A hierarchical optimization framework for simulation-based architectural design.

This thesis presents a rigorous simulation-based optimization framework that enables concurrent and consistent decision-making in building design. The framework facilitates computational decision support for meeting building performance goals, allows autonomy of specialized design tasks with timely and efficient use of analysis tools, and preserves dependencies between possibly competing building performance goals. Approaching simulation-based building design as system design problem, the thesis extends Analytic Target Cascading (ATC) to thermal and HVAC design. Previously used only for automotive design, ATC is a hierarchical optimization methodology for achieving compatible design targets in large engineering systems at early product development stages. Two case studies illustrate the use of the ATC framework in the context of building design. The first case study demonstrates how hierarchical optimization methodology functions in the context of building design. The second case study portrays a realistic and challenging design scenario derived from a survey of an existing health care facility on the University of Michigan campus. The case study reveals both the benefits and challenges associated with formalizing a simulation-based design scenario and demonstrates how dependencies between HVAC, energy and thermal comfort, and CFD analysis tools can be coordinated within the ATC process. Thus, a new approach is contributed to ongoing work on integrated simulation-based building design. Experience gained from both studies is used for deriving a generalized framework for organizing and solving goal-driven, analytic design tasks in architecture. Work presented in this thesis is built on the foundation laid by continuous non-linear optimization models. This thesis demonstrates how problem-specific selection and optimization methods can be used for solving simulation-based design tasks in architecture. Furthermore, application of optimization methods in building design is extended not only to computing efficient solutions for specific problems, but also to deriving compatible values of dependencies between multiple interconnected problems.PhDArchitectureCommunication and the ArtsUniversity of Michigan, Horace H. Rackham School of Graduate Studieshttp://deepblue.lib.umich.edu/bitstream/2027.42/124032/2/3121905.pd