CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CFX (http://www.cfd-online.com/Forums/cfx/)
-   -   Mesh Deformation (http://www.cfd-online.com/Forums/cfx/26486-mesh-deformation.html)

 Ed October 7, 2008 10:43

Mesh Deformation

Hey all,

I'm trying to set up a simulation in which I have 1 fluid domain and a wall is pivoting inside of the domain. The problem is similar to Tutorial 20: Fluid Structure Interaction and Mesh Deformation, except instead of a ball I have a wall that pivots 90 degrees.

My problem is that after a small rotation, the run fails. This is because the wall pulls the mesh and elements become highly distorted and have negative volumes.

Is there a way to remesh my fluid so that the mesh moves around the wall rather than with it? It wont let me use mesh adaptation because the problem is transient, and it needs to be transient to get the mesh deformation (if there is a way around this, that would be nice).

My pivot motion is defined using CEL in which the wall rotates 0.675 degrees for each substep of .0075 seconds, until it ends at 90 degrees at 1 second.

Also, I cannot rotate the fluid around the wall because it pivots toward another stationary wall and I need to see how the fluid reacts as the 2 walls converge (there is space for it to move between the pivoting wall and the boundary wall).

Thanks, Ed M.

 Glenn Horrocks October 7, 2008 20:23

Re: Mesh Deformation

Hi,

Motions of this magnitude are difficult to do using mesh motion. I have two recommendations:

1) You can stop the simulation after a small angle (maybe 20 degrees) then start a new simulation with the geometry at that angle and a nice mesh and interpolate the results onto it.

2) If the motion is only a rotation you can put the rotating bit in a rotating frame of reference and connect it to the remainder of the domain with a transient rotor/stator GGI. The approach, if applicable will be the most efficient as you don't need to use moving mesh and that's a considerable saving. Rotating frames of reference are much simpler than full moving mesh.

Glenn Horrocks

 All times are GMT -4. The time now is 00:10.