Analysis and optimum design of cold-formed steel arch structures

Abstract

This research presents the development of a unified framework for cold-formed steel warehouse design at variations of clear roof spans. The work involves the advanced finite element analyses of nonlinear arch-shape warehouse structures, namely one that considers 2nd-order nonlinear geometry effects-sufficiently accurate approximation of large deformation responses. An optimum design procedure of 3D arch steel warehouse structures is conducted with the help of a novel algorithm, called mixed ESO-PSO approach. The generic idea is based on the implementation of a well-known evolutionary structural optimization algorithm (ESO), and further integrates a mapping, underpinning particle swarm optimization (PSO), technique of design variables to the closest upper or lower integer numbers listed in the available steel sections. In essence, the use of ESO algorithm during the preliminary design provides good initial points by eliminating those ineffective or even infeasible design domains, and hence reduces the sizes of discrete variable entries prior to performing the PSO searches in the final step. The objective function is to minimize suitable arch geometry that yields minimum cold-formed steel hollow sections employed specifically in warehouse applications. Both ultimate strength and serviceability criteria of designed structures comply with AISC-LRFD and AISI-LRFD specifications. The algorithm and various constraints were coded within a Microsoft Visual Basic environment interfaced with the commercial SAP2000 structural analysis codes by direct OAPI communication. Whilst there is still no guarantee for (leaving alone the global one) optimal solution in view of the intrinsic non-convex and non-smooth optimization problems, a number of successfully solved benchmarks highlighted efficiency and accuracy of the present design method in steel structure applications. The optimum procedure is automatically performed as an iterative process, and incorporated advanced analysis until the minimum cross-sections obtained without violating its given constraints.