Copied from http://members.aol.com/gmagnetics/changes.htm:
Changes from 2.1 to 2.1a - September 11, 1999
* This is mostly a bug fix version. No obvious functionality has been added to the program. However, quite a lot of work has gone on ``under the hood'' to fix some bugs that cropped up and to give more accurate answers.
* Several integrals had errors when applied to axisymmetric problems. These errors have now been fixed. The specific integrals modified are:
- Total losses--forgot to multiply by 2*Pi*r
- Total current--modified to include induced currents in the integration. Previously, only source currents were included.
* The formulation used to solve axisymmetric problems has been changed. The old formulation gave good performance almost all of the time, but occasionally, a problem would come along in which the scheme would exhibit poor convergence. The formulation has been changed to one similar to the axisymmetric formulation suggested by Henrotte et al. in their paper ``A new method for axisymmetric linear and nonlinear problems,'' IEEE Trans. Mag. 29(3):1332-1335, March 1993. This formulation has the same well-behaved characteristics as the old formulation in the region close to the r=0 line, but it also successfully deals with the problems in which the old formulation gave spurious results. The new formulation has been applied to both static and harmonic problems in femm. The wrong eddy currents were reported in response to inquiries about solution properties at a particular point in the axisymmetric case. Some typographical errors in the manual were fixed.
This new release has been tested on a few problems, and it was found, that the postprocessing tool femmview does not display the results properly any more. The eddy current density seems not to be proportional to the radius (for small distances of a few millimetres) any more, but to be almost constant with a non-vanishing value for r =0. However, the the old version (2.1) of femmview can be used with the new release of femme andfkern. The preprocessing and FE programs femme and fkern seem to work properly. For the solid iron sphere and the small copper sphere identical results have been obtained with the old and new versions.
The remark in the above given notes for the new release on the behaviour of the solution in the region close to r =0 seems to be noteworthy. Probably this is a known problem in the solution of axisymmetric electromagnetic problems, because the solution ''far away'' from the axis (as is the case for the 5 cm iron sphere) is in good agreement with the analytical solution. Close to the symmetry axis both, the FE solution of the iron sphere and that of the copper spheres shows rather large deviations for the analytical solution.
Next: Conclusions
Up: Calculation of eddy currents
Previous: Eddy currents in a solid sphere
Contents