This work presents a computational design of optimal chemical concentrations of chosen alloying elements in creating new magnetic alloys without rare earth elements that have their multiple desired macroscopic properties extremized. The design process is iterative and uses experimental data and a multi-objective evolutionary optimization algorithm combined with a robust response surface generation algorithm. Chemical concentrations of each of the alloying elements in the initial set of candidate alloys were created using a quasi-random sequence generation algorithm. The candidate alloys were then examined for phase equilibria and associated magnetic properties using a thermodynamic database. The most stable candidate alloys were manufactured and tested for macroscopic properties, which were then fitted with response surfaces. The desired magnetic properties were maximized simultaneously by using a multi-objective optimization algorithm. The best predicted Pareto-optimal alloy compositions were manufactured, synthesized and tested thus increasing a set of experimentally verified alloys. This design process converges in a few cycles resulting with alloy chemistries that produce significantly improved desired macroscopic properties, thus proving efficiency of this combined meta-modelling and experimental/computational alloy design method.
