Mixed variable optimization using Taguchi's orthogonal arrays

Taguchi's orthogonal arrays for robust design are used in this paper in a nontraditional way to solve a mixed continuous-discrete structural optimization problem. The factors of an orthogonal array correspond to the members of a structure and the levels of each factor correspond to the material choices of each member. Based on the number of factors to be studied and the number of levels of each factor, an appropriate orthogonal array is selected for each specific problem. The number of rows of the orthogonal array correspond to the number of experiments (i.e. continuous sizing optimizations) to be conducted. The response of these experiments, which are the weight of the optimal designs corresponding to different material settings, are then used to calculate the mean effect of each factor level. Some possible optimal material settings can then be determined. Three examples are presented in this paper. Analysis using Taguchi's orthogonal arrays was able to isolate several near optimal or optimal designs. The accuracy and efficiency of the proposed method compared to more traditional solution methodologies are also discussed.