Degree Type


Date of Award


Degree Name

Doctor of Philosophy


Electrical and Computer Engineering

First Advisor

William Lord


A major advantage of electromagnetic nondestructive evaluation (NDE) methods, particularly those associated with eddy current or magnetic flux leakage inspection, is the use of non-contacting probes, which allow rapid moving inspection. However, the output signals generated by moving probes are affected by the resulting motionally induced currents. To model actual testing situations, it is therefore necessary to include such probe velocity effects in the energy/defect interaction model;Motional induction causes a loss of self-adjointness in the governing equations. In such cases with high magnetic Reynolds numbers, the standard finite element and finite difference results show spurious oscillations. Although such oscillations can be removed by severe mesh refinement, this increases the burden on the computer resource and may limit the practical use of the numerical methods. To overcome this difficulty, an upwinding technique, originally developed in fluid mechanics, is applied to the uniform geometry cases associated with variable reluctance, eddy current, and remote field eddy current probes. However, most NDE methods are concerned with the detection of abnormalities in the testing specimen. In such non-uniform geometries, the distribution of motional induction currents changes with time in all exposed conducting surfaces. Consequently, a transient analysis is required. Upwinding techniques developed from steady state equations show some numerical dissipation, and time stepping with upwinding has too much dissipation due to its accumulation at each time step. Therefore, a new time step method is employed that uses separate time weighting factors and an artificial reluctivity term;The fact that the probe movement in a non-uniform geometry causes a transient situation prevents the same analysis of AC steady state based eddy current methods. Therefore, the probe velocity in eddy current testing should be restricted to a very low speed. Also, new output variables for eddy current signals seem to be necessary for high speed inspection that are not defined under the AC steady state assumption. These points are illustrated by applying the Leismann and Frind's time step method, which is validated by a successful reproduction of upwinding results, to magnetic flux leakage testing with a non-uniform geometry.



Digital Repository @ Iowa State University,

Copyright Owner

Young-Kil Shin



Proquest ID


File Format


File Size

246 pages