16 shows impedance responses of the smart rocks SR-A and SR-B (in the frequency range
480 100-600 kHz) under the compressive forces F1-F11. The three resonant impedance signals (i.e., three
481 peaks in the figure) were measured as follows: about 210 kHz for Peak 1, 265 kHz for Peak 2, and 470
482 kHz for Peak 3. Among these, Peak 1’s impedance signals show more sensitive to force variations.
483 For SR-A (see Figure 16a), the impedance signals were insignificantly varied under the forces
484 F1-F6. The impedance signals were suddenly changed under the force F7, thus revealing the
485 transformation of the concrete medium around SR-A (e.g., inner concrete damaged [40]). The
486 impedance responses were continuously altered under the force F8, and it had the other abrupt
487 change under the force F9 (incipient surface crack close to SR-A). The impedance signals kept varied
488 under the force F10 (crack propagation), and it had the most variation under the force F11 (concrete
489 failure). For SR-B (see Figure 16b), impedance signatures had insignificant variations under the forces
490 F1-F10. The substantial alteration only occurred as the anchorage zone was collapsed (under force
491 F11).
492 Figure 17 shows the impedance responses of the surface-mounted PZT-S1 and PZT-S2 (in the
493 frequency range of 100-600 kHz) under the eleven loading cases. There were several resonant peaks
494 in the examined ranges, and Peak 1’s impedance also shows the most variation. For PZT-S1 (see
495 Figure 17a), impedance signals were insignificant changes under the forces F1-F8. The impedance
496 signals were suddenly changed under the force F9. The signals keep changing under the forces F10-
497 F11. For PZT-S2 (see Figure 17b), impedance signatures show ignorable changes under the forces F1-
498 F10, and it had rushed change when the anchorage zone was partially damaged (the force F11).
499 By matching the impedance signatures of the embedded-PZT sensors and the surface-mounted
500 PZT sensors to crack appearance, the observations were made as follows. First, the crack occurrence
501 caused rushed changes in the impedance responses, that is, SR-A and PZT-S1’s impedance were
502 significant changes, but those of SR-B and PZT-S2 were ignorable changes (under the force F9).
503 Second, the incipient crack and damage in concrete, which early occurred on the left side of the
504 anchorage zone, could come from the unequal applied force (via four steel blocks) and the nature of
505 concrete structure (i.e., an inelastic material). In Figure 16a, a clear peak frequency shift is observed,
506 so does the bandwidth.
507
508
 Sensors 2020, 20, x FOR PEER REVIEW 16 of 21

Peak 1 Failure Crack propagation

Surface crack
Sudden change

Peak 3
Peak 2

Selected range

509
510 a) SR-A

Failure
Peak 1

Peak 3
Peak 2

Selected range

511
512 b) SR-B
513 Figure 16. Impedance responses of PZT-embedded smart rocks in anchorage zone.
514

Crack Surface crack

propagation
Peak 1

Failure

Selected range

515
516 a) PZT-S1

Failure

Peak 1

Selected range

517
518 b) PZT-S2
519 Figure 17. Impedance responses of surface-bonded PZT sensors on anchorage zone.
 Sensors 2020, 20, x FOR PEER REVIEW 17 of 21

