EVOLUTIONARY OPTIMISATION OF URBAN STREET PATTERNS USING AXIAL LINE ANALYSIS

Abstract

This dissertation examines the feasibility of automating iteratively the axial line analysis of Space Syntax software with generative modelling to evolve and evaluate a population of urban street patterns through Evolutionary Algorithms. The aim of this dissertation is to assist decision-making in the early "sketch" design of urban street patterns by attempting to use axial line analysis as a design tool. Urban design is a complex process involving various considerations and time-consuming. However, the concept of automating design options has been prevalent in the architectural design nowadays. To date, Space Syntax is one of the few tools which is capable of analysing urban street patterns especially the axial line analysis technique. It is still unable to be automated for evaluating design options. In the area of urban street modelling, generative technique is capable of producing a large number of street options in a short period of time. Hence, a special framework needs to be developed to incorporate axial line analysis into generative modelling of street patterns for evolutionary optimisation. The dissertation framework consists of two parts in general. The first part reviews the development of generative modelling and the potential of axial line analysis of urban street patterns. Case studies of similar structure are examined to investigate the most suitable methods of generating and evaluating urban street patterns. The second part of the framework is the design experiment. The experiment demonstrates the hypothesis in a software environment using a combination of computer softwares. Two generative models based on street grid and junction types are developed in the preliminary tests to set up the basis for the experiment. Based upon the findings, another model is developed to run for the experiment. The results are fed to the evolutionary system that evolved and optimised the design population following certain performance criteria. The results show that the hypothesis is valid and the feasibility of using axial line analysis as a design tool is successfully demonstrated despite certain limitations. It is hope that findings of the experiment provide a discussion point with regard to the underlying opportunity of axial line analysis technique in the Space Syntax tool. It also shows that Evolutionary Algorithms are capable of evaluating design options and facilitating the decision-making in the early stage urban street design process.