New Method For Reliability-Based Optimization

A new mono-level iterative approach to optimal structural designs with mixing design variables and stochastic variables in a hybrid space is presented and discussed. Unlike the classical bi-level method, the method proposed here employs a new iterative approach. In the new approach during each outer optimization step the reliability analysis process performs only once, rather than completes the whole process till convergence. By doing so, the reliability analysis and optimization process converge simultaneously and respectively. Sequential unconstrained minimization problem is formulated using exterior penalty function approach and BFGS algorithm is taken for solution to the problem. Based on standard commercial FEA cods-ANSYS, a general purpose Reliability-Based Optimization system is developed by seamlessly integration with the analysis modulus of ANSYS. Truss examples are presented to illustrate the efficiency and accuracy of the methodology