Hybrid PO and MoM SEP solvers

Shan25_22239

In the attached I am modelling a reflector with a dielectric lens. The reflector is modeled using large element PO solver and homogeneous ci-convex is modeled by MOM SEP method. The permittivity of lens is set to be equal air, such that the nearfield data must be identical to the no-lens case. The result predicted by hybrid PO-MoM SEP is not correct. In the solver high frequency settings I tried to decouple PO and MoM solutions but the results are identical to the full coupling case. 

I have attached models for your reference. Can you please have a look at the solver settings?

 

Reflector_BiConvex_epsR_1.0_HybridSolve_Decoupled.cfx

Reflector_BiConvex_epsR_1.0_IterativeCoupled.cfx

Reflector_BiConvex_epsR_1.0_Test_PO_No_Lens.cfx

Rene Fiedler_22147

Hello @Shan25 ,

 

PO is not suiteable for transmission through dielectrics. It is a good and useful method for single reflection placement analysis. It only models a single reflection. It can also model a transmission in the case where we use a layered medium but otherwise it is not suitable for transmission.

For this we have RL-GO. So you could repeat the calculation with RL-GO and compare to MLFMM. Note in some cases there is a penalty in the runtime when using many observation points. 

Otherwise it seems you are using 2020.0 - if you update FEKO we have replaced the default solver in MLFMM and made it more efficient for more cores. So your runtime and memory should both decline after update. 

Please test this as well if you think the runtime currently is too high. Tough in this case I would skip the local mesh refinement and mesh overall with "coarse" - using a coarser than recommended mesh can lead to longer runtimes in some cases.