Automated Design of Steel Frames Using Advanced Analysis and Object-Oriented Evolutionary Computation

Abstract

An automated design algorithm for partially restrained and fully restrained steel frames is presented. Advanced analysis based design is utilized and inelastic modeling of member behavior is based upon the distributed plasticity (plastic zone) model. Nonlinear connection behavior is simulated with trilinear moment-rotation curves. The automated design process is implemented using an evolutionary algorithm (EA). Object-oriented (OO) data structures are used to represent building frame components and implement the reproductive operations in the evolutionary algorithm. An optimized design problem is developed using an objective function which includes frame member weight and connection cost/complexity. Constraints related to both service and strength load levels are included. Two frame designs with varying topology are presented and discussions of the OO-EA performance are provided. It is shown that the OO-EA is a robust procedure for the automated design of steel frames using advanced analysis, which suggest that the EA can be a useful methodology upon which to pursue the development of automated performance-based design algorithms.