Application of Infrared Thermography for Damage Detection in Structural
Concrete  a Preliminary Report
J. K. C. Shih, Ming Hsin
Institute of Technology
R. Delpak, C. W. Hu, P. Plassmann
University of Glamorgan, UK
Wawrzynek, M. Kogut
Technical University of Silesia, Poland
 International Symposium on NDT Contribution to
the Infrastructure Safety Systems, 1999 NOV 2226 Torres,
published by UFSM, Santa Maria, RS, Brazil 
Abstract
This paper explores the possibility of detecting geometrical defects in concrete specimens using thermal imaging. The method is based on the characteristics of heat flow phenomenon, in a conductive medium of specific geometry, which is intended to model predetermined boundary conditions. A range of crack widths, representing mechanical damage, has been induced under controlled laboratory conditions. In all cases the cracks have been located successfully. Numerical modelling results, using finite element method, are present. The current analysis is developed as a twodimensional problem. The numerical model may be used to confirm the experimental results, once material physical and mechanical parameters are available. Experimental data, based on selected shape parameters, show a satisfactory agreement with the numerical results, hence encouraging the scope of the present study to be widened.KEYWORDS: NDT damage detection, IR thermography, finite element analysis and concrete fine crack detection
Introduction
The current financial climate in the construction industry throughout Europe and the industrial countries dictates the early detection and evaluation of faults with respect to remediation and repair. Not all faults become potentially harmful to structures. It is therefore essential to identify faults and cracks of specified length and width, both as a part of repair measures and as a means of preventative and preemptive strengthening and maintenance strategy.
In recent years, the detection of defects by nondestructive testing (NDT) has gained prominence Burger and Raj Babak[1], Buyukozturk[2], Inagaki et al[3], Maldague[4], Maldague[5]. Amongst the nonmechanical methods are use of radar, ultrasonics, ultrasound and infrared thermography. In a perfect heat conducting medium, with simple boundary conditions, the lines of heat flow have geometrically predictable trajectories. Any interference with the original geometry distorts the lines of heat flow. In thermal NDT methods, the surface temperature distribution is recorded and analysed using an infrared (IR) camera. This is due to interference in the lines of heat flow through an originally perfect and isotropic/homogenous continuum. In general all geometrical anomalies or material discontinuities influence the propagation of the heat flow is detected by a suitable IR camera. In the current investigation, the geometrical parameter is only the crack width which presents a resistance to heat flow the extent of which is to be evaluated.
The aim of this work is to identify the surface temperature distribution captured by the IR camera in order to determine the position of the crack/delamination and other geometrical imperfections. In this paper, the IR camera is applied as a temperature monitoring device, to observe temperature variations sourced from a continuous and constant heat supply. Concrete specimens with specified geometrical discontinuities have been prepared. The cracked material simulated is plain concrete. The present authors have encountered published reports relating to the use of Thermography for metals and plastics Arya and Muralidhar[6], Cielo at el[7], Luthi et al[8]. There is also published research related to moisture movement in structures due to possible damp etc. Published reports on use of IR Thermography in structural concrete for crack and fault detection is more restricted.
Experimental Setup
Concrete test pieces, 80x400x100mm with known maximum crack widths of 0.5 and 1 mm, are mounted vertically and heated by a uniform output heating element. The concrete mix complies to British Standard BS8110 specification and is designed to an approximate cube strength of 40 MPa. This is a generic mix which is meant to represent many of the concrete structures in the UK and EU, hence adding realism to the experiments. The concrete mix can be procured from most contractors for industrial use. The surface temperature distribution of the specimen is registered by the IR camera which has a resolution of 240 x 360 pixels, and a sensitivity of 0.15 °C. The IR camera is capable of capturing consecutive images using a Frame Grabber through an image acquisition expansion card for PC. Appropriate software is used to control the grabber, and many frames of recorded images may be displayed simultaneously in pseudocolours. A schematic diagram of the experimental setup is shown in Fig. 1.
Fig 1: The Schematic of the Experimental Setup

Experimental Results
The experimental results are analysed using a bespoke software package. Surface temperature distribution is presented in the form of the images. Results for cracks with a width 1 mm are shown in Fig. 2 (images 18).
Fig 2: A Set of Infrared Images Which Show the Extent of Heat Flow Blocking due to Surface Discontinuity Caused by A Mechanical Crack. Image 7 is at SteadyState Condition after 394 Minutes of Test Starting

