Original by the forlink team, please indicate the reprint.
Good meshing is the premise and basis for electromagnetic equipment analysis using electromagnetic field finite element software or programs, which is especially true for transient field and 3D models.
Both the static field and the eddy current field contain adaptive solution processes. The software has grid self-encryption capabilities, so in most cases, a good grid can be obtained without manual division. The transient field mesh must be manually divided, the initial mesh (initial)
mesh) The quality is very poor. Adaptive grid provides a good idea for transient fields, which is to import the adaptive grid of static magnetic field and eddy current field, and is used for transient fields. This function is analyzed by Analysis/setup1/advanced/import.
Implemented in mesh.
Anyone who has done 3D analysis knows that mesh division is the top priority of 3D analysis. Sometimes the model can be divided and calculated, but the results are much different from the design or experiment. At this time, the problem often occurs in the segmentation process, just change the segmentation. For example, if you need to do eddy current analysis, you need to consider the division based on the depth of the skin effect. This function is in mesh
operations/assign/on selection/skin depth based
refinement. For example, if you want to analyze the cogging of permanent magnet synchronous motors
Torque, then multiple layers of air gaps are required. At this time, the so-called "dumb" method needs to be used to achieve it by building several more air gap layers. For example, for the common stacked windings or wave windings in motors, in order to better divide the curved parts of the end of the motor, the surface approximate splitting is required. This function is in mesh
Set it in operations/assign/surface approximation.
There are many ways to divide the classification, mainly the difference between on selection and in selection. The former is based on surface dissection, while the latter is based on internal dissection. Note, on
selection does not mean not to dissect the internal content,
Selection does not mean not to dissect the surface, but the two dissecting methods have different focuss, as their literal meanings show. Generally speaking, on
Selection is more suitable for the dissection of solids in high-frequency analysis, because the eddy current effect is very prominent in high-frequency analysis. in
Selection is more suitable for the analysis of DC and power frequency. For most physical models, using in selection is sufficient.
Finally, let’s talk about the issue of grid length when setting the segmentation. For the same or similar physical models, they can be divided into the same group and specify the same grid length. For larger models, the mesh length can be appropriately enlarged. For example, for permanent magnet synchronous motors, the motor stator core is divided into one group, the stator windings are formed into one group, the permanent magnets are formed into one group, and the band is formed into one group. Among the above four groups, the recommended grid length is as follows: stator core>permanent magnet>stator winding>band. Select the rotor core at the same time and use in
Selection segmentation, the system will specify a default length. Generally, 1/2 of the default length is enough to be accurate. Others can be similar.
Note: This article is based on maxwell explanation, but the meshing method is not limited to maxwell. It is suitable for all electromagnetic field finite element analysis software, such as flux, jmag, magnet, opera, comsol
multiphysics, quickfield, etc. I hope I can learn from it and apply it to it.