Hi
I could be wrong but shouldn't there be checks for negative densities in FlowSolver and for negative concentrations in AdvectionDiffusionSolver just to protect against potential numerical inaccuracies ?
Regards
Check for negative field values
-
raback
- Site Admin
- Posts: 4832
- Joined: 22 Aug 2009, 11:57
- Antispam: Yes
- Location: Espoo, Finland
- Contact:
Re: Check for negative field values
Hi
There are soft and hard limiters in the library that you can use. Checking for negative values is akin to finite volume method where you do local tricks more often. If the negative value minimizes the energy functional in finite element method it is not ideal just to cut of those automatically. One should analyze what is happening and perhaps improve the mesh etc.
In some cases the negative values result to serious convergence issues even though at convergence those values would vanish. For example, in RANS turbulence models where the negative values (or zeros) are in the nominator it may be best to cut them off.
-Peter
There are soft and hard limiters in the library that you can use. Checking for negative values is akin to finite volume method where you do local tricks more often. If the negative value minimizes the energy functional in finite element method it is not ideal just to cut of those automatically. One should analyze what is happening and perhaps improve the mesh etc.
In some cases the negative values result to serious convergence issues even though at convergence those values would vanish. For example, in RANS turbulence models where the negative values (or zeros) are in the nominator it may be best to cut them off.
-Peter