Measurement 121 (2018) 144–159

Contents lists available at ScienceDirect

Measurement
journal homepage: www.elsevier.com/locate/measurement

Experimental and numerical studies for suitable infrared thermography T

implementation on concrete bridge decks
⁎
Shuhei Hiasaa,b, Recep Birgulc,d, Masato Matsumotoe, F. Necati Catbasa,
a
Department of Civil, Environmental, and Construction Engineering, University of Central Florida, 12800 Pegasus Drive, Suite 211, Orlando, FL, USA
b
West Nippon Expressway Company Limited (NEXCO-West), Dojima Avanza 19F, 1-6-20 Dojima, Kita-ku, Osaka 530-0003, Japan
c
Mugla Sitki Kocman University, Turkey
d
University of Central Florida, 12800 Pegasus Drive, Suite 211, Orlando, FL, USA
e
NEXCO-West USA Inc. 8300, Boone Blvd., Suite 240, Vienna, VA 22182, USA

A R T I C L E I N F O A B S T R A C T

Keywords: Capturing the temperature difference between sound and defective parts under ambient conditions is key for
Infrared thermography infrared thermography (IRT) on concrete bridges. This study explores the favorable time windows for concrete
Non-destructive evaluation bridge deck inspections by IRT through field experiment and finite element model simulations. Based on the
Bridge deck inspection numerical simulations and experimental IRT results, the preferable thermal contrast to detect defects occurs
Data collection time
during both daytime and nighttime. However, available time span during daytime is much shorter than that of
FE model simulation
nighttime due to interchange periods between cooling and heating cycles in the morning and in the evening.
Furthermore, IRT is affected by sunlight during the daytime resulting in possible misdetections. Moreover, ef-
fects of clouds and radiative cooling are observed, and it is found that the clear sky is a preferable condition for
IRT. Therefore, optimal conditions for IRT implementation on concrete bridge decks can be concluded that
nighttime application under the clear sky condition. In addition, the effect of obstacles on a bridge surface such
as gravel, wood chips that bring additional challenges to IRT are also evaluated experimentally.

1. Introduction
Degradation of concrete bridges, especially concrete bridge decks, is
a widespread problem in the United States. Most bridge decks, 93%
(346 km2 out of 371 km2) in bridge deck area [1], in the USA are made
of concrete. In addition, concrete bridge decks deteriorate faster than
other bridge components due to direct exposure to traffic. Furthermore,
most State Departments of Transportation (DOTs) spend 50–80% of
their bridge maintenance budgets on concrete bridge decks due to in-
sufficiency of detecting hidden defects until they become serious pro-
blems [2]. Thus, the Federal Highway Administration’s (FHWA’s) Long
Term Bridge Performance (LTBP) Program regards bridge decks as the
highest priority issues for bridge performance [2,3]. In the US, every
bridge as defined by the National Bridge Inspection Standards (23 CFR
650 Subpart C) is required to have a routine inspection conducted, vi-
sual inspection at a distance to identify obvious defects, at regular in-
tervals not to exceed twenty-four months. In addition, in-depth in-
spections to identify any deficiencies not readily detectable using
routine inspection procedures are also conducted. Generally in-depth
inspections are performed over longer intervals than routine
inspections; the level and frequency are determined by the bridge ad-
ministrators. In-depth inspections consist of hands-on inspections by
chain drag and/or hammer sounding to fully ascertain the existence of
or the extent of any deficiencies, and close-up visual assessment of
every defective surface at a distance no further than an arm’s length
[4]. These methods, however, require a great deal of field labor and
lane closures, especially for bridge deck inspection. Thus, better
methods are needed to detect defects and to quantify the extent and
severity of bridge deck conditions early, accurately, and rapidly with
minimal traffic impact, ideally, without lane closures for inspections
[2,3].
Under these circumstances, Non-Destructive Evaluation (NDE)
techniques such as computer-vision based crack detection, impact echo
(IE), ultrasonic surface waves (USW), electrical resistivity (ER), ground-
penetrating radar (GPR) and infrared thermography (IRT) have been
developed to inspect and monitor aging and deteriorating structures
rapidly and effectively [5,6]. Most NDE methods aim to achieve the
highest quality of visual imaging of the relevant internal and/or ex-
ternal features of structures [7,8]. However, no single method can de-
tect all types of defects in concrete structures as well as the traditional

⁎
Corresponding author.
E-mail addresses: hiasa615@Knights.ucf.edu, s.hiasa.aa@w-nexco.co.jp (S. Hiasa), rbirgul@mu.edu.tr (R. Birgul), m.matsumoto@w-nexco-usa.com (M. Matsumoto),
catbas@ucf.edu (F. Necati Catbas).

