You are on page 1of 18

Construction and Building Materials 114 (2016) 719–736

Contents lists available at ScienceDirect

Construction and Building Materials


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

Continuous health monitoring of pavement systems using smart


sensing technology
Amir H. Alavi ⇑, Hassene Hasni, Nizar Lajnef, Karim Chatti
Department of Civil and Environmental Engineering, Michigan State University, East Lansing, MI 48823, USA

h i g h l i g h t s

 A self-powered sensing approach is proposed for health monitoring of pavement systems.


 Damage detection performance is evaluated with numerical and experimental studies.
 A new miniaturized spherical packaging system is designed for the protection of embedded sensing system.
 Damage localization and quantification is investigated.

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

Article history: Recently, significant attention has been devoted to the utilization of new sensing technologies for pave-
Received 4 November 2015 ment maintenance and preservation systems. This study presents a new approach for the continuous
Received in revised form 9 March 2016 health monitoring of asphalt concrete pavements based on piezoelectric self-powered sensing technol-
Accepted 23 March 2016
ogy. The beauty of this technology is that the signal sensed by the piezoelectric transducers from traffic
Available online 6 April 2016
loading can be used both for empowering the self-powered sensors and damage diagnosis. Numerical and
experimental studies were carried out to evaluate the damage detection performance of the proposed
Keywords:
self-sustained sensing system. A three-dimensional finite element analysis was performed to obtain
Pavement health monitoring
Self-powered wireless sensor
the pavement responses under moving tire loading. Damage was introduced as bottom-up fatigue cracks
Fatigue cracking at the bottom of the asphalt layer. Thereafter, features extracted from the dynamic strain data for a
Damage number of sensing nodes were used to detect the damage progression. The laboratory tests were carried
out on an asphalt concrete specimen in three point bending mode. For the protection of the embedded
sensors, a new miniaturized spherical packaging system was designed and tested. Based on the results
of the numerical study, the sensing nodes located along the loading path are capable of detecting the
damage progression. Besides, the experimental study indicates that the proposed method is efficient in
detecting different damage states including crack propagation. Finally, the possibility of localizing the
damage and quantifying its severity was investigated and discussed.
Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction [1]. The external evaluation methods have been commonly used
for the evaluation of surface distresses. Typical examples in this
Pavement health monitoring plays a key role in pavement man- context are using image analysis techniques to analyze the pave-
agement systems. Early repair and maintenance scheduling ment distress [2,3], or stereo-imagery for measuring pavement
increase the safe operation and in-service performance of pave- deformation [4]. Besides, there are numerous nondestructive eval-
ment. This can be achieved through an accurate and consistent uation (NDE) methods for the assessment of assess the behavior of
monitoring of pavement condition. In general, the existing pavements and other infrastructures [5–18]. The in-situ pavement
approaches for pavement health monitoring can be divided into sensing methods have been the focus of many studies for the last
external evaluation technologies and in situ pavement sensors decades as alternatives to the traditional monitoring [19–22]. Dif-
ferent type of sensors can be used in this domain such as pressure
cell, deflectometer, strain gauge, thermocouple, moisture sensor,
⇑ Corresponding author. fiber-optic sensors, etc [23–34]. Moreover, there are several full-
E-mail addresses: alavi@msu.edu, ah_alavi@hotmail.com (A.H. Alavi), hasniha1 scale test studies to measure the in situ pavement responses under
@msu.edu (H. Hasni), lajnefni@egr.msu.edu (N. Lajnef), chatti@egr.msu.edu
(K. Chatti).
traffic load [35–37].

http://dx.doi.org/10.1016/j.conbuildmat.2016.03.128
0950-0618/Ó 2016 Elsevier Ltd. All rights reserved.
720 A.H. Alavi et al. / Construction and Building Materials 114 (2016) 719–736

A major limitation of the traditional wired sensors pertains to compressed as a function of cumulative time at each load level. This
their deployment and maintenance. More, managing huge amount drawback results in a difficulty in the interpretation of the data
of data generated by a dense array of wired sensors is very chal- generated by SWS [64]. More recently, Alavi et al. [65] proposed a
lenging and costly [38]. To cope with these limitations, wireless data interpretation system integrating finite element method
sensor networks (WSNs) are increasingly utilized as alternatives (FEM) and probabilistic neural network (PNN) for the detection
to traditional structural engineering monitoring systems. The sig- damage progression in gusset plates based on the SWS data.
nificant capability of WSNs for sensing the physical state of the This study presents a new system for the continuous long-term
structural systems has attracted considerable attention in recent health monitoring of pavement structures based on the SWS data.
years [39–46]. WSNs are not merely monitoring systems but also The proposed approach uses features extracted from the cumula-
autonomous data acquisition nodes [47–49]. Dense arrays of tive time strain distributions at preselected discrete levels. The
low-cost smart wireless sensors can offer useful data about the finite element model of an asphalt concrete pavement layer was
structural deterioration. Such information can be efficiently used used as the representative of the real structure. The main goal
to enhance the performance of the pavement condition monitoring was to detect the fatigue cracking due to excessive tensile strain
systems [38,50,51]. Recent development and applications of smart at the bottom of the asphalt concrete. In order to analyze the
sensors and sensing systems in infrastructural engineering can be response of the sensing system embedded within the asphalt layer,
found in [38,52–56]. However, a significant concern for the appli- a series of tests were conducted on an asphalt concrete beam under
cation of wireless sensors is about their power supply. Nearly all a three point bending configuration. The results indicate that the
of the commercially viable sensors for structural health monitoring proposed system can be efficiently used for the pavement health
(SHM) require an external power source, either battery or solar monitoring.
power [57]. Periodic replacement of batteries for embedded sen-
sors or use of solar power technology would be cost-prohibitive
2. The proposed pavement health monitoring system
and in some cases impractical. This issue becomes more challeng-
ing for the continuous long-term monitoring of pavement struc-
Damage detection algorithms play a key role in the SHM sys-
tures. Harvesting ambient energy seems to be an attractive
tems. These algorithms are used for the analysis of raw sensor data
solution for tackling this problem [54,58–61]. Energy harvesting
and subsequently for damage diagnosis [67]. The SWS-based dam-
devices can convert mechanical energy into electrical energy
age detection procedure proposed in this study includes three
[62]. These micro-power generators can be used and integrated
major phases: (1) structural simulation with finite element method
with the monitoring system. Among various self-powering energy
(FEM) (2) generation of data and feature extraction based on the
sources, piezoelectric transducers are proved to be one of the most
outputs of the SWS memory cells, and (3) finding a reasonable rela-
efficient choices [57–64]. For pavement health monitoring, piezo-
tionship between the probability density function (PDF) parame-
electric transducers can be used for the self-powering of wireless
ters obtained from strain distribution, and damage progression.
sensors by harvesting energy from the mechanical loading experi-
First, a damage scenario was defined for the given pavement struc-
enced by the pavement [64]. Recently, the authors at Michigan
ture via the FE simulations. Subsequently, the cumulative time of
State University (MSU) have developed a new class of self-
occurrences at predetermined strain levels are determined for
powered wireless sensors (SWS) [63–65]. The designed SWS is a
the data acquisition points (sensors). The strain distribution in
small size battery-less sensor. The prototype of this miniaturized
each sensing node was used for defining damage indicators.
strain-senor is shown in Fig. 1.
Besides, a major challenge in application of wireless sensors results
This unique sensor is based on the integration of the piezoelec-
from the fact that damage in structures is an intrinsically local
tric transducers with an array of ultra-low power floating gate
phenomenon. Thus, sensors that are close to the damaged site
computational circuits [64]. By embedding these sensors inside
are more influenced than those remote to the damage site. The
the pavement, it is possible to monitor the localized strain
only existing solution to effectively detect damage at an arbitrary
statistics. The recorded information can be used for early damage
location in a structure is to densely distribute the sensors through-
detection and future condition evaluation. Research in the previous
out the structure [38]. Obviously, this is not an optimal and
FHWA funded project revealed the applicability of the SWS for con-
economic solution. To cope with this issue, this study proposes
tinuous monitoring of infrastructures [63–66]. That project was
another strategy based on the effect of array of sensors.
basically focused on the manufacturing of the sensor electronics,
and design of a packaging system to withstand loading and envi-
ronmental conditions for the pavement implementation. A limited 2.1. Smart sensor
study was done on developing a method for predicting remaining
fatigue life of pavement and generating missing data from a set The new smart SWS made at MSU is capable of continuously
of measurements by a classical statistical technique [64]. Despite monitoring of strain events within the host structure. It uses only
several advantages of using SWS, there would be a considerable self-generated electrical energy harvested directly from the sens-
loss of information. In fact, a part of the sensed information is ing signal induced by a piezoelectric transducer connected to the
pavement. This sensor can be economically attached to pavement
structures either during construction, or anytime during routine
Interface Piezoelectric maintenance operations. The communication between the sensor
Board Transducer
and a service vehicle is done using Radio Frequency Identification
(RFID) technology. Previous study showed that the sensor can be
read using an RFID scanner from a distance of 16 in. (406.4 mm)
D = 12mm
[64]. For a sensing system embedded inside an asphalt specimen,
the thickness of the asphalt layer affects the accuracy of the trans-
mitted and recovered signal due to its viscoelastic properties. In
Sensor
Electronics this case, the sensor and RFID scanner can communicate from a
distance of about 12 in. (304.8 mm) [64]. The convertible electrical
power levels in structures are typically less than 1 lW. Given the
Fig. 1. The prototype of the SWS system. strain levels observed in pavements, it is believed that the available
A.H. Alavi et al. / Construction and Building Materials 114 (2016) 719–736 721

