Computationally unifying urban masterplanning

• Main Author: Birch, David Alan
• Other Authors: Field, Anthony ; Kelly, Paul
• Published: Imperial College London 2013
• Subjects: 004
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.587359

Urban masterplanning is the process of creating a coherent design for developing a campus, suburb, city or region. Unfortunately these design and analysis teams face challenges which prevent rapid quantitative analysis of design iterations; precluding potential design improvement. These include limited automation, poor integration of modelling disciplines and, in particular, very limited scope for design space exploration. This thesis investigates these challenges and their solutions. A computational frame- work HierSynth is presented to help computationally unify the design and analysis sides of the urban masterplanning community. The key contribution of this thesis is HierSynths data model. This presents a reconceptualization of the workflow graph by composing it with tree based design-decompositions commonly found in architectural interoperability formats. This is achieved through a hierarchy of design queries, templates and analyses which when executed form a design hierarchy annotated with evaluated analyses. This enables detailed multi-scale analysis directly on design elements whilst supporting scenario generation and design space exploration capabilities and techniques to explore design improvements. The HierSynth framework is evaluated by application to a major commercial masterplanning project with Arup North America and is used to explore the most effective techniques for generating design insight. HierSynth enabled an order-of-magnitude more analysis iterations and previously infeasible design space exploration to answer design questions. During this collaboration an unexpected challenge was identified in maintaining and debugging complex, highly interrelated analysis models implemented as spreadsheets. A toolkit to address this is developed and applied to several generations of complex multi-disciplinary sustainability models. In summary this thesis presents evidence of the need for, implementation of, and practical benefits from, computationally unifying urban masterplanning design and analysis. The key contribution is a compositional data model supporting this unification. Finally avenues for further work are explored to further aid this community including data provenance and supporting smart cities.