Reliability-based optimization of steel structures using genetic algorithms and nonlinear finite elements

Abstract

Uncertainties are inherent in material properties, geometry parameters and loading in structural design problems. In a realistic design, it is necessary to consider these types of uncertainties to ensure safety and quality. Design constraints are formulated in probabilistic terms such as probability of failure or reliability index. The process of design optimization enhanced by the addition of reliability constraints is referred as Reliability-Based Design Optimization (RBDO). Most of RBDO methods use classical mathematical optimization algorithms and require the gradients of objective function and constraints. This task sometimes can be cumbersome and hard because reliability constraints are implicit functions of design variables. However, the increased power of computers has made possible to apply heuristic methods, especially Genetic Algorithms in RBDO problems. In this paper Genetic Algorithm is combined with OpenSees, a nonlinear Finite Element Reliability Analysis software, to salve RBDO problems. Two numerical examples show the performance of the implementation.