Overview

Test Cell 1a (TC1a) contains half-hourly whole house electricity consumption data from October 2012 to September 2013 for 9201 domestic customers with basic smart metering. This data has been analysed to provide insight into typical domestic electricity use patterns and the relationship between demographic indicators and energy use. Additionally, TC1a is used as the control group or starting point against which the other test cells can be compared. The demographic composition of the participants in this test cell is representative of the UK population.

The dataset displays some expected behaviour:

• The daily energy use profile is broadly divided into night/early morning (low demand), daytime (mid-level demand) and evening (higher demand).
• In the winter months, overall consumption and average daily peak demand are higher than in other seasons. Also in winter there is a greater variability in energy use between customers.
• Most households have their peak daily demand during the 4-8pm period. However, there is a significant minority of households across all demographic groups whose peak demand occurs earlier in the day, around midday.

It is important to note when considering the population statistics below that variability in electricity consumption has been shown to be significant even within any specific group. This means that general trends reported below are not a prediction of the energy consumption of an individual household. However, this level of diversity in electricity consumption is a positive finding for networks as the heterogeneity of the customer base connected to a given network should produce a lower coincident peak.

Household income is the demographic attribute with the clearest impact on energy use, although the correlation is at best weak. On average, high income households (above £30k p.a.) have the highest overall energy consumption and highest average peak power demands, which is in line with broader industry understanding of customer profiles. On average, a high income household was found to consume:

• 40% more energy than a low (<£15k p.a.) income household.
• 15% more energy than a medium income (£15k-£30k) household.

In TC1a, over 34% of the energy was consumed by high income households despite this group forming only 29% of the sample. Additionally, for the high income group, the proportion of electricity use concentrated in the evening peak period (4pm-8pm) is higher than for any other group.

Noting that other correlations are weaker, the following demographic indicators were also found, on average, to be relevant when determining energy consumption:

• The existence of dependents (people aged below 5 or over 65) in the household was found to be correlated with lower overall energy consumption and peak demand.
• Home ownership (rather than renting) was found to be positively correlated with increasing energy consumption and peak demand.
• Living in a rural area was observed to correspond with higher energy use and higher demand peak compared to urban households.

However, it should be noted that the last two attributes above may largely be proxies/secondary to household income, because home ownership and rurality are arguably positively correlated with household income. Additionally, the analysis did not investigate whether the link between income and electricity use was, in turn, related to other characteristics such as household size or behaviour, and of course the trends identified do not necessarily hold for every individual household.

While electricity use was found to be linked to demographic indicators, other variables such as ambient temperature and time of the year have a much greater impact on electricity use, in particular when predicting the peak demand on any particular day.

As a comprehensive dataset of residential half-hourly electricity use, TC1a can be used for the following:

• The demand profiles can be compared with and potentially update standard consumption profiles currently used in industry for network planning.
• The demographic breakdown of the data could be useful when targeting the deployment of future interventions, for example to identify the customers most responsible for a particular network challenge (such as peak demand), or to understand the distributional impact of energy price tariffs

 

Back to Project Library