Enhancing computational efficiency in nuclear fusion through reduced order modelling: Applications in magnetohydrodynamics

Magnetohydrodynamics (MHD) studies the dynamics of electrically conducting fluids under the influence of a magnetic field and it is relevant in several nuclear applications. However, the high computational cost of multi-physics MHD simulations poses a challenge. Reduced Order Modelling (ROM) offers a promising alternative, enabling lower-dimensional approximations while preserving accuracy. This allows for a reduction in the computational time and, at the same time, accurate approximations of the intricate physics involved in fusion reactors. However, ROM techniques are relatively new within the MHD framework, and benchmark test cases should be considered in this first stage for verification and validation. Therefore, this study applies the ROM methodology to a MHD scenario to study their potentialities (and eventual criticalities) for this class of problems. The benchmark test case considered in this work is the Backward-Facing Step. The obtained results contribute to assessing the capabilities of ROM methodologies in MHD scenarios, demonstrating their potential to enhance computational efficiency in this field and representing a critical step towards advancing the computational modelling of complex systems in nuclear fusion. • Application of Reduced Order Modelling to Magnetohydrodynamics. • Lead-lithium magnetohydrodynamic flow in a Backward-Facing Step geometry. • Dynamic Mode Decomposition method for forecasting the evolution of the system. • Evaluation of the accuracy of the method for state reconstruction.