Convergence issue with MLFMM

Christian Ballesteros

Hello,

I am trying to simulate a vehicle with two LTE antennas mounted on top at relatively low frequencies (700-900 MHz) but the MLFMM solver is not converging. The number of triangles is between 200k and 300k (depending on mesh). I tried using the additional stabilisation and it seems that the residuum converges but it never stops (crossing the typical 3E-3 limit). Where could I find some more information about how to control it? I have already checked the manual instructions and played with the mesh, box size, etc. I need some practical guidelines and explanation about how to control the parameters and what is their meaning. Also, the advanced iterative solver panel is a bit confusing - what does the stopping criteria mean and how to modify it -.

Thank you very much.

 

BR,

Christian

Torben Voigt

Hi @Christian Ballesteros ,

MLFMM works best if the mesh is "appropriate". This means that the mesh should have edge lengths in the range of lambda/12 or lambda/10. Wenn ein großer Teil deines Mesh sehr viel kleiner als lambda/10 ist, dann wird nicht nur der Speicherbedarf und die Laufzeit deutlich steigen, sondern es wird auch die Konvergenz darunter leiden.

Is it possible that your model contains a lot of geometric details, resulting in many very small meshelements? In the automotive field, it is common to first simplify the CAD data to be simulated. Usually this is done in HyperMesh, but in principle it is also possible in Feko.

Best regards,
Torben