Using two SteadyPhaseChange solvers in one simulation

Numerical methods and mathematical models of Elmer
Post Reply
arved
Posts: 27
Joined: 06 Aug 2020, 15:05
Antispam: Yes

Using two SteadyPhaseChange solvers in one simulation

Post by arved »

Dear Elmer community,

I am trying to solve a case involving two phase boundaries using SteadyPhaseChange. My axisymmetric setup consists of three blocks stacked on top of each other with top and bottom solid and liquid in the center. When using only one SteadyPhaseChangesolver it works well, with the case attached to this post I ran into the following error:

Code: Select all

SteadyPhaseChange: --------------------------------------------
SteadyPhaseChange: Using steady algorithm to find the isotherm
SteadyPhaseChange: --------------------------------------------
SteadyPhaseChange: Steady state isoterm formulation
SteadyPhaseChange: Melting point found:                5.0500E+02
Note: The following floating-point exceptions are signalling: IEEE_UNDERFLOW_FLAG
STOP 1
ERROR:: SteadyPhaseChange: Isotherm is empty thus cannot map phase change surface
Is it generally possible to run such a case or would it require a modification of the solver?

I played a bit around and tried different configurations but nothing is working. It can be all found here, including a more extensive description and a couple of screenshots: https://github.com/nemocrys/elmer-double-phasechange

Thanks for your help and best regards
Arved
Attachments
elmersolver.log
(28.31 KiB) Downloaded 67 times
case.sif
(4.52 KiB) Downloaded 68 times
case.msh
(29.96 KiB) Downloaded 70 times
raback
Site Admin
Posts: 4812
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: Using two SteadyPhaseChange solvers in one simulation

Post by raback »

Hi

Looking at line ~183 in
https://github.com/ElmerCSC/elmerfem/bl ... Change.F90

You can see that there is some stuff being allocated & initialized the 1st time. These should be different for the two instances so some modifications would be needed here. Also there should be a different flag for the two different instances. You might make a copy with different name & change the flag as the most simple solution.

-Peter
arved
Posts: 27
Joined: 06 Aug 2020, 15:05
Antispam: Yes

Re: Using two SteadyPhaseChange solvers in one simulation

Post by arved »

Hi Peter,

thanks a lot for this hint, I adjusted it and it works!

Attached to this post you can find my modified 'SteadyPhaseChange2.F90'. I'll also upload the whole setup here later today: https://github.com/nemocrys/elmer-double-phasechange

Best regards
Arved
Attachments
SteadyPhaseChange2.F90
(37.33 KiB) Downloaded 70 times
Post Reply