Drivers of variability in energy consumption

Meeting uncertain supply conditions with adequate demandside measures is becoming increasingly central to the day-to-day operations of energy utility companies, as variability is one of the main drivers of cost in operating the grid. In this paper we provide the first characterization of demand variability and the factors influencing it. We propose that heterogeneity in energy consumption is driven by users’ daily schedules, and mine a large sample of smart meter data from 30,000 residential users in CA („ 17M daily load profiles) to extract a small alphabet of typical schedules that people follow throughout the day. We next relate variability in consumption to features pertaining to the individuals and their neighborhoods using data on customer engagement with the utility company and block-level U.S. Census demographics. Our analysis shows that certain high-level attributes such as income level or tenancy situation are robust predictors of schedule stability.