Model-assisted probability of detection validation for immersion ultrasonic application

Abstract

Model assisted probability of detection (MAPOD) is a method of calculating probability of detection versus flaws size curves which utilize physics-based models to reduce the time, cost, and complexity of POD studies with respect to those conducted by purely empirical approaches. An approach to utilizing models combined with empirical data in calculating POD curves based on immersion ultrasonics as the example case is described. Before MAPOD calculations can be performed, the applicable physics-based model must be validated to accuracy consistent with the desired accuracy of the resulting POD curve. In this case, the Thompson-Gray ultrasonic measurement model was used. For the purposes of validating the MAPOD approach, a specimen was configured that would allow empirical data to be collected that would serve the dual purpose of providing input to the MAPOD calculation as well as providing input to empirical POD calculations (MIL-HDBK-1823). Validation of the MAPOD approach consisted of directly comparing the results of conventional empirical calculation of POD curves with MAPOD-generated POD curves for the same inspection situation. The results of the two POD calculation approaches agreed very well. Based on the successful validation activity, the MAPOD approach has been implemented by Pratt & Whitney for many internal programs. © 2007 American Institute of Physics.