Franklin Hi-res Run

This page outlines the 0.5cm (4 level) run on the NERSC machine, Franklin. The grid for this problem is 2048cm x 2048cm which allows for a large "buffer" region below the He4 interface. The average time per timestep is currently about 17.5 seconds on 384 processors. Perhaps this buffer is too large; once we are all set on Intrepid, I'll run a smaller domain run there.

Evolution of enucdot and tfromp

The image below shows the most recent plot of the evolution of the burning layer - the maximum in enucdot should occur at the location that has the highest density, temperature, and He4 mass fraction. The purpose of this image was to extrapolate an estimate as to how much longer until things become "interesting", i.e. nonlinear. It seems, however, that the burning rate is leveling off at about 8.0e12 erg/g/s - note that this is qualitatively similar to what is in the coarse (10cm) resolution image where enucdot initially peaks and then levels off for an extended amount of time before taking off in the "interesting" regime. Unfortunately, I think it is too early to say anything about when we will go nonlinear.

Hi-res	Low-res

Because this data was generated before I added the runtime diagnostics to the XRB test problem, the max{enucdot} information above was post-processed using the fextrema.f90 routine in fParallel/data_processing/ which defaults to scan the finest level data for the extrema in a plotfile. However, using the tfromp or tfromh data from that routine would give us the max temperature which occurs at the bottom of the domain, whereas we are interested in the maximum temperature in the burning layer. The runtime diagnostic appreciates this fact, but for post-processing I needed to do something clever with raw data in VisIt. Below is an image of the evolution of the max tfromp in the burning layer - it closely follows the trend of enucdot as expected.

Hi-res	Low-res

Carbon Mass Fraction

Snapshots of the data are also not very useful at this time. One really has to look at the data on the finest level to see much evolution - this makes for a very tiny patch of the interface. fsnapshots2d.f90, when compiled with my version of gfortran, does not generate an image file (or an error) for some reason. I'm currently looking into this, but for now I have some logscale plots generated with VisIt.

0.56 ms

1.13 ms

1.79 ms

2.52 ms

3.29 ms

4.09 ms

4.88 ms