1 COMBINED ENERGY HARVESTING AND STRUCTURAL HEALTH 1 MONITORING POTENTIAL OF 2 EMBEDDED PIEZO CONCRETE VIBRATION SENSORS 3 Naveet Kaur 1 and Suresh Bhalla 2 4 ABSTRACT 5 Piezoelectric materials have proven their efficacy for both energy harvesting and structural 6 health monitoring (SHM) individually. Piezoelectric ceramic (PZT) patches, operating in d 31 - 7 mode, are considered best for SHM. However, for energy harvesting, built up configurations 8 such as stack actuators are more preferred. The proposed study in this paper provides a proof- 9 of-concept experimental demonstration of achieving both energy harvesting and structural 10 health monitoring from the same PZT patch in the form of concrete vibration sensor (CVS), 11 designed specifically for reinforced concrete (RC) structures. This packaged sensor (CVS), 12 composite in nature, has better compatibility with surrounding concrete and can withstand the 13 harsh conditions encountered during construction. The paper covers experiments carried out 14 in the laboratory environment to measure the voltage and the power generated by a CVS 15 embedded in a life-sized simply supported RC beam subjected to harmonic excitations. An 16 analytical model is developed to compute the power output from the embedded CVS, duly 17 considering the effect of the shear lag associated with the bonding layers between the 18 encapsulated PZT sensor and the surrounding concrete. The performance of the CVS is 19 compared with the surface-bonded PZT patch. Utilization of the same patch for SHM through 20 a combination of the global vibration and the local EMI techniques is also covered. 21 Harvesting potential of vibration energy by PZT sensors during idle time is experimentally 22 demonstrated and extended to real-life structures based on the validated analytical model. 23 Keywords: Energy harvesting, structural health monitoring, d 31 mode, real life structure, embedded 24 PZT patches, concrete vibration sensor (CVS), traffic loads. 25 1 Research Scholar, Department of Civil Engineering, Indian Institute of Technology (IIT) Delhi, Hauz Khas, New Delhi ‐ 110 016, (India). E‐Mail: [email protected], Phone: +91‐8130‐332121 2 Associate Professor (Corresponding Author), Department of Civil Engineering, Indian Institute of Technology (IIT) Delhi, Hauz Khas, New Delhi ‐ 110 016, (India). Email: [email protected],Phone: (91)‐11‐2659‐1040, Fax : (91)‐11‐2658‐1117
Transcript
1
COMBINED ENERGY HARVESTING AND STRUCTURAL HEALTH 1 MONITORING POTENTIAL OF 2
Piezoelectric materials have proven their efficacy for both energy harvesting and structural 6
health monitoring (SHM) individually. Piezoelectric ceramic (PZT) patches, operating in d31-7
mode, are considered best for SHM. However, for energy harvesting, built up configurations 8
such as stack actuators are more preferred. The proposed study in this paper provides a proof-9
of-concept experimental demonstration of achieving both energy harvesting and structural 10
health monitoring from the same PZT patch in the form of concrete vibration sensor (CVS), 11
designed specifically for reinforced concrete (RC) structures. This packaged sensor (CVS), 12
composite in nature, has better compatibility with surrounding concrete and can withstand the 13
harsh conditions encountered during construction. The paper covers experiments carried out 14
in the laboratory environment to measure the voltage and the power generated by a CVS 15
embedded in a life-sized simply supported RC beam subjected to harmonic excitations. An 16
analytical model is developed to compute the power output from the embedded CVS, duly 17
considering the effect of the shear lag associated with the bonding layers between the 18
encapsulated PZT sensor and the surrounding concrete. The performance of the CVS is 19
compared with the surface-bonded PZT patch. Utilization of the same patch for SHM through 20
a combination of the global vibration and the local EMI techniques is also covered. 21
Harvesting potential of vibration energy by PZT sensors during idle time is experimentally 22
demonstrated and extended to real-life structures based on the validated analytical model. 23
Keywords: Energy harvesting, structural health monitoring, d31 mode, real life structure, embedded 24 PZT patches, concrete vibration sensor (CVS), traffic loads. 25
1 Research Scholar, Department of Civil Engineering, Indian Institute of Technology (IIT) Delhi, Hauz Khas, New Delhi ‐ 110 016, (India). E‐Mail: [email protected], Phone: +91‐8130‐332121 2 Associate Professor (Corresponding Author), Department of Civil Engineering, Indian Institute of Technology (IIT) Delhi, Hauz Khas, New Delhi ‐ 110 016, (India). Email: [email protected],Phone: (91)‐11‐2659‐1040, Fax : (91)‐11‐2658‐1117
2
INTRODUCTION 26
Replacing the batteries employed for running the monitoring process of reinforced concrete 27
(RC) structures by embedded PZT patches is somewhat complex. Hence, it becomes essential 28
to search upon a pragmatic solution which can act as an alternative for the replacement of the 29
batteries. This paper proposes a possible solution by integrating the structural health 30
monitoring (SHM) with energy harvesting, using same embedded PZT sensor. SHM is 31
defined as the measurement of the operating and loading environment and the critical 32
responses of a structure to track and evaluate the symptoms of operational incidents, 33
anomalies, and/or deterioration or damage indicators, which may affect operation, 34
serviceability or safety reliability (Aktan et al. 2000). The SHM techniques reported in the 35
literature can be broadly classified into (i) Global dynamic techniques (ii) Local techniques. 36
Curvature mode shape has been used extensively in literature (Pandey et al. 1991 and Zhou et 37
al. 2007) for damage identification in conjunction with the global dynamic technique. It is 38
based on the fact that with reduction in the flexural stiffness resulting from damage, the 39
curvature of a flexural structural member increases abruptly locally. Using surface bonded/ 40
embedded PZT patches, curvature mode shape can be directly obtained since the voltage 41
response across the patch in such configuration is directly proportional to strain, and hence 42
the curvature (Shanker et al. 2011). Among the various local techniques, the electro-43
mechanical impedance (EMI) technique has established its niche for SHM of structural 44
engineering systems (Bhalla and Soh, 2004a; Bhalla et al. 2012). The EMI technique is based 45
on change in the mechanical impedance of the structure with change in structural parameters, 46
i.e., as stiffness, the damping and the mass due to any damage. It works in very high 47
frequency range (30 kHz to 400 kHz), which makes the technique highly sensitive to damage, 48
with the sensitivity typically reaching of the order of the ultrasonic techniques (Park et al. 49
2000). 50
3
Energy harvesting is converting the ambient energy (sunlight, thermal gradient, human 51
motion and body heat, vibration, and ambient radio-frequency energy) into useful forms for 52
direct/future use. In the present work, mechanical vibrations of the structure are considered as 53
the ambient source for energy. During the past two decades, the piezo sensors have already 54
established their potential in detecting, locating and estimating the severity levels of 55
structural damages. Employing these sensors for energy harvesting along with SHM has 56
attracted researchers after the recent advent of modern low power consuming electronics. 57
Several review articles can be found covering the wide variety of mechanisms and techniques 58
for energy harvesting (Priya 2007; Anton and Sodano 2007; Beeby et al. 2006, Roundy and 59
Wright 2004; Sodano et al. 2004a; Priya and Inman 2009) based on various configurations of 60
piezos, such as multilayer, macro fibre composite (MFC), bimorphs, quick pack etc. Goldfarb 61
and Jones (1999) explained that the basic problem with harvesting electrical power using the 62
piezoelectric materials is that the majority of the energy produced by it is returned back to the 63