The heat 'front' is allowed to propagate at various time intervals and its progress recorded as isotherms by the FrameGrabber software. The temperature field variation graph indicates the sharp change in temperature between the leading and the lagging cracktip, due to interference with the natural heat conduction of the continuum. Images 16 are taken at given intervals in Table 1 for a total time lapse of 354minutes. This constitutes a transient heat flow condition. Image 7 and 8 are at steadystate after 394 minutes. Table 1 includes the complete time and image recording details for this experiment at which the images were taken.
Image Number
 1
 2
 3
 4
 5
 6
 7
 8

Lapse time ts in minutes(t start = t 0 = 0 )
 0
 26
 90
 210
 266
 354
 394
 ¥

Transient (Tn)Steadystate (Ss)
 Tn
 Tn
 Tn
 Tn
 Tn
 Tn
 Ss
 Ss

Table 1: Image and Time Lapse Specified 
Numerical Modelling
Fig 3: Typical FE Crack Geometry Mod

Fig 4: Temperature Distribution for a T dimensional Model

A commercially available finite element (FE) software is used to model the geometry of the test specimens as well as determine the temperature variation field in the material due to a specified heat source (ANSYS, Release 5.4, 1997). There are standard text books discussing the use of FE method for heat transfer, for example see Bonet and Henwood[9], Zienkiewicz[10].
All present analytical models are formulated in twodimensions, see Fig. 3. The 2D model which includes the crack, is generated automatically using 2480 triangular FE's from which the temperature distributions are obtained. Material parameters such as the thermal conductivity k, specific heat C, and the heat transfer coefficient a are used based on the referenced literature such as Carslaw[11], Chapman[12], DeWitt and Incropera[13], Jaluria and Torrance[14], and Minkowycz et al [15]. The density of the specimen is obtained accurately from the standard laboratory measurements.
The characteristic of the temperature distribution is identical to the images gained from the experiments. Figure 4 is an example representing the shape of the predicted temperature field.
As the heat is continuously delivered by the source, the temperature distribution along the concrete surface, at steadystate conditions, can be plotted by a single curve. A group of such curves are shown in Fig. 5. This shows surface temperature distributions due to a concentrated heat source (generating heat) as multiples of the ambient temperature (T). Similar results are shown in Fig. 6, where each curve represents the surface temperature variation due to input of heat, in this instance assumed to be uniformly distributed as a plateau, also as multiples of the ambient temperature (T).
Fig 5: Surface Temperature Distribution due to Different Temperatures of an Applied Point Heat Source (curves relate to various source temperatures, from T to 5T), S/d = 0.5

Fig 6: Surface Temperature Distribution due to Different Temperatures of the Distributed Heat Source Applied (S is the distance of the leading edge of heat source to crack centre, refer to Fig. 1), S/d = 0.5

A series of calculations are carried out with both heat source models in order to study the effect of varying geometric parameters of the crack on heat flow characteristics. As an example, the temperature distributions along the specimen for different widths of the cracks are presented in Fig. 7, where the source temperature is 3T due to a plateautype heat source.
Fig 7: Surface Temperature Variation for Different Crack Width
(S = leading edge heat source to crack centre, refer to Fig1) 
It is surmised that if there were no cracks present, the temperature distribution on the concrete surface would be expected to be symmetrical and continuous. In the section of the surface with at least 1 crack, there would be a sharp local change in the temperature distribution. The temperature value is seen to reduce with the increase of distance from the heat source.
To summarise, the theoretical study indicates that:
 As expected, the temperature distribution on the surface is symmetrical relative to the source centreline, in the absence of surface discontinuities (assuming that the edge effects are negligible).
 Even very small cracks have significant effects which interfere with the heat flow.
 The difference of crack geometry would have a varying impact on the temperature difference between the leading and trailing edges across the crack width.
 Geometrical parameters of the crack influence surface temperatures only in close vicinity of damage.
 It can be seen in Fig 6 that on the left side of heating element temperature variation has a fixed variation irrespective of different crack widths.
 Increasing the source temperature (within limits) exaggerates the temperature difference across the crack width, for the same damage geometry.
Comparison
Both experimental and numerical results display clear trends in identifying the position of the crack. Temperature drops across the crack width are easily identifiable in both the thermographs and FE plots. Surface temperature distributions of the concrete specimens vary with different types of configurations and dimensions of the defects. Qualitatively, the obtained results show remarkable consistency of variation. However, the experimental data differ from the numerical values within acceptable limits. These differences are thought to be caused by unconfirmed values of the thermal properties (such as those cited in Section 4) used for FE analysis, and the simplification of the heat transfer phenomena in numerical modelling. For the followon research, the additional experiments in evaluating the uncertain thermal properties will be implemented by making appropriate adjustments to the numerical model. Comparison between few experimental data and the remainder of the theoretical results, is shown in Fig. 8.
Fig 8: Theoretical and Observed Influence of Crack Widths on Percentage Difference of Cracktip Temperatures over than Ambient Value, Refer to Specimen Geometry in Fig 1.

