The potential for energy reduction through building retrofit projects is huge in the U.S. This research aims to devise a novel method for optimizing post-retrofit building performance in future climates. The proposed method is implemented on Penn campus buildings. The proposed decision-making support framework is demonstrated by its robustness to the problem of deep retrofit optimization and can provide support for brainstorming and enumerating various possibilities during the process of decision-making. The study is structured into three sections:
a) A data-driven approach for projecting future hourly energy use, considering global climate change (GCC). Machine learning models predict future energy consumption, revealing how GCC influences optimal energy conservation measures.
b) Development of SimBldPy - a simplified building performance simulation tool based on dynamic hourly algorithms. This tool offers rapid modeling and assessment of various energy conservation measure options.
c) Implementation of a non-dominated sorting technique for multi-objective optimization, complemented by a visualization schema to aid decision-making. By integrating the simplified simulation model and random forest models, deep retrofit problems can be swiftly optimized. The method's efficacy is demonstrated through a case study on a Penn campus building, showcasing its robustness in deep retrofit optimization and its potential to guide decision-making processes.