Predicting Nusselt number in turbulent Rayleigh–Benard convection using different machine learning models.
Developed a Deep neural network to predict the Nusselt number, Nu in turbulent RBC convection
for Nusselt number and Prandtl number given as input features and compared its performance with simpler models like linear and random forest regressor.
Modelling of Nusselt number in turbulent Rayleigh–
Bénard convection
In Rayleigh-Bénard Convection (RBC), flow parameters such as Rayleigh number (\(Ra\)) and Prandtl number (\(Pr\)) determine the bulk properties like Reynolds number (\(Re\)) and Nusselt number (\(Nu\)). We attempted to describe the dependence of \(Nu\) on \(Ra\) for a fixed \(Pr\) by numerical simulation using a finite difference solver. We determined the exponent (\(\alpha\)) in the power law relation \(Nu \propto Ra^{\alpha}\) by conducting Direct Numerical Simulations (DNS) of two-dimensional (2D) RBC at Rayleigh numbers ranging from \(10^5\) to \(10^8\) at \(Pr=1\). We also computed the spectra and fluxes for all the cases to understand the effect of aspect ratio on energy transfers in the flow.