Theoretical Models of Eddy Current Interaction with Defects.

Available from UMI in association with The British Library. Quantitative nondestructive evaluation (NDE) using eddy current techniques should be based on a good physical understanding and good physical models. Once these models have been developed, they can be applied to the two broad classes of tasks in NDE: the forward problem, predicting the probe response from a known physical situation, and the inverse problem, predicting some of the physical parameters from a measured probe response. The theory and the numerical methods used to find approximate solutions to these problems are presented in this thesis for a broad class of geometries, probes and defects. The structure of typical workpieces divides naturally into two groups, planar and cylindrical. The planar media types are presented first. One of the contributions of this thesis is the general description of the field interactions for planar media with any number of layers, with the source any layer. This general theory developed for planar types is then extended in a consistent way to include cylindrical structures. Once in place, this theory was used to develop a layered media forward model which predicts probe responses to a large class of stratified conductors. Validation exercises are then presented, along with a brief applications section. The theory developed for unblemished media is extended to model probe/flaw interactions in planar and cylindrical stratified conductors based on a simple flaw model. The volume integral method is then used to find approximate solutions to the three-dimensional forward problem. The implementation of this general purpose forward model, capable of predicting probe responses to a wide variety of defects: intergranular attack (IGA), surface breaking cracks, embedded cracks, inclusions, etc., represents one of the major contributions of this work. The model has been validated by comparing model predictions with analytical results, experimental results and international benchmarks. The power of the models is then presented in a brief applications section. A physical model of IGA corrosion is then presented, along with a novel approach for solving the inverse problem, that is, predicting the depth of corrosion from a measured probe response. The analytical and numerical analysis used in this inverse model is presented. Model validation work based on both experimental and numerical data is presented and indicates that this approach is both robust and accurate.