Optimizing building retrofit through data analytics: A study of multi-objective optimization and surrogate models derived from energy performance certificates

The building stock is responsible for a large share of global energy consumption and greenhouse gas emissions, therefore, it is critical to promote building retrofit to achieve the proposed carbon and energy neutrality goals. One of the policies implemented in recent years was the Energy Performance Certificate (EPC) policy, which proposes building stock benchmarking to identify buildings that require rehabilitation. However, research shows that these mechanisms fail to engage stakeholders in the retrofit process because it is widely seen as a mandatory and complex bureaucracy. This study makes use of an EPC database to integrate machine learning techniques with multi-objective optimization and develop an interface capable of (1) predicting a building’s, or household’s, energy needs; and (2) providing the user with optimum retrofit solutions, costs, and return on investment. The goal is to provide an open-source, easy-to-use interface that guides the user in the building retrofit process. The energy and EPC prediction models show a coefficient of determination (R${}^2$) of 0.84 and 0.79, and the optimization results for one case study EPC with a 2000€ budget limit in Évora, Portugal, show decreases of up to 60% in energy needs and return on investments of up to 7 in 3 years.