infinite MeshVelocities

Extension of Elmer in computational glaciology

infinite MeshVelocities

Postby Martina » 24 Feb 2014, 12:45

Hi all,
I have a problem with Mesh Velocities in the following setup:
a) I do a surface relaxation run (mainly Navier-Stokes, MeshUpdate, FreeSurfaceSolver - I am not using the StructuredMeshMapper since this is an old project which was started earlier and I didn't wanted to switch in the middle),
b) I restart from this, reset the Mesh Velocities to zero at the beginning, execute MeshUpdate Solver and calculate a Temperature Steady state,
c) I restart from b) and want to run the Temperature Solver as a transient simulation (only temperature, so no changes in velocities or geometry) after executing MeshUpdate.

In c) I run the MeshUpdate Solver as very first Solver "Before Simulation" and obtain infinite Mesh Velocities (in x, y and z). I tried to put them zero in the Initial Conditions and even as a Body Force, I tried with and without "Restart Before Initial Conditions=True". Whatever I do, MeshUpdate Solver leads to infinite MeshVelocity values and then my Temperature Solver crashes because of this. If I skip the MeshUpdate, the temperature field is fine.
Also all variables are correctly read during the restart and the MeshVelocities are 0 at the end of simulation b).

Does anyone have an idea why I get these infinite values? And what should they be? Zero as in simulation b) or some real value?

Thanks a lot,
Martina
Martina
 
Posts: 39
Joined: 17 Apr 2013, 14:06

Re: infinite MeshVelocities

Postby tzwinger » 24 Feb 2014, 12:58

Martina,
if you anyhow do not need the Mesh Velocity, why don't you set:
Code: Select all
Compute Mesh Velocity = Logical False

?
Why does it produce non-sense?:
In the code it reads as follows:
Code: Select all
MeshVelocity = ( MeshUpdate - Solver % Variable % PrevValues(:,1) ) / dt

My suspicion is, that you have dt=0 in case of steady state run.
Regards,

Thomas
tzwinger
Site Admin
 
Posts: 79
Joined: 24 Aug 2009, 12:20

Re: infinite MeshVelocities

Postby Martina » 24 Feb 2014, 13:25

Thanks for the quick reply.

Run c) is a transient simulation, so dt shouldn't be zero, besides if Elmer is using the value of dt from the simulation b) (which is steady-state) since I execute the Solver "Before Simulation"?

Code: Select all
Compute Mesh Velocity = Logical False

The problem is, that I do not know if I need them or not. The conclusion of some older discussions here in the forum was that they should be zero in run b), but I'm unsure about the situation in run c)!

Martina
Last edited by Martina on 24 Feb 2014, 15:18, edited 1 time in total.
Martina
 
Posts: 39
Joined: 17 Apr 2013, 14:06

Re: infinite MeshVelocities

Postby tzwinger » 24 Feb 2014, 13:48

I think, by keeping the geometry constant in c), you also should omit any ALE related corrections, i.e., keep Mesh Velocity at zero.

-Thomas
tzwinger
Site Admin
 
Posts: 79
Joined: 24 Aug 2009, 12:20

Re: infinite MeshVelocities

Postby raback » 24 Feb 2014, 14:07

Hi

My guess is that the "before simulation" slot does not itilialize the "dt" at all. After all what would the timestep be? It makes sense only within the timestepping loop.

-Peter
raback
Site Admin
 
Posts: 3110
Joined: 22 Aug 2009, 11:57
Location: Espoo, Finland

Re: infinite MeshVelocities

Postby Martina » 24 Feb 2014, 15:17

Ok, thanks Thomas and Peter.
I'm using
Code: Select all
Compute Mesh Velocity = Logical False
now and simulations do look ok!
Martina
Martina
 
Posts: 39
Joined: 17 Apr 2013, 14:06


Return to Elmer/Ice

Who is online

Users browsing this forum: No registered users and 1 guest