Automated Detection and Location of Leaks in Water MainsUsing Infrared Photography
Mohamed Fahmy, P.Eng., A.M.ASCE1; and Osama Moselhi, P.Eng., F.ASCE2
Abstract: Leakage of water distribution networks is the most common reason of undesirable losses of potable water. Problems associatedwith water main leaks pose growing concern around the globe. These problems include water and energy loss, in addition to the risk itposes to structural damage of adjacent properties. In current practice, not all water leaks can be detected in a timely and cost effectivemanner. This paper presents a study conducted for detection of water leaks in underground pipelines, and identification of their respectivelocations using thermography infrared camera. The paper describes the field work and the experimental protocol which were carried outover 2 years in three different locations in greater Montréal �Canada� area in order to investigate factors that affect the applicability andlimitations of using IR camera in water leak detection. These factors beyond those studied in previous work carried out by American WaterWorks Association and National Research Council. The paper presents a model developed to determine approximate location of leaks inwater mains. The developed model was then applied successfully to detect and locate leaks in water mains in fall and spring seasons. Itfailed, however, to detect leaks in the summer and winter due to high pavement temperature and the snow coverage, respectively.
DOI: 10.1061/�ASCE�CF.1943-5509.0000094
CE Database subject headings: Water pipelines; Leakage; Failure investigations; Photography; Canada.
Author keywords: Water pipelines; Thermography; Water leak detection; Failure investigations.
Introduction
Current methods for water leak detection have number of limita-tions: �1� they are considered to be satisfactory for metallic pipeslocated in urban areas; �2� they are relatively time-consuming andconsequently expensive, particularly for short length pipeline; and�3� they are normally affected by traffic and population densities.The main objective of this research was to develop methodologythat can detect water leaks effectively and pinpoint their respec-tive locations, in addition to overcoming limitations of currentmethods.
Thermography �infrared� camera measures and images theemitted IR radiation from an object. It can detect thermal con-trasts at pavement surface due to water leaks of pipes below thatsurface. In addition, it enables relatively large areas to be inves-tigated effectively in less time and consequently less cost com-paring to currently leak detection methods, which are either point-testing or line-testing methods. It is also independent of pipematerial type and size. Also, it can be used in day or night time.These advantages make use of the IR camera overcomes limita-tions associated with currently leak detection methods. However,
1Ph.D. Candidate, Dept. of Building, Civil, and Environmental Engi-neering, Concordia Univ., 1455 Blvd. de Maisonneuve W, Montreal QC,Canada H3G1M8 �corresponding author�. E-mail: [email protected]
2Professor, Dept. of Building, Civil, and Environmental Engineering,Concordia Univ., 1455 Blvd. de Maisonneuve W, Montreal QC, CanadaH3G1M8. E-mail: [email protected]
the radiation measured by IR camera does not only depend on thetemperature of the object but also is a function of the emissivity,which is affected by many factors such as weather conditions,soil, and pavement surface conditions. It should be noted that theuse of IR camera as a leak detector method is not new. Weil�1998� used it in detecting leaks in sewer pipes and identifiedtheir respective locations as an affected area rather than pinpointlocation of leaks. Previous study carried out by National ResearchCouncil �NRC� in collaboration with American Water Works As-sociation in detecting water leaks �Hunaidi et al. 2000�. Theyperformed thermography survey of a simulated leak area at theNRC leak detection facility during cloudless night in fall season.This paper presents a study conducted to investigate factors be-yond those considered in previous research. The presented studywas carried out in two successive years which represents a widerange of weather and prevailing light conditions in three differentlocations with varied groundwater level. The study also consid-ered the existence of adjacent sewer pipes, setup of IR camera,and vehicle speed �i.e., on which the camera is mounted�, alongwith their impact on the accuracy of the results obtained. A caseexample is presented to illustrate the use of the proposed meth-odology.
Thermography „IR… Camera System
The ThermaCAM S 60 IR condition monitoring system was usedin conducting a set of field experiments. The system consists of anIR camera with a built in 24° lens, a visual color camera, a laserpointer, and IR communications link �FLIR Systems AB 2004�.This system provides real time high resolution color images inboth IR and visual modes. The visual mode was used to check theexistence of any foreign bodies on the pavement surface whichmight affect thermal contrast. To document the thermal variationon pavement surface due to water leaks of pipes below ground it
is possible to capture and store images on a removable flash card.
