Why identical two HFSS models have large different S-parameters?

[Jim1901]

Could someone figure out why the same HFSS projects have large different S-parameters? As shown in atttached figure, symboled lines are from one project, no-symboled are from another projects. They are very different in terms of the accuracy of HFSS. The results are very weird. Maybe I missed some details. Can some one find out what happens for these two projects?
Thanks!
The model is a partially filled rectangular waveguide with the partial part filled with the material we defined.
The attached zipped file has two HFSS projects.

 

[JonathanC]

maybe double check the delta s value and from the message box is it saying the adaptive mesh is not converged?

 

[Jim1901]

I have checked that. I am pretty sure that the results are converged...

 

[volker@muehlhaus]

If you put both files into a text compare tool, you will find differences like this:

< 'Calculate Lossy Dielectrics'=false
> 'Calculate Lossy Dielectrics'=true

< DoLambdaRefine=true
> DoLambdaRefine=false

Geometry might be the same, but simulation settings are not identical.

 

[Jim1901]

If you put both files into a text compare tool, you will find differences like this:

< 'Calculate Lossy Dielectrics'=false
> 'Calculate Lossy Dielectrics'=true

< DoLambdaRefine=true
> DoLambdaRefine=false

Geometry might be the same, but simulation settings are not identical.

Thanks so much. It is solved. This is related to the defined materials. The software maybe set this unnoticeably depending on your modelling process.