heat & NS with phase change

Numerical methods and mathematical models of Elmer
Post Reply
mark smith
Posts: 215
Joined: 26 Aug 2009, 18:20
Location: Peterborough, England

heat & NS with phase change

Post by mark smith »

Hi All,
I have been trying heat + NS and phase change for a phase change material like paraffin wax, I used the excellent work posted by Julien Givernaud in post viewtopic.php?f=3&t=3846&hilit=phase+change (thanks very much) with the appropriate material properties for my case and I am getting some good results.
I would like to do the problem as an axi-symmetric problem rather than 2D Cartesian but I get a NAN fault in the heat solver when I change the coordinate system, can this type of problem be solved as axi-symmetric (y axis as the axis of rotation) using the standard solvers? I also tried to do it in parallel with the cartesian coordinate system but get an error in the flowsolver run, is the heat with latent heat only usable in serial mode?
Thanks in advance
Regards
Mark
mzenker
Posts: 1999
Joined: 07 Dec 2009, 11:49
Location: Germany

Re: heat & NS with phase change

Post by mzenker »

Hi,

quickly, without digging further into your problem: there is also cylindric symmetric, see this post for an explanation of the difference.

HTH,

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

Re: heat & NS with phase change

Post by raback »

Hi Mark,

The problem should be solvable also as axisymmetric.

The reinitiaization i.e. the computation for signed distance needed for the levelset probably does not work in parallel. The standard PDE solvers such as N-S and Heat should otherwise be fine. You might have to tweak the linear strategies there since the standard ILUn is only done partitionwise and therefore usually leads to inferior convergence in parallel. But as said, the reinitialization will probably be a problem.

-Peter
mark smith
Posts: 215
Joined: 26 Aug 2009, 18:20
Location: Peterborough, England

Re: heat & NS with phase change

Post by mark smith »

Hi Peter and Elmer users,
In the models manual for the heat equation, under the Equation section it has various settings for Phase Change Model, Spatial 1, Spatial 2 and Temporal, now in my cartesian problem I use Spatial 2 as that was what Julien Givernaud used in his simulation, now when I change the problem to Axi Symmetric Spatial 2 throws a NAN error in the heat solver, but if I change to Spatial 1 it runs without a problem. The manual throws no light on what the settings Spatial 1, Spatial 2 and Temporal actually mean and when each type should be used so I'm not certain that although Spatial 1 seems to work that the results are valid, could you please enlighten?
I've looked at the source code but not being a Fortran person it doesn't help me ;-(

Thanks in advance.
Best regards
Mark
annier
Posts: 1168
Joined: 27 Aug 2013, 13:51
Antispam: Yes

Re: heat & NS with phase change

Post by annier »

mzenker wrote:Hi,

quickly, without digging further into your problem: there is also cylindric symmetric, see this post for an explanation of the difference.

HTH,

Matthias
Matthias, The discussion you started and we participated in the thread seems to be very important for us ( Elmer users) (especially, making nonlinear Newton iteration = 0)


Yours Sincerely,
Anil Kunwar
Anil Kunwar
Faculty of Mechanical Engineering, Silesian University of Technology, Gliwice
annier
Posts: 1168
Joined: 27 Aug 2013, 13:51
Antispam: Yes

Re: heat & NS with phase change

Post by annier »

Mark,
Please see the thread on spatial 1, spatial 2 and temperoral started by Ehooi.
viewtopic.php?f=3&t=3357


Yours Sincerely,
Anil Kunwar
Anil Kunwar
Faculty of Mechanical Engineering, Silesian University of Technology, Gliwice
mark smith
Posts: 215
Joined: 26 Aug 2009, 18:20
Location: Peterborough, England

Re: heat & NS with phase change

Post by mark smith »

Hi All,
Thanks Anil for the link on the Spatial posts and other helpful suggestions.
As I posted before I am getting some useful results for a cylinder of paraffin wax like material melting in serial mode (axi symmetric simulation) with a fully quadrilateral mesh but when I try to solve in parallel I get an error and exit ;-)
Attached is my sif file, material property file, the run log, gmsh geometry file and an image of the problem domain, can anyone explain why I should get the error in the NAVIER-STOKES ITERATION section and what I can try to get it to run sucessfully?
Should you want to repeat the simulation the mesh can be built from the .geo file using gmsh.

Thanks in advance
Best regards
Mark
Attachments
Elmer_pcm60_axi_2.zip
zipped problem files
(30.22 KiB) Downloaded 267 times
annier
Posts: 1168
Joined: 27 Aug 2013, 13:51
Antispam: Yes

Re: heat & NS with phase change

Post by annier »

Hi Mark,
Recently, I just read a discussion between TFLOPS and Peter, in which Peter has explained some codes written to be run in serial only.
viewtopic.php?f=3&t=4495&p=16454&hilit= ... 034#p16454
Regarding your case, Peter may explain which part of your input in SIF leads to error in parallel exection.

Cheers,
Yours Sincerely,
Anil Kunwar
Anil Kunwar
Faculty of Mechanical Engineering, Silesian University of Technology, Gliwice
Post Reply