https://doi.org/10.1016/j.measurement.2018.02.019
Received 19 June 2016; Received in revised form 8 February 2018; Accepted 13 February 2018
Available online 25 February 2018
0263-2241/ © 2018 Elsevier Ltd. All rights reserved.
S. Hiasa et al.
inspection combination of visual and sounding inspections; hence, there
is still no international standard NDE methods for concrete bridges,
although significant progress has been made up to the present [8]. In
terms of assessing the external condition based on visual inspections,
the application of computer vision-based defect detection for concrete
infrastructure has been exponentially increasing over the last decade,
and automation of detection and measurements of defects has been
developing [6]. Regarding the assessment of internal conditions
through the use of hammer sounding and/or chain drag, several NDE
technologies for bridge deck damage detection were compared and
evaluated at in-service bridges [9–13].
Among several NDE technologies, image-based bridge inspection
technologies, consisting of high-definition (HD) imaging and IRT, are
the promising candidates to conduct rapid and in-depth bridge in-
spections by attaching these cameras on vehicles, such as cars and
unmanned aerial vehicles (UAVs). Both techniques allow for non-con-
tact application and can instantly portray a wide range of concrete
structures at one time [14–17]. Especially for concrete bridge deck
inspections, bridge decks can be scanned at a highway speed without
lane closures with reasonable accuracy by attaching these cameras on a
car [9,18]. The data collection speed of the integrated IRT and HD
systems is a great advantage for periodic bridge inspections compared
to current routine and in-depth inspections. Actually, the combination
of IRT and HD imaging system has been applied for some bridges at
normal driving speeds for several years, although the ASTM standard
suggests an IRT data collection speed of no greater than 16 km/h
(10 mph) [19]. Hiasa et al. [9,18] conducted a comparative study on
infrared (IR) camera specifications with three different cameras and
concluded that if devices with shorter integration time (such as the
commonly equipped cooled type cameras) are used, high-speed in-
spection at normal driving speeds (e.g. 80 km/h) without any lane
closures is possible with similar or better level of accuracy compared to
other NDE methods such as IE, GPR and chain drag.
Regarding the accuracy of IRT, capturing the temperature difference
between sound and defective parts under ambient conditions is key for
IRT on concrete bridges. Effective utilization of the IRT on civil en-
gineering structures also depends on the application time window on
structures. The main objective of this study is to explore the favorable
time windows for concrete bridge deck inspections by IRT through field
experiments and finite element (FE) model simulations. As part of this
objective, development of a reproducible numerical model is discussed.
Finally, the effect of obstacles on bridge surface such as gravel, wood
chips that bring additional challenges to IRT are also evaluated ex-
perimentally.
Previously, Watase et al. [20] used specially designed concrete
plates and thermocouples to assess the favorable time for IRT. However,
they defined the time windows by measuring temperature difference
between sound and delaminated areas by thermocouples. This requires
access to the bridge and installation of the calibration test setups on the
bridges for 1–2 days of data collection prior to the actual field IRT scan
of the bridge decks. Hiasa et al. [21] developed an FE model, yet, the
number of reference data to check the reproducibility was limited.
Therefore, this study focused on collecting IRT data more frequently, on
an hourly basis from morning to midnight, and compared IRT data and
simulation data to validate the correlation. Based on the simulation
model and experimental data, the data are interpreted for more effec-
tive IRT implementation on real bridges by considering the favorable
time windows along with some of the challenges such as debris on
bridge decks, which has been presented as an issue in the literature.
2. Challenges of infrared thermography
NDE Web Manual of FHWA [22] lists the limitations of IRT, such as
(1) it is difficult to select the optimal conditions for effective applica-
tion, (2) obstacles, such as water, stains, or debris, can cause mis-
detection, (3) unable to provide information about delamination depth,
145
Measurement 121 (2018) 144–159
and (4) difficulty of detecting defects located at deeper than 10–13 cm
(4–5 in.). As for (1), Hiasa et al. [9,18,21,23] and Watase et al. [20]
have conducted several studies to identify the uncertainties and chal-
lenges of IRT, such as data collection time, delamination size, camera
specifications and data collection speed, and explored proper methods,
ideal conditions, and data interpretation methods for more effective
application of IRT. In terms of (2), those misleading thermal contrasts
can be distinguished by comparing IR images with HD images; thus,
integrated use of IRT and HD imaging is recommended [9]. Regarding
(3), delamination depth estimation method was developed using finite
element (FE) modeling [24]. Approximate depth of a delamination
found by IRT at an in-service bridge deck was estimated, and the
method was validated from a core sample at that delaminated area. In
regard to (4), Hiasa et al. [21,24] found the correlation between size of
delamination and detectable depth. Actually, Kee et al. [12] were able
to detect a 15.2 cm (6 in.) depth of delamination by IRT. Therefore, IRT
can detect delamination that is deeper than 10–13 cm if the delami-
nation size is large enough. In addition to these limitations, Hiasa et al.
[23,25] have been developing more objective data interpretation
method than just judging the color contrast of IR images by in-
corporating FE model simulations. Even though more data is needed to
enhance the reliability and usability of IRT for bridge inspection, sig-
nificant improvements within the IRT method have already been
achieved through past studies.
However, IRT still has some challenges to overcome such as the
practical time for data collection to enhance reliability, accuracy and
usability. It is essential to clarify the specific ideal time for IRT data
collection since there are contradictory reports regarding the appro-
priate time window for IRT measurements, and this factor might be the
main cause that several researchers reported different evaluations on
IRT. Washer et al. proposed 5–9 h after sunrise to detect subsurface
delamination for the solar loading part depending on the defect loca-
tions, 51 mm (2 in.) to 127 mm (4 in.) deep delaminations [26,27],
while Gucunski et al. [10] reported that a thermal image recorded
40 min after sunrise yielded a much clearer image than another one
recorded around noon. They also assumed that there is a small time
window during the day during which IRT works well since IRT is very
sensitive to the environmental conditions [10]. Additionally, Yehia
et al. [28] found that the responses of delaminations were described as
weaker in IR images as the time approached 3 PM [28]. Moreover, Kee
et al. [12] also reported that no indication was found from the IR image
taken 3 h and 45 min after sunrise (with the shallowest delamination
located at 6.35 cm depth) while the best results were achieved using the
cooling cycle in which even 15.24 cm deep delaminations could be
detected. Furthermore, Watase et al. [20] proposed favorable time
windows for inspections depending on the parts of a bridge; noontime
for the deck top, and midnight for the deck soffit. Therefore, it can be
assumed that IRT has practical time windows for data collection;
however, a generalized, consistent ideal time window for IRT data
collection has not been defined yet.
3. Test settings
3.1. Test specimens
IRT for detecting subsurface defects is based on differences of
thermal characteristics such as thermal conductivity and thermal effu-
sivity between the defect and the host material [29]. Thermal effusivity
(e) is determined by the product of density ( ρ ), specific heat capacity
(c) and thermal conductivity (k) as follows [30];
e= ρck (1)
Since thermal effusivity is part of thermal conductivity as shown in
Eq. (1), the mechanism of IRT for concrete structures is briefly ex-
plained using a difference in thermal conductivity as follows: sunlight
S. Hiasa et al.
heats up the concrete surface. If there is a delamination, that part is
filled by air and the air acts as a thermal insulator (thermal conductivity
of air: 0.0241 W/m °C) and prevents heat from penetrating to the con-
crete (thermal conductivity of concrete: 1.6 W/m °C) beneath the de-
lamination. Thus, the concrete above the delamination becomes
warmer than the surroundings during the daytime. During the night-
time, the heat escapes to the sky and the area over the delamination
appears cooler [31]. IRT detects interior defects by capturing those
temperature differences of concrete surface by reading the emitted
electromagnetic radiation from the concrete surface and converting it
to a temperature [32].
Previously Hiasa et al. [18] conducted a field test with manu-
factured concrete blocks which had artificial delaminations. In the ex-
periment, a typical concrete bridge deck encountered in the USA was
reproduced as the test specimen, 91.4 × 91.4 × 20.3 cm in size, and
two types of artificial delamination made by styrofoam and cardboard
wrapped in plastic were installed inside the concrete. Styrofoam and
cardboard were chosen to generate an artificial delamination to simu-
late thermal conductivity properties similar to that of air. The thermal
conductivity of the styrofoam is 0.024 W/m °C; the cardboard has a lot
of air internally. According to Cotič et al. [33], no significant differ-
ences were observed between the surface temperatures above poly-
styrene (thermal conductivity: 0.033 W/m °C) and air-filled defects.
Therefore, the artificial delamination which has similar thermal con-
ductivity can be regarded as an actual delamination. Styrofoam was
installed at the center part of the specimen and wrapped cardboard was
embedded at a corner. Each size of delamination is 10.2 cm
(4 in.) × 10.2 cm × approximately 0.3 cm (1/8 in.), and they were
placed at different depths from the surface, 1.3 cm (0.5 in.), 2.5 cm
(1 in.), 5.1 cm (2 in.) and 7.6 cm (3 in.). In this study, the concrete slabs
which have delaminations at 1.3 cm and 2.5 cm were used as displayed
in Fig. 1 since delaminations at 5.1 cm and 7.6 cm were not detected in
the previous tests under the given conditions using cameras attached to
a movable vehicle [18].
In order to investigate the effect of obstacles on the concrete surface
for IRT, several types of obstacles, such as gravel and wood chips, and
tapes of different colors and sizes were installed on the concrete slab
which has 2.5 cm deep delamination as shown in Fig. 1. Even though
ASTM standard suggests to remove any debris on the bridge decks prior
to the utilization of IRT [19], it requires extra work for bridge deck
inspections. Furthermore, Gucunski et al. [10] also suggests that the
effect of debris and markings on bridge decks is needed to be taken into
account. Therefore, this study examines how debris affects the result of
IRT for bridge deck inspections.
3.2. Temperature data
During the experiment, thermocouples were attached on the surface
Fig. 1. Concrete test specimens (Delamination - left: 1.3 cm, right: 2.5 cm).
146
Measurement 121 (2018) 144–159
of both concrete slabs as can be seen in Fig. 1, and the other thermo-
couples were put between the concrete blocks and wooden pallet to
measure the air temperature while avoiding direct sun-loading. Fur-
thermore, this study utilized the weather record at the closest clima-
tological substation, about 700 m away from the test site, as depicted in
Fig. 2. Fig. 3 shows each air temperature record during the test;
“Thermocouple (1.3 cm)” and “Thermocouple (2.5 cm)” in the graph
indicate air temperatures between the concrete blocks and wooden
pallets. As can be seen in the figure, the temperatures of both ther-
mocouples increase in the afternoon. This can be due to the sun shining
on these thermocouples during those times, even though they were
placed under the concrete slabs; thus, they were heated up by sun-
loading and measured higher temperatures than the ambient tem-
perature. Compared to the temperature record of the climatological
substation, these two thermocouples indicate very close values except
for these peak periods; hence, the data of the climatological substation
can be substituted during the period which thermocouples were af-
fected by sun-loading. The temperature data of “Thermocouple
(2.5 cm)” was chosen as ambient air temperature for FE model simu-
lation, and temperature record of the climatological substation was
substituted during 2:48 PM to 4:27 PM.
3.3. Photography situation
In this experiment, an infrared camera, T420 manufactured by FLIR
Systems, Inc., was utilized. The specifications are listed in Table 1. The
field test was conducted on December 19, 2015 at the University of
Central Florida as shown in Fig. 2. As depicted in Fig. 4, IR images were
taken at a height of about 213 cm from the concrete surface, and each
concrete slab was photographed separately from the same location. The
distance from the closest edges was 122 cm from the 1.3 cm deep de-
lamination and 15 cm from the 2.5 cm deep delamination specimens. IR
images were taken every hour from 7 AM to 12 AM. Furthermore, in
order to capture the interchange point from the nighttime cooling effect
to the daytime heating effect, IRT data were collected at 30-min in-
tervals from 7 AM to 10 AM.
4. IRT test results
4.1. Results of 1.3 cm deep delamination
Fig. 5 shows IR images of the concrete surface, which has 1.3 cm
deep delaminations at the center and lower left corner in the images. As
mentioned in Section 3.3, IRT data were collected at 30-min intervals
from 7 AM to 10 AM to capture the interchange point from the night-
time cooling effect to the daytime heating effect, and then IRT data
were taken every hour from 10 AM to 12 AM. The temperature range of
the IR images is set up to 5 K for all images. In this study, when the
S. Hiasa et al. Measurement 121 (2018) 144–159

Fig. 2. Location of test site and the closest climatological substation (Google Maps).