520 4.2.2 Damage Monitoring in Anchorage Zone using PZT’s Impedance Features
521 As observed in Figures 16-17, Peak 1’s impedance signals show the most variations, so the
522 frequency range of 100-250 kHz was selected for the impedance-based damage detection. The RMSD
523 index was used to quantify impedance changes, and the control threshold UCL was utilized for
524 decision-making. Figure 18 shows the RSMD indices of SR-A and SR-B calculated for 11 loading cases.
525 Also, the error bars calculated from four ensembles of the impedance signals were also depicted. The
526 RSMD magnitudes were ignorable (0.2%) under the intact case, and they were gradually increased
527 with respect to the loading cases. Standard deviations of impedance signals were comparatively small
528 (see error bar in the figures), thus suggesting a low dispersion of the data.
529 For SR-A (see Figure 18a), the RMSD indices were insignificant for the forces F1-F6 (i.e., 0.2-
530 4.3%). Under the force F7, the RMSD index (i.e., 7.9%) became significant (about twice times larger
531 than that under the force F6), thus suggesting sudden changes in the concrete surrounding SR-A.
532 Under force F9, the RMSD index (29.5%) was about three times the one under the force F8. The change
533 could be caused by inner damage developed to incipient the surface crack. Then, the RMSD value
534 was the maximum (i.e., 117.0%) under the force F11. For SR-B (see Figure 18b), the RMSD values were
535 insignificantly small (0.2-3.2%) under the forces F1-F10. Under the force F11, the RMSD index (i.e.,
536 19.8%) became six times higher than those ones under the force F10. The result again confirmed the
537 variations of PZT-embedded smart rocks’ impedance signatures were induced by concrete damage.
538 Figure 19 shows the RSMD indices of PZT-S1 and PZT-S2 computed for 11 loading cases. The
539 RSMD magnitudes were ignorable (i.e., 0.2%) under the intact case. For PZT-S1 (see Figure 19a), the
540 RMSD values (0.2-8.7%) were gradually increased from under the forces F1-F10, and its value (i.e.,
541 6.1%) reduced under the force F11. For PZT-S2 (see Figure 19b), the RMSD values were stably small
542 (<2.0%) under the forces F1-F9, and the RMSD index was a significant value (15.0%) when the
543 concrete was damaged.
544
Stage 1: intact state Stage 2: damage state

Failure

Crack propagation Failure

Surface crack
Sudden change
UCL UCL
(Inner crack)

545
546 a) SR-A b) SR-B
547 Figure 18. RMSD indices of impedance responses of PZT-embedded smart rock in anchorage zone.
Stage 1: intact state Stage 2: damage state

Failure
Crack propagation
Failure
Surface crack
UCL UCL

548
549 a) PZT-S1 b) PZT-S2
550 Figure 19. RMSD indices of impedance responses of surface-bonded PZT sensors on anchorage zone.

551 4.2.3 Discussion on Damage Monitoring on Anchorage Zone using Smart Rock
552 When the concrete was damaged, impedance responses were progressively increased in
553 magnitudes and reduced in the impedance frequencies (see Figure 16a), thus evidencing the
554 mechanism of the PZT-embedded smart rock (as described in Section 2). The rushed changes in
 Sensors 2020, 20, x FOR PEER REVIEW 18 of 21

555 impedance signatures revealed that the concrete domain was damaged. Thus, the anchorage zone
556 was classified into the two states (see Figure 18a), including an intact state (for the force lower than
557 F6) and a damage state starting with the incipient damage (under the force F7), the damage
558 propagation (under the force F8), the initial surface crack (under the force F9), the concrete failure
559 (under the force F10).
560 The impedance features of the PZT-embedded smart rock yielded a better damage indicator than
561 those of the surface-mounted PZT sensor. This is because the PZT-embedded smart rocks directly
562 received stress variations from the anchorage zone rather than the surface-mounted sensors. Previous
563 studies [33, 57] demonstrated that the more stress changes yielded, the more variations in impedance
564 responses. As a result, the embedded PZT sensors yielded higher sensitivities in impedance
565 responses. Furthermore, it is known that PZT’s impedance responses were insensitive to compressive
566 forces [33, 58], and the variations in impedance features came from changes in boundary conditions
567 [33] or stiffness reduction [58]. In this study, the impedance responses of the surface-mounted sensors
568 were also relatively less sensitive when the compressive force lower than F6 (intact case, see Figure
569 19a). The impedance signals of PZT-S1 (close to SR-A) show relatively higher sensitivity than those
570 of PZT-S2. It could come from unequal applied forces (via the four steel blocks). When the forces
571 increased (larger than F6), it could cause changes in boundary conditions (e.g., concrete damage)
572 surrounding the sensors, thus leading the differences in their impedance features.