harvestable energy is also around 1 lW. The sensor uses novel ana- relationship between the PDF parameters (l and r) and damage
log signal processing circuits that require less than 1 lW of power state. The major advantage of the proposed method is that it is
[66]. More details about this smart sensor can be found in [64]. based on relative variations in the strain response distributions.
However, it is known that the piezoelectric transducers have This means that there is no need to measure damage directly. The
the ability to convert the mechanical applied charge to an electrical effect of damage is ‘‘sensed” by the sensors, and it is evaluated
charge using the direct piezoelectricity effect. The open source based on relative shifts of PDFs over time. This is schematically
voltage (V) generated across the piezoelectric ceramic transducer shown in Fig. 2 and will be proved in this study. When the sensors
material is given by the following equation [63]: are deployed in a network, the sensors that are closer to critical
locations will experience more prominent shifts in the distributions
SYd31 h compared to other sensors. Knowing the exact location of these sen-
V¼ ð1Þ
e sors gives the locations of the critical damage areas.
where S, Y, d31, h and e, are the applied strain, Young’s modulus of The analyses carried out in this study were divided into two
the piezoelectric material, piezoelectric constant, thickness, and the main stages. First, a 3D FE model was developed to obtain the
electrical permittivity, respectively. The sensor has a series of mem- response of the asphalt pavement under moving load. The main
ory cells that cumulatively store the duration of strain events, at a goal was to detect the fatigue cracking due to excessive tensile
preselected level discretization. If the voltage generated by the strain at the bottom of the asphalt concrete. Then, an experimental
piezoelectric transducer due to pavement loading exceeds a prese- study was carried out to evaluate the performance of the sensors
lected threshold, the memory cell corresponding to that threshold embedded inside an asphalt concrete slab.
will start measuring the duration of the event. At constant loading
frequency, the sensor output is the cumulative histogram of the 2.2. Finite element modeling of pavement
loading strain distribution [65]. In fact, the sensor does not directly
measure the absolute value of strain in order to estimate damage. A 3D FE model was developed to analyze the dynamic response
The rate of variation of strain distributions is related to the rate of of the pavement under a moving truck tire loading. ABAQUS was
damage. The whole methodology is based on relative damage. A used for the modeling and post-processing of the results. A series
representation of the level crossing cumulative time counting of acquisition nodes were considered at the bottom of the asphalt
implemented by the developed SWS can be found in [65]. It has layer as the potential sensors. The studied pavement is composed
been shown that the sensor output can be characterized by the fol- of 3 layers: asphalt, base and subgrade layers. The FE model is
lowing cumulative density function (CDF) [65]: shown in Fig. 3.
  
a el
FðeÞ ¼ 1  erf pffiffiffi ð2Þ 2.2.1. Geometry
2 r 2 The model has a dimension of 6200 mm (244 in.) along the
direction of traffic and 4170 mm (164 in.) along the transverse
where, l, r and a are mean of cumulative time distribution, stan-
direction (the width of one lane). A model of such size was used
dard deviation accounting for the load and frequency variability,
to minimize any edge effect errors, especially on the longitudinal
and total cumulative time of the applied strain. l and r of the strain
tensile strain.
distribution can be regarded as indicators of the damage progres-
sion [65]. In fact, l and r are the only viable tools to define the
SWS output data. These parameters can be obtained by a curve fit- 2.2.2. Material properties and thickness of layers
ting of the sensor output distribution collected from the entire Table 1 shows the material properties and thickness of layers of
memory cells. More descriptions about the curve fitting procedure the pavement system.
to determine l and r can be found in [65]. Consequently, the dam-
age state is logically considered to be a function of l and r. In the 2.2.3. Mesh
present study, the cumulative time of occurrences at predetermined A schematic illustration of the FE mesh is illustrated in Fig. 4.
strain levels are determined for the sensors. Then, the correspond- The model was idealized with linear hexahedral element of type
ing l and r values for each damage scenario were obtained through (C3D8R). On the basis of a sensitivity analysis, the mesh dimen-
the fitting of a CDF (Eq. (2)) to the cumulative duration of strain sions were considered about 12.7 mm (0.5 in.) in the loading area
events. Then, more in-depth analyses were conducted to find the and 25.4 mm far from the contact zone. The total number of

Fig. 2. A schematic representation of the PDF shifts due to damage progression.


722 A.H. Alavi et al. / Construction and Building Materials 114 (2016) 719–736

Tire

Step 1

Step 2

Step 3 Moving Direction

Fig. 5. A schematic representation of the moving load modeling.

Asphalt Layer
Fig. 3. The 3D FE model for the pavement structure. Strain measurement

2 inches
0.5 inch
Table 1
Material properties and layer thickness. Crack Zone

Layer E m Thickness
Base Layer
Asphalt layer 2757.9 MPa (400 ksi) 0.35 152.4 mm (6 in.)
Base 344.7 MPa (50 ksi) 0.35 127 mm (5 in.)
Subgrade 34.55 MPa (5 ksi) 0.48 4826 mm (190 in.)

Subgrade

Fig. 6. Crack zone and measurement location.

the element can be used to calculate the step time. In this study,
it was supposed that the vehicle runs at a constant speed of
6 mph (about 10 km/h). The tire pressure was taken 0.69 MPa
(100 psi). The load was moved a 76.2-mm (3 in.) in each increment
over 6 elements. The total number of increments (locations of the
load) required to achieve one full passage of the tire over the entire
Fig. 4. The FE model mesh. model was 32. More details about the procedure followed in this
study to simulate the moving load can be found in Al-Qadi and
elements was 615,076 elements of type C3D8R. Due to the high Wang [68].
number of degrees of freedom of the model, a supercomputer
was used to run the Abaqus simulations for both the intact and
damaged models. The available servers with high performance 2.2.6. Location of sensors and damage zone
computing at Division of Engineering Computing Services (DECS) Fig. 6 shows the location of the sensors. As it is seen, the sensors
at MSU were used for this aim. Using a server with 384 GB RAM were located at a distance of 50.8 mm (2 in.) from the bottom of
and 72 logical RAM, the simulation time for the intact configura- the asphalt layer. Fig. 7 illustrates the layout of sensors inside
tion took around 42 h. the pavement. The distance between two consecutive sensors is
304.8 mm (12 in.). Also, distance of Sensors 1 and 9 from vertical
boundaries was equal to 1880 mm (74 in.). In this analysis, 55 sen-
2.2.4. Boundary conditions sors were used to measure the longitudinal and transversal strains.
The vertical or horizontal movements at the bottom of the layer
were restrained. Also, there was no horizontal displacements per-
pendicular to the perimeters of the pavement while the rotation
was free. Furthermore, to simulate a new pavement condition,
the interface between different layers had no relative movement.