excitation source as the reactive energy. Various efforts on analytical modeling of 64
piezoelectric energy harvesting, considering the effects of mechanical and electrical loads and 65
electrical circuits, can also be found in the literature (Roundy and Wright 2004; Goldfarb and 66
Jones 1999; duToit et al. 2005; Sodano et al. 2004b; Lu et al. 2004; Twiefel at al. 2006; Kim 67
et al. 2005; Sohn et al. 2005). The power density (W/cm3 or W/kg) or the efficiency 68
parameter was suggested by duToit et al. (2005) as a good indicator for comparison. 69
70
Although extensive research in using piezoelectric sensors for energy harvesting can be found 71
in literature, the integrated use of these sensors both for SHM and energy harvesting is new in 72
its kind. Possibility of energy harvesting using the d31-mode (the stress applied in axial 73
direction and the electric voltage developed in perpendicular (thickness) direction), which is 74
generally employed for SHM, has recently been explored in depth by Kaur and Bhalla (2014) 75
for surface bonded PZT patches. This configuration provides an advantage of being simple 76
4
and most natural excitation from ambient sources in civil structures (Ramsey and Clark 2001; 77
Mateu and Moll 2005). Additionally, the PZT patch in this configuration adequately serves 78
for SHM, utilizing either the EMI technique or the global vibration technique (based on strain 79
mode shape) or a combination of the two (Shaker et al. 2011). This paper is an extension of 80
the earlier developments of the authors on surface bonded sensors (Kaur and Bhalla 2014) to 81
embedded sensors. The main objective of this paper is to explore the possibility of energy 82
harvesting from embedded PZT patches operating in the d31-mode owing to their suitability 83
for SHM. 84
85
The principle of integrated SHM and energy harvesting is illustrated in Fig. 1. The structure 86
is assumed to be operating in two stages, idle state and SHM state. During the idle state 87
(when SHM is not being performed), the PZT patches embedded inside the structure will 88
harvest the energy and store it in an appropriate storage device like battery or capacitor. In 89
the SHM state, the same stored energy will be utilized for the SHM of the host structure using 90
the same PZT patch, either in the global mode or the local mode or both. In this paper, an 91
analytical model has been developed for beams to estimate the power output from embedded 92
patches. The losses associated with the PZT patch are included in the model and quantified. 93
An estimation of energy harvesting from real life structures is made based on the validated 94
analytical model. 95
96
PZT PATCH AS EMBEDDED CONCRETE VIBRATION SENSOR (CVS) 97
Concrete Vibration Sensor (CVS), shown in Fig. 2(a), is a packaged sensor, designed 98
especially for monitoring RC structures. CVS is composite in nature, has better compatibility 99
with the surrounding concrete, and can withstand the harsh conditions encountered typically 100
encountered in the RC structures during casting. It is a proprietary product developed by 101
Bhalla and Gupta (2007) in the Smart Structures and Dynamics Laboratory (SSDL), IIT 102
5
Delhi (SSDL 2014). It consists of a PZT sensor patch encapsulated in a proprietary 103
configuration suitable for casting along with the structure, thereby permanently embedding 104
the patch in the host RC structure. The packaging provides an additional advantage of 105
protecting the sensing element against ambient environmental conditions, hence, enhancing 106
its life expectancy. 107
108
In this section, an analytical model is developed for the prediction of the voltage generated by 109
the PZT patch embedded as CVS in a simply supported RC beam with the configuration 110
shown in Fig. 2(a). The beam is assumed to be under a concentrated sinusoidal load at the 111
centre with an operating frequency ( )ω . The theoretical amplitude ( )a of the dynamic 112
vibration of the beam, considering first n modes, is given by (Chopra 1995; Kaur and Bhalla 113
2014), 114
( )∑∞
= ⎥⎥
⎦
⎤
⎢⎢
⎣
⎡
⎟⎟
⎠
⎞
⎜⎜
⎝
⎛−=
1
2/
sinn
nDn
Lx
PtRK
a n θωφ
(1) 115
where, ( )nφ denotes the mode shape and nω the cyclic natural frequency, RD the dynamic 116
magnification factor, θ the phase angle and Pn the generalized force, respectively given by, 117
m
EIL
nn 2
22πω =
and
( )
Lxnxnπφ sin= (2) 118
( )[ ] ( )[ ]222 21
1
nn
DRωωζωω +−
= , (3) 119
( )[ ]( )[ ]2
1
12
tann
n
ωωωωζ
θ−
= − , (4) 120
and ⎟⎠⎞
⎜⎝⎛=
2sinsino
πω ntpPn (5) 121
6
where, ζ denotes the damping ratio. The configuration of the embedded PZT patch 122
[Fig. 2(a)] is different from that of the surface bonded PZT patch earlier considered by Kaur 123
and Bhalla (2014). Stress here is transferred from the structure to the patch from the top, the 124
bottom, as well as the side faces, unlike the surface bonded configuration where only the face 125
solely interacts with the structure. The potential difference (Vp) across the terminals of the 126
PZT patch of thickness h, undergoing an axial strain (S1) can be derived in line with Kaur and 127
Bhalla (2014) as, 128
( )∑∞
= ⎥⎥
⎦
⎤
⎢⎢
⎣
⎡
⎟⎟
⎠
⎞
⎜⎜
⎝
⎛−
⎟⎟
⎠
⎞
⎜⎜
⎝
⎛=
1
2/
233
'31 sin
12n
nDn
Lx
T
E
p PtRKL
dYhdV n θω
φ
ε (6) 129
where, d’ denotes the distance of the centre line of the PZT patch from the neutral axis (refer 130
Fig. 2), d31 the piezoelectric strain coefficient, ( )jYY EE η+= 1 the complex Young’s 131
modulus of elasticity of the PZT patch at constant electric field and ( )jTT δεε −= 13333 the 132
complex electric permittivity (in direction ‘3’) at constant stress, 1−=j and η and δ 133
denote the mechanical loss factor and the dielectric loss factor of the PZT material, 134
respectively. It may also be noted that the above equation assumes perfect bonding between 135
the PZT patch and the structure. The losses considered here are the mechanical and the 136
dielectric losses (resulting from the heat generated by PZT material) and the shear lag loss 137
due to the interaction of PZT patch with the host structure via the adhesive layer. Based on 138
the background covered by Kaur and Bhalla (2014), the absolute value of the voltage, MpV 139
and DpV generated by the PZT including the effect of mechanical loss and dielectric loss, 140
respectively, can be expressed as, 141
aLdhYd
V T
EM
p⎥⎥⎦
⎤
⎢⎢⎣
⎡ += 2
33
2'31 112
εη
(7) 142
7
and aL
dhYdVT
ED
p⎥⎥⎦
⎤
⎢⎢⎣
⎡
+=
2233
'31
112
δε (8) 143
where, ‘a’ is given by Eq. (1). 144
145
The PZT patch is encapsulated inside the CVS, which is in turn embedded inside the host 146
structure. Unlike the surface bonded PZT patch (Sirohi and Chopra 2000; Bhalla and Soh 147
2004b), the adhesive (such as epoxy) in this case, forms a permanent finitely thick interfacial 148
layer between the host structure and the PZT patch on both sides of the patch (see Fig. 3), 149
thus inflicting greater shear lag effect. In addition, the boundary conditions encountered at the 150
two ends are different from stress free conditions in the case of surface-bonded patch. Here, a 151
close form analytical solution is derived here for embedded PZT patch, duly considering the 152
shear lag induced by the adhesive layer. The typical configuration of the system is shown in 153
Fig. 3(a). The patch has a length Lp, width wp and thickness h, while the bonding layer has a 154
thickness of ts (both top and bottom), and the adhesive encasing PZT patch is located at a 155
depth tc from the surface of the beam. The beam has an overall depth D and width wb. Let Tp 156
denote the axial stress in the PZT patch and τ the interfacial shear stress. Let up be the 157
displacement at the interface between the PZT patch and the bond layer and u the 158
corresponding displacement at the interface between the bond layer and the beam at a 159
distance ‘x’ from the centre of the patch. Considering the static equilibrium of the differential 160
element of the PZT patch in the x-direction, as shown in Fig. 3(a), we can derive, along the 161
lines of Sirohi and Chopra (2000) and Bhalla and Soh (2004b) 162
hx
Tp
∂∂
=τ2 (9) 163