573 5. Concluding Remarks

574 In this paper, the impedance-based damage detection in the anchorage zone using the
575 impedance features via the PZT-embedded smart rock was presented. Firstly, the impedance
576 measurement model was designed for impedance monitoring of the concrete member using the PZT-
577 embedded smart rock. Secondly, the smart rock embedded with the PZT sensor was designed to
578 sensitively catch impedance variations induced by concrete damage in the anchorage zone. Thirdly,
579 the numerical analysis and the experimental tests were performed to predetermine the sensitive
580 frequency band for impedance monitoring. Lastly, the experiment on the anchorage zone embedded
581 with the smart rocks and the surface-mounted PZT sensors was conducted for impedance monitoring
582 under a series of the loading cases to evaluate the feasibility of the smart rock for detecting internal
583 concrete damage.
584 From the numerical and experimental analyses, the following remarks can be drawn: (1) the
585 sensitive frequency range of the smart rock was 150-250 kHz; (2) the impedance features had
586 ignorable changes under the intact state due to the pre-stress on the smart rock during concrete
587 strength development; and (3) the inner concrete damage was successfully detected using impedance
588 features via PZT-embedded smart rock.
589 The experimental results showed the promising applicability of the smart rock for the detection
590 of incipient concrete damage in the prestressed anchorage zone. For example, smart rocks are
591 embedded during construction, and their impedance responses are measured and utilized as a
592 baseline. The measured signals overtime are compared to baseline signals to determine the health
593 monitoring of a monitored structure. As a future study, a real-scale experiment on a prestressed
594 anchorage zone should be conducted to evaluate the practicality of the proposed smart rock. Also,
595 recent advanced techniques based on machine learning should be employed to analyze multiple
596 parameters (e.g., peak frequency, RMSD, and bandwidth) for damage prediction.
597 Author Contributions: Q.Q Pham and N.L Dang developed the methodology, Q.Q Pham performed the
598 simulation, and prepared the manuscript, N.L Dang and J.T Kim revised the manuscript. J.T Kim supervised the
599 whole work. All authors read and approved the final manuscript.

600 Funding: This work was supported by Basic Science Research Program through the National Research
601 Foundation of Korea (NRF 2019R1A2B5B01069719). The post-doctoral researcher and graduate students were
602 also supported by the 4th BK21 program of Korean Government.

