Practical Experiences in POD Determination for Airframe ET Inspection

Evaluation of NDT reliability has received increasing emphasis in recent times. In particular, quantifying the probability of detection (POD) attained in routine inspections have become more widespread. Although there are good guidelines and standards for POD determination, the process is still far from trivial. Various choices made during the experimental set-up may have significant effect on the results. Also, the cracked samples used are often limited necessitating various compromises in the analysis.
Patria performed a set of POD studies for eddy-current inspections performed on various parts of typical metal airframe. The project included manufacturing of cracked samples, organizing the inspection of these samples and final analysis of the results. Several inspectors from different organizations took part in the exercise. The project was done in collaboration with Finnish and international partners.
The data showed various unlikely events (small hits, big misses and poor separation), which necessitated adjustment for the standard methodology. Contrary to expectation, the false call rate did not show significant correlation with the inspection performance. When the â vs. a and hit/miss analyses could be directly compared, they showed surprisingly poor correlation and caution is advised in using â vs. a analysis for manual inspections such as the ones shown here.