clarifications on resolution for my 3D cooling jet runs

Here is an updated table from my previous blog post. Again, run A was not submitted to bluegene since it requires a lot of resolution. The required x res is the fixed resolution that I would need to guarantee 10 cells per cooling length.

run n jet eta chi required x res base grid levels actual cells per cooling length cells per jet radius
A 0.1 0.01 0.0015 78,335 … … … …
B 1 0.1 0.034 3495 56 x 28 x 28 6 10.15 298.67
C 10 1 0.358 335 42 x 21 x 21 3 10.02 28
D 100 10 1.367 88 44 x 22 x 22 1 10.02 7.33
E 1000 100 2.428 50 50 x 25 x 25 0 10.12 4.17
… … … … … … … … …
C' 10 1 0.358 335 48 x 24 x 24 3 11.46 32
D' 100 10 1.367 88 48 x 24 x 24 3 43.74 32
E' 1000 100 2.428 50 48 x 24 x 24 3 77.70 32

I also figured out that if I want at least 32 cells per jet radius, then I need a minimum fixed x resolution of 348 which corresponds to a base grid of 48 x 24 x 24 with 3 levels of AMR. Unfortunately, when I set up these runs I was only considering cooling length to set the resolution, so runs C, D, and E are below this minimum resolution based on jet radius although C is very close.

The primed runs are new ones that meet the minimum resolution requirement. This essentially puts "extra" resolution on the cooling length. Fortunately, the old runs were still waiting in the queue, so I updated their global.data files to have the higher resolution. So C, D, and E were successfully replaced by C', D', and E'. I'll have to take a look at the results when I get back from vacation.

Comments

No comments.