The bending moment at any cross section of the beam, where the PZT patch is embedded, is 164
given by 165
( )httDhwTM scpp 5.05.0 −−−= (10) 166
8
Further, using Bernoulli’s theorem, we can derive 167
⎟⎠⎞
⎜⎝⎛−=
DIM b 5.0
σ (11) 168
where, σb is the bending stress of the beam at its extreme top fibre and ‘I’ is the second 169
moment of inertia of the beam cross-section. The negative sign signifies the fact that the 170
sagging moment and the tensile stresses are considered positive. Comparing Eqs. (10) and 171
(11), and solving, 172
04
'
=+IhDDwT pp
bσ (12) 173
Substituting ( )httD sc −−− 22 by D’ differentiating with respect to x, and comparing with 174
Eq. (9), we get 175
( ) 024
'
=+∂∂ τσ
IDDw
xpb (13) 176
From Fig. 3(b), the shear strain ( )γ in the bond layer can be expressed as, 177
s
p
tuu −
=γ (14) 178
Using Hooke’s law ( )γτσ spe
pbb GSYTES === and ; and substituting Eq. (14) in Eqs. (9) 179
and (13), then differentiating with respect to x, we get Eqs. (15) and (16), respectively. 180
Es
Ebsp
htYSG
xS
ξ⎟⎟⎠
⎞⎜⎜⎝
⎛=
∂
∂ 22
2
(15) 181
Es
bpsb
EItSDDwG
xS
ξ⎟⎟⎠
⎞⎜⎜⎝
⎛−=
∂∂
42 '
2
2
(16) 182
where, ⎟⎟⎠
⎞⎜⎜⎝
⎛−=
b
pE S
S1ξ (17) 183
Here, E and YE denote the Young’s modulus of elasticity of the beam and the PZT patch (at 184
zero electric field for the patch), respectively, and Sb and Sp respectively the corresponding 185
9
strains. Gs denotes the shear modulus of elasticity of the bond layer and γ the shear strain 186
undergone by it. Subtracting Eq. (15) from Eq. (16), we get 187
022
2
=Γ+∂∂
EEE
xξξ (18) 188
where, ⎪⎭
⎪⎬⎫
⎪⎩
⎪⎨⎧
+=Γs
ps
sE
sE EIt
DDwGhtY
G4
2'
2 (19) 189
The parameter ΓE (unit m-1) is the modified shear lag parameter and the ratio ξE is the strain 190
lag ratio for embedded PZT patch. The ratio ξE is a measure of the differential PZT patch’s 191
strain (patch being in embedded condition) relative to the strain on the host substrate 192
surrounding the PZT patch, caused by the shear lag effect. The general solution for Eq. (18) 193
can be written as, 194
xBxA EEE Γ−Γ= sinhcoshξ (20) 195
Since, the PZT patch is embedded inside the concrete, hence, the shear lag ratio will be equal 196
on the both the ends of embedded PZT patch ( )( )lELxELxE ppξξξ == −=+= i.e. . Further, contrary 197
to the surface-bonded PZT patch whose ends are stress-free, the ends of the PZT patch for 198
CVS in the present configuration experience non-zero stress. Using Eq. (17) and assuming 199
that at the ends of PZT patch, the stress in the beam (σb=ESb) will be same as the stress in the 200
embedded PZT patch (σp=YESp), we can derive that, 201
( ) ⎟⎠⎞
⎜⎝⎛ −== = ElxElE Y
E1)( ξξ (21) 202
Now, substituting the above equation in Eq. (20), we get 203
At x=+Lp; ( ) pEpElE LBLA Γ−Γ= sinhcoshξ (22) 204
At x=-Lp; ( ) pEpElE LBLA Γ+Γ= sinhcoshξ (23) 205
Solving Eqs. (22) and (23) for constants A and B, we get 206
10
( )
pE
lE
LA
Γ=
coshξ
and 0=B (24) 207
Substituting the values of constants A and B in Eq. (20), we get 208
( )pE
ElEE L
xΓΓ
=coshcosh
ξξ (25) 209
Also, from Eq. (17) and Eq. (25), the PZT to beam strain ratio can be derived as 210
( )pE
ElE
b
p
Lx
SS
ΓΓ
−=coshcosh
1 ξ (26) 211
The effective/equivalent length ( )Eeffl of the embedded PZT patch can thus be derived as 212
defined by Sirohi and Chopra (2000). It is that length which possesses a constant strain, equal 213
to Sb (the strain on the beam surface), such that the patch produces the same voltage output. It 214
is mathematically given by the area under the curve between (Sp/Sb) and (x/Lp) for half length 215
of the patch, that is, 216
( )∫=
=
=2/
0
pLx
xbp
Eeff dxSSl (27) 217
Substituting Eq. (26) into Eq. (27) and upon integrating, we can derive effective length 218
fraction EeffX as 219
( )
( )22tanh
12 pE
pEl
p
EeffE
eff LL
Ll
XΓ
Γ−== ξ (28) 220
The voltage ( )SpV
E generated by the PZT patch embedded in the concrete beam, duly 221
considering the effect of shear lag, can be expressed as (Sirohi and Chopra 2000) 222
( ) ∗= qbpESp SKKSV 1 (29) 223
where, S1 denotes the longitudinal strain developed in the beam at the level of the PZT patch, 224
Kb the correction factor to take care of the shear lag effect in the bond layer, Kp the correction 225
11
factor due to Poisson’s effect and ∗qS the circuit sensitivity, representing the output voltage 226
per unit strain input. Kp, Kb and ∗qS are given by (Sirohi and Chopra 2000), 227
⎟⎟⎠
⎞⎜⎜⎝
⎛−=
32
311dd
K p ν (30)228
Eeff
Eeffb BXK = (31) 229
T
E
qhYd
S33
31
ε=∗ (32) 230
where, ν is the Poisson’s ratio of the host structure material and d31=d32 for the square PZT 231
patch used in the present set of experiments. EeffX and E
effB are the effective length and width 232
fractions, respectively, as expressed by Eq. (28). The value of Kb is independent of the 233
material properties of the sensor, and is dependent only on its geometry and the properties of 234
the adhesive layer. Ignoring the shear lag effect along the direction of the width of the PZT 235
patch, EeffB can be considered as unity. Substituting 2'
1 12 LdS = (here d’ denotes the 236
distance of the centre line of embedded PZT patch from the neutral axis) in Eq. (29), the 237
following final expression can be derived for the voltage generated by PZT patch, duly 238
considering the shear lag effect 239
( ) aSKKLdV qbpE
Sp
∗= 2
'12 (33) 240
241
The voltage generated by the PZT patch, ( )E
SpV , embedded at midpoint of the RC beam 242
subjected to sinusoidal concentrated load, has been compared with the voltage generated by 243
surface bonded PZT patch, ( )S
SpV , both experimentally and analytically. Kaur and Bhalla 244
(2014) derived expressions for determining the shear lag factor and the effective length 245
(based on shear lag consideration) for a PZT patch surface bonded on beam at its midpoint. 246
12
The terms used in the following expressions for the surface bonded patch hold similar 247
explanations as given in the previous sections (note that subscript/ superscript ‘s’ signified 248
‘surface bonded’ configuration. 249
⎪⎭
⎪⎬⎫
⎪⎩
⎪⎨⎧
+=Γs
ps
sE
sS EIt
DwGhtY
G4
22 (34) 250
( )
( )22tanh
12 pS
pS
p
SeffS
eff LL
Ll
XΓ
Γ−== (35) 251
( ) aSKKLDV qbpS
Sp
∗= 2
6 (36) 252
The numerical values of the properties of the RC beam and the PZT patch / bond layer used 253
for calculating the shear lag parameters and the effective length and width fractions for both 254
the embedded and the surface bonded PZT patches are listed in Tables 1 and 2, respectively. 255
The adhesive layer is assumed to consist of two part araldite epoxy adhesive with shear 256
modulus of elasticity of 1GPa (Moharana and Bhalla 2012). The values of Kp = 0.9 and 257
V 10.197741=∗qS were considered. The amplitude (a) was calculated in accordance with Eq. 258
(1), considering damping (ζ) of 1.7% (determined experimentally using half-power band 259
width method). The comparison of the voltage generated by the embedded and the surface 260
PZT patches incorporating the effect of mechanical, dielectric and shear lag loss determined 261
theoretically with varying forcing frequency is shown in Fig.4. It may be noted that the 262
values of η and δ are so small that there is not much variation in absolute modulus values of 263
voltages, MpV and D
pV [see Eqs. (7) and (8)], for both the embedded and the surface bonded 264
PZT patches. For comprehensive description, refer Kaur and Bhalla (2014). It can be 265
observed from Fig. 4 that the theoretical voltage generated by the surface bonded PZT patch 266
is somewhat higher than the embedded CVS at the same location. The ratio of the voltage 267
generated by embedded CVS to the surface bonded PZT patch (Vemb/VSurf) was found to be 268
0.79 theoretically. This can be attributed due to the fact that the shear lag effect is playing 269
13
double effect (due to presence of bond layer at both top and bottom) since, the PZT patch is 270
surrounded by the adhesive on all its sides. 271
272
For experimental comparison of the voltage, a simply supported real-life sized RC beam, 273
with properties as listed in Table 1, was chosen as the experimental host structure. The 274
concrete of the beam confirmed to a self-compacting M40 grade with 30% fly ash. Ultimate 275
