Learning Forum

[saeed.sedigh]

I am simulating the capacitance between two conductive plates. The plate-to-plate distance is swept from:

Hmin = 0.0011202 mm
to
Hmax = 4 mm

The geometry, material setup, boundary conditions, and distance sweep are the same in all cases. However, I tested three different automatic mesh configurations and compared the capacitance result:

Case Study	Mesh Refinement	Dynamic Surface Resolution	TAU Mesh	Hmin	Hmax	Maxwell Result (pF)	Q3D Result (pF)
Case 1	Maximum	Disabled	Disabled	0.0011202 mm	4 mm	20.16	14.96
Case 2	Maximum	Enabled	Disabled	0.0011202 mm	4 mm	12.25	10.04
Case 3	Maximum	Enabled	Enabled	0.0011202 mm	4 mm	23.18	10.00

Issue:
The calculated capacitance changes significantly depending on the automatic mesh settings, although the model setup is unchanged. For example, the Maxwell result changes from 12.25 pF to 23.18 pF, while the Q3D result changes from 10.00 pF to 14.96 pF over the same distance range.

Question:

Why do the capacitance results vary so much between these automatic mesh configurations? For capacitance extraction between two conductive plates, which software should I trust more, ANSYS Maxwell or ANSYS Q3D, and which mesh configuration should be considered the most reliable? Also, what is the recommended procedure to verify mesh convergence for this model?

https://drive.google.com/file/d/1UkA5H-PwPPw-MLbwMty-_IaUeAUvFFgo/view?usp=drive_link

 

[Miranda Hourihan]

Different solvers, different meshing: Maxwell 3D is a general EM field solver; Q3D is a quasi‑static extractor tuned specifically for RLGC extraction. They use different element types, refinement criteria, and error estimators, so the same geometry with “automatic” settings can land at very different effective mesh qualities.

For results with large variation, I would suspect it's not converged: If changing only the automatic mesh options moves C from ~12 pF to ~23 pF (Maxwell), the solution is still mesh‑dependent. That’s the definition of a non‑converged result.