PSI - Issue 77

João Queirós et al. / Procedia Structural Integrity 77 (2026) 475–483

479

The experimental setup shown in Figure 3 was used to measure the plate's strain field. A uniform thermal pulse was applied using the same two halogen lamps placed symmetrically to the plate surface, delivering 500 W. The strain field was then monitored dur ing the plate’s cooling stage by measuring the phase of the interference pattern at 8 second intervals over 96 seconds. This involved the application of the temporal phase modulation technique, which requires the acquisition of four intensity images with a phase step of π/2 of the laser wavelength (Kreis, 2000). This is achieved by the translation of one of the Michelson interferometer mirrors using integrated piezoelectric transducers, actuated by a piezoelectric controller. The second mirror of this interferometer was slightly tilted to create a shearing amount of 5 mm in the x-direction (Yang, 2016). The set of acquired images is stored in the computer and later used to evaluate the phase map corresponding to the strain field at an 8-second interval.

Fig. 3. Experimental setup for the measurements of the plate deformation gradient.

4. Enhancement of Damage Identification Raw thermographic images and DS raw phase maps often contain subtle anomalies that indicate damage. In thermography, these small thermal variations can be difficult to distinguish from the background temperature, particularly for smaller or deeper flaws. Similarly, phase maps contain minute phase variations that are obscured by the global deformation phase. Therefore, post-processing of raw data is crucial for isolating these faint signals and improving the identification of internal structural damage. The following sections detail key post-processing methodologies for thermography and shearography data. 4.1 Post-processing of Thermograms Post-processing thermographic images is a crucial step in extracting meaningful information, especially when looking for subtle temperature changes caused by subsurface damage. This process significantly improves the signal to-noise ratio (SNR), enhances damage detectability, and enables more quantitative analysis. In PT, raw thermograms typically exhibit a logarithmic decay of temperature over time. However, this decay can suffer from a poor SNR, particularly towards the end of the recording. Additionally, artifacts from reflections and air convection can obscure the temperature changes induced by damage. To enhance thermograms for damage identification, Thermographic Signal Reconstruction (TSR) is a commonly applied technique. TSR works by fitting a 4th or 5th order polynomial to the thermal signal at each individual pixel (Shepard et al., 2003). This mathematical modeling effectively mitigates noise and reduces artifacts, significantly improving the SNR and allowing for clearer identification of damage-induced temperature variations. However, TSR is not appropriate or beneficial for Lock-in Thermography (LT). This is because LT deals with a fundamentally periodic and demodulated signal, which differs significantly from the transient thermal decay that TSR is designed to process. The two main algorithms for advanced thermogram post-processing are the Fast Fourier Transform (FFT) (Maldague and Marinetti, 1996) and Principal Component Thermography (PCT) (Rajic, 2002; Ebrahimi et al., 2021).