2.2.5. Loading
To simulate the movement of the load at the desired speed, a
quasi-static analysis was adopted [68]. The location of the load
and its amplitude were gradually shifted over the loading path at
each step until a single wheel pass is completed (Fig. 5). Based
on the concept of a continuously moving load, an element is
increasingly loaded when the load approaches and then unloaded
as the load leaves it [68,69]. The traffic speed and the length of Fig. 7. Layout of data acquisition nodes.
A.H. Alavi et al. / Construction and Building Materials 114 (2016) 719–736 723

(a) 100 (b) 100


Intact Intact
80 80
Damaged Damaged

Transverse Strain (μɛ)


Transverse Strain (μɛ)
60 60

40 40

20 20

0 0
0 0.25 0.5 0.75 1 0 0.25 0.5 0.75 1
-20 -20
Time (sec) Time (sec)
(a) Sensor 3 (b) Sensor 4
(c) 100 (d) 100
Intact Intact
80 80
Damaged Damaged

Transverse Strain (μɛ)


Transverse Strain (μɛ)

60 60

40 40

20 20

0 0
0 0.25 0.5 0.75 1 0 0.25 0.5 0.75 1
-20 -20
Time (sec) Time (sec)
(c) Sensor 5 (d) Sensor 6

(e) 100
Intact
80 Damaged
Transverse Strain (μɛ)

60

40

20

0
0 0.25 0.5 0.75 1
-20
Time (sec)
(e) Sensor 7
Fig. 8. Comparison of the transverse tensile strains for the intact and damaged pavement.

Based on the FE simulations, the strain values for sensing nodes far 2.3. Laboratory testing of the embedded system
from the loading path are very low. Thus, it is not feasible for the
sensors to detect the change. Consequently, out of the available The performance of the sensors was tested through an experi-
sensors, only those along the loading path which experienced mental study on an asphalt concrete specimen. The sample was
strains higher than 20 le were kept for the analysis. These are tested under a three point bending configuration. The test setup
Sensors 1–9 located under the loading path. Since the tire loading is schematically shown in Fig. 10. As it seen, the test fixture is sim-
imprint area was started from the location of Sensor 1 and termi- ilar to that of Single Edge Notched Beam SE(B) Test which is one of
nated by Sensor 9, only the results for Sensors 3 and 7, located at an the most documented methods to determine fracture properties
adequate distance from loading points, were included in the [70,71]. However, the size of the sensor packaging (about 1 in.)
analysis. Later, a new FE model was developed with a damage zone limits the use of recommend by ASTM E1820-09 [70] for the beam
at the bottom of the asphalt layer. The length of the crack zone was geometry in the SE(B) method. In this study, the fixture was devel-
0.5 in. from the bottom of the asphalt layer (Fig. 6). The location of oped for an asphalt concrete sample with a span length of 381 mm
the sensors was the same as the intact model. Sensor 5 is above the (15 in.), thickness of 165.1 mm (6.5 in.) and a width of 152.4 mm
damaged area. (6 in.) (Fig. 11). The rollers have a diameter of 30 mm (1.2 in.)
and are setup to be free in rotation. The slab was built using the
HMA, 4E1 mixture type provided by Michigan Department of
2.2.7. FE simulation results Transportation (MDOT). The sample was compacted using PReS-
Figs. 8 and 9 show the transverse and longitudinal strains for BOX Asphalt Shear-box Compactor. The temperature was around
the intact and damaged pavements. As it is seen, the damaged 150 °C (302 °F). The weight of the hot mix asphalt (HMA) was
pavement experiences higher strains in all sensor locations. This 25 kg and the length of the slab was equal to 450 mm (17.75 in.).
is more evident for Sensor 5 which is located above the damage The center load point was designed to provide a uniform loading.
zone. The three point bend tests were carried out using an MTS machine.
724 A.H. Alavi et al. / Construction and Building Materials 114 (2016) 719–736

(a) 80 (b) 80
Intact Intact

Longitudinal Strain (μɛ)

Longitudinal Strain (μɛ)


60 60
Damaged Damaged
40 40

20 20

0 0
0 0.25 0.5 0.75 1 0 0.25 0.5 0.75 1
-20 -20

-40 -40
Time (sec) Time (sec)
(a) Sensor 3 (b) Sensor 4

(c) 80 (d) 80
Intact Intact
Longitudinal Strain (μɛ)

Longitudinal Strain (μɛ)


60 60
Damaged Damaged
40 40

20 20

0 0
0 0.25 0.5 0.75 1 0 0.25 0.5 0.75 1
-20 -20

-40 -40
Time (sec) Time (sec)
(c) Sensor 5 (d) Sensor 6
(e) 80
Intact
Longitudinal Strain (μɛ)

60 Damaged

40

20

0
0 0.25 0.5 0.75 1
-20

-40
Time (sec)
(e) Sensor 7
Fig. 9. Comparison of the longitudinal tensile strains for the intact and damaged pavement.

Load
Intact : Intact plateða ¼ 0 mmÞ

Damage 1 : a ¼ 22:2 mmð7=8 in:Þ

Damage 2 : a ¼ 31:75 mmð1 1=400 Þ


For this stage, different damage states were defined to evaluate
the sensitivity of the sensors to the damage severity. After introduc-
ing the second damage phase, the displacement was increased to
Fig. 10. Single edge notched beam test [71]. 0.3 mm to evaluate the behavior of the sample for higher
amplitudes. After a number of cyclic loadings, a crack propagation
phenomenon was observed. The crack length was 12.7 mm
All tests were conducted under constant axial displacement rate (0.5 in.). This new damage phase was considered as Damage 3.
control. The tests were done at 2 and 5 Hz loading frequency for Accordingly, the total of length of crack for Damage 3 was
0.1, 0.15 and 0.2 mm amplitudes. Before starting the test, a preload 44.45 mm (1 3/400 ). Fig. 12 illustrates the captured crack growth in
equal to 0.5 kN was applied to the sample to ensure it is seated on the test.
the fixture. Thereafter, the cyclic displacements were applied.
The main goal was to evaluate the response of the embedded 2.3.1. Design and implementation of a small scale packaging system
piezoelectric transducers due to damage progression. Damage Current installation procedures demand considerable care dur-
was introduced by making a notch at the bottom of the asphalt ing construction. This is to insure that the commonly-used H-
layer using a circular saw. The damage states were defined by gages are properly bonded to the pavement surface layer, and are
increasing the notch size (a) as follows: properly aligned in both horizontal and vertical directions. For the
A.H. Alavi et al. / Construction and Building Materials 114 (2016) 719–736 725

(a) Three point bending configuration

(b) A typical notched specimen representing Damage 2


Fig. 11. Test setup and sensor locations.

