no creation of matrix system

[Daniela]

hello there,

I tried extending my 2D simulations to 3D, but somehow the combination of ElmerGrid and gmsh doesn't create a good file.
I tried a really simple setup, a qubic box which I genereated in gmsh, but when I run ElmerSolver, I get this output:

• ELMER SOLVER (v 6.0) STARTED AT: 2010/11/29 15:50:20
  MAIN:
  MAIN: ==========================================
  MAIN: E L M E R S O L V E R S T A R T I N G
  MAIN: Library version: 6.0 (Rev: exported)
  MAIN: ==========================================
  MAIN:
  MAIN:
  MAIN: -----------------------
  MAIN: Reading Model ...
  Loading user function library: [/viscosity_T]...[viscosity_T]
  Model Input: Unlisted keyword: [molecular diameter] in section: [material 1]
  Model Input: Unlisted keyword: [calculate weights] in section: [solver 1]
  MAIN: Done
  MAIN: -----------------------
  MAIN:
  MAIN: -------------------------------------
  MAIN: Steady state iteration: 1
  MAIN: -------------------------------------
  MAIN:
  ComputeNodalWeights: Computing weights for solver temperature
  ComputeNodalWeights: All done
  ComputeChange: SS (ITER=1) (NRM,RELC): ( 0.0000000 0.0000000 ) :: heat equation
  ComputeChange: SS (ITER=1) (NRM,RELC): ( 0.10000000E-05 2.0000000 ) :: navier-stokes
  WriteToPost: Saving results in ElmerPost format to file ./test.ep
  MAIN:
  MAIN: -------------------------------------
  MAIN: Steady state iteration: 2
  MAIN: -------------------------------------
  MAIN:
  ComputeNodalWeights: Computing weights for solver temperature
  ComputeNodalWeights: All done
  ComputeChange: SS (ITER=2) (NRM,RELC): ( 0.0000000 0.0000000 ) :: heat equation
  ComputeChange: SS (ITER=2) (NRM,RELC): ( 0.10000000E-05 0.0000000 ) :: navier-stokes
  ElmerSolver: *** Elmer Solver: ALL DONE ***

Since I used nearly the same sif file for the 3D than for the 2D I assume that there is something wrong with the generation of the mesh in the gmsh program. But I have no idea where my mistake might be.
Gmsh seems to work fine, this is my cube,

wcube.jpeg (348.63 KiB) Viewed 3950 times

,
the same this command "ElmerGrid 14 2 cube.msh"
but in the end, ElmerSolver doesn't work properly.

Do yo have any clue, where my mistake might be?

Thanks a lot in advance,
Daniela

[Daniela]

just in case you need more information,

I attach my used sif File, the geo file for creating the mesh with gmsh and the resulted .msh file.

thanks again,

Daniela

[raback]

Hi Daniela

Your boundary indexes are incorrect. Change 102 -> 28 and you should be in the game.

-Peter

[Daniela]

Hi Peter,

thanks,

I realized that just now and I changed it to 27, (since otherwise the boundary 28 is twice in the file)
but it didn't help.

So, do you have any other clue?

thanks, Daniela

[Daniela]

hello there,
just in case anybody else gets my problem:

I solved it by defining each surface with its own number and the number of boundaries have to go from 1 up to needed boundary, in my case 6.
I had only 27 and 28 and this didn't work.

So, if you are using gmsh and if you are defining your boundaries with help of gmsh, then take care of numbering of boundaries afterwards.

greetings,
Daniela

[mzenker]

Hi Daniela,

since I also use gmsh and was not aware of the problem you describe, I have tried to reproduce it. I cannot confirm that defining each surface as physical with an individual index starting from 1 makes a difference compared to your initial physical surface definition in gmsh. For all cases the solver runs for me, it only spits a

SolveSystem: Solution trivially zero!

in the Navier Stokes section of the log.
I must admit that I have never used the Navier Stokes solver before and do not have your viscosity_T function. So I defined the viscosity as Variable Temperature with 2 points. Apart from that and the boundary conditions, I have used your sif file. I did not find the cause of the "trivially zero" problem, but I did not search for such a long time either.
Anyway, since the numbering of the boundaries made no difference, I doubt that this was really the cause of the problem you had.

Best regards,

Matthias

[Daniela]

Hi Matthias,

I tested it again: My case works as soon as I use before generating the mesh:

• Physical Volume(1) = {26};

you're right, I don"t need to define the physical surfaces from one upwards, but Elmer ran properly only after setting the volume to 1.

The Navier Stokes Solver is trivially zero because in this setup there is no real flow, I was only too lazy to remove this part from the sif file, so don't worry.

Greetings,
Daniela

[mzenker]

Hi Daniela,

your case works for me even when the volume has another physical number, e.g. 29.

Best regards,

Matthias

[Daniela]

Hi Matthias,

not with me.

maybe the problem is that I compiled Elmer directly for unix and I didn't do the same for gmsh since I found a precompiled version.

Anyway, I know how to get my program working, so I'm fine.

Greetings,
Daniela