CE / MAY/JUNE 2010
2010.24:242-248.
Dow
nloa
ded
from
asc
elib
rary
.org
by
SOU
TH
ER
N C
AL
IFO
RN
IA U
NIV
ER
SIT
Y o
n 08
/24/
13. C
opyr
ight
ASC
E. F
or p
erso
nal u
se o
nly;
all
righ
ts r
eser
ved.
The captured images that have sequential numbers can then beanalyzed in the field using the developed methodology to deter-mine approximate locations of leaks. Fig. 1 shows the thermalcontrast between areas with lower temperature �dark areas� thatrepresent pavement surface temperature in its natural state �i.e.,without leaks� and the bright areas that indicate water leak.
Proposed Methodology
The methodology presented in this study is based on intensiveliterature review, meeting with experts, and on the analysis ofactual data collected from fields located in three municipalities;Pierrefonds, southwest, and downtown Montreal �Canada�. Thedevelopment of the methodology involves six major steps: �1�identification of factors that affect thermal contrast at pavementsurface; �2� field investigation and on-site experimental work; �3�analysis of the data obtained; �4� determine most suitable condi-tions of using IR camera for the detecting and locating waterleaks; �5� establish the relationship between the detected leakagearea at pavement surface and the location of leak in the watermain being tested; and �6� validation of the proposed methodol-ogy by comparing leak locations detected by the proposed systemand by acoustic-based methods. These steps described below fol-lowing the chart shown in Fig. 2.
Step 1: Factors That Affect Thermal Contrast atPavement Surface
Based on a comprehensive literature review, seven interviewswith experts, and preliminary field investigation it became clearthat heat balance at pavement surface in addition to heat andmoisture transfer in soil are two major factors that affect the ther-mal contrast at pavement surface.
At the pavement surface, four modes of heat transfer are con-sidered: conduction into the pavement layer, convection, solarabsorption, and gray-body irradiation to the surrounding. The heatflow contribution due to conduction assuming steady-state heat
Fig. 1. IR camera shows thermal contrast at pavement surface due towater leak
flow in one direction can be expressed as �American Society
JOURNAL OF PERFORMANC
J. Perform. Constr. Facil.
of Heating, Refrigerating and Air-Conditioning Engineers�ASHRAE� 1981; Hutcheon and Handegord 1983; Bentz 2000�
Qcond = Kcond+ �Tp − Ts�/L �W/m2� �1�
where Kcond=average thermal conductivity of the soil and pave-ment in �w/m K�; Tp and Ts=pipe temperature and surface tem-perature, respectively; and L=length of the flow path �i.e., burialdepth�.
For convection, the heat flow can be expressed as �Hutcheonand Handegord 1983; Schlangen 2000�
Qconv = hconv� �Ts − Tambiant� �W/m2� �2�
where hconv=convection coefficient in �w/m K� and Tambient
=ambient temperature. The convection coefficient can be calcu-lated based on wind speed as follows:
For radiative heat transfer at the top pavement surface, twocontributions are considered the first is radiation absorbed fromthe incoming sunlight. For incoming heat flow due to this sourcecan be expressed as �Mcullough and Rasmussen 1999�
Qsun = �abs � Qinc�W/m2� �5�
where Qinc= incident solar radiation �W /m2� and �abs=solar ab-sorptivity of the pavement. In this study Qinc was taken fromweather data files while a value of 0.8 was used for �abs �Loomanset al. 2003�. The second is the emission of radiation from thepavement to the sky, which can be expressed as �American Soci-ety of Heating, Refrigerating and Air-Conditioning Engineers�ASHRAE� 1981; Hutcheon and Handegord 1983; Bentz 2000�
Qsky = � € � �Ts4 − Tsky
4 � �W/m2� �6�
where �=Stefan-Boltzmann constant �5.669�10−8 W /m2 °C4�and €=emissivity of the pavement. In this research values of €
=0.94 for asphalt pavement was used �FLIR Systems AB 2004;Moher and Osborne 2006�. Ts�pavement surface temperature �inK� and Tsky�calculated sky temperature in K.
