The objective of this work is to study the effects of geometry on structural performance of free-form steel space-frame roof structures and to optimize the structures without compromising overall architectural forms. Minimum weight optimization is performed to better study the effects of geometric alterations on overall structural performance. The intent is to achieve a strong optimum shape with superior load-carrying capacity allowing for the smallest and lightest structural members to be used. A two-phase genetic algorithm (GA) is developed to perform minimum weight design of the roof structures which consist of rectangular hollow structural sections (HSS). The new methodology is applied to two example roof structures subjected to the AISC LRFD code (AISC, 2005) and ASCE-10 snow, wind, and seismic loading (ASCE, 2010). Both are train station roofs for the Ottawa Light Rail Transit (OLRT) system to be built in Ottawa, Canada, in 2018. The structures are made up of a diamond-shaped grid pattern and their members are subjected to torsion in addition to bending and axial forces.
The GA was developed to perform simultaneous dimension, topology, and shape optimization and resulted in final designs which are 22% and 24% lighter than the initial designs created in a design office for the two roof structures. This global optimum solution was achieved in less than 19 hours on a standard workstation machine with a 2.83 GHZ dual core processor, a relatively short amount of time considering the complexity of both the structures and the optimization problem.