I try to simulate zero-index-material in RF module, so the permittivity and permeability should be very small value, like 10-4 or even smaller. But when the permittivity and permeability are too small, the comsol always encounters a problem that "Failed to find a solution for the initial parameter. The relative error (0.62) is greater than the relative tolerance". The calculation continues when increase the relative tolerance and element size, but this way will cause inaccuracy of calculation.
So is there any way to solve this promblem that setting permittivity and permeability of the material to extremely small value (like 10-10) in a frequency range, moreover without reducing the computational accuracy.
Dear Chuan Liu,
u can try it by ratio. try it by bigger number. how it is working? or else perhaps by checking proper initial parameter setting.
best wishes.
Dear Chuan Liu,
how if u try integer number. it is compulsory for u to use Comsol for RF module?
my experience for RF, i use CST and HFSS. sorry i just share my experience.
it is possible also with comsol but its need careful with the sensitive issue in terms of number.
Noor HIDAYAH Mohd Yunus
Thank you again for your support. Using integer number is fine, extremely small value problem may relate to "singularity" in calculation I guess, but I do not know how to solve the problem.
According to the information I have, only Comsol can analyze 2D model, in CST or HFSS the problem is always 3D. (Can they simulate 2D model?)
Chuan Liu
Normally COMSOL rounds up numbers toward a certain accuracy whose value could be changed in the setting. The small value (like e-10) is far away beyond the default accuracy. Have you tried to change it? I think if you change the accuracy to be like e-12, you should be able to resolve the problem.
A problem that I expect will hit the simulation is that the electric and magnetic fields in your material will approach an infinite value as the relative permeability and relative permittivity approach zero. I expect that the calculation is carried out using electric and magnetic fields, rather than flux densities (which would be better behaved) and so the simulation may become very badly conditioned.
You must increase your permittivity and permeability. Try values like 0.01 and 0.001 and see if the results differ much.
- Badges
- Science topic