Use of In-Service Inspection Data in the Performance Measurement of Non-Destructive Inspections (Mise en oeuvre de donn es resultant de visites d'inspection en service pour l'evaluation des performances des visites d'inspection non destructives)

摘要

Most available non-destructive inspection (NDI) reliability data results from analyses under laboratory or simulated in-service conditions. Adding analyses of in-service NDI findings could improve NDI. There is no organized process whereby in service data are collated for NDI reliability studies. The extent of this data and its usefulness to the NDI reliability program was analysed and processes for collection were defined as well as analytical methods to calculate NDI reliability from in-service data. Findings: One major contribution is a detailed summary of the close relationship between NDI, fracture mechanics and airworthiness including an important review of the statistical basis for many of current approaches to inspection. NDI maintenance records were reviewed for their usability to pooling of data from different sources in order to obtain statistically significant numbers. It was concluded that such records vary considerably in quality and fidelity and specific recommendations for improvement were made. Of the three approaches for using in- service inspection data to characterize the capability of an inspection system two aiming at characterizing NDI capability in terms of the probability of detection were rejected (POD) because in maintenance too many cracks of a detectable size are not detected rendering the basis data unreliable. A third approach summing inspection results in terms of the cumulative distribution function (CDF) of the sizes of the detected cracks does provide information about the capability of the NDI system in the in-service environment. The CDF does not directly yield the reliably detectable crack size (at a given confidence level) but it gives a first estimate of this size.