ERROR 33132/ FEM setup
I want to simulate the electromagnetic back scattering response from half of a biconical antenna over an anisotropic half plane. Since FEKO does not support the use of a half plane with an anisotropic dielectric function I am instead trying to use a large box with the FEM. I have surrounded my system with an air buffer, but am receiving the following warning:
ERROR 33132: Not enough memory available for an in-core solution. Out-of-core storage of the MoM matrix is not supported with the FEM
See also message in the output file Model.out
So are my options essentially to simulate with more memory such that the FEM-MoM matrix can be stored locally, or reduce the size of my simulation? Is it possible to simulate purely with the FEM with absorbing boundary conditions? If so how would one set this up? I have attached my model below
Answers
-
Hi pmcardle:
When surrounding an FEM simulation with an airbox, you most likely did not intend to also have Method-of-Moments (MoM) activated. By default, Feko will solve FEM simulations with a MoM/FEM hybridization. I would suggest navigating to the Solve/Run tab in CADFEKO, selecting Solver Settings and switching to the FEM tab of the respective dialog box. There is a checkbox to decouple MoM. The resulting simulation will be a pure FEM analysis.
Regards,
Derek
0 -
Thank you for the response. As a follow up, what is the boundary condition for the FEM? and how does it
differ from the MoM boundary condition?
Cheers,
Patrick
0 -
FEM uses an airbox that is called an Absorbing Boundary Condition (ABC). The air is responsible for absorbing all frequencies, all polarizations and all angles of arrival. MoM exploits the tangential component of the electric field converging to 0 as the field approaches a perfect electric conductor (PEC). The rule can also be massaged to account for dielectric materials and finite conductivity metals.
The FEM airbox can have difficulty completely absorbing energy across all scenarios and especially grazing angles. Energy then reflects off the airbox and back into the simulation, which corrupts the results. The solution is to then make the airbox bigger at the cost of additional computational resources. MoM does not have this limitation and can handle propagation analytically exact.
0 -
In my simulation I am considering the scattered far-field from my antenna over a box. If I am using the absorbing boundary condition, is the scattered field being absorbed by the air box? Or am I not conceptualizing this correctly.
0 -
The airbox is definitely absorbing energy. The user is still required to verify accuracy. Users should always first work through a mesh convergence study to verify accuracy. Sometimes during this process the results do not converge regardless of how small the mesh elements become. One solution during this scenario is to enlarge the airbox. If the results change with respect to airbox dimensions, then this behavior is typically a clue that the airbox was too small.
0