Proposing energy performance indicators to identify energy-wasting operations on big time-series data

Abstract

With the widespread of the internet of things (IoT) and smart metering, more and more time-series data are being collected. However, building energy managers cannot easily interpret the fluctuating energy data. There is a need for readable indicators of building operations to identify energy wasting with time-series data. Improving the management of energy-consuming equipment is a low-cost strategy to reduce building energy consumption. This study proposes three energy performance indicators (EPIs) for operations of office buildings with domain knowledge on time-series data. Then, for each building, these EPIs are calculated on its time-series energy data. To demonstrates the actual performance of the building among its peer buildings, these proposed EPIs are further rated by comparing with the values of its similar buildings. Based on the psychological motivation that once acknowledged that they were just a little behind of their peers, people would improve their behavior, this study calculates the energy savings of small improvement in operations for low-performance buildings with a linear method. This study is fulfilled on BDG2 datasets that contain not only the time-series data but also the meta-data of 1636 non-residential buildings. The results show that the proposed EPIs can uncover the energy performance of a building in different time segments. The case study of low-performance buildings demonstrates that the improvements of EPIs to medium levels can save 13.0% to 29.1% energy for according periods. The proposed EPIs can serve as an extension of energy benchmarking and help building managers to understand the energy efficiency of operations on time dimension intuitively.