Worst case analysis in robust design of NMR magnets

The optimal design of electromagnetic devices is often performed by balancing conflicting objectives, including performance and other relevant quality figures such as cost and robustness. The balance of conflicting partial objectives can be tackled with the Pareto approach. Unfortunately, assessment of robustness typically implies a relevant increase in the computational cost of optimization procedures; in the paper, it is proposed an approach allowing to include robustness in Pareto optimization with a reduced increase in the computational cost.