Modelling retail store energy consumption for improved accuracy of consumption forecasting

PI: David Wallom

Department: Engineering Science

Traditional building energy usage estimation methods rely on engineering models. This methodology misses the main driver for consumption within any building: the occupants. The data driven method developed within the EPSRC WICKED project, as proof of concept, instead looks within the property portfolio of the user (or in future a repository of anonymised properties) for those most similar in location, type, internal layout and configuration and uses this to build a composite model. The model is created combining the historical one-hour energy consumption readings of the similar properties. The output is a predicted usage model along with a window of uncertainty.

The developed prototype will demonstrate how to gain a clearer understanding of possible energy consumption, overall and in profile, from new stores based on the operating model of current stores. It will also develop understanding of the sensitivity of an individual store consumption profile to these parameters.