[CIG-LONG] Static vs. Wrap Top, Hydrostatic Top BC, Static Left Boundary

[Walter Landry]

Karen Paczkowski <karen.paczkowski at yale.edu> wrote:
> Hi,
> 
> I am having difficulty with the static vs. wrap top boundary setting,
> the hydrostatic top BC, and a static side boundary setting.  I am
> trying to simulate a vertical subduction zone, so that the velocity
> along the left edge is Vy = -Vslab and Vx=0.  I would like to keep the
> top of the simulation flat, so Vy=0 and the left and bottom just a
> hydrostatic pressure condition (as to simulate an infinite domain).

If you set Vy=0 on the top, then that will, by itself, force the top
to be flat.

As for your other questions:

1) Hydrostatic BC

   If you add a HydrostaticTerm, then you should always add a
   hydrostatic BC.  This is independent of whether you use static or
   wrap top.

2) Static Top

   You would generally use this only if you have something like an air
   layer.  That is, you have a material on top with negligible mass.
   This is so that you do not have to simulate the entire atmosphere,
   which does not affect your simulation anyway.

3) Wrap Top

   This is what you usually want to use on the top.

In your case, you are setting Vy=0 on the top.  That should give
identical results for static top and wrap top, except at the left
side, where Vy=-Vslab.  In fact, since you are specifying Vy=0, you do
not need to specify either static top or wrap top.

My guess is that what you really want is a static left as well.  That
will fix the top and bottom left corners.

Also, I noticed that, in your input file, you specified the left BC
after the top BC.  The corner is set by whatever comes last, so it
seems that you added a fixedCornerShape to set it back to zero.  You
may want to switch that.  I am attaching a new input file that does
what I think you want and a screenshot after 20 steps.

Note if you set Vy=0 on the top and Vy=-Vslab on the left, then you
will have a step function in the velocity at the corner.  To solve
Stokes flow, we have to take a derivative of the velocity, resulting
in a delta function.

Numerical codes do not like delta functions.  It may not be entirely
fatal.  The Geomod 2004 shortening experiment, for example, has a step
function at the corner.  But it does make it difficult to get reliable
results.  If you have an analytic solution for the corner flow, it may
help to specify that directly.

Cheers,
Walter Landry
walter at geodynamics.org
-------------- next part --------------
A non-text attachment was scrubbed...
Name: SideSlab_HydrostaticStaticTop.xml
Type: text/xml
Size: 21403 bytes
Desc: not available
Url : http://geodynamics.org/pipermail/cig-long/attachments/20100716/109795f9/attachment-0001.bin 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Side_Slab.png
Type: image/png
Size: 74918 bytes
Desc: not available
Url : http://geodynamics.org/pipermail/cig-long/attachments/20100716/109795f9/attachment-0001.png