[CIG-LONG] Crash Error in TimeIntegrand

[Walter Landry]

María Helga Guðmundsdóttir <mariahg at stanford.edu> wrote:
> Hi Walter,
> 
> This seems to have resolved the TimeIntegrand error issue – thanks!
> 
> However, I have another question that relates to one of the
> modifications you and George made to the model. The way the model
> ends are set up now, there is effectively no inflow of material,
> because the two ends of the model move at a rate equal to the
> velocity of the inflowing material. As a result, the models end up
> shortening more and more the longer they run, and with some of the
> smaller/shorter models, we practically run out of material in the
> late stages of the run. (Obviously you can't try to simulate 70 km
> of displacement on a 60 km model if the model is going to shorten by
> the full amount of the displacement.)
> 
> George and I experimented a bit with the Front/BackEquations, but we
> were unable to modify them to allow the model to actually retain its
> dimensions or change them more slowly; we always got the "wall is
> moving in the wrong direction" error. Is there any workaround that
> will allow us to keep the models at a constant size, but have
> material flow in and out of the two ends? Or should we try something
> else, like just defining the models to be 70+ km longer than we need
> them to be at the start, and allow that material to be gradually
> lost as the simulation progresses?

I would expect that setting the front and back equations to a constant
(so get rid of the time component) would fix them, letting material
flow in and out.

However, that is what you had originally, and that was causing a lot
of distortion.  I think a better route is to make the model 70 km
longer than the final result.

Cheers,
Walter Landry