Longitudinal analysis of energy metering data from non-domestic buildings

To identify, understand and reduce energy wastage in buildings, a significant indicator is time patterns of consumption linked to building occupancy. Advanced metering can log energy data at short half-hourly intervals or less. However, analysis of these data may still often follow traditional monitoring and targeting techniques developed previously for daily or weekly energy data. To explore the potential for advanced metering more fully, and to understand the energy consumption patterns and energy wastage in non-domestic buildings, a longitudinal study was made of energy data collected from approximately 300 buildings in Leicester, UK, between 2001 and 2008. This was the first such study of its kind. Evidence was gathered from gas, electricity and water meters, with water consumption being used as a proxy for building occupancy. Algorithms for cleaning the data are described. Four principal building failure modes for gas space heating were identified that cause excessive and wasteful energy consumption. In 2004, 34% of buildings were heated during unoccupied periods at weekends, although this reduced to 22% by 2008. Longitudinal analysis of night-time electricity baseloads showed an average annual rate of increase of around 8%, although wide variations exist between buildings. The method of identifying building control phenomena may be applied to any metered energy data set with a comparable sampling period.