Data-Driven Research on Energy-Efficient Retrofit and Multi-Objective Optimization of Urban Building Clusters

Abstract

Urban energy-efficient retrofit is an important path to reduce energy consumption and carbon emission. However, balancing environmental and economic benefits and making choices among numerous retrofit packages is challenging for decision-makers. This study proposes a data-driven framework that integrates physical UBEM and machine learning model to evaluate the energy retrofit performance of urban building clusters and assists decision-makers in rapidly selecting the optimal energy-efficient retrofit packages through NSGA-II. The feasibility of the proposed framework is validated using the building clusters of Sipailou campus in Southeast University as a case study and identified 25, 19, and 13 optimal retrofit packages for office, public and education building clusters that minimizes the total energy consumption and carbon emissions while maximizing economic benefits from 343 retrofit packages of each cluster in a 30-year retrofit period.