Probability of Detection analysis for transient thermography in discontinuous long fibre composites using flat-bottom hole specimens

Probability of Detection (POD) analysis in composites commonly uses artificial defects and assumes spatially uniform background responses to define defect contrast relative to a sound region. This assumption does not hold for discontinuous long fibre-reinforced polymer composites (DLFRPCs), such as prepreg platelet moulded composites (PPMC), where stochastic mesostructures produce locally varying thermal responses even in defect-free material. This study introduces a pre/post differential thermographic POD methodology that addresses this limitation. Transient infrared thermography is applied to PPMC specimens using flat-bottom holes (FBHs) as controlled defect simulants. Teflon inserts, commonly used in continuous-fibre laminates, were found unsuitable for PPMC due to uncontrolled movement and distortion during moulding. To eliminate the need for a uniform sound-region reference, specimens are inspected before and after FBH machining under identical conditions. Thermographic Signal Reconstruction (TSR) is used to derive a defect-sensitive response metric from changes in temperature-derivative fields, separating defect-induced contrast from mesostructural variability. Detection reliability is quantified using both continuous-response and hit/miss POD frameworks. The proposed approach establishes a practical thermographic POD methodology for DLFRPCs, where stochastic mesostructures prevent conventional sound-region normalisation.