As modern tall buildings are getting increasingly taller and more complex, the search for the most cost-effective structural form for such buildings is the most challenging task in the design synthesis process. The objective of this research is to develop an automated computer-based technique for simultaneous topology and element sizing design optimisation of practical tall building structures....[ Read more ]

As modern tall buildings are getting increasingly taller and more complex, the search for the most cost-effective structural form for such buildings is the most challenging task in the design synthesis process. The objective of this research is to develop an automated computer-based technique for simultaneous topology and element sizing design optimisation of practical tall building structures.

An effective building representation model that defines the relationships of various structural subsystems and basic elements in a building structure is first developed. The subsystem-based representation allows structural engineers to describe the alternative designs of such subsystems according to the demands of practical design purposes. The use of the proposed representation model is capable of not only generating practical building layouts, but also facilitating effective searching for alternative structural forms.

A novel evolutionary technique incorporating a genetic algorithm (GA) with a local search (LS) operator is developed for the solution of the optimal design problem. An LS operator based on a rigorously derived optimality criteria (OC) method is developed and embedded in the framework of a GA, the resulting hybrid is called the hybrid OC-GA method. The hybridisation of the OC-GA method is strategically designed such that the GA is aimed for the global exploration of optimal topologies, whereas the OC is used to serve as an efficient local search operator for optimally sizing elements of selected topologies. Key parameters affecting the robustness and efficiency of the OC-GA method are identified and extensively tested. The superiority of the hybrid OC-GA method to the pure GA method in terms of both the quality of the optimised designs and the computational efficiency is evident in the design examples.

Example buildings ranging from a 40-storey planar frame to a 56-storey three-dimensional residential building are presented to illustrate the applicability, efficiency and practicality of the OC-GA method.