Fig. 3. Temperature record during the test.

Table 1 concrete blocks were casted, carrying handles were set up at the two
Camera specifications for T420. corners on a diagonal line to a reversed position of the corner delami-
nation by reinforcing bars (upper left and lower right corner for 1.3 cm
Camera type T420
depth, upper right and lower left corner for 2.5 cm depth). Further-
Detector type Uncooled microbolometer more, thermocouples were attached on the surface of concrete slabs
Thermal sensitivity (NETD) < 0.045 K at 30 °C (right side of the middle part in IR images). Since they have a different
Accuracy ± 2 °C or ± 2%
temperature distribution from the concrete surface, they were not
Resolution 320 × 240 pixels
Spectral range 7.5–13 µm
considered as delamination even though IR images indicated different
Frame rate 60 Hz colors. Moreover, since the temperature distribution of the concrete
Field of view 25° × 19° surface was not entirely homogeneous due to reasons such as boundary
Integration time/time constant (electronic shutter 12 ms conditions and the effect of sunlight, some temperature differences
speed)
were judged as noise or errors, and in turn, they were evaluated as
sound areas.

147
S. Hiasa et al. Measurement 121 (2018) 144–159

Fig. 4. Photography situation.

Fig. 5. IR images at each time for 1.3 cm deep delamination specimen.

148
S. Hiasa et al.
Fig. 6. Sun loading situation at characteristic times.
In the images from 7 AM to 8:30 AM, both delaminations can be
distinguished clearly and they indicate a lower temperature than the
surroundings. Even though those parts on the edge also show a lower
temperature than the areas in the middle, this can be considered as the
effect of the boundary condition. The indication becomes unclear at 9
AM and disappears at 9:30 AM. Then, those delaminated areas become
hotter than the surroundings after 10 AM. This result proved that there
is an interchange period from the nighttime cooling effect to the day-
time heating effect, and IRT cannot detect delamination during this
period.
Regarding the opposite interchange period, from daytime heating to
the nighttime cooling effect, IR images at 3 PM and 4 PM do not show
indications at the delaminated areas. Therefore, the period of these
interchanges are within 1 h in the morning and within 2 h in the eve-
ning in this experiment under the given conditions. The noteworthy
points other than the interchange points are the effect of the boundary
condition from 9:30 AM to 6 PM, and the IR image taken at 12 AM
which became unclear compared to the image taken at 11 PM.
Regarding the boundary condition, the area on the edge which re-
ceived sunlight is much hotter than the other area during the period.
Fig. 6 displays pictures of the test site at some characteristic times, and
Table 2 summarizes weather conditions and IRT results. In this table,
149
Measurement 121 (2018) 144–159
each mark indicates following:
• Circle: overall irradiation to the surface for “Sun-loading”; cloudy
sky for “Cloud”; clear indication in the IR image for “Indication”
• Triangle: partial irradiation to the surface for “Sun-loading”; clear
sky but some clouds in the sky for “Cloud”; obscure indication in the
IR image for “Indication”
• X-mark: no irradiation to the surface for “Sun-loading”; clear sky
without cloud for “Cloud”; no indication in the IR image for
“Indication”
According to the record of the climatological substation, sunrise was
at 7:12 AM and sunset was at 5:32 PM on that day. However, the sun
started to shine on the concrete slabs from 9 AM due to interference of a
tree and a building near the test site as can be seen in Fig. 6. In IR
images of Fig. 5, the sun shined from the left upper side to the right
upper side as time passes from 9 AM, and those edges show much
higher temperatures than the temperature range of the IR images which
is set up to 5 K. Therefore, those parts require another temperature
range setting, and it makes more challenging to evaluate IRT results.
Furthermore, by setting several temperature ranges for the same IR
image, it might yield misdetections. Additionally, according to
S. Hiasa et al. Measurement 121 (2018) 144–159

Table 2
Summary of weather conditions and IR results (1.3 cm depth).

Time 7:00 AM 7:30 AM 8:00 AM 8:30 AM 9:00 AM 9:30 AM 10:00 AM 11:00 AM 12:00 PM 1:00 PM

Ambient temperature (°C) 9.9 9.7 10 10.4 10.9 11.7 12.9 15.8 18 19.6
Humidity (%) 72 73 74 73 72 71 69 66 59 57
Sun-loading (1.3 cm) × × × × △ ○ △ ○ ○ ○
Cloud × × × × × × × × × ×
Indication (1.3 cm) ○ ○ ○ ○ △ × ○ ○ ○ ○
Cooling/heating Cooling Cooling Cooling Cooling Cooling N/A Heating Heating Heating Heating

Time 2:00 PM 3:00 PM 4:00 PM 5:00 PM 6:00 PM 7:00 PM 8:00 PM 9:00 PM 10:00 PM 11:00 PM 12:00 AM

Ambient temperature (°C) 20.3 18.9 17.8 16.7 15.1 14.5 13.7 13.2 12.9 12.8 12.7
Humidity (%) 59 58 60 65 71 74 78 80 82 84 86
Sun-loading (1.3 cm) ○ △ ○ × × × × × × × ×
Cloud × × × × × × × × × × ○
Indication (1.3 cm) ○ × × ○ ○ ○ ○ ○ ○ ○ △
Cooling/heating Heating N/A N/A Cooling Cooling Cooling Cooling Cooling Cooling Cooling Cooling

Note: Sunrise; 7:12 AM, Sunset; 5:32 PM.

○:overall irradiation (sunloading); cloudy sky (cloud); clear indication (indication).
△: partial irradiation (sunloading); clear sky but some clouds (cloud); obscure indication (indication).
×: no irradiation (sunloading); clear sky without cloud (cloud); no indication (indication).

Gucunski et al. [10], an IR image taken around noon indicated sub-
stantial noise compared to another one taken in the morning. Therefore,
the preferable time period to apply IRT for concrete bridge deck in-
spection might be during the nighttime cooling effect in order to reduce
the possibility of misdetection, even though delaminated areas are also
indicated clearly during the daytime heating effect period in this study.
In terms of the image from 12 AM, clouds might have affected IRT
since the difference between 12 AM and before 11 PM is only the sky
condition, whether it was clear or cloudy. On that day, the sky was
always clear during the experiment except at 12 AM when the sky was
covered with clouds.
4.2. Discussion on the effect of cloud
Fig. 7 depicts the schematic diagram of the flow of thermal energy.
The atmospheric temperature changes depending on the heat balance.
There are two types of radiation: visible radiation and invisible infrared
radiation. The first type, visible radiation, includes direct solar radia-
tion, diffuse solar radiation which is diffused in the atmosphere/clouds,
and reflected solar radiation which is reflected radiation of both direct
and diffuse solar radiation by the ground/clouds. Some of the diffused
solar radiation returns to space and the rest of it reaches the ground.
Reflected solar radiation is diffused in the atmosphere and again returns
to the ground or space. This total amount of radiation is called global
solar radiation. The second type of radiation is invisible infrared ra-
diation. Infrared radiation is emitted from the atmosphere such as

Fig. 7. The flow of thermal energy (Translated from Japanese and reproduced the Figure 1.10 of [34]).

