Reliability-based optimization of anisotropic cylinders with response surface approximation of axial buckling load.

摘要

The reliability analysis and reliability-based optimization of laminated circular cylinders under axial buckling instability are studied. Structural reliability is measured in terms of Hasofer-Lind reliability index. The response surface models are used in both the calculation of the reliability index and the reliability-based optimization. In the reliability analysis, both deterministic and probabilistic sensitivity factors are investigated; the results show that the reliability index is most sensitive to the applied load and Young's modulus of the material. Two cases are considered in the optimization study. In the first case, the cylinder weight is minimized subject to a reliability constraint whereas in the second case, cylinder reliability is maximized subject to a weight constraint. In addition, two different optimization techniques are studied. In the first technique, a global response surface model of the buckling response based on 3000 Monte Carlo simulations is used for the design optimization whereas in the second technique, multiple local regression models, with each based on approximately 20 simulations, are used in sequential search of an optimum design. An optimum design is found. The results based on sequential application of multiple local regression models are close to those from global optimization while the former is much more efficient in terms of computational cost.