Altair RISE

A program to recognize and reward our most engaged community members

Nominate Yourself Now!

Getting error 38878 from the exported model

Hi, I am exporting one patch array (4x4) from CST to FEKO 2023.1. I recreated port and did the union operation after defining material properties. The model status mentioned the model is valid. But I am getting error 38878 . My model geometry file and the error screenshot is attached here. Kindly fix my error so that I can run my file. File is attached here.

Feko_array_try.cfx

Find more posts tagged with

English

Feko

v2023.1

Accepted answers

Torben Voigt

Hi @Srabonty Soily,

Due to the size of the model, memory requirements and runtime are quite large. I will therefore use MLFMM instead of standard MoM. However, I could imagine that FEM could be the most efficient solver here. I will try that as well.

Best regards,
Torben

All comments

Torben Voigt

Hi Srabonty Soily,

The recently implemented mesh engine seems to have problems with the very thin PEC regions of the waveguides. If I switch to "legacy mesher", it works.

However, I would really advise simplifying the model. It doesn't really make sense to model 35 µm thick PEC patches with phyisic thickness. Instead, you should just use surfaces, which significantly reduces the number of mesh elements and thus memory and runtime.

By the way, there is another error in the model: In the waveguides, the exciting wave hits a PEC face (and is reflected), which is not intended:

I will change the model a little and get back to you later.

Best regards,
Torben

sswh2

@Torben Voigt sir thank you! I will wait for your corrected model.

Torben Voigt

Hi @Srabonty Soily,

Best regards,
Torben

Torben Voigt

Hi @Srabonty Soily,

I would recommend MLFMM with my simplified model, see attached. The model seems to be a little large for FEM. With MLFMM and e.g. 16 parallel processes, only 35 GByte RAM are required, and 11 frequencies are calculated in 15 minutes.

Best regards,
Torben

Feko_array_try_MLFMM.zip

sswh2

Hi @Srabonty Soily,

I would recommend MLFMM with my simplified model, see attached. The model seems to be a little large for FEM. With MLFMM and e.g. 16 parallel processes, only 35 GByte RAM are required, and 11 frequencies are calculated in 15 minutes.

Best regards,
Torben

Thank you@ Torben Voigt sir. But I found the problem in my array is there is only one waveguide source defined in my model, there should be 16 waveguide source I think.

Torben Voigt

Thank you@ Torben Voigt sir. But I found the problem in my array is there is only one waveguide source defined in my model, there should be 16 waveguide source I think.

Hi @Srabonty Soily,

In the original model you attached there were 16 waveguide ports but only one waveguide source. I qickly added the other sources as well now, see attached.

Best regards,
Torben

Feko_array_try_16_sources_MLFMM.cfx

sswh2

Hi @Srabonty Soily,

In the original model you attached there were 16 waveguide ports but only one waveguide source. I qickly added the other sources as well now, see attached.

Best regards,
Torben

@Torben Voigt sir I appreciate it! Thanks! Let me run it.