150
S. Hiasa et al.
clouds, water vapor, and carbon dioxide, and the infrared radiation
returns to space and reaches the ground. This is called atmospheric
radiation. Furthermore, the ground surface also emits infrared radiation
which is called earth radiation. The heat balance is decided by these
two types of radiation, sensible heat, latent heat and underground
conductive heat [34].
Fig. 7 depicts the heat balance during the daytime. If it is nighttime,
the global solar radiation becomes zero. Therefore, the infrared radia-
tion emitted from the ground surface goes to space; consequently, air
temperature decreases during the nighttime. However, if there is a
cloud as depicted in the figure, the cloud absorbs the infrared radiation
from the ground and emits it again to the ground and space; thus, the
air temperature near the ground is warmed up when the sky is cloudy.
On the other hand, when the sky is clear, most of the infrared radiation
emitted from the ground goes to space; consequently, the air tem-
perature falls more than that of a cloudy day. Even though, some ra-
diation is diffused by water vapor and carbon dioxide, the effect is
much lower than the effect of clouds. This phenomenon is known as
radiative cooling, and on sunny days, the night tends to be cooler than
on cloudy days. This phenomenon was observed from a long term ex-
periment by Worzewski et al. [35]. In their experiment, a rotor blade
segment of a wind turbine was monitored by an IR camera. On a cloudy
day, 100% of the sky was covered by clouds; when the clouds were
cleared suddenly in the early morning, around 3 AM, the surface tem-
perature of the rotor blade was decreased due to the effect of radiative
cooling [35].
In order to validate the effect of radiative cooling on IRT, the
measured temperatures are investigated. In this experiment, the con-
crete’s surface temperature was measured by two devices, an IR camera
and a thermocouple. Fig. 8 describes the results; “Del. (IRT)” depicts
temperatures of center part of the middle delamination; “Sound (IRT)”
shows temperatures of sound part where it is not affected by the
boundary condition (approximately in-between of the edge of the slab
and the middle delamination); “Air Temp.” indicates the temperatures
measured by the thermocouple which was installed between the con-
crete blocks and wooden pallet; “Surface (thermocouple)” shows sound
concrete surface’s temperatures measured by the thermocouple which
was attached on the concrete slab as shown in Fig. 1. Regarding the
temperature of delaminated area, since the middle delamination was
installed so as not to be affected by boundary conditions, the center part
of the middle delamination is regarded as the temperature of
Fig. 8. Temperature results from IRT camera and thermocouples (1.3 cm depth).
151
Measurement 121 (2018) 144–159
delaminated area in this study. The surface temperature measured by
the thermocouple shows a much higher temperature than the tem-
peratures measured by the IR camera. A possible explanation for this
behavior is that thermocouples used in this experiment registered
temperature values that are a combination of the concrete’s surface
temperature, the ambient air temperature and the temperature of the
thermocouple itself which heated up by direct sunlight since the sensor
part is covered by a black material and it was exposed to the sunlight
directly during the experiment. Even the thermocouples which were
installed under concrete blocks so as to avoid direct sunlight were af-
fected significantly when they were illuminated by the sun and the
temperature suddenly increased to around 26–28 °C as shown in Fig. 3.
Thus, it can be considered that the effect of sun loading is larger and the
surface’s temperature measured by the thermocouple is much higher
than the IRT data especially during the daytime. Therefore, this com-
bination gives rise to a higher temperature reading on a concrete sur-
face than the actual temperature value of the surface.
Another possibility is that IRT data was affected by reflection of sky
temperature due to the long wavelength of the spectrum range of the IR
camera, 7.5–13 µm. Since concrete surface temperature is always lower
than the air temperature, the temperature captured by the IR camera
might be lowered by reflected temperature. Hashimoto and Akashi [36]
reported that IR cameras with more than 8 µm wavelength detection
region were affected by the reflection of the sky, and the effect increases
when the angle between IR camera and concrete surface becomes
shallower, especially when the surface is smoother. They took IR
images of the sky with three types of cameras, and short/medium
spectral range, 3–5 µm, IR cameras captured the temperature as around
20 °C, while long wavelength IR camera, 8–13 µm, showed the tem-
perature from 0 to −35 °C [36]. Furthermore, when Hiasa et al. [9]
compared three IR cameras with different specifications for high-speed
scanning of a bridge deck, this camera, T420, indicated much lower
temperature, about 16–18 °C, than other two IR cameras. In another
experiment, however, Hiasa et al. [21] utilized three different IR
cameras with the same concrete blocks as this experiment, and this
camera captured similar temperatures to the other two cameras all the
time. Although whether the IRT data was affected by reflection is not
certain since this experiment utilized only one IR camera, the tem-
peratures measured by IRT were consistently changing throughout the
day. Therefore, this study considers that IRT measures concrete sur-
face’s temperature correctly.
S. Hiasa et al.
Despite these two devices indicating different temperatures, both
devices provide the consistent results that the concrete surface tem-
perature increased suddenly from 11 PM to midnight (12 AM). There
was a decrease in air temperature until approximately 11:30 PM, at
which point the air temperature began increasing slightly until 12 AM.
However, the temperature of the concrete surface changed more than
the air temperature. This can be due to the atmospheric temperature
being affected by infrared radiation from the ground, which decreases
air temperature, and atmospheric radiation from the clouds, which
increases air temperature. As shown in Table 2, the sky was always
clear until 11 PM on that day, thereafter, the sky condition became
cloudy. Thus, atmospheric radiation from the clouds can be considered
to warm up the air temperature near the ground by reflecting infrared
radiation from the cloud to the ground again. Consequently, the air
temperature began to increase slightly from 11:30 PM. Regarding the
concrete’s surface temperature, the surface temperature can be con-
sidered more sensitive than the air temperature near the ground due to
much higher thermal conductivity of concrete than that of air. There-
fore, the change in air temperature was smaller than the change in
temperature of the concrete surface. This is an opposite phenomenon to
what was observed by Worzewski et al. [35]. In their case, a cloud
covered sky was suddenly cleared and then the surface temperature of a
rotor blade decreased. This experimental data shows the effects of
clouds and radiative cooling on IRT for concrete decks, and indicates
that the clear sky is a preferable condition for IRT even for nighttime
applications. The damage indication during the clear sky condition was
much clearer than the indication under cloudy condition as shown in
Fig. 5.
4.3. Results of 2.5 cm deep delamination
Fig. 9 shows IR images of the concrete surface which has 2.5 cm
deep delaminations at the center and left upper corner of the image,
and Table 3 summarizes the weather conditions and IRT results. These
IRT data were also collected at 30-min intervals from 7 AM to 10 AM,
and then taken every hour from 10 AM to 12 AM. The temperature
range of the IR images was set up to 5 K for all images. As mentioned in
Section 3.1, several types of obstacles, such as gravel, wood chips, and
tape of different color and size, were installed on the concrete surface to
investigate the effect of obstacles on the concrete surface for IRT as
shown in Fig. 1. Locations of each obstacle is displayed in Fig. 10, and
whether each obstacle can be distinguished from the IR image is sum-
marized in Table 4.
Regarding wood chips, two chips were placed around the middle
part of the delaminated area, and both wood chips can be distinguished
from IR images at any time. In terms of gravels, every gravel can also be
distinguished from IR images at any time, although some of them are
not clearly distinguished sometimes due to the effect of boundary
condition and the much smaller sizes than the wood chips. The reason
can be considered that the debris have much smaller surface area than
the concrete slab, so they were cooled down/heated up much faster
than the concrete surface. Moreover, these debris have different
thermal conductivity compared to the concrete block so that they
generate different temperature than concrete surface. Therefore, debris
on bridge decks can be distinguished from IR images even they are not
removed before IRT implementation, although distinguishing debris
from HD images is much easier and more accurate.
As for the effect of these debris on IRT, the delaminated area hidden
by the wood chips could not be detected because IR images show
temperatures of the wood chips’ surface instead of concrete surface’s
temperature. However, if the delaminated area is larger than debris,
delamination can be detected even those debris are not removed prior
to IRT application. Usually, such large debris must be removed from
bridge decks for safety reason, thereby some small debris might not be a
problem, and cleaning of bridge decks is not always required (Even
though, removing debris before IRT may provide better results).
152
Measurement 121 (2018) 144–159
In terms of the tapes, light colored tapes such as white (T-1, 3, 11,
14), pale pink (T-4, 9) and silver (T-7) were always cooler than the
concrete surface when they are distinguished from IR images. This can
be considered that these colors reflect sunlight without observing the
heat. On the other hand, dark colored tapes such as black (T-6, 8, 12),
green (T-5, 10, 13) and brown (T-2) show higher temperatures than the
surroundings from 11 AM to 2 PM while they yielded lower tempera-
tures during other time windows. It can be considered that this is also
caused by the difference of heat absorption depending on colors.
However, it can be observed that these different colors can also be
distinguished by IRT. Furthermore, it can be assumed that if visual
high-definition images are also taken at the same time, most of the color
differences as well as debris can be distinguished by comparing IR and
HD images easily. Therefore, removing debris on the bridge deck is not
always required for inspections of concrete bridge decks by IRT.
5. FE model simulation
5.1. FE model development
The FE models of the concrete test specimens were developed and
the test environment was simulated using Heat Transfer Module of
COMSOL Multiphysics software. A concrete model of the same size as
the concrete blocks used in the experiment, 91.4 × 91.4 × 20.3 cm,
was established on a larger ground, 6 × 6 × 1 m, and two pieces of
styrofoam (delamination), 10.2 × 10.2 × 0.3 cm, were installed inside
the concrete as described in Fig. 11. The depth from the surface is the
same as the experiment, 1.3 cm and 2.5 cm. The orientation of the
concrete block was also set up in almost the same way as the experi-
ment. The picture on the left shows the real concrete blocks used in the
experiment (the right two concrete slabs), and the images on the right
depict the modeled concrete block. In the experiment, concrete blocks
were placed on wooden stands to make space through which wind
blows, and those were set up on wooden pallets as shown in the picture.
However, in the model, a concrete block was put on concrete stands,
which have the same height (20 cm) as wooden stands and the pallet for
simplifying the modeling and simulation. In terms of the mesh size, this
software automatically assigns mesh sizes of each component by
choosing predefined element sizes. The “Finer” element size (3rd finest
size out of 9 levels) was selected for the concrete block and the Styr-
ofoam in this simulation for more accurate characterization. In this
mesh size selection, the software automatically models mesh sizes of
selected components within elements ranging in size from 2.4 cm to
33 cm depending on the component’s size and the boundaries. As for
the stands of the concrete block and the ground, a “Normal” element
size (5th finest size out of 9 levels), within elements ranging in size from
10.8 cm to 60 cm, was selected for computational economy since it can
be considered that those mesh sizes are less sensitive for concrete sur-
face temperature than other two components. The material properties
were set up as shown in Table 5 by referring to [37,38] and calibrating
the model repeatedly with experimental results.
Solar radiation, air temperature and convective heat transfer were
considered for each side of the concrete block in the FE model. The
primary heat source was solar radiation in this model, and the or-
ientation of sunlight (zenith angle and the solar elevation) over the
simulation time was automatically computed from the longitude, lati-
tude, time zone, date, and time by the software. The solar position is
estimated by the Julian Day calendar calculation, and the zenith angle
(zen) and the azimuth (azi) angles of the sun are converted into a di-
rection vector (isx, isy, isz) in Cartesian coordinates assuming that the
North, the West, and the up directions correspond to the x, y, and z
directions, respectively, in the FE model as follows [39];
S. Hiasa et al. Measurement 121 (2018) 144–159