when the epoxy is strained, an induced axial loading is applied


to the piezo. Therefore, choosing an appropriate epoxy resin to
manufacture the spherical casing is an important issue. There
should be stiffness compatibility between the epoxy and the host
material (HMA). Stiffness compatibility can be achieved using
epoxy with a lower stiffness than the host material. Also, non-
metallic materials should be used so that it allows for wireless
12.7 mm communication. Among different epoxies tested (e.g. AralditeÒ
GY-6010 and EPIC S7514), ConathaneÒ TU-981 epoxy was found
to perform better. The robustness of the ConathaneÒ TU-981 epoxy
to withstand harsh environmental conditions has been already
verified [64]. Thus, this epoxy was used for encasing the proposed
spherical packaging system. ConathaneÒ TU-981 is a
two-component liquid casting system that produces a 65-Shore
Fig. 12. The crack propagation phase during the test (Damage state 3).
D elastomer of exceptional toughness and extraordinary process-
ing flexibility. Table 2 shows the basic properties of this epoxy.
purpose of this study, the packaging system was miniaturized so The elastic modulus of the ConathaneÒ TU-981 epoxy was
that it can be tossed in the pavement material during construction obtained through an experimental study. The MTS machine was
or can be used within a mesh network distributed over the base layer to this aim. The dimension of the Epoxy TU-981 sample was
(Fig. 13). The size of the designed spherical packaging system was of 40  35  16 mm (length  width  height). The tests were done
the same order of a coarse aggregate particle. All the electronics using cyclic loading on two directions of epoxy specimens
already have a small size and the antenna can be miniaturized to (Fig. 14). Fig. 16 shows the stress and strain curves for the epoxy
fit the desired size. The spherical packaging systems were designed material. The calculated elastic modulus is shown in Table 3. As
using SolidWorks software and built by a 3D printer. The manufac- it is seen, the elastic modulus of Epoxy TU-981 is 258.8 MPa.
tured mould is shown in Fig. 14. The top part has an opening to Besides, the survivability of the epoxy in high pressure and tem-
put epoxy into the mould. The two parts combine together as a perature during asphalt pavement compaction was verified using
sphere with six anchor legs in three directions (see Fig. 15). the Asphalt Shear-box Compactor. For brevity, the results for this
The RFID, antenna and electronics are fragile and sensitive phase are not presented.
parts. They also have very low melting temperature. As a result, Fig. 17 presents the manufacturing process of the packaging
they cannot survive alone in asphalt pavement because asphalt system with embedded piezoelectric transducers. The epoxy resin
pavement is under high pressure and temperature during con- was cured for 24 h at room temperature. A total of 3 lead zirconate
structions. Epoxy materials have relatively high melting tempera- titanate (PZT) ceramic transducers with spherical packaging were
ture and excellent flexibility. Thus, epoxy materials can be used embedded within the asphalt layer in slab compactor. The PZT
as the protectors for the sensor electronics. On the other hand, properties are shown in Table 4.
726 A.H. Alavi et al. / Construction and Building Materials 114 (2016) 719–736

Fig. 13. The designed sensing system in a spherical casing.

The spheres were located at a distance of approximately 2 in.


from the bottom of the layer. The first PZT was placed at mid-
span and the remaining ones were located at a distance of
76.2 mm (3 in.) from the middle PZT. The layout of the PZTs is
shown in Fig. 18. After compaction, the PZT output voltage was
read on NI 9220 data acquisition system (with 1 GX impedance)
in parallel with a resistor with an impedance similar to the self-
powered wireless sensors (50 MX).

2.3.2. Comparison of the maximum voltage delivered by PZTs for


different damage states
The preliminary tests showed that PZT 1 located at the middle
of the slab was not delivering voltage. A probable reason was
Fig. 14. Top part (right) and bottom part (left) mould.
failure of the PZT or the connection wires during the compaction

Fig. 15. Elastic modulus test (Direction 1 and direction 2).


A.H. Alavi et al. / Construction and Building Materials 114 (2016) 719–736 727

Table 2
The basic properties of ConathaneÒ TU-981 epoxy.

TU-981 Part A Part B PART B Black Test method


Color Light Amber Dark Amber Black Visual
Appearance Translucent Translucent Opaque Visual
Viscosity 25C, cps 2800 190 190 ASTM D2393
% NCO 12 N/A N/A N/A
Specific gravity @ 25C 1.05 1.06 1.06 ASTM D792
Cured flex strength 13.1 MPa

Fig. 16. Stress strain curve of Epoxy TU-981.

Table 3
Test results for the elastic modulus of Epoxy TU-981.
minimum level of strains to be captured by piezoelectric transduc-
ers was about 20.00 le. On the other hand, the maximum of the
Epoxy TU-981 Test 1 Test 2 Test 3 strain value extracted from the FE simulations was about 97 le.
Elastic modulus (MPa) 248.7 265.7 262.1 Considering a reasonable value of 10 strain levels for the gates,
Average (MPa) 258.8 the difference between the strain levels is 8.56 le. Table 5 presents
the preselected strain levels for the pavement layer. Using the
strain histories of all elements, a script is written in MATLAB to
process. Accordingly, the results for the other PZTs were recorded perform the following tasks:
and presented in Figs. 19 and 20. As can be observed in these fig-
ures, the PZTs are experiencing higher strains and therefore out- a) Takes the strain-time data from Abaqus and measures the
putting higher voltage by increasing the notch size. duration of events at the strain levels defined in Table 5.
b) Fits CDF given in Eq. (2) to the cumulative time of occur-
rences at predetermined strain levels obtained from the first
3. Damage detection using smart sensing technology
step.
c) Reports the l and r for the data acquisition nodes for the
This section presents the damage detection results based on the
intact and damaged models.
information sensed by the SWS. First, the results obtained by the
FE model are analyzed. Thereafter, the performance of the sensor
Fig. 21 shows the cumulative time at predefined strain levels
is verified using the laboratory test results.
versus the gate numbers, along with the fitted CDF curve. Using
the l and r values, the PDF plot corresponding to each sensor
3.1. Damage detection: FE model was obtained and shown in Figs. 22 and 23. As can be seen in these
figures, l decreases and r increases by transitioning from intact to
As discussed before, a series of sensing nodes was considered damaged mode. Accordingly, the PDFs shift to left, their width
near the bottom of the asphalt layer to detect the damage progres- increases and their height decreases due to damage progression.
sion. Fatigue cracking is one of these distresses and it is caused by It is evident that for Sensor 5 which located above the crack, there
excessive tensile strain at the bottom of the asphalt concrete is a distinct change in the PDF shape. It is worth mentioning that
[72,73]. Considering the importance of tensile strain in the predic- even for sensors far from the damage, the PDFs still shift to left
tion of fatigue cracking, this type of strain was measured for the and expand due to damage progression.
analysis. Referring to Section 2.2.6, the results for sensors with a As discussed in Sections 3.1, it is possible to detect the
reasonable distance from loading points were included, i.e., Sen- damage progression with individual (point) sensors. The point
sors 3 and 7. A smart pebble wireless sensor has a series of memory sensors provide discrete and localized measurements. Although
cells (gates). Each of these gates cumulatively stores the duration they deliver valuable quantitative information, measurements
of strain events at a preselected levels, experienced at the sensing at a single location might not be sufficient for accurate damage
node location. In general, the number of gates is dependent on the detection [74,75]. Rationally, the closer the sensors to the notch
nature of the problem and the material. In this study, 10 strain are, the higher the damage detection accuracy is. For the cases
levels were defined to efficiently cover the lower and upper limits where the potential sensors are at a far distance from the
of the strain values extracted from the FE simulations. The imposed damage, the sensor may not have a sense of damage
728 A.H. Alavi et al. / Construction and Building Materials 114 (2016) 719–736

Fig. 17. Manufacturing process of spherical packaging.

Table 4
The PZT properties.

