Explaining daily total energy demand in British housing using linked smart meter and socio-technical data in a bottom-up statistical model

Abstract

This paper investigates factors associated with variation in daily total energy consumption in domestic buildings using linked pre-COVID-19 smart meter, weather, building thermal characteristics, and socio-technical survey data covering appliance ownership, demographics, behaviours, and attitudes for a sub-sample of 617 British households selected from the Smart Energy Research Laboratory (SERL) Observatory panel.Linear mixed effects modelling resulted in marginal/conditional R2 of 0.68/0.83 and root mean squared error of 17.7 kWh/day, for daily gas and electricity use combined. Increased daily energy consumption was significantly associated (p-value<0.05) with: households living in buildings with larger floor area, more rooms, that are older, have lower energy efficiency, and experience colder or less sunny weather; households with more adult occupants, more children, older adult occupants, fewer adults with qualifications, higher heating temperature setpoints, that do not try to save energy, and that do not put on more clothes rather than turning the heating on. The results demonstrate the value of smart meter data linked with contextual data for improving understanding of energy demand in British housing. Accredited UK researchers are invited to apply to access the data which has recently been updated to include over 13,000 households from across Great Britain.