Rayleigh Taylor instability

Results from RT experiments run with different parameters

In these experiments, we change various parameters and study the effects in resultant density. We use both the Ramses solver (ramses/mhd_eos from the Dispatch-Bifrost repository) and Bifrost (ideal_mhd from the Bifrost <https://github.com/ITA-Solar/Bifrost>_ repository) to simulate the experiment. Simulation snapshot are taken after 5 seconds, and at this time we can compare results from different runs.

Initially, experimental parameters follow Abel. T (2011)

Experiment Parameters

Quantity

Value

y0

0.5

d1

1

d2

2

deltay

0.025

deltavy

0.025

dymin

0.3

dymax

0.7

with gas constant gamma = 1.4

Ramses

First, we study the Ramses solver

Varying Ramses slope type

slope_type is a parameter for the ramses solver. We tried the different values 1, 2, and 3.5

The following simulations are run on grids with resolution 128x256 in x, y.

_images/density_slope_1_50.png

slope_type = 1.0
_images/density_slope_2_50.png

slope_type = 2.0
_images/density_run3_50.png

slope_type = 3.5

Some details in the results are clearly dependent on the slope_type variable. Larger values of this parameter gives the results finer/smaller scale patterns. Larger slope_type also introduces more non-linearity, and the flow becomes more turbulent.

Effects of grid in Ramses

Falling back to the default slope_type=3.5, we study how grid resolution affects results.

_images/density_ramses_small_50.png

nx=64, ny=128
_images/density_run3_50.png

nx=128, ny=256
_images/density_ramses_medium_20.png

nx=256, ny=512
_images/density_ramses_large_20.png

nx=512, ny=1024

Finer resolution gives more details to the results, and patterns exists on smaller scales. It also makes the density profiles less symmetric and more chaotic.

Bifrost

Now, we move on to the Bifrost solver. We have the same intial conditions, and the Bifrost parameters are by default

mhd_template['nu1'] = 0.1
mhd_template['nu2'] = 0.1
mhd_template['nu3'] = 0.5
mhd_template['nu_r'] = 0.3

The default grid is nx=128, ny=256.

Effects of grid in Bifrost

_images/rho_bifrost_64x128.png

nx=64, ny=128
_images/rho_bifrost_128x256.png

nx=128, ny=256
_images/rho_bifrost_256x512.png

nx=256, ny=512
_images/rho_bifrost_512x1024.png

nx=512, ny=1024

With Bifrost, the solution does not posess the same details as the Ramses solver. There are less whirls and non-linearity in these solutions.

Varying Bifrost parameters

Changing the Bifrost parameters one at a time while other parameters have the default values.

_images/rho_bifrost_nu1_02.png

nu1 = 0.1
_images/rho_bifrost_nu2_02.png

nu2 = 0.1
_images/rho_bifrost_nu3_03.png

nu3 = 0.3
_images/rho_bifrost_nu_r_01.png

nu_r = 0.1
_images/rho_bifrost_nu_r_03.png

nu_r = 0.3
Nothinghere

Placeholder just so the figures render

Changes to the Rayleigh Taylor scripts

I made some changes to the make_snap.py, such that the intial state has the correct symmetry. The np.roll function is used on pz, shifting the array to the right such that the two fingers are side by side. I did not fix the boundaries dymin and dymax.

In make_plot.py, it is possible to plot time evolution with the make_plot() function, just uncomment the lines at the bottom of the document.