load carrying capacity of the beam was determined to be 9650 N, much greater than the total 276
weight of inertial shaker (about 800 N). The RC beam is shown in Fig. 5 (a) before casting 277
and (b) during casting. The schematic diagram of the beam showing the reinforcement, the 278
location of the embedded CVS and the notch for creating the damaged condition (during 279
SHM state) is shown in Fig. 6. The complete details of the damage induction and detection 280
are covered in latter sections. The beam consisted of two layers of 19 CVS each at top and 281
bottom, flushing with the surface. The complete experimental set-up under excitation is 282
shown in Fig. 7. Three PZT patches were surface bonded on the top of the beam, first one at 283
the centre (just above CVS10) and other two at an offset of 195 mm to left and right each 284
(above CVS9 and CVS11, respectively). The thickness of the bonding layer for the surface 285
bonded PZT patches was maintained equal to 150 microns with the help of optical fibers 286
while bonding the PZT patch on the beam surface, as done earlier by Bhalla and Soh (2004c). 287
The bond layer for the embedded CVS was maintained at 2.5 mm on either side of PZT patch 288
(one third of the thickness of CVS). The beam was excited using LDS V406 series portable 289
dynamic shaker. A function generator (Agilent 33210A) was used to generate an electrical 290
signal, which was amplified by a power amplifier (LDS PA500L) and transmitted to the 291
shaker, which converted the signal into mechanical force. Pure harmonic signal (sinusoidal in 292
nature) was applied to the structure via the function generator. Sinusoidal signals, with 293
different monotonic frequencies, at an excitation level of 5 V were applied in the experiment. 294
In the initial stage, the experiments were conducted using contact type dynamic shaker (LDS 295
14
V406), with the arrangement shown in Fig. 8(a). However, due to limitation of force the 296
generated by the contact type arrangement, it was converted into inertial-type shaker using a 297
simple rearrangement [Fig. 8(b)], with the help of four springs and cover plates. The 298
schematic diagram of the inertial-type shaker is shown in Fig. 8(c)-(e) with all the essential 299
details. The bottom plate and the mid plate were connected using four 16 mm diameter bolts 300
to facilitate secure connection with the RC beam and also to allow the movement of the 301
shaker along the length of the beam, whenever needed. LDSV406 shaker was connected to 302
the mid plate resting on the top of beam using four bolts. In this arrangement, the mechanical 303
force generated by the LDSV406 shaker is transferred to the top plate via seven stringers as 304
shown in Fig. 8(e). The inertial force generated by the vibration of the top plate and the 305
additional plates (which acted as the inertial mass) was finally transferred to the beam via the 306
four springs. Hence, the force transmission can be varied by varying the number of additional 307
mass plates at the top. The force generated by the inertial-type shaker was experimentally 308
quantified by measuring the acceleration of the top plate of inertial-type shaker. In the present 309
study, the force measured by an accelerometer attached at the top plate of inertial-type shaker 310
was measured to be 75 N. The inertial shaker arrangement is capable of exerting force in 311
excess of 110 N, much higher than the contact type shaker (refer Pal, 2013 for further 312
details).The output voltages generated by the surface bonded PZT patches and the embedded 313
CVS were compared in four different cases illustrated in detail in Fig. 9. These are briefly 314
described below: 315
Case (a): Force applied by hitting an impact hammer, with the contact type shaker present 316
[Fig. 9(a)]; 317
Case (b): harmonic force applied by contact type shaker, at 20 Hz frequency [Fig. 9(b)]; 318
Case (c): force applied by inertial-type shaker, at 20 Hz frequency, [Fig. 9(c)]; and 319
Case (d): force applied by inertial-type shaker, with the frequency sweeping from 5 Hz to 70 320
Hz, within a duration of 4 seconds [Fig. 9(d)]. 321
15
For better understanding of the type of signal applied and voltage output observed in the 322
sensors for above mentioned four cases, the typical overlapped voltage plots for the surface 323
bonded and the embedded CVS at location 11 varying with time are shown in Fig. 9 itself for 324
all the four cases described above. The nature of the voltage signal clearly depicts the type of 325
the force applied on the concrete beam. Various sets of readings were recorded for the above 326
mentioned four cases. For case (a), the voltage readings were recorded for both the surface 327
bonded and the embedded sensors at location 9, 10 and 11, with the hammer hitting at the 328
beam at locations 1, 8, 12 and 19 (Refer Fig. 7). The ratio (Vemb/VSurf) was experimentally 329
determined and averaged for the four sets of readings. On the similar lines, in rest of the three 330
cases, two sets of readings were recorded. Unfortunately for case (d), due to unexpected 331
malfunctioning of embedded CVS at locations 9 and 10, the comparison for sensor at location 332
11 only is possible. Given the rigorous nature of the experiments and consistency observed in 333
other results, trustworthy conclusions can be drawn without the readings of CVS at location 9 334
and 10 for case (d). The voltages generated by the embedded CVS and the surface bonded 335
sensors along with the average of the ratio (Vemb/VSurf ) for the four cases and their subsequent 336
sets of readings are provided in Table 3. 337
338
The ratio (Vemb/VSurf), when averaged over all the cases, comes out to be 0.967 (against 0.79 339
from theoretical analysis), which strengthens the conclusions deduced by the proposed 340
analytical model and the related observation that the voltage generated by surface bonded 341
sensor is higher than that of the embedded CVS. However, exceptions can be observed from 342
the table in three sets of readings. Two of these sets (with average voltage ratios 1.39 and 343
1.82 at locations 10 and 11 respectively) were recorded by hitting the beam with hammer 344
[Case(a)], which not only implied inconsistency of the applied force, but possibly caused 345
direct compression of the embedded patch (and hence d33 effect), leading to somewhat higher 346
voltage. In other cases [Case (b) to (d)] shaker was used to maintain the consistency in 347
16
applied force. However, an exception can be observed here too for Case (b) at location 11, 348
where the average ratio is 1.06. The possible reason is that the proximity to the stinger in the 349
contact type shaker configuration, which caused direct compression of the embedded CVS. 350
Further, the possible reasons for the experimental voltage ratio (0.967) being higher than the 351
theoretical value (0.79) are the non-inclusion of possible localized 3D stress effects and the 352
idealized modeling of the shear lag effect in the theoretical analysis. In one isolated case of 353
shaker excitation [case (c)], somewhat lower value (0.41) can be observed from Table 3. 354
Possible reason could be that the mid base plate [refer Fig. 8(c)] of the inertial-type shaker 355
was touching the surface bonded sensor, thereby inducing d33 effect and thus leading to bit 356
higher value of the voltage in the surface bonded sensor. However, in most other cases, where 357
the sensor was far from the loading system, the voltage ratio (Vemb/VSurf) lied in the range 0.67 358
to 0.94, implying consistency of the observations and hence strengthening the conclusions of 359
the model. 360
361
STRUCTURAL HEALTH MONITORING OF RC BEAM USING CVS 362
Continuous long term monitoring of the strength gain and fatigue characteristics of the RC 363
beam were first investigated using two techniques, namely, the global dynamic technique and 364
the local EMI technique via the embedded CVS, commencing immediately after casting and 365
