Occupancy analysis in commercial building cooling energy modelling with domestic water and electricity consumption

Abstract

Occupancy is known to play a crucial role in building energy consumption, but its application in building energy calculations has been much simplified. In whole-building energy statistical models, occupancy is rarely considered if at all. One hurdle is that many buildings in which energy efficiency projects are implemented do not have occupancy sensors for pre-project measurement. This work explores a potential workaround for such buildings by comparing the consumption of domestic cold water and electricity as proxies for occupancy in building energy statistical models. They are tested using a clustering approach and a linear approach. Using the chilled water consumption data of a classroom building, the use of domestic cold water in the clustering approach was able to reduce the model’s coefficient of variation of the root mean square error (CV-RMSE) by 2.9 percentage points to an average of 11.3% from the already-good 14.2% produced by the traditional weekday-and-weekend method. The case study data also covered a period when the COVID-19 pandemic forced the building closure. This sudden change of utility consumption caused problems for all modelling methods and this study highlights the special care that must be taken to treat such events.