ELMER crash

The graphical user interface of Elmer
Tonio
Posts: 8
Joined: 12 Mar 2017, 20:35
Antispam: Yes

ELMER crash

Post by Tonio »

Hello,

I get a crash of Elmer solver when solving heat diffusion equation in a system made of 6 bodies in contact, with heat generation in one of these bodies.

Crash is generated by ResultOutput which I want to use to generate a vtu output file for further analysis in Paraview.

Error message is:
"SingleSolver: Attempting to call solver
SingleSolver: Solver Equation string is: result output
ResultOutputSolver: -------------------------------------
ResultOutputSolve: Saving with prefix: case130umcu20um
ResultOutputSolver: Dimension of mesh: 3
ResultOutputSolver: Creating list for saving - if not present
CreateListForSaving: Field Variables for Saving
CreateListForSaving: Scalar Field 1: temperature
ResultOutputSolver: Saving in unstructured VTK XML (.vtu) format
VtuOutputSolver: Saving results in VTK XML format with prefix: case130umcu20um
VtuOutputSolver: Saving number of partitions: 1


Program received signal SIGSEGV: Segmentation fault - invalid memory reference.

Backtrace for this error:
#0 ffffffffffffffff
VtuOutputSolver: Number of geometry nodes to save: 26238
VtuOutputSolver: Number of dof nodes to save: 26238
VtuOutputSolver: Number of elements to save: 143634"

Modifying SIF file so that ResultOutput is never executed result in a succesfull solve by ELMER, but of course without vtu files being generated.

I used ResultOutput successfully on a simpler geometry (5 bodies, same heat diffusion equation).

Any hint? Might this be a problem with my geometry (although prostprocessing by Elmer VTK seems to show expected heat diffusion and temperature distribution in the geometry)?

Mesh was generated by gmsh if this helps.

Antoine
mzenker
Posts: 1999
Joined: 07 Dec 2009, 11:49
Location: Germany

Re: ELMER crash

Post by mzenker »

Hi,

could you post your sif file?
What happens if, instead of calling the ResultOutput solver explicitly, you say

Post File = case130umcu20um.vtu

in the Simulation section (Model -> Setup in ElmerGUI)?

HTH,

Matthias
Tonio
Posts: 8
Joined: 12 Mar 2017, 20:35
Antispam: Yes

Re: ELMER crash

Post by Tonio »

Hello,

Same failure occurs when requesting output file to be in vtu format instead of ep format (Model ->Setup choose case130umCu20um.vtu as output file). Elmer crashes after first timestep with error message : "Program received signal SIGSEGV: Segmentation fault - invalid memory reference.

Backtrace for this error:
#0 ffffffffffffffff"

I attach the SIF file.

Thank you for your help.

Antoine
Attachments
case130umCu20um.sif
(4.48 KiB) Downloaded 383 times
mzenker
Posts: 1999
Joined: 07 Dec 2009, 11:49
Location: Germany

Re: ELMER crash

Post by mzenker »

Hi,

so it's the result output solver indeed.
I don't know what happens here. However, since you define an equation for every solver and only use equation 1, meaning that the other solvers are never called anyway, I don't know if that can cause the problem.
In the GUI, you don't need to add an "equation" for every solver. You can only have a single "equation" per body. Instead, check all solvers "active" in one equation. Use different equations only if you want a solver for one body, but not for the other one.

HTH,

Matthias
Tonio
Posts: 8
Joined: 12 Mar 2017, 20:35
Antispam: Yes

Re: ELMER crash

Post by Tonio »

Thank you for the tip.

I removed all équations but the first (heat diffusion), and made active both ResultsOutput and SaveScalars in addition to heat equation.

Crash is still there when executing ResultsOutput, but my SIF file is smaller and I learned something.

If somebody already experienced crash of ResultsOutput and can chime in ....

Thanks again

Antoine
mzenker
Posts: 1999
Joined: 07 Dec 2009, 11:49
Location: Germany

Re: ELMER crash

Post by mzenker »

Hi Antoine,

could you post the new sif file that produces the crash?

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

Re: ELMER crash

Post by mark smith »

Hi Antoine,
Would you by any chance be doing a parallel solve with a partitioned mesh?
I have had Elmer crash when writing out the vtu results if I am writing out the materials, if all the materials are not present in each of the partitions then this could happen at least in earlier versions. i you are doing a serial solve then sorry this won't help ;-(
Regards
Mark
Tonio
Posts: 8
Joined: 12 Mar 2017, 20:35
Antispam: Yes

Re: ELMER crash

Post by Tonio »

Thank you Mark for your comments.

I'm a beginner with Elmer, so I do not yet use parallel solve.

Matthew, I uploaded the new SIF file should you like to examine it.

Crashes after first time step with message indicated in the log file attached.

Thanks again

Antoine
Attachments
solver060417.log
(5.8 KiB) Downloaded 386 times
case130umCu20um.sif
(4.36 KiB) Downloaded 383 times
raback
Site Admin
Posts: 4812
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: ELMER crash

Post by raback »

Hi

You seem to request heater control but don't have a heater. This is a feedback mechanism that sets the temperature to be the desired one by scaling the given r.h.s. approrpiately. But if your r.h.s. is zero it cannot scale it except with NaN resulting to your problems.

-Peter
Tonio
Posts: 8
Joined: 12 Mar 2017, 20:35
Antispam: Yes

Re: ELMER crash

Post by Tonio »

Hello Peter,

I try to resolve the temperature over time Inside this composite system. Heat is generated Inside Body 6. Body force (i.e. heat generation) is described in the SIF file (Body Force 2) applied to Body 6.

Would you be kind enough to explain me in more detail what is wrong? I do not understand your statement.

Best regards

Antoine
Post Reply