continuing for 108 days. In the global dynamic technique, the global characteristics, here the 366
fundamental natural frequency, were examined. The natural frequency and the damping ratio 367
of the beam were determined via impact hammer test. Effect of conversion of the shaker from 368
contact-type to inertial-type on the natural frequency of the beam was also investigated. 369
When the shaker was kept under the beam, the natural frequency of the RC beam was 370
measured to be 22.32 Hz (28th day after casting), which matched well with the theoretical 371
value (fn=22.27 Hz) of first fundamental natural frequency. However, when the contact type 372
shaker was converted into an inertial-type shaker and placed at the top of the beam, a 373
17
reduction in natural frequency to fn=20.18 Hz was observed, which can be attributed to the 374
increase in mass of the system due to addition of approximately 80 kg additional dead weight 375
of the shaker system. The natural frequency of the beam from the 20th day to the 108th day 376
from casting at varying intervals of time is shown in Fig. 10(a). It can be observed that 377
natural frequency remained almost constant (22.32 Hz) till day 40 and started reducing for 378
interval-II (day 40 to day 108). In the local dynamic technique, the equivalent stiffness of the 379
RC beam, based on Bhalla et al. (2012), using conductance and susceptance signatures 380
(frequency range: 50 kHz to 250 kHz), was monitored from the 6th day to the 108th day using 381
Agilent E4980 LCR meter. The equivalent stiffness was determined from the signature of the 382
CVS at location 16 (top) using the computational procedure outlined in Bhalla et al. (2012) 383
and Talakokula et al. (2014). The PZT patch identified the host structure as a Kelvin-Voigt 384
system. The equivalent stiffness exhibited an increasing trend, as observed from Fig. 10(b), 385
during the curing period (day 6 to day 14) similar to the expected behavior of physical 386
stiffness. After this curing period, the equivalent stiffness became constant over the interval-I 387
(day 14 to day 40) as can be observed from Fig. 10(c). Thereafter, it started exhibiting similar 388
behavior (reducing trend) as shown by natural frequency, for interval-II (day 40 to day 108). 389
After interval-I, the contact type shaker was converted to inertial-type shaker, which weighed 390
around 80 kg, and was installed above the beam (refer preceding Section) and operated on 391
daily basis. Rigorous shaking of beam under fatigue loading of 1.368 × 105 cycles per day 392
(overall 9.303 × 106 cycles for interval II) ended up in development of micro-cracks and 393
reduction in fatigue strength, and hence reduced the equivalent stiffness and the natural 394
frequency of the RC beam. 395
396
Detailed structural health monitoring (SHM) of the RC beam was carried out using embedded 397
CVS as an integral part of the study. Controlled damage was induced by chipping off the 398
concrete at a specific location (one third length of the beam, between CVS location 7 and 8) 399
18
where a notch was created at the time of casting (see Fig. 7 and 11 for details). Curvature 400
mode shape (global vibration technique) and EMI signature (local vibration technique) of the 401
RC beam were compared in the undamaged and damaged state. The damage was induced in 402
three levels as explained using Fig. 11. State 1 represents the undamaged condition, and in 403
State 2, concrete was chipped off from the notch, which was specifically designed to create 404
damage. For states 3 and 4, 50% and 100% of the bottom reinforcement was curtailed, 405
respectively. Detailed observations of the two techniques are covered below. 406
407
Global Vibration Technique 408
The SHM of the beam using the global vibration technique was carried out by obtaining its 409
curvature mode shape in the undamaged state and comparing it with the damaged state. The 410
key feature of this approach is that curvature mode shape is directly obtainable from the 411
response of the PZT patches embedded near the surface (Shanker et al. 2011). For this 412
purpose, both the PZT patches located at a given section were connected in series. From 413
fundamental of structural analysis, curvature ( )ϕ at a given point of beam can be derived as 414
( ) DSS bt +=ϕ (37) 415
where St and Sb are the flexural strains at the top and the bottom fibres of the beam, 416
respectively and D is overall depth of beam. As evident from fundamentals of 417
piezoelectricity, the voltage across PZT patch is proportional to the strain at the location of 418
sensor on the beam. Hence, the combined voltage given by two sensors (each top and 419
bottom), when connected in series, can be derived as 420
( )btq SSSV += *
(38) 421
From Eqs. (37) and (38), it can be concluded that the curvature ( )ϕ of the beam is 422
proportional to the combined voltage (V) generated by the PZT sensors located at the top and 423
bottom of the beam and connected in series. The experimental set-up used for obtaining the 424
19
curvature mode shape is shown in Fig. 12. Impact force was applied on the RC beam by 425
hitting it with a wooden hammer at the alternate CVS locations, the force being measured 426
with Dytran force sensor mounted over the beam [see Fig. 12(a)]. The voltage signal across 427
the series connected CVS at locations 2, 4, 8..,16 and 18 was acquired using the eight channel 428
QDA1008 (Quazar Technologies 2013) and a computer in two rounds of operation. The 429
voltage response (frequency domain) was normalized with that of the measured force. The 430
curvature mode shapes for the undamaged and the damaged states were obtained by plotting 431
the relative amplitudes (Pal 2013; Shaker et al. 2011) of the curvature along the length of the 432
beam, as shown in Fig. 13(a). The damage index for State 2 to State 4 was computed for the 433
elements between the different sensor locations using the following relation (Talwar 2011) 434
( )d
i
nnu
i
nnjiDI ⎟
⎠⎞
⎜⎝⎛ +
−⎟⎠⎞
⎜⎝⎛ +
= ++
2211 φφφφ (39) 435
where, j represents the damage state, i denotes the element number between nth and (n+1)th 436
sensor node location, where n varies as 2, 4, 6..,16 and 18. The variation of the damage index 437
for the three damage states is shown in Fig. 13(b). It can be observed from the figure that the 438
damage has been correctly and effectively located using the CVS sensors for damage states 3 439
and 4, with maximum damage index value for the most severe damage state 4. The error lies 440
only with the reference to State 2, in which case the level of the induced damage is incipient 441
in nature. This is a well accepted fact for global vibration technique (Shanker et al. 2011)and 442
the short coming is nullified with the aid of the EMI technique, as described below. 443
444
EMI Technique 445
The three levels of damage in the RC beam were also monitored via the EMI technique, 446
which is the local vibration technique, typically operating in the high (kilohertz) frequency 447
range. The EMI signature of the beam was acquired using the new cost effective and low 448
power consuming miniature impedance analyzer AD5933 (Analog Devices 2013) and 449
20
verified using the conventional LCR meter, model E4980A (Agilent Technologies 2013). The 450
signature was acquired for the 19 CVS at top and 19 at bottom separately, using both the 451
equipment. The experimental setup used for acquiring the EMI signature using AD5933 is 452
shown in Fig. 14. For the operation of AD5933, the frequency range chosen was 80 kHz to 453
100 kHz. The procedure for using AD5933 is based on the Shanker (2013). The calibration of 454
the circuit was done before using it for the CVS sensors. The output of the circuit is in form 455
of the real part (resistance, R) and the imaginary (reactance, X) part of impedance; from 456
which, the conductance value was determined as 457
22 XRRG+
= (40) 458
A typical plot of the conductance for various states using AD5933 and LCR meter is shown 459
in Fig. 15(a) and Fig. 15(b) respectively. Similar conclusions can be deducted from the two 460
