D. Arnett: Progress in 3D simulations of Stellar Convection (Review, Wed, Aug. 23)
This was one of the review talks of session D in the Convection Symposium: "Stellar evolution, nucleosynthesis, and convective mixing".
3D simulations with modest resources were presented, where oxygen burning allows thermal relaxation for quasi-adiabatic convection. The computational domain is a subsection of the star. Multi-fluid, realistic physics allows astronomical connections to be made. A careful treatment of boundary conditions and initial models allows long times to be simulated. Finally, extensive graphical and theoretical analysis was used to extract physics. Much of the work was done by Casey A. Meakin (PhD student).
A comparison to 2D simulations shows that 2D simulations overestimate velocities (angular momentum constraint) and incorrectly give inefficient turbulent mixing. Movies of the core showed that in 2D one gets much higher velocities and a less homogeneous core.
After a discussion of 2D burning of C, Ne, O (flames develop because of entrainment of nuclear fuel), the buoyancy frequency, and density fluctuations (occur mostly at interfaces), a comparison of oxygen in 2D and 3D was presented. In 3D, the oxygen abundance is much smoother, better behaved, and there are less fluctuations.
Next, it was pointed out that the solar convection zone can be seen in a plot of superadiabaticity vs. radius as a tiny spike at the surface, and that "Stein and Nordlund country" includes only 3% of the Sun.
Waves are generated at convective interfaces. This can be seen in simulations of turbulent, compressible convection: Velocities do not go to zero at the boundaries (as when using MLT). Those are not convection, but waves (g and p modes).
The convective core grows by entrainment. In a graphic of abundance gradients (e.g. oxygen) as a function of radius and time one could see the convective core growing with time. Patrick Young is another collaborator and has incorporated a first version of the entrainment in the TYCHO code. Previously successful results for wide, double-line eclipsing variables (which used a simpler model) are reproduced.
Implications for the standard solar model:
- It has a problem - it does helioseismology too well.
- It is static - adding even small dynamic effects may spoil it.
- Dynamic effects on opacity diminish the helioseismologic discrepancies for the new abundances and increase them for the old composition.
- The increased diffusion and the He surface abundance are at odds - rotational stirring may be required.
- John Bahcall would have loved this new challenge.