Numerical Analysis of Nonlocal Convection --- Comparison with Three-dimensional Numerical Simulations of Efficient Turbulent Convection

TL;DR

This study compares 1D nonlocal turbulent convection models with 3D simulations, evaluating closure models and calibrating coefficients to improve the accuracy of simplified models in capturing key features of turbulent convection.

Contribution

The paper introduces a calibration method for 1D down-gradient models using 3D simulation data, enhancing their robustness for deep convection zones.

Findings

1D models capture key features like the U-shape of temperature gradients.

Calibrated coefficients are more reliable in deep convection zones.

Prediction of kinetic energy flux remains unsatisfactory.

Abstract

We compare 1D nonlocal turbulent convection models with 3D hydrodynamic numerical simulations. We study the validity of closure models and turbulent coefficients by varying the Prandtl number, the P$\acute{e}$clet number, and the depth of the convection zone. Four closure models of the fourth-order moments are evaluated with the 3D simulation data. The performance of the closure models varies among different cases and different fourth-order moments. We solve the dynamic equations of moments together with equations of the thermal structure. Unfortunately, we cannot obtain steady-state solutions when these closure models of fourth-order moments are adopted. The numerical solutions of the down-gradient approximations of the third-order moments, on the other hand, are robust. We calibrate the coefficients of the 1D down-gradient model from the 3D simulation data. The calibrated coefficients are more robust in the cases of deep convection zones. Finally we have compared the 1D steady-state solutions with the 3D simulation results. The 1D model has captured many features appearing in the 3D simulations : (1) $\nabla-\nabla_{a}$ has a U-shape with a minimum value at the lower part of the convection zone. (2) There exists a bump for $\nabla-\nabla_{a}$ near the top of the convection zone when the P$\acute{e}$clet number is large. (3) The temperature gradient can be sub-adiabatic due to the nonlocal effect. Apart from these similarities, the prediction on the kinetic energy flux, however, is unsatisfactory.