The sky temperature is estimated based on an algorithm pre-sented by Walton �1985�, using the following series of equations:
N= “tenth cloud cover,” taking values between 0.0 and 1.0As to the second factor, migration of heat and moisture in soil
is a coupled energy and mass transport process, which is affectedby the field distribution of temperature, pressure, and velocity�Liu et al. 2005�. Soil heat transfer in the unsaturated zone of thesoil is the sum of fluxes due to heat conduction and convection.The soil heat flow can be expressed as �Sung et al. 2002�
where T=soil temperature in °C; ����=thermal conductivity ofsoil in W cm−1 K−1; Cs and Cw=volumetric specific heats inJ cm−3 K−1 for soil porous media and water, respectively; q=volumetric flux of water �or Darcy velocity� in cm day−1.
The volumetric specific heat Cs��� of the soil can be calculatedby the addition of the heat capacities of the various phases �DeVris and Afgan 1975�, as follows:
Cs = Cw�w+Ca�a+Co�o+Cm�m �11�
where Ci and �i=specific heat and the volumetric content of eachith component, respectively. The subscripts w, a, o, m indicatewater, air, organic matter, and mineral of the soil, respectively.
Horton and Chung �1991� simplified Eq. �11� as follows:
Cs = �1 − �s� � 1.92 � 106 + 4.18 � 106� �12�
where �s=soil water content at saturation in cm3 cm−3; � is thesoil water content in cm3 cm−3; and Cs is the volumetric specific
Step 2: Field Investigation and On-Site ExperimentalWork
Field investigation was conducted using thermography IR cameraand the results obtained were compared to the results obtainedusing acoustic-based system which will be referred to in thispaper as “leak finder.” Leak finder locates water leaks by detect-ing the sound or vibration induced by water leaking from pres-surized pipes. The leak finder system consists of acoustic sensorssuch as accelerometers and hydrophones, wireless signal trans-mitters and receivers, and an electronic processing unit. The sen-sors are attached at two contact points with the pipe �normally firehydrants� that bracket a suspected leak. The signals are transmit-ted from the sensors to the processing unit wirelessly. The pro-cessing unit computes the cross-correlation function of the twoleak signals to determine the time lag between them. It then cal-culates the location of the leak based on a simple algebraic rela-tionship between the time lag, sensor-to-sensor spacing, andsound propagation velocity in the pipe �Hunaidi et al. 2004�.
Preliminary field experiments showed that the moisture levelnear the pavement surface affects the pavement surface tempera-ture because of its major influence on the thermal properties ofthe soil �Fahmy and Moselhi 2009�. Furthermore, it was foundthat the thermal contrast detected by IR camera was close to theexact location of the leak detected by the acoustic-base leak finderdevice. Following the preliminary survey, detailed field investiga-tion and experimental work were conducted in order to determinethe thermal performance of water leaks in underground pipelinesand establish relationship between the detected leaking areas andthe accurate location of the leaks.
In order to attain these objectives 42 water pipelines werescanned using IR camera. The diameter of these pipes rangedfrom 150 to 200 mm and their length ranged from 48 to 300 m.The field tests were conducted in downtown Montreal, south-westMontreal, and the Pierrefonds municipalities in Canada. Thestudy presented in this research was carried out over 24-monthperiod from July 2005 to August 2007, and the timing of thefieldwork was selected to represent a wide range of weather con-ditions in terms of prevailing light and ambient air temperature.The inspection was also executed throughout a range of cloudcover from clear sky to overcast. This allowed testing the effec-tiveness of energy transfer between the sky and the investigatedpavements.
It should be noted that the measured temperatures using IRcamera were compared to those measured using thermocoupledevice. The average difference in measured temperature was2°C.
In order to obtain obvious color contrast in acquired images,the IR camera setup was adjusted based on number of trials. Thedistance from the pavement surface to the camera ranged from1.20 to 12.0 m. Combinations of various ranges of vehicle speed,on which the camera was mounted, and time intervals of captur-ing images were carried out. The vehicle speed ranged from 5 to20 km/h and the rate of capturing images ranged from image/2 sto image/10 s.
