Defect Characterization from Lock-in Vibrothermography Data

A method to characterize vertical cracks which combines lock-in vibrothermography experiments and a stabilized inversion algorithm is presented. The capability of the method to retrieve the shape and location of square cracks buried at increasing depths is analyzed by inverting synthetic data with added white noise. Moreover, the ability of the method to resolve two square buried cracks as a function of depth is studied. A relationship between the depth of the cracks and the minimum distance between them in order to obtain separate reconstructions is found. The predictions of the inversions with synthetic data have been checked by inverting experimental data obtained from steel samples containing calibrated cracks. In order to prevent the sample from being damaged and to reach the steady state faster, low ultrasound excitation powers are used. The signal-to-noise ratio corresponding to the resulting weak signal is improved by processing a large number of infrared images using the lock-in method. The experimental results confirm that the method is able to reconstruct square heat sources whose depth is twice their lateral dimension, and that the resolution condition holds.