Estimating residential load balancing potential

Shefali Khanna¹, Felix Bracht², Ralf Martin¹ and Mirabelle Muûls¹

¹Imperial College London, UK, ²London School of Economics, UK

Background

The deployment of smart meters has provided us with much more visibility into households’ energy use patterns. These devices also enable utilities to introduce various forms of dynamic pricing, which can be leveraged to incentivise consumers to shift their electricity consumption to the hours when prices are low that will typically also coincide with the times when there is more zero carbon energy available. The returns from dynamic pricing could be even larger for consumers that have made complementary investments in decentralised technologies such as rooftop solar and electric vehicles since these consumers are now less reliant on the grid to meet their energy demand, and can earn a large return from selling electricity to the grid when prices are high. Finally, to the extent these pricing incentives also reduce peak demand, they could significantly lower the average cost of power procurement since utilities may forward contract large amounts of generation to meet the expected demand of only a small fraction of the hours in the year. However, if households are inattentive to energy prices, there may be a need for smart technologies that enable automated demand management, where households’ energy use responds automatically to price signals. Small changes in energy demand facilitated by automation technologies could result in sharp supply cost reductions. In this project, we focus on measuring the underlying load balancing potential of households so that we can decide how to better target both pricing incentives and technologies that facilitate automated energy demand management.

Methods

This study combines detailed half-hourly electricity and gas smart meter data for several thousand residential consumers in the UK with hourly ERA5-Land data, which covers a range of weather variables, including temperature, precipitation, wind speed and relative humidity. We use the data to estimate household-specific regression models of electricity and gas consumption as a function of weather separately by season and time-of-day. Premised on the assumption that the demand for heating is driven by weather, we use the variation in electricity and gas consumption explained by weather as a measure of the load balancing potential of households. We show how this potential varies by property value, floor area, location and the type of heating system the households use.

Results

The results suggest that weather fluctuations explain more of the variation in gas consumption than the variation in electricity consumption and is especially pronounced in the autumn and spring seasons. Residential load balancing potential appears to be largest for more energy-intensive households and concentrated during the hours when more members are present at home. We are also using the smart meter data to disaggregate demand by individual end uses and to examine how the potential for demand flexibility will evolve when heating is electrified.

Conclusion

This study puts together a detailed picture of the conditions under which households are more responsive to changes in weather in order to assess where the largest gains can be made through load balancing. Our parameter estimates can also be used to inform how to target long-term investments in electrification vs short-term incentives for demand flexibility.