International Symposium (NDTCE 2003) NonDestructive Testing in Civil Engineering 2003  
Start > Contributions >Posters > Thermography: 
POSSIBILITIES OF DETERMINATION AND CARACTERIZATON OF SUBSURFACE DEFECTS IN MATERIAL BY MEANS OF THERMOGRAPHY AND 3D NUMERICAL MODELIvanka Boras, Faculty of Mechanical Engineering and Naval Architecture, Zagreb, CroatiaSrecko Svaic, Faculty of Mechanical Engineering and Naval Architecture, Zagreb, Croatia AbstractThe paper presents the method for determination and characterization of subsurface defects in material by means of IR thermography and 3D nonstationary numerical model of heat conduction. The numerical model is based on the control volume numerical method with control volume net adjustable to the problem. The software developed enable the record of temperatures field in different planes and time as well as changing the duration of thermal stimulation time. Inverse method is also included and enables the estimation of defect length on the base of experimentally obtained thermograms. The method was applied on the specimen made of steel with the defects in form of cylinders containing air. The comparison of temperature fields obtained experimentally and numerically was done for the same time expired as well as the dependence of the current contrast vs. time curves related to the defects depth. The mutual interaction of defects on the surface temperature field was analyzed and their influence on the current contrast determined with the aim to recognize the possible errors in determination of defect geometry and position. 1. INTRODUCTIONThe research presented includes experiments and numerical simulation, with the goal to contribute in solving the problems of visualization of the object subsurface structure. The temperature field, which appears on the surface of an object stimulated with heat flux is the result of heat conduction through its structure. It represents a 3D distribution of temperature parameters and mutual interaction of the object and its surrounding. According to that if the geometry of the object is known, as well as its interaction with the surrounding it is possible to investigate its invisible structure. In the paper, beside the analysis of the nonstationary temperature fields which appears on the surface of the specimen including subsurface defects as a result of heat stimulation, the mutual interaction of defects itself and influence on the temperature contrast was also observed. Applying the inverse method in numerical simulation the length of the defect was estimated and influence of the neighboring defects discussed. 2. DESCRIPTION OF THE SPECIMENThe specimen used in testing consists of two parts (see fig. 1) both made of the same quality steel. The defects are located in the base plate, which was covered by the cover plate. The dimensions of the plates are as follows:
All six defects are in the form of cylinders having diameter of 18 mm, but located on different depth (see fig. 2).
The thermodynamic properties of the specimen material and air are given in table 1.
3. MATHEMATICAL DESCRIPTION OF THE MODEL, START AND BOUNDARY CONDITIONSIn numerical analysis the control volume method was applied. By means of implicit discretization of 3D nonstationary heat conduction equation a certain number of algebraic equations was obtained for each control volume and they are solved iteratively. In this equations the thermal properties were assumed to be independent on temperature.
The control volume net was made much closer in the area where greater temperature gradients have been expected. The total number of control volumes was 38x30x12 having base dimensions between 1 and 12 mm. The time step was 0,04 sec. The heat transfer through defects was assumed as conduction and the heat conduction coefficient for the boundary control volumes (steelair) calculated according to equation 3.
Interaction between specimen and surrounding was defined through boundary conditions shown on fig.4. In numerical analysis the heat transfer coefficient (free convection) was calculated for each control volume placed on monitored and stimulated surface from previous calculating step and used for calculation of new values of current temperatures with iteratively made corrections. The analysis of numerically obtained temperature field on specimen surface was done by means of current contrast as a time dependent value using equation 4.
Where: T_{nd}  is a surface temperature over homogeneous material and T_{d}  is a surface temperature over defect For developed model and particular boundary conditions the numerical analysis was first carried out for the case of only one defect in the specimen. The reason was to eliminate the influence of neighboring defects. By varying the defect depth (distance from the observed surface) the relation between time of maximal current contrast and defect depth was determined (see. fig. 5).
It can be shown that the time of maximal current contrast depends only on the defect depth. The defect length can be found by means of inverse method and experimentally obtained results, thermograms recorded for identical start, boundary conditions and expired time. The inverse method is based on iteration, which start with chosen defect length. In each step the chosen value is corrected till the satisfied accuracy. The principle of the inverse method is shown on fig. 6. and iteration on fig. 7.
It must be denoted that when six defects are located in specimen its mutual influence on maximum current contrast must be taken into consideration. If not, the results obtained by inverse method are less accurate. It is very important to choose the right points on the surface for recording the data, which will be used for calculation of current contrast. The influence of one and six defects on the surface temperature distribution obtained by simulation can be seen on fig. 8 and 9.
4. EXPERIMENTThe experiment was performed on the rig shown on fig. 10. The heat source of 500 W was used for heat stimulation of the specimen. The temperature distribution on the observed surface was recorded by means of IR camera AGA 680 Standard and referent temperature measured by thermocouples. Measurement uncertainty for temperature was 0,5 ^{o}C, emission coefficients was determined as e =0,31 for heated surface and e =0,7 for monitored surface.
5. Comparison of the results obtained by simulation and experimentOn fig. 11. the thermogram obtained by experiment and temperature distribution obtained by simulation can be seen. They are recorded in the same expired time and for the same start and boundary conditions. The differences between the temperatures obtained by simulation and measurement are within 7%.
6. CONCLUSIONCombination of thermographic measurement and numerical simulation gives possibility for development of new TNDT methods which will help in solving many problems relating to nonhomogeneities in the material structure without its destruction. Using a new IR camera and power computer the accuracy of methods can be essentially improved. LITERATURE