plots. Significant difference in the conductance values for State 1 (undamaged condition) and 461
state 2 (incipient level of damage) can be easily observed. This state was not correctly located 462
using the global vibration technique. For locating the damage, RMSD value (Shanker 2011) 463
for each sensor node was calculated using the following equation, 464
( )
( )100(%) 2
2
×−
=∑
∑uk
uk
dk
G
GGRMSD (41) 465
where, ukG denotes the undamaged conductance value and d
kG the conductance value after 466
damage for the kth frequency. RMSD value for the ith element (between nth and (n+1)th sensor 467
node location) for the 19 CVS at top and bottom was calculated each using 468
21++
= nni
RMSDRMSDRMSD
(42) 469
RMSD values were determined for three different damage states for the 19 CVS at top and 470
bottom each and plotted for bottom 19 CVS sensors for AD5933 and LCR meter in Fig. 16(a) 471
and Fig. 16(b), respectively. It can be observed from the figure that the damage in all the 472
21
three stages has been effectively located. However, the difference in the magnitudes of the 473
RMSD index corresponding to States 3 and 4 is not very high. This fact can also be well 474
corroborated with Fig. 15, where significant shift is observed for State 2 but not thereafter. 475
Hence, as damage level grows from incipient to moderate, the higher level of damages 476
became less and less distinguishable. Again, this is a well accepted fact for the EMI 477
technique (Shanker et al. 2011) and therefore, makes case for the integrated use of the global 478
vibration technique and EMI technique. The next section highlights the possibility of energy 479
harvesting along with SHM by the embedded CVS. 480
481
ENERGY HARVESTING POTENTIAL OF EMBEDDED CVS 482
Experiments were performed in the laboratory for harvesting and storing the energy 483
generated by the CVS embedded in RC beam. Based on vibration data reported in the 484
literature for seven existing real life bridges/flyovers, the power generated by a typical CVS 485
embedded in these structures is also computed. 486
487
Laboratory Experiments 488
The experimental set-up is similar as in Fig. 7. The power measurement was done for two 489
different locations of the concentrated dynamic loads, (a) shaker at the centre of the RC beam 490
(b) shaker at an offset of 600 mm from the beam centre. The PZT patch in the form of CVS, 491
embedded inside the RC beam, just below the shaker location was considered for power 492
measurement. Electrical signal in form of sine waves with varying frequency was generated 493
by the function generator, then amplified by the power amplifier (PA 500L) [Fig. 17(a)] and 494
further propelled to dynamic shaker (LDS V406), which transformed the signal to mechanical 495
force. The excitation frequency was varied monotonically at a step interval of 2 Hz and 5 Hz, 496
respectively for the study when the shaker was at the centre and at an offset of 600 mm from 497
the centre. The power generated by the PZT patch was measured via oscilloscope and the 498
22
simple in-house circuit shown in Fig. 17(b) (Kaur and Bhalla 2014). Here, the PZT patch acts 499
as the source of voltage generation (Vin). The value of load resistances, R1= 494.7 kΩ and R2 500
= 471.66 kΩ, were chosen by trial and error such that these values are close to impedance of 501
the source, the PZT patch. The peak current (IPeak) flowing in the circuit (which is very small, 502
hence difficult to directly measure) is determined by measuring voltage V2 as 503
12 RVIPeak = (43) 504
The maximum power (PPeak) generated by the embedded PZT patch based on Kaur and 505
Bhalla (2014) is given by 506
( )212 RRIP PeakPeak += (44) 507
The root mean square (RMS) values of the current and power can be determined using 508
IIRMS 707.0= and ( )212 RRIP RMSRMS += (45) 509
The total energy (U) generated can be computed by determining the area under the curve of 510
power (P) and time (t) and thus the average power (Pavg) over time t1 can be derived as 511
1tUPavg = (46) 512
The variation of three different forms of power, PPeak, PRMS and Pavg, with varying forcing 513
frequency for the above considered two cases is shown in Fig. 18. It can be observed that the 514
maximum power (in all forms) is achieved at the first natural frequency ( )Hz 19=f of the 515
vibrating structure (RC beam), when it experiences maximum deflections in the first case. In 516
the latter case, maximum values are observed at an operating frequency of 20 Hz. This is due 517
to adopting higher frequency interval (5 Hz) in latter case, resulting in missing the peak value 518
in its close proximity. Also, it can be observed that higher modes of vibration were 519
effectively captured (at a frequency of 75 Hz) in the later case when the shaker operated at an 520
offset of 600m from the centre. This mode was missed out in the first case because the sensor 521
location (centre of beam) was at the nodal point for second mode (Avitabile 2001). The 522
23
maximum power generated by CVS sensor embedded in the concrete is found to be 0.177 523
µW and 0.02 µW for the PZT at the centre of beam [Fig. 18(a)] and another PZT at an offset 524
of 600 mm from the centre [Fig. 18(b)], respectively at the natural frequency of the RC beam. 525
The maximum power generated by an embedded PZT patch is thus in the microwatt range. 526
Also, Table 4 provides an idea of the power generated by the embedded CVS at different 527
locations when the shaker was positioned at the centre of the beam and operated at the natural 528
frequency (19 Hz) of the beam. The maximum RMS power (0.09 µW) was found to be at the 529
centre for CVS at location 11. As expected, the power drops as we move away from the 530
centre of the beam towards the support. The normalized value of the power with respect to 531
acceleration was computed to be 0.027 µW/ms-2, against 0.046 µW/ms-2, which was observed 532
for the experimental steel beam with surface bonded PZT patch (Kaur and Bhalla 2014). 533
Although the yield is lower than steel structures, the results confirm that a CVS sensor 534
embedded in RC structures possesses the potential for energy harvesting suitable for low 535
power electronics. The embedded CVS has additional advantages that the sensor is 536
unobtrusively encased inside the concrete and well protected against environmental 537
degradation, thereby ensuring maintenance free operation and higher longevity. In addition, 538
being packaged sensor, its installation is relatively simpler than surface bonded sensors, 539
which requires a minimum skill level from the user. 540
541
Energy harvesting circuit [shown in Fig. 19(a)], consisting of a full wave bridge rectifier built 542
of Zener diodes and a 1000 µF capacitor, was used for harvesting and storing the energy 543
generated by the CVS embedded in the RC beam. The variation of the voltage across the 544
capacitor during its charging and discharging is shown in Fig. 19(b). It can be observed that 545
the capacitor charged to a maximum voltage of 97 mV in 187 seconds. Using the relation, 546
221 CVEc = , the energy stored in the capacitor (Ec) was computed as 4.753 µJ. It can be 547
24
derived that a continuous harvesting for 15 days is sufficient for one time operation of 548
AD5933 (refer subsequent section), which requires energy of 33 mJ . Hence, SHM of the real 549
life structure can be performed twice a month using the same CVS for SHM and energy 550
harvesting. The harvesting time is expected to reduce in the near future with the development 551
of further low power consuming electronics. The results of the lab study are extended to real-552
life structures in the next section. 553
554
Extention to Real-Life Structures 555
The dynamic vibrations, under traffic loads and ambient conditions, experienced by seven 556
existing real life bridges/flyovers have been considered in this study. Based on vibration data 557
reported in the literature, the power generated by a surface bonded/ embedded CVS in these 558
structures is computed. The seven cases are summarized in Table 5. The voltage generated by 559
