Physics informed deep learning for fusion thermal hydraulics

A significant challenge in magnetic confinement fusion is the high heat loading of the plasma-facing
components. High energy neutrons bombard these components, creating a non-uniform volumetric heat
loading which must be transferred to prevent thermal damage. While numerical simulation offers an
attractive route forward for the design of these components, its high cost, breadth and high
dimensionality limits efficacy in design space exploration.
In this project, the potential of physics aware deep learning surrogate models as candidates for full fluid
flow PDE modelling will be investigated. Our motivation is efficient exploration of parameter space for
fusion thermal hydraulics. Physics informed neural network (PINN) techniques have already shown
significant promise in isothermal flows without magnetohydrodynamic (MHD) effects but need further
developing to account for fusion relevant conditions. PINNs will be used for broad and rapid parameter
space exploration while selected high-order Computational Fluid dynamics (CFD) studies (generated as
part of this PhD) will be used in training and testing of the model.
This work will focus on turbulent buoyant MHD, with conjugate heat transfer and inhomogeneous
volumetric heating. The cases studies are a heated cavity and serpentine passage configuration, both
being central to blanket / divertor design in fusion. This will provide a vital understanding of the complex
flow physics present in the plasma-facing components and is expected to lead to improved designs in
future fusion plants.