Data-driven modeling of MHD turbulence for fusion device blankets

A Large-eddy simulation database of MHD turbulent channel flow with a wall-normal magnetic field has been generated using the open-source tool, OpenFOAM. The Lorentz force in the momentum equation and the Poisson equation for electric potential have been implemented and validated against existing Direct Numerical Simulation results. The range of Reynolds (Re) and Hartmann (Ha) numbers covered are relevant to fusion device blankets. Presently a range of 5,000 < Re < 50,000 and 10 < Ha < 500 is explored relevant to molten salt blanket concepts. This will be extended to higher Ha up to about 10,000 relevant for liquid metals later. The database is used to train machine learning (ML) models for predicting the source terms for the turbulent kinetic energy (k) and turbulence dissipation rate (ε) that can be used for the k-ε Reynolds averaged Navier Stokes model. This model is then exercised in fusion blankets to assess the difference in predictions with and without a turbulence model accounting for MHD effects. Presently, the database and the ML models are being extended to include the effect of three-dimensional interactions between the fluid flow and the external magnetic field.



Keyword: MHD turbulence, fusion blankets, AI-ML models, OpenFOAM



This project is funded through the Vertex LDRD project at Oak Ridge National Laboratory managed by UT-Battelle, LLC, under contract DE-AC05-00OR22725 with the US DOE.