Hi all,
There seem to be an error in the mfix_eb_hopper.cpp file, as compared to what the 2022 MfiX exa paper (and trigonometry) reports (see https://journals.sagepub.com/doi/full/10.1177/10943420211009293).
l 70 normal[1] = -orifice_radius; → normal[1] = orifice_radius - funnel_radius;
Anna
I don’t know why this is true… but it makes no difference. This is volfrac (ie EB covered volume) from two plot files; left is your suggested change and right is develop. Maybe @rangarad can explain exactly why this isn’t a bug, but from a user perspective I don’t think it is.
Hi and thanks for looking into this 
If you’re using the 07 benchmark, then it makes no difference because funnel_radius is twice orifice_radius (it is true of both Size0001 and Size0008). So, in these cases it works. We noticed the problem when we started using that “hopper” geometry with other values.
mfix_eb_hopper.cpp seems to create the hopper a bit differently than what the pseudo-code in the paper mentions. It does a 90 degree rotation at the end to align to the correct axis. Is there an set of inputs for the hopper where you are seeing it produce a geometry different from what you are expecting? Could you please share that case? Thanks.
Yes we did. I am attaching the input file if you want to test it, and a screenshot of the results without modifs (the green cone) vs with modifs (the dark hopper shape, what we want to obtain). I don’t think the rotation affects that computation.
inputs.dat (5.6 KB)
@AMJF irrespective of this issue, just an FYI, this and all benchmarks are just “dummy” or toy problems. Personally, I build geometry as a csg for anything that’s not a box or cylinder.
Thanks for sharing the inputs for when the geometry generated is not as expected. I will create a merge request with the fix.