Step 3: Analysis of Data Obtained
Figs. 3–6 show the relationships found between pipe temperature,average ambient air temperature, and average pavement tempera-ture. As shown in Figs. 3–6 there is a strong relationship betweenthe averaged ambient air temperatures and observed daily pave-
ment surface temperatures. Figs. 3, 4, and 6 show a warmer pipe
CE / MAY/JUNE 2010
2010.24:242-248.
Dow
nloa
ded
from
asc
elib
rary
.org
by
SOU
TH
ER
N C
AL
IFO
RN
IA U
NIV
ER
SIT
Y o
n 08
/24/
13. C
opyr
ight
ASC
E. F
or p
erso
nal u
se o
nly;
all
righ
ts r
eser
ved.
temperature from Week Number 1 �i.e., first week of January� toWeek Number 22 �i.e., first week of June�, also from Week Num-ber 42 �i.e., middle of October� to Week Number 52 �i.e., end ofDecember�; that indicates high possibility in detecting water leaksusing IR camera according to Eq. �1�.
Step 4: Determine Most Suitable Conditions for UsingProposed System
Cloud Cover and Prevailing LightData collected in this research showed that pavement tempera-tures under clear sky and/or during day time were consistentlywarmer than pavement under cloudy condition and/or at night andearly morning. As a result, heat and moisture flow toward pave-ment surface generated from leak in warmer pipe will be de-creased; consequently, detection of leaks will be more accurateunder overcast condition between 11 p.m. and 6 a.m.
Change in Thermal Characteristics of Soil and PavementSurfaceSoils close to water leaks experience increase in moisture contentand may become saturated. Such change in moisture contentchanges the thermal characteristics of the soil and makes it moreconductive to heat relative to dry soil away from the leak. Thesoil temperature variation observed in this research through thefour seasons indicates that the soil temperature presents highervariation in shallow than in deeper depth. During summer, thetemperature difference at pavement surface between summer andwinter was 38°C, while at the depth of 1.00 m that difference intemperature decreased to about 14°C and at the depth of 1.80 m�i.e., the average depth of water mains� it further decreased toapproximately 2°C. During winter, the average soil temperatureof deeper layers is higher than that at the shallow soil depths. Itmeans that, during winter the heat is transferred from the deeper
-20
-15
-10
-5
0
5
10
15
1 2 3 4 5 6 7 8 9 10 11 12 13
TemperatureoC
Time/week
Winter
Pipe Temperature
AverageAmbientTemperature
AveragePavementTemperature
Fig. 3. Comparison between temperatures in winter season
0
5
10
15
20
25
30
14 15 16 17 18 19 20 21 22 23 24 25 26
TemperatureoC
Time/Week
Spring
Pipe Temperature
AverageAmbientTemperature
AveragePavementTemperature
Fig. 4. Comparison between temperatures in spring season
JOURNAL OF PERFORMANC
J. Perform. Constr. Facil.
soil depths to surface, while during the summer months it changesdirection. This is in agreement with the recent study conducted byAntonopoulos �2006�. Also during the winter period, the rate ofchange in soil temperature under snow cover was less due to lowthermal diffusivity and high albedo of the snow. It was found thatareas detected that have a thermal contrast on pavement surfacedecreased with soil surface evaporation during daytime andslightly increases during nighttime.
Infiltration into Adjacent Sewer PipesThe field investigation and experimental work carried out in thisresearch revealed that more than 40% of the water leaks detectedwere infiltrated into adjacent sewer pipes that prevent the mois-ture movement from reaching pavement surface, consequently,that type of leaks could not be detected using IR camera.
Ground Water TableThe groundwater table has a great influence on the use of IRcamera; experimental work conducted in the vicinity of SaintLaurence River in Montreal showed that the groundwater tablewas higher than the pipe level. The IR images captured to thepavement surface at this area showed no variation in the thermalproperties of the pavement surface.
