Metodología eficiente de optimización de diseño basada en fiabilidad aplicada a estructuras

Resumen

The main objective of this thesis is to analyze several Reliability-Based Design Optimization methods. These optimization techniques are applied to solving design problems where the cost function of a structure, like the weight or the volume must be minimized subject to probabilistic constraints written as limit state functions. Actions, material properties and geometric variables of the structural members are considered like random variables. These methods can be classified as double-loop methods, single loop methods and decoupled methods. A computer toolbox has been developed including procedures in the areas of structural analysis, structural reliability, design sensitivity analysis, mathematical programming and Monte Carlo simulation. The program solves both analytical problems and structural problems like plane and spatial structures. Computational efficiency, convergence, accuracy, generality and robustness of the methods have been discussed. Many options can be set by de user: displacement constraints, stress constraints and buckling constraints, prescribed reliability indices, distribution function for random variables, correlation coefficients between random variables, design sensitivity analysis methods. Probabilistic optimum designs are verified by Importance Sampling-based Monte Carlo Simulation.