Bayesian Updating of Structural Health and Integrity Assessments using Real-Time Inspection Results

Abstract

Conventional methods for assessing the structural health and integrity of an airframe typically involve performing damage tolerance analysis of critical fracture locations (CFLs) and establishing inspection interval for those locations based on the analysis. Some structural health assessment programs are also implementing probabilistic failure analysis methods to determine the time to fracture and the likelihood of fracture of a critical component during a given time frame. Both the deterministic and probabilistic analysis methods usually start with an estimation of the initial flaw size (along with its size distribution for probabilistic methods) in order to define initial and recurring inspection intervals for the critical structural component. These inspection intervals are established to ensure the airworthiness of the airframe, and nondestructive inspection (NDI) procedures are developed for the various critical structural components on an airframe. The NDI procedures are conducted per the prescribed inspection intervals, but it is not common practice to use the inspection results to make real-time updates to the initial flaw size estimate used in the analyses for the original structural health and integrity assessment. A suitable method for combining and updating available information is embodied in the Bayesian approach. This approach provides a logical basis for updating the initial flaw size estimate and its distribution based on the inspection and detection of material defects. While various NDI methods are used to minimize the risk of fracture, these NDI methods are invariably imperfect; thereby leaving the possibility that all flaws are not detected in an inspection. The detection capability of an NDI method and its probability of detecting a flaw generally improves with the flaw size, and probability of detection (POD) curves are defined for a variety of NDI methods. The Bayesian updating process can be used with a specified NDI method and its defined POD curve to provide improved estimates of flaw size and its size distribution. With the improved flaw size estimate, analysts can provide more accurate assessments of a structure’s health and integrity based on real-time inspection results. In this case study, the Bayesian updating process is applied to a hypothetical CFL on an airframe component where a NDI method is used to inspect the location in the structure. The NDI method is the manual bolt hole eddy current (BHEC) technique with a defined POD curve. The Bayesian approach is used to update flaw growth and probability of fracture predictions for the critical location at defined inspection intervals based on the given NDI method. For comparison, a structural health monitoring method using magnetostrictive sensors (MsS) is also employed to monitor and provide detection capability of flaw nucleation and growth at the critical location. Structural health and integrity assessments using the NDI and SHM methods are compared in order to illustrate how the Bayesian updating process can provide improved accuracy to flaw growth and probability of fracture predictions. The comparison also demonstrates how either the NDI or SHM method can impact the maintenance and inspection burden for ensuring a structure’s health and integrity.