Exploration and Optimization of Building Design Solutions using Computational Design

Sammanfattning: The focus of building design is increasingly moving towards considering performance as a driver, which can especially be seen in the realm of sustainable or green building design due to the advent of ambitious goals regarding energy consumption and emissions. Accounting for these concerns is an important target of the building design process, where there is a potential to make valuable contributions. To realize this, command of interconnected and sometimes contradictory requirements is needed, and it necessitates an ability to create novel solutions that fulfill demanding requirements. This entails an iterative process to find design solutions that meet the requirements and needs imposed by clients and regulations on space, costs, energy, etc. To account for the multitude of criteria in building design and aid practitioners in making well-informed design decisions, potential design solutions need to be explored to find feasible solutions and optimized to find solutions that perform as best we can. Even though the potential of further supporting both exploration and optimization can provide significant benefits to improving buildings’ performance, a common challenge is that the act of generating and evaluating larger sets of solutions can be restricted by time and resources. The use of computer-based methods has been suggested to facilitate this need. Through building information modeling (BIM), relevant information can be encapsulated and organized, and computational design approaches can leverage computers’ computational capability to efficiently generate, represent, and evaluate solutions. Together, these can represent an approach that facilitates the need for exploration and optimization of building design solutions. However, to achieve this further iterations, refinement, and evaluation with different design problems and contexts are needed to better leverage and understand its potential. This includes contributing to a further understanding of how and when different computational design approaches can be used to support the building design process.Therefore, the overall purpose of this thesis was to explore computational design for building design. The aim was to develop frameworks for exploring and optimizing building design solutions in a BIM-based workflow. The research design was based on an exploratory approach, in which frameworks were developed and evaluated in different building design settings. In this approach, a problem drives an iterative development of frameworks, where cycles of objectives, development, demonstration, and evaluation of the frameworks lead to suggestions that are communicated. The frameworks were then demonstrated and evaluated through real-world cases in building design settings. This process was iterated until the identified problems were addressed and the research purpose was achieved. The main findings in the research presented are:The inclusion of interactive exploration with computational design, where progression is not solely reliant on algorithms, facilitates the addition of qualitative preferences and criteria to be used in guiding the exploration of design solutions. This may be particularly useful in the early stages of design and when the emphasis is on finding novel design solutions.The inclusion of multi-objective optimization techniques (e.g., genetic algorithms) in computational design approaches can be used in situations where exploitation is of interest to optimize design solutions with multiple or conflicting objectives. This may be particularly useful when distinct and measurable objectives are available and targeted, such as the minimization of a building’s energy use, or costs from a life cycle perspective.Sensitivity analysis can be used to provide additional information on the impact of relevant parameters on a building solution’s performance. This can be applied to assist the analysis of alternative solutions from a computational design approach and can be useful in supporting the selection of solutions to bring forward in the design process.A master model approach can be used to structure and contain the constituent components needed to link inputs, outputs, and processes used in computational design approaches. This provides a framework to define a product, its design variables, constraints, and objectives, and to support the generation of representations and models necessary for performance evaluation.To facilitate information exchange and interoperability for the systems involved in a computational design approach, a middleware approach can be used to create interfaces between components. This facilitates the integration of existing computer-based methods and tools in BIM-based workflows into computational design approaches.Overall, the research presented in this thesis highlights different choices of computational design approaches and possible applications supporting the design process of buildings depending on the design problem’s characteristics and objectives. The proposed frameworks facilitate this through two different objectives: one targeted at the exploration of design solutions, and one targeted at the optimization of design solutions. Both are focused on facilitating and strengthening the role of computers as collaborative partners in the design process, rather than solely for information organization or increasing efficiency. By purposefully choosing an approach and through thoughtful application in different design problems, practitioners could be supported in making well-informed decisions regarding multiple design criteria, for example, in relation to environmental sustainability. The work in this thesis also presents approaches for mitigating some of the shortcomings of interoperability between a BIM-based workflow and components related to the proposed frameworks, which are necessary to overcome to maximize the effectiveness of a computational design approach.