The building façade plays a key role in the entire building’s energy performance. In commercial buildings, energy demand is dominated by space heating, cooling, and artificial lighting. Façade design variables for these three factors have always been interacting and sometimes even in conflict with each other. For different climates, adaptive façade design solutions should be implemented to achieve optimal design objectives, such as energy performance, human comfort, and life cycle cost. While the optimal solution is traditionally identified through “trial-and-error”, for complex optimization problems that contain a great number of design variables, it might require extensive hours of computation at early design stage, a condition that is increasingly infeasible in practice due to cost or time constraints. Since 2008, there has been a significant trend in building performance optimization techniques (that used to emphasize solely on simulation) being implemented, instead of building simulation techniques, to obtain design solutions for building performance optimization problems. Among widely implemented optimization algorithms, the genetic algorithm (GAs) have proven effective with its robustness in dealing with discontinuous variables. However, for complex optimization problems with a great number of variables, such as façade performance optimization (FPO) problems, GAs are still too time-consuming to be implemented at the early design stage, thus efficiency becomes the main area for its augmentation. The main objective of this study is to develop a new evolutionary algorithm method, adaptive radiation (AR), based on simple GAs to solve complex optimization problems relative to the design approach of the climate-responsive façades. AR is derived from the biological process of adaptation where specific species are evolutionarily adapted to their immediate ecological niches. This process can obtain optimal solutions of façade design variables (infiltration, window-to-wall ratio, shading geometry, glazing types, wall insulation, etc.) in significantly less computation time than GA. In this study, AR is implemented in three different climates in the United States to demonstrate its robustness and efficiency. The results validated the potential of AR through façade design scenarios. The procedure can also be extended towards a broad field of complex simulation-based architectural optimization problems. This thesis can be downloaded from: https://deepblue.lib.umich.edu/handle/2027.42/133475
