Integrated two-stage optimization of façade adaptation and renewable systems for indoor comfort in NZEBs

While adaptive façades have been extensively studied, computational platforms for designing nearly-zero energy buildings with optimally controlled adaptive façades remain limited. This study introduces a two-stage optimization platform to achieve thermo-visual comfort with minimal energy consumption in the first stage, followed by the optimal sizing of a renewable energy system to meet the corresponding energy demand in the second stage. The performances of the Opossum and deep reinforcement learning algorithms are compared for controlling façade openness to maintain indoor thermo-visual comfort and reduce energy demand. Once the first-stage optimization converges, a black box model is used to optimally size a standalone rooftop solar system or a hybrid solar/wind energy system. The proposed platform was validated through simulations of an office building using real-world data. Results indicate that controlling the façade with a deep reinforcement learning algorithm enhanced the annual daylight glare probability and the predicted percentage of dissatisfied by 5.9% and 2.01%, respectively, compared to Opossum. However, the off-grid solar and hybrid systems designed using the Opossum solver achieved 13.73% and 11.11% lower levelized costs of electricity, respectively, compared to those designed by the deep reinforcement algorithm.