21.2.1.6. Robust Design Optimization Formulation

From the viewpoint of design optimization, the robust design represents that the performance of a design is insensitive to the variations of design variables and preassigned parameters. Thus, the robust design optimization finds the mean values of random design variables to minimize the variances of analysis responses while satisfying the design constraints for the mean responses.

There are two types of random parameters such as random design variables and random constant. The latter is called ‘Noise Factors’, which are the fixed values but their deviation effect on the variation of analysis responses. The formers are changeable during optimization process while their deviations are constants but effect on the variation of analysis responses.

Figure 21.4 shows a typical formulation of robust design optimization. As the statistical analysis requires many analyses, the robust design optimization process requires so many analyses. Thus, it is very important to minimize the total number of analyses in the robust design optimization process. RecurDyn/AutoDesign can reduce the number of analyses dramatically.