Multi-objective optimization of magnetic ON-OFF mode in magnetically fluidized beds

This study presents a computational framework for optimizing magnetically fluidized bed (MFB) ON-OFF mode systems through integrated CFD-DEM simulations, surrogate modelling, and multi-objective optimization. Its aim is to achieve precise control over periodic particle segregation and/or mixing in MFBs for industrial applications, requiring sustained segregation or mixing. A physics-based CFD-DEM model for fluidization with incorporated magnetic field gradient effects was developed to capture the dynamics of magnetic and non-magnetic particle interactions under gradient ON-OFF switching. Given the computational expense of extensive parametric studies involving two hundred operational scenarios, neural network-based surrogate models were evaluated to predict segregation behaviour as a function of magnetic ON time, OFF time, and inlet gas velocity. The Non-dominated Sorting Genetic Algorithm II (NSGA-II) was employed for multi-objective optimization, revealing a well-defined Pareto front illustrating trade-offs between segregation effectiveness and cycle times. Optimization results demonstrate a characteristic staircase pattern spanning penalty values from 0.0 to 12.0 and cycle times from 4.0 to 16.0 s. Two critical operating points were identified: a shorter cycle time solution (4.0 s cycle time, 6.6 penalty) and a performance-optimized solution (9.5 s cycle time, 0.0 penalty). Validation using two representative solutions confirmed successful achievement of optimization targets with sustained segregation and mixing. The integrated framework demonstrates controlled gradient ON-OFF mode operation feasibility in MFBs, providing foundation for future experimental investigations.