Methods Of Structural Robust Design And Their Applications

The variation of various uncertain factors in practice may have negative effects on structural performance. If structural performance is sensitive to variability of the uncertainties, the possible variability of factors can affect the normal function of structures, worsen structural performance, and even make it disabled. However there are probabilistic uncertainties and non-probabilistic uncertainties commonly in engineering design, it is therefore necessary to study methods of robust design of structures with probabilistic and non-probabilistic uncertainties and theirs applications, in which the structural performance is least sensitive to the variation of uncertain factors. Robust design is a new concept. Its fundamental principle is to select suitable design variables so that the design performance is insensitive to the various causes of variation without eliminating possible variation of variables. Robust design addresses both the objective robustness and the robustness of constraint functions (or the feasibility robustness). In the study, new robust design methods have been proposed to consider variation of both probabilistic and non-probabilistic uncertainties. Various strategies to achieve the objective robustness and ones to achieve the robustness of constraint functions have been proposed. Designers may select suitable methods to formulate a suitable mathematical model for robust design according to special questions under consideration. Robust design of structure may be converted into one or several non-linear optimization problems with constraints. Therefore optimal algorithms from the optimization toolbox and the genetic algorithms toolbox of the Matlab software may be properly chosen to obtain the optimum solution to the robust design depending on continuous or discrete design variables.Furthermore, the proposed methods are applied to the design of two typical engineering structures. One is the robust design of a simple truss, and another is the robust design of a submarine cylindrical pressure hull. The computational results show that the design whose performance is insensitive to the variation of uncertainties can be achieved using proposed mathematical models of robust design. Compared to the design scheme from traditional design, the weight of the design scheme from proposed robust design increases. However, the increased weight is required to improve the ability of resisting the variation of various uncertain factors. The reliability and robustness of the structure have been improved accordingly. Studies in the thesis provide a new approach for engineering structural design.