Fig. 9. IR images at each time for 2.5 cm deep delamination specimen.

⎧ isx = −cos (azi ) sin (zen ) −81.189103, latitude; 28.591897, time zone (GMT (Greenwich Mean
is y = sin (azi ) sin (zen ) Time)); −5 h. The location is the test field, and the location information
⎨ was referred to Google Maps. Regarding the ambient temperature
⎩ isz = −cos (zen ) (2)
(Tamb), the measured temperature data was input as mentioned in
In this simulation, the information was set up as follows: longitude; Section 3.2, and solar irradiance (Is ) and heat transfer coefficient (h )

Table 3
Summary of weather conditions and IR results (2.5 cm depth).

Time 7:00 AM 7:30 AM 8:00 AM 8:30 AM 9:00 AM 9:30 AM 10:00 AM 11:00 AM 12:00 PM 1:00 PM

Ambient temperature (°C) 9.9 9.7 10 10.4 10.9 11.7 12.9 15.8 18 19.6
Humidity (%) 72 73 74 73 72 71 69 66 59 57
Sun-loading (2.5 cm) × × × × △ × × ○ ○ ○
Cloud × × × × × × × × × ×
Indication (2.5 cm) ○ ○ ○ △ × × △ ○ ○ ○
Cooling/heating Cooling Cooling Cooling Cooling Cooling N/A Heating Heating Heating Heating

Time 2:00 PM 3:00 PM 4:00 PM 5:00 PM 6:00 PM 7:00 PM 8:00 PM 9:00 PM 10:00 PM 11:00 PM 12:00 AM

Ambient temperature (°C) 20.3 18.9 17.8 16.7 15.1 14.5 13.7 13.2 12.9 12.8 12.7
Humidity (%) 59 58 60 65 71 74 78 80 82 84 86
Sun-loading (2.5 cm) ○ △ ○ × × × × × × × ×
Cloud × × × × × × × × × × ○
Indication (2.5 cm) ○ × × ○ ○ ○ ○ ○ ○ ○ △
Cooling/heating Heating N/A N/A Cooling Cooling Cooling Cooling Cooling Cooling Cooling Cooling

Note: Sunrise; 7:12 AM, Sunset; 5:32 PM.

○: overall irradiation (sunloading); cloudy sky (cloud); clear indication (indication).
△: partial irradiation (sunloading); clear sky but some clouds (cloud); obscure indication (indication).
×: no irradiation (sunloading); clear sky without cloud (cloud); no indication (indication).

153
S. Hiasa et al. Measurement 121 (2018) 144–159

Fig. 10. Locations of gravel, wood chips, and tape.

were defined based on the record of the nearest climatological substa-
tion [40]. Since the maximum solar irradiance was 696 W/m2 on that
day, Is = 700 W/m2 was set in the software. In terms of convective heat
transfer, Kumar and Mullick [41] summarized several equations of
wind heat transfer coefficient provided by past research and compared
it to their experimental results. In their comparative study, Sharples and
Charlesworth [42] provided the closest value of wind induced con-
vective heat transfer coefficient h w to their experimental values as Eq.
(3), and Kumar et al. [43] established highest coefficient h w as Eq. (4);
h w = 6. 5 + 3. 3Vw (W/m2·K) Vw ⩽ 6 m/ s
h w = 10. 03 + 4. 687Vw (W/m2·K) Vw ⩽ 5 m/ s
The maximum wind speed was 4.9 m/s, while the maximum wind
gust speed was 7.1 m/s on that day according to the climatological
substation data [40]. In this software, the relational expression between
temperature and heat flux boundary conditions is defined as Eq. (5);
q0 = h (Text −Tamb)
where, q0 is inward heat flux normal to the boundary (W/m ) and Text
is the temperature far away from the modeled domain [39], which is
assumed as the sky temperature. Based on Eqs. (3), (4) and the wind
speed data, Text = Tamb−50 °C and h = 30 W/m2 K for all exposed sur-
faces as the heat flux boundary condition were utilized in the software
through iterative trials.
5.2. FE model simulation results of 1.3 cm deep delamination
Fig. 12 displays FE model simulation results for the concrete slab
with delaminations at 1.3 cm depth. In this figure, simulation results at
(3) some characteristic times were chosen. The temperature range of each
simulation result is set up to 5 K to keep consistency with IR images of
(4)
Fig. 5. Similar to the IRT results, delaminated areas at 9 AM and 4 PM
are obscure. The FE modeling could also find the interchange periods
between cooling effect and heating effect cycles for passive IRT, and
those time periods are same as the time periods that observed from the
field experiment. Furthermore, as can be seen in the IRT results, the
effect of the boundary condition can also be observed in the simulation
(5)
results at the same edges of the concrete slab as the experimental results
2
depending on the time.
In addition, Fig. 13 compares both temperatures from experimental

Table 4
Results of which obstacles were distinguished from the IR images.

Obstacle G-1 G-2 G-3 G-4 G-5 G-6 W-1 W-2 T-1 T-2 T-3 T-4 T-5 T-6 T-7 T-8 T-9 T-10 T-11 T-12 T-13 T-14

7:00 AM ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ × ○ ○ ○ ○ ○ ○
7:30 AM ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ × ○ ○ ○ ○ ○ ○
8:00 AM ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ × ○ ○ ○ ○ ○ ○
8:30 AM ○ ○ ○ ○ ○ △ ○ ○ ○ ○ ○ ○ ○ △ ○ × ○ △ ○ ○ ○ ○
9:00 AM ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ × ○ ○ ○ △ ○ ○ ○ ○ ○
9:30 AM ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ △ ○ × ○ △ ○ ○ ○ ○
10:00 AM ○ ○ ○ △ ○ ○ ○ ○ ○ ○ ○ ○ × × ○ × ○ ○ ○ △ ○ ○
11:00 AM ○ ○ ○ ○ ○ ○ ○ ○ ○ △ ○ ○ ○ ○ ○ ○ △ ○ ○ △ × △
12:00 PM ○ ○ ○ ○ ○ ○ ○ ○ ○ △ ○ ○ ○ ○ ○ ○ △ ○ ○ ○ × ○
1:00 PM ○ ○ ○ ○ ○ ○ ○ ○ ○ △ ○ ○ ○ ○ ○ ○ △ ○ ○ ○ × ○
2:00 PM ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ × ○ ○ × △
3:00 PM ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ △ ○ △ ○ × ○ ○ ○ ○
4:00 PM △ △ ○ ○ ○ △ ○ ○ ○ ○ ○ ○ × × ○ × ○ △ ○ × ○ ○
5:00 PM ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ △ ○ ○ ○ × ○ ○
6:00 PM ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○
7:00 PM ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○
8:00 PM ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○
9:00 PM ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ △ ○ ○ ○ ○ ○ ○
10:00 PM ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ △ ○ ○ ○ ○ ○ ○
11:00 PM ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ ○ △ ○ ○ ○ ○ ○ ○
12:00 AM ○ ○ ○ ○ ○ ○ ○ ○ △ △ × △ × △ ○ × × △ △ × × ×

○: clearly distinguished, △: not clear, but somehow distinguished, ×: not distinguished.

