Defect representation using the electromagnetic tensor formulation for eddy current NDT

A difficulty for eddy current inspection is in the interpretation of data from either individual detectors or arrays. For individual detectors, it is common to monitor signal variations in the impedance plane and then detect signal variations outside a pre-defined envelope. For eddy current arrays (ECA), the data is often displayed as an image as the array is passed over the test piece. The image often has a qualitative grey / colour scale selected to emphasise anomalies caused by defects. Quantitative information such as crack sizing is often either based on calibrations of empirical responses from test pieces or relationships derived from much simplified analytic models. Quantitative eddy current inspection relies on an accurate model of the sensor system. This requires a solution to the full 3D eddy current problem describing the particular application. Unfortunately, standard techniques to solve eddy current problems, such as the finite element method (FEM) or boundary element method (BEM) or the method of auxiliary sources (MAS) are numerically intensive; and therefore, are typically used for off-line studies. This research aims to improve modelling techniques for eddy current inspection and inversion, and illustrate a different approach to modelling which is highly computationally efficient. This is achieved by representing the defect by its eddy current signature in the form of an equivalent tensor (a 3 by 3 matrix representing the X Y Z responses of the defect to each of the X Y Z components of the applied field) allowing the sensor response to be determined on-line with relatively modest computing hardware. These results will be used as a basis for the further development of an eddy current inspection system which can supply online quantitative information about defect depth and orientation using real time tensor calculations.