[CIG-LONG] Gale Boundary Conditions & Output

[Walter Landry]

Rachel Headley <rheadley at uw.edu> wrote:
> I'm attempting to model glacial flow which is a nonlinear rheology for
> a tilted block. Right now, I'm using the power law in Gale, but I have
> essentially turned off the temperature component by setting T_0 = 0. I
> have n = 3, and the rest of the components set for a glacier (except
> all variables/constants that are part of the temperature conditions,
> which I have left as defaults since I think they shouldn't matter;
> eventually, I would like to re-incorporate a temperature
> dependence). I have not scaled anything, as it seems to converge
> without the scaling. I am also only running for one time step, as I
> just want the steady state solution without any further deformation. I
> will be the first to admit that this might not be the most elegant way
> to solve this specific problem.
> 
> My velocity boundary conditions are as follows:
> Left: vx = 30 m/yr, vy = 0 (set up as a flux boundary, not a moving
> boundary)
> Bottom: vx = vy = vz = 0
> Top: vy = 0
> Back & Front: vz = 0

FYI, since you are only running for 1 time step, it does not matter if
you have flux boundaries or moving boundaries.

> When I use the specific boundaries, the flow follows how I might
> expect. However, when I set no conditions on Top, Back, and Front, the
> flow shows velocity vectors normal to these boundaries when close to
> Left and mostly negligible at other locations. This behavior surprises
> me, and I'm not sure the best way to work beyond it without setting
> the boundary conditions (Top, Back, and Front) that I have set, which
> feels a bit kloogey to me. Any suggestions for the best way to
> approach this?

I do not quite understand what is happening.  If you do not set
boundary conditions on a wall, that is equivalent to a free surface.
If it is on the side, then I would expect the material to flow out.

> I have a few questions about the boundary conditions.
> 1) What does it mean if I do not set velocity boundary conditions
> (i.e. for the right side)?
> 
> 2) Is there a simple way to set the boundaries as a free surface?

If you do not set vx for the right side, then it is a free surface and
material will flow through that boundary.  This would be appropriate
if you want to look at how a blob of ice deforms under gravity.  Is
that what you are trying to do?

> I've attached my xml file that I'm using and also a PDF showing the
> comparison between the Gale output and Comsol. I have set up a similar
> problem in Comsol for steady flow, except I'm using an approximation
> for stokes flow (Pattyn, JGR, 2003) to solve for ux and uz, then I use
> conservation of mass to find the vertical velocity (uy). I also have
> the left boundary with an input velocity of 30 m/yr, and all
> velocities are 0 at the base. The other sides are free surfaces.

I could not run the input file.  I need rampB.txt.

> Also, as I'm attempting to do gauge the performance of Gale versus
> other models, I'm interested in the run time and so on. First off,
> what information does the file timeinfo.*.dat hold? Is it run time in
> seconds or something else?

It is time in the simulation and the difference in time from the
previous step.

> Finally, and perhaps this is more from my inexperience using the
> command prompt than anything, but is there a simple way to save the
> verbose output as a text file?

  ./Gale g2c.xml | tee gale_out

will output to both the screen and the file "gale_out".  If you do not
want the screen output, you can use

  ./Gale g2c.xml > gale_out

Cheers,
Walter Landry