154
S. Hiasa et al. Measurement 121 (2018) 144–159

Fig. 11. Concrete blocks and the FE simulation model.

Table 5 temperature than the simulation result in the morning. In terms of the
Material properties of the simulation model. difference at 12 AM, the effect of clouds can be assumed as mentioned
in Section 4.2.
Material properties Unit Concrete Styrofoam Ground
Furthermore, Fig. 14 compares temperature differences between
Thermal conductivity W/(m K) 1.6 0.024 0.6 sound and delaminated parts of IRT and FE model simulation at each
Heat capacity at constant pressure J/(kg K) 880 1130 800 time. Since the most important factor for IRT is the temperature dif-
Density kg/m3 2300 25 1500
ference between sound and delaminated parts, the reproducibility of it
Solar absorptivity 0.6 –
Surface emissivity 0.88 –
by FE model simulation was compared with the IRT result. As can been
seen in the graph, the reproducibility of FE model simulation is very
close to the IRT result even during the morning. It should be noted that
IRT scans and FE simulation results at delaminated and sound parts of the temperature differences varied at most approximately 0.7 °C de-
the concrete surface. Since the temperature of the concrete surface pending on the camera specifications even when using the same IRT
measured by the thermocouple indicated a much higher temperature method in previous experiment to investigate the effect of camera
than the IR camera as mentioned above, results from the IR camera are specifications [21]. Therefore, the small differences between the two
used for the validation of FE model simulation. It is obvious that the results can be considered within the measurement error range. In this
simulation result shows very close values for both delaminated and result, temperature differences during the morning are also simulated
sound parts except during the morning and at 12 AM. Regarding the accurately even though the concrete surface was shaded and the si-
difference in the morning, it can be considered that the FE model si- mulation result indicated a higher temperature than the real surface
mulation assumed a clear sky the whole day, while the concrete surface temperature recoded by IRT at that time. Thus, it can be concluded that
was shaded until 9 AM as can be seen in Fig. 6, and it caused a lower this FE model simulation can reproduce temperature difference

Fig. 12. Simulation results of temperature distribution of the concrete slab surface (1.3 cm depth).

155
S. Hiasa et al.
Fig. 13. Temperatures from experimental IRT and FE model simulation results (1.3 cm depth).
Fig. 14. Comparison of temperature differences (1.3 cm depth).
between sound and delaminated areas regardless of the surface being
shined or shaded.
5.3. FE model simulation results of 2.5 cm deep delamination
Fig. 15 depicts the ambient air temperature and both the tempera-
tures from the experimental IRT scan and simulation results at the
delaminated and sound parts of the concrete surface. This simulation
result also shows values that are very close to the IRT results, even
though the temperatures are higher than the IRT results during the
morning. The surface temperatures of the IRT results at 12 AM also
increased for this concrete slab in the same way as the 1.3 cm depth
specimen. Therefore, the same reasons can be assumed; the surface was
shaded in the morning, until 11 AM for this specimen, and clouds
prevented radiative cooling at 12 AM. Furthermore, temperature dif-
ferences between sound and delaminated parts of IRT and FE model
simulation at each time were also compared as displayed in Fig. 16. As
can been seen in the graph, the reproducibility of the FE model simu-
lation is also high through the data collection time.
Based on the results of FE model simulation for the two concrete
slabs, it can be concluded that this FE model simulation can reproduce
temperature differences between sound and delaminated areas with
high accuracy.
156
Measurement 121 (2018) 144–159
6. Discussion on suitable IRT application time window for bridge
deck inspections
Through the field experiment and FE model simulation, this study
investigated time windows that should be avoided to inspect bridge
decks using IRT technique. The field experiment proves that there are
interchange periods between the nighttime cooling effect and the
daytime heating effect, and IRT cannot detect delamination during
these periods, 9–9:30 AM and 3–4 PM in this case, as shown in Tables 2
and 3. In terms of FE modeling, the simulation result points out the
interchange points between the nighttime cooling effect and the day-
time heating effect around 9 AM and 4 PM. These times match with the
IRT results, from 9 AM to 9:30 AM and from 3 PM to 4 PM. Thus,
through the FE model simulations and the results from IRT data, the
period of these interchanges were assumed to be about 1–2 h under the
given conditions as shown in Fig. 17. Therefore, IRT should not be
applied before and after at least 1 h from the interchange points of the
simulation result.
Furthermore, higher temperature differences in the day can be ob-
served from both results around 11 AM to 2 PM and 6 PM to 11 PM, and
those time windows must be preferable for IRT. However, as can be
seen in Fig. 17, the available time during daytime is shorter than the
early morning or nighttime in this case; thus, if IRT is used during
daytime, there is a possibility that first or last few hours are un-
detectable time zones. On the other hand, if IRT is implemented during
nighttime, the maximum temperature difference occurs at night,
around 7 PM in this case (one and a half hours after sunset), and then it
decreases gradually from that time to the next morning. Subsequently,
it reaches to the interchange point a few hours after the sunrise. Hence,
if IRT is applied after sunset, there is enough time until the interchange
point of the next morning. It should be noted that if IRT inspection
starts early in the morning, there is a high possibility to reach the in-
terchange period during the bridge inspection.
Moreover, surface temperature distribution during daytime varies
depending on the location due to sunlight, even though the delaminated
areas were also indicated clearly during the daytime. Fig. 18(a) shows
an example of the effect of sunlight around noontime; the left image
was recorded 40 min after sunrise while the right image was taken
around noon [10]. The left image indicates delaminated areas very
clearly. On the other hand, as can be seen in the right image of
Fig. 18(a), even sound areas indicate different temperature distribution
depending on the location. This ununiformity of concrete surface
S. Hiasa et al. Measurement 121 (2018) 144–159

Fig. 15. Temperatures from IRT and FE model simulation results (2.5 cm depth).

(a) 1.3 cm depth of delamination

Fig. 16. Comparison of temperature differences (2.5 cm depth).

temperature tends to cause misdetection for IRT application. In the left

image of Fig. 18(a), the delaminated areas show cooler temperatures
than the surrounding areas, indicating this IR image was taken during
the cooling cycle. Thus, this example supports the conclusion that
cooling cycle is better than heating cycle to conduct bridge deck in-
spections using IRT technique.
Fig. 18(b) shows another example of the effect of sunlight during
daytime. This IR image was taken at a sidewalk of a bridge deck in
Jacksonville, FL, around noon time [44]. The enclosed area was de-
tected by IRT, and the delaminated area was validated by hammer
sounding. The delaminated area of the shined area was easily detected
from the IR image while the shaded area by the railing was difficult to
evaluate the condition from the IR image since the surface temperature (b) 2.5 cm depth of delamination
was heavily influenced by the shade. Thus, daytime IRT application to
bridge deck inspections is also affected by shade, and this example also Fig. 17. Available time window for IRT application.

support the conclusion that nighttime is better than daytime regarding

bridge deck inspections using IRT in order to reduce the possibility of
misdetection, even though daytime application can also detect dela-
minated areas.
Through field experiments and FE modeling, it can be concluded
that nighttime IRT application is the most suitable time window for
concrete bridge deck inspections under the natural environment (pas-
sive condition) in order to reduce the possibility of misdetection.
7. Conclusions
Application time window is one of the major concerns of effective
utilization of the IRT on civil engineering structures. This study ex-
plores the favorable time windows for concrete bridge deck inspections
by IRT through field experiment and FE model simulations. In this
experiment, two concrete slabs, which have delaminations at 1.3 cm
and 2.5 cm depths were used. In addition, how debris affects the result

