[CIG-LONG] Dike hill problem

[Walter Landry]

Taichi SATO <taichix at aori.u-tokyo.ac.jp> wrote:
> Dear walter,
> 
> Thank you very much!
> I attached recent my input file. It may be useful for solving the
> problem.

Sorry this has taken so long.  It turns out that the way MeshShapeVC
is used is not the best.  This is my fault, and I apologize.  In the
past, I thought that the area that MeshShapeVC affects is constant as
the grid distorts.  It turns out that MeshShapeVC chooses the points
at the beginning and never changes.  So as the surface moves up and
down, your divergence region expands and compresses.

The way to solve this is to put in an 'air' layer and fix the top of the
simulation.  Air layers often cause problems when there is subduction,
because air can get advected underground and spuriously lubricate the
interface between the plates.  I do not think that is an issue for
this simulation.

However, you will see some sub-grid behaviour that seems odd.  For
example, you will see small structures that should collapse.  A small
part of this is because the viscosity of the 'air' is not as close to
zero as it should.  But most of it is because of the finite resolution
of the grid.  As you increase the resolution, the size of the features
gets smaller.

Also, because the air layer has a fairly low viscosity, the time step
needs to be relatively small.  You can try modifying the viscosity of
the air to see how large you can make it.

I am attaching a copy of the new input file.  Let me know if this
works for you.

Cheers,
Walter Landry
walter at geodynamics.org

-------------- next part --------------
A non-text attachment was scrubbed...
Name: magma-tecto_new1_taichi.xml
Type: text/xml
Size: 37107 bytes
Desc: not available
Url : http://geodynamics.org/pipermail/cig-long/attachments/20100531/eb88fa37/attachment-0001.bin