Quantitative thermoelastic stress analysis by means of low-cost setups

Abstract

A low-cost Thermoelastic Stress Analysis (TSA) experimental setup is proposed which uses an ordinary micro-bolometer and in-house developed signal processing scripts. The setup is evaluated by analysing the thermoelastic signal from a tensile and a SENT specimen made of stainless steel AISI 304L, and the bolometer performances are compared with those of a state of the art photon detector. Signal processing is based on off-line cross-correlation, using a self-reference signal which is retrieved from the acquired thermal data. Procedures are in particular developed to recognise, quantify and correct errors due to spectral leakage and loss of streamed frames. The thermoelastic signal amplitude/phase, the thermoelastic constant and the Mode I Stress Intensity Factor (SIF) from the bolometer and photonic cameras are evaluated considering the influence of loading frequency, sampling frequency, detector array sub-windowing and acquisition interval duration. A camera-specific linear calibration procedure is applied to correct the thermoelastic signal obtained with the bolometer. The procedure is extended to correct also SIF values, finding a good match with the SIFs obtained by the photon detector. An automatic iterative algorithm, based on the least square fitting of Williams’ series functions, is proposed to identify the crack tip position. An estimation of processing times of the developed signal processing scripts has been carried out, finding that a full crack characterisation (TSA maps, crack tip position, SIF) can be performed with a data acquisition time of 10-20 s, a post-processing time of less than 2 s and an overall hardware cost under 10 k€.