Hello Elmer gurus -
I run into a strange problem.
I;m doing some testbeds for simple HVAC ducts, nothing complicated.
The goal is to illustrate to students how pressure and speed levels of the air are in different ducts.
To simplify things I use one of the examples that are supplied by ELMER team (LaminarIncompressibleStepFlow) and then load a different
mesh inside the project, reapply boundary conditions, material, etc.
The problem that I encounter is that in one case I get the error ERROR:: IterSolve: Failed convergence tolerances while the other one is running the solver with no error ...
So the simple question is why is this happening (the meshes for HVAC ducts are generated using same software, same steps, etc.)
I can upload the files if that could help.
Thanks for help!
ERROR:: IterSolve: Failed convergence tolerances
-
- Site Admin
- Posts: 4832
- Joined: 22 Aug 2009, 11:57
- Antispam: Yes
- Location: Espoo, Finland
- Contact:
Re: ERROR:: IterSolve: Failed convergence tolerances
Hi
You might improve on the preconditioner (with the expense of memory) or try some other iterative solver. For example ILU1 or ILU2 with BiCGstabl and "Bicgstabl Polynomial Degree = 4" works often rather well. The structured mesh in the tutorial is rather optimal and you may require a much denser mesh for similar accuracy using Delaunay meshes.
-Peter
You might improve on the preconditioner (with the expense of memory) or try some other iterative solver. For example ILU1 or ILU2 with BiCGstabl and "Bicgstabl Polynomial Degree = 4" works often rather well. The structured mesh in the tutorial is rather optimal and you may require a much denser mesh for similar accuracy using Delaunay meshes.
-Peter
Re: ERROR:: IterSolve: Failed convergence tolerances
Peter -
thanks for your suggestions.
I'll try ILU1 and ILU2 with BiCGstabl.
If won't converge I'll try to improve the mesh density.
I'll post my results.
Thanks for helping.
best,
Adrian
thanks for your suggestions.
I'll try ILU1 and ILU2 with BiCGstabl.
If won't converge I'll try to improve the mesh density.
I'll post my results.
Thanks for helping.
best,
Adrian
raback wrote:Hi
You might improve on the preconditioner (with the expense of memory) or try some other iterative solver. For example ILU1 or ILU2 with BiCGstabl and "Bicgstabl Polynomial Degree = 4" works often rather well. The structured mesh in the tutorial is rather optimal and you may require a much denser mesh for similar accuracy using Delaunay meshes.
-Peter