Parameter Value
Diameter 20 mm (a) 1.00
Thickness 3 mm 0.90 PZT 2 PZT 3
Piezo material SM111 (PZT-4) 0.80
Voltage (volt)

Resonant frequency 690 KHz ± 3% 0.70


0.60
Resonant impedance (Zm) 63.6 X 0.50
Static capacitance (Cs) 1265 pF ± 10% 0.40
Piezoelectric constant d33 320  1012 m/v 0.30
d31 140  1012 m/v 0.20
0.10
Density 7.9 g/cm3 0.00
Electromechanical coupling coefficient Kp 0.58 Intact Damage 1 Damage 2 Damage 3
Kt 0.45
K31 0.34
(a) Displacement: 0.1 mm

(b) 1.50
1.30 PZT 2 PZT 3
Voltage (volt)

1.10
0.90
PZT 2 PZT 1 PZT 3 0.70
0.50
0.30
0.10
-0.10
3'' Intact Damage 1 Damage 2 Damage 3

(b) Displacement: 0.15 mm


(c) 2.00
1.80 PZT 2 PZT 3
1.60
Voltage (volt)

1.40
1.20
1.00
0.80
0.60
0.40
0.20
0.00
Intact Damage 1 Damage 2 Damage 3

(c) Displacement: 0.2 mm


Fig. 19. A comparison of maximum delivered voltage by PZT discs for 2 Hz loading
Fig. 18. Layout of the spherical packaging in the compactor. frequency.
A.H. Alavi et al. / Construction and Building Materials 114 (2016) 719–736 729

(a) 1.00 variation of the STD of l and r of group of sensors was taken
0.90 PZT 2 PZT 3 into account. Fig. 23 presents the variation of the STD of l
0.80
and r of all of the sensors. As it is seen, the STD of l and r
Voltage (volt)

0.70
0.60 of group of sensors increases with damage progression.
0.50
0.40
0.30
0.20
3.2. Damage detection: laboratory testing
0.10
0.00 The damage detection results are provided for amplitude of
Intact Damage 1 Damage 2 Damage 3 0.2 mm and loading frequencies of 2 and 5 Hz. In this study, 7 volt-
(a) Displacement: 0.1 mm age levels were defined to efficiently cover the lower and upper
limits of the strains experienced by the specimen. The minimum
(b) 1.50 and maximum voltage values outputted by piezoelectric transduc-
1.30 PZT 2 PZT 3 ers were 0.35 and 1.73 V for the tested loading amplitudes. 7 volt-
1.10 age levels were taken for the gates. The minimum and maximum
Voltage (volt)

0.90
voltage thresholds were set to 0.2 and 2 V, respectively. Accord-
0.70
0.50
ingly, the difference between the strain levels is 0.3 V. Table 7 pre-
0.30 sents the preselected voltage levels for the pavement layer. It
0.10 should be noted that the voltage supplied by piezoelectric trans-
-0.10 ducers is proportional to experienced strain [54]. Consequently,
Intact Damage 1 Damage 2 Damage 3
using voltage or corresponding strain values does not affect the
(b) Displacement: 0.15 mm damage analysis trends. However, the recorded voltage histories
of PZTs 2 and 3 were inputted into a script in MATLAB to find
(c) 2.00 the l and r of the strain distribution in the intact and damaged
1.80 PZT 2 PZT 3
1.60 models. The l and r values were then used to plot the PDFs corre-
Voltage (volt)

1.40 sponding to each sensor (Figs. 24 and 25). For brevity, the cumula-
1.20
1.00 tive time versus gate numbers and the fitted CDF curves are not
0.80 presented. Also, since there were merely 2 sensors were involved,
0.60
0.40
performing the group effect analysis was not done. As can be seen
0.20 in Figs. 24 and 25, the values of l and r, respectively, decrease and
0.00
increase due to damage progression. This is in accordance with
Intact Damage 1 Damage 2 Damage 3 what previously observed for the FE model. Interestingly, the PDFs
(c) Displacement: 0.2 mm changes notably by transiting from Intact to Damage 1 compared
to other stages. This is because the first introduced notch was dee-
Fig. 20. A comparison of maximum delivered voltage by PZT discs for 5 Hz loading per than those considered for Damage states 2 and 3. Additionally,
frequency. the crack propagation process was accurately detected by the
sensors.

Table 5
The preselected strain levels considered for the analysis. 4. Discussion
Gate number Strain level (le)
A major problem with utilization of the existing wireless sen-
1 20.00
2 28.56
sors is providing a reliable power source for them. To cope with
3 37.11 this concern, a novel approach has been developed in this study
4 45.67 for the continuous battery-less monitoring of structures. The
5 54.22 results for the numerical and experimental study clearly indicate
6 62.78
that the outputs of the developed Self-powered sensor are damage
7 71.33
8 79.89 indicators. The main observation from both of the FE and labora-
9 88.44 tory studies was that the PDFs obtained from the cumulative time
10 97.00 histograms change reasonably with damage progression. The
expansion of PDFs and their shift to left side was shown as a sign
of damage occurrence. In general, changes of r seem to be a better
or the results may not have a reasonable trend. Thus, a poten- sign of damage progression. On the other hand, the experimental
tially more effective strategy was defined to improve the damage study proved that the designed miniaturized sensor packaging
detection performance through spatial measurements over an withstands temperature and pressure during compaction. Such
area of the plate. On this basis, it was decided to use the infor- small size system can be easily used and integrated in large-scale
mation provided by a group of sensors, termed as ‘‘group effect”. sensor networks by State highway agencies. However, in addition
In this case, even if one sensor does not sense the damage, the to its damage detection capability, the proposed approach gives
group effect will help detect the damage. The main goal was an insight into the damage localization and quantification. To clar-
to find the relationship between the PDF parameters of a group ify this issue, the percentage of variation of damage indicator
of sensors and damage progression. To this aim, average, stan- parameters for the intact and damaged states in the FE simulations
dard deviation (STD), range, minimum, maximum, skewness, is presented in Table 6 and Fig. 26. As it is seen, the highest varia-
and kurtosis of the PDF parameters (l and r) for the group of tion belongs to Sensor 5 which is located above the crack zone. The
sensors (Sensors 3–7) were obtained and normalized. Based on variation has a proportionally increasing trend as the load
preliminary analyses, the average, range, minimum, maximum, approaches the damage zone (either from Sensors 3 and 4, or from
skewness, and kurtosis of the PDF parameters did not have a Sensors 7 and 6 up to Sensor 5). This is valid for both of the l and r
sound relationship with the damage progression. Thus, only the trends.
730 A.H. Alavi et al. / Construction and Building Materials 114 (2016) 719–736

Fig. 21. Cumulative time versus gate numbers for intact and damaged states.

Damage quantification performance can be evaluated for the and 5 Hz loading frequencies are, respectively, shown in Tables 8
experimental results as more damage states are involved. To this and 9. Furthermore, the percentage of changes of these values by
aim, variations of damage indicator parameters (l and r) for 2 transiting from a mode to the consecutive mode is illustrated in
A.H. Alavi et al. / Construction and Building Materials 114 (2016) 719–736 731

(a) Sensor 3 (b) Sensor 4

(c) Sensor 5 (d) Sensor 6

(e) Sensor 7
Fig. 22. Changes of PDFs due to damage.

Figs. 27 and 28. It should be noted that the notch size was = 12.7 mm for Damage 2 to Damage 3 (crack propagation) mode.
increased 22.2 mm for Intact to Damage 1 mode, (31.75–22.2) Accordingly, the most severe damage state belonged to the
= 9.55 mm for Damage 1 to Damage 2 mode, and (44.45–31.75) Intact-Damage 1 mode followed by the Damage 2-Damage 3 and
732 A.H. Alavi et al. / Construction and Building Materials 114 (2016) 719–736

(a) (b)
1.4 1.4

Normalized STD of σ of All Sensors

Normalized STD of μ of All Sensors


1.2 1.2

1.0 1.0

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0.0 0.0
Intact Damaged Intact Damaged
Fig. 23. The STD of r and l of group of sensors.

(a) Sensor 2 (a) Sensor 2

(b) Sensor 3 (b) Sensor 3

Fig. 24. Changes of PDFs due to damage progression for 2 Hz loading frequency. Fig. 25. Changes of PDFs due to damage progression for 5 Hz loading frequency.
A.H. Alavi et al. / Construction and Building Materials 114 (2016) 719–736 733

(a) designate the difference between increasing the notch size by


800%
12.7 mm or 9.55 mm. Another observation from Figs. 27
(a) and 28(a) is that variations of l do not provide reasonable infor-
Variation Percentage of μ

600% mation for the damage quantification.

