To perform design optimization, the same objective, constraint, and side constraints defined in trade-off determination are used. DOT, ANSYS, and sensitivity computation and model update programs similar to those discussed in Section
3.8.2 are integrated to perform design optimization. After four iterations, a local minimum is achieved. The optimization histories for objective, constraint, and design variables are shown in
Figures 3.31a, b, and c, respectively.
From
Figure 3.31a, the objective function starts around 362 in.
3 and jumps to 382 in.
3 immediately to correct constraint violation. Then, the objective function is reduced further until a minimum point, 354 in.
3 is reached. Also, the constraint function history graph shows that 80% violation is found at the initial design, and the violation is reduced significantly to 65% below the bound at the first iteration. Then, the constraint function is stabilized and stays feasible for the rest of iterations. At optimum, the constraint is 4% below the bound, the maximum displacement becomes 0.09950 in., and the design is feasible. The most interesting observation is that, from
Figure 3.31c, all design variables are decreasing in the design iterations, except for dv1 and dv7. Design variable dv1 increases from 0.58 to 0.65 in. at the optimum. However, the most significant design change is dv7 from 1.25 to 1.44 in., which contributes largely to the reduction in the deformation, as listed in
Table 3.7. Decrement of the rest of the design variables contributes to the volume reduction.