Session: 26-02 Applications using Probabilistic and Machine Learning Methods

Paper Number: 129117

129117 - Integrating Location-Specific Non-Destructive Inspection Simulation With Probabilistic Damage Tolerance Assessment 

Probabilistic damage tolerance (PDT) analysis has become a standard practice in the aircraft gas turbine engine industry.  The US Federal Aviation Administration (FAA) has issued a number of Advisory Circulars that prescribe the use of probabilistic approaches for certification assessment of critical rotating components and the potential risk reduction associated with non-destructive evaluation (NDE) involving probability of detection (POD) curves.

The conventional approach to POD curve development relies on testing.  This approach is expensive, and the generalized curves that are developed may not account for variations in component geometry and local features that may influence POD outcomes. NDE simulation is an emerging technology that has the potential to significantly reduce the effort that is required for POD curve development.  Unlike the conventional approach, NDE simulation includes the capability to consider variability in all of the parameters that may influence the accuracy of simulated POD curves. One of the challenges to the NDE simulation approach is the anomaly size and orientation information that is needed for POD curve development.  PDT can identify anomaly sizes and orientations at  specific locations in a critical component that are most likely to fail (and would benefit most from NDE inspection). Application of this location-specific information to NDE simulation may substantially reduce the computation time to create POD curves, and may significantly reduce the number of POD curves that must be created to satisfy PDT assessment requirements.

In this paper, a methodology is presented for integrating NDE simulation with PDT analysis of turbine engine components.  An approach is presented for identifying risk-critical locations in FE component geometries using PDT analysis (via the DARWIN^® software) and the associated anomaly sizes and orientations that may lead to failure. This information is used to guide the development of location-specific ultrasonic inspection POD curves using NDE simulation (via the CIVA software).  The resulting POD curves are applied to PDT assessments that enable the analyst to identify the locations that may benefit most from an NDE inspection.  The methodology is illustrated for a representative gas turbine engine component.  The results are used to quantify the influence of location-specific NDE on component fracture risk in comparison with conventional NDE.

Presenting Author: Michael Enright Southwest Research Institute

Presenting Author Biography: Dr. Michael Enright specializes in reliability-based life prediction with an emphasis on probabilistic fatigue and fracture. He has published over 100 peer-reviewed journal articles and conference papers (including four award winning articles) and organized a number of international conferences focused on reliability-based life prediction of gas turbine engine materials. Dr. Enright is currently responsible for development of the DARWIN probabilistic fracture mechanics software developed by Southwest Research Institute. He also serves as Adjunct Professor at the University of Texas and Trinity University in San Antonio.

Authors:

Michael Enright Southwest Research Institute
Yasin Zaman Southwest Research Institute
Fabrice Foucher Extende
Sebastien Lonné Extende
Philippe Dubois Extende
Stephane Leberre CEA DISC
Pierre Calmon CEA DISC