Use this URL to cite or link to this record in EThOS:
Title: Performance driven design systems in practice
Author: Joyce, Sam
ISNI:       0000 0004 5923 1794
Awarding Body: University of Bath
Current Institution: University of Bath
Date of Award: 2016
Availability of Full Text:
Access from EThOS:
Access from Institution:
This thesis is concerned with the application of computation in the context of professional architectural practice and specifically towards defining complex buildings that are highly integrated with respect to design and engineering performance. The thesis represents applied research undertaken whilst in practice at Foster + Partners. It reviews the current state of the art of computational design techniques to quickly but flexibly model and analyse building options. The application of parametric design tools to active design projects is discussed with respect to real examples as well as methods to then link the geometric definitions to structural engineering analysis, to provide performance data in near real time. The practical interoperability between design software and engineering tools is also examined. The role of performance data in design decision making is analysed by comparing manual work-flows with methods assisted by computation. This extends to optimisation methods which by making use of design automation actively make design decisions to return optimised results. The challenges and drawbacks of using these methods effectively in real deign situations is discussed, especially the limitations of these methods with respect to incomplete problem definitions, and the design exploration resulting in modified performance requirements. To counter these issues a performance driven design work flow is proposed. This is a mixed initiative whereby designer centric understanding and decisions are computer assisted. Flexible meta-design descriptions that encapsulate the variability of the design space under consideration are explored and compared with existing optimisation approaches. Computation is used to produce and visualise the performance data from these large design spaces generated by parametric design descriptions and associated engineering analysis. Novel methods are introduced that define a design and performance space using cluster computing methods to speed up the generation of large numbers of options. The use of data visualisation is applied to design problems, showing how in real situations it can aid design orientation and decision making using the large amount of data produced. Strategies to enable these work-flows are discussed and implemented, focusing on re-appropriating existing web design paradigms using a modular approach concentrating on scalable data creation and information display.
Supervisor: Williams, Christopher ; Shepherd, Paul Sponsor: Foster + Partners
Qualification Name: Thesis (Eng.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available
Keywords: Design Exploration ; Optimisation ; Parametric Design ; Computational design ; data visualization