Rad Feedback from Sink Particles Update
This last week I spent testing the feedback code. To do this, I started with a sink particle at the center of the grid and prescribed some energy to be injected into the grid each timestep. Here are some results.
Run 1.
Final time - 1e-4
Energy injection rate - 28052.48 (chosen so that by the end of the simulation a total of 2 Erad will be injected into the grid, where Erad is the total Erad of the simulation at t=0)
Opacities, both set to 1d-6.
This run produced the following graph of Erad vs t, verifying the code is behaving as expected:
That is, we recover the energy injection rate, as given by the 'predicted' curve.
Here is how the radiative, thermal, and total energy behave over time.
A pseudo-color plot of Erad over time seems a little odd though, and I am trying to understand the behavior here. As this movie shows, the min and max remain very close over the simulation (a query shows they only differ in their 7th decimal place), and that by frame 1 there is already a large sphere on the grid. The legend isn't fixed (if it were, would be hard to capture the sphere at all), and so the breathing may be an artifact of the plotting. Why is the min/max so close like this?
I then lowered the final simulation time while keeping the energy injection rate the same. Here is a psuedo plot movie of that, but I see the same behavior. Perhaps the rate is small compared to the diffusion rate so that it all smoothes out within a timestep? Need to look at the different rates in more detail.
In the meantime, I kept the final simulation time small but increased the energy injection rate to play with different opacities. These results I will go through next.
Run 2.
tfinal - 1e-9
energy injection rate - 5,610,800,000 (such that 4*Erad is injected over course of simulation, where Erad is initial total Erad in the grid)
opacities - both set to 1d-6
So as you can see, AstroBEAR is losing some of the energy (supposedly the rad transfer routines aren't conservative??) over the course of the simulation, but performing quite well still.
For this case, the pseudo plot looks quite nice. It is shown in my previous blog post under 'Run 1'. I made a movie that shows how the q variables behave over time for a radial cut of the data. As the pseudo plot suggests, seeing little coupling between the gas and the radiation. For the next run, I will increase the roseland opacity to try and confine the radiation to a small region around the sink particle, but not change the planck opacity so that the gas will remain uneffected by the radiation.
Run 3.
tfinal - 1e-9
energy injection rate - 5,610,800,000
opacities - roseland=1d-3, planck=1d-6
The q-variables are shown in this movie here. For kicks, here are some images of the kernel:
All looks as expected.
Run 4.
Now am trying to couple the gas to the radiation, so boosted the Planck opacity. I found it had to be increased many orders above the Roseland so that it would couple.
tfinal - 1e-9
energy injection rate - 5,610,800,000
opacities - roseland=1d-3, planck=1000
Here is a movie of the q-variables again. And a movie of just Erad here. The pressure, temperature, and total energy grow at the same rate, but density remains flat. I am curious why density is not being affected at all.
Attachments (14)
- EthermVsTime.png (18.1 KB) - added by 9 years ago.
- EVsTime.png (17.3 KB) - added by 9 years ago.
- EradVsTime.png (16.3 KB) - added by 9 years ago.
- movie1.gif (5.8 MB) - added by 9 years ago.
- rates.png (22.8 KB) - added by 9 years ago.
- rates_lowlow.png (21.3 KB) - added by 9 years ago.
- rates_orig.png (20.4 KB) - added by 9 years ago.
- qvstime_lowlow.gif (1.3 MB) - added by 9 years ago.
- qvstime_highlow.gif (1.3 MB) - added by 9 years ago.
- kernel.png (17.6 KB) - added by 9 years ago.
- kernel_lineout.png (21.8 KB) - added by 9 years ago.
- erad_highhigh.gif (1.1 MB) - added by 9 years ago.
- qvstime_highhigh.gif (1.3 MB) - added by 9 years ago.
- erad_smalltime.gif (1.3 MB) - added by 9 years ago.
Comments
No comments.