[Urgent ]CST higher meschcells is running my results

[zA123]

So far I was simulating my structure at around 800 000 meschcells , I am limited to 1000000 by the student version .
so 800 000 is still pretty high compared to my limit .
Then I re simulated at 900 000 and my results are drastically different . should they be that different ?
please help ...

 

[johnjoe]

Please provide more information. What are you simulating, TD, FD and post some pics and results...

 

[zA123]

sad;

Please provide more information. What are you simulating, TD, FD and post some pics and results...

sadly for agreement purposes I cannot send much info but I am simulating in time domain , and my structure gives the best results at a certain mesh point and the results start to get ruined onwards... does more mesh cells = more accuracy or there is a best mesch cell number compatible with each structure depending on the structure ?

 

[johnjoe]

The result should not depend on mesh size or number of mesh cells. Depending on your structure, you should start with a low mesh cell number and use adaptive meshing. The simulator will stop to increase mesh cell numbers as soon as results will be unchanged with a pre-specified error.

 

[zA123]

The result should not depend on mesh size or number of mesh cells. Depending on your structure, you should start with a low mesh cell number and use adaptive meshing. The simulator will stop to increase mesh cell numbers as soon as results will be unchanged with a pre-specified error.

I tried adaptive meshing but it didn't yield any effective results since the values on average of my S-parameters are not wrong as values but they face some reflection after a certain frequency , can this meshing be done by manual trials?

 

[volker@muehlhaus]

does more mesh cells = more accuracy

Yes! More mesh density = more accurate. If results break with higher mesh density, they are not real.

 

[johnjoe]

Maybe it would possible to use some kind of subgridding for the mesh. But it is hard to say what can help without knowing what you are simulating.

 

[volker@muehlhaus]

I guess the student version is just too limited to solve this problem accurately.