Optimization of product in dynamic design space and selection through the arc-elasticity concept

In the industrial design process, "trial-and-error" loops are usually used between design departments and simulation departments, to design and validate a candidate solution. Such a solution is defined by instantiated design variables, characterizing the main parameters of the product. This paper proposes to replace the trial-and-error process by an optimization method; two bottlenecks were identified: the importance of the initial design solution which has already engaged delays and costs, and the formalization of designers preferences to select high-performance solutions. To take into account the initial solution, a strategy is proposed to explore the global design space (which represents all the candidate solutions) starting from the initial one, and using a hierarchical organization of the design variables. Then, the Observation-Interpretation-Aggregation method is used first to formalize designer knowledge related to the product to design and then to rank every design solution using a single performance value. Moreover, the concept of arc-elasticity is used to qualify a solution through its neighborhood, and to characterize technological breakthroughs, compared to the initial solution. This method is discussed and illustrated through the design of an aeronautical riveted assembly. © 2011 Springer-Verlag.