Pavement or Road model representation in FEKO

Jey

Hi,

I use a car model, I want to model the Pavement or Road for an EM simulation.- using MLFMM. Model approaches used 

a) Pavement modeled as dielectric medium

b) Plane or Array Planar multiplayer substrate (Layer 0 - +inf with Free space; Layer-1 with 10mm thk+FS; Layer 2 with -inf +FS) & Solver settings: Domain decomposition : solve model with DGFM enabled)

Model (a) has more run time, so preferring Model approach (b). In model (b) below error received. Any suggestions ?

' ERROR    2580: No special Green's function such as the planar multi-layer substrate may be activated with the MLFMM'

Mel

Usually pavements or real earth are approximately modeled with an infinite half space.

There are two options, reflection coefficient ground plane, which supports MLFMM. Here the car and other geometry cannot touch the half space.

The other half space option is the Sommerfeld half space (based on the multilayer Green's Function) which does not support MLFMM but geometry can touch and penetrate the half space.

 

The DGFM is used mainly for arrays to reduce computational resources between array elements by approximating the coupling between the elements. In the case of a car and some plane, the DGFM should not be used.

But in any case the MLFMM together with planar multilayer substrate are not supported.

 

If the frequency is very high another option would be to model the pavement as a thin dielectric sheet.

 

 

Jey

Thanks for the explanation. With Planar multi-layer substrate I cannot use MLFMM. 

To simulate method to represent the reality,

High frequency problem, with pavement 15 mm thick & 4m X 4m in size below the car @ atleast 0.5*lamda distance, assuming I have correct  dielectric properties fro pavement.

1) MOM method with pavement (15t), dielectric sheet has out-of core issue, bcoz Pavement itself ~0.5 million # of elements

2) MLFMM method with dielectric, can handle problem size. But pavement sheet cannot be thk>2mm & high computation time.

3) MLFMM method with halfspace (ref-coeff), can handle problem size also less computation time, but thickness cannot be accounted (if can be let me know)

 

Assuming method (1) as ideal case, can the (2) dielectric property can be tuned for a thin thk? or (3) can be used w/o thk consideration. What method can be adopted, for good computational time with less compromise in solution accuracy ? if any other approach available let me know..

 

Mel

I see that you are not a University user. If you are a commercial user please contact your local Hyperworks representative for further support and provide as much details of your model as possible.