[CIG-LONG] Corner Flow Velocity unstable at initial condition

[mspieg]

Hi Karen,
    When you say time step, do you mean iterations of the uzawa  
solver?  Stokes flow should be time-independent but the standard  
uzawa solver used in Gale iterates for both pressure and velocity  
(where the pressure field is used to enforce the zero divergence  
constraint).  You should also show your pressure fields and some norm  
of the residual of the full V,P problem.  I suspect the problem is  
with the pressure.

Also,  if you are specifying velocities on all boundaries, Stokes  is  
actually singular which will lead to an inconsistent solution if the  
boundary conditions are not exactly divergence free.  If possible you  
should also calculate the divergence of the velocity field over the  
domain.

just some thoughts
marc


On Feb 25, 2011, at 3:21 PM, Karen Paczkowski wrote:

> Hi,
>
> I have been trying to write a benchmark code for a simple corner  
> flow model using GALE.  For a 90 degree corner I have found that by  
> adding a small velocity gradient near the corner and by using the  
> analytic solution as boundary conditions the velocity field  
> converges to the analytic solution.  However, this is only true of  
> the velocity field at the 1st time step.  After that the numerical  
> answer seems to go unstable, producing internal velocities a couple  
> orders of magnitude higher than any of the boundary conditions.
>
> Has anyone encountered anything like this?
>
> What might be happening?
>
>
> I have attached my .xml file and the File1, File2, File3 and File4  
> files used as the input for the analytic solution boundary conditions.
>
> The top BC is vx = 1 and vy =0, the left is vx = 0 and vy = -1, the  
> boundary conditions for the right and bottom use the input files to  
> give them the values of the analytic solution.  The velocity  
> gradient in the top left corner spans the length of 10% of the box  
> size along the top and left side.
>
> I have also attached pictures of the velocity field at the 1st time  
> step and then at the second time step.  The background color is  
> velocity magnitude.
>
> Thank you,
> Karen
>
>>
>>
> <Initial.png>
> <AfterInitial.png>
> <CornerFlow_2-25-2011.xml>
> <vxBottomCornerFlow>
> <vxRightCornerFlow>
> <vyBottomCornerFlow>
> <vyRightCornerFlow>
>
> _______________________________________________
> CIG-LONG mailing list
> CIG-LONG at geodynamics.org
> http://geodynamics.org/cgi-bin/mailman/listinfo/cig-long

----------------------------------------------------
Marc Spiegelman
Lamont-Doherty Earth Observatory
Dept. of Applied Physics/Applied Math
Columbia University
http://www.ldeo.columbia.edu/~mspieg
tel: 845 704 2323 (SkypeIn)
----------------------------------------------------


-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://geodynamics.org/pipermail/cig-long/attachments/20110225/5854483d/attachment.htm