EXPLORING URBAN CONFIGURATIONS FOR A WALKABLE NEW TOWN USING EVOLUTIONARY ALGORITHM

Abstract

Today’s cities and towns has grown from Ebenezer Howard’s simple ‘hub-and-spoke’(Curtis 2006) planning typology to complex and interlinked configurations. Modern planning could be more often in conflict than before and requires more empirical results to be substantial. This paper proposes an evolutionary approach to designing an urban configuration for a walkable new town. The urban configuration typically refers to the urban structure as defined by Wegener and Fürst( 1999). The walkable new town refers to a town whereby people can ride public transport to within the neighbourhood of their destinations and then walk to the final destination. This paper describes the evolutionary procedures used in order to evolve an urban configuration optimised for multiple conflicting performance criteria, including issues related to density, open space, walkability, and travel times. When multiple conflicting performance criteria are simultaneously being evaluated, there can be no single optimal solution(Fonseca and Fleming 1995). Multi-objective evolutionary optimisation therefore results in a population of solutions, where each solution represents alternative trade-offs between the conflicting performance criteria. The evolutionary algorithm developed in this research evaluates four key performance criteria: • the amount of open spaces adjacent to urbanised areas, • average built density of the urbanised areas, • travelling time needed to get around the town by public transportation and walking, and • cost of establishing the public transportation network. In total, 20,000 urban configurations were evolved. They are analysed in graphs. The evaluation scores for these urban configurations were plotted on a series of Pareto graphs, each graph comparing two performance criteria. The graphs highlight the inherent conflict between the performance criteria, as no urban configurations exist that are able to optimize all performance criteria simultaneously. Based on the Pareto graphs, certain designs were identified for more detailed analysis. This analysis highlighted certain urban typologies that were able to achieve a good trade-off between the four performance criteria. We refer to the best typology as ‘stretched’ urban configurations. Other interesting typologies were classified as ‘compact’ urban configurations, and ‘segregated’ urban configurations. The significance of this experiment is that it demonstrates how urban configurations can be explored empirically using evolutionary algorithm. The aim of such an exploration process is not to evolve a single ‘best’ design. Rather, this type of process will allow urban planners or designers to discover certain typologies and patterns that perform well with respect to a number of narrowly defined performance criteria. These typologies and patterns could then be used as a starting point for a more detailed process of design that takes into account a much broader set of social, environmental, and economic issues.