Meeting Update 03.07

clump project

So I did some cooling shock problem using the clump settings, just removed the clumps. The shock front looked fine when hviscosity and periodic boundaries are used.
So the clump runs are being restarted from frame 0 using those settings. The following animations show Bx Strong, Bx Weak, By Strong, By Weak respectively. They are not finished yet, but are running. The final frame should be 60. So By runs are almost finished, Bx runs are half way through.
​Bx Strong

​Bx Weak

​By Strong

​By Weak

cloud collapse project

Eric and I thought it will be very interesting to do the BE collapse problem as mentioned in last week's journal club:

a paper about dynamo generations at various scales in the cloud collapse problem and the impact of a realistic prandtl number.

The runs will be similar to those of the Federrah 2011 paper, but put in physical viscosity, that is, Braginskii when gyrofrequency >> collision frequency, Spitzer when the opposite is true. The resistivity will be Spitzer, which we already have.

Based on these thoughts, I implemented Braginskii viscosity into the code. The numerical method is based on Parrish et al. 2012 paper:

​http://arxiv.org/abs/1201.0754

They did the HBI/MTI simulations with explicit Braginskii viscosity.

I then went on to test 2D Braginskii viscosity in a MHD KH instablitiy problem using James Stone's parameters:

​MHD Kelvin Helmholtz Instability with Braginskii type Viscosity
​MHD Kelvin Helmholtz Instability with Braginskii type Viscosity, Comparison

It is best to have the code switching between Spitzer and Braginskii viscosity based on the relation between local gyrofrequency and collision frequency. Unfortunately we do not have that yet. Not far away though.

code dev

A problem we discussed before: when considering the energy equation for the operator split induction equation, do we need to apply the induction flux to the energy equation explicitly as well? I now doubt to explicitly calculate the energy equation is the correct way. So I added some additional calculations to apply energy flux induced by multiphysical processes in both the braginskii viscosity and resistivity solvers.

I think we are at a point where sub-cycling is becoming a bottle-neck. Some thoughts are needed. The problem is:

How to do subcyclings in AMR without a significant increase in the size of ghost zones and at the same time maintain accuracy.

Comments

No comments.