A trade-off function to tackle robust design problems in engineering

In this paper, an original trade-off function is proposed to deal with robust design in engineering. Robust design in engineering aims at sizing systems with a low variation in all performance measures under uncertainties. It is mainly related to the generation and evaluation of candidate solutions. However, it is also suitable to consider the robustness in the decision-making process to ensure the selection of the most preferred design in the early stages of the design process. The purpose of our research work consists in tackling the robust design problem as a trade-off between two design objectives: (i) improve the global level of performance of the system and (ii) reduce the sensitivity of the performance with regard to uncertainty sources. The robustness of the candidate solutions is evaluated by a trade-off function modelling the designer's intention with an a priori articulation of preferences. Design objective formulation is based on the concept of desirability function and indices. Candidates solutions are ranked according to their ability of reducing the design sensitivity while remaining highly desirable. The developed approach is illustrated on an example of a truss structure design problem which is numerically solved by a genetic algorithm.