The reliability of nondestructive evaluation (NDE) techniques is currently determined through probability-of-detection (POD) studies that rely on measurements of actual hardware copies of representative flaws. There is now a need to eliminate the inefficiencies and expenses of this method by using model-based standards derived from sophisticated engineering computer codes that are, in turn, derived from rigorous applications of physical theories. This proposal applies the concepts of model-based and voxel-based inversion that is based upon Victor Technologies'''' well-known eddy-current code, VIC-3D(c), to solve the problem of reliably assessing NDE techniques. We will introduce the notion of `Probability of Inversion,'''' and show that the classical model-assisted POD (MAPOD) becomes a simple corollary to it. Furthermore, we will develop a rigorous electromagnetic model for anisotropic media, such as Ti-6Al-4V, which is a titanium alloy that is widely used in aerospace structures and engines. With these concepts we effectively enter the `digital age'''' of eddy-current NDE, leaving traditional emphasis on analog instruments behind.
Benefits: The technology that we develop in this proposal will be applicable to the aerospace, nuclear power, materials characterization and many other industries, so our research will have commercial benefits that extend far beyond military applications.
Keywords: Model-Assisted Probability Of Detection, Model-Based Inversion, Electromagnetic Inverse Methods, Volume-Integral Equations, Computational Electromagne