Non-Boussinesq Convection

Numerical methods and mathematical models of Elmer
MrChips
Posts: 33
Joined: 12 Jul 2016, 00:16
Antispam: Yes
Location: Canada

Re: Non-Boussinesq Convection

Post by MrChips »

raback wrote:Hi
You could perhaps sent the case for inspection. It should not be so much more difficult from the Boussinesq case.
-Peter
Hi Peter,

I'm not really sure how, but I was able to get ILU preconditioning to work. I typically set up my models in ElmerGUI and then run things from the command line. I think ElmerGUI had some trouble after I made copies of the project files and started experimenting with different meshes. After copying the project files and opening them in the GUI, a different sif file would generate. Some, but not all settings would carry over. Loading in different meshes in the GUI without setting up a new project file seemed to have caused problems as well even after the sif file was corrected.

Anyways, I started the project from scratch and was able to get convergence in my first go.

Here are the results. These seem to be close enough. I have attached all the project files so that, if you wish, you can inspect it. Hopefully this helps others as well.
Image
Attachments
Non-Boussineq Run 1.rar
(98.91 KiB) Downloaded 283 times
raback
Site Admin
Posts: 4812
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: Non-Boussinesq Convection

Post by raback »

Hi

I ran your case and got into trouble myself. I think that the problem is in the undefined pressure level. Try adding line

Code: Select all

Pressure Single Node = Real 0.0
to the Body Force definition.

This problem was just added as test case to NaturalConvection3.

-Peter
MrChips
Posts: 33
Joined: 12 Jul 2016, 00:16
Antispam: Yes
Location: Canada

Re: Non-Boussinesq Convection

Post by MrChips »

What problems did you run into Peter? I did not have any convergence issues.

Could you explain the pressure level addition? Is this similar to defining a reference pressure for the solver?

Cheers!
raback
Site Admin
Posts: 4812
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: Non-Boussinesq Convection

Post by raback »

Hi

My case was sligthly different but it had a pressure level at around 1e9 after 3rd iteration. The problem is that pressure is not directly a variable, only the gradient of pressure is affected. Hence there can be an offset. This additional keyword defines the pressure level. I guess one could be lucky in that the iterative process does not lead to diverging pressure level. Also I had inconsistencies depending on the linear solver strategy and preconditioner used.

-Peter
annier
Posts: 1168
Joined: 27 Aug 2013, 13:51
Antispam: Yes

Re: Non-Boussinesq Convection

Post by annier »

This is just a note to the heat solver divergence in coupled heat transfer and fluid flow analysis:
Message in Solver Run:

Code: Select all

Heat Solver: ILU0 ...
NUMERICAL ERROR:: IterSolve: System diverged over tolerance.
NUMERICAL ERROR:: ComputeChange: Norm of solution appears to be NaN
Heat Solver ...
I have changed the linear solver settings from Iterative to Direct.

Code: Select all

Solver 1
      Equation = Heat Equation
      Procedure = "HeatSolve" "HeatSolver"
      Calculate Loads = True
      Variable = Temperature
      Exec Solver = Always
      Stabilize = True
      Bubbles = False
      Lumped Mass Matrix = False
      Optimize Bandwidth = True
      Steady State Convergence Tolerance = 5.0e-7
      Nonlinear System Convergence Tolerance = 1.0e-7
      Nonlinear System Max Iterations = 20
      Nonlinear System Newton After Iterations = 3
      Nonlinear System Newton After Tolerance = 1.0e-3
      Nonlinear System Relaxation Factor = 1
      !Linear System Solver = Iterative
      !Linear System Iterative Method = BiCGStab !BiCGStabl
      !BiCGstabl polynomial degree = 10
      !Linear System Max Iterations = 500
      !Linear System Convergence Tolerance = 1.0e-10 ! 1.0e-5
      !Linear System Preconditioning = ILU0
     ! Linear System ILUT Tolerance = 1.0e-3
      !Linear System Abort Not Converged = False
      !Linear System Residual Output = 1
      !Linear System Precondition Recompute = 1
      Linear System Solver = Direct
      Linear System Direct Method = UMFPack
      Steady State Convergence Tolerance = 1.0e-4
    End
yours sincerely,
anil Kunwar
Anil Kunwar
Faculty of Mechanical Engineering, Silesian University of Technology, Gliwice
Post Reply