Author(s)

E. Sandgren

Abstract

Design Optimization of structures is considered under variation of loading, geometry and material properties. A hybrid, genetic/nonlinear programming algorithm is coupled to a Monte Carlo simulation technique which seeks to locate an optimal design which can function adequately under the specified range of variation. Crossectional, geometric and topological changes are considered in the formulation. A traditional finite element solution is performed for each simulation where a specific value is selected for each design parameter from a statistical distribution which defines the range of variation. A design evaluation is the result of a number of simulations which produces an output distribution for each constraint imposed upon the design. The solution is executed in a parallel computing environment due to the large number of finite element an

Keywords