Energy conversion devices using solid-state magnetocaloric materials have the potential to reduce energy consumption and mitigate environmental pollutants. To overcome the limited magnetic entropy change of magnetocaloric materials, magnetic refrigeration devices typically use the active magnetic regenerator (AMR) cycle. AMR devices have demonstrated promising performance, however costs must be reduced for broad market penetration. Although the magnet cost is of greatest importance for commercialization, literature has decoupled magnet design from AMR optimization. And while multilayered regenerators can improve performance without increasing cost, a number of questions remain unanswered as a result of the prohibitive parameter space.

This dissertation explores methods of improving AMR performance and decreasing cost both at the subsystem level, namely the magnetocaloric regenerator, fluid flow system and magnetic field source, and the device level by coupling the regenerator and magnet design problems in a cost optimization framework. To improve AMR performance, multilayered regenerators with second-order magnetocaloric materials are experimentally and numerically investigated, yielding insight on how individual layers behave and interact over a wide range of regenerator compositions and operating parameters. An efficient AMR modeling approach is presented where individual layers are treated as cascaded AMR elements, and simulations are in excellent agreement with experiments. Insights from the computationally efficient model are used to inform device modifications, and a no-load temperature span of 40 K is measured in close proximity to the simulated optimum;​ one of the highest in literature.

To simultaneously decrease AMR costs, a permanent magnet optimization framework is explored that is conducive to nonlinear objectives and constraints. This is used to investigate the optimal design of permanent magnet structures with reduced rare-earth permanent magnet materials. The regenerator and magnet design problems are then coupled in a permanent magnet topology optimization to minimize the combined capital and operating costs of an AMR. The optimal magnetic field waveform and the optimal means of producing this waveform are simultaneously obtained. The lifetime ownership costs of the optimized AMR device are shown to be in the realm of existing entry-level cooling devices. The presented cost optimization framework is of interest to both scientists and engineers, and demonstrates the importance of fast AMR models in identifying system designs, regenerator compositions and operating regimes that increase AMR performance and decrease cost.