400% 5. Conclusions and future directions

200% This paper presents the performance analysis of a newly


developed self-powered wireless sensor for the continuous
health monitoring of asphalt concrete pavements. The unique-
0% ness of this sensor pertains to the fact that there is no need to
Sensor 3 Sensor 4 Sensor 5 Sensor 6 Sensor 7 directly measure the absolute value of strain in order to estimate
damage. The rate of variation of strain distributions is related to
(b) the rate of damage. Thus, the whole methodology is based on
140%
relative damage. However, a 3D FE model was developed to ana-
Variation Percentage of σ

120% lyze the response of intact and damaged asphalt pavement struc-
tures under moving vehicular loading. The moving load was
100%
simulated using a quasi-static approach. An extensive experi-
80% mental study was further conducted to assess the damage detec-
60% tion performance of the self-sustained long term sensing system.
The laboratory tests were carried out on an asphalt slab in three
40%
point bending mode. A new miniaturized spherical packaging
20% system was designed and manufactured to encase the sensors.
0% Different nonmetallic epoxies were tested to find the most
Sensor 3 Sensor 4 Sensor 5 Sensor 6 Sensor 7 appropriate one for manufacturing of the packaging system.
The piezoelectric transducers were placed inside the spherical
Fig. 26. Percentage of variation of damage indicator parameters for the intact and packaging and the entire package was embedded inside the
damaged states. specimen. The following conclusions can be derived from the
results presented in this research:
Damage 1-Damage 2 modes. As can be observed from Figs. 27
(b) and 28(b), variation of r can be regarded as a good indicator I. The FE simulation results indicate that the proposed system
of damage severity. The variations for the Intact-Damage 1 mode is capable of detecting the bottom-up fatigue cracks intro-
with a deeper notch size are notably higher than the other modes. duced at the bottom of the asphalt layer. The same results
Even for the 5 Hz loading frequency, both of the sensors show a were observed from the laboratory testing. In this case, the
higher variation for the Damage 2-Damage 3 mode compared to sensors could detect different damage states including a
the Damage 1-Damage 2 mode. That is to say, the sensors could crack propagation phase.

(a) 1600%
1400%
Sensor 2
Sensor 3
Variation Percentage

1200%
1000%
of μ

800%
600%
400%
200%
0%
Intact-Damage 1 Damage 1-Damage 2 Damage 2-Damage 3

(b) 200%
180% Sensor 2
Variation Percentage

160% Sensor 3
140%
120%
100%
of σ

80%
60%
40%
20%
0%
Intact-Damage 1 Damage 1-Damage 2 Damage 2-Damage 3

Fig. 27. Percentage of changes of l and r values due to damage progression for 2 Hz loading frequency.
734 A.H. Alavi et al. / Construction and Building Materials 114 (2016) 719–736

(a) 300%
Sensor 2
250% Sensor 3

Variation Percentage of
200%

150%

μ
100%

50%

0%
Intact-Damage 1 Damage 1-Damage 2 Damage 2-Damage 3

(b) 80%
70%
Sensor 2
Variation Percentage of

Sensor 3
60%
50%
40%
σ

30%
20%
10%
0%
Intact-Damage 1 Damage 1-Damage 2 Damage 2-Damage 3

Fig. 28. Percentage of changes of l and r values due to damage progression for 5 Hz loading frequency.

Table 6
Variation of damage indicator parameters of l and r.

1st feature: l 2nd feature: r


Intact Damaged Variation percentage Intact Damaged Variation percentage
Sensor 3 1.43 1.87 31% 4.67 5.09 9%
Sensor 4 0.95 1.82 92% 4.23 5.04 19%
Sensor 5⁄ 0.84 7.10 750% 4.12 9.44 129%
Sensor 6 1.08 1.89 75% 4.36 5.10 17%
Sensor 7 1.54 1.98 29% 4.78 5.19 9%

Sensor 5 is located above the crack zone.

Table 7
The preselected strain levels considered for the analysis. II. It was found that damage progression can be considered as a
function of PDF parameters (l and r) obtained from the
Gate number Voltage level (V)
strain distribution. l and r were derived by curve fitting
1 0.20
of the sensor output distribution collected from the entire
2 0.50
3 0.80 memory cells of the sensor. The important observation from
4 1.10 the individual sensor analysis was that the PDFs shift to left
5 1.40 (l decreases) and their width increases (r increases) due to
6 1.70 the damage progression. Based on the results, variations of r
7 2.00
can be regarded as a better indicator of damage.

Table 8
Variation of damage indicator parameters of l and r for 2 Hz loading frequency.

Displacement 1st feature: l 2nd feature: r


Intact Damage 1 Damage 2 Damage 3 Intact Damage 1 Damage 2 Damage 3
0.2 mm Sensor 2 3.38E01 2.29E02 2.38E01 4.44E01 3.12E01 8.87E01 1.13E+00 1.34E+00
Sensor 3 3.39E01 1.02E02 1.29E01 3.48E01 3.06E01 8.47E01 1.01E+00 1.24E+00

Table 9
Variation of damage indicator parameters of l and r for 5 Hz loading frequency.

Displacement 1st Feature: l 2nd Feature: r


Intact Damage 1 Damage 2 Damage 3 Intact Damage 1 Damage 2 Damage 3
0.2 mm Sensor 2 2.77E01 7.26E02 1.41E01 5.10E01 4.54E01 7.70E01 1.02E+00 1.41E+00
Sensor 3 3.13E01 1.57E01 2.59E02 3.85E01 3.83E01 6.59E01 8.93E01 1.28E+00
A.H. Alavi et al. / Construction and Building Materials 114 (2016) 719–736 735

