Performance-based Building Massing Design Generation And Optimization System

In order to leverage the capability of computational and artificial intelligent in architectural design,particularly for solving the design challenge related to building performance,computational design optimization has been considered a powerful and promising tool for assisting architects in achieving a performance-based architectural design.With the development computational design and optimization techniques,parametric design,optimization algorithms,and building performance simulation become widely accessible in architectural design,and researchers and architects can integrate these components to establish automated optimization workflows aimed at performance-based building design optimization.Aimed at various performance factors,such workflows can help architects search from building design alternatives with excellent performance.Such design approaches are often referred to as performance-based design optimization.Even though many successful cases in research show that performance-based design optimization is able to play a positive role in improving building performance,it is still challenging to use this approach in real-world design tasks.It is due to the fact that the user(architects)have to spend a considerable amount of time and effort in using performative design optimization,while the return from the optimization is very low in terms of design information feedback.On the one hand,the process of parametric modelling and establishing an optimization workflow is typically time-consuming and technical demanding even to skillful researchers or architects.On the other hand,information produced from performance-based design optimization is typically limited,which cannot give sufficient feedback in architects’ design ideation and decision-making.Considering these limitations,The aim of this project is to develop a computer-aided design approach for integrating building performance optimization into concept architectural design exploration,by providing a more integrated building massing design generation and optimization system that can produce site-specific and task-specific high-performing design solutions while with minimal interruption to architects’ design processes.In order to fulfil the aim,the research proposes a Performance-based Building Massing Design Generation and Optimization System(PBMDGO system)which is feasible for architects to use in conceptual design phases.The development of PBMDGO system is also aimed to explore possible methods and approaches to addressing the mismatch between the need for architectural design and the application of performative design optimization.In order to make PBMD satisfy the need for architectural design,the research re-designs the standard optimization workflow and develops PBMDGO system based on the re-designed workflow on the Rhino-Grasshopper parametric modelling platform.The system simplifies the operation and the establishing of the optimization workflow,while enhancing the design information feedback from the optimization result.To make PBMDGO system functional,the major task of this research is to rebuild and upgrade the two essential components in the re-designed optimization workflow,which are the design generator(the parametric model)and the optimization solver(the optimization algorithm).With these two components,architects can establish customized PBMDGO system aimed at different performance objective by integrating different third-party building performance simulation tools.In order to investigate the efficacy of the PBMDGO system,testing design optimization problems with different conditions and constraints are optimized.The result indicates that the two PBMDGO systems produces task-specific design variants for every design optimization problem.Moreover,comparing with other simple building design generative models/algorithms or generative models/algorithms based on free-form or curved geometry,the optimization result produced by PBMD based on orthogonal geometry reflects the design strategies and design information in a clearer way.In addition,compared with using multi-objective Pareto optimization,the optimization result produced by PBMD based on multiple single-objective design optimization tasks can provide design information more relevant and specific to the given performance factor.The development and implementation of PBMDGO system provide a new possible approach to integrating performance-based design optimization into architectural design.In(conceptual)architectural design,performance-based design optimization should help the architect widely and systematically explore different design possibilities,while providing timely and essential design information feedback to the architect for helping them overcome design fixation and achieve a performance-informed and-aware design process.PBMDGO system actively responds to these needs and can provide design information feedback to the architect’s dynamic and iteratively explorative-exploitative design synthesis processes.Thus,PBMDGO system can be an independent“machine designer” who is externalized to human designs but can also collaborate with human designers during the design synthesis process.Such design processes can be characterized as a human-machine co-evolution design synthesis process and is able to significantly strengthen the design competence of architects.In the context of rapid development of digitalization and intelligence,such co-evolution design synthesis processes are also an essential to achieve step changes in performance-driven architectural design.The whole thesis contains about 118,000 Chinese characters,233 pictures and charts,and around 8300 lines of codes.