Colloquium: Alejandro Montoya Santamaria (FPT)

In the Liquid Metal Breeding Blankets (LMBB) of fusion reactors, the magnetic field interacting with liquid metal induces magnetohydrodynamic (MHD) effects on the flow. This results in large pressure losses and unconventional turbulence states that  can only be captured in higher fidelity Computational Fluid Dynamics (CFD) simulations such as Large Eddy Simulation (LES), but the high computational cost of these simulations make them impractical for industrial applications compared to Reynolds Averaged Navier-Stokes (RANS). This work presents a data driven approach to model MHD turbulence in RANS. Two correction fields are obtained through a frozen RANS simulation in which the mean LES fields are inserted into the RANS equations. The Reynolds stress anisotropy correction term is approximated with a modified Tensor Basis Neural Network (TBNN). Moreover, for modelling the turbulence production correction a Scalar Basis Neural Network (SBNN) is proposed and compared to a Sparse Algebraic Regression using the SpaRTA approach.

Supervisor: Dr. N.A.K. Doan