603 Conflicts of Interest: The authors declare no conflict of interest.

604
 Sensors 2020, 20, x FOR PEER REVIEW 19 of 21

605 References
606 1. Mehrabi, A.B., Ligozio, C.A., Ciolko, A.T., and Wyatt, S.T. (2010), "Evaluation, rehabilitation planning, and
607 stay-cable replacement design for the hale boggs bridge in Luling, Louisiana," Journal of Bridge Engineering,
608 vol. 15(4), pp. 364-372.
609 2. Hou, D.W., Zhao, J.L., Shen, J.S.L., and Chen, J. (2017), "Investigation and improvement of strut-and-tie
610 model for design of end anchorage zone in post-tensioned concrete structure," Construction and Building
611 Materials, vol. 136, pp. 482-494.
612 1. Mehrabi, A.B., Ligozio, C.A., Ciolko, A.T., and Wyatt, S.T. (2010), "Evaluation, rehabilitation planning, and
613 stay-cable replacement design for the hale boggs bridge in Luling, Louisiana," Journal of Bridge Engineering,
614 vol. 15(4), pp. 364-372.
615 2. Hou, D.W., Zhao, J.L., Shen, J.S.L., and Chen, J. (2017), "Investigation and improvement of strut-and-tie
616 model for design of end anchorage zone in post-tensioned concrete structure," Construction and Building
617 Materials, vol. 136, pp. 482-494.
618 3. Liu, Y., Cho, S., Spencer, B.F., Jr., and Fan, J. (2014), "Automated assessment of cracks on concrete surfaces
619 using adaptive digital image processing," Smart Structures and Systems, vol. 14(4), pp. 719-741.
620 4. Kim, J.T., Park, J.H., and Lee, B.J. (2007), "Vibration-based damage monitoring in model plate-girder
621 bridges under uncertain temperature conditions," Engineering Structures, vol. 29(7), pp. 1354-1365.
622 5. Ho, D.D., Kim, J.T., Stubbs, N., and Park, W.S. (2012), "Prestress-force estimation in PSC girder using modal
623 parameters and system identification," Advances in Structural Engineering, vol. 15(6), pp. 997-1012.
624 6. Lee, S., and Kalos, N. (2014), "Non-destructive testing methods in the US for bridge inspection and
625 maintenance," KSCE Journal of civil engineering, vol. 18(5), pp. 1322-1331.
626 7. Grosse, C.U. (2009), "Acoustic emission localization methods for large structures based on beam forming
627 and array techniques," Proceedings of the NDTCE, vol. 9.
628 8. Aryan, P., Sampath, S., and Sohn, H. (2018), "An Overview of Non-Destructive Testing Methods for
629 Integrated Circuit Packaging Inspection," Sensors (Basel), vol. 18(7).
630 9. Henault, J.M., Quiertant, M., Delepine-Lesoille, S., Salin, J., Moreau, G., Taillade, F., et al. (2012),
631 "Quantitative strain measurement and crack detection in RC structures using a truly distributed fiber optic
632 sensing system," Construction and Building Materials, vol. 37, pp. 916-923.
633 10. Du, C., Dutta, S., Kurup, P., Yu, T., and Wang, X. (2020), "A review of railway infrastructure monitoring
634 using fiber optic sensors," Sensors and Actuators A: Physical, vol. 303.
635 11. Lacidogna, G., Piana, G., Accornero, F., and Carpinteri, A. (2020), "Multi-technique damage monitoring of
636 concrete beams: Acoustic Emission, Digital Image Correlation, Dynamic Identification," Construction and
637 Building Materials, vol. 242.
638 12. Lacidogna, G., Piana, G., and Carpinteri, A. (2017), "Acoustic Emission and Modal Frequency Variation in
639 Concrete Specimens under Four-Point Bending," Applied Sciences, vol. 7(4).
640 13. Kim, J.T., Park, J.H., Hong, D.S., and Park, W.S. (2010), "Hybrid health monitoring of prestressed concrete
641 girder bridges by sequential vibration-impedance approaches," Engineering Structures, vol. 32(1), pp. 115-128.
642 14. Dang, N.L., Huynh, T.C., and Kim, J.T. (2019), "Local Strand-Breakage Detection in Multi-Strand Anchorage
643 System Using an Impedance-Based Stress Monitoring Method-Feasibility Study," Sensors (Basel), vol. 19(5).
644 15. Yang, Y., Hu, Y., and Lu, Y. (2008), "Sensitivity of PZT impedance sensors for damage detection of concrete
645 structures," Sensors, vol. 8(1), pp. 327-346.
646 16. Negi, P., Chhabra, R., Kaur, N., and Bhalla, S. (2019), "Health monitoring of reinforced concrete structures
647 under impact using multiple piezo-based configurations," Construction and Building Materials, vol. 222, pp.
648 371-389.
649 17. Ai, D., Luo, H., Wang, C., and Zhu, H. (2018), "Monitoring of the load-induced RC beam structural
650 tension/compression stress and damage using piezoelectric transducers," Engineering Structures, vol. 154, pp.
651 38-51.
652 18. Narayanan, A., and Subramaniam, K.V.L. (2016), "Experimental evaluation of load-induced damage in
653 concrete from distributed microcracks to localized cracking on electro-mechanical impedance response of
654 bonded PZT," Construction and Building Materials, vol. 105, pp. 536-544.
655 19. Baptista, F.G., Budoya, D.E., De Almeida, V.A., and Ulson, J.A. (2014), "An experimental study on the effect
656 of temperature on piezoelectric sensors for impedance-based structural health monitoring," Sensors (Basel), vol.
657 14(1), pp. 1208-27.
658 20. Huynh, T.C., and Kim, J.T. (2018), "RBFN-based temperature compensation method for impedance
659 monitoring in prestressed tendon anchorage," Structural Control and Health Monitoring, vol. 25(6).
660 21. Gianesini, B.M., Cortez, N.E., Antunes, R.A., and Vieira Filho, J. (2020), "Method for removing temperature
661 effect in impedance-based structural health monitoring systems using polynomial regression," Structural Health
662 Monitoring, vol.
 Sensors 2020, 20, x FOR PEER REVIEW 20 of 21

