This paper introduces a new bi-objective optimization approach for treating black-box models. Black-box models coming from the industry are often complex and some cannot always be treated with techniques that are solely deterministic. To overcome this issue whilst avoiding opting for a stochastic solution that would result in high computational costs, a combination of mono-objective solvers under a reference-point based scheme is proposed. Validation is performed against a nonlinearly constrained black-box model of a slotless permanent-magnet actuator. Satisfactory results reflect the effectiveness and utility of the optimizer.