From axi-symmetric case to 3D

Mesh generators, CAD programs, and other tools
Post Reply
jarij
Posts: 5
Joined: 15 Apr 2011, 10:28
Antispam: Yes

From axi-symmetric case to 3D

Post by jarij » 15 Apr 2011, 10:48

Hi Elmer Team/Fans,

I have nice Elmer results for axi-symmetric liquid flow in coupled Navier-Stokes and heat equation system. I'm proceeding to rotationally symmetric 3D from the axi-symmetric case, and was wondering, what kinds of tools I could have in Elmer to use my axi-symmetric results as initial guess in 3D. I'm able to create a mesh and .sif in/for 3D, but the challenge is to rotate the axi-symmetric results for 3D. Would the tools be available in Elmer Solver or Elmer Post, or should I go directly to self-written subroutine?

With kind regards,

Jari

raback
Site Admin
Posts: 3264
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: From axi-symmetric case to 3D

Post by raback » 15 Apr 2011, 22:00

Hi Jari,

and welcome to the forum.

For getting data from 2D to 3D there is no existing tool. What we have done in a similar case (coupling between global 2D and local 3D) is to save the temperature BCs in the global model and then use them in the local model.

In the open source version this can be done using SaveLine to save say [x,T] dependence. If you prescribe the variables to be saved explicitely you can get just these two variables.

The next step is then to use this in BC of the 3D case. You should probably introduce a dummy variable, say Radius (using Exported Variables) and set for this initial condition

Code: Select all

Radius = Variable Coordinate
  Real MATC "sqrt(tx(0)^2+tx(2)^2)"
Assuming that axis of symmetry is y-coordinate. Then you can give the 3D condition as

Code: Select all

Temperature = Variable Radius
  Real
    include "myline.dat"
  End
Or something like that, you might check the syntax.

One may argue what is the best way to transfer BCs, Dirichlet, Neumann or Robin...

When you then start the 3D simulation you could use long timestep to start with and in order to get a nice initial guess you may say

Code: Select all

Viscosity = Variable "timestep size"
  Real MATC "c*tx"
where c should be such that for the shortest timestep you obtain the true viscosity.

-Peter

-Peter

Post Reply