Crack detections in fatigue loaded structures by means of low-cost Thermoelastic Stress Analysis setups

The work presents the results of an experimental evaluation of the thermoelastic signal on fatigue loaded single-edge-notched and tensile samples made of stainless steel. The thermographic signal is acquired by a micro-bolometer sensor Flir A655sc. The evaluation is performed varying the loading frequency and sampling frequency. The thermoelastic signal is evaluated by a Discrete Fourier Transform algorithm accounting for the influence of spectral leakage. Two different crack tip search algorithms are applied and compared: a grid method with pixel and sub-pixel accuracy and a sub-pixel pattern search method, both based on least square fitting of the Williams’ Series stress function. The two algorithms are also evaluated in terms of accuracy, automation level and computation timing. The SIF evaluated by the Micro-Bolometer is corrected by a specific camera dependent calibration procedure, and the influence of the number of points used in the least square fitting on the final value of the SIF is investigated in order to obtain the fastest and reliable setup conditions for an automatic full fatigue crack growth characterisation with a non-contact and low cost Infrared camera system.