the CVS embedded in these real life structures was estimated using the proposed analytical 560
model [Eq. (29)]. Further, the power generated was computed using Eq. (38) with R1= 494.7 561
kΩ and R2 = 471.66 kΩ. The voltage and the power generated by the embedded CVS under 562
vibrations experienced at the mid span of the real life bridges are listed in Table 5. The list 563
includes both steel (surface bonded PZT patches) and RC (embedded CVS). It can be 564
observed that the average power which can be achieved from the mechanical vibrations of the 565
steel (based on Kaur and Bhalla 2014) and RC bridges is 0.406 μW and 0.032 μW, 566
respectively. In one isolated case (Zuo et al. 2012), a maximum power of 26.154 μW has 567
been estimated. This power can be harvested during the idle state using the energy storage 568
circuits and used for SHM of the structure via same embedded CVS using AD5933 (Analog 569
Devices 2013), or any other equivalent circuit. Assuming that AD5933 consumes 33 mW of 570
power (Analog Devices 2013) and considering the average power generated by the embedded 571
CVS in real life bridge, it is estimated, that under traffic loads, a period of 22 hours for steel 572
bridges and 12 days for RC bridges (as CVS) will be needed for the thin piezo based 573
25
harvester to harvest sufficient energy so as to operate AD5933 for one second. With further 574
advancements in digital electronics (with new commercial versions of low power consuming 575
circuits), significant reduction in the energy harvesting time is expected in the future. 576
577
CONCLUSIONS 578
This paper has presented the feasibility of combined SHM and energy harvesting using 579
specially designed embedded PZT patch (CVS) operating in the axial mode. Experiments 580
have been carried out in the laboratory environment to measure the voltage and the power 581
generated by a PZT patch embedded in life sized RC beam. A coupled electro-mechanical 582
model for embedded CVS duly incorporating the losses associated with PZT patches, 583
especially the shear lag loss, has been derived and validated with experimental 584
measurements. Comparison of the voltage generated by the surface bonded PZT patch and 585
embedded CVS has been studied analytically and validated experimentally. Long term 586
continuous monitoring of the natural frequency and equivalent stiffness of the RC beam for 587
108 days showed that both the natural frequency and equivalent stiffness remained constant 588
for the initial period, however started reducing after day 40 due to loss of fatigue strength, 589
thus proving suitability for fatigue damage monitoring. SHM via the global vibration 590
technique and the local EMI technique of the RC beam was also performed using CVS as an 591
integrated part of this paper. It is concluded that the embedded CVS can effectively detect 592
damage ranging from incipient to severe nature using both global vibration technique and 593
local EMI technique in integration. Comparison of the power generated by the surface 594
bonded PZT patch and the embedded CVS suggests that the CVS is capable of generating 595
about 60% of its counterpart bonded in a steel structure subjected to same order of magnitude 596
of acceleration. Experimental demonstration for harvesting and storage of energy has also 597
been described. The coupled electro-mechanical model has been extended to seven real-life 598
bridges across the world. It is estimated that a period of less than a day for steel bridges and 599
26
about 12 days for RC bridges (as CVS) will be needed for the thin piezo based harvester to 600
harvest sufficient energy so as to enable one time operation of AD5933. With the ongoing 601
developments in electronics, as lesser power consuming circuits are emerging, it is believed 602
that the energy scavenging time will drastically come down. Hence, using PZT patch in the 603
form of CVS both for SHM and energy harvesting in real life structures is expected to prove 604
as a new and useful contribution. This piezo will act as energy harvester when not in use (idle 605
state) and shall carry out SHM utilizing its own energy harvesting when needed. 606
27
REFERENCES 607
Abdessemed, M., Kenai, S., Bali, A., and Kibboua, A. (2011). “Dynamic analysis of a bridge 608
repaired by CFRP: Experimental and numerical modeling.” Construction and 609
Building Materials, 25, 1270–1276. 610
Agilent Technologies (2013). “Test and Measurement Catalogue.” USA. 611
<http://www.home.agilent.com/>. 612
Aktan, A. E., Catbas, F. N., Grimmelsman, K. A., and Tsikos, C. J. (2000). “Issues in 613
infrastructure health monitoring for management.” J. Eng. Mech., 126(7), 711-724. 614
Anton, S. R., and Sodano, H. A. (2007). “A review of power harvesting using piezoelectric 615
Figure 9. Voltage generated by surface bonded and embedded CVS at location 11 741
varying with time for four cases namely case (a) to case (d). 742
Figure 10. Variation of (a) natural frequency, equivalent stiffness with increasing number 743
of days for (b) curing period and (c) after curing period. 744
Figure 11. The front view of the notch showing undamaged and different states of 745
damage [see this Fig. in conjunction with Fig. 6(c)]. 746
Figure 12. Experimental set-up for measuring mode shape of the concrete beam showing 747
(a) element numbers and (b) data acquisition system. 748
Figure 13. (a) Comparison of mode shape and (b) corresponding damage index for 749
undamaged and three damaged states. 750
Figure 14. Experimental set-up for local EMI technique done using AD5933 (a low-cost 751
alternative of conventional LCR meter). 752
Figure 15. Comparison of typical conductance (G) signature acquired using (a) AD5933 753
and (b) LCR meter for undamaged and different states of damage. 754
Figure 16. Variation of RMSD of conductance plots acquired using (a) AD5933 and (b) 755
LCR meter at different elements of concrete beam for three states of damage. 756
Figure 17(a). Overall set up for power quantification. 757
Figure 17(b). Details of energy quantification circuit. 758
33
Figure 18(a). Variation of PPeak, PRMS and Pavg with varying forcing frequency when: Shaker 759
and PZT patch at center and 760
Figure 18(b). Shaker and PZT patch at an offset of 600 mm from centre. 761
Figure 19(a). Charging and discharging voltage across capacitor. 762
Figure 19(b). Bridge rectifier circuit used for storing energy in capacitor. 763
764
LIST OF TABLES CAPTIONS 765
Table 1. Properties of RC beam. 766
Table 2. Properties of PZT patch (PI Ceramic 2013) and bond layer. 767
Table 3. Voltage generated by surface bonded and embedded CVS for four cases. 768
Table 4. Power generated by the embedded CVS at different locations 769
Table 5. Voltage and power generated by the embedded CVS under vibrations 770
experienced at the mid span of a real life bridge. 771
34
Figure 1: Principle of integrated structural health monitoring and energy harvesting.
STRUCTUREPIEZO
IDLE STATE
ENERGY HARVESTING DEVICE
SHM STATE
STRUCTURAL HEALTH MONITORING
+ ++ ++ ++- - - -- - -
35
Figure 2: (a) Dynamic load acting on a RC beam housing a CVS. Strain distribution across depth is also shown.
(b) Variation of excitation load with respect to time.
(b)
p0
-p0
Pn
t
(a)
Strain, S1PZT patchBeam Pn
d'
15 mm
25 mm
Concrete Vibration Sensor (CVS) (Bhalla and Gupta 2007)
36
(a)
(b)
Figure 3: (a) A PZT patch embedded inside concrete beam and bonded using adhesive layer. (b) Deformation in bonding layer and PZT patch embedded in concrete beam.
A B
D C
A’B’
D’C’
x
y
x u
up
After Deformation
uo
upoPZT Patch
BondingLayer
BEAM
Dh ts
ts
x
y
dx
BEAM
PZT Patch Bond Layer
Differential Element
2pL
tc
2pL
τ
τdx
Tp Tp + ∂Tp ∂x
dx
37
Figure 4: Theoretical comparison of voltage generated by embedded and surface PZT patch considering mechanical, dielectric and shear lag loss.
0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
8.0
9.0
0 5 10 15 20 25 30 35 40
Theo
retic
al V
olta
ge, (
V)
Frequency, (Hz)
Surface Bonded PZT Patch
Embedded PZT Patch
38
(a)
(b)
Figure 5: RC beam (a) before casting and (b) during casting.