Scope for future research
The future research methodology in this field is expected to use numerical solution techniques of the inverse problems more extensively. This will lead to determine the required geometrical parameters quantitatively, see Wawrzynek et al [16].
Conclusion
Numerical results obtained using a commercial FE program show that the geometrical imperfections can be clearly identified. Temperature distribution on the surface of the element, for a given source heat input, depends on many factors such as the material mechanical and thermal properties, the geometry of the considered body, as well as the manner of applying the heat source and the time after which the image is taken. Although the location of a typical crack is reasonably simple to determine, the geometrical parameters of the damage are difficult to quantify without knowing the thermal properties of the specimen under investigation. In order to determine the geometrical parameters in engineering practice, some preliminary tests should be carried out to calibrate the system parameters before applying IR monitoring. The experimental study demonstrates that in using IR thermography, an unmistakable image of crack/damage, through heat conduction blocking mechanism emerges. It is seen that using the limited set of experimental data so far, the matches with the analytical results are remarkably close. In this study, it is concluded that identical trend variations exist for both physical and analytical data with satisfactory agreement between both approaches.
Acknowledgment
The authors would wish to express their appreciation to Directorate of Queening HiTech Corporation in providing the research grant and equipment for the present report.
They are also indebted to the management of British Trade and Cultural Office in Taiwan who has been generous with mobility and the associated expenses directly linked with this project. Finally they acknowledge the following UK companies Rockwool UK (Bridgend) and Microtherm Int. Ltd. (Upton) for generous provision of specialist insulation materials without which this project would not have progressed.
References
 Burger C. and Raj Babak, Nondestructive Evaluation Through Transient Thermographic Imaging (TTI)., 15^{th} Symp. On NDT San Antonio, 4, 5667, 1985
 Buyukozturk O., Imaging of Concrete Structures. , NDT&E International, 31, 233243, 1999
 Inagaki T., Ishii T., Iwamoto T., On the NDT and E for the Diagnosis of Defects Using Infrared Thermography, NDT&E International, 32, 247257, 1999
 Maldague X.P.V., Infrared Methodology and Technology, Gordon and Breach Science Publishers, 7, 3319,199
 Maldauge X.P.V., Nondestructive Evaluation of Material By Infrared Thermography, SpringerVerlag London Ltd., 12110,1993
 Arya N.K., Muralidhar C., Evaluation of Defects in Axisymmetric Composite Structures by Thermography, NDT&E International, 26, 189193, 1993
 Cielo P., Maldague, Deom A.A., Lewak, Thermographic Nondestructive Evaluation of Industrial Materials and Structures , The American Society for Nondestructive Testing, Inc, 453365 , 1987
 Luthi T. Meier H. Primas R., Zogmal O., Infrared Inspection of External Bonded CFPRSheets, International Symposium NonDestructive Testing in Civil Engineering (NDTCE), 26, 689704,1995
 Bonet J., Henwood D., Finite Elements. Gentle introduction., Macmillan Press Ltd., London, 1996
 Zienkiewicz O.C., The Finite Element Method, 3^{rd} Edition, McGrawHill Book Company (UK) Ltd., Berkshire, 423449,1977
 Carslaw H. S., Jaeger J. C., Conduction of Heat in Solids, Oxford University Press, Oxford, 1959
 Chapman A.J., Heat Transfer, Macmillan, New York, 1984
 DeWitt D.P., Incropera F.P., Introduction to Heat Transfer, John Wiley &
Sons, New York, 1985
 Jaluria Y., Torrance K. E., Computational Heat Transfer, Springer Verlag, Berlin, 1986
 Minkowycz W.J., Sparrow E.M., Schneider G.E., Pletcher R.H.,Handbook of Numerical Heat Transfer, John Wiley & Sons, NY, 1988
 Wawrzynek A., Delpak R., Kogut M., Use of Inverse Geometry Thermal Solution Technique for damage Identification, XI International Scientific Conference, Brno, 1999
/DB:Article /SO:NDTISS /AU:Shih_J /AU:Hsin_M /AU:Delpak_R /AU:Hu_C /AU:et al /CN:UK /CT:IRT /CT:concrete /CT:civil /ED:200002