157
S. Hiasa et al.
(a) IR images taken at different time (Source:[10])
Fig. 18. Potential sunlight effect for daytime IRT application to bridge decks.
of IRT for bridge deck inspections was investigated by placing several
types of obstacles on the concrete slab, which has 2.5 cm deep dela-
minations. In this study, IR images were taken every hour from 7 AM to
12 AM. Furthermore, in order to capture the interchange point from the
nighttime cooling effect to the daytime heating effect, IRT data were
collected at 30-min intervals from 7 AM to 10 AM.
In terms of IRT results of 1.3 cm depth delaminations on the test
specimen, both delaminations can be detected except for 9:30 AM, 3 PM
and 4 PM data sets. This result proved that there are interchange per-
iods between the nighttime cooling effect and the daytime heating ef-
fect, and IRT cannot detect delamination during these periods. In ad-
dition, the period of these interchanges were within 1 h in the morning
and within 2 h in the evening for this experiment under the given
conditions. Furthermore, it was observed that the sides, which were
exposed to sunlight (including the edge) showed much higher tem-
peratures than the other parts of the surface. Therefore, those parts
require another temperature range setting, making it more challenging
to evaluate the IRT results. Moreover, by setting several temperature
ranges for the same IRT image, it might yield misdetections. Therefore,
it can be concluded that the preferable time period to apply IRT for
concrete bridge deck inspection is during the nighttime cooling cycle in
order to reduce the possibility of misdetection, even though the dela-
minated areas are also observed clearly during the daytime heating
period in this study.
As for the FE model simulation, a highly-reproducible model was
constructed by considering the phenomenon of radiative cooling. In this
model, the sky temperature (external air temperature) was set up as
−50 °C from the ambient air temperature through iterative trials.
Especially for the temperature differences between the sound and de-
laminated parts (the most important factor for IRT), the FE model si-
mulation performed high reproducibility compared with the experi-
mental IRT results. Based on the simulation and IRT results, higher
temperature differences in the day can be observed from both results
around noontime and nighttime, and those time windows must be
preferable for the field implementation of the IRT. However, con-
sidering the IRT results, IRT is affected by sun loading during the
daytime heating cycle and it might cause misdetections while IRT is not
affected by the sun during the nighttime cooling cycle. Furthermore, it
can be assumed from the simulation result that the maximum effect in
cooling cycle is at night (around 7 PM in this case) and then the tem-
perature difference decreases gradually from that time to the next
morning; subsequently, it reaches to the interchange point a few hours
after sunrise. Therefore, the available time during cooling cycle is much
longer than that during heating cycle in the simulated case. Moreover,
if IRT is used during daytime, there is a possibility that first or last few
hours are undetectable time zones due to the interchange periods.
Regarding the time window during cooling cycle, although IRT would
show clear indications of the delaminated areas during the morning, it
would be weaker than the indications of the previous night. In addition,
there is a high possibility to reach to the interchange period during the
158
Measurement 121 (2018) 144–159
(b) IR image taken around noon
(Source: [44])
bridge inspection if IRT is utilized for real bridge deck inspections in the
early morning. On the other hand, if IRT is applied after sunset, there is
an enough time till the interchange period (next morning). Thus, it can
be concluded that nighttime application of IRT is the most suitable time
window for concrete bridge deck inspections.
In addition to the optimal data collection time, this experimental
study observed the effects of clouds and radiative cooling on IRT for
concrete decks. The damage indication during the clear sky condition
was much clearer than the indication under cloudy condition, in-
dicating that the clear sky is also the preferable condition for IRT ap-
plication under passive condition. Therefore, optimal conditions for IRT
implementation on concrete bridge decks can be concluded that
nighttime under the clear sky condition.
Since FE model is consistently agreeing with the IRT data collected
from the field, it might be possible to develop such a model to de-
termine the favorable times as opposed to visiting the structure to in-
stall test specimens [20] to determine the best times for field scans.
Considering that the authors are focusing on rapid bridge deck scans
with instrumented vehicles operating at highway speeds [9,18], elim-
ination of prior access and installation of test specimens will tre-
mendously increase the rate of structures being scanned and evaluated.
In addition, such models validated with field data, will allow for further
simulations of the bridge deck under environmental and mechanical
inputs.
Regarding the effect of obstacles on concrete bridge deck surfaces, it
can be concluded that debris on the concrete surfaces can be dis-
tinguished on the IRT images since the debris have much smaller sur-
face areas, so that they are cooled down/heated up much faster than the
concrete surface. Furthermore, it can be assumed that if visual high-
definition images are taken at the same time, most of the debris can be
easily distinguished by comparing IR and HD images. Therefore, re-
moving debris on the bridge deck may not always be required for
concrete bridge deck inspections by IRT.
Acknowledgments
This work was supported by West Nippon Expressway Company
Limited (NEXCO-West), Scientific and Technological Research Council
of Turkey (TÜBİTAK), and National Science Foundation (NSF CMMI
#1463493). The authors would like to express our sincere gratitude to
Mr. Juan Cruz and our other research group members for their support
with manufacturing test specimens and setting up equipment. The au-
thors also wish to express sincere appreciation to Andy Derewiany for
providing editing help.
References
[1] FHWA, Highway Bridges by Deck Structure Type 2016, 2017. https://www.fhwa.
dot.gov/bridge/nbi/no10/deck16.cfm (Accessed November 20, 2017).
[2] N. Gucunski, S. Kee, H. La, B. Basily, A. Maher, Delamination and concrete quality
assessment of concrete bridge decks using a fully autonomous RABIT platform,
S. Hiasa et al.
Struct. Monit. Maint. 2 (2015) 19–34.
[3] M.C. Brown, J.P. Gomez, M.L. Hammer, J.M. Hooks, Long-Term Bridge
Performance High Priority Bridge Performance Issues, FHWA-HRT-1, McLean, VA,
2014.
[4] GPO, Electronic Code of Federal Regulations, U.S. Gov. Publ. Off., 2017. http://
www.ecfr.gov/cgi-bin/text-idx?rgn=div5&node=23:1.0.1.7.28 (Accessed
September 28, 2017).
[5] S. Kashif Ur Rehman, Z. Ibrahim, S.A. Memon, M. Jameel, Nondestructive test
methods for concrete bridges: A review, Constr. Build. Mater. 107 (2016) 58–86,
http://dx.doi.org/10.1016/j.conbuildmat.2015.12.011.
[6] C. Koch, K. Georgieva, V. Kasireddy, B. Akinci, P. Fieguth, A review on computer
vision based defect detection and condition assessment of concrete and asphalt civil
infrastructure, Adv. Eng. Informatics 29 (2015) 196–210, http://dx.doi.org/10.
1016/j.aei.2015.01.008.
[7] M.C. Forde, Differences in international strategy for the NDT of concrete, in:
O. Günes, Y. Akkaya (Eds.), Nondestruct. Test. Mater. Struct. Springer Netherlands,
Dordrecht, 2013, pp. 1217–1227, , http://dx.doi.org/10.1007/978-94-007-0723-8.