Distance of Sensor from SourceThe impact of the distance from the pavement surface to thecamera was studied. Tests were conducted over a range from 1.20to 12.0 m as shown in Table 1. The experimental works revealedthat the more the distance between sensor �IR camera� and pave-ment increases the more the thermal contrast enhances and viseversa as shown in Figs. 7�a and b�. Therefore, more distinction ofleakage area were obtained from 12.0 m from the pavement sur-face.
0
5
10
15
20
25
30
27 28 29 30 31 32 33 34 35 36 37 38 39
TemperatureoC
Time/Week
Summer
Pipe Temperature oC
AverageAmbientTemperature
AveragePavementTemperature
Fig. 5. Comparison between temperatures in summer season
-10
-5
0
5
10
15
20
40 41 42 43 44 45 46 47 48 49 50 51 52
TemperatureoC
Time/Week
Fall
Pipe Temperature
AverageAmbientTemperature
AveragePavementTemperature
Fig. 6. Comparison between temperatures in fall season
Vehicle Speed and Rate of Capturing ImagesCombinations of 12 sets represent various ranges of vehicle speedand periodic capturing of images were carried out. The vehiclespeed ranged from 5 to 20 km/h and the rate of capturing imagesranged from image/2 s to image/10 s. the best results obtained interms of distinguishing thermal contrast and accuracy when thevehicle speed was set at 5 km/h and the rate of capturing imageswas set at image/2 s as shown in Fig. 8.
Effect of IR Camera SetupIn order to obtain obvious color contrast in acquired images, cam-era setup was adjusted based on sets of thirty six trials, it was
Table 1. Relationship between Distance of IR Camera from Pavement S
Distance �m� 1.2 3
Contrast Minimum Below average
(b)
(a)
Fig. 7. �a� Image captured 1.2 m from pavement surface �poor ther-mal contrast�; �b� image captured 6 m from pavement surface �aver-age thermal contrast�
0102030405060708090100
5 10 15 20
A ccuracy
ofResults
(%)
Vehicle Speed ( Km/hr)image/10 sec.
image/5 sec.
image/2 sec.
Fig. 8. Relationship between vehicle speed and accuracy of results
found that the emissivity, palette type, and noise reduction func-tion that reduce clutters were the most effective parameters. Thefinal selection of these parameters was as following: emissivitywas selected based on pavement status ranging from 0.85 forsnow cover, 0.90 for dry pavement surface, and 0.94 for wetpavement surface. Palette iron was selected which provides finestcontrast and color degradation ranged from blue �i.e., representslowest temperature� to white �i.e., represents highest tempera-ture�. Noise reduction function was activated.
Step 5: Establish Approximate Location of Leak
The approximate location of water leak carried out in this re-search was based on two major steps:1. Determination of areas that indicating thermal change at
pavement surface �i.e., water leaks�; and2. Establish the relationship between detected leaking area and
pipe burial depth.
Determination of Areas Indicating Water LeaksTwenty-five pipelines experienced water leaks were tested usingIR camera in order to develop Eq. �13�. This equation can be usedto approximately locate detected leaks. Then, the user has tomove to that location, and then determine the entire area thatexperience thermal contrast �i.e., to measure the diameter of cir-cular base of virtual cone as described below�
X =�N − 1� � 0.28S
R�13�
where X=approximate location of water leak from the originpoint �m�; N=chronological image number; S=average vehiclespeed �km/h�; and R=rate of capturing IR image �image/s�.
Establish the Relationship between Detected Leaking Areaand Pipe Burial DepthField observations conducted in this research revealed that thedetected thermal contrast due to water leak on pavement surfaceapproximately represents a circular base of a cone where its headrepresents the location of the leak in the pipe being tested �seeFig. 9�. Regression analysis was applied on the data collected inorder to establish the relationship between the dependant �� andthe independent variables, which are burial depth of pipe �B� andaverage diameter of the area experience thermal contrast at pave-ment surface due to water leak �D� as shown in
and Thermal Contrast
6 9 12
Average Above average Maximum
� B
D
Leak in Pipe
Pavement Surface
Fig. 9. Virtual cone that determines leak location
urface
CE / MAY/JUNE 2010
2010.24:242-248.
