Quantitative characterisation of near and subsurface defects using active thermographic techniques

Abstract

A mobile Line Scanning Thermography (LST) setup combined with a novel post-processing method that can autonomously characterise different defect types across different materials has been developed and validated. This newly developed system facilitates a more complete inspection of large structures than currently available by removing subjectivity from data analysis, minimising operator input and reducing cost. Inspection of large structures for defects is increasingly challenging due to the demand for complete evaluation. While a more complete evaluation is possible although likely time consuming, many non-destructive evaluation techniques and their data analysis require highly trained operators, making large scale inspecting costly. Automation is critical to improve inspection speed and reliability while reducing human error. Active thermography relies on a difference in thermal properties between defective and non-defective areas to locate defects by monitoring the surface temperature of the sample with an infrared camera after thermal excitation. Particularly LST is promising as it offers rapid, non-contact inspection, however, typically uses lasers, leading to bulky equipment and increased safety measures. An elliptical reflector was designed using optical simulation to provide a compromise between maximum intensity, line width and acceptable variations for out of focus measurements. A thermal model, validated with experimental data, investigated the effect of different scanning parameters on defect visibility and can be used to tailor inspection parameters for the experimental setup. A polished aluminium reflector and halogen light were used which greatly reduced safety requirements compared to traditional laser based systems and demonstrating applicability towards real world inspection scenarios. Using the developed LST system and a proposed novel processing method, flat bottom holes in acrylic were autonomously detected and characterised with a minimum detected diameter to depth ratio of 3.2. The processing estimated depths with an accuracy of 0.1 mm for near and subsurface defects and lateral sizing had a median error of less than 4%. The Fourier-based filtering was introduced as an additional term to the post-processing method to allow accurate defect identification in composite samples, such as carbon fibre reinforced polymers. Artificial delamination defects were successfully identified at a maximum depth of 1.97 mm with diameter to depth ratios ranging from 10 to 3.8. Limitations were observed for deeper defects and in areas with surface non-uniformities, highlighting opportunities for further work to fully understand the detection limits of LST. When combined with the post-processing method the developed LST system allowed for a more complete inspection of large areas by reducing subjectivity and operator time thereby lowering costs. This work extends the applicability of LST with an elliptical reflector to a variety of materials and defect types, while the rule based post-processing enables reliable defect characterisation across multiple scenarios without extensive training datasets, advancing the technique for a wide range of industrial applications.