III. The effect of group of sensors was studied for the numerical [15] M.A.A. Aldahdooh, N. Muhamad Bunnori, Crack classification in reinforced
concrete beams with varying thicknesses by mean of acoustic emission signal
simulations with a number of active sensors. On this basis, it
features, Constr. Build. Mater. 45 (2013) 282–288.
was shown that the STD of l and r of group of sensors has a [16] D. Wang, H. Song, H. Zhu, Numerical and experimental studies on damage
sound relationship with the damage progression. Based on detection of a concrete beam based on PZT admittances and correlation
the results, this parameter increases with the progression of coefficient, Constr. Build. Mater. 49 (2013) 564–574.
[17] D. Xu, S. Banerjee, Y. Wang, S. Huang, X. Cheng, Temperature and loading
damage. effects of embedded smart piezoelectric sensor for health monitoring of
IV. Another important observation was the possibility of localiz- concrete structures, Constr. Build. Mater. 76 (2015) 187–193.
ing the damage and quantifying its severity through the [18] E. Kassem, W. Liu, T. Scullion, E. Masad, A. Chowdhury, Development of
compaction monitoring system for asphalt pavements, Constr. Build. Mater. 96
analysis of the PDF shifts. The variations of both l and r (15) (2015) 334–345.
can be used to localize the damage. In order to evaluate [19] J.F. Potter, H.C. Mayhew, A.P. Mayo, Instrumentation of the Full Scale
the damage severity, the changing trend of r was found to Experiment on A1 Trunk Road at Conington Huntingdonshire, Transport
Research Laboratory (Road Research Laboratory), Wokingham, Berkshire, U.K.,
be very informative. 1969.
[20] A. Badr, A.G. Karlaftis, Using the asphalt pavement dynamic stiffness modulus
Although the efficiency of the proposed concept is shown for in assessing falling weight deflectometer test results, Adv. Mater. Res. 685
(2013) 233–239.
the pavement structures but its applicability is not limited to this [21] A. Badr, A.G. Karlaftis, Duration model estimation for pavement rehabilitation
case. However, there are still some challenges that are the focus and service life, Adv. Appl. Stat. 31 (No. 1) (2012) 1–19.
of future research as follows: [22] A.G. Karlaftis, A. Badr, Predicting asphalt pavement crack initiation following
rehabilitation treatments, Transp. Res. C 55 (2015) 510–517.
[23] R. Huff, C. Berthelot, B. Daku, Continuous primary dynamic pavement response
a. Considering different type of damages, as well as different system using piezoelectric axle sensors, Can. J. Civ. Eng. 32 (1) (2005) 260–269.
type of structures for the case study of damage identification. [24] M. Malekzadeh, G. Atia, F.N. Catbas, A hybrid data interpretation framework
b. Although the FE simulation and laboratory results still for automated performance monitoring of infrastructure, in: Structures
Congress 2015, 2015.
remain satisfactory, verification of the proposed approach [25] M. Malekzadeh, M. Gul, I.B. Kwon, N. Catbas, An integrated approach for
with full-scale experiments would be an interesting topic structural health monitoring using an in-house built fiber optic system and
for future study. Also, the performance of the whole sensing non-parametric data analysis, Smart Struct. Syst. 14 (5) (2014) 917–942.
[26] H. Salehi, S. Das, S. Chakrabartty, S. Biswas, R. Burgueno, Structural Assessment
system including wireless communication needs to be stud- and Damage Identification Algorithms Using Binary Data: Asme 2015
ied more in-depth. Conference on Smart Materials, Adaptive Structures and Intelligent Systems
(SMASIS2015), Colorado Springs, Colorado, 2015.
[27] P.E. Sebaaly, N. Tabatabaee, B. Kulakowski, Evaluation of the Hall-effect sensor
for pavement instrumentation, J. Test. Eval. 23 (3) (1995) 189–195.
Acknowledgment [28] P.E. Sebaaly, N. Tabatabaee, B. Kulakowski, T. Scullion, Instrumentation for
Flexible Pavements—Field Performance of Selected Sensors, Federal Highway
Administration, Washington, DC, 1991.
The presented work is supported by a research grant from the [29] J.M. Signore, J.R. Roesler, Using Fiber-optic Sensing Techniques to Monitor
Federal Highway Administration (FHWA) (DTFH61-13-C-00015). Behavior of Transportation Materials, Transportation Research Record 1478,
Transportation Research Board, Washington, DC, 1995, pp. 37–43.
[30] M. Malekzadeh, G. Atia, F.N. Catbas, Performance-based structural health
References monitoring through an innovative hybrid data interpretation framework, J.
Civ. Struct. Health Monit. (2015) 1–19.
[1] W. Xue, L. Wang, D. Wang, C. Druta, Pavement health monitoring system based [31] W.J. Xue, E. Weaver, Pavement shear strain response to dual and wide-base
on an embedded sensing network, J. Mater. Civ. Eng. 26 (10) (2014) 04014072. tires, J. Transp. Res. Board 2225 (2011) 155–164.
[2] M.H. Mohajeri, P.J. Manning, ARIA: An Operating System of Pavement Distress [32] Z. Zhou, W. Liu, Y. Huang, H. Wang, H. Jianping, M. Huang, O. Jinping, Optical
Diagnosis by Image Processing, Transportation Research Record 1311, fiber Bragg grating sensor assembly for 3D strain monitoring and its case study
Transportation Research Board, Washington, DC, 1991, pp. 120–130. in highway pavement, Mech. Syst. Signal Process. 28 (2012) 36–49.
[3] H. Koutsopoulos, A. Downey, Primitive based classification of pavement [33] T. Yiqiu, W. Haipeng, M. Shaojun, X. Huining, Quality control of asphalt
cracking images, J. Transp. Eng. (1993), http://dx.doi.org/10.1061/(ASCE)0733 pavement compaction using fibre Bragg grating sensing technology, Constr.
-947X(1993) 119:3(402), 402-418. Build. Mater. 54 (2014) 53–59.
[4] J.P. Mills, I. Newton, G.C. Peirson, Pavement deformation monitoring in a [34] M. Sun, R.J.Y. Liew, M.H. Zhang, W. Li, Development of cement-based strain
rolling load facility, Photogramm. Rec. 17 (97) (2001) 7–24. sensor for health monitoring of ultra high strength concrete, Constr. Build.
[5] A. Goel, A. Das, Nondestructive testing of asphalt pavements for structural Mater. 65 (2014) 630–637.
condition evaluation: a state of the art, Nondestr. Test. Eval. 23 (2) (2008) 121– [35] I.L. Al-Qadi, A. Loulizi, M. Elseifi, S. Lahouar, The Virginia smart road: the
140. impact of pavement instrumentation on understanding pavement
[6] H. Ceylan, K. Gopalakrishnan, S. Kim, Looking to the future: the next generation performance, J. Assoc. Asphalt Paving Technol. 73 (2004) 427–465.
hot-mix asphalt dynamic modulus prediction models, Int. J. Pavement Eng. 10 [36] D.H. Timm, A.L. Priest, Dynamic Pavement Response Data Collection and
(5) (2009) 341–352. Processing at the NCAT Test Track NCAT Rep. No. 04-03, National Center for
[7] A. Loizos, C. Plati, Accuracy of pavement thicknesses estimation using different Asphalt Technology, Auburn Univ, Auburn, AL, 2004.
ground penetrating radar analysis approaches, NDT and E Int. 40 (2) (2007) [37] R.S. Rollings, D.W. Pittman, Field instrumentation and performance
147–157. monitoring of rigid pavements, J. Transp. Eng. (1992), http://dx.doi.org/
[8] C. Plati, P. Georgiou, A. Loizos, Use of infrared thermography for assessing HMA 10.1061/(ASCE)0733-947X(1992) 118:3(361), 361-370.
paving and compaction, Transp. Res. C 46 (2014) 192–208. [38] B.A. Sundaram, K. Ravisankar, R. Senthil, S. Parivalla, Wireless sensors for
[9] C.P. Plati, A. Loizos, Estimation of in-situ density and moisture content in HMA structural health monitoring and damage detection techniques, Curr. Sci. 104
pavements based on GPR trace reflection amplitude using different (11) (2013) 1496–1505.
frequencies, J. Appl. Geophys. 97 (2013) 3–10. [39] J.P. Lynch, K.J. Loh, A summary review of wireless sensors and sensor networks
[10] S. Tapkın, A. Çevik, Ü. Usßar, Accumulated strain prediction of polypropylene for structural health monitoring’, Shock Vib. Digest 38 (2006) 91–128.
modified marshall specimens in repeated creep test using artificial neural [40] A. Islam, F. Li, P. Kolli, Monitoring structural health of a bridge using wireless
networks, Expert Syst. Appl. 36 (8) (2009) 11186–11197. sensor network, in: Structures Congress, 2010, pp. 3634–3644.
[11] S. Tapkın, A. Çevik, Ü. Usßar, E. Gülsßan, Rutting prediction of asphalt mixtures [41] H. Salehi, T. Taghikhany, A.Y. Fallah, Seismic protection of vulnerable
modified by polypropylene fibers via repeated creep testing by utilising equipment with semi-active control by employing robust and clipped-
genetic programming, Mater. Res. 16 (2) (2014) 277–292. optimal algorithms, Int. J. Civ. Eng. 12 (4) (2014) 413–428.
[12] M. Bagherifaez, A. Behnia, A. Majeed, H.K. Chai, Acoustic emission monitoring [42] S. Ukkusuri, L. Du, Geometric connectivity of vehicular ad hoc networks:
of multi-cell reinforced concrete box girders subjected to torsion, Sci. World J. analytical characterization, Transp. Res. C 16 (5) (2008) 615–634.
567619 (2014). [43] S. Ukkusuri, D. Kumar, S.T. Waller, S. Viswanath, Sensor protocols for asset
[13] A. Behnia, H.K. Chai, M. BagheriFaez, T. Shiotani, Advanced structural health routing via constrained energy (SPARCE): problem definition and example, in:
monitoring of concrete structures with the aid of acoustic emission, Constr. Vehicular Technology Conference, 2005, p. 1323. VTC-2005-Fall. 2005 IEEE
Build. Mater. 65 (29) (2014) 282–302. 62nd 2.
[14] A. Behnia, H.K. Chai, M. Yorikawa, S. Momoki, M. Terazawa, T. Shiotani, [44] S. Ukkusuri, Y. Wang, T. Chigan, Special issue on exploiting wireless
Integrated non-destructive assessment of concrete structures under flexure by communication technologies in vehicular transportation networks, IEEE
acoustic emission and travel time tomography, Constr. Build. Mater. 67 (Part Trans. Intell. Transp. Syst. 12 (3) (2011) 633–634.
B) (2014) 202–215.
736 A.H. Alavi et al. / Construction and Building Materials 114 (2016) 719–736

