This presentation was made at CAASE18, The Conference on Advancing Analysis & Simulation in Engineering. CAASE18 brought together the leading visionaries, developers, and practitioners of CAE-related technologies in an open forum, to share experiences, discuss relevant trends, discover common themes, and explore future issues.

Resource Abstract

A robust design can be characterized by low sensitivities due to parameter changes, where the changes are small and typically within the tolerances of the design parameters. For designs with high safety factors, robustness is often postulated and not studied in detail. But robustness is becoming very important, when optimization is used to widely exploit the material and safety factors are drastically reduced. Then, small parameter changes can have a significant influence on the results and failures become more likely. So, with the wider use of design optimization, robustness of designs has to be considered more systematically than in the past.

From a methodological point of view, optimization and reliability methods have to be available for robust design simulations. Beyond that, the key point is the integration of both methods, i.e. the use of reliability constraints in optimization methods, to directly achieve a robust optimum.

From the parameter point of view, there are design variables for the optimization, and there are basic (uncertain) variables for the reliability analysis. A design variable may be uncertain or not. The selection of design variables depends on possible design variations like material, element properties (e.g. shell thickness, beam cross section), and geometry. The selection of uncertain variables depends on tolerance specifications and manufacturing conditions, which influence the product properties. The right ranges and distributions of uncertain variables need additional model input, which best should be obtained from product and manufacturing quality measurements. In addition, also load factors and load directions can be uncertain.

As an industrial example, the paper shows a charge air cooler, where an optimization is performed to reduce weight and stresses. Then, uncertain parameters are introduced and failure modes are defined. Their influence on the optimized design is studied and sensitivities are evaluated. Finally, an integrated optimization with reliability constraints is applied to directly achieve a robust optimum. Analysis, optimization, and reliability are performed with the industrial FEA code PERMAS.