[FLASH-USERS] Problem forcing derefinement in AMR routine

[Nicholas Omahen]

Dear flash users,

I am having trouble forcing derefinement in my simulations using FLASH 4.5.

My simulation is setup as follows: I am embedding a binary system that is
orbiting in a "wind-tunnel" domain, such that the binary system is being
continuously fed gas. Each binary member is represented by an "accretor",
which has a boundary within which mass is accreted. The boundary radius of
each binary member depends linearly on their mass, and in these simulations
I am using disparate mass ratios  for my binary components (down to a ratio
of q ~ 0.01).

I would like to refine the system such that each accretor boundary has the
same amount of cells (~10) across its radius. As their boundary radii will
be up to a factor of 100 different, I want to refine the lower mass
(smaller radius) accretor at a higher level than the higher mass (larger
radius) accretor.

To do this, I implemented an AMR routine that is a modification of the
subroutine gr_markInRadius, where I use "lref1 = lrefine_max" as my maximum
refinement, and the quantity "lref2 = lrefine_max - lrefine_delta", which
is my "intermediate refinement". I select all cells within a radius
encompassing the whole binary system, exclude the cells near the lower-mass
component, and flag these for intermediate refinement. Then, for the
excluded cells near the lower-mass component, I flag these for higher
refinement. The goal is for the lower-mass secondary to be orbiting in a
"pocket" of maximal refinement.

So, usually, this works; in the snapshot I have uploaded here
<https://i.imgur.com/ZVJYbWh.png>, the lower-mass binary component is
embedded in a maximally refined region the follows along its orbit. The
rest of the region embedding the binary system is at the intermediate
refinement level. The problem is throughout the simulation, other regions
end up reaching the maximum refinement; see here
<https://i.imgur.com/OB8tenI.png>. The region trailing the lower-mass
secondary component did not derefine, and the higher-mass primary suddenly
becomes maximally refined. This is periodic, and these regions will
eventually derefine successfully.

I generally also have the code refining on internal energy, however when I
turn this off such that gr_markInRadius is the only refinement criterion,
the anomalous refinement still occurs.

Is there away I can force my simulation to derefine more reliably within
the regions I've defined?

Please let me know if there's any other information I should provide

Thanks,
Nick
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://flash.rochester.edu/pipermail/flash-users/attachments/20190605/d5ae9c12/attachment.htm>