PI: Matthew Bate

Project: High Performance Computing Support for Exeter Astrophysics

Our team has performed numerical simulations of Boussinesq convection using DIaL-3 and Cosma-8, considering 2D and 3D Cartesian box, and 3D Spherical shell geometries. These simulations are intended to be idealised representations of stellar interior convection. In this type of model, where convection is typically driven by imposing a fixed flux or temperature at the boundaries, many properties of the flow are dependent on the diffusivities of the fluid. In stellar convection zones, these molecular diffusivities are very small, and so are not expected to influence the bulk dynamics of the convection zone. Our simulations use a different approach to drive convection, namely by applying a distributed heating and cooling to the fluid internally, which allows us to bypass the formation of thermal boundary layers which typically throttle convective heat transport. This approach has enabled us to obtain diffusion-free behaviour at relatively modest parameter values. In rotating spherical shell convection, we find that diffusion-free heat transport and radial temperature contrasts are obtained, as is the transition from retrograde to prograde equatorial flow (which is generated by momentum transport associated with the convection). Other results, such as the convective velocity amplitudes, have a diffusivity dependence that varies with the degree of rotational constraint. Ongoing work in our group is looking at the influence of the structure of the heating and cooling on the simulated convection, as well as the application of this set-up to more complicated problems, such as magnetoconvection and convection in the anelastic approximation. The fact that many aspects of our simulations are diffusion-free has promising implications for the development of realistic models for stellar and giant planet convection.

Publications resulting from this work: Joshi-Hartley et al. (2025 MNRAS 541 2291), Lewis et al. (2025 ApJ 998 52), Joshi-Hartley et al. (in prep)

Left: Snapshot of the perturbation temperature  for a rapidly rotating spherical shell simulation, where the hot pole and cool equator are clear. Right: Plots of azimuthally-averaged normalised angular rotation rates for a set of simulations at a fixed Rossby number. Diffusivity decreases from left to right and top to bottom. The ‘anti-solar’ behaviour of slow equator, fast pole is seen, and does not vary with diffusivity. Images from Lewis et al. (2025)