Dow
nloa
ded
from
asc
elib
rary
.org
by
SOU
TH
ER
N C
AL
IFO
RN
IA U
NIV
ER
SIT
Y o
n 08
/24/
13. C
opyr
ight
ASC
E. F
or p
erso
nal u
se o
nly;
all
righ
ts r
eser
ved.
o = 100.0 − 32.6B + 12.5D �14�
where D=average diameter of area indicating water leak �m� andB=burial depth of pipe �m�.
Based on t-test results, all t values �t/2 which is 2.26, and allp values � which is 0.05, the hypothesis H0 ��i=0� can berejected and Ha ��i�0� is true. A significant individual relation-ship is present for the model in Eq. �14�. Based on Table 2,because the test statistic F=86.03�F=F0.05=1.8799 and p val-ues �=0.05 which is the targeted level of significance for thisstudy, the hypothesis H0 ��i=0� can be rejected and Ha �one ormore of the factors are not equal to 0� is true. A significant overallrelationship is present for the model in Eq. �14�. According to theabove F-test and t-test results, the model has both an overallstatistical significance and individual statistical significance.
Step 6: Validation of Proposed Methodology
The leak locations detected using IR camera for 25 water leakswere compared to those detected using the acoustic-based leak
Fig. 10. Leak locations �m� using acoustic leak finder versus IRcamera
0
0.5
1
1.5
2
2.5
3
3.5
0 5 10 15 20 25 30 35
Abs
olut
eEr
ror(
m)
Distance(m) From Fire Hydrant toLeak LocationUsingIRCamera
AbsoluteError…
Fig. 11. Absolute error �m� using IR camera
JOURNAL OF PERFORMANC
J. Perform. Constr. Facil.
finder method. The results are shown in Figs. 10 and 11. Asshown in these figures the difference ranged from 1.01 to 2.30 m
Case Example
In this example we consider a 6-in. diameter CI water main seg-ment, 48.7-m length �i.e., the distance between two fire hydrants�,located 1.80 m below the ground in a residential area of the city.The municipality asks for performing a condition assessmentwork on that pipeline using thermography method �IR camera�and verifying the results by using the acoustic-based leak finder.The inspection team used a vehicle with speed 6 km/h and therate of capturing images was 0.5 image/s.
Applying the methodology described above as shown in Fig.12�a� the user found that
MS F P
2,139.4 86.03 0.000
24.9
1.80m
Fire hydrant 1Fire hydrant 2Image #6
16.80
IR Method
Acoustic Method15.41 m
3.04 m30.76 m
33.29 m
14.9 m
Leak
(b)
Determine location of suspected pipeline usingmetal detector, and then Mark its location at
pavement surface
Move over pipeline parallel to the marked lineusing the vehicle on which the camera is mounted
to scan pipeline being tested
Setup the IR camera including but not limited tospeed of capturing frame (image/sec), and then
define vehicle speed accordingly
Download the saved images to the laptop, andthen apply developed model to determine the
approximate location of leak
Approximate Location of Leak
Pipeline Experience Water Leak
(a)
Fig. 12. �a� Steps of detecting water leak using proposed methodol-ogy; �b� approximate location of leak
1. Image Number 6 showed thermal change as shown in Fig.12�b�; and
2. The user moved to the location represents Image Number 6and perform further investigation to determine boundaries ofarea indicating leak.
Applying Eq. �13�
X =�N − 1� � 0.28S
R=
�6 − 1� � 0.28 � 6
0.50= 16.80 m
Then the user moved 16.80 m from origin �i.e., Fire Hydrant #1�in order to determine boundaries of leaking area at pavementsurface, consequently determine average diameter which was 3.04m. The distance from origin to place of leak in pipeline beingtested using the developed methodology is 16.42 m and by usingacoustic-based method is 15.41 m. The results obtained is shownin Fig. 12�b�.
