Closed abarret closed 6 months ago
I've done one test with this so far. There seems to be an issue with the solver when I use input2d.cf.relax. Using the input2d.cf.relax file causes the run to be terminated at t = 0.01 with the following error:
P=00000:ERROR MESSAGE:
P=00000:Internal RefineSchedule<DIM> error...
P=00000: In finishScheduleConstruction() --
P=00000: No coarser levels...could not fill from coarser.
The logs at this time step indicate:
change in domain volume detected (volume is computed by summing control volumes)
old volume = 1.0006
new volume = 1
this may indicate overlapping patches in the AMR grid hierarchy.
And the AMR mesh at t = 0.01 is shown below, which is different from the results I get from the cf_frm branch for the same inputs:
Uhoh. What's going on with that domain generation? I'll take a look.
@bindi-nagda it should be fixed now.
Now the solver runs for longer, but I think the mesh generation still needs some investigation. The first image below is from this branch while the second image is from the cf_frm branch and the mesh in the second one looks better. Also, the AMR mesh doesn't seem to capture the refinement criteria, which is more apparent at later times. For example, the AMR mesh at the center of the "blob" takes the coarsest value (see image 3).
P.S: I'm traveling so I haven't been able to properly look at the code or do more testing but I will try to get to it on Thursday,
Results from new code at t =0.25;
Result from cf_frm at t=0.25:
Result from new code at t=0.296:
And another thing, the option output_conformation_tensor
is currently broken. You should set this to false
in the input file for now.
These are new changes to get the multiphase code working with IBAMR master. This is the last set of changes needed before
cf_frm
can be merged intomain
. Note that you must use a version of IBAMR from after February 9.@bindi-nagda when you get a chance, can you check that the complex fluid code still produces reasonable values? I only checked that the code runs.