A Review of Probability of Detection Evaluation for Acoustic Emission and Ultrasonic Testing in Composite Structures: Context-Dependent Reliability for Inspection Planning

Abstract

The probability of detection (POD) is widely adopted as a central reliability metric in non-destructive evaluation (NDE). However, existing review literature often treats POD as a comparable performance indicator across inspection modalities without sufficiently accounting for the contextual variability. This review advances a reliability-oriented synthesis by integrating detection physics, statistical modeling assumptions, and contextual drivers into a unified framework for interpreting POD in Acoustic Emission Testing (AET) and Ultrasonic Testing (UT). A fundamental distinction is established between the event-driven “Monitoring POD” characteristic of passive AET and the stimulus–response “Inspection POD” associated with active UT. These modalities differ in both sensing mechanisms and statistical formulations, such as hit–miss and â versus a approaches. They also exhibit varying sensitivities to contextual factors, including loading regimes, defect morphology, and environmental conditions. From this perspective, context is defined as a combined set of physical, operational, and human variables that condition the detection probability. Rather than ranking techniques by nominal a₉₀/₉₅ values, this review proposes a Context-Dependent POD (CDPOD) framework that clarifies how similar numerical metrics may arise from fundamentally different detection environments. A comparative synthesis of representative studies demonstrates that apparent performance differences frequently reflect contextual misalignment rather than intrinsic technique superiority, thereby limiting direct cross-modality comparison and transferability of laboratory-derived POD values to field applications. By reframing the POD as a context-sensitive reliability construct rather than a standalone scalar metric, this review provides a structured basis for inspection planning, uncertainty interpretation, and engineering decision-making in composite systems.