Summary and Concluding Remarks
This paper presented a study on the use of thermography IR cam-era for detecting and locating leaks of water mains. The studyencompassed field investigation and testing as well as modelingdevelopment. The field work was conducted on water mains inthree locations in the greater Montreal area. The IR camera de-tected successfully number of leaks as a thermal contrast at pave-ment surface that occurred in fall and spring seasons, while itfailed in detecting leaks occurred in summer and winter due tohigh pavement temperature and the snow coverage, respectively.The thermal contrast due to water leaks take a shape of nearcircular cone base with an angle that ranges from 80.4° to123.4°. The head of the cone represents the approximate locationof leak. However, using IR camera in vicinity of sewer pipe wasnot reliable. The near optimum diurnal time of using the camerawas between 6–8 a.m. The leaks detected using IR camera wascompared to those detected using acoustic-based leak findermethod. A case example is presented to demonstrate the use andaccuracy of the developed methodology.
Acknowledgments
The writers acknowledge the financial support provided by theNatural Sciences and Engineering Research Council of Canadaand the internal research grant provided by the Faculty of Engi-
neering and Computer Science, Concordia University.
American Society of Heating, Refrigerating and Air-Conditioning Engi-neers �ASHRAE�. �1981�. ASHRAE handbook fundamentals, N.Y.
Antonopoulos, V. �2006�. “Water movement and heat transfer simulationsin a soil under ryegrass.” Journal of Biosystems Engineering, 95�1�,127–138.
Bentz, D. �2000�. A computer model to predict the surface temperatureand time-of- wetness of concrete pavements and bridge decks, Vol.6551, National Institute of Standards and Technology �NISTIR�, U.S.Department of Commerce, 1–29.
De Vris, D., and Afgan, N. �1975�. Heat and mass transfer in the bio-sphere, Wiley, New York, 5–28.
Fahmy, M., and Moselhi, O. �2009�. “Detecting and locating leaks inunderground water mains using thermography.” Proc., 26th Int. Symp.on Automation and Robotic in Construction, 10.
FLIR Systems AB. �2004�. ThermaCAM S60 operator’s manual, Dan-deryd, Sweden, 2–40.
Horton, R., and Chung, S. �1991�. “Soil heat flow.” Modeling plant andsoil systems: Agronomy monograph no. 31, American Society ofAgronomy-Crop Science Society of America-Soil Science Society ofAmerica, Madison, 397–438.
Hunaidi, O., Chu, W., Wang, A., and Guan, W. �2000�. “Detecting leaksin plastic pipes.” J. AWWA, 92�2�, 82–94.
Hunaidi, O., Wang, A., Bracken, M., Gambino, T., and Fricke, C. �2004�.“Acoustic methods for locating leaks in municipal water pipe net-works.” Proc., Int. Water Demand Management Conf., Dead Sea, Jor-dan, 1–14.
Hutcheon, N., and Handegord, G. �1983�. Building science for a coldclimate, Wiley, Toronto.
Liu, B., Liu, W., and Peng, S. �2005�. “Study of heat and moisture trans-fer in soil with a dry surface layer.” Int. J. Heat Mass Transfer, 48,4579–4589.
Loomans, M., Oversloot, H., De Bondt, A., Jansen, R., and Van Rij, H.�2003�. “Design tool for the thermal energy potential of asphalt pave-ments.” Proc., 8th Int. IBPSA Conf., 745–752.
Mcullough, B., and Rasmussen, R. �1999�. Fast track paving: Concretetemperature control and traffic opening criteria for bonded concreteoverlays, Federal Highway Administration, Mclean, Va.
Moher, A., and Osborne, L. �2006�. Analysis of pavement radiation load-ing through pavement temperature, Regional Weather InformationCenter, University of North Dakota, N.D., 1–5.
Schlangen, E. �2000�. Heat of FEMMASSE manual, INTRON BV, TheNetherlands.
Sung, K., Yavuz, C., and Drew, M. �2002�. “Heat and mass transfer in thevadose zone.” J. Contam. Hydrol., 57, 99–127.
Walton, G. �1985�. “Thermal analysis research program—Referencemanual.” NBSIR 83-655, Department of Commerce.
Weil, G. �1998�. “Infrared thermography—Eyes below the surface.”Trenchless Technology Magazine, October 1998, 48–49.
Zarr, R. �1998�. “Analysis study of residential buildings with reflective
