Design optimization for robustness using quadrature factorial models

This paper describes a robust optimization methodology for designs involving either complex simulations or actual experiments. The methodology adopts a new objective function that consists of the Expected Performance and the weighted Deviation Index. The proposed Quadrature Factorial Model estimates the expected performance and the standard deviation of a design. This scheme greatly reduces the number of experiments and provides superior results for systems with significant interaction effects and nonlinear variations. The proposed methodology is applied to the design of helical gears with minimum transmission error. The robust optimum shows a significant reduction of the expected transmission error compared with previous studies, while maintaining the insensitivity to manufacturing errors and load variation.