Memory problem in the MLFMM simulation

Altair Forum User

Hi,friends and experts,

 

When I am simulating a not that electric large simulation model,  FEKO always give me some error: not dynamic memory allocation.

I do check the reason. It was found that when goes to the precondioner calculation, it will need a a lot of memory. Also, different preconditioners may lead to different memory requirement.

So after goes to the No precondioner settings, there will be no problem at this stage. However, when goes to the iteration procedure, it will also have that problem.

 

Could anyone can help me with this problem? The model is attached here.

 

Thanks.

 

Unable to find an attachment - read this blog

Mel

Hi XuWang

 

You can try MoM with higher order basis functions. A standard mesh requires 31 GByte of memory.

If you don't have enough memory, this solver can use out-of-core solving. 

 

For MLFMM, you can set the combined field integral equation solution on the faces bounding the closed volumes in your model.

This improves convergence. 

 

A further improvement in convergence is obtained if you change the precision to double precision.

Since the source is some wavelengths away from the model you can use the 'Coarse' setting for create mesh.