[45] D.G. Watters, P. Jayaweera, A.J. Bahr, D.L. Huestis, N. Priyantha, R. Meline, R. [61] N. Lajnef, S. Chakrabartty, N. Elvin, A piezo-powered floating-gate sensor array
Reis, D. Parks, Smart pebble: wireless sensors for structural health monitoring for long-term fatigue monitoring in biomechanical implants, IEEE Trans.
of bridge decks’, Proc. SPIE 5057 (2003) 20–28. Biomed. Circuits Syst. 2 (3) (2008) 164–172.
[46] H. Salehi, T. Taghikhany, Application of robust-optimum algorithms in semi- [62] N. Elvin, A. Elvin, M. Spector, A self-powered mechanical strain energy sensor,
active control strategy for seismic protection of equipment, in: 15th World Smart Mater. Struct. 10 (2001) 293–299.
Conference on Earthquake Engineering, Sep. 24-28, Lisbon, Portugal, 2012, pp. [63] N. Lajnef, M. Rhimi, K. Chatti, L. Mhamdi, Toward an integrated smart sensing
11–20. system and data interpretation techniques for pavement fatigue monitoring,
[47] S. Cho, C. Yun, J.P. Lynch, A.T. Zimmerman, B.F. Spencer Jr., T. Nagayama, Smart Comput. Aided Civ. Infrastruct. Eng. 26 (2011) 513–523.
wireless sensor technology for structural health monitoring of civil structures’, [64] N. Lajnef, K. Chatti, S. Chakrabartty, M. Rhimi, P. Sarkar, Smart Pavement
Int. J. Steel Struct. 8 (2008) 267–275. Monitoring System, Report: FHWA-HRT-12-072, Federal Highway
[48] S. Yang, K. Shen, H. Ceylan, S. Kim, D. Qiao, K. Gopalakrishnan, Integration of a Administration (FHWA), Washington, DC, 2013.
prototype wireless communication system with micro-electromechanical [65] A.H. Alavi, H. Hasni, N. Lajnef, K. Chatti, F. Faridazar, An intelligent structural
temperature and humidity sensor for concrete pavement health monitoring, damage detection approach based on self-powered wireless sensor data,
Cogent Eng. 2 (1) (2015) 1014278. Autom. Constr. (2015), http://dx.doi.org/10.1016/j.autcon.2015.10.001 (in
[49] C.B. Yun, J. Min, Smart sensing, monitoring, and damage detection for civil press).
infrastructures, KSCE J. Civ. Eng. 15 (1) (2011) 1–14. [66] M. Rhimi, N. Lajnef, K. Chatti, F. Faridazar, A self-powered sensing system for
[50] M. Sackin, A Feasibility Study of Embedded Microdevices for Infrastructure continuous fatigue monitoring of in-service pavements, Int. J. Pavement Res.
Monitoring: Smart Aggregate for Concrete (a thesis submitted in partial Technol. 5 (5) (2012) 303–310.
fulfillment of the requirements for the degree of Master of Science), 1999. [67] N.G. Singh, M. Joshi, Optimization of location and number of sensors for
[51] R. Bennet, Wireless monitoring of highways, SPIE Conf. Smart Syst. Bridges structural health monitoring using genetic algorithm, Mater. Forum 33 (2009)
Struct. Highways 3671 (1999). 273-182. 359–367.
[52] J.P. Lynch, Overview of wireless sensors for real-time health monitoring of civil [68] I.L. Al-Qadi, H. Wang, Pavement Damage Due to Different Tire and Loading
structures, in: Proceedings of the 4th International Workshop on Structural Configurations on Secondary Roads, NEXTRANS University Transportation
Control, New York, NY, 2004, pp. 189–194. Center, West Lafayette, Ind, 2010.
[53] J.P. Lynch, Design of a wireless active sensing unit for localized structural [69] P.J. Yoo, I.L. Al-Qadi, Effect of Transient Dynamic Loading on Flexible
health monitoring, J. Struct. Control Health Monit. 12 (3–4) (2005) 405–423. Pavements, Transportation Research Record, 1990, Transportation Research
[54] J. Sirohi, I. Chopra, Fundamental understanding of piezoelectric strain sensors, Board of the National Academies, Washington, DC, 2007, pp. 129–140.
Intell. Mater. Syst. Struct. 11 (2000) 246–257. [70] ASTM E1820–09, Standard Test Method for Measurement of Fracture
[55] H. Salehi, S. Das, S. Chakrabartty, S. Biswas, R. Burgueno, Structural Assessment Toughness, American Society for Testing and Materials, USA, 2009.
and Damage Identification Algorithms Using Binary Data: ASME 2015 [71] M.O. Marasteanu, W. Buttlar, H. Bahia, C. Williams, K.H. Moon, E.Z. Teshale, A.
Conference on Smart Materials, Adaptive Structures and Intelligent Systems C. Falchetto, M. Turos, Investigation of Low Temperature Cracking in Asphalt
(SMASIS2015), Colorado Springs, Colorado, 2015, p. 2015. Pavements, National Pooled Fund Study – Phase II, Final Report 2012–232012,
[56] B.F. Spencer Jr., M.E. Ruiz-Sandoval, N. Kurata, Smart sensing technology: University of Minnesota, 2012.
opportunities and challenges, J. Struct. Control Health Monit. 11 (4) (2004) [72] W. Hafiang, Fatigue Performance Evaluation of Westrack Asphalt Mixtures
349–368. Based on Viscoelastic Analysis of Indirect Tensile Test (Ph.D. dissertation),
[57] N. Elvin, A. Elvin, D.H. Choi, A self-powered damage detection sensor, J. Strain North Carolina State University, 2001.
Anal. 38 (2) (2003) 115–124. [73] K.A. Ghuzlan, H. Carpenter, Fatigue damage analysis in asphalt concrete
[58] I. Korhonen, R. Lankinen, Energy harvester for a wireless sensor in a boiler mixtures using the dissipated energy approach, Can. J. Civ. Eng. 33 (7) (2006)
environment, Measurement 58 (2014) 241–248. 890–901.
[59] M. Rahimi, H. Shah, G.S. Sukhatme, J. Heideman, D. Estrin, Studying the [74] A.R. Burton, K. Minegishi, M. Kurata, J.P. Lynch, Free-standing carbon nanotube
feasibility of energy harvesting in a mobile sensor network, in: Proceedings of composite sensing skin for distributed strain sensing in structures, SPIE 9061
the 2003 IEEE International Conference on Robotics & Automation, Taipei. (2014) 906123.
Taiwan, September 14–19, 2003, pp. 19–24. [75] A.H. Alavi, H. Hasni, N. Lajnef, K. Chatti, F. Faridazar, Damage detection using
[60] S. Roundy, P.K. Wright, J. Rabaey, A study of low level vibrations as a power self-powered wireless sensor data: an evolutionary approach, Measurement
source for wireless sensor nodes’, Comput. Commun. 26 (2002) 1131–1144. 82 (2016) 254–283.

You might also like