Two-Level Design Optimization for Compression Stiffened Panel

Abstract

In this paper, a two-level design optimization approach is presented for compression stiffened panel sizing. First, the stiffened panel design procedure is described by means of a structural index concept. This method is based on the analytical formulations for structures and is used for the preliminary design level. In this procedure, the final results have a general form without being restricted by the geometrical parameters. The final results can be used for sizing different types of compression stiffened panels. Using the results obtained in the first level, the surrogate-based optimization is employed for design optimization in the second level. The analysis tool in the second level is the finite element method. Using Sequential Quadratic Programming algorithm, a novel technique is developed to find the global optimum of the surrogate model. The proposed two-level design approach is employed for design optimization of a flanged stiffened panel. In this panel design problem, which has six design variables, an initial training set with 57 points is created. Using 173 infill points, the optimum solution is obtained. In comparison with the conventional optimization methods, the surrogate modeling reduces the required nonlinear buckling analyses significantly. An estimate of the required number of analyses in the conventional methods is 57,201. The results of both levels are compared with each other. The final results of the optimization process are the optimum design non-dimensional ratios for plate elements of the flanged stiffened panel.