[aspect-devel] Couette flow solution

[FELIPE ORELLANA ROVIROSA]

 Hi all,

   I have a question on Aspect Stokes solver.

   I am simulating something that includes a Couette flow, where my model
box is a control volume and the top boundary (plate) moves creating a
velocity boundary layer by viscous dragging.
   Velocity BCs: I am keeping the front (y=0) and back (y=8e5) walls as
tangential, the bottom (z=0) fixed, and prescribing a velocity on top
(z=5e5, where V= 0.1 m/yr ; 0 ; 0). My experiment: I am NOT writing
anything about the left and right walls (upstream receiver (x=0) and
downstream window (x=1e6) of the flow directed by the X velocity BC on top.

   What I get are fluid velocities ~2 orders of magnitude above the top BC
imposed velocity, and what happens at both the left and right walls can be
interpreted as local mass conservation: the upper halves has an inflow, the
lower halves an outflow, they vary continuously having a zero in the middle.

  What is Aspect doing here? correct me if I am wrong, as I am trying to
understand:

   Because I did not provide a BC on those walls, the system is
UNDER-determined? Then, it solves the unknowns via a compensated solution
containing inflow and outflow along the X coordinate (and 0 s on the other
coordinates)?

   But why does not Aspect compensates the inflow /outflow between the
walls?, say inflow at left, outflow at right, and instead it does
inflow/outflow on each wall.

   Is there any way to instead, provide BCs for pressure (p), plus Vy=0,
Vz=0, and let Aspect find Vx on those walls (and on the interior)? or maybe
Aspect just can not compute variables on the boundaries..

thanks,

Felipe
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.geodynamics.org/pipermail/aspect-devel/attachments/20170105/ba245b1a/attachment.html>