Data-driven approach for electricity consumption benchmarking of multi residential buildings

Building energy benchmarking is a crucial process for comparing a building's energy performance with that of others to determine its energy consumption profile and identify potential savings. This approach aims to enhance energy efficiency, provide a foundation for policy development, and encourage conservation efforts. However, current regulatory framework for building energy benchmarking primarily focuses on newly constructed buildings. The complexity of required paperwork and procedures also creates barriers to broader adoption. This study introduces a streamlined, data-driven framework for building energy benchmarking that utilizes publicly available data to simplify the process. Our framework integrates databases encompassing building information, electricity consumption data, and weather data, and incorporates models for predicting monthly electricity consumption and evaluating energy efficiency. By synthesizing these data sources, we developed a predictive model that estimates monthly electricity consumption, producing results that are used to calculate energy use scores and classify buildings into five energy-efficiency grades. We applied this model to evaluate 1,768 buildings in Seoul and Daegu, classifying 235 as Grade 1 (highest efficiency), 728 as Grade 2, 355 as Grade 3, 326 as Grade 4, and 124 as Grade 5 (lowest efficiency). Analysis of buildings with similar specifications in Seoul showed that older buildings tended to have lower energy efficiency than newer constructions, suggesting potential gains from retrofitting and other improvements. This framework, by providing clear explanations of energy efficiency grades and demand predictions, aims to assist building owners, managers, and policymakers in understanding energy performance patterns and identifying opportunities for energy conservation.