Multi-Objective Optimisation of Urban Design Using a Genetic Algorithm

In the current fast-growing cities, urban planners are continually required to optimize city designs to improve the urban environment whilst meeting ever more stringent regulations. Thus, the demand for optimization is increased continuously. However, real-world optimization problems often have a large search space and have objective functions with many conflicting objectives that are expensive to evaluate. The decision-making process to select a solution is equally demanding. These difficulties could be relieved by tools supporting the urban planners and decision-makers throughout the entire process of urban planning. This paper aims to show the potential and share insight gained using a genetic algorithm in such a problem area. It proposes an approach for parametric urban design at a neighbourhood level that optimizes the buildings’ ground floor area (GFA) and green area in relation to the demography of the population. The approach is based on dynamic parameters such as demography, and urban morphology, which are linked to people’s needs rather than fixed rules. It is automated in a tool implemented using a nondominated sorting genetic algorithm (NSGA-II). The tool proves to be useful for urban planning when decisions influencing the development of the city for a long time are taken.