Learning Forum

[tpaplham]

I have recently been running an eddy current model in 3D that generally converges, but to try to reduce both error and runtime, I tried to take advantage of the rotational symmetry of my model and use Maxwell 2D in RZ mode. However, rather than the quick, accurate solutions I was expecting, 2D seems to really struggle with convergence. Even after multiple passes, the error is still changing by several orders of magnitude per pass, and eventually I run into an error where "one side of the equation resulted in infinity or NaN".

I have attached a screenshot of my geometry. Other than the fact that the top piece is relatively thin (0.152 mm), the geometry is extremely simple, and given that the 3D model, given enough passes/mesh refinement per pass can handle well, I have no clue why the 2D model is having so much trouble. It's definitely the thickness of the sheet that is the source of the issue; removing the sheet entirely, simulating just the axisymmetric coil results in convergence after seconds. I added the sheet back in with a much greater thickness of 0.5 cm, which still resulted in convergence, but even reducing to 0.25 cm (more than an order of magnitude thicker than I need to be able to model) resulted in convergence issues?

Why is the 2D model having such a problem with convergence for this relatively simple model?

Screen Shot 2021-09-10 at 9.35.57 AM.png

[NKC]

Hi @tpaplham Change the initial mesh setting to classic and check.
If this didn't solve the issue please share the screenshots of the design and error message.
Ansys employees can't access attachments. Please use the image option and not the Attachment option while sharing the screenshots.
Regards Navya

[tpaplham]

Navya I tried switching the initial mesh setting to Classic but it doesn't appear to have had any noticeable effect. While the latest run has not yet run into the infinity NaN error, it is displaying the same incredibly large energy error and huge jumps to higher and lower errors between passes as before. I have added images of the design and the solution data I am seeing. For reference, I am using the RZ symmetry configuration for the 2D Maxwell simulation, and the boundary condition applied to the region edges (as visible in the more zoomed out image) is the Balloon condition.

-Tyler

[tpaplham]

Here is the final energy error report and error message when the simulation terminated after 13/20 passes.

[NKC]

Hi @tpaplham Select and Right mouse click on the thin object(plate)> Assign mesh operation> inside selection> length based and limit the max element length less than the thickness of the sheet.
If this did not work.
Please share a screenshot of excitation, boundary, and other design data, so that I can check.

Regards Navya

[tpaplham]

Still doesn't work. The boundary condition is the balloon condition along the polyline highlighted in blue in the first image. The second image shows the Current excitation being applied to the copper wire cross-section.