Influence of the Computational Domain and Mesh on the Signal in Eddy Current Testing

Introduction. This paper investigates the influence of numerical simulation parameters for eddy current testing on the characteristics of the output signal during the simulation of a defect’s movement relative to the measuring coil. The objective of this work is to determine optimal parameters for the computational domain, the local finite-element mesh, and the convergence criterion that ensure a smooth output signal in eddy current testing problems when a defect moves relative to the coil. The relevance of this research stems from the widespread use of the finite element method for analyzing electromagnetic processes in nondestructive testing and the need to ensure high reliability of numerical simulation results. Materials and Methods. The study was conducted using numerical modeling of electromagnetic processes in eddy current testing with the application of the finite element method. The influence of the computational domain dimensions and the parameters of the finite-element mesh on the magnetic flux through the coil and the smoothness of the output signal was investigated as the defect position varied relative to the coil. The dimensions of the computational domain relative to the coil diameter were varied, as was the maximum mesh element size in the coil and defect regions. The convergence of the numerical solution was evaluated using the energy error and energy change criteria. To quantitatively assess signal smoothness, a criterion based on the second difference of discrete magnetic flux values was used. Results. Our research has shown that one of the determining factors is the size of the computational domain. It has been established that the size of the region along the (Z) axis — which corresponds to the direction of field propagation from the coil into the surrounding air — has the greatest influence. An analysis of the discretization parameters showed that it is not the overall mesh density that is decisive, but rather its local refinement in regions with high field gradients. Conclusions. Rational parameters for the computational domain have been established for the model under study. Further increasing the range beyond the specified values does not significantly improve signal quality. To quantitatively control numerical noise, an additional smoothness criterion based on the second derivative of the signal is proposed, which allows for the evaluation of local oscillations in the output signal. The results obtained can be used to select optimal parameters for finite-element models of eddy current testing, taking into account signal quality and the required computational cost.