correction on cooling length resolution in pulsed jet sims

So I actually made a mistake in my cooling length calculations… I was modeling a 100 km/s shock in my 1D code to get a feel for what the cooling length in my pulsed jet sims should be. I thought this was correct since the injection velocity varies from 150 - 250 km/s. However, the maximum shock velocity is actually half of that, so 50 km/s.

By comparison, a 50 km/s shock will have a smaller cooling length, so this is actually good news. It means that my cooling length resolution is better than I thought.

Below is an updated table of the resolutions used from other papers.

Paper Base Grid AMR Physical Size [AU] Jet Radius [AU] Lcool [AU] Cells/AU Cells/Rj Cells/Lcool
Mine (2013) 42 x 420 7 2005 x 20054 334 7.29 2.68 896 19.53
de Colle (2006) 180 x 1800 0 2005 x 20054 334 7.29 0.09 30 0.66
Raga (2007) 16 x 64 10 1671 x 6685 134 10 9.80 1314 98.05
Tesileanu (2012) 128 x 384 6 400 x 1200 20 0.46 40.96 819.2 18.81

This also has implications for the 3D runs that we want to do. The extremely high cpu hour estimates are for higher cooling length resolutions which we may not need. Below is an updated table of the final results I had previously posted.

Base Grid Levels Cells/Lcool SUs (millions)
84 x 420 x 84 7 19.53 645.0
84 x 420 x 84 6 9.77 116.6
84 x 420 x 84 5 4.88 19.79
21 x 105 x 21 7 4.88 19.79
21 x 105 x 21 6 2.44 3.028
21 x 105 x 21 5 1.22 0.424

I tried a run on bluestreak with the parameters of the lowest resolution listed in the table above. I ran the model and got 12 frames in about 18.55 hours on 4096 cores. That is approximately 76,000 SUs.

Comments

No comments.