Numerical modeling is an useful help to design and optimise probes: it reduces the number of costly experiments (such as test blocks) and allows to easily and separately study the influence of various parameters.

Before using a code for optimising or designing a probe, it has to be properly validated and it has to show its capabilities of modeling a real inspection. We chose to deal with the inspection of a 6 mm vertical crack (an artificial crack) on a steam generator tube from a 900 MW French nuclear power plant. The probe is a differential pancake probe composed of 2 ferrite core coils. Its movement is helicoidal (1 mm each 360°).

Only a 3D code can simulate such a problem. EDF Research Department has developed for 10 years a 3D Finite Element Method / Boundary Integral Element Method code to solve electromagnetic problems which is called TRIFOU.

This paper illustrates the application of a TRIFOU modeling to this NDT problem. In the tube and in the ferrite core, magnetic field and eddy currents are computed using Finite Elements. The inducing field is computed only on the boundary of the Finite Element mesh using Boundary Integral Elements. The crack is supposed to be impermeable to eddy currents. This assumption is realistic because we inspect a tube with an artificial crack.

Principles of the TRIFOU modeling will be briefly presented as well as the particular modules developed for coil impedance computation.

Comparisons with experimental data (on a steam generator tube with an artificial crack) will be presented on 2 different probe movements: a rotation along the vertical axis and a translation along the crack.

The experimental results are fitting closely with numerical results.