663 22. Gu, H., Song, G., Dhonde, H., Mo, Y.L., and Yan, S. (2006), "Concrete early-age strength monitoring using
664 embedded piezoelectric transducers," Smart Materials and Structures, vol. 15(6), pp. 1837-1845.
665 23. Feng, Q., Liang, Y., and Song, G. (2019), "Real-Time Monitoring of Early-Age Concrete Strength Using
666 Piezoceramic-Based Smart Aggregates," Journal of Aerospace Engineering, vol. 32(1).
667 24. Saravanan, T.J., Balamonica, K., Priya, C.B., Reddy, A.L., and Gopalakrishnan, N. (2015), "Comparative
668 performance of various smart aggregates during strength gain and damage states of concrete," Smart Materials
669 and Structures, vol. 24(8).
670 25. Breen, J.E. (1994), Anchorage zone reinforcement for post-tensioned concrete girders. Transportation
671 Research Board.
672 26. Theryo, T.S. (2019), Anchorage zone design and detailing from practical perspective, FDOT Transportation
673 Symposium vol.
674 27. Mao, W., Gou, H., He, Y., and Pu, Q. (2018), "Local Stress Behavior of Post-Tensioned Prestressed
675 Anchorage Zones in Continuous Rigid Frame Arch Railway Bridge," Applied Sciences, vol. 8(10).
676 28. Hanoon, A.N., Jaafar, M.S., Hejazi, F., and Aziz, F.N.a.A. (2017), "Strut-and-tie model for externally bonded
677 CFRP-strengthened reinforced concrete deep beams based on particle swarm optimization algorithm: CFRP
678 debonding and rupture," Construction and Building Materials, vol. 147, pp. 428-447.
679 29. He, Z.Q., and Liu, Z. (2010), "Optimal three-dimensional strut-and-tie models for anchorage diaphragms
680 in externally prestressed bridges," Engineering Structures, vol. 32(8), pp. 2057-2064.
681 30. Wu, J., Xian, G., and Li, H. (2018), "A novel anchorage system for CFRP cable: Experimental and numerical
682 investigation," Composite Structures, vol. 194, pp. 555-563.
683 31. Dang, N.L., Pham, Q.Q., and Kim, J.T. (2021), "Piezoelectric-based Hoop-Type Interface for Impedance
684 Monitoring of Local Strand Breakage in Prestressed Multi-Strands Anchorage," Structural Control and Health
685 Monitoring, vol. 28(1), pp. 1-20.
686 32. Min, J., Yun, C.B., and Hong, J.W. (2016), "An electromechanical impedance-based method for tensile force
687 estimation and damage diagnosis of post-tensioning systems," Smart Structures and Systems, vol. 17(1), pp. 107-
688 122.
689 33. Lim, Y.Y., and Soh, C.K. (2012), "Effect of varying axial load under fixed boundary condition on admittance
690 signatures of electromechanical impedance technique," Journal of Intelligent Material Systems and Structures,
691 vol. 23(7), pp. 815-826.
692 34. Kim, J.T., Nguyen, K.D., and Huynh, T.C. (2013), "Wireless health monitoring of stay cable using
693 piezoelectric strain response and smart skin technique," Smart Structures and Systems, vol. 12(3_4), pp. 381-397.
694 35. Park, S., Shin, H.H., and Yun, C.B. (2009), "Wireless impedance sensor nodes for functions of structural
695 damage identification and sensor self-diagnosis," Smart Materials and Structures, vol. 18(5).
696 36. Huynh, T.C., Nguyen, T.D., Ho, D.D., Dang, N.L., and Kim, J.T. (2020), "Sensor Fault Diagnosis for
697 Impedance Monitoring Using a Piezoelectric-Based Smart Interface Technique," Sensors (Basel), vol. 20(2).
698 37. Liang, C., Sun, F.P., and Rogers, C.A. (1994), "Coupled Electro-Mechanical Analysis of Adaptive Material
699 Systems-Determination of the Actuator Power Consumption and System Energy Transfer," Journal of Intelligent
700 Material Systems and Structures, vol. 5(1), pp. 12-20.
701 38. Kundu, T., Narayanan, A., and Subramaniam, K.V.L. (2016), "Early age monitoring of cement mortar using
702 embedded piezoelectric sensors" Proceedings of Health Monitoring of Structural and Biological Systems 2016,
703 39. Wang, D., and Zhu, H. (2011), "Monitoring of the strength gain of concrete using embedded PZT impedance
704 transducer," Construction and Building Materials, vol. 25(9), pp. 3703-3708.
705 40. Kocherla, A., and Subramaniam, K.V.L. (2020), "Embedded smart PZT-based sensor for internal damage
706 detection in concrete under applied compression," Measurement, vol. 163.
707 41. Nawy, E.G. (2010), Prestressed Concrete: A Fundamental Approach. Prentice Hall.
708 42. Giurgiutiu, V., and Rogers, C.A. (1998), Recent advancements in the electromechanical (E/M) impedance
709 method for structural health monitoring and NDE, Smart Structures and Materials 1998: Smart Structures and
710 Integrated Systems, International Society for Optics and Photonics vol. 3329, pp. 536-547.
711 43. Huynh, T.C., Lee, K.S., and Kim, J.T. (2015), "Local dynamic characteristics of PZT impedance interface on
712 tendon anchorage under prestress force variation," Smart Structures and Systems, vol. 15(2), pp. 375-393.
713 44. Song, G., Gu, H., and Mo, Y.L. (2008), "Smart aggregates: multi-functional sensors for concrete structures—
714 a tutorial and a review," Smart Materials and Structures, vol. 17(3), pp. 1-17.
715 45. Negi, P., Chakraborty, T., Kaur, N., and Bhalla, S. (2018), "Investigations on effectiveness of embedded PZT
716 patches at varying orientations for monitoring concrete hydration using EMI technique," Construction and
717 Building Materials, vol. 169, pp. 489-498.
718 46. Huynh, T.C., Dang, N.L., and Kim, J.T. (2017), "Advances and challenges in impedance-based structural
719 health monitoring," Structural Monitoring and Maintenance, vol. 4(4), pp. 301-329.
 Sensors 2020, 20, x FOR PEER REVIEW 21 of 21

