Context Navigation

← Previous Post
Next Post →

Dust in AstroBEAR - Update 2019/10/21

Current Objective

Debugging the dust drag advection routines (collisional drag)
Finishing the dust sputtering routine
Finishing the total mass calculation routine

Current Status

Drag Advection

(See also report from last week https://www.pas.rochester.edu/astrobear/blog/fschmidt10142019)

dqdt for the dust velocity is way too large (in the order of 1e42) which (apart from being quite obviously incorrect) causes an issue with the RK integrators
I found a couple of bugs already (e.g. there was an issue with the transformation into the dust grain rest frame) but it's still giving me values that are too big. The frame of reference is changed multiple times throughout the calculation so I suspect more errors in there somewhere. Currently still debugging.

Sputtering Routine

Started implementing the sputtering routines based on Dwek & Arendt (1992), Tielens et al. (1994), and Nozawa et al. (2006). This is a work in progress.

Total Mass Calculation

Did some more work on this internal processing routine which calculates the total mass of dust in each bin and across all the bins for each timestep (not just each output timestep) and stores them in a file. This additional dust-related output can be turned on or off and will mainly be used for 3D simulations (in case we don't get a new HPC award next year; our local facilities can't handle many AMR 3D Chombo output files. So using this output option would allow us to cut down on the Chombo output while still getting dust mass information with high temporal resolution). Still a work in progress but should be done soon-ish.

Questions

Feedback on Gas

Accelerating the dust should remove energy from the system and growing or destroying grains should either generate or remove gas mass from the system. I've been thinking about how best to implement that:

For the grain destruction or grain growth, the most obvious choice would be to simply add or remove material to/from the gas during the source step. So

\sum_{i=1}^{nDustBins+2} \left( \bigtriangleup m_{\textrm{dust evaporation}, i} - \bigtriangleup m_{\textrm{dust growth}, i} \right)= \bigtriangleup m_{\textrm{gas}}

For the gas drag the defining quantity that's calculated is the force F_{\textrm{dust}} the dust grains experience. F_{\textrm{dust}} is then used to calculate the acceleration that each grain experiences. My current approach is to set F_{\textrm{gas}} = - F_{\textrm{dust}}, where F_{\textrm{gas}} is the total force experienced by the gas particles in one cell. The force which a single gas particles experiences is then F_{\textrm{gas}}/n and that then results in a declaration. All of that is currently calculated in the source term routine alongside the dust acceleration. Questions: Will this result in energy conservation problems?

Stiffness of Equations

I'm very certain that the current issue with the RK integrator error is down to a bug and not down to an intrinsic issue with the stiffness of the drag equations (since by definition the acceleration of the dust grains cannot exceed the acceleration of the gas particles). However, it is, in theory, possible that the destruction routines are very stiff under certain conditions (very high densities, very high velocities etc.). Question: I was wondering what could be done in that case?

astrobear