Hello!
I am pretty sure I am doing something silly here, or I am not understanding syntax very well.
I have a model with an air-filled flux trap in the center. I want to calculate the flux in the flux trap. The preferred method would be a cylindrical mesh over the flux trap.
Comparing the same tally with a cylindrical and rectilinear mesh, I get substantially different flux values after normalization. Some of the cells are practically identical between the two meshes (or at least so I think) and yet I’m seeing significantly different flux values (units of source/cm^2-s). These two tallies are from the same calculation.
Find several plots attached, as well as my tally.out and tally.xml files.
Any help you can provide is appreciated!
Valerio
tallies.xml (1.2 KB)
Link to tallies.out: tallies.out
I’ve found this topic: About cylindrical meshes visulization - #11 by Artur
Which is releated to this bug fix: Fix bug in cylindrical and spherical meshes when grid boundary is coincident with cell boundary by paulromano · Pull Request #2439 · openmc-dev/openmc · GitHub
Although it seems like that was resolved and released, this problem looks very similar. I have tried using the collision estimator, and I get more similar values for flux to those of the rectilinear mesh. However, since the area in which I’m tallying is transparent to neutrons, the convergence is extra slow and it’s difficult to make final determinations.
If anyone has any extra feedback, it’d be greatly appreciated.
Thought I’d give some more info. I’ve re-run these by splitting the central flux trap into separate cells and tallied the flux using a cell-based filter. I’ve then compared the volume-normalized tallies for the rectilinear mesh, the cylindrical mesh, and the cell filter with cylindrical cells.
You can note how, at y=0, the flux for the cylindrical mesh is extremely different from the equivalent one obtained from the rectilinear mesh and cell-based filter. A 2-D heatmap of the cylindrical mesh tally shows the unphysical behavior for the last slice, close to x=0.
I have attached a minimal version of the .xml file that includes the three tallies that have been compared (20, 21, and 22).
tallies_minimal.xml (12.6 KB)
Hi!
Have you tried “autoscaling” or changing the min and max values in openmc plotter? Looks like the values in the rectilinear grid tally are above you maximum limit in the plotter. The results of the two tallies should definitely match each other – have you tried normalizing the tally results by dividing the scores of the tallies by the volumes of the mesh cells? In general, the sizes of the cells in your rectilinear mesh are bigger than the ones in the curvilinear mesh. At least, when scaled properly, the distributions should looks similar.
I can’t comment on whether you’ve found a bug or not.
Best, J