720 47. Yang, Y., Lim, Y.Y., and Soh, C.K. (2008), "Practical issues related to the application of the electromechanical
721 impedance technique in the structural health monitoring of civil structures: I. Experiment," Smart Materials and
722 Structures, vol. 17(3).
723 48. Ai, D., Lin, C., and Zhu, H. (2020), "Embedded piezoelectric transducers based early-age hydration
724 monitoring of cement concrete added with accelerator/retarder admixtures," Journal of Intelligent Material
725 Systems and Structures, vol.
726 49. Bagherzadeh, R., Sadeghi, A.H., and Latifi, M. (2011), "Utilizing polypropylene fibers to improve physical
727 and mechanical properties of concrete," Textile Research Journal, vol. 82(1), pp. 88-96.
728 50. Epoxy, Fix your things with PC-products, https://www.pcepoxy.com/wp-
729 content/uploads/2017/07/superepoxydatabackcolor.pdf, Accessed on, Dec, 2019.
730 51. Zhao, S., Fan, S., Yang, J., and Kitipornchai, S. (2020), "Numerical and experimental investigation of electro-
731 mechanical impedance based concrete quantitative damage assessment," Smart Materials and Structures, vol.
732 29(5).
733 52. Bhalla, S., and Soh, C.K. (2004), "Electromechanical Impedance Modeling for Adhesively Bonded Piezo-
734 Transducers," Journal of Intelligent Material Systems and Structures, vol. 15(12), pp. 955-972.
735 53. Vsl 2018. VSL Strand Post-tensioning systems. In: (ed.)^(eds.) ed.
736 54. Huynh, T.C., Dang, N.L., and Kim, J.T. (2018), "Preload Monitoring in Bolted Connection Using
737 Piezoelectric-Based Smart Interface," Sensors (Basel), vol. 18(9).
738 55. Aashto, L. 2007. AASHTO LRFD bridge design specifications, SI units. In: (ed.)^(eds.) ed.: Washington, DC:
739 American Association of State Highway and Transportation.
740 56. Roberts, C. (1990), "Behavior and design of the local anchorage zone in post-tensioned concrete", Thesis,
741 The University of Texas, Austin, Texas.
742 57. Dang, N.L., Huynh, T.C., Pham, Q.Q., Lee, S.Y., and Kim, J.T. (2020), "Damage‐sensitive impedance sensor
743 placement on multi‐strand anchorage based on local stress variation analysis," Structural Control and Health
744 Monitoring, vol. 27, pp. e2547.
745 58. Zhang, C., Yan, Q., Panda, G.P., Wu, W., Song, G., and Vipulanandan, C. (2020), "Real-time monitoring
746 stiffness degradation of hardened cement paste under uniaxial compression loading through piezoceramic-
747 based electromechanical impedance method," Construction and Building Materials, vol. 256.
748
749
© 2020 by the authors. Submitted for possible open access publication under the terms
and conditions of the Creative Commons Attribution (CC BY) license
(http://creativecommons.org/licenses/by/4.0/).
750