AI-driven Monte Carlo uncertainty analysis of Curie temperature effects on active magnetic regenerator performance.

Magnetic refrigeration is a promising alternative to conventional vapor compression systems, offering the potential for higher efficiency and environmental benefits. However, its widespread adoption is hindered by several aspects. This study evaluates the impact of one of them, the Curie temperature uncertainty, on the cooling capacity of multilayered active magnetic regenerators. That is achieved by Monte Carlo simulations performed by Artificial Neural Networks trained on experimentally validated numerical data. The results show that the current variability of the material poses significant challenges for both prototype development and large-scale production. To ensure reliable performance, prototype designs tend to be oversized, leading to excess cooling capacity ranging from 30 to 80 %, as well as increased costs due to larger magnetic circuits and active magnetic regenerators. Furthermore, industrial-scale production remains unfeasible, as current magnetocaloric materials lack the quality control required to meet manufacturing standards.