PhaseChange, mesh update plus linear elasticity

Numerical methods and mathematical models of Elmer
Post Reply
wmiller
Posts: 20
Joined: 09 May 2017, 18:05
Antispam: Yes

PhaseChange, mesh update plus linear elasticity

Post by wmiller »

Dear all

I like to run an axisymmetric calculation for Czochralski crystal growth, i.e. using
phase change "SteadyPhaseChange" and mesh update together with computing stress
via Linear elasticity "StressSolver".

Without the linear elasticity solver the mesh update works fine.
The phase change takes into account both melt and crystal and in both domains the
mesh is updated.

However, when adding the linear elasticity solver, which is acting only in the crystal, the
mesh update becomes nonsense, because it seems that I will use the displacement variable
for updating.
Is there something to be set to avoid that the mesh update solve is using the displacements
computed by the linear elasticity solver ?

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

Re: PhaseChange, mesh update plus linear elasticity

Post by raback »

Hi

Maybe give the StressSolver a different name for the "Variable". I think that the MeshUpdate solver only checks for "Displacement".

-Peter
wmiller
Posts: 20
Joined: 09 May 2017, 18:05
Antispam: Yes

Re: PhaseChange, mesh update plus linear elasticity

Post by wmiller »

Thank's Peter.

Indeed this works and the mesh update solver works correctly.

Wolfram
wmiller
Posts: 20
Joined: 09 May 2017, 18:05
Antispam: Yes

Re: PhaseChange, mesh update plus linear elasticity

Post by wmiller »

The problem runs fine if the material parameter for elasticity is given by a single value for Youngs Modulus.

If a matrix is given for Youngs Modulus in the Material section the Mesh Update Solver fails in the second global iteration ( some where NaN value ).
(In the first iteration step the solution is trivial because there is no change in geomety - the interface is flat)

There is no difference if Mesh Youngs Modulus and Mesh Poisson ratio are set in the Materials section.

So it seems that the Mesh Adaption Solver tries to use the matrix and fails internally (may be using a component, which is zero).

Is there any solution ?
Attachments
run-RGZ69conv-80.sif
anisotropic case
(22.63 KiB) Downloaded 284 times
run-RGZ69conv-80.sif
isotropic case
(22.09 KiB) Downloaded 271 times
raback
Site Admin
Posts: 4812
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: PhaseChange, mesh update plus linear elasticity

Post by raback »

Hi, a small test case would help. -Peter
wmiller
Posts: 20
Joined: 09 May 2017, 18:05
Antispam: Yes

Re: PhaseChange, mesh update plus linear elasticity

Post by wmiller »

I used the wrong keyword.

It is
---------------------------------------------------------
Mesh Elastic Modulus = 1e11
--------------------------------------------------------
This is working.
Post Reply