Stellar turbulent convection: A self-consistent model

We present a self-consistent model for stellar turbulent convection that is similar in spirit to the CM model (Canuto & Mazzitelli 1991) since it accounts for the full spectrum of the turbulent eddies rather than only one eddy, as done in the mixing length theory (MLT). The model differs from the CM model in the treatment of the rate of energy input n(s)(k) from the source that generates the turbulence. In the present model, n(s)(k) is controlled by both the source and the turbulence it ultimately generates, thus ensuring a self-consistent modeling of the turbulence. This improves the CM model in which n(s)(k) was taken to be equal to the growth rate of the linear unstable convective modes. However, since the formulation of a self-consistent treatment is far from simple, we were forced to use a representation of the nonlinear interactions less complete than the one in the CM model. The ensuing equations were solved numerically for a wide range of convective efficiencies. The results are the convective flux, the mean square turbulent velocity, the root mean squared turbulent pressure and the turbulent viscosity. We implemented the model in the ATON stellar structure code and computed the evolution of a solar model. The results are generally similar to those of the CM model and thus quite different from the MLT. The present model requires a smaller overshoot into the upper radiative zone than does the CM model, in accord with recent empirical estimates. Application to Population II stars and comparison with the very metal-poor globular cluster M68 yields an age in the range 11-12 Gyr. This is somewhat younger than the CM age, which in turn is younger than the corresponding MLT age, a result of possible cosmological interest.