Substraction of thin volumes (Thermal simulation of a chip) 1

[Patyuchenko]

I am trying to implement thermal simulation of a simplified model of a microchip. The geometry of the chip includes very thin volumes (their thickness is small in comparison with their length and width) which represent depletion regions of transistors, i.e. the heat sources. These volumes are surrounded by another volume which represents the shallow trench isolation.

The problem is in the following: when I try to substract these thin volumes from the big volume I get a message that the obtained entity is identical to the initial one. I tried to adjust tolerances for Boolean operations but it gave no success.

The only thing which helped is the change of the thicknesses of the thin volumes: initialy it is about 10 nm and when I change it into 30 nm then the volumes can be substracted (the width and length of the volumes are 1 um and 15 um correspondenly) but doing so I destroy the real geometry of the model what is not desirable.

Has anyone solved a similar problem? You would help me a lot.

Here, with the same simulation, I encountered another problem connected with existence of gray areas in Results Viewer. I will create a new thread on this issue. If anyone faced the same problem I will be very grateful to you for your help.

 

[cbrn]

Hi,
I can understand your worrying very well: Ansys is very disappointing as regards boolean operations (as regards solid modeling as a whole, I'd say...). I'd suggest you to build the geometry in a CAD system, if you can.
But however afterwards it's very unlikely that you will be able to mesh without problems, due to the enormous difference in the size of the various edges / faces.
Perhaps the way out is build separate pieces of solids instead of hollowing one single solid, and then spend a lot of patience with manual mesh adjusting (transitions, etc...)
Hope this helps...

Regards

 

[Patyuchenko]

Thank you very much cbrn for your help.

I tried to surround the thin volumes with another volumes which all together form the initialy desired bigger volume. Then step by step I meshed each of these small volumes. However, it takes really a lot of time and therefore it is not always suitable for the structures geometry of which should often be changed and resimulation performed.

This is probably is a real problem in ANSYS.

Thank you again.

Regards

 

[cbrn]

Hi,
yes, the modeling process is a pain in Ansys: its modeler was born in the early eighties and afterwards, with the birth of the modern solid modelers, the intent seemed to push users building their geometry in CAD systems and then import, especially for very particular (as in your case) or complex geometry.
But the topology you are dealing with would be critical with every FEM on the market, you can be sure of it!!! Thanks to the mesh controls in Ansys you can work it out, but in others you would have to rely on the automatic mesher and you could just pray that it would be able to handle the situation...

Regards

 

[Patyuchenko]

Hi,

Thank you for support cbrn.

Best regards