19 CVS atBottom
19 CVS at Top CVS at Top
Wooden plank for leaving a pocket for damage
CVS at Bottom
CVS flushing with beam top
39
(a)
(b) (c)
Figure 6: (a) Concrete beam showing reinforcement, embedded CVS and notch for damage. (b) Detail of reinforcement (cross-section). (c) Damage states.
State 1
Section 2-2
State 4
70 mm
State 2 State 3
2 nos.-10 mm Dia
160 mm
2 nos.-16 mm Dia
8 mm @ 180 mm c/c
Section 1-1
20 mm
20 mm
170 mm
50 mm Notch
Span=4 m
Clear Span = 3.9 m 19 CVS each at top and bottom layer @ 195mm c/c
2
2
1
1
40
Figure 7: Complete experimental set up with CVS location.
Function Generator (Agilent 33210A)
Oscilloscope (TDS 2004B)
Notch after filling concrete
Embedded CVS flushing at top (typ.)
12
3 4
56 7 8 9
10
11-19
Dynamic Shaker (LDS V406)
Amplifier (LDS PA 500L)
Laptop
RC Beam
Connectors for CVS
Figurty
Sp
Sprin
Ci
re 8: (a) Coype shaker s
prings (4 No
ngs (4 Nos.; 3
ircular Hollohold spring
RC
(a)
ntact type Sshowing (c)
(d)
Bolts (Shakeros.)
460
1
380 mm dia)
ow Plate togs (4 Nos.)
C Beam
Shaker; (b) i) elevation;
(4 Nos; 16 mr Base
2
2
41
(c)
inertial-type(d) View 1-
mm dia)
250
F(
2
A
e Shaker; sc-1 and (e) V
(e
Top of
Top P(250×4
1
Four Additio(mP1= mP2=1.
Shake
Mid and Bo
Additional Ma
(b)
chematic diView 2-2 (un
e)
Shaker
LoStr
late (mt=19 k460×16)
onal Mass Pla78 kg; mP3=
r (LDS V406
ottom Plate (2
Bolts (4 N
ass Plates
agram of innits in ‘mm
cation of ringers (7 No
kg)
ates mP4=2.15 kg
6)
250×460×16
Nos.; 16 mm
7 Stringers
nertial-’).
os.)
)
6)
dia)
42
(a) (b)
(c) (d)
Figure 9: Voltage generated by surface bonded and embedded CVS at location 11 varying with time for four cases namely case (a) to case (d).
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
810 860 910 960
Vol
tage
(V)
Time (sec)
Embedded Surface
-0.45
-0.35
-0.25
-0.15
-0.05
0.05
0.15
0.25
1 1.1 1.2 1.3 1.4 1.5
Vol
tage
(V)
Time (sec)
Embedded Surface
-0.9
-0.7
-0.5
-0.3
-0.1
0.1
0.3
0.5
0.5 0.6 0.7 0.8 0.9 1
Vol
tage
(V)
Time (sec)
Embedded Surface
-0.52
-0.44
-0.36
-0.28
-0.20
-0.12
-0.04
0.04
0.12
0.20
3.6 4.8 6 7.2 8.4
Vol
tage
(V)
Time (sec)
Embedded Surface
Sine wave Freq= 20Hz
Inertial-Type Shaker
Sine wave Sweep Freq= 5Hz-70Hz
Inertial-Type Shaker
Impact Hammer
Contact Type Shaker Contact Type Shaker
Sine wave Freq= 20Hz
43
(a)
(b) (c) Figure 10: Variation of (a) natural frequency, equivalent stiffness with increasing number of
days for (b) curing period and (c) after curing period.
0
5
10
15
20
25
18 38 58 78 98 118
f n(H
z)
Day
Interval-I Interval-II
0
5
10
15
20
25
30
5 7 9 11 13 15
k(k
N/m
)
Day
Curing Period0
5
10
15
20
25
30
10 32 54 76 98 120
k(k
N/m
)
Day
After Curing Period
Interval-I Interval -II
44
Figure 11: The front view of the notch showing undamaged and different states of damage [see this Fig. in conjunction with Fig. 6(c)].
State 2 State 3 State 4 State 1
45
(a)
Figure 12: Experimental set-up for measuring mode shape of the concrete beam showing (a) element numbers and (b) data acquisition system.
Computer Oscilloscope
PCB Rectifier
QDA1008
Dytran Force Sensor
Beam
Hammer
Element 2
Element 4
Element 9Element 8Element 7
Element 6
Element 5
Element 3
Element 1 1
2
4
6
8
10
12 14 16 18
Element 10
(b)
46
(a)
(b)
Figure 13: (a) Comparison of mode shape and (b) corresponding damage index for undamaged and three damaged states.
Figure 14: Experimental set-up for local EMI technique done using AD5933 (a low-cost alternative of conventional LCR meter).
AD5933 Circuit Board
48
(a) (b) Figure 15: Comparison of typical conductance (G) signature acquired using (a) AD5933 and
(b) LCR meter for undamaged and different states of damage.
0.045
0.046
0.047
0.048
0.049
85 86 87 88 89 90
Con
duct
ance
(G, m
S)
Frequency (kHz)
0.43
0.44
0.45
0.46
0.47
0.48
0.49
85 86 87 88 89 90
Con
duct
ance
(G, m
S)
Frequency (kHz)
Undamaged (State 1)
Damaged (State 3) Damaged (State 2)
Damaged (State 4)
Undamaged (State 1)
Damaged (State 2)
Damaged (State 3)Damaged (State 4)
49
(a)
(b)
Figure 16: Variation of RMSD of conductance plots acquired using (a) AD5933 and (b) LCR
meter at different elements of concrete beam for three states of damage.
1.5
1.7
1.9
2.1
2.3
2.5
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
RM
SD (%
)
Element
Damaged-State 2Damaged-State 3Damaged-State 4
Damage Location
3.0
3.5
4.0
4.5
5.0
5.5
6.0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
RM
SD (%
)
Element
Damaged-Stage 2
Damaged-Stage 3
Damaged-Stage 4Damage Location
State 2
State 3
State 4
50
(a) (b)
Figure 17: (a) Overall set up for power quantification. (b) Details of energy quantification circuit.
R2
R1 PZT V2
I
~ Vin
Power Measuring Circuit (Kaur and Bhalla 2014)
Function Generator (Agilent 33210A)
Amplifier (LDS PA 500L)
Oscilloscope
51
(a) (b)
Figure 18: Variation of PPeak, PRMS and Pavg with varying forcing frequency when (a) Shaker and PZT patch at center and (b) Shaker and PZT patch at an offset of 600 mm from centre.
0.00
0.04
0.08
0.12
0.16
0.20
14 18 22 26 30
Pow
er (µ
W)
Forcing Frequency (f, Hz)
PPeakPRMS
Pavg
0.000
0.005
0.010
0.015
0.020
0.025
10 25 40 55 70
Pow
er (µ
W)
Forcing Frequency (f, Hz)
PPeak
PRMS
Pavg
52
(a) (b)
Figure 19: (a) Full-wave bridge rectifier circuit used for storing energy in capacitor. (b) Charging and discharging voltage across capacitor.
0
20
40
60
80
100
120
0 100 200 300 400 500
Vol
tage
acr
oss C
apac
itor (
mV
)
Time (s)
Capacitor Zener Diode
53
Table 1: Properties of RC beam.
Property Unit Value Length, L m 4.0 Cross-section 0.210 m × 0.160 m Characteristic strength of concrete, fck N/mm2 40 Characteristic strength of reinforcement, fy N/mm2 415 Flexural rigidity modulus (based on fck), EI N-m2 3.9x106
Mass per unit length, m kg/m 84 Poison’s ratio of beam, ν 0.20 Ultimate load carrying capacity, Mu kN-m 10.86
54
Table 2: Properties of PZT patch (PI Ceramic 2013) and bond layer.
Property Unit Value
PZT Size, Lp × wp m2 0.010 × 0.010 Thickness, h m 3.000 × 10-4
