Optimization of a layered regenerator inside a magnetocaloric cooling system using an evolutionary algorithm.

Magnetocaloric (MC) refrigeration systems have to implement MC Materials (MCM) with differentiated Curie temperatures (TC) inside a layered regenerator in order to reach temperature spans required for commercial applications. Magnetic and thermal interactions between MCM with different TC and the number of free parameters related to the dimensioning of the system lead to numerous computational difficulties to reach optimal designs. In this paper, we present an optimization process of a MC cooling system from the points of view of both thermal power density and exergy efficiency. A 3D magnetic - 2D thermal - 1D fluidic multiphysics numerical model of parallel plates Active Magnetic Regenerator (AMR) is used as an evaluation function in an evolutionary algorithm which is coupled with massively parallelized computing capabilities. The solutions are wanted to be resilient with respect to variable operating conditions. They converge towards an optimal design and without calculating the overall Pareto’s front.