Surrogate-assisted coordinated design optimization of building and microclimate considering their mutual impacts
Zeming Zhao, Hangxin Li, Shengwei Wang
Abstract
Building energy performance and pedestrian thermal comfort are strongly related and crucial in urban development. However, the mutual impacts of building design and local microclimate are often not quantitatively considered in design optimization due to their complex quantification. This study proposes a coordinated design optimization method that enables simultaneous optimization of both a building and its microclimate within a practically affordable time frame through an effective quantification method. SVR-based local air temperature and LightGBM-based local wind velocity surrogate models, along with automated building simulations, are integrated with the optimizer to enhance efficiency and generalizability. A total of eleven essential building design variables are optimized to minimize both total building energy consumption and pedestrian thermal discomfort. Global optimal solutions (i.e., Pareto front) identified by NSGA-II are evaluated using the entropy-TOPSIS method to determine the best solution. The proposed method is validated through a case study of a mixed-use building in Hong Kong. Results indicate that using the surrogate-assisted coordinated optimal design method can reduce the total building energy consumption by up to 63.34 % and pedestrian thermal discomfort degree by up to 1.88 K in subtropical regions. Additionally, computation time for design optimization is reduced by 99.98 % (i.e., from 42,684.44 to 8.89 h) compared to conventional simulation methods. This study fills a critical gap in the simultaneous design optimization to enhance building energy performance while balancing local microclimate impacts efficiently. • A coordinated design optimization method is proposed for a building and its microclimate. • Mutual impacts between building design and local microclimate are considered. • Pareto optimal solutions balance building energy efficiency and pedestrian thermal comfort. • Energy consumption reduced by up to 63.34 %; thermal discomfort decreased by 1.88 K. • 99.98 % reduction in computation time for surrogate-assisted design optimization over conventional simulation methods.