[8] M.C. Forde, International practice using NDE for the inspection of concrete and
masonry arch bridges, Bridg. Struct. 6 (2010) 25–34, http://dx.doi.org/10.3233/
BRS-2010-004.
[9] S. Hiasa, F. Necati Catbas, M. Matsumoto, K. Mitani, Considerations and issues in
the utilization of infrared thermography for concrete bridge inspection at normal
driving speeds, J. Bridg. Eng. 22 (2017) 4017101, http://dx.doi.org/10.1061/
(ASCE)BE.1943-5592.0001124.
[10] N. Gucunski, A. Imani, F. Romero, S. Nazarian, D. Yuan, H. Wiggenhauser, et al.,
Nondestructive Testing to Identify Concrete Bridge Deck Deterioration,
Transportation Research Board, Washington, D.C., 2013. doi: 10.17226/22771.
[11] K. Vaghefi, T. (Tess) M. Ahlborn, D.K. Harris, C.N. Brooks, Combined imaging
technologies for concrete bridge deck condition assessment, J. Perform. Constr.
Facil. 29 (2015) 4014102, http://dx.doi.org/10.1061/(ASCE)CF.1943-5509.
0000465.
[12] S.-H. Kee, T. Oh, J.S. Popovics, R.W. Arndt, J. Zhu, Nondestructive bridge deck
testing with air-coupled impact-echo and infrared thermography, J. Bridg. Eng. 17
(2012) 928–939, http://dx.doi.org/10.1061/(ASCE)BE.1943-5592.0000350.
[13] T. Oh, S. Kee, R.W. Arndt, J.S. Popovics, M. Asce, J. Zhu, Comparison of NDT
methods for assessment of a concrete bridge deck, J. Eng. Mech. 139 (2013)
305–314, http://dx.doi.org/10.1061/(ASCE)EM.1943-7889.0000441.
[14] F. Khan, M. Bolhassani, A. Kontsos, A. Hamid, I. Bartoli, Modeling and experimental
implementation of infrared thermography on concrete masonry structures, Infrared
Phys. Technol. 69 (2015) 228–237, http://dx.doi.org/10.1016/j.infrared.2015.02.
001.
[15] D.M. McCann, M.C. Forde, Review of NDT methods in the assessment of concrete
and masonry structures, NDT E Int. 34 (2001) 71–84, http://dx.doi.org/10.1016/
S0963-8695(00)00032-3.
[16] X.P.V. Maldague, Introduction to NDT by active infrared thermography, Mater.
Eval. 60 2002, pp. 1060–1073 http://w3.gel.ulaval.ca/∼maldagx/r_1221t.pdf.
[17] R.C. Waugh, Development of infrared techniques for practical defect identification
in bonded joints, Dev. Infrared Tech. Pract. Defect Identif. Bond. Joints, 1st,
Springer International Publishing, 2016, pp. 21–37 10.1007/978-3-319-22982-9..
[18] S. Hiasa, F.N. Catbas, M. Matsumoto, K. Mitani, Monitoring concrete bridge decks
using infrared thermography with high speed vehicles, Struct. Monit. Maintenance,
An Int. J. 3 (2016) 277–296, http://dx.doi.org/10.12989/smm.2016.3.3.277.
[19] ASTM, D4788, Standard Test Method for Detecting Delaminations in Bridge Decks
Using Infrared Thermography, 3rd, ASTM International, West Conshohocken, PA,
USA, 2014, , http://dx.doi.org/10.1520/D4788-03R13.2.
[20] A. Watase, R. Birgul, S. Hiasa, M. Matsumoto, K. Mitani, F.N. Catbas, Practical
identification of favorable time windows for infrared thermography for concrete
bridge evaluation, Constr. Build. Mater. 101 (2015) 1016–1030, http://dx.doi.org/
10.1016/j.conbuildmat.2015.10.156.
[21] S. Hiasa, R. Birgul, F.N. Catbas, Investigation of effective utilization of infrared
thermography (IRT) through advanced finite element modeling, Constr. Build.
Mater. 150 (2017) 295–309, http://dx.doi.org/10.1016/j.conbuildmat.2017.05.
175.
[22] FHWA, Federal highway administration research and technology: nondestructive
evaluation (NDE) Web Manual, 2015. https://fhwaapps.fhwa.dot.gov/ndep/
159
Measurement 121 (2018) 144–159
DisplayTechnology.aspx?tech_id=14 (Accessed September 28, 2017).
[23] S. Hiasa, R. Birgul, F.N. Catbas, Infrared thermography for civil structural assess-
ment: demonstrations with laboratory and field studies, J. Civ. Struct. Heal. Monit.
6 (2016) 619–636, http://dx.doi.org/10.1007/s13349-016-0180-9.
[24] S. Hiasa, R. Birgul, F.N. Catbas, Effect of defect size on subsurface defect detect-
ability and defect depth estimation for concrete structures by infrared thermo-
graphy, J. Nondestruct. Eval. 36 (2017) 57, http://dx.doi.org/10.1007/s10921-
017-0435-3.
[25] S. Hiasa, R. Birgul, F. Necati Catbas, A data processing methodology for infrared
thermography images of concrete bridges, Comput. Struct. 190 (2017) 205–218,
http://dx.doi.org/10.1016/j.compstruc.2017.05.011.
[26] G. Washer, R. Fenwick, N. Bolleni, Development of Hand-held Thermographic
Inspection Technologies, Report No. OR10-007, 2009.
[27] G. Washer, R. Fenwick, N. Bolleni, Effects of solar loading on infrared imaging of
subsurface features in concrete, J. Bridg. Eng. 15 (2010) 384–390, http://dx.doi.
org/10.1061/(ASCE)BE.1943-5592.0000117.
[28] S. Yehia, O. Abudayyeh, S. Nabulsi, I. Abdelqader, Detection of common defects in
concrete bridge decks using nondestructive evaluation techniques, J. Bridg. Eng. 12
(2007) 215–225, http://dx.doi.org/10.1061/(ASCE)1084-0702(2007) 12:2(215).
[29] C. Meola, A new approach for estimation of defects detection with infrared ther-
mography, Mater. Lett. 61 (2007) 747–750, http://dx.doi.org/10.1016/j.matlet.
2006.04.120.
[30] D. Almond, P. Patel, Photothermal Science and Techniques, 1st editio, Chapman &
Hall, London, UK, 1996.
[31] G.C. Holst, Common sense approach to thermal imaging, JCD Publishing, Winter
Park, FL, USA, 2000.
[32] J. Tashan, R. Al-mahaidi, A. Mamkak, Defect size measurement and far distance
infrared detection in CFRP-concrete and CFRP-steel systems, Aust. J. Struct. Eng.
7982 (2015), http://dx.doi.org/10.1080/13287982.2015.1116177 In press.
[33] P. Cotič, D. Kolarič, V.B. Bosiljkov, V. Bosiljkov, Z. Jagličić, Determination of the
applicability and limits of void and delamination detection in concrete structures
using infrared thermography, NDT E Int. 74 (2015) 87–93, http://dx.doi.org/10.
1016/j.ndteint.2015.05.003.
[34] J. Kondo, Atmospheric Science near the Ground Surface, 3rd editio, University of
Tokyo Press, Tokyo, Japan, 2000.
[35] T. Worzewski, R. Krankenhagen, M. Doroshtnasir, Thermographic inspection of
wind turbine rotor blade segment utilizing natural conditions as excitation source,
Part II: The effect of climatic conditions on thermographic inspections – A long term
outdoor experiment, Infrared Phys. Technol. 76 (2016) 767–776, http://dx.doi.org/
10.1016/j.infrared.2016.04.012.
[36] K. Hashimoto, Y. Akashi, Points to consider for photography by infrared cameras
with different wavelength detection region, in: 65th JSCE Annu. Meet., Japan
Society of Civil Engineers (JSCE), Sapporo, Japan, 2010, pp. VI-160.
[37] T.E. ToolBox, The Engineering ToolBox: Absorbed Solar Radiation, 2017, https://
www.engineeringtoolbox.com/solar-radiation-absorbed-materials-d_1568.html
(Accessed December 26, 2017).
[38] T.E. ToolBox, The Engineering ToolBox: Emissivity Coefficients of some common
Materials, 2017, https://www.engineeringtoolbox.com/emissivity-coefficients-d_
447.html (Accessed December 26, 2017).
[39] COMSOL, Heat Transfer Module User’s Guide Version 5.1, 2015.
[40] WeatherUnderground, Weather History for Orlando, FL [KFLORLAN179], 2015,
http://www.wunderground.com/personal-weather-station/dashboard?ID=
KFLORLAN179#history/tdata/s20151219/e20151219/mdaily.
[41] S. Kumar, S.C. Mullick, Wind heat transfer coefficient in solar collectors in outdoor
conditions, Sol. Energy 84 (2010) 956–963, http://dx.doi.org/10.1016/j.solener.
2010.03.003.
[42] S. Sharples, P.S. Charlesworth, Full-scale measurements of wind-induced convective
heat transfer from a roof-mounted flat plate solar collector, Sol. Energy 62 (1998)
69–77.
[43] S. Kumar, V.B. Sharma, T.C. Kandpal, S.C. Mullick, Wind induced heat losses from
outer cover of solar collectors, Renew. Energy 10 (1997) 613–616.
[44] S. Hiasa, E. Karaaslan, F.N. Catbas, Considerations for infrared thermography field
implementations and data analysis for bridge inspections, in: TRB 97th Annu.
Meet., Transportation Research Board, Washington, D.C., USA, 2018.