MOGLI: Model for Multiphase Gas using Multifluid hydrodynamics
Hello there! 👋
We would like to introduce our recently published, MOGLI subgrid model for capturing multiphase gas dynamics!
Following the age-old tradition of contrived Astrophysical abbreviation, MOGLI stands for MOdel for Multiphase Gas using MulLIfluid hydrodynamics.
The preprint version of the paper is now available on arXiv at 2412.03751.
Multiphase gas, with hot (∼\(10^6\)K) and cold (∼\(10^4\)K) gas, is ubiquitous in astrophysical media across a wide range of scales. However, simulating multiphase gas has been a long-standing challenge, due to the large separation between the size of cold gas structures and the scales at which such gas impacts the evolution of associated systems.
This is where MOGLI comes into the picture. Implemented with multifluid AREPO, the model is based on first principles and theoretical results from previous studies with resolved small-scale simulations, leading to a minimal number of free parameters in the formulation.
Multifluid methods have long been used for engineering applications for water-air interactions, combustions, and meteorolgy.
In this scheme, fluid properties like density, momenta and energy of multiple fluid are tracked on the same grid and the different fluids are tracked via their volume fractions in each cell. We build upon the implementation of multifluid methods in AREPO, showcased in this paper by Weinberger & Hernquist (2023).
For example, with two fluids, each cell would contain information about density \((\rho)\), momenta \((\vec{p})\) and energy \((E)\) of each fluid, in addition to the volume fraction \((\alpha)\) of one of the fluids.
We use the theoretical understanding from previous studies which use small-scale resolved simluations as the foundation of MOGLI. We divide the interactions between the hot and cold gas fluid in the model into three sources: drag, turbulent mixing and cold gas growth.
We verify the different components of the framework through extensive comparison with benchmark single-fluid simulations across different simulation parameters, such as how resolved the cold gas is initially, the turbulent Mach number, spatial resolution, and random initialisation of turbulence.
We find that MOGLI can reproduce behaviour like the cold gas survival criteria as an emergent property, without it being explicitly included in the model.
Additionally, we demonstrate its capability by running a simulation which would be computationally prohibitive to run as a resolved single-fluid simulation, and following are two such simulations.
For more about the details and verification of the model check our paper, now on arXiv at 2412.03751.
MOGLI with 100 unresolved clouds
Here’s showcase an simulation which would required a resolution upwards of \(3000^3\) cells, if done with the tranditional resolved single-fluid method.
But, with MOGLI, we can get an equivalent simulation with just \(64^3\) cells.
We are simulating a turbulent box with \(64^3\) cells and 100 unresolved clouds of radius \(L_\mathrm{box}/256\):
Reduced MOGLI, and 1000 unresolved clouds
In this simulation, we are showing the evolution of a similar simulation setup as the previous section. But this time with a reduced MOGLI mode, i.e. cold gas growth is turned off. The cold gas is just destroyed via turbulent mixing and there are drag forces.
We are simulating a turbulent box with \(64^3\) cells and 1000 unresolved clouds of radius \(L_\mathrm{box}/256\):