Analyzing energy consumption patterns of an educational building through data mining

Green building standards were adopted around the world to minimize energy consumption and carbon emission from the building sector. However, designing buildings following the standards does not guarantee a low-energy building. Actual energy consumptions in buildings were reported, on average, 2.5 times higher than predictions during the design stage, which is known as the performance gap. Most building operation managers do not have the tool and knowledge to understand the reasons for such differences. Moreover, current building management systems (BMS) do not have the intelligence to identify such differences and their sources. Understanding the root causes of the performance gap is even more difficult in an educational building because of its different usage patterns and various space types. This research aims to analyze the energy consumption patterns of different spaces in a mixed-use educational buildings and identify possible sources of energy waste using an unsupervised data mining approach. A 5-star Green Star-rated educational building in Melbourne, Australia, was considered for the case study. The results showed electrical and gas energy performance gaps of 2.4 and 3.1 times, respectively, in the studied building. Further analysis revealed that energy consumption during non-working hours was 48% of total energy consumption during the one-year studied period, which is very high and was one of the possible sources of waste. During the holidays, the mechanical system and plug loads ran as per the weekday operating schedule in an empty building, resulting in energy waste. Actual hourly lighting and plug load consumption profiles differed significantly from the predicted profile during design. Based on the findings, several recommendations were made to minimize the performance gap in an educational building.