[CIG-LONG] Gale 2.0 intermittent bug?

[Walter Landry]

George Hilley <hilley at stanford.edu> wrote:
> Hi Walter,
> 
> I just noticed they are one and the same in Gale 2.0.  Sorry about that.
> 
> I attach a gzipped file that contains the output for the last two
> time steps that were registered by the model.  I took a quick look
> at the velocities and strain rate fields, but did not notice
> anything strange in this case.

If you look at the inflow and outflow sections of the grid, you can
see that the mesh is getting strongly distorted.  I am attaching a
screenshot with the vertical axis exaggerated by 10.  On the outflow
section, this is happening because the boundary conditions are pulling
faster than the rest of the simulation is pushing.  This leads to
thinning.  But the boundary has its height fixed, leading to severe
mesh distortion at the boundary.

So the solution to this is to let the height of the outflow boundary
move.  That should work, but there seems to be a bug preventing this.
I will fix this.

The second problem is in the inflow section.  The inflowing material
is entraining material next to it that is not moving.  This also
causes a thinning.  The solution to this would be to have the boundary
actually move.  This is not implemented, but it should be.

So, I have one bug to fix and one minor feature to implement, and then
your model should work.

Cheers,
Walter Landry

-------------- next part --------------
A non-text attachment was scrubbed...
Name: boundary.png
Type: image/png
Size: 117648 bytes
Desc: not available
Url : http://geodynamics.org/pipermail/cig-long/